22
99 ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99–120 Keywords – Fire evacuation; layout improvement; shape grammar; station design INTRODUCTION Underground stations can broadly be divided into two methods of construction: cut and cover (i.e. excavating a trench and recovering) and tunnelling. The station layout is mostly comprised of the areas that are common to all stations and their spatial planning must be set out with a thorough understanding of passenger circulation patterns. The station components, namely the ticket hall, escalators, concourses, platforms, etc. all interact and connect in a coherent, logical manner with the principal aim of transporting customers to/away from trains as quickly and efficiently as possible. However, achieving this aim must not be detrimental to passenger safety, which is paramount in station design (Smith et al, 2003). London has the largest, oldest and one of the most complex underground railway systems in the world, dating back to 1863. It has an excellent passenger safety record, with its most notable tragic events taking place at the King’s Cross fire in November 1987 and in the terrorist attacks of July 2005. The standards and regulations through which London Underground (LU) operates its fire strategies for station design approvals are as follows: LU Station Planning Standards and Guidelines (SPSG, 2007) Building Regulations (exempt, but still meet functional requirements) 1-080: The application of fire safety engineering principles to LU premises 1-081: The design and installation of fire protection systems Abstract Spatial layout planning is an important part of the underground railway station design process, taking into account its need-driven nature and the resulting infrastructure that is sized on demand. The predicted passenger flow rates are the underlining factor and are divided into ‘levels of service’ for space planning considerations. This paper reports on the development of a knowledge-based computer design environment capable of generating multiple spatial layout solutions, thus providing for more effective fire evacuation analysis when compared with the traditional design process. The developed programme, titled SGEvac in this research, utilizes shape grammar theory to provide for automatic generation of solutions at the reference (preliminary) design level, based on visual rules of shape recognition and replacement, their connectivity and spatial relationships. Although it has been developed to meet London Underground Station Planning Standards and Guidelines (SPSG) and related codes of practice, it has both the scope and potential for redevelopment to any other country’s design legislation. Novel shape grammar and functional logic design rules that incorporate station planning design knowledge and guidance are developed and specified, along with the theoretical research. Validation of the results thus far is discussed, with a ‘train on fire in a station’ evacuation scenario analysed. ARTICLE Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation Graham Smith and Boris Ceranic doi:10.3763/aedm.2008.0078 © 2008 Earthscan ISSN: 1745-2007 (print), 1752-7589 (online) www.earthscanjournals.com

Spatial Layout Planning in Sub-Surface Rail Station …fbe-studiocollaboration.unsw.wikispaces.net/file/view/UG+Station... · common to all stations and their spatial planning must

Embed Size (px)

Citation preview

99

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Keywords ndash Fire evacuation layout improvement shapegrammar station design

INTRODUCTIONUnderground stations can broadly be divided into twomethods of construction cut and cover (ie excavatinga trench and recovering) and tunnelling The stationlayout is mostly comprised of the areas that arecommon to all stations and their spatial planning mustbe set out with a thorough understanding of passengercirculation patterns The station components namelythe ticket hall escalators concourses platforms etcall interact and connect in a coherent logical mannerwith the principal aim of transporting customerstoaway from trains as quickly and efficiently aspossible However achieving this aim must not bedetrimental to passenger safety which is paramount instation design (Smith et al 2003)

London has the largest oldest and one of the mostcomplex underground railway systems in the worlddating back to 1863 It has an excellent passengersafety record with its most notable tragic eventstaking place at the Kingrsquos Cross fire in November 1987and in the terrorist attacks of July 2005

The standards and regulations through whichLondon Underground (LU) operates its fire strategiesfor station design approvals are as follows

LU Station Planning Standards and Guidelines(SPSG 2007)

Building Regulations (exempt but still meetfunctional requirements)

1-080 The application of fire safety engineeringprinciples to LU premises

1-081 The design and installation of fireprotection systems

AbstractSpatial layout planning is an important part of the underground railway station design process taking intoaccount its need-driven nature and the resulting infrastructure that is sized on demand The predictedpassenger flow rates are the underlining factor and are divided into lsquolevels of servicersquo for space planningconsiderations This paper reports on the development of a knowledge-based computer design environmentcapable of generating multiple spatial layout solutions thus providing for more effective fire evacuationanalysis when compared with the traditional design process The developed programme titled SGEvac in thisresearch utilizes shape grammar theory to provide for automatic generation of solutions at the reference(preliminary) design level based on visual rules of shape recognition and replacement their connectivity andspatial relationships Although it has been developed to meet London Underground Station PlanningStandards and Guidelines (SPSG) and related codes of practice it has both the scope and potential forredevelopment to any other countryrsquos design legislation Novel shape grammar and functional logic designrules that incorporate station planning design knowledge and guidance are developed and specified alongwith the theoretical research Validation of the results thus far is discussed with a lsquotrain on fire in a stationrsquoevacuation scenario analysed

ARTICLE

Spatial Layout Planning in Sub-SurfaceRail Station Design for Effective FireEvacuationGraham Smith and Boris Ceranic

doi103763aedm20080078 copy 2008 Earthscan ISSN 1745-2007 (print) 1752-7589 (online) wwwearthscanjournalscom

1-082 Active fire protection systems and portablefire extinguishers

1-083 Passive fire protection systems 1-084 Maintenance of fire protection systems 1-085 Fire safety performance of materials 1-066 Emergency lighting HMRI Railway Safety Principles and Guidance

Part 2 Section B Regulatory Reform (Fire Safety) Order 2005 BS 5588 Fire precautions in the design

construction and use of buildings BS 7974 Application of fire safety engineering

principles to the design of buildings BS 8300 Design of buildings and their approaches

to meet the needs of disabled people

The figures for predicted passenger flow rates areobtained from Transport for London (TfL) According toTfL 2007 figures London Underground had over 1 billionpassenger journeys for the first time in its history whichis more than the entire national rail network Thus theresearch to improve both the evacuation and normalcirculation routes with regard to emergency evacuation(means of escape travelling distances position of fireexits circulation etc) could have an important benefitwhen designing large interchange stations under majorrefurbishment or new build consideration

THEORY AND APPLICATION OF SHAPEGRAMMAR The concept of shape grammar formalism is primarilyattributed to Stiny and Gips (1971) with majorcontributions made by Stiny (1977 1980) Flemming(1987) Knight (1995 1999) Chase (1996) Tapia(1999) Li (2003) and Liew (2003)

Shape grammar is a method of describing andcreating languages of designs where shapes aredevices for visual expression Their composition iscontrolled by preconceived rules in a very similar wayas to how grammar is used to arrange written wordsinto sentences The theory provides an approach forthe generation of composite shapes based onprinciples of recognition and replacement of aparticular shape Primitive shapes are transformedand combined to create more complex compositesthrough the application of predefined rules and soshape language emerges (Cha and Gero 1999)

Loomis (2003) demonstrates an exercise of a two-dimensional shape grammar with one rule The shapeon the left-hand side is replaced by the shape on theright which following recursive application leads tothe generation of simple shape grammar designs asshown in Figures 1a and 1b respectively

As Loomis demonstrates further it is possible tohave the shape rule in Figure 1a applied in fourpositions (see Figure 2) which would lead to differentshape grammar designs To remove any uncertaintyassociated with rule application it is necessary tointroduce a deterministic approach where shapegrammar rule application is managed by a set oflabels These labels (or rules) constitute a set fromwhich alternative designs are derived Figure 3 showssome of the possible outcomes that have beenderived from the rule set specified in Figure 2

Tapia (1999) explains how shape grammarsoriginally defined by hand lsquonaturally lend themselvesto computer implementation [handling] therepresentation and computation of shapes rulesgrammars and the presentation of alternativeshellipthedesigner specifies explores develops the languageand selects [generated] alternativesrsquo

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

100 G SMITH AND B CERANIC

(a) (b)

RULE 1 RULE 2

RULE 4RULE 3

FIGURE 1 A shape grammar rule and its recursive applications

(Loomis 2003)

FIGURE 2 Labelling shape rule for deterministic grammar

designs (Loomis 2003)

Stiny and Gips (1971) define shape grammar as

S = [VT VM R I] (1)

where VT is a finite set of shapes VM is a finite setof labels R is a finite set of shape rules and I is aninitial shape

In this research parameter VT is the finite set ofthe station infrastructure components shapes andVM is the finite set of labels that comprise thepossible station area permutations mapping twolsquolegally connectedrsquo components into a spatial relationwithin the overall layout R is the finite set of theshape connectivity rules composed of informationcontained in regulations and guidance documents interms of the area assignments adjacency and theirconnectivity I is an initial shape (ticket hall in theSGEvac programme)

In summary a proposed shape grammar systemconsists of a set of vocabulary elements (stationcomponents) a set of rewrite rules and a start stateStation layout can be considered in terms of compositeshape generation and its design process in terms oflogical sequence and deductive reasoning It is thisapproach based on the function of the building thatallows for a transition from lsquotraditionalrsquo layout planningto the computational shape grammar layoutgeneration that is proposed in this research Howeverits application depends directly upon the expertrsquosknowledge filtered into this rule-based approach

An additive shape grammar approach (Knight1999) is proposed as a suitable technique for

generating station layouts This approach is useful forthe irregular station boundaries derived from sitespecific conditions where designs are generatedfrom a core (the ticket hall) to which othercomponents are successively added

Knight (1995) remarks that the rules detailed inStinyrsquos Kindergarten paper (1980) can be preacutecised toexternalize a studentrsquos design ideas so that they canbe examined changed and communicated morereadily It is the choice of credible shape grammaralternatives that is important not the multitude ofsolutions themselves Good analytic grammars areboth efficient and descriptive revealing simplicity orregularities behind seemingly complex designs

STATION LAYOUT DESIGNCONSIDERATIONSPassenger space planning needs to be considered interms of peoplersquos procedures when they use theUnderground Initially customers will enter thecomplex from either street level or via a surfacestation stairwell A one- or two-way passage will thentake passengers to the unpaid side of the ticket hallwhere they will purchase a ticket before passingthrough to the paid side

Customers then descend using stairsescalatorslifts to platform level At both the top and bottom ofstairs and escalators there is a station componenttermed the run-off area The length of the run-offdepends on the peak passenger flow capacity of theescalatorstairwell and whether it connects to aplatform passageway or street It is an area wherepeople move away from the change in level orientatethemselves and make a decision of where to go next(SPSG 2007)

Generally at platform level people will journey tothe desired line to take them to their destination via afurther passageway or lobby area Some escalatorsstairs can exit directly onto platforms The travelprocess is reversed when alighting from a train andcontinuing to the surface

The predicted passenger flow rates and sitespecific conditions dictate both the number of railwaylines and the platform(s) configuration (side centre orstacked platform) Station composition must alsoconsider areas that are not sized by passengervolume demand but are required to ensure that the

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 101

INITIAL SHAPE

INITIAL SHAPE

RULE 1 RULE 1 RULE 1 RULE 1

RULE 2RULE 2RULE 2RULE 3

FIGURE 3 Labelling shape rule for deterministic grammar

designs (Loomis 2003)

station operates satisfactorily on a daily basis Thestation managerrsquos office plant and staff roomscomputer rooms train technician rooms etc all needaccommodating in the proposed shape grammarformalism In addition ancillary accommodationsuch as the cleanerrsquos store police room publicitystore retail etc also have to be catered for

To aid designing and sizing infrastructure levelsof service as shown in Figure 4 are used by LondonUnderground These are based on the work of Fruin(1971) where a detailed study of crowd movementwas undertaken Level A represents free circulationwith subsequent levels growing progressively morerestricted in terms of people movement from B to Eculminating with level F that represents a completebreakdown in flow with numerous stoppages

The approach proposed in this research definesthe generic station layout arrangements from thelsquohigh-levelrsquo spatial site organization to the subsequent

specified components and their logical spatialallocations as illustrated in Figure 5 The consequentshape grammar rules can then be deduced taking intoaccount the predicted passenger flow rates fire safetystrategy and relevant design regulations

FIRE SAFETY DESIGN STRATEGYBefore any design commences a clear identificationand appraisal of the key constraints and parametersat the site layout planning stage is required Forexample deep foundations mains servicesarchaeological discoveries and even the RiverThames will affect station design decisionsFurthermore structural appraisals geotechnicalsurveys Statutory Undertaker searches groundcontamination testing environmental impactassessments and land acquisition etc all need to becarried out as part of a feasibility study before stationlayout planning at reference level commences

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

102 G SMITH AND B CERANIC

Level of service A

Level of service D Level of service E Level of service F

Complete breakdown in trafficflow with many stoppages

Free circulationDescription (forqueuing areaswalkways andstairways)

Uni-directional flows andfree circulation Reverseand cross-flows withonly minor conflicts

Restricted circulation formost pedestriansSignificant difficulty forreverse and cross-flows

Restricted circulation for allpedestrians Intermittentstoppages and seriousdifficulties for reverse andcross-flows

Sligthly restricted circulationdue to difficulty inpassing others Reverse andcross-flows with difficulty

Level of service B Level of service C

FIGURE 4 London Underground levels of service (SPSG 2007)

Such constraints may be viewed in a positivecontext in terms of services already being providedhelping to justify the business case for a new stationDesign procurement strategies may include decisionssuch as locating toilet areas within x metres of anexisting sewer route or locating electrical plant roomswithin y metres of an existing electrical sub-station

In the real world parameters such as these constrict layout production and the choice ofinfrastructure component placement as opposed toleading to the production of many solutions that arelikely to be discarded when for example projectcosts are reviewed This level of regularity andrestriction typical in station design is very suitable

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 103

SITE

CUT AND COVER STATION TUNNELLED STATION

TICKET HALL

UTS GATELINE (gateline dimensionsdetermine minimumticket hall width)

PASSAGEWAYS ONE- or TWO-WAY

(entrances exits determinea minimum uniform width)

ESCALATOR

STAIRS

RAMPS

TOP RUN-OFF

TRAVELATOR

PAID CONCOURSE SIDE

BOTTOM RUN-OFF

PASSAGEWAYS ONE -WAY or TWO-WAY

NON-PASSENGERSPACE ELEMENTS ATTICKET HALL LEVEL (egStation Managerrsquos Office)

PLATFORM LEVEL

SPACE ELEMENTS ATPLATFORM LEVEL (egplant rooms toilets etc)

SIDE PLATFORM CENTRE PLATFORM STACKED PLATFORM

No OF RAILWAY LINES

MEZZANINEOptional

MEZZANINE

DEEP

SHALLOW

LIFTS

NON-PASSENGER

FIGURE 5 lsquoHigh-levelrsquo layout hierarchy for generic station composition

for the application of a prescriptive shape grammarapproach

In parallel with the shape grammar developmentfunctional logic statements are proposed analogousto the qualitative design review as described in BS7974 Application of Fire Safety EngineeringPrinciples to the Design of Buildings The intention isto develop SGEvac procedures in a manner that isrecognized by fire safety engineers and designers inpractice The Department for the Environment (DOE)(1996) states that lsquothe architectural layout can have asignificant influence on peoplersquos movementComplexity in circulation routes may not be overcomenecessarily by signage these may not have beennoticed or they may have been forgotten as a resultof stressrsquo Hence station spatial requirements can bereviewed alongside a number of functional logicstatements devised in conjunction with the layoutplanning

For example lsquothe probability of successful escapethrough smoke diminishes sharply with (an increasein the) number of decision elements included in theescape pathrsquo (Hesketad 1999)

A functional logic statement response to theabove would be to minimize such decision elementsthus enabling faster egress Other examples offunctional logic statements employed are

no corridor or staircase will reduce in width alongescape routes

length of the passages are to be minimized maximize level areas to facilitate faster travel

speeds for escapees maintain uniformity to aid way-finding and relative

orientation evacuation and normal circulation routes are to

be the same exits will be prominent easily identifiable and

straightforward to reach

Such statements are intuitive by their nature and arepart of the qualitative reviews in the traditional designprocess Due to time constraints only a limitednumber of possible alternatives in terms of the spatiallayout improvements are ever considered if any Incontrast SGEvac offers not only the opportunity toconsider a significant number of possible alternatives

but also the means of quantitative assessment via fireevacuation simulations

Garling et al (1983) contend that newcomers to anenvironment first learn a number of salient locationsthen the paths in between and finally organize theacquired knowledge in a spatial system Indeedavoiding symmetry and lack of differentiation fromthe userrsquos viewpoint is a good practice for stationdesign considering perception of relative orientationin the station This is where interaction between thedesigner and computer program becomes essentialand it is in this context that the research argues therole of SGEvac as a computer-aided design tool notas a substitute for the designer

A comparative methodology overview is given inTable 1 The processes of BS 7974 the procurementof a general logic model (Galle and Kovacs 1992) andthe shape grammar development stages arebenchmarked against SGEvac procedures Thisperhaps has most relevance to fire safetyengineering since it is with this objective that theproposed shape grammar system is to be used inpractice

Research studies such as one by Magnusson et al(1995) note that fire safety engineering has motivateddesigners and regulators to examine alternative codecompliance approaches based on a greater relianceupon quantitative analyses This has led toreassessment of prescriptive approaches andconsequent adjustments as required It is nowaccepted that existing prescriptive practices shouldbe evaluated by quantitative performance-basedmethods as a part of the overall fire safety strategyas encouraged in the Regulatory Reform (Fire Safety)Order (RRFSO) 2005 SGEvac follows this trendaugmenting the array of tools available forperformance-based fire safety engineering design

There is obviously an inevitable need for designerinput and control at all stages of the design processDue to the potential necessity for deviation fromstandards via fire safety engineering judgement theproposed shape grammar approach cannot alonemodel project-specific concessions

Still the movement away from prescription viaBS 7974 and the RRFSO to risk-based judgementplaces computer-aided design tools such as SGEvacin a position of interdependency with other design

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

104 G SMITH AND B CERANIC

criteria neither can be formulated in isolation Thisreinforces the view that computers will in future playan ever-increasing role in aiding and enabling thedesign teams but will never be able to replace them

SYSTEM DEVELOPMENTFigure 6 describes the modular approach proposedfor the system development offering the addedpotential of dual-purpose application in the designprocess It allows the designer to use the systemeither as a whole or to take advantage of itsindependent modules The system offers designerscomplete control over the shape grammarcomposition and layout modifications together withthe constant monitoring facilities and ability todiscard any solutions deemed to be lsquoinferiorrsquo

The proposed system is partitioned into threemain modules ndash lsquoexpert knowledgersquo lsquolayoutrsquo andlsquotestingrsquo modules The expert knowledge modulecaptures the expertise available to facilitate accurateshape grammar generations of layout design Thismodule provides the designer with the ability tointeractively reference Station Planning Standards

and Guidelines (and other relevant fire safetyregulations) allowing himher to make necessary site-specific decisions and to control the lsquohigh levelrsquo shapegrammar composition rules

The designer specifies passenger flow rates anddesignates the overall site boundary dimensions thatdefine the permissible station footprint The final exitpositions are also stipulated at this stage with theseoften being constrained by the existing builtenvironment

With this data supplied the shape grammargenerations commence in the layout module Thearea assignments adjacency requirements and layoutcomposition rules are defined in detail at this stageOnce the designer has made decisions with regard tothe layout composition the computer becomes thecentral tool for generating searching and logicallysorting through the permissible permutations ofpossible layouts The user has the option to auto-generate layouts (usually for sites without manyconstraints) or to manually control the layoutassembly (a decision normally associated with heavilyconstrained sites)

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 105

TABLE 1 Methodology comparison BS 7974 PROCESSES GALLE AND KOVACS SHAPE GRAMMAR SGEvac

LOGIC MODELLING DEVELOPMENT IMPLEMENTATION

Assess benefits Identification Shape grammar Assess constraints

and constraints system identification identify benefitssite

boundariesfinal exit

positions

Qualitative design Conceptualization Initial state Station component

review identification and

interactions

Quantitative analysis Formalization Rules and shape Legislative compliance

language passenger flow figures

area assignments

Satisfactory Implementation Shape grammar Alternative station

unsatisfactory executions layouts generation

Report and review Testing Design analysis Testingvalidation

iterationsevolution of layouts generation

End Experimentation Experimentation

iterationsimprovements

The testing module is used to evaluate potentiallysuitable layouts at reference level by exporting them toAutoCAD via standard drawing data exchange formatand analysing them in the context of project-specificsite constraints and requirements A level of drawingdetail is then added and the layout solution is exportedto Simulex a building evacuation simulation software

package Once the occupant numbers and persongroups are assigned and the escape and verticalcirculation routes defined the simulation software iscapable of the accurate measurement of all travellingdistances creation of distance maps and calculation ofevacuation times for different fire scenarios (egdifferent group affiliations persons exiting via different

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

106 G SMITH AND B CERANIC

EGY D ESIGNDesigner

Feasibility studiesProject brief requirements

Passenger flow data

DDE

DDE

Layout Production

Shape grammar rulesMultiple layout production

Layout Analysis

Layout interrogationsDesign team judgement

No

Yes

DXF

DXF

Yes

No

AutoCAD

Feed lsquoblock reference planrsquoresults into overall project

Improvedsolution identified

EXPERT KNOWLEDGE MODULE

Fire Strategy Design

LU SPSG4FSE legislation amp guidance

Infrastructure size calculationHigh level brief development

LAYOUT MODULE

Layout Composition

Area and adjacency planningLayout option appraisal

Layoutimproved

TESTING MODULE

AutoCAD

SGEvac export into CAD

Simulex

Fire evacuation scenarios

DesignerSGEvac

DDE ndash Database Data Exchange

DXF ndash Standard Drawing Data Exchange

FIGURE 6 Proposed SGEvac algorithm

routes and having delayed movement in response to afire alarm signal) The design team then assesses theresults against the qualitative fire safety design goalsset earlier to establish whether improvements toevacuation times have been made

DETAILED ANALYSIS OF STATIONCOMPONENTSIn conjunction with the site boundary final exitpositions and passenger flow specificationconsideration must also be given to the vertical sitedimension The shape grammar system assumes allfinal exits are at ground level with the ticket hall andplatform(s) being the other two elements that requirethe user to determine their depth relative to the finalexits In so doing SGEvac selects the correctinterchange components by an event-drivenmechanism that evaluates the height change andpassenger volume demand Thresholds for bothcriteria have been defined as shown in Table 2 andEquations 2ndash5

If H (height in mm) = 0 ndash 500 then I (infrastructure) = Ramp (2)

If H = 500 ndash 3000 and P (passengers) le 3000 then I = Stairway (3)

If H = 3000 ndash 5000 And P le 3000 then I = StairwayElse I = Escalator (4)

If H gt 5000 and P le 1000 then I = StairwayElse I = Escalator and Lift (5)

(where lsquoElsersquo denotes the mechanism for the choiceof the alternative infrastructure component if theoriginal condition is not satisfied)

The above constructs also condition the run-offdimensions associated with each of theinterchange components (between 4 m and 12 mas stipulated by SPSG) This being the case run-offs are used to size the ticket hall length Figure 7shows the paid side of the ticket hall with variabledistances shown between the escalator or staircaseand the gateline

The paid side of the ticket hall is always dictatedby run-off distance while the unpaid side is governedby two parameters the run-off and the concoursearea required for the heaviest 15-minute passengerflow If the maximum 15-minute passenger flowfigure multiplied by 1 m2 (say answer A) is larger inplan area than the product of the gateline widthmultiplied by the relevant run-off (say answer B) thenanswer A will be incorporated into the layout Thereverse is true if B is the greater number This ensuresthat the station is always sized appropriately

Figure 8 shows a ticket hall layout with thegateline dictating the width (denoted as Tick_H_wd)itself conditioned by the passenger flow figurescalculated as shown in Table 3 The table onlyevaluates entry flow requirements but SGEvac willrepeat the same process for exit gates to produce acorrectly sized gateline

In this case the maximum 15-minute passengerflow capacity size prevailed over the run-off to createthe unpaid side overall

The divide between the paid and unpaid side ofthe ticket hall is the sized according to Table 4

SGEvac shape rules mandate that entry gates arealways located adjacent to the ticket office queuingspace (designated as Q_TIW) to avoid conflictingpassenger entry and exit flows The LU standard is to

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 107

TABLE 2 Interchange infrastructure selectionINFRASTRUCTURE REASON FOR SELECTION

Ramp Vertical rise between 0ndash05 m preferred to steps due to disabled access issues generally considered

people will travel faster on ramps than up or down steps

Stairs Vertical rise between 05ndash5 m but greater height available for selection if station has a light passenger flow

Escalator Vertical rise is minimum of 3 m for heavy passenger flow stations but used for lightmedium flow stations

above 5 m

Lift Vertical rise above 5 m also required for accessibility of disabled persons Lifts always used with escalators

not as a substitute

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

108 G SMITH AND B CERANIC

Con_Area

Exit

UTS gates

6601900 2450

2400

EFW

Ticketoffice

POMcounter

4000

Tick_H_wd

1000

Run

-off

Entry

Q_space

Q_TIW

FIGURE 8 SGEvac ticket hall arrangement

Heavy flow escalator

Escalator

Staircase

Medium to Heavy flow

Medium to Heavy flow

Light flow escalator

Light flow

Heavy flow 12mMedium flow 8ndash12m

Light flow

Light flow 8m

Light flow 6m

Heavy flow

staircase to gateline

staircase to gateline 10m

Light flow

to gateline12mGateline

Gateline

EscalatorndashGateline

Total Run-off

Heavy flow 10mMedium flow 6ndash10m

StaircasendashGateline

Total Run-off

to gateline 8m

2m 4m 6m 8m 10m

2m 4m 6m

6m

8m 10m

12m

FIGURE 7 Passenger flow rate run-off requirements (SPSG 2007)

allow for 4 m (or a five-person queue) as shown inFigure 8 The overall gateline is defined by summingboth exit and entry gate numbers obtained from user-inputted passenger figures

The length of the ticket office interface with theticket hall is stipulated by the number of ticket issuingwindows (TIWs) plus excess fare windows (EFWs)TIWs always belong on the unpaid side of the tickethall and EFWs always on the paid side both forrevenue protection reasons This situation facilitates aprecise shape grammar labelling for the ticket officeto ticket hallqueue space connection

Passenger operated machine (POM) enclosures arenormally connected to the ticket hall ticket office andits own queue space (noted as Q_space in Figure 8)

To make shape rule specification manageable allstation areas are divided into groups arranged to suitthe most probable connection patterns in the stationTable 5 shows the group station managerrsquos officeexample with all other station components followingthe same concept

London Underground views its stations as largebasements (Weller 2007) and escape distances fromplatforms are based on agreed dispensations with theLondon Fire Brigade SGEvac respects thesedistances by placing restrictive labels on the platformso that they cannot be exceeded as demonstrated bythe example shown in Figure 9

SGEvac escape stair infrastructure may bepositioned up to 20 m from the end of platforms toavoid undesirable lsquodead endrsquo conditions The 45 mmaximum platform travel distance starts from thenearside junction of the two 33 m wide escapestaircases with the 45 m distance rule creating ashortfall in the platform centre of 34 m Generallylocating three interchange components (eg stairs

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 109

TABLE 3 Gateline width calculationMETHOD WORKED EXAMPLE

Obtain peak hourly flow rate through Say 6500 peoplehr

ticket hall

Calculate 5-minute peak passenger flow 650012 = 541

Allowance for ticket rejection 541 + 20 = 561

Allowance for growth in use demand 561 x 110 = 617

Assume 125 people use each gate over

the 5-minute peak passenger flow period 617125 = 493

Round up gates to nearest whole number 5 gates required

TABLE 4 London Underground ticket hall gateline parametersregulations (SPSG 2007)EQUIPMENT DIMENSIONS (W X D) POSITIONARRANGEMENT

Combined manualequipment gate 1750 x 270 mm Height ndash 1300 mm Adjacent to ticket office Can include

1400 mm clear opening luggage facility

Manual gate ndash passenger 1750 x 270 mm Height ndash 1300 mm Adjacent to ticket office With more than 6

gateluggage port 900 mm ndash clearance and walkways in a single gateline or significant

outward opening interchange with mainline stations airport or

tourist traffic provide separate luggage chutes

Manual gate ndash single leaf 1394 mm Height ndash 1300 mm Beside the dedicated excess gates

with luggage port 695 mm clearance

Gateline attendance report 990 x 1240 mm Within the controlled zone but not necessarily

part of the gateline Usually close

Station control unit (SCU) 432 x 430 mm Should be installed within a secure enclosure

(which could be a ticket office) and also with

a clear view of the gates Controls and monitors

the station underground ticketing system

equipment

Manual emergency exit gates 1900 x 125 mm Adjacent to ticket office Add these gates rather

than adding extra UTS gates for evacuation

purposes

or escalators) at these distances will result incompliance with the Underground standards but itmay not always be possible (or practical for aninterchange situation) to procure such a layout

Lengths of station components are controlled bya range of criteria For the platform it is the length ofthe train vehicle for interchange components it isthe vertical height in the terms of the depth to theplatform level for passages it is the designated finalexit positions and the existing station areas alreadygenerated

Vector direction labels are underpinned by thepermissible union of compatible spaces two points ofcontact will be maintained so as not to restrict thewidth of connecting passenger areas This undueconstriction of passenger flow at the head of apassage also satisfies realistic construction practiceNon-passenger and ancillary area vector directions areas per project requirements and the successive areaadditions managed by the SGEvac rule groupings

To create alternative layouts SGEvac uses alsquorandomizersquo rule In this shape grammar rule initialstate infrastructure is generated as per guidelinesbut it can randomly change the location ofcomponent arrangements within the predefined

boundaries so that a number of possible alternativelayouts are produced The designs can then beinterrogated and ranked by their travelling distancesand against the qualitative functional logicstatements

The SGEvac design tool developed in this researchis based on a highly restrictive additive grammar ruleset due to SPSGrsquos prescriptive guidance and thenature of the layout planning problem As Grimsdaleand Chang (1996) argue many published grammarslack high-level reasoning during the iteration processsince they are often concerned with the developmentof shape grammar theories rather than their practicalapplications By contrast SGEvac being a practicalapplication has a complete set of primitive shapesonly possible to arrange in a certain way It thusprovides viable reference design layout compositionsbased on SPSG with further interventions possible bythe designer

SYSTEM DEVELOPMENT TO DATEThe SGEvac prototype is compiled within the Clanguage environment allowing for modular object-oriented programming (suitable for shape grammar)Figure 10 shows the main interface where the stationname and relevant LU fare zone are entered (whichimpacts on sizing infrastructure) as are the ticket halland platform depths the overall site boundaries andfinal exitsrsquo positions The peak passenger flow figuresover the relevant time interval of normally 3 hours arealso specified SGEvac is then capable of accuratelycalculating all passenger area infrastructure based onthe SPSG guidelines as already discussed

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

110 G SMITH AND B CERANIC

TABLE 5 Station manager suite ndash shape rule connectionsELEMENT SHAPE RULE DESIGNATION CONNECTED COMPONENT GROUP ALLOCATION

Group station managerrsquos office SR_GSM_01 Ticket hall paid side GSM

SR_GSM_02 Ticket hall unpaid side

SR_GSM_03 GSM administration

SR_GSM_04 Duty station managerrsquos office

SR_GSM_05 Business managerrsquos office

SR_GSM_06 Administration

SR_GSM_07 Ticket office

SR_GSM_08 Police room

SR_GSM_09 Revenue protection examiner

SR_GSM_10 Interview room

140

Stair Stair

100 2020

4533

34

45

FIGURE 9 Platform escape distance SGEvac label restrictions

The history of auto-generated layouts is accessiblefrom this screen and the user is given the capabilityto delete and copy them (useful when minoralterations are required to produce an alternativelayout)

Shape labelling in the programme is multi-faceted which is necessary since many stationcomponents have various possible connections Anescalator for example may connect to a passagetravelator ramp escalator run-off stair run-off lift orthe ticket hall so all such connections need adding tothe component properties The SGEvac grouping ofcomponent interactions is accessed from menutoolbars that allow insertion of legislative compliantstation area connections as shown in Figure 11

Once the position and the depth of the ticket halland platform are specified the railway track is inserteddependent on the platform configuration The platformis the only possible connection to the track but thetrack can be positioned either side of platforms so thata side or centre (island) platform layout is possible

Non-passenger and ancillary areas are alsoselectable from the shape grammar library via property

windows for each station component eg the ticketoffice properties window as shown in Figure 12

Figure 13 shows examples of the layout modulestation components available in the SGEvacprogramme

Figure 14 shows the layout module interfaceInformation on the components (relative position inthe overall site layout and its absolute sizing) isdisplayed at the bottom of the screen once thecomponent is selected from the list In addition ameasurement bar (in metres) at the base of theinteractive design environment provides a sense ofscale for the designer

A checklist summarizes the station areas includedin the layout but a more detailed analysis is possiblevia a report function if required In addition to thisdetailed list a summary screen accessed from themain interface collates all specified components ineach generated layout This screen lists all passengernon-passenger and ancillary areas in their respectivecategories with infrastructure width data ofpassenger areas also summarized It also has anevacuation summary where the train frequency is

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 111

FIGURE 10 SGEvac main interface

specified and the train evacuation load (taken as peaklsquorush hourrsquo loading) is displayed This section of thesummary screen is useful for input into Simulexwhere fire evacuation test runs of the building layoutcan be implemented

The lsquoshow hierarchyrsquo function in Figure 15 allowsdesign team members to view the progression ofdesigns by superimposing connecting lines betweenstation elements in the order of their compositionThis also serves as a safeguard against deleting

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

112 G SMITH AND B CERANIC

FIGURE 12 SGEvac ticket office properties window

FIGURE 11 SGEvac properties window

components out of sequence that could lead to aninvalid solution

Once auto-generated layouts are complete withcorrectly sized and legitimately connected stationcomponents manual insertion of non-passenger areaswould then commence Upon completion thegenerated layout is transferred to AutoCAD via dxf fileformat as shown in Figure 16 Once in AutoCAD thelayout has a level of detail added by the designer suchas door openings passage connections fire escapeopenings etc so that passenger flows may beassessed It is then exported to Simulex for fireevacuation simulation analysis again via dxf file format

Simulex understands any voids as potentialcirculation routes treating the rest as solid andpermanent obstructions ie walls internal structuralelements permanent fixtures etc Since its solutionoperates in a two-dimensional environment for whatin effect is a three-dimensional problem the means

of vertical circulation (eg staircases) have to beestablished True lengths of staircases must bespecified based on the total length of all flights(sloping length) and the intermediate rests The fireexits need to be defined both in terms of theirlocation and width The links have to be establishedto define the point of the connection between thefloors and staircasesescalatorsramps The travellingdistance maps are then calculated type and numberof occupants specified building populated and theevacuation simulation performed

SYSTEM TESTING AND VALIDATIONKuligowski and Peacock (2005) define and recommendthe following ways of validating evacuation models

validation against code requirements validation against fire drills or other peoplersquos

movement experimentstrials

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 113

Gateline Vertical

Platform Right

Stairs Horizontal Stairs Vertical

Travelator Vertical

Ticket Hall

BMOAdmin Reception

Platform Top Ram Up Bottom Vertical

Lift Passage Horizontal

FIGURE 13 Examples of SGEvac station components

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

114 G SMITH AND B CERANIC

FIGURE 15 SGEvac hierarchy function

FIGURE 14 Layout module graphical user interface

validation against literature on past evacuationexperiments

validation against other models third-party validation

Furthermore Galea (1999) deconstructs the validationprocess into four main areas ndash component testingfunctional testing qualitative validation andquantitative validation This approach is more suitablefor models such as SGEvac than for those based oncompliance with code requirements and hence it isthe one adopted for the validation in this research

Component testing is regarded as testingcomputer code that drives the model checking forelementary errors Functional validation involveschecking that the model possesses the ability to

exhibit the range of capabilities required to performthe desired simulations Component testing andfunctional validation for SGEvac is carried outagainst code requirements insofar that allcalculations for infrastructure sizing are evaluatedmanually so that the method and answer are knownto be correct It uses the Simulex program to aid themodelling of people movement and to test thehypothesis of improvement to evacuation timesthus relying on Simulexrsquos own rigorous validationprocess Simulex has been used extensively inpractice as well as a validating model for otherevacuation softwares (Kuligowski and Peacock2005) SGEvac has gone through its initialcompletion and validation process with finalvalidation proposed for the end of the year

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 115

FIGURE 16 Auto-generated SGEvac station layout export file for fire

evacuation simulation in Simulex

FIRE EVACUATION SIMULATION RESULTSThe considered lsquotrain on fire in stationrsquo scenario takesinto account the service interval of 333 minutes(based on 18 trains per hour) the crush capacity of afour-car train together with the number of peoplewaiting on the platform after cancellation of oneservice (ie after 666 minutes) On the oppositeplatform the evacuation load is taken to be thenormal boardingalighting load assuming nocancellations to services (50 of all boarders andalighters)

Evacuation load predictions were based on TfL2006 projected passenger flows with an averageflow of 36 peopleminute for boarders and 123peopleminute for boarders and alighters combinedThe train crush capacity was averaged at 165peoplecar The simulation also models evacuationfrom the ticket hall level to assess the impact ofdispersing crowds at peak times in this part of thestation From these figures the total evacuation loadwas calculated as follows

Platform 1 Boarders (36 x 666 min) = 240Train crush load (165 x 4) = 660

Platform 2 Boarders and alighters

(123 x 05 x 333 min) = 205Ticket hall = 470Total station evacuation load = 1575 people

The simulation results have been captured as imagesat 25 50 100 and 300 second intervals respectivelyfor both platform and ticket hall levels as shown inFigures 17 to 20

The provided means of escape follow LU SPSG4requirements insofar that the maximum travellingdistance on a platform to an exit is no more than 45 m (ie exits must be no more than 90 m apart) andthat the dead-end travelling distance is no more than20 m to an exit Hence each platform is provided withthree exits The MIPsrsquo (mobility impaired persons)means of escape is provided from both ends ofplatforms via emergency evacuation lifts Totalevacuation time (ie all people reaching an exit onticket hallground level) was 5 minutes and

24 seconds with the platform and train load clearedwithin 3 minutes and 5 seconds This is within therequirements to evacuate the platform and train loadwithin 4 minutes with all people reaching a fire-separated route within 6 minutes (the ticket hall is notclassed as a fire-separated route)

The performed simulation was based on theminimum travelling distance maps resulting in eachperson exiting via the nearest fire exit route Theaverage 10 second lsquotime to start to moversquo wasassigned to all occupants in order to accommodatefor real world people behaviour commonly reportedby fire evacuation research studies The boarders onthe end of both platforms (including most peoplefrom end carriages of the train) evacuated viastaircasesMIP lifts directly to the ground level pointof escape The boarders situated around the middleof platforms together with the people from the twomiddle carriages used the central escalators to reachticket hall fire exits Additionally some people fromthe end carriages and some staffretail unitcustomers used this route as means of escape

In this lsquotrain on fire in stationrsquo scenario it wasassumed that none of the escapes routes areunavailable due to maintenance However furtherevacuation studies will investigate possible scenarioswhere one of the escalator routes may be unavailabledue to maintenance reasons

FUTURE RESEARCHIntended future work will concentrate on the furtherconditioning developing and modelling of shapegrammar design rules based on the outcomes of thefinal system validation and LU expertsrsquo feedbackFurther research is proposed on the development ofa novel layout optimization approach based on theshape grammar rules to investigate the evacuation-related parametric improvements in subsurfacesurface railway station design (eg minimizingtravelling distances)

The fire safety strategy in terms of performance-based fire risk design is an important aspirationalaspect of SGEvac requiring further research andliaison with experts to determine how such variedcriteria could be allied with layout planning This isparticularly important where deviation from standardsoccurs Thus further development of the expert

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

116 G SMITH AND B CERANIC

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 117

FIGURE 17 Station evacuation after 25 seconds

FIGURE 18 Station evacuation after 50 seconds

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

118 G SMITH AND B CERANIC

FIGURE 19 Station evacuation after 100 seconds

FIGURE 20 Station evacuation after 300 seconds

knowledge database would be required particularlyin the light of frequent changes to legislation

On a more practical note it could be beneficial todevelop the system on a stand-alone modular basislinked to each generated station component A list offire safety strategy objectives and functional logicstatements can then be specified for commonlyrecurring items (such as emergency lighting fire alarmdetectors sprinkler heads etc) and spaced accordingto the designerrsquos wishes or via shape rule formulation

Site constraint issues would also benefit fromfuture development as well as a move towardsdevelopment of the 3D shape grammar version Oncedevised other relevant 3D modelling systems such ascomputational fluid dynamics (CFD) smoke movementsoftware could be utilized in conjunction with SGEvac

CONCLUSIONSThe computer-aided modelling environment underdevelopment is intended to act as a layout planning aidfor those who practice in sub-surfacesurface railwaystation design The system under developmentcarefully considers London Underground StationPlanning Standards and Guidelines together withother relevant fire safety regulations that influenceoverall fire strategy design The proposed computer-aided design environment (SGEvac) is based on highlyrestrictive additive shape grammar rules driven by akernel of key passenger flow information to generatestation plan solutions

The major benefit to the design team is the ability toreview multiple layouts of accurately sized and correctlyplaced infrastructure at the reference or preliminarystage of the station layout planning design process Timeis hence saved regarding different option appraisals andquantitative testing using evacuation software shouldlead to a choice of more effective layouts in relation tomeans of escape travelling distances and improvementin the total evacuation times Modifications to layoutsshould be straightforward and there is scope toefficiently assess travelling distances and evacuationtimes for each of the considered alternatives comparedwith a lsquotraditionalrsquo approach

The proposed novel shape grammar approach ismost suited to new-build stations or largerefurbishment projects where more design flexibilityexists Sites with many site-specific constraints have

an array of predetermined parameters and will mostlikely benefit from the manual operation of SGEvac

Use in other rail systems due to recurringinfrastructure components and inherent functionallogic related to the laying out of the stationcomponents is also feasible provided that shapeconnections and shape rules are corrected to reflectthat countryrsquos legislative requirements

The potential for station layout improvementswith regard to emergency evacuation is an area thatmerits research especially in scenarios that concernthe safety of the general public and form a vital partof Londonrsquos transportation system Potential layoutimprovements could not only benefit evacuationsituations but also enhance the through-flow ofpeople during peak travelling periods

ACKNOWLEDGEMENTS The authors wish to express their gratitude to JacobsEngineering (UK) Ltd which has sponsored theresearch and to London Underground Ltd forproviding the design guidance that is reproduced inthis paper and for authorising the use of the stationplans London Underground has not reviewed thispaper and the content is solely the authorsrsquo work Mr Robert Hutchinson is also thanked for hiscomputer programming assistance

AUTHOR CONTACT DETAILSGGrraahhaamm SSmmiitthh (corresponding author) Senior Architectural

Technologist Jacobs (UK) Ltd Cardinal Square North Point

10 Nottingham Road Derby DE1 3QT UK Tel +44 (0)1332 234482

fax +44 (0)1332 344558 e-mail grahampsmithjacobscomBBoorriiss CCeerraanniicc Programme Leader for BScBA (Hons) Architectural

Programmes Room Number 105 Faculty of Arts Design and

Technology University of Derby Markeaton Street Derby DE22

3AW UK Tel +44 (0)1332 593136 e-mail bceranicderbyacuk

REFERENCES BS 7974 2001 Application of Fire Safety Engineering Principles to the

Design of Buildings London BSI

Cha MY and Gero JS 1999 Style Learning Inductive Generalisation of

Architectural Shape Patterns Paichai University and Sydney University

co-paper

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 119

Chase SC 1996 lsquoUsing logic to specify shapes and spatial relations in

design grammarsrsquo in Workshop notes Grammatical Design Fourth

International Conference on Artificial Intelligence in Design Stanford

University US June

Department of the Environment 1996 Design Principles of Fire Safety

London The Stationery Office Books

Flemming U 1987 lsquoMore than the sum of its parts the grammar of

Queen Anne housesrsquo in Environment and Planning B Planning and

Design 14 323ndash350

Fruin JJ 1971 Pedestrian Planning and Design New York Metropolitan

Association of Urban Designers and Environmental Planners

Galea ER 1999 lsquoA general approach to validating evacuation models with

an application to EXODUSrsquo in Journal of Fire Sciences 16 414ndash436

Galle P and Kovacs LB 1992 lsquoThe logic of worms a study in

architectural knowledge representationrsquo in Environment and Planning B

Planning and Design 19 5ndash31

Garling T Lindberg E and Mantyla T 1983 lsquoOrientation in buildings ndash

effects of familiarity visual access and orientation aidsrsquo in Journal of

Applied Psychology 68(1) 177ndash186

Grimsdale RL and Chang CW 1996 lsquoThe layout design language A

technique for generating layout plansrsquo in Computer Graphics Forum

15(2) 97ndash106

Heskestad AW 1999 lsquoPerformance in smoke of way guidance systemsrsquo

in Fire and Materials 23(6) 375ndash381

Knight T 1995 lsquoConstructive symmetryrsquo in Environment and Planning B

Planning and Design 22 419ndash450

Knight TW 1999 lsquoShape grammars six typesrsquo in Environment and

Planning B Planning and Design 26 15ndash31

Kuligowski ED and Peacock RD 2005 A Review of Building Evacuation

Models Technical Note 1471 National Institute of Standards and

Technology (NIST) Technology Administration US Department of

Commerce

Li A I-Kang 2003 lsquoThe Yingzao fashi in the information agersquo conference

paper for the Beaux-Arts Paul Philippe Cret and Twentieth Century

Architecture in China The University of Pennsylvania October

Liew H 2003 lsquoSGML ndash A shape grammar meta-languagersquo in Digital

Design 21st eCAADe Conference Proceedings Graz Austria September

639ndash648

Loomis B 2003 lsquoA note on generative design techniques SGGA a user

driven genetic algorithm for evolving non-deterministic shape grammarsrsquo

working paper Massachusetts Institute of Technology (MIT)

Magnusson SE Drysdale DD Fitzgerald RW Mowrer F

Quintiere JG Williamson RB and Zalosh RG 1995

lsquoA proposal for a model curriculum in fire safety engineeringrsquo

in Fire Safety Journal 25 9ndash10

Smith G Ceranic B and Fryer C 2003 lsquoDevelopment of CAD

environment for spatial layout optimisation in sub-surfacesurface rail

station designrsquo in Proceedings of International BIAT Conference The

British Museum London January 2003

SPSG 2007 London Underground Station Planning Standards and

Guidelines SPSG4 (1ndash371) London

Stiny G and Gips J 1971 lsquoShape grammars and the generative

specification of painting and sculpturersquo in Information Processing 71

1460ndash1465

Stiny G 1977 lsquoIce-ray a note on the generation of Chinese lattice

designsrsquo in Environment and Planning B Planning and Design 4 89ndash98

Stiny G 1980 lsquoKindergarten grammars designing with Froebelrsquos building

giftsrsquo in Environment and Planning B 3 409ndash462

Tapia M 1999 lsquoA visual implementation of a shape grammar systemrsquo in

Environment and Planning B Planning and Design 26(1) 59ndash73

Transport for London 2007 Tube Carries One Billion Passengers for First

Time wwwtflgovukcorporatemedianewscentre4770aspx

(accessed 24 March 2008)

Weller M 2007 London Underground private correspondence

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

120 G SMITH AND B CERANIC

1-082 Active fire protection systems and portablefire extinguishers

1-083 Passive fire protection systems 1-084 Maintenance of fire protection systems 1-085 Fire safety performance of materials 1-066 Emergency lighting HMRI Railway Safety Principles and Guidance

Part 2 Section B Regulatory Reform (Fire Safety) Order 2005 BS 5588 Fire precautions in the design

construction and use of buildings BS 7974 Application of fire safety engineering

principles to the design of buildings BS 8300 Design of buildings and their approaches

to meet the needs of disabled people

The figures for predicted passenger flow rates areobtained from Transport for London (TfL) According toTfL 2007 figures London Underground had over 1 billionpassenger journeys for the first time in its history whichis more than the entire national rail network Thus theresearch to improve both the evacuation and normalcirculation routes with regard to emergency evacuation(means of escape travelling distances position of fireexits circulation etc) could have an important benefitwhen designing large interchange stations under majorrefurbishment or new build consideration

THEORY AND APPLICATION OF SHAPEGRAMMAR The concept of shape grammar formalism is primarilyattributed to Stiny and Gips (1971) with majorcontributions made by Stiny (1977 1980) Flemming(1987) Knight (1995 1999) Chase (1996) Tapia(1999) Li (2003) and Liew (2003)

Shape grammar is a method of describing andcreating languages of designs where shapes aredevices for visual expression Their composition iscontrolled by preconceived rules in a very similar wayas to how grammar is used to arrange written wordsinto sentences The theory provides an approach forthe generation of composite shapes based onprinciples of recognition and replacement of aparticular shape Primitive shapes are transformedand combined to create more complex compositesthrough the application of predefined rules and soshape language emerges (Cha and Gero 1999)

Loomis (2003) demonstrates an exercise of a two-dimensional shape grammar with one rule The shapeon the left-hand side is replaced by the shape on theright which following recursive application leads tothe generation of simple shape grammar designs asshown in Figures 1a and 1b respectively

As Loomis demonstrates further it is possible tohave the shape rule in Figure 1a applied in fourpositions (see Figure 2) which would lead to differentshape grammar designs To remove any uncertaintyassociated with rule application it is necessary tointroduce a deterministic approach where shapegrammar rule application is managed by a set oflabels These labels (or rules) constitute a set fromwhich alternative designs are derived Figure 3 showssome of the possible outcomes that have beenderived from the rule set specified in Figure 2

Tapia (1999) explains how shape grammarsoriginally defined by hand lsquonaturally lend themselvesto computer implementation [handling] therepresentation and computation of shapes rulesgrammars and the presentation of alternativeshellipthedesigner specifies explores develops the languageand selects [generated] alternativesrsquo

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

100 G SMITH AND B CERANIC

(a) (b)

RULE 1 RULE 2

RULE 4RULE 3

FIGURE 1 A shape grammar rule and its recursive applications

(Loomis 2003)

FIGURE 2 Labelling shape rule for deterministic grammar

designs (Loomis 2003)

Stiny and Gips (1971) define shape grammar as

S = [VT VM R I] (1)

where VT is a finite set of shapes VM is a finite setof labels R is a finite set of shape rules and I is aninitial shape

In this research parameter VT is the finite set ofthe station infrastructure components shapes andVM is the finite set of labels that comprise thepossible station area permutations mapping twolsquolegally connectedrsquo components into a spatial relationwithin the overall layout R is the finite set of theshape connectivity rules composed of informationcontained in regulations and guidance documents interms of the area assignments adjacency and theirconnectivity I is an initial shape (ticket hall in theSGEvac programme)

In summary a proposed shape grammar systemconsists of a set of vocabulary elements (stationcomponents) a set of rewrite rules and a start stateStation layout can be considered in terms of compositeshape generation and its design process in terms oflogical sequence and deductive reasoning It is thisapproach based on the function of the building thatallows for a transition from lsquotraditionalrsquo layout planningto the computational shape grammar layoutgeneration that is proposed in this research Howeverits application depends directly upon the expertrsquosknowledge filtered into this rule-based approach

An additive shape grammar approach (Knight1999) is proposed as a suitable technique for

generating station layouts This approach is useful forthe irregular station boundaries derived from sitespecific conditions where designs are generatedfrom a core (the ticket hall) to which othercomponents are successively added

Knight (1995) remarks that the rules detailed inStinyrsquos Kindergarten paper (1980) can be preacutecised toexternalize a studentrsquos design ideas so that they canbe examined changed and communicated morereadily It is the choice of credible shape grammaralternatives that is important not the multitude ofsolutions themselves Good analytic grammars areboth efficient and descriptive revealing simplicity orregularities behind seemingly complex designs

STATION LAYOUT DESIGNCONSIDERATIONSPassenger space planning needs to be considered interms of peoplersquos procedures when they use theUnderground Initially customers will enter thecomplex from either street level or via a surfacestation stairwell A one- or two-way passage will thentake passengers to the unpaid side of the ticket hallwhere they will purchase a ticket before passingthrough to the paid side

Customers then descend using stairsescalatorslifts to platform level At both the top and bottom ofstairs and escalators there is a station componenttermed the run-off area The length of the run-offdepends on the peak passenger flow capacity of theescalatorstairwell and whether it connects to aplatform passageway or street It is an area wherepeople move away from the change in level orientatethemselves and make a decision of where to go next(SPSG 2007)

Generally at platform level people will journey tothe desired line to take them to their destination via afurther passageway or lobby area Some escalatorsstairs can exit directly onto platforms The travelprocess is reversed when alighting from a train andcontinuing to the surface

The predicted passenger flow rates and sitespecific conditions dictate both the number of railwaylines and the platform(s) configuration (side centre orstacked platform) Station composition must alsoconsider areas that are not sized by passengervolume demand but are required to ensure that the

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 101

INITIAL SHAPE

INITIAL SHAPE

RULE 1 RULE 1 RULE 1 RULE 1

RULE 2RULE 2RULE 2RULE 3

FIGURE 3 Labelling shape rule for deterministic grammar

designs (Loomis 2003)

station operates satisfactorily on a daily basis Thestation managerrsquos office plant and staff roomscomputer rooms train technician rooms etc all needaccommodating in the proposed shape grammarformalism In addition ancillary accommodationsuch as the cleanerrsquos store police room publicitystore retail etc also have to be catered for

To aid designing and sizing infrastructure levelsof service as shown in Figure 4 are used by LondonUnderground These are based on the work of Fruin(1971) where a detailed study of crowd movementwas undertaken Level A represents free circulationwith subsequent levels growing progressively morerestricted in terms of people movement from B to Eculminating with level F that represents a completebreakdown in flow with numerous stoppages

The approach proposed in this research definesthe generic station layout arrangements from thelsquohigh-levelrsquo spatial site organization to the subsequent

specified components and their logical spatialallocations as illustrated in Figure 5 The consequentshape grammar rules can then be deduced taking intoaccount the predicted passenger flow rates fire safetystrategy and relevant design regulations

FIRE SAFETY DESIGN STRATEGYBefore any design commences a clear identificationand appraisal of the key constraints and parametersat the site layout planning stage is required Forexample deep foundations mains servicesarchaeological discoveries and even the RiverThames will affect station design decisionsFurthermore structural appraisals geotechnicalsurveys Statutory Undertaker searches groundcontamination testing environmental impactassessments and land acquisition etc all need to becarried out as part of a feasibility study before stationlayout planning at reference level commences

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

102 G SMITH AND B CERANIC

Level of service A

Level of service D Level of service E Level of service F

Complete breakdown in trafficflow with many stoppages

Free circulationDescription (forqueuing areaswalkways andstairways)

Uni-directional flows andfree circulation Reverseand cross-flows withonly minor conflicts

Restricted circulation formost pedestriansSignificant difficulty forreverse and cross-flows

Restricted circulation for allpedestrians Intermittentstoppages and seriousdifficulties for reverse andcross-flows

Sligthly restricted circulationdue to difficulty inpassing others Reverse andcross-flows with difficulty

Level of service B Level of service C

FIGURE 4 London Underground levels of service (SPSG 2007)

Such constraints may be viewed in a positivecontext in terms of services already being providedhelping to justify the business case for a new stationDesign procurement strategies may include decisionssuch as locating toilet areas within x metres of anexisting sewer route or locating electrical plant roomswithin y metres of an existing electrical sub-station

In the real world parameters such as these constrict layout production and the choice ofinfrastructure component placement as opposed toleading to the production of many solutions that arelikely to be discarded when for example projectcosts are reviewed This level of regularity andrestriction typical in station design is very suitable

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 103

SITE

CUT AND COVER STATION TUNNELLED STATION

TICKET HALL

UTS GATELINE (gateline dimensionsdetermine minimumticket hall width)

PASSAGEWAYS ONE- or TWO-WAY

(entrances exits determinea minimum uniform width)

ESCALATOR

STAIRS

RAMPS

TOP RUN-OFF

TRAVELATOR

PAID CONCOURSE SIDE

BOTTOM RUN-OFF

PASSAGEWAYS ONE -WAY or TWO-WAY

NON-PASSENGERSPACE ELEMENTS ATTICKET HALL LEVEL (egStation Managerrsquos Office)

PLATFORM LEVEL

SPACE ELEMENTS ATPLATFORM LEVEL (egplant rooms toilets etc)

SIDE PLATFORM CENTRE PLATFORM STACKED PLATFORM

No OF RAILWAY LINES

MEZZANINEOptional

MEZZANINE

DEEP

SHALLOW

LIFTS

NON-PASSENGER

FIGURE 5 lsquoHigh-levelrsquo layout hierarchy for generic station composition

for the application of a prescriptive shape grammarapproach

In parallel with the shape grammar developmentfunctional logic statements are proposed analogousto the qualitative design review as described in BS7974 Application of Fire Safety EngineeringPrinciples to the Design of Buildings The intention isto develop SGEvac procedures in a manner that isrecognized by fire safety engineers and designers inpractice The Department for the Environment (DOE)(1996) states that lsquothe architectural layout can have asignificant influence on peoplersquos movementComplexity in circulation routes may not be overcomenecessarily by signage these may not have beennoticed or they may have been forgotten as a resultof stressrsquo Hence station spatial requirements can bereviewed alongside a number of functional logicstatements devised in conjunction with the layoutplanning

For example lsquothe probability of successful escapethrough smoke diminishes sharply with (an increasein the) number of decision elements included in theescape pathrsquo (Hesketad 1999)

A functional logic statement response to theabove would be to minimize such decision elementsthus enabling faster egress Other examples offunctional logic statements employed are

no corridor or staircase will reduce in width alongescape routes

length of the passages are to be minimized maximize level areas to facilitate faster travel

speeds for escapees maintain uniformity to aid way-finding and relative

orientation evacuation and normal circulation routes are to

be the same exits will be prominent easily identifiable and

straightforward to reach

Such statements are intuitive by their nature and arepart of the qualitative reviews in the traditional designprocess Due to time constraints only a limitednumber of possible alternatives in terms of the spatiallayout improvements are ever considered if any Incontrast SGEvac offers not only the opportunity toconsider a significant number of possible alternatives

but also the means of quantitative assessment via fireevacuation simulations

Garling et al (1983) contend that newcomers to anenvironment first learn a number of salient locationsthen the paths in between and finally organize theacquired knowledge in a spatial system Indeedavoiding symmetry and lack of differentiation fromthe userrsquos viewpoint is a good practice for stationdesign considering perception of relative orientationin the station This is where interaction between thedesigner and computer program becomes essentialand it is in this context that the research argues therole of SGEvac as a computer-aided design tool notas a substitute for the designer

A comparative methodology overview is given inTable 1 The processes of BS 7974 the procurementof a general logic model (Galle and Kovacs 1992) andthe shape grammar development stages arebenchmarked against SGEvac procedures Thisperhaps has most relevance to fire safetyengineering since it is with this objective that theproposed shape grammar system is to be used inpractice

Research studies such as one by Magnusson et al(1995) note that fire safety engineering has motivateddesigners and regulators to examine alternative codecompliance approaches based on a greater relianceupon quantitative analyses This has led toreassessment of prescriptive approaches andconsequent adjustments as required It is nowaccepted that existing prescriptive practices shouldbe evaluated by quantitative performance-basedmethods as a part of the overall fire safety strategyas encouraged in the Regulatory Reform (Fire Safety)Order (RRFSO) 2005 SGEvac follows this trendaugmenting the array of tools available forperformance-based fire safety engineering design

There is obviously an inevitable need for designerinput and control at all stages of the design processDue to the potential necessity for deviation fromstandards via fire safety engineering judgement theproposed shape grammar approach cannot alonemodel project-specific concessions

Still the movement away from prescription viaBS 7974 and the RRFSO to risk-based judgementplaces computer-aided design tools such as SGEvacin a position of interdependency with other design

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

104 G SMITH AND B CERANIC

criteria neither can be formulated in isolation Thisreinforces the view that computers will in future playan ever-increasing role in aiding and enabling thedesign teams but will never be able to replace them

SYSTEM DEVELOPMENTFigure 6 describes the modular approach proposedfor the system development offering the addedpotential of dual-purpose application in the designprocess It allows the designer to use the systemeither as a whole or to take advantage of itsindependent modules The system offers designerscomplete control over the shape grammarcomposition and layout modifications together withthe constant monitoring facilities and ability todiscard any solutions deemed to be lsquoinferiorrsquo

The proposed system is partitioned into threemain modules ndash lsquoexpert knowledgersquo lsquolayoutrsquo andlsquotestingrsquo modules The expert knowledge modulecaptures the expertise available to facilitate accurateshape grammar generations of layout design Thismodule provides the designer with the ability tointeractively reference Station Planning Standards

and Guidelines (and other relevant fire safetyregulations) allowing himher to make necessary site-specific decisions and to control the lsquohigh levelrsquo shapegrammar composition rules

The designer specifies passenger flow rates anddesignates the overall site boundary dimensions thatdefine the permissible station footprint The final exitpositions are also stipulated at this stage with theseoften being constrained by the existing builtenvironment

With this data supplied the shape grammargenerations commence in the layout module Thearea assignments adjacency requirements and layoutcomposition rules are defined in detail at this stageOnce the designer has made decisions with regard tothe layout composition the computer becomes thecentral tool for generating searching and logicallysorting through the permissible permutations ofpossible layouts The user has the option to auto-generate layouts (usually for sites without manyconstraints) or to manually control the layoutassembly (a decision normally associated with heavilyconstrained sites)

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 105

TABLE 1 Methodology comparison BS 7974 PROCESSES GALLE AND KOVACS SHAPE GRAMMAR SGEvac

LOGIC MODELLING DEVELOPMENT IMPLEMENTATION

Assess benefits Identification Shape grammar Assess constraints

and constraints system identification identify benefitssite

boundariesfinal exit

positions

Qualitative design Conceptualization Initial state Station component

review identification and

interactions

Quantitative analysis Formalization Rules and shape Legislative compliance

language passenger flow figures

area assignments

Satisfactory Implementation Shape grammar Alternative station

unsatisfactory executions layouts generation

Report and review Testing Design analysis Testingvalidation

iterationsevolution of layouts generation

End Experimentation Experimentation

iterationsimprovements

The testing module is used to evaluate potentiallysuitable layouts at reference level by exporting them toAutoCAD via standard drawing data exchange formatand analysing them in the context of project-specificsite constraints and requirements A level of drawingdetail is then added and the layout solution is exportedto Simulex a building evacuation simulation software

package Once the occupant numbers and persongroups are assigned and the escape and verticalcirculation routes defined the simulation software iscapable of the accurate measurement of all travellingdistances creation of distance maps and calculation ofevacuation times for different fire scenarios (egdifferent group affiliations persons exiting via different

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

106 G SMITH AND B CERANIC

EGY D ESIGNDesigner

Feasibility studiesProject brief requirements

Passenger flow data

DDE

DDE

Layout Production

Shape grammar rulesMultiple layout production

Layout Analysis

Layout interrogationsDesign team judgement

No

Yes

DXF

DXF

Yes

No

AutoCAD

Feed lsquoblock reference planrsquoresults into overall project

Improvedsolution identified

EXPERT KNOWLEDGE MODULE

Fire Strategy Design

LU SPSG4FSE legislation amp guidance

Infrastructure size calculationHigh level brief development

LAYOUT MODULE

Layout Composition

Area and adjacency planningLayout option appraisal

Layoutimproved

TESTING MODULE

AutoCAD

SGEvac export into CAD

Simulex

Fire evacuation scenarios

DesignerSGEvac

DDE ndash Database Data Exchange

DXF ndash Standard Drawing Data Exchange

FIGURE 6 Proposed SGEvac algorithm

routes and having delayed movement in response to afire alarm signal) The design team then assesses theresults against the qualitative fire safety design goalsset earlier to establish whether improvements toevacuation times have been made

DETAILED ANALYSIS OF STATIONCOMPONENTSIn conjunction with the site boundary final exitpositions and passenger flow specificationconsideration must also be given to the vertical sitedimension The shape grammar system assumes allfinal exits are at ground level with the ticket hall andplatform(s) being the other two elements that requirethe user to determine their depth relative to the finalexits In so doing SGEvac selects the correctinterchange components by an event-drivenmechanism that evaluates the height change andpassenger volume demand Thresholds for bothcriteria have been defined as shown in Table 2 andEquations 2ndash5

If H (height in mm) = 0 ndash 500 then I (infrastructure) = Ramp (2)

If H = 500 ndash 3000 and P (passengers) le 3000 then I = Stairway (3)

If H = 3000 ndash 5000 And P le 3000 then I = StairwayElse I = Escalator (4)

If H gt 5000 and P le 1000 then I = StairwayElse I = Escalator and Lift (5)

(where lsquoElsersquo denotes the mechanism for the choiceof the alternative infrastructure component if theoriginal condition is not satisfied)

The above constructs also condition the run-offdimensions associated with each of theinterchange components (between 4 m and 12 mas stipulated by SPSG) This being the case run-offs are used to size the ticket hall length Figure 7shows the paid side of the ticket hall with variabledistances shown between the escalator or staircaseand the gateline

The paid side of the ticket hall is always dictatedby run-off distance while the unpaid side is governedby two parameters the run-off and the concoursearea required for the heaviest 15-minute passengerflow If the maximum 15-minute passenger flowfigure multiplied by 1 m2 (say answer A) is larger inplan area than the product of the gateline widthmultiplied by the relevant run-off (say answer B) thenanswer A will be incorporated into the layout Thereverse is true if B is the greater number This ensuresthat the station is always sized appropriately

Figure 8 shows a ticket hall layout with thegateline dictating the width (denoted as Tick_H_wd)itself conditioned by the passenger flow figurescalculated as shown in Table 3 The table onlyevaluates entry flow requirements but SGEvac willrepeat the same process for exit gates to produce acorrectly sized gateline

In this case the maximum 15-minute passengerflow capacity size prevailed over the run-off to createthe unpaid side overall

The divide between the paid and unpaid side ofthe ticket hall is the sized according to Table 4

SGEvac shape rules mandate that entry gates arealways located adjacent to the ticket office queuingspace (designated as Q_TIW) to avoid conflictingpassenger entry and exit flows The LU standard is to

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 107

TABLE 2 Interchange infrastructure selectionINFRASTRUCTURE REASON FOR SELECTION

Ramp Vertical rise between 0ndash05 m preferred to steps due to disabled access issues generally considered

people will travel faster on ramps than up or down steps

Stairs Vertical rise between 05ndash5 m but greater height available for selection if station has a light passenger flow

Escalator Vertical rise is minimum of 3 m for heavy passenger flow stations but used for lightmedium flow stations

above 5 m

Lift Vertical rise above 5 m also required for accessibility of disabled persons Lifts always used with escalators

not as a substitute

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

108 G SMITH AND B CERANIC

Con_Area

Exit

UTS gates

6601900 2450

2400

EFW

Ticketoffice

POMcounter

4000

Tick_H_wd

1000

Run

-off

Entry

Q_space

Q_TIW

FIGURE 8 SGEvac ticket hall arrangement

Heavy flow escalator

Escalator

Staircase

Medium to Heavy flow

Medium to Heavy flow

Light flow escalator

Light flow

Heavy flow 12mMedium flow 8ndash12m

Light flow

Light flow 8m

Light flow 6m

Heavy flow

staircase to gateline

staircase to gateline 10m

Light flow

to gateline12mGateline

Gateline

EscalatorndashGateline

Total Run-off

Heavy flow 10mMedium flow 6ndash10m

StaircasendashGateline

Total Run-off

to gateline 8m

2m 4m 6m 8m 10m

2m 4m 6m

6m

8m 10m

12m

FIGURE 7 Passenger flow rate run-off requirements (SPSG 2007)

allow for 4 m (or a five-person queue) as shown inFigure 8 The overall gateline is defined by summingboth exit and entry gate numbers obtained from user-inputted passenger figures

The length of the ticket office interface with theticket hall is stipulated by the number of ticket issuingwindows (TIWs) plus excess fare windows (EFWs)TIWs always belong on the unpaid side of the tickethall and EFWs always on the paid side both forrevenue protection reasons This situation facilitates aprecise shape grammar labelling for the ticket officeto ticket hallqueue space connection

Passenger operated machine (POM) enclosures arenormally connected to the ticket hall ticket office andits own queue space (noted as Q_space in Figure 8)

To make shape rule specification manageable allstation areas are divided into groups arranged to suitthe most probable connection patterns in the stationTable 5 shows the group station managerrsquos officeexample with all other station components followingthe same concept

London Underground views its stations as largebasements (Weller 2007) and escape distances fromplatforms are based on agreed dispensations with theLondon Fire Brigade SGEvac respects thesedistances by placing restrictive labels on the platformso that they cannot be exceeded as demonstrated bythe example shown in Figure 9

SGEvac escape stair infrastructure may bepositioned up to 20 m from the end of platforms toavoid undesirable lsquodead endrsquo conditions The 45 mmaximum platform travel distance starts from thenearside junction of the two 33 m wide escapestaircases with the 45 m distance rule creating ashortfall in the platform centre of 34 m Generallylocating three interchange components (eg stairs

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 109

TABLE 3 Gateline width calculationMETHOD WORKED EXAMPLE

Obtain peak hourly flow rate through Say 6500 peoplehr

ticket hall

Calculate 5-minute peak passenger flow 650012 = 541

Allowance for ticket rejection 541 + 20 = 561

Allowance for growth in use demand 561 x 110 = 617

Assume 125 people use each gate over

the 5-minute peak passenger flow period 617125 = 493

Round up gates to nearest whole number 5 gates required

TABLE 4 London Underground ticket hall gateline parametersregulations (SPSG 2007)EQUIPMENT DIMENSIONS (W X D) POSITIONARRANGEMENT

Combined manualequipment gate 1750 x 270 mm Height ndash 1300 mm Adjacent to ticket office Can include

1400 mm clear opening luggage facility

Manual gate ndash passenger 1750 x 270 mm Height ndash 1300 mm Adjacent to ticket office With more than 6

gateluggage port 900 mm ndash clearance and walkways in a single gateline or significant

outward opening interchange with mainline stations airport or

tourist traffic provide separate luggage chutes

Manual gate ndash single leaf 1394 mm Height ndash 1300 mm Beside the dedicated excess gates

with luggage port 695 mm clearance

Gateline attendance report 990 x 1240 mm Within the controlled zone but not necessarily

part of the gateline Usually close

Station control unit (SCU) 432 x 430 mm Should be installed within a secure enclosure

(which could be a ticket office) and also with

a clear view of the gates Controls and monitors

the station underground ticketing system

equipment

Manual emergency exit gates 1900 x 125 mm Adjacent to ticket office Add these gates rather

than adding extra UTS gates for evacuation

purposes

or escalators) at these distances will result incompliance with the Underground standards but itmay not always be possible (or practical for aninterchange situation) to procure such a layout

Lengths of station components are controlled bya range of criteria For the platform it is the length ofthe train vehicle for interchange components it isthe vertical height in the terms of the depth to theplatform level for passages it is the designated finalexit positions and the existing station areas alreadygenerated

Vector direction labels are underpinned by thepermissible union of compatible spaces two points ofcontact will be maintained so as not to restrict thewidth of connecting passenger areas This undueconstriction of passenger flow at the head of apassage also satisfies realistic construction practiceNon-passenger and ancillary area vector directions areas per project requirements and the successive areaadditions managed by the SGEvac rule groupings

To create alternative layouts SGEvac uses alsquorandomizersquo rule In this shape grammar rule initialstate infrastructure is generated as per guidelinesbut it can randomly change the location ofcomponent arrangements within the predefined

boundaries so that a number of possible alternativelayouts are produced The designs can then beinterrogated and ranked by their travelling distancesand against the qualitative functional logicstatements

The SGEvac design tool developed in this researchis based on a highly restrictive additive grammar ruleset due to SPSGrsquos prescriptive guidance and thenature of the layout planning problem As Grimsdaleand Chang (1996) argue many published grammarslack high-level reasoning during the iteration processsince they are often concerned with the developmentof shape grammar theories rather than their practicalapplications By contrast SGEvac being a practicalapplication has a complete set of primitive shapesonly possible to arrange in a certain way It thusprovides viable reference design layout compositionsbased on SPSG with further interventions possible bythe designer

SYSTEM DEVELOPMENT TO DATEThe SGEvac prototype is compiled within the Clanguage environment allowing for modular object-oriented programming (suitable for shape grammar)Figure 10 shows the main interface where the stationname and relevant LU fare zone are entered (whichimpacts on sizing infrastructure) as are the ticket halland platform depths the overall site boundaries andfinal exitsrsquo positions The peak passenger flow figuresover the relevant time interval of normally 3 hours arealso specified SGEvac is then capable of accuratelycalculating all passenger area infrastructure based onthe SPSG guidelines as already discussed

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

110 G SMITH AND B CERANIC

TABLE 5 Station manager suite ndash shape rule connectionsELEMENT SHAPE RULE DESIGNATION CONNECTED COMPONENT GROUP ALLOCATION

Group station managerrsquos office SR_GSM_01 Ticket hall paid side GSM

SR_GSM_02 Ticket hall unpaid side

SR_GSM_03 GSM administration

SR_GSM_04 Duty station managerrsquos office

SR_GSM_05 Business managerrsquos office

SR_GSM_06 Administration

SR_GSM_07 Ticket office

SR_GSM_08 Police room

SR_GSM_09 Revenue protection examiner

SR_GSM_10 Interview room

140

Stair Stair

100 2020

4533

34

45

FIGURE 9 Platform escape distance SGEvac label restrictions

The history of auto-generated layouts is accessiblefrom this screen and the user is given the capabilityto delete and copy them (useful when minoralterations are required to produce an alternativelayout)

Shape labelling in the programme is multi-faceted which is necessary since many stationcomponents have various possible connections Anescalator for example may connect to a passagetravelator ramp escalator run-off stair run-off lift orthe ticket hall so all such connections need adding tothe component properties The SGEvac grouping ofcomponent interactions is accessed from menutoolbars that allow insertion of legislative compliantstation area connections as shown in Figure 11

Once the position and the depth of the ticket halland platform are specified the railway track is inserteddependent on the platform configuration The platformis the only possible connection to the track but thetrack can be positioned either side of platforms so thata side or centre (island) platform layout is possible

Non-passenger and ancillary areas are alsoselectable from the shape grammar library via property

windows for each station component eg the ticketoffice properties window as shown in Figure 12

Figure 13 shows examples of the layout modulestation components available in the SGEvacprogramme

Figure 14 shows the layout module interfaceInformation on the components (relative position inthe overall site layout and its absolute sizing) isdisplayed at the bottom of the screen once thecomponent is selected from the list In addition ameasurement bar (in metres) at the base of theinteractive design environment provides a sense ofscale for the designer

A checklist summarizes the station areas includedin the layout but a more detailed analysis is possiblevia a report function if required In addition to thisdetailed list a summary screen accessed from themain interface collates all specified components ineach generated layout This screen lists all passengernon-passenger and ancillary areas in their respectivecategories with infrastructure width data ofpassenger areas also summarized It also has anevacuation summary where the train frequency is

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 111

FIGURE 10 SGEvac main interface

specified and the train evacuation load (taken as peaklsquorush hourrsquo loading) is displayed This section of thesummary screen is useful for input into Simulexwhere fire evacuation test runs of the building layoutcan be implemented

The lsquoshow hierarchyrsquo function in Figure 15 allowsdesign team members to view the progression ofdesigns by superimposing connecting lines betweenstation elements in the order of their compositionThis also serves as a safeguard against deleting

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

112 G SMITH AND B CERANIC

FIGURE 12 SGEvac ticket office properties window

FIGURE 11 SGEvac properties window

components out of sequence that could lead to aninvalid solution

Once auto-generated layouts are complete withcorrectly sized and legitimately connected stationcomponents manual insertion of non-passenger areaswould then commence Upon completion thegenerated layout is transferred to AutoCAD via dxf fileformat as shown in Figure 16 Once in AutoCAD thelayout has a level of detail added by the designer suchas door openings passage connections fire escapeopenings etc so that passenger flows may beassessed It is then exported to Simulex for fireevacuation simulation analysis again via dxf file format

Simulex understands any voids as potentialcirculation routes treating the rest as solid andpermanent obstructions ie walls internal structuralelements permanent fixtures etc Since its solutionoperates in a two-dimensional environment for whatin effect is a three-dimensional problem the means

of vertical circulation (eg staircases) have to beestablished True lengths of staircases must bespecified based on the total length of all flights(sloping length) and the intermediate rests The fireexits need to be defined both in terms of theirlocation and width The links have to be establishedto define the point of the connection between thefloors and staircasesescalatorsramps The travellingdistance maps are then calculated type and numberof occupants specified building populated and theevacuation simulation performed

SYSTEM TESTING AND VALIDATIONKuligowski and Peacock (2005) define and recommendthe following ways of validating evacuation models

validation against code requirements validation against fire drills or other peoplersquos

movement experimentstrials

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 113

Gateline Vertical

Platform Right

Stairs Horizontal Stairs Vertical

Travelator Vertical

Ticket Hall

BMOAdmin Reception

Platform Top Ram Up Bottom Vertical

Lift Passage Horizontal

FIGURE 13 Examples of SGEvac station components

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

114 G SMITH AND B CERANIC

FIGURE 15 SGEvac hierarchy function

FIGURE 14 Layout module graphical user interface

validation against literature on past evacuationexperiments

validation against other models third-party validation

Furthermore Galea (1999) deconstructs the validationprocess into four main areas ndash component testingfunctional testing qualitative validation andquantitative validation This approach is more suitablefor models such as SGEvac than for those based oncompliance with code requirements and hence it isthe one adopted for the validation in this research

Component testing is regarded as testingcomputer code that drives the model checking forelementary errors Functional validation involveschecking that the model possesses the ability to

exhibit the range of capabilities required to performthe desired simulations Component testing andfunctional validation for SGEvac is carried outagainst code requirements insofar that allcalculations for infrastructure sizing are evaluatedmanually so that the method and answer are knownto be correct It uses the Simulex program to aid themodelling of people movement and to test thehypothesis of improvement to evacuation timesthus relying on Simulexrsquos own rigorous validationprocess Simulex has been used extensively inpractice as well as a validating model for otherevacuation softwares (Kuligowski and Peacock2005) SGEvac has gone through its initialcompletion and validation process with finalvalidation proposed for the end of the year

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 115

FIGURE 16 Auto-generated SGEvac station layout export file for fire

evacuation simulation in Simulex

FIRE EVACUATION SIMULATION RESULTSThe considered lsquotrain on fire in stationrsquo scenario takesinto account the service interval of 333 minutes(based on 18 trains per hour) the crush capacity of afour-car train together with the number of peoplewaiting on the platform after cancellation of oneservice (ie after 666 minutes) On the oppositeplatform the evacuation load is taken to be thenormal boardingalighting load assuming nocancellations to services (50 of all boarders andalighters)

Evacuation load predictions were based on TfL2006 projected passenger flows with an averageflow of 36 peopleminute for boarders and 123peopleminute for boarders and alighters combinedThe train crush capacity was averaged at 165peoplecar The simulation also models evacuationfrom the ticket hall level to assess the impact ofdispersing crowds at peak times in this part of thestation From these figures the total evacuation loadwas calculated as follows

Platform 1 Boarders (36 x 666 min) = 240Train crush load (165 x 4) = 660

Platform 2 Boarders and alighters

(123 x 05 x 333 min) = 205Ticket hall = 470Total station evacuation load = 1575 people

The simulation results have been captured as imagesat 25 50 100 and 300 second intervals respectivelyfor both platform and ticket hall levels as shown inFigures 17 to 20

The provided means of escape follow LU SPSG4requirements insofar that the maximum travellingdistance on a platform to an exit is no more than 45 m (ie exits must be no more than 90 m apart) andthat the dead-end travelling distance is no more than20 m to an exit Hence each platform is provided withthree exits The MIPsrsquo (mobility impaired persons)means of escape is provided from both ends ofplatforms via emergency evacuation lifts Totalevacuation time (ie all people reaching an exit onticket hallground level) was 5 minutes and

24 seconds with the platform and train load clearedwithin 3 minutes and 5 seconds This is within therequirements to evacuate the platform and train loadwithin 4 minutes with all people reaching a fire-separated route within 6 minutes (the ticket hall is notclassed as a fire-separated route)

The performed simulation was based on theminimum travelling distance maps resulting in eachperson exiting via the nearest fire exit route Theaverage 10 second lsquotime to start to moversquo wasassigned to all occupants in order to accommodatefor real world people behaviour commonly reportedby fire evacuation research studies The boarders onthe end of both platforms (including most peoplefrom end carriages of the train) evacuated viastaircasesMIP lifts directly to the ground level pointof escape The boarders situated around the middleof platforms together with the people from the twomiddle carriages used the central escalators to reachticket hall fire exits Additionally some people fromthe end carriages and some staffretail unitcustomers used this route as means of escape

In this lsquotrain on fire in stationrsquo scenario it wasassumed that none of the escapes routes areunavailable due to maintenance However furtherevacuation studies will investigate possible scenarioswhere one of the escalator routes may be unavailabledue to maintenance reasons

FUTURE RESEARCHIntended future work will concentrate on the furtherconditioning developing and modelling of shapegrammar design rules based on the outcomes of thefinal system validation and LU expertsrsquo feedbackFurther research is proposed on the development ofa novel layout optimization approach based on theshape grammar rules to investigate the evacuation-related parametric improvements in subsurfacesurface railway station design (eg minimizingtravelling distances)

The fire safety strategy in terms of performance-based fire risk design is an important aspirationalaspect of SGEvac requiring further research andliaison with experts to determine how such variedcriteria could be allied with layout planning This isparticularly important where deviation from standardsoccurs Thus further development of the expert

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

116 G SMITH AND B CERANIC

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 117

FIGURE 17 Station evacuation after 25 seconds

FIGURE 18 Station evacuation after 50 seconds

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

118 G SMITH AND B CERANIC

FIGURE 19 Station evacuation after 100 seconds

FIGURE 20 Station evacuation after 300 seconds

knowledge database would be required particularlyin the light of frequent changes to legislation

On a more practical note it could be beneficial todevelop the system on a stand-alone modular basislinked to each generated station component A list offire safety strategy objectives and functional logicstatements can then be specified for commonlyrecurring items (such as emergency lighting fire alarmdetectors sprinkler heads etc) and spaced accordingto the designerrsquos wishes or via shape rule formulation

Site constraint issues would also benefit fromfuture development as well as a move towardsdevelopment of the 3D shape grammar version Oncedevised other relevant 3D modelling systems such ascomputational fluid dynamics (CFD) smoke movementsoftware could be utilized in conjunction with SGEvac

CONCLUSIONSThe computer-aided modelling environment underdevelopment is intended to act as a layout planning aidfor those who practice in sub-surfacesurface railwaystation design The system under developmentcarefully considers London Underground StationPlanning Standards and Guidelines together withother relevant fire safety regulations that influenceoverall fire strategy design The proposed computer-aided design environment (SGEvac) is based on highlyrestrictive additive shape grammar rules driven by akernel of key passenger flow information to generatestation plan solutions

The major benefit to the design team is the ability toreview multiple layouts of accurately sized and correctlyplaced infrastructure at the reference or preliminarystage of the station layout planning design process Timeis hence saved regarding different option appraisals andquantitative testing using evacuation software shouldlead to a choice of more effective layouts in relation tomeans of escape travelling distances and improvementin the total evacuation times Modifications to layoutsshould be straightforward and there is scope toefficiently assess travelling distances and evacuationtimes for each of the considered alternatives comparedwith a lsquotraditionalrsquo approach

The proposed novel shape grammar approach ismost suited to new-build stations or largerefurbishment projects where more design flexibilityexists Sites with many site-specific constraints have

an array of predetermined parameters and will mostlikely benefit from the manual operation of SGEvac

Use in other rail systems due to recurringinfrastructure components and inherent functionallogic related to the laying out of the stationcomponents is also feasible provided that shapeconnections and shape rules are corrected to reflectthat countryrsquos legislative requirements

The potential for station layout improvementswith regard to emergency evacuation is an area thatmerits research especially in scenarios that concernthe safety of the general public and form a vital partof Londonrsquos transportation system Potential layoutimprovements could not only benefit evacuationsituations but also enhance the through-flow ofpeople during peak travelling periods

ACKNOWLEDGEMENTS The authors wish to express their gratitude to JacobsEngineering (UK) Ltd which has sponsored theresearch and to London Underground Ltd forproviding the design guidance that is reproduced inthis paper and for authorising the use of the stationplans London Underground has not reviewed thispaper and the content is solely the authorsrsquo work Mr Robert Hutchinson is also thanked for hiscomputer programming assistance

AUTHOR CONTACT DETAILSGGrraahhaamm SSmmiitthh (corresponding author) Senior Architectural

Technologist Jacobs (UK) Ltd Cardinal Square North Point

10 Nottingham Road Derby DE1 3QT UK Tel +44 (0)1332 234482

fax +44 (0)1332 344558 e-mail grahampsmithjacobscomBBoorriiss CCeerraanniicc Programme Leader for BScBA (Hons) Architectural

Programmes Room Number 105 Faculty of Arts Design and

Technology University of Derby Markeaton Street Derby DE22

3AW UK Tel +44 (0)1332 593136 e-mail bceranicderbyacuk

REFERENCES BS 7974 2001 Application of Fire Safety Engineering Principles to the

Design of Buildings London BSI

Cha MY and Gero JS 1999 Style Learning Inductive Generalisation of

Architectural Shape Patterns Paichai University and Sydney University

co-paper

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 119

Chase SC 1996 lsquoUsing logic to specify shapes and spatial relations in

design grammarsrsquo in Workshop notes Grammatical Design Fourth

International Conference on Artificial Intelligence in Design Stanford

University US June

Department of the Environment 1996 Design Principles of Fire Safety

London The Stationery Office Books

Flemming U 1987 lsquoMore than the sum of its parts the grammar of

Queen Anne housesrsquo in Environment and Planning B Planning and

Design 14 323ndash350

Fruin JJ 1971 Pedestrian Planning and Design New York Metropolitan

Association of Urban Designers and Environmental Planners

Galea ER 1999 lsquoA general approach to validating evacuation models with

an application to EXODUSrsquo in Journal of Fire Sciences 16 414ndash436

Galle P and Kovacs LB 1992 lsquoThe logic of worms a study in

architectural knowledge representationrsquo in Environment and Planning B

Planning and Design 19 5ndash31

Garling T Lindberg E and Mantyla T 1983 lsquoOrientation in buildings ndash

effects of familiarity visual access and orientation aidsrsquo in Journal of

Applied Psychology 68(1) 177ndash186

Grimsdale RL and Chang CW 1996 lsquoThe layout design language A

technique for generating layout plansrsquo in Computer Graphics Forum

15(2) 97ndash106

Heskestad AW 1999 lsquoPerformance in smoke of way guidance systemsrsquo

in Fire and Materials 23(6) 375ndash381

Knight T 1995 lsquoConstructive symmetryrsquo in Environment and Planning B

Planning and Design 22 419ndash450

Knight TW 1999 lsquoShape grammars six typesrsquo in Environment and

Planning B Planning and Design 26 15ndash31

Kuligowski ED and Peacock RD 2005 A Review of Building Evacuation

Models Technical Note 1471 National Institute of Standards and

Technology (NIST) Technology Administration US Department of

Commerce

Li A I-Kang 2003 lsquoThe Yingzao fashi in the information agersquo conference

paper for the Beaux-Arts Paul Philippe Cret and Twentieth Century

Architecture in China The University of Pennsylvania October

Liew H 2003 lsquoSGML ndash A shape grammar meta-languagersquo in Digital

Design 21st eCAADe Conference Proceedings Graz Austria September

639ndash648

Loomis B 2003 lsquoA note on generative design techniques SGGA a user

driven genetic algorithm for evolving non-deterministic shape grammarsrsquo

working paper Massachusetts Institute of Technology (MIT)

Magnusson SE Drysdale DD Fitzgerald RW Mowrer F

Quintiere JG Williamson RB and Zalosh RG 1995

lsquoA proposal for a model curriculum in fire safety engineeringrsquo

in Fire Safety Journal 25 9ndash10

Smith G Ceranic B and Fryer C 2003 lsquoDevelopment of CAD

environment for spatial layout optimisation in sub-surfacesurface rail

station designrsquo in Proceedings of International BIAT Conference The

British Museum London January 2003

SPSG 2007 London Underground Station Planning Standards and

Guidelines SPSG4 (1ndash371) London

Stiny G and Gips J 1971 lsquoShape grammars and the generative

specification of painting and sculpturersquo in Information Processing 71

1460ndash1465

Stiny G 1977 lsquoIce-ray a note on the generation of Chinese lattice

designsrsquo in Environment and Planning B Planning and Design 4 89ndash98

Stiny G 1980 lsquoKindergarten grammars designing with Froebelrsquos building

giftsrsquo in Environment and Planning B 3 409ndash462

Tapia M 1999 lsquoA visual implementation of a shape grammar systemrsquo in

Environment and Planning B Planning and Design 26(1) 59ndash73

Transport for London 2007 Tube Carries One Billion Passengers for First

Time wwwtflgovukcorporatemedianewscentre4770aspx

(accessed 24 March 2008)

Weller M 2007 London Underground private correspondence

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

120 G SMITH AND B CERANIC

Stiny and Gips (1971) define shape grammar as

S = [VT VM R I] (1)

where VT is a finite set of shapes VM is a finite setof labels R is a finite set of shape rules and I is aninitial shape

In this research parameter VT is the finite set ofthe station infrastructure components shapes andVM is the finite set of labels that comprise thepossible station area permutations mapping twolsquolegally connectedrsquo components into a spatial relationwithin the overall layout R is the finite set of theshape connectivity rules composed of informationcontained in regulations and guidance documents interms of the area assignments adjacency and theirconnectivity I is an initial shape (ticket hall in theSGEvac programme)

In summary a proposed shape grammar systemconsists of a set of vocabulary elements (stationcomponents) a set of rewrite rules and a start stateStation layout can be considered in terms of compositeshape generation and its design process in terms oflogical sequence and deductive reasoning It is thisapproach based on the function of the building thatallows for a transition from lsquotraditionalrsquo layout planningto the computational shape grammar layoutgeneration that is proposed in this research Howeverits application depends directly upon the expertrsquosknowledge filtered into this rule-based approach

An additive shape grammar approach (Knight1999) is proposed as a suitable technique for

generating station layouts This approach is useful forthe irregular station boundaries derived from sitespecific conditions where designs are generatedfrom a core (the ticket hall) to which othercomponents are successively added

Knight (1995) remarks that the rules detailed inStinyrsquos Kindergarten paper (1980) can be preacutecised toexternalize a studentrsquos design ideas so that they canbe examined changed and communicated morereadily It is the choice of credible shape grammaralternatives that is important not the multitude ofsolutions themselves Good analytic grammars areboth efficient and descriptive revealing simplicity orregularities behind seemingly complex designs

STATION LAYOUT DESIGNCONSIDERATIONSPassenger space planning needs to be considered interms of peoplersquos procedures when they use theUnderground Initially customers will enter thecomplex from either street level or via a surfacestation stairwell A one- or two-way passage will thentake passengers to the unpaid side of the ticket hallwhere they will purchase a ticket before passingthrough to the paid side

Customers then descend using stairsescalatorslifts to platform level At both the top and bottom ofstairs and escalators there is a station componenttermed the run-off area The length of the run-offdepends on the peak passenger flow capacity of theescalatorstairwell and whether it connects to aplatform passageway or street It is an area wherepeople move away from the change in level orientatethemselves and make a decision of where to go next(SPSG 2007)

Generally at platform level people will journey tothe desired line to take them to their destination via afurther passageway or lobby area Some escalatorsstairs can exit directly onto platforms The travelprocess is reversed when alighting from a train andcontinuing to the surface

The predicted passenger flow rates and sitespecific conditions dictate both the number of railwaylines and the platform(s) configuration (side centre orstacked platform) Station composition must alsoconsider areas that are not sized by passengervolume demand but are required to ensure that the

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 101

INITIAL SHAPE

INITIAL SHAPE

RULE 1 RULE 1 RULE 1 RULE 1

RULE 2RULE 2RULE 2RULE 3

FIGURE 3 Labelling shape rule for deterministic grammar

designs (Loomis 2003)

station operates satisfactorily on a daily basis Thestation managerrsquos office plant and staff roomscomputer rooms train technician rooms etc all needaccommodating in the proposed shape grammarformalism In addition ancillary accommodationsuch as the cleanerrsquos store police room publicitystore retail etc also have to be catered for

To aid designing and sizing infrastructure levelsof service as shown in Figure 4 are used by LondonUnderground These are based on the work of Fruin(1971) where a detailed study of crowd movementwas undertaken Level A represents free circulationwith subsequent levels growing progressively morerestricted in terms of people movement from B to Eculminating with level F that represents a completebreakdown in flow with numerous stoppages

The approach proposed in this research definesthe generic station layout arrangements from thelsquohigh-levelrsquo spatial site organization to the subsequent

specified components and their logical spatialallocations as illustrated in Figure 5 The consequentshape grammar rules can then be deduced taking intoaccount the predicted passenger flow rates fire safetystrategy and relevant design regulations

FIRE SAFETY DESIGN STRATEGYBefore any design commences a clear identificationand appraisal of the key constraints and parametersat the site layout planning stage is required Forexample deep foundations mains servicesarchaeological discoveries and even the RiverThames will affect station design decisionsFurthermore structural appraisals geotechnicalsurveys Statutory Undertaker searches groundcontamination testing environmental impactassessments and land acquisition etc all need to becarried out as part of a feasibility study before stationlayout planning at reference level commences

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

102 G SMITH AND B CERANIC

Level of service A

Level of service D Level of service E Level of service F

Complete breakdown in trafficflow with many stoppages

Free circulationDescription (forqueuing areaswalkways andstairways)

Uni-directional flows andfree circulation Reverseand cross-flows withonly minor conflicts

Restricted circulation formost pedestriansSignificant difficulty forreverse and cross-flows

Restricted circulation for allpedestrians Intermittentstoppages and seriousdifficulties for reverse andcross-flows

Sligthly restricted circulationdue to difficulty inpassing others Reverse andcross-flows with difficulty

Level of service B Level of service C

FIGURE 4 London Underground levels of service (SPSG 2007)

Such constraints may be viewed in a positivecontext in terms of services already being providedhelping to justify the business case for a new stationDesign procurement strategies may include decisionssuch as locating toilet areas within x metres of anexisting sewer route or locating electrical plant roomswithin y metres of an existing electrical sub-station

In the real world parameters such as these constrict layout production and the choice ofinfrastructure component placement as opposed toleading to the production of many solutions that arelikely to be discarded when for example projectcosts are reviewed This level of regularity andrestriction typical in station design is very suitable

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 103

SITE

CUT AND COVER STATION TUNNELLED STATION

TICKET HALL

UTS GATELINE (gateline dimensionsdetermine minimumticket hall width)

PASSAGEWAYS ONE- or TWO-WAY

(entrances exits determinea minimum uniform width)

ESCALATOR

STAIRS

RAMPS

TOP RUN-OFF

TRAVELATOR

PAID CONCOURSE SIDE

BOTTOM RUN-OFF

PASSAGEWAYS ONE -WAY or TWO-WAY

NON-PASSENGERSPACE ELEMENTS ATTICKET HALL LEVEL (egStation Managerrsquos Office)

PLATFORM LEVEL

SPACE ELEMENTS ATPLATFORM LEVEL (egplant rooms toilets etc)

SIDE PLATFORM CENTRE PLATFORM STACKED PLATFORM

No OF RAILWAY LINES

MEZZANINEOptional

MEZZANINE

DEEP

SHALLOW

LIFTS

NON-PASSENGER

FIGURE 5 lsquoHigh-levelrsquo layout hierarchy for generic station composition

for the application of a prescriptive shape grammarapproach

In parallel with the shape grammar developmentfunctional logic statements are proposed analogousto the qualitative design review as described in BS7974 Application of Fire Safety EngineeringPrinciples to the Design of Buildings The intention isto develop SGEvac procedures in a manner that isrecognized by fire safety engineers and designers inpractice The Department for the Environment (DOE)(1996) states that lsquothe architectural layout can have asignificant influence on peoplersquos movementComplexity in circulation routes may not be overcomenecessarily by signage these may not have beennoticed or they may have been forgotten as a resultof stressrsquo Hence station spatial requirements can bereviewed alongside a number of functional logicstatements devised in conjunction with the layoutplanning

For example lsquothe probability of successful escapethrough smoke diminishes sharply with (an increasein the) number of decision elements included in theescape pathrsquo (Hesketad 1999)

A functional logic statement response to theabove would be to minimize such decision elementsthus enabling faster egress Other examples offunctional logic statements employed are

no corridor or staircase will reduce in width alongescape routes

length of the passages are to be minimized maximize level areas to facilitate faster travel

speeds for escapees maintain uniformity to aid way-finding and relative

orientation evacuation and normal circulation routes are to

be the same exits will be prominent easily identifiable and

straightforward to reach

Such statements are intuitive by their nature and arepart of the qualitative reviews in the traditional designprocess Due to time constraints only a limitednumber of possible alternatives in terms of the spatiallayout improvements are ever considered if any Incontrast SGEvac offers not only the opportunity toconsider a significant number of possible alternatives

but also the means of quantitative assessment via fireevacuation simulations

Garling et al (1983) contend that newcomers to anenvironment first learn a number of salient locationsthen the paths in between and finally organize theacquired knowledge in a spatial system Indeedavoiding symmetry and lack of differentiation fromthe userrsquos viewpoint is a good practice for stationdesign considering perception of relative orientationin the station This is where interaction between thedesigner and computer program becomes essentialand it is in this context that the research argues therole of SGEvac as a computer-aided design tool notas a substitute for the designer

A comparative methodology overview is given inTable 1 The processes of BS 7974 the procurementof a general logic model (Galle and Kovacs 1992) andthe shape grammar development stages arebenchmarked against SGEvac procedures Thisperhaps has most relevance to fire safetyengineering since it is with this objective that theproposed shape grammar system is to be used inpractice

Research studies such as one by Magnusson et al(1995) note that fire safety engineering has motivateddesigners and regulators to examine alternative codecompliance approaches based on a greater relianceupon quantitative analyses This has led toreassessment of prescriptive approaches andconsequent adjustments as required It is nowaccepted that existing prescriptive practices shouldbe evaluated by quantitative performance-basedmethods as a part of the overall fire safety strategyas encouraged in the Regulatory Reform (Fire Safety)Order (RRFSO) 2005 SGEvac follows this trendaugmenting the array of tools available forperformance-based fire safety engineering design

There is obviously an inevitable need for designerinput and control at all stages of the design processDue to the potential necessity for deviation fromstandards via fire safety engineering judgement theproposed shape grammar approach cannot alonemodel project-specific concessions

Still the movement away from prescription viaBS 7974 and the RRFSO to risk-based judgementplaces computer-aided design tools such as SGEvacin a position of interdependency with other design

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

104 G SMITH AND B CERANIC

criteria neither can be formulated in isolation Thisreinforces the view that computers will in future playan ever-increasing role in aiding and enabling thedesign teams but will never be able to replace them

SYSTEM DEVELOPMENTFigure 6 describes the modular approach proposedfor the system development offering the addedpotential of dual-purpose application in the designprocess It allows the designer to use the systemeither as a whole or to take advantage of itsindependent modules The system offers designerscomplete control over the shape grammarcomposition and layout modifications together withthe constant monitoring facilities and ability todiscard any solutions deemed to be lsquoinferiorrsquo

The proposed system is partitioned into threemain modules ndash lsquoexpert knowledgersquo lsquolayoutrsquo andlsquotestingrsquo modules The expert knowledge modulecaptures the expertise available to facilitate accurateshape grammar generations of layout design Thismodule provides the designer with the ability tointeractively reference Station Planning Standards

and Guidelines (and other relevant fire safetyregulations) allowing himher to make necessary site-specific decisions and to control the lsquohigh levelrsquo shapegrammar composition rules

The designer specifies passenger flow rates anddesignates the overall site boundary dimensions thatdefine the permissible station footprint The final exitpositions are also stipulated at this stage with theseoften being constrained by the existing builtenvironment

With this data supplied the shape grammargenerations commence in the layout module Thearea assignments adjacency requirements and layoutcomposition rules are defined in detail at this stageOnce the designer has made decisions with regard tothe layout composition the computer becomes thecentral tool for generating searching and logicallysorting through the permissible permutations ofpossible layouts The user has the option to auto-generate layouts (usually for sites without manyconstraints) or to manually control the layoutassembly (a decision normally associated with heavilyconstrained sites)

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 105

TABLE 1 Methodology comparison BS 7974 PROCESSES GALLE AND KOVACS SHAPE GRAMMAR SGEvac

LOGIC MODELLING DEVELOPMENT IMPLEMENTATION

Assess benefits Identification Shape grammar Assess constraints

and constraints system identification identify benefitssite

boundariesfinal exit

positions

Qualitative design Conceptualization Initial state Station component

review identification and

interactions

Quantitative analysis Formalization Rules and shape Legislative compliance

language passenger flow figures

area assignments

Satisfactory Implementation Shape grammar Alternative station

unsatisfactory executions layouts generation

Report and review Testing Design analysis Testingvalidation

iterationsevolution of layouts generation

End Experimentation Experimentation

iterationsimprovements

The testing module is used to evaluate potentiallysuitable layouts at reference level by exporting them toAutoCAD via standard drawing data exchange formatand analysing them in the context of project-specificsite constraints and requirements A level of drawingdetail is then added and the layout solution is exportedto Simulex a building evacuation simulation software

package Once the occupant numbers and persongroups are assigned and the escape and verticalcirculation routes defined the simulation software iscapable of the accurate measurement of all travellingdistances creation of distance maps and calculation ofevacuation times for different fire scenarios (egdifferent group affiliations persons exiting via different

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

106 G SMITH AND B CERANIC

EGY D ESIGNDesigner

Feasibility studiesProject brief requirements

Passenger flow data

DDE

DDE

Layout Production

Shape grammar rulesMultiple layout production

Layout Analysis

Layout interrogationsDesign team judgement

No

Yes

DXF

DXF

Yes

No

AutoCAD

Feed lsquoblock reference planrsquoresults into overall project

Improvedsolution identified

EXPERT KNOWLEDGE MODULE

Fire Strategy Design

LU SPSG4FSE legislation amp guidance

Infrastructure size calculationHigh level brief development

LAYOUT MODULE

Layout Composition

Area and adjacency planningLayout option appraisal

Layoutimproved

TESTING MODULE

AutoCAD

SGEvac export into CAD

Simulex

Fire evacuation scenarios

DesignerSGEvac

DDE ndash Database Data Exchange

DXF ndash Standard Drawing Data Exchange

FIGURE 6 Proposed SGEvac algorithm

routes and having delayed movement in response to afire alarm signal) The design team then assesses theresults against the qualitative fire safety design goalsset earlier to establish whether improvements toevacuation times have been made

DETAILED ANALYSIS OF STATIONCOMPONENTSIn conjunction with the site boundary final exitpositions and passenger flow specificationconsideration must also be given to the vertical sitedimension The shape grammar system assumes allfinal exits are at ground level with the ticket hall andplatform(s) being the other two elements that requirethe user to determine their depth relative to the finalexits In so doing SGEvac selects the correctinterchange components by an event-drivenmechanism that evaluates the height change andpassenger volume demand Thresholds for bothcriteria have been defined as shown in Table 2 andEquations 2ndash5

If H (height in mm) = 0 ndash 500 then I (infrastructure) = Ramp (2)

If H = 500 ndash 3000 and P (passengers) le 3000 then I = Stairway (3)

If H = 3000 ndash 5000 And P le 3000 then I = StairwayElse I = Escalator (4)

If H gt 5000 and P le 1000 then I = StairwayElse I = Escalator and Lift (5)

(where lsquoElsersquo denotes the mechanism for the choiceof the alternative infrastructure component if theoriginal condition is not satisfied)

The above constructs also condition the run-offdimensions associated with each of theinterchange components (between 4 m and 12 mas stipulated by SPSG) This being the case run-offs are used to size the ticket hall length Figure 7shows the paid side of the ticket hall with variabledistances shown between the escalator or staircaseand the gateline

The paid side of the ticket hall is always dictatedby run-off distance while the unpaid side is governedby two parameters the run-off and the concoursearea required for the heaviest 15-minute passengerflow If the maximum 15-minute passenger flowfigure multiplied by 1 m2 (say answer A) is larger inplan area than the product of the gateline widthmultiplied by the relevant run-off (say answer B) thenanswer A will be incorporated into the layout Thereverse is true if B is the greater number This ensuresthat the station is always sized appropriately

Figure 8 shows a ticket hall layout with thegateline dictating the width (denoted as Tick_H_wd)itself conditioned by the passenger flow figurescalculated as shown in Table 3 The table onlyevaluates entry flow requirements but SGEvac willrepeat the same process for exit gates to produce acorrectly sized gateline

In this case the maximum 15-minute passengerflow capacity size prevailed over the run-off to createthe unpaid side overall

The divide between the paid and unpaid side ofthe ticket hall is the sized according to Table 4

SGEvac shape rules mandate that entry gates arealways located adjacent to the ticket office queuingspace (designated as Q_TIW) to avoid conflictingpassenger entry and exit flows The LU standard is to

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 107

TABLE 2 Interchange infrastructure selectionINFRASTRUCTURE REASON FOR SELECTION

Ramp Vertical rise between 0ndash05 m preferred to steps due to disabled access issues generally considered

people will travel faster on ramps than up or down steps

Stairs Vertical rise between 05ndash5 m but greater height available for selection if station has a light passenger flow

Escalator Vertical rise is minimum of 3 m for heavy passenger flow stations but used for lightmedium flow stations

above 5 m

Lift Vertical rise above 5 m also required for accessibility of disabled persons Lifts always used with escalators

not as a substitute

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

108 G SMITH AND B CERANIC

Con_Area

Exit

UTS gates

6601900 2450

2400

EFW

Ticketoffice

POMcounter

4000

Tick_H_wd

1000

Run

-off

Entry

Q_space

Q_TIW

FIGURE 8 SGEvac ticket hall arrangement

Heavy flow escalator

Escalator

Staircase

Medium to Heavy flow

Medium to Heavy flow

Light flow escalator

Light flow

Heavy flow 12mMedium flow 8ndash12m

Light flow

Light flow 8m

Light flow 6m

Heavy flow

staircase to gateline

staircase to gateline 10m

Light flow

to gateline12mGateline

Gateline

EscalatorndashGateline

Total Run-off

Heavy flow 10mMedium flow 6ndash10m

StaircasendashGateline

Total Run-off

to gateline 8m

2m 4m 6m 8m 10m

2m 4m 6m

6m

8m 10m

12m

FIGURE 7 Passenger flow rate run-off requirements (SPSG 2007)

allow for 4 m (or a five-person queue) as shown inFigure 8 The overall gateline is defined by summingboth exit and entry gate numbers obtained from user-inputted passenger figures

The length of the ticket office interface with theticket hall is stipulated by the number of ticket issuingwindows (TIWs) plus excess fare windows (EFWs)TIWs always belong on the unpaid side of the tickethall and EFWs always on the paid side both forrevenue protection reasons This situation facilitates aprecise shape grammar labelling for the ticket officeto ticket hallqueue space connection

Passenger operated machine (POM) enclosures arenormally connected to the ticket hall ticket office andits own queue space (noted as Q_space in Figure 8)

To make shape rule specification manageable allstation areas are divided into groups arranged to suitthe most probable connection patterns in the stationTable 5 shows the group station managerrsquos officeexample with all other station components followingthe same concept

London Underground views its stations as largebasements (Weller 2007) and escape distances fromplatforms are based on agreed dispensations with theLondon Fire Brigade SGEvac respects thesedistances by placing restrictive labels on the platformso that they cannot be exceeded as demonstrated bythe example shown in Figure 9

SGEvac escape stair infrastructure may bepositioned up to 20 m from the end of platforms toavoid undesirable lsquodead endrsquo conditions The 45 mmaximum platform travel distance starts from thenearside junction of the two 33 m wide escapestaircases with the 45 m distance rule creating ashortfall in the platform centre of 34 m Generallylocating three interchange components (eg stairs

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 109

TABLE 3 Gateline width calculationMETHOD WORKED EXAMPLE

Obtain peak hourly flow rate through Say 6500 peoplehr

ticket hall

Calculate 5-minute peak passenger flow 650012 = 541

Allowance for ticket rejection 541 + 20 = 561

Allowance for growth in use demand 561 x 110 = 617

Assume 125 people use each gate over

the 5-minute peak passenger flow period 617125 = 493

Round up gates to nearest whole number 5 gates required

TABLE 4 London Underground ticket hall gateline parametersregulations (SPSG 2007)EQUIPMENT DIMENSIONS (W X D) POSITIONARRANGEMENT

Combined manualequipment gate 1750 x 270 mm Height ndash 1300 mm Adjacent to ticket office Can include

1400 mm clear opening luggage facility

Manual gate ndash passenger 1750 x 270 mm Height ndash 1300 mm Adjacent to ticket office With more than 6

gateluggage port 900 mm ndash clearance and walkways in a single gateline or significant

outward opening interchange with mainline stations airport or

tourist traffic provide separate luggage chutes

Manual gate ndash single leaf 1394 mm Height ndash 1300 mm Beside the dedicated excess gates

with luggage port 695 mm clearance

Gateline attendance report 990 x 1240 mm Within the controlled zone but not necessarily

part of the gateline Usually close

Station control unit (SCU) 432 x 430 mm Should be installed within a secure enclosure

(which could be a ticket office) and also with

a clear view of the gates Controls and monitors

the station underground ticketing system

equipment

Manual emergency exit gates 1900 x 125 mm Adjacent to ticket office Add these gates rather

than adding extra UTS gates for evacuation

purposes

or escalators) at these distances will result incompliance with the Underground standards but itmay not always be possible (or practical for aninterchange situation) to procure such a layout

Lengths of station components are controlled bya range of criteria For the platform it is the length ofthe train vehicle for interchange components it isthe vertical height in the terms of the depth to theplatform level for passages it is the designated finalexit positions and the existing station areas alreadygenerated

Vector direction labels are underpinned by thepermissible union of compatible spaces two points ofcontact will be maintained so as not to restrict thewidth of connecting passenger areas This undueconstriction of passenger flow at the head of apassage also satisfies realistic construction practiceNon-passenger and ancillary area vector directions areas per project requirements and the successive areaadditions managed by the SGEvac rule groupings

To create alternative layouts SGEvac uses alsquorandomizersquo rule In this shape grammar rule initialstate infrastructure is generated as per guidelinesbut it can randomly change the location ofcomponent arrangements within the predefined

boundaries so that a number of possible alternativelayouts are produced The designs can then beinterrogated and ranked by their travelling distancesand against the qualitative functional logicstatements

The SGEvac design tool developed in this researchis based on a highly restrictive additive grammar ruleset due to SPSGrsquos prescriptive guidance and thenature of the layout planning problem As Grimsdaleand Chang (1996) argue many published grammarslack high-level reasoning during the iteration processsince they are often concerned with the developmentof shape grammar theories rather than their practicalapplications By contrast SGEvac being a practicalapplication has a complete set of primitive shapesonly possible to arrange in a certain way It thusprovides viable reference design layout compositionsbased on SPSG with further interventions possible bythe designer

SYSTEM DEVELOPMENT TO DATEThe SGEvac prototype is compiled within the Clanguage environment allowing for modular object-oriented programming (suitable for shape grammar)Figure 10 shows the main interface where the stationname and relevant LU fare zone are entered (whichimpacts on sizing infrastructure) as are the ticket halland platform depths the overall site boundaries andfinal exitsrsquo positions The peak passenger flow figuresover the relevant time interval of normally 3 hours arealso specified SGEvac is then capable of accuratelycalculating all passenger area infrastructure based onthe SPSG guidelines as already discussed

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

110 G SMITH AND B CERANIC

TABLE 5 Station manager suite ndash shape rule connectionsELEMENT SHAPE RULE DESIGNATION CONNECTED COMPONENT GROUP ALLOCATION

Group station managerrsquos office SR_GSM_01 Ticket hall paid side GSM

SR_GSM_02 Ticket hall unpaid side

SR_GSM_03 GSM administration

SR_GSM_04 Duty station managerrsquos office

SR_GSM_05 Business managerrsquos office

SR_GSM_06 Administration

SR_GSM_07 Ticket office

SR_GSM_08 Police room

SR_GSM_09 Revenue protection examiner

SR_GSM_10 Interview room

140

Stair Stair

100 2020

4533

34

45

FIGURE 9 Platform escape distance SGEvac label restrictions

The history of auto-generated layouts is accessiblefrom this screen and the user is given the capabilityto delete and copy them (useful when minoralterations are required to produce an alternativelayout)

Shape labelling in the programme is multi-faceted which is necessary since many stationcomponents have various possible connections Anescalator for example may connect to a passagetravelator ramp escalator run-off stair run-off lift orthe ticket hall so all such connections need adding tothe component properties The SGEvac grouping ofcomponent interactions is accessed from menutoolbars that allow insertion of legislative compliantstation area connections as shown in Figure 11

Once the position and the depth of the ticket halland platform are specified the railway track is inserteddependent on the platform configuration The platformis the only possible connection to the track but thetrack can be positioned either side of platforms so thata side or centre (island) platform layout is possible

Non-passenger and ancillary areas are alsoselectable from the shape grammar library via property

windows for each station component eg the ticketoffice properties window as shown in Figure 12

Figure 13 shows examples of the layout modulestation components available in the SGEvacprogramme

Figure 14 shows the layout module interfaceInformation on the components (relative position inthe overall site layout and its absolute sizing) isdisplayed at the bottom of the screen once thecomponent is selected from the list In addition ameasurement bar (in metres) at the base of theinteractive design environment provides a sense ofscale for the designer

A checklist summarizes the station areas includedin the layout but a more detailed analysis is possiblevia a report function if required In addition to thisdetailed list a summary screen accessed from themain interface collates all specified components ineach generated layout This screen lists all passengernon-passenger and ancillary areas in their respectivecategories with infrastructure width data ofpassenger areas also summarized It also has anevacuation summary where the train frequency is

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 111

FIGURE 10 SGEvac main interface

specified and the train evacuation load (taken as peaklsquorush hourrsquo loading) is displayed This section of thesummary screen is useful for input into Simulexwhere fire evacuation test runs of the building layoutcan be implemented

The lsquoshow hierarchyrsquo function in Figure 15 allowsdesign team members to view the progression ofdesigns by superimposing connecting lines betweenstation elements in the order of their compositionThis also serves as a safeguard against deleting

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

112 G SMITH AND B CERANIC

FIGURE 12 SGEvac ticket office properties window

FIGURE 11 SGEvac properties window

components out of sequence that could lead to aninvalid solution

Once auto-generated layouts are complete withcorrectly sized and legitimately connected stationcomponents manual insertion of non-passenger areaswould then commence Upon completion thegenerated layout is transferred to AutoCAD via dxf fileformat as shown in Figure 16 Once in AutoCAD thelayout has a level of detail added by the designer suchas door openings passage connections fire escapeopenings etc so that passenger flows may beassessed It is then exported to Simulex for fireevacuation simulation analysis again via dxf file format

Simulex understands any voids as potentialcirculation routes treating the rest as solid andpermanent obstructions ie walls internal structuralelements permanent fixtures etc Since its solutionoperates in a two-dimensional environment for whatin effect is a three-dimensional problem the means

of vertical circulation (eg staircases) have to beestablished True lengths of staircases must bespecified based on the total length of all flights(sloping length) and the intermediate rests The fireexits need to be defined both in terms of theirlocation and width The links have to be establishedto define the point of the connection between thefloors and staircasesescalatorsramps The travellingdistance maps are then calculated type and numberof occupants specified building populated and theevacuation simulation performed

SYSTEM TESTING AND VALIDATIONKuligowski and Peacock (2005) define and recommendthe following ways of validating evacuation models

validation against code requirements validation against fire drills or other peoplersquos

movement experimentstrials

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 113

Gateline Vertical

Platform Right

Stairs Horizontal Stairs Vertical

Travelator Vertical

Ticket Hall

BMOAdmin Reception

Platform Top Ram Up Bottom Vertical

Lift Passage Horizontal

FIGURE 13 Examples of SGEvac station components

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

114 G SMITH AND B CERANIC

FIGURE 15 SGEvac hierarchy function

FIGURE 14 Layout module graphical user interface

validation against literature on past evacuationexperiments

validation against other models third-party validation

Furthermore Galea (1999) deconstructs the validationprocess into four main areas ndash component testingfunctional testing qualitative validation andquantitative validation This approach is more suitablefor models such as SGEvac than for those based oncompliance with code requirements and hence it isthe one adopted for the validation in this research

Component testing is regarded as testingcomputer code that drives the model checking forelementary errors Functional validation involveschecking that the model possesses the ability to

exhibit the range of capabilities required to performthe desired simulations Component testing andfunctional validation for SGEvac is carried outagainst code requirements insofar that allcalculations for infrastructure sizing are evaluatedmanually so that the method and answer are knownto be correct It uses the Simulex program to aid themodelling of people movement and to test thehypothesis of improvement to evacuation timesthus relying on Simulexrsquos own rigorous validationprocess Simulex has been used extensively inpractice as well as a validating model for otherevacuation softwares (Kuligowski and Peacock2005) SGEvac has gone through its initialcompletion and validation process with finalvalidation proposed for the end of the year

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 115

FIGURE 16 Auto-generated SGEvac station layout export file for fire

evacuation simulation in Simulex

FIRE EVACUATION SIMULATION RESULTSThe considered lsquotrain on fire in stationrsquo scenario takesinto account the service interval of 333 minutes(based on 18 trains per hour) the crush capacity of afour-car train together with the number of peoplewaiting on the platform after cancellation of oneservice (ie after 666 minutes) On the oppositeplatform the evacuation load is taken to be thenormal boardingalighting load assuming nocancellations to services (50 of all boarders andalighters)

Evacuation load predictions were based on TfL2006 projected passenger flows with an averageflow of 36 peopleminute for boarders and 123peopleminute for boarders and alighters combinedThe train crush capacity was averaged at 165peoplecar The simulation also models evacuationfrom the ticket hall level to assess the impact ofdispersing crowds at peak times in this part of thestation From these figures the total evacuation loadwas calculated as follows

Platform 1 Boarders (36 x 666 min) = 240Train crush load (165 x 4) = 660

Platform 2 Boarders and alighters

(123 x 05 x 333 min) = 205Ticket hall = 470Total station evacuation load = 1575 people

The simulation results have been captured as imagesat 25 50 100 and 300 second intervals respectivelyfor both platform and ticket hall levels as shown inFigures 17 to 20

The provided means of escape follow LU SPSG4requirements insofar that the maximum travellingdistance on a platform to an exit is no more than 45 m (ie exits must be no more than 90 m apart) andthat the dead-end travelling distance is no more than20 m to an exit Hence each platform is provided withthree exits The MIPsrsquo (mobility impaired persons)means of escape is provided from both ends ofplatforms via emergency evacuation lifts Totalevacuation time (ie all people reaching an exit onticket hallground level) was 5 minutes and

24 seconds with the platform and train load clearedwithin 3 minutes and 5 seconds This is within therequirements to evacuate the platform and train loadwithin 4 minutes with all people reaching a fire-separated route within 6 minutes (the ticket hall is notclassed as a fire-separated route)

The performed simulation was based on theminimum travelling distance maps resulting in eachperson exiting via the nearest fire exit route Theaverage 10 second lsquotime to start to moversquo wasassigned to all occupants in order to accommodatefor real world people behaviour commonly reportedby fire evacuation research studies The boarders onthe end of both platforms (including most peoplefrom end carriages of the train) evacuated viastaircasesMIP lifts directly to the ground level pointof escape The boarders situated around the middleof platforms together with the people from the twomiddle carriages used the central escalators to reachticket hall fire exits Additionally some people fromthe end carriages and some staffretail unitcustomers used this route as means of escape

In this lsquotrain on fire in stationrsquo scenario it wasassumed that none of the escapes routes areunavailable due to maintenance However furtherevacuation studies will investigate possible scenarioswhere one of the escalator routes may be unavailabledue to maintenance reasons

FUTURE RESEARCHIntended future work will concentrate on the furtherconditioning developing and modelling of shapegrammar design rules based on the outcomes of thefinal system validation and LU expertsrsquo feedbackFurther research is proposed on the development ofa novel layout optimization approach based on theshape grammar rules to investigate the evacuation-related parametric improvements in subsurfacesurface railway station design (eg minimizingtravelling distances)

The fire safety strategy in terms of performance-based fire risk design is an important aspirationalaspect of SGEvac requiring further research andliaison with experts to determine how such variedcriteria could be allied with layout planning This isparticularly important where deviation from standardsoccurs Thus further development of the expert

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

116 G SMITH AND B CERANIC

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 117

FIGURE 17 Station evacuation after 25 seconds

FIGURE 18 Station evacuation after 50 seconds

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

118 G SMITH AND B CERANIC

FIGURE 19 Station evacuation after 100 seconds

FIGURE 20 Station evacuation after 300 seconds

knowledge database would be required particularlyin the light of frequent changes to legislation

On a more practical note it could be beneficial todevelop the system on a stand-alone modular basislinked to each generated station component A list offire safety strategy objectives and functional logicstatements can then be specified for commonlyrecurring items (such as emergency lighting fire alarmdetectors sprinkler heads etc) and spaced accordingto the designerrsquos wishes or via shape rule formulation

Site constraint issues would also benefit fromfuture development as well as a move towardsdevelopment of the 3D shape grammar version Oncedevised other relevant 3D modelling systems such ascomputational fluid dynamics (CFD) smoke movementsoftware could be utilized in conjunction with SGEvac

CONCLUSIONSThe computer-aided modelling environment underdevelopment is intended to act as a layout planning aidfor those who practice in sub-surfacesurface railwaystation design The system under developmentcarefully considers London Underground StationPlanning Standards and Guidelines together withother relevant fire safety regulations that influenceoverall fire strategy design The proposed computer-aided design environment (SGEvac) is based on highlyrestrictive additive shape grammar rules driven by akernel of key passenger flow information to generatestation plan solutions

The major benefit to the design team is the ability toreview multiple layouts of accurately sized and correctlyplaced infrastructure at the reference or preliminarystage of the station layout planning design process Timeis hence saved regarding different option appraisals andquantitative testing using evacuation software shouldlead to a choice of more effective layouts in relation tomeans of escape travelling distances and improvementin the total evacuation times Modifications to layoutsshould be straightforward and there is scope toefficiently assess travelling distances and evacuationtimes for each of the considered alternatives comparedwith a lsquotraditionalrsquo approach

The proposed novel shape grammar approach ismost suited to new-build stations or largerefurbishment projects where more design flexibilityexists Sites with many site-specific constraints have

an array of predetermined parameters and will mostlikely benefit from the manual operation of SGEvac

Use in other rail systems due to recurringinfrastructure components and inherent functionallogic related to the laying out of the stationcomponents is also feasible provided that shapeconnections and shape rules are corrected to reflectthat countryrsquos legislative requirements

The potential for station layout improvementswith regard to emergency evacuation is an area thatmerits research especially in scenarios that concernthe safety of the general public and form a vital partof Londonrsquos transportation system Potential layoutimprovements could not only benefit evacuationsituations but also enhance the through-flow ofpeople during peak travelling periods

ACKNOWLEDGEMENTS The authors wish to express their gratitude to JacobsEngineering (UK) Ltd which has sponsored theresearch and to London Underground Ltd forproviding the design guidance that is reproduced inthis paper and for authorising the use of the stationplans London Underground has not reviewed thispaper and the content is solely the authorsrsquo work Mr Robert Hutchinson is also thanked for hiscomputer programming assistance

AUTHOR CONTACT DETAILSGGrraahhaamm SSmmiitthh (corresponding author) Senior Architectural

Technologist Jacobs (UK) Ltd Cardinal Square North Point

10 Nottingham Road Derby DE1 3QT UK Tel +44 (0)1332 234482

fax +44 (0)1332 344558 e-mail grahampsmithjacobscomBBoorriiss CCeerraanniicc Programme Leader for BScBA (Hons) Architectural

Programmes Room Number 105 Faculty of Arts Design and

Technology University of Derby Markeaton Street Derby DE22

3AW UK Tel +44 (0)1332 593136 e-mail bceranicderbyacuk

REFERENCES BS 7974 2001 Application of Fire Safety Engineering Principles to the

Design of Buildings London BSI

Cha MY and Gero JS 1999 Style Learning Inductive Generalisation of

Architectural Shape Patterns Paichai University and Sydney University

co-paper

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 119

Chase SC 1996 lsquoUsing logic to specify shapes and spatial relations in

design grammarsrsquo in Workshop notes Grammatical Design Fourth

International Conference on Artificial Intelligence in Design Stanford

University US June

Department of the Environment 1996 Design Principles of Fire Safety

London The Stationery Office Books

Flemming U 1987 lsquoMore than the sum of its parts the grammar of

Queen Anne housesrsquo in Environment and Planning B Planning and

Design 14 323ndash350

Fruin JJ 1971 Pedestrian Planning and Design New York Metropolitan

Association of Urban Designers and Environmental Planners

Galea ER 1999 lsquoA general approach to validating evacuation models with

an application to EXODUSrsquo in Journal of Fire Sciences 16 414ndash436

Galle P and Kovacs LB 1992 lsquoThe logic of worms a study in

architectural knowledge representationrsquo in Environment and Planning B

Planning and Design 19 5ndash31

Garling T Lindberg E and Mantyla T 1983 lsquoOrientation in buildings ndash

effects of familiarity visual access and orientation aidsrsquo in Journal of

Applied Psychology 68(1) 177ndash186

Grimsdale RL and Chang CW 1996 lsquoThe layout design language A

technique for generating layout plansrsquo in Computer Graphics Forum

15(2) 97ndash106

Heskestad AW 1999 lsquoPerformance in smoke of way guidance systemsrsquo

in Fire and Materials 23(6) 375ndash381

Knight T 1995 lsquoConstructive symmetryrsquo in Environment and Planning B

Planning and Design 22 419ndash450

Knight TW 1999 lsquoShape grammars six typesrsquo in Environment and

Planning B Planning and Design 26 15ndash31

Kuligowski ED and Peacock RD 2005 A Review of Building Evacuation

Models Technical Note 1471 National Institute of Standards and

Technology (NIST) Technology Administration US Department of

Commerce

Li A I-Kang 2003 lsquoThe Yingzao fashi in the information agersquo conference

paper for the Beaux-Arts Paul Philippe Cret and Twentieth Century

Architecture in China The University of Pennsylvania October

Liew H 2003 lsquoSGML ndash A shape grammar meta-languagersquo in Digital

Design 21st eCAADe Conference Proceedings Graz Austria September

639ndash648

Loomis B 2003 lsquoA note on generative design techniques SGGA a user

driven genetic algorithm for evolving non-deterministic shape grammarsrsquo

working paper Massachusetts Institute of Technology (MIT)

Magnusson SE Drysdale DD Fitzgerald RW Mowrer F

Quintiere JG Williamson RB and Zalosh RG 1995

lsquoA proposal for a model curriculum in fire safety engineeringrsquo

in Fire Safety Journal 25 9ndash10

Smith G Ceranic B and Fryer C 2003 lsquoDevelopment of CAD

environment for spatial layout optimisation in sub-surfacesurface rail

station designrsquo in Proceedings of International BIAT Conference The

British Museum London January 2003

SPSG 2007 London Underground Station Planning Standards and

Guidelines SPSG4 (1ndash371) London

Stiny G and Gips J 1971 lsquoShape grammars and the generative

specification of painting and sculpturersquo in Information Processing 71

1460ndash1465

Stiny G 1977 lsquoIce-ray a note on the generation of Chinese lattice

designsrsquo in Environment and Planning B Planning and Design 4 89ndash98

Stiny G 1980 lsquoKindergarten grammars designing with Froebelrsquos building

giftsrsquo in Environment and Planning B 3 409ndash462

Tapia M 1999 lsquoA visual implementation of a shape grammar systemrsquo in

Environment and Planning B Planning and Design 26(1) 59ndash73

Transport for London 2007 Tube Carries One Billion Passengers for First

Time wwwtflgovukcorporatemedianewscentre4770aspx

(accessed 24 March 2008)

Weller M 2007 London Underground private correspondence

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

120 G SMITH AND B CERANIC

station operates satisfactorily on a daily basis Thestation managerrsquos office plant and staff roomscomputer rooms train technician rooms etc all needaccommodating in the proposed shape grammarformalism In addition ancillary accommodationsuch as the cleanerrsquos store police room publicitystore retail etc also have to be catered for

To aid designing and sizing infrastructure levelsof service as shown in Figure 4 are used by LondonUnderground These are based on the work of Fruin(1971) where a detailed study of crowd movementwas undertaken Level A represents free circulationwith subsequent levels growing progressively morerestricted in terms of people movement from B to Eculminating with level F that represents a completebreakdown in flow with numerous stoppages

The approach proposed in this research definesthe generic station layout arrangements from thelsquohigh-levelrsquo spatial site organization to the subsequent

specified components and their logical spatialallocations as illustrated in Figure 5 The consequentshape grammar rules can then be deduced taking intoaccount the predicted passenger flow rates fire safetystrategy and relevant design regulations

FIRE SAFETY DESIGN STRATEGYBefore any design commences a clear identificationand appraisal of the key constraints and parametersat the site layout planning stage is required Forexample deep foundations mains servicesarchaeological discoveries and even the RiverThames will affect station design decisionsFurthermore structural appraisals geotechnicalsurveys Statutory Undertaker searches groundcontamination testing environmental impactassessments and land acquisition etc all need to becarried out as part of a feasibility study before stationlayout planning at reference level commences

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

102 G SMITH AND B CERANIC

Level of service A

Level of service D Level of service E Level of service F

Complete breakdown in trafficflow with many stoppages

Free circulationDescription (forqueuing areaswalkways andstairways)

Uni-directional flows andfree circulation Reverseand cross-flows withonly minor conflicts

Restricted circulation formost pedestriansSignificant difficulty forreverse and cross-flows

Restricted circulation for allpedestrians Intermittentstoppages and seriousdifficulties for reverse andcross-flows

Sligthly restricted circulationdue to difficulty inpassing others Reverse andcross-flows with difficulty

Level of service B Level of service C

FIGURE 4 London Underground levels of service (SPSG 2007)

Such constraints may be viewed in a positivecontext in terms of services already being providedhelping to justify the business case for a new stationDesign procurement strategies may include decisionssuch as locating toilet areas within x metres of anexisting sewer route or locating electrical plant roomswithin y metres of an existing electrical sub-station

In the real world parameters such as these constrict layout production and the choice ofinfrastructure component placement as opposed toleading to the production of many solutions that arelikely to be discarded when for example projectcosts are reviewed This level of regularity andrestriction typical in station design is very suitable

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 103

SITE

CUT AND COVER STATION TUNNELLED STATION

TICKET HALL

UTS GATELINE (gateline dimensionsdetermine minimumticket hall width)

PASSAGEWAYS ONE- or TWO-WAY

(entrances exits determinea minimum uniform width)

ESCALATOR

STAIRS

RAMPS

TOP RUN-OFF

TRAVELATOR

PAID CONCOURSE SIDE

BOTTOM RUN-OFF

PASSAGEWAYS ONE -WAY or TWO-WAY

NON-PASSENGERSPACE ELEMENTS ATTICKET HALL LEVEL (egStation Managerrsquos Office)

PLATFORM LEVEL

SPACE ELEMENTS ATPLATFORM LEVEL (egplant rooms toilets etc)

SIDE PLATFORM CENTRE PLATFORM STACKED PLATFORM

No OF RAILWAY LINES

MEZZANINEOptional

MEZZANINE

DEEP

SHALLOW

LIFTS

NON-PASSENGER

FIGURE 5 lsquoHigh-levelrsquo layout hierarchy for generic station composition

for the application of a prescriptive shape grammarapproach

In parallel with the shape grammar developmentfunctional logic statements are proposed analogousto the qualitative design review as described in BS7974 Application of Fire Safety EngineeringPrinciples to the Design of Buildings The intention isto develop SGEvac procedures in a manner that isrecognized by fire safety engineers and designers inpractice The Department for the Environment (DOE)(1996) states that lsquothe architectural layout can have asignificant influence on peoplersquos movementComplexity in circulation routes may not be overcomenecessarily by signage these may not have beennoticed or they may have been forgotten as a resultof stressrsquo Hence station spatial requirements can bereviewed alongside a number of functional logicstatements devised in conjunction with the layoutplanning

For example lsquothe probability of successful escapethrough smoke diminishes sharply with (an increasein the) number of decision elements included in theescape pathrsquo (Hesketad 1999)

A functional logic statement response to theabove would be to minimize such decision elementsthus enabling faster egress Other examples offunctional logic statements employed are

no corridor or staircase will reduce in width alongescape routes

length of the passages are to be minimized maximize level areas to facilitate faster travel

speeds for escapees maintain uniformity to aid way-finding and relative

orientation evacuation and normal circulation routes are to

be the same exits will be prominent easily identifiable and

straightforward to reach

Such statements are intuitive by their nature and arepart of the qualitative reviews in the traditional designprocess Due to time constraints only a limitednumber of possible alternatives in terms of the spatiallayout improvements are ever considered if any Incontrast SGEvac offers not only the opportunity toconsider a significant number of possible alternatives

but also the means of quantitative assessment via fireevacuation simulations

Garling et al (1983) contend that newcomers to anenvironment first learn a number of salient locationsthen the paths in between and finally organize theacquired knowledge in a spatial system Indeedavoiding symmetry and lack of differentiation fromthe userrsquos viewpoint is a good practice for stationdesign considering perception of relative orientationin the station This is where interaction between thedesigner and computer program becomes essentialand it is in this context that the research argues therole of SGEvac as a computer-aided design tool notas a substitute for the designer

A comparative methodology overview is given inTable 1 The processes of BS 7974 the procurementof a general logic model (Galle and Kovacs 1992) andthe shape grammar development stages arebenchmarked against SGEvac procedures Thisperhaps has most relevance to fire safetyengineering since it is with this objective that theproposed shape grammar system is to be used inpractice

Research studies such as one by Magnusson et al(1995) note that fire safety engineering has motivateddesigners and regulators to examine alternative codecompliance approaches based on a greater relianceupon quantitative analyses This has led toreassessment of prescriptive approaches andconsequent adjustments as required It is nowaccepted that existing prescriptive practices shouldbe evaluated by quantitative performance-basedmethods as a part of the overall fire safety strategyas encouraged in the Regulatory Reform (Fire Safety)Order (RRFSO) 2005 SGEvac follows this trendaugmenting the array of tools available forperformance-based fire safety engineering design

There is obviously an inevitable need for designerinput and control at all stages of the design processDue to the potential necessity for deviation fromstandards via fire safety engineering judgement theproposed shape grammar approach cannot alonemodel project-specific concessions

Still the movement away from prescription viaBS 7974 and the RRFSO to risk-based judgementplaces computer-aided design tools such as SGEvacin a position of interdependency with other design

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

104 G SMITH AND B CERANIC

criteria neither can be formulated in isolation Thisreinforces the view that computers will in future playan ever-increasing role in aiding and enabling thedesign teams but will never be able to replace them

SYSTEM DEVELOPMENTFigure 6 describes the modular approach proposedfor the system development offering the addedpotential of dual-purpose application in the designprocess It allows the designer to use the systemeither as a whole or to take advantage of itsindependent modules The system offers designerscomplete control over the shape grammarcomposition and layout modifications together withthe constant monitoring facilities and ability todiscard any solutions deemed to be lsquoinferiorrsquo

The proposed system is partitioned into threemain modules ndash lsquoexpert knowledgersquo lsquolayoutrsquo andlsquotestingrsquo modules The expert knowledge modulecaptures the expertise available to facilitate accurateshape grammar generations of layout design Thismodule provides the designer with the ability tointeractively reference Station Planning Standards

and Guidelines (and other relevant fire safetyregulations) allowing himher to make necessary site-specific decisions and to control the lsquohigh levelrsquo shapegrammar composition rules

The designer specifies passenger flow rates anddesignates the overall site boundary dimensions thatdefine the permissible station footprint The final exitpositions are also stipulated at this stage with theseoften being constrained by the existing builtenvironment

With this data supplied the shape grammargenerations commence in the layout module Thearea assignments adjacency requirements and layoutcomposition rules are defined in detail at this stageOnce the designer has made decisions with regard tothe layout composition the computer becomes thecentral tool for generating searching and logicallysorting through the permissible permutations ofpossible layouts The user has the option to auto-generate layouts (usually for sites without manyconstraints) or to manually control the layoutassembly (a decision normally associated with heavilyconstrained sites)

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 105

TABLE 1 Methodology comparison BS 7974 PROCESSES GALLE AND KOVACS SHAPE GRAMMAR SGEvac

LOGIC MODELLING DEVELOPMENT IMPLEMENTATION

Assess benefits Identification Shape grammar Assess constraints

and constraints system identification identify benefitssite

boundariesfinal exit

positions

Qualitative design Conceptualization Initial state Station component

review identification and

interactions

Quantitative analysis Formalization Rules and shape Legislative compliance

language passenger flow figures

area assignments

Satisfactory Implementation Shape grammar Alternative station

unsatisfactory executions layouts generation

Report and review Testing Design analysis Testingvalidation

iterationsevolution of layouts generation

End Experimentation Experimentation

iterationsimprovements

The testing module is used to evaluate potentiallysuitable layouts at reference level by exporting them toAutoCAD via standard drawing data exchange formatand analysing them in the context of project-specificsite constraints and requirements A level of drawingdetail is then added and the layout solution is exportedto Simulex a building evacuation simulation software

package Once the occupant numbers and persongroups are assigned and the escape and verticalcirculation routes defined the simulation software iscapable of the accurate measurement of all travellingdistances creation of distance maps and calculation ofevacuation times for different fire scenarios (egdifferent group affiliations persons exiting via different

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

106 G SMITH AND B CERANIC

EGY D ESIGNDesigner

Feasibility studiesProject brief requirements

Passenger flow data

DDE

DDE

Layout Production

Shape grammar rulesMultiple layout production

Layout Analysis

Layout interrogationsDesign team judgement

No

Yes

DXF

DXF

Yes

No

AutoCAD

Feed lsquoblock reference planrsquoresults into overall project

Improvedsolution identified

EXPERT KNOWLEDGE MODULE

Fire Strategy Design

LU SPSG4FSE legislation amp guidance

Infrastructure size calculationHigh level brief development

LAYOUT MODULE

Layout Composition

Area and adjacency planningLayout option appraisal

Layoutimproved

TESTING MODULE

AutoCAD

SGEvac export into CAD

Simulex

Fire evacuation scenarios

DesignerSGEvac

DDE ndash Database Data Exchange

DXF ndash Standard Drawing Data Exchange

FIGURE 6 Proposed SGEvac algorithm

routes and having delayed movement in response to afire alarm signal) The design team then assesses theresults against the qualitative fire safety design goalsset earlier to establish whether improvements toevacuation times have been made

DETAILED ANALYSIS OF STATIONCOMPONENTSIn conjunction with the site boundary final exitpositions and passenger flow specificationconsideration must also be given to the vertical sitedimension The shape grammar system assumes allfinal exits are at ground level with the ticket hall andplatform(s) being the other two elements that requirethe user to determine their depth relative to the finalexits In so doing SGEvac selects the correctinterchange components by an event-drivenmechanism that evaluates the height change andpassenger volume demand Thresholds for bothcriteria have been defined as shown in Table 2 andEquations 2ndash5

If H (height in mm) = 0 ndash 500 then I (infrastructure) = Ramp (2)

If H = 500 ndash 3000 and P (passengers) le 3000 then I = Stairway (3)

If H = 3000 ndash 5000 And P le 3000 then I = StairwayElse I = Escalator (4)

If H gt 5000 and P le 1000 then I = StairwayElse I = Escalator and Lift (5)

(where lsquoElsersquo denotes the mechanism for the choiceof the alternative infrastructure component if theoriginal condition is not satisfied)

The above constructs also condition the run-offdimensions associated with each of theinterchange components (between 4 m and 12 mas stipulated by SPSG) This being the case run-offs are used to size the ticket hall length Figure 7shows the paid side of the ticket hall with variabledistances shown between the escalator or staircaseand the gateline

The paid side of the ticket hall is always dictatedby run-off distance while the unpaid side is governedby two parameters the run-off and the concoursearea required for the heaviest 15-minute passengerflow If the maximum 15-minute passenger flowfigure multiplied by 1 m2 (say answer A) is larger inplan area than the product of the gateline widthmultiplied by the relevant run-off (say answer B) thenanswer A will be incorporated into the layout Thereverse is true if B is the greater number This ensuresthat the station is always sized appropriately

Figure 8 shows a ticket hall layout with thegateline dictating the width (denoted as Tick_H_wd)itself conditioned by the passenger flow figurescalculated as shown in Table 3 The table onlyevaluates entry flow requirements but SGEvac willrepeat the same process for exit gates to produce acorrectly sized gateline

In this case the maximum 15-minute passengerflow capacity size prevailed over the run-off to createthe unpaid side overall

The divide between the paid and unpaid side ofthe ticket hall is the sized according to Table 4

SGEvac shape rules mandate that entry gates arealways located adjacent to the ticket office queuingspace (designated as Q_TIW) to avoid conflictingpassenger entry and exit flows The LU standard is to

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 107

TABLE 2 Interchange infrastructure selectionINFRASTRUCTURE REASON FOR SELECTION

Ramp Vertical rise between 0ndash05 m preferred to steps due to disabled access issues generally considered

people will travel faster on ramps than up or down steps

Stairs Vertical rise between 05ndash5 m but greater height available for selection if station has a light passenger flow

Escalator Vertical rise is minimum of 3 m for heavy passenger flow stations but used for lightmedium flow stations

above 5 m

Lift Vertical rise above 5 m also required for accessibility of disabled persons Lifts always used with escalators

not as a substitute

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

108 G SMITH AND B CERANIC

Con_Area

Exit

UTS gates

6601900 2450

2400

EFW

Ticketoffice

POMcounter

4000

Tick_H_wd

1000

Run

-off

Entry

Q_space

Q_TIW

FIGURE 8 SGEvac ticket hall arrangement

Heavy flow escalator

Escalator

Staircase

Medium to Heavy flow

Medium to Heavy flow

Light flow escalator

Light flow

Heavy flow 12mMedium flow 8ndash12m

Light flow

Light flow 8m

Light flow 6m

Heavy flow

staircase to gateline

staircase to gateline 10m

Light flow

to gateline12mGateline

Gateline

EscalatorndashGateline

Total Run-off

Heavy flow 10mMedium flow 6ndash10m

StaircasendashGateline

Total Run-off

to gateline 8m

2m 4m 6m 8m 10m

2m 4m 6m

6m

8m 10m

12m

FIGURE 7 Passenger flow rate run-off requirements (SPSG 2007)

allow for 4 m (or a five-person queue) as shown inFigure 8 The overall gateline is defined by summingboth exit and entry gate numbers obtained from user-inputted passenger figures

The length of the ticket office interface with theticket hall is stipulated by the number of ticket issuingwindows (TIWs) plus excess fare windows (EFWs)TIWs always belong on the unpaid side of the tickethall and EFWs always on the paid side both forrevenue protection reasons This situation facilitates aprecise shape grammar labelling for the ticket officeto ticket hallqueue space connection

Passenger operated machine (POM) enclosures arenormally connected to the ticket hall ticket office andits own queue space (noted as Q_space in Figure 8)

To make shape rule specification manageable allstation areas are divided into groups arranged to suitthe most probable connection patterns in the stationTable 5 shows the group station managerrsquos officeexample with all other station components followingthe same concept

London Underground views its stations as largebasements (Weller 2007) and escape distances fromplatforms are based on agreed dispensations with theLondon Fire Brigade SGEvac respects thesedistances by placing restrictive labels on the platformso that they cannot be exceeded as demonstrated bythe example shown in Figure 9

SGEvac escape stair infrastructure may bepositioned up to 20 m from the end of platforms toavoid undesirable lsquodead endrsquo conditions The 45 mmaximum platform travel distance starts from thenearside junction of the two 33 m wide escapestaircases with the 45 m distance rule creating ashortfall in the platform centre of 34 m Generallylocating three interchange components (eg stairs

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 109

TABLE 3 Gateline width calculationMETHOD WORKED EXAMPLE

Obtain peak hourly flow rate through Say 6500 peoplehr

ticket hall

Calculate 5-minute peak passenger flow 650012 = 541

Allowance for ticket rejection 541 + 20 = 561

Allowance for growth in use demand 561 x 110 = 617

Assume 125 people use each gate over

the 5-minute peak passenger flow period 617125 = 493

Round up gates to nearest whole number 5 gates required

TABLE 4 London Underground ticket hall gateline parametersregulations (SPSG 2007)EQUIPMENT DIMENSIONS (W X D) POSITIONARRANGEMENT

Combined manualequipment gate 1750 x 270 mm Height ndash 1300 mm Adjacent to ticket office Can include

1400 mm clear opening luggage facility

Manual gate ndash passenger 1750 x 270 mm Height ndash 1300 mm Adjacent to ticket office With more than 6

gateluggage port 900 mm ndash clearance and walkways in a single gateline or significant

outward opening interchange with mainline stations airport or

tourist traffic provide separate luggage chutes

Manual gate ndash single leaf 1394 mm Height ndash 1300 mm Beside the dedicated excess gates

with luggage port 695 mm clearance

Gateline attendance report 990 x 1240 mm Within the controlled zone but not necessarily

part of the gateline Usually close

Station control unit (SCU) 432 x 430 mm Should be installed within a secure enclosure

(which could be a ticket office) and also with

a clear view of the gates Controls and monitors

the station underground ticketing system

equipment

Manual emergency exit gates 1900 x 125 mm Adjacent to ticket office Add these gates rather

than adding extra UTS gates for evacuation

purposes

or escalators) at these distances will result incompliance with the Underground standards but itmay not always be possible (or practical for aninterchange situation) to procure such a layout

Lengths of station components are controlled bya range of criteria For the platform it is the length ofthe train vehicle for interchange components it isthe vertical height in the terms of the depth to theplatform level for passages it is the designated finalexit positions and the existing station areas alreadygenerated

Vector direction labels are underpinned by thepermissible union of compatible spaces two points ofcontact will be maintained so as not to restrict thewidth of connecting passenger areas This undueconstriction of passenger flow at the head of apassage also satisfies realistic construction practiceNon-passenger and ancillary area vector directions areas per project requirements and the successive areaadditions managed by the SGEvac rule groupings

To create alternative layouts SGEvac uses alsquorandomizersquo rule In this shape grammar rule initialstate infrastructure is generated as per guidelinesbut it can randomly change the location ofcomponent arrangements within the predefined

boundaries so that a number of possible alternativelayouts are produced The designs can then beinterrogated and ranked by their travelling distancesand against the qualitative functional logicstatements

The SGEvac design tool developed in this researchis based on a highly restrictive additive grammar ruleset due to SPSGrsquos prescriptive guidance and thenature of the layout planning problem As Grimsdaleand Chang (1996) argue many published grammarslack high-level reasoning during the iteration processsince they are often concerned with the developmentof shape grammar theories rather than their practicalapplications By contrast SGEvac being a practicalapplication has a complete set of primitive shapesonly possible to arrange in a certain way It thusprovides viable reference design layout compositionsbased on SPSG with further interventions possible bythe designer

SYSTEM DEVELOPMENT TO DATEThe SGEvac prototype is compiled within the Clanguage environment allowing for modular object-oriented programming (suitable for shape grammar)Figure 10 shows the main interface where the stationname and relevant LU fare zone are entered (whichimpacts on sizing infrastructure) as are the ticket halland platform depths the overall site boundaries andfinal exitsrsquo positions The peak passenger flow figuresover the relevant time interval of normally 3 hours arealso specified SGEvac is then capable of accuratelycalculating all passenger area infrastructure based onthe SPSG guidelines as already discussed

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

110 G SMITH AND B CERANIC

TABLE 5 Station manager suite ndash shape rule connectionsELEMENT SHAPE RULE DESIGNATION CONNECTED COMPONENT GROUP ALLOCATION

Group station managerrsquos office SR_GSM_01 Ticket hall paid side GSM

SR_GSM_02 Ticket hall unpaid side

SR_GSM_03 GSM administration

SR_GSM_04 Duty station managerrsquos office

SR_GSM_05 Business managerrsquos office

SR_GSM_06 Administration

SR_GSM_07 Ticket office

SR_GSM_08 Police room

SR_GSM_09 Revenue protection examiner

SR_GSM_10 Interview room

140

Stair Stair

100 2020

4533

34

45

FIGURE 9 Platform escape distance SGEvac label restrictions

The history of auto-generated layouts is accessiblefrom this screen and the user is given the capabilityto delete and copy them (useful when minoralterations are required to produce an alternativelayout)

Shape labelling in the programme is multi-faceted which is necessary since many stationcomponents have various possible connections Anescalator for example may connect to a passagetravelator ramp escalator run-off stair run-off lift orthe ticket hall so all such connections need adding tothe component properties The SGEvac grouping ofcomponent interactions is accessed from menutoolbars that allow insertion of legislative compliantstation area connections as shown in Figure 11

Once the position and the depth of the ticket halland platform are specified the railway track is inserteddependent on the platform configuration The platformis the only possible connection to the track but thetrack can be positioned either side of platforms so thata side or centre (island) platform layout is possible

Non-passenger and ancillary areas are alsoselectable from the shape grammar library via property

windows for each station component eg the ticketoffice properties window as shown in Figure 12

Figure 13 shows examples of the layout modulestation components available in the SGEvacprogramme

Figure 14 shows the layout module interfaceInformation on the components (relative position inthe overall site layout and its absolute sizing) isdisplayed at the bottom of the screen once thecomponent is selected from the list In addition ameasurement bar (in metres) at the base of theinteractive design environment provides a sense ofscale for the designer

A checklist summarizes the station areas includedin the layout but a more detailed analysis is possiblevia a report function if required In addition to thisdetailed list a summary screen accessed from themain interface collates all specified components ineach generated layout This screen lists all passengernon-passenger and ancillary areas in their respectivecategories with infrastructure width data ofpassenger areas also summarized It also has anevacuation summary where the train frequency is

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 111

FIGURE 10 SGEvac main interface

specified and the train evacuation load (taken as peaklsquorush hourrsquo loading) is displayed This section of thesummary screen is useful for input into Simulexwhere fire evacuation test runs of the building layoutcan be implemented

The lsquoshow hierarchyrsquo function in Figure 15 allowsdesign team members to view the progression ofdesigns by superimposing connecting lines betweenstation elements in the order of their compositionThis also serves as a safeguard against deleting

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

112 G SMITH AND B CERANIC

FIGURE 12 SGEvac ticket office properties window

FIGURE 11 SGEvac properties window

components out of sequence that could lead to aninvalid solution

Once auto-generated layouts are complete withcorrectly sized and legitimately connected stationcomponents manual insertion of non-passenger areaswould then commence Upon completion thegenerated layout is transferred to AutoCAD via dxf fileformat as shown in Figure 16 Once in AutoCAD thelayout has a level of detail added by the designer suchas door openings passage connections fire escapeopenings etc so that passenger flows may beassessed It is then exported to Simulex for fireevacuation simulation analysis again via dxf file format

Simulex understands any voids as potentialcirculation routes treating the rest as solid andpermanent obstructions ie walls internal structuralelements permanent fixtures etc Since its solutionoperates in a two-dimensional environment for whatin effect is a three-dimensional problem the means

of vertical circulation (eg staircases) have to beestablished True lengths of staircases must bespecified based on the total length of all flights(sloping length) and the intermediate rests The fireexits need to be defined both in terms of theirlocation and width The links have to be establishedto define the point of the connection between thefloors and staircasesescalatorsramps The travellingdistance maps are then calculated type and numberof occupants specified building populated and theevacuation simulation performed

SYSTEM TESTING AND VALIDATIONKuligowski and Peacock (2005) define and recommendthe following ways of validating evacuation models

validation against code requirements validation against fire drills or other peoplersquos

movement experimentstrials

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 113

Gateline Vertical

Platform Right

Stairs Horizontal Stairs Vertical

Travelator Vertical

Ticket Hall

BMOAdmin Reception

Platform Top Ram Up Bottom Vertical

Lift Passage Horizontal

FIGURE 13 Examples of SGEvac station components

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

114 G SMITH AND B CERANIC

FIGURE 15 SGEvac hierarchy function

FIGURE 14 Layout module graphical user interface

validation against literature on past evacuationexperiments

validation against other models third-party validation

Furthermore Galea (1999) deconstructs the validationprocess into four main areas ndash component testingfunctional testing qualitative validation andquantitative validation This approach is more suitablefor models such as SGEvac than for those based oncompliance with code requirements and hence it isthe one adopted for the validation in this research

Component testing is regarded as testingcomputer code that drives the model checking forelementary errors Functional validation involveschecking that the model possesses the ability to

exhibit the range of capabilities required to performthe desired simulations Component testing andfunctional validation for SGEvac is carried outagainst code requirements insofar that allcalculations for infrastructure sizing are evaluatedmanually so that the method and answer are knownto be correct It uses the Simulex program to aid themodelling of people movement and to test thehypothesis of improvement to evacuation timesthus relying on Simulexrsquos own rigorous validationprocess Simulex has been used extensively inpractice as well as a validating model for otherevacuation softwares (Kuligowski and Peacock2005) SGEvac has gone through its initialcompletion and validation process with finalvalidation proposed for the end of the year

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 115

FIGURE 16 Auto-generated SGEvac station layout export file for fire

evacuation simulation in Simulex

FIRE EVACUATION SIMULATION RESULTSThe considered lsquotrain on fire in stationrsquo scenario takesinto account the service interval of 333 minutes(based on 18 trains per hour) the crush capacity of afour-car train together with the number of peoplewaiting on the platform after cancellation of oneservice (ie after 666 minutes) On the oppositeplatform the evacuation load is taken to be thenormal boardingalighting load assuming nocancellations to services (50 of all boarders andalighters)

Evacuation load predictions were based on TfL2006 projected passenger flows with an averageflow of 36 peopleminute for boarders and 123peopleminute for boarders and alighters combinedThe train crush capacity was averaged at 165peoplecar The simulation also models evacuationfrom the ticket hall level to assess the impact ofdispersing crowds at peak times in this part of thestation From these figures the total evacuation loadwas calculated as follows

Platform 1 Boarders (36 x 666 min) = 240Train crush load (165 x 4) = 660

Platform 2 Boarders and alighters

(123 x 05 x 333 min) = 205Ticket hall = 470Total station evacuation load = 1575 people

The simulation results have been captured as imagesat 25 50 100 and 300 second intervals respectivelyfor both platform and ticket hall levels as shown inFigures 17 to 20

The provided means of escape follow LU SPSG4requirements insofar that the maximum travellingdistance on a platform to an exit is no more than 45 m (ie exits must be no more than 90 m apart) andthat the dead-end travelling distance is no more than20 m to an exit Hence each platform is provided withthree exits The MIPsrsquo (mobility impaired persons)means of escape is provided from both ends ofplatforms via emergency evacuation lifts Totalevacuation time (ie all people reaching an exit onticket hallground level) was 5 minutes and

24 seconds with the platform and train load clearedwithin 3 minutes and 5 seconds This is within therequirements to evacuate the platform and train loadwithin 4 minutes with all people reaching a fire-separated route within 6 minutes (the ticket hall is notclassed as a fire-separated route)

The performed simulation was based on theminimum travelling distance maps resulting in eachperson exiting via the nearest fire exit route Theaverage 10 second lsquotime to start to moversquo wasassigned to all occupants in order to accommodatefor real world people behaviour commonly reportedby fire evacuation research studies The boarders onthe end of both platforms (including most peoplefrom end carriages of the train) evacuated viastaircasesMIP lifts directly to the ground level pointof escape The boarders situated around the middleof platforms together with the people from the twomiddle carriages used the central escalators to reachticket hall fire exits Additionally some people fromthe end carriages and some staffretail unitcustomers used this route as means of escape

In this lsquotrain on fire in stationrsquo scenario it wasassumed that none of the escapes routes areunavailable due to maintenance However furtherevacuation studies will investigate possible scenarioswhere one of the escalator routes may be unavailabledue to maintenance reasons

FUTURE RESEARCHIntended future work will concentrate on the furtherconditioning developing and modelling of shapegrammar design rules based on the outcomes of thefinal system validation and LU expertsrsquo feedbackFurther research is proposed on the development ofa novel layout optimization approach based on theshape grammar rules to investigate the evacuation-related parametric improvements in subsurfacesurface railway station design (eg minimizingtravelling distances)

The fire safety strategy in terms of performance-based fire risk design is an important aspirationalaspect of SGEvac requiring further research andliaison with experts to determine how such variedcriteria could be allied with layout planning This isparticularly important where deviation from standardsoccurs Thus further development of the expert

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

116 G SMITH AND B CERANIC

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 117

FIGURE 17 Station evacuation after 25 seconds

FIGURE 18 Station evacuation after 50 seconds

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

118 G SMITH AND B CERANIC

FIGURE 19 Station evacuation after 100 seconds

FIGURE 20 Station evacuation after 300 seconds

knowledge database would be required particularlyin the light of frequent changes to legislation

On a more practical note it could be beneficial todevelop the system on a stand-alone modular basislinked to each generated station component A list offire safety strategy objectives and functional logicstatements can then be specified for commonlyrecurring items (such as emergency lighting fire alarmdetectors sprinkler heads etc) and spaced accordingto the designerrsquos wishes or via shape rule formulation

Site constraint issues would also benefit fromfuture development as well as a move towardsdevelopment of the 3D shape grammar version Oncedevised other relevant 3D modelling systems such ascomputational fluid dynamics (CFD) smoke movementsoftware could be utilized in conjunction with SGEvac

CONCLUSIONSThe computer-aided modelling environment underdevelopment is intended to act as a layout planning aidfor those who practice in sub-surfacesurface railwaystation design The system under developmentcarefully considers London Underground StationPlanning Standards and Guidelines together withother relevant fire safety regulations that influenceoverall fire strategy design The proposed computer-aided design environment (SGEvac) is based on highlyrestrictive additive shape grammar rules driven by akernel of key passenger flow information to generatestation plan solutions

The major benefit to the design team is the ability toreview multiple layouts of accurately sized and correctlyplaced infrastructure at the reference or preliminarystage of the station layout planning design process Timeis hence saved regarding different option appraisals andquantitative testing using evacuation software shouldlead to a choice of more effective layouts in relation tomeans of escape travelling distances and improvementin the total evacuation times Modifications to layoutsshould be straightforward and there is scope toefficiently assess travelling distances and evacuationtimes for each of the considered alternatives comparedwith a lsquotraditionalrsquo approach

The proposed novel shape grammar approach ismost suited to new-build stations or largerefurbishment projects where more design flexibilityexists Sites with many site-specific constraints have

an array of predetermined parameters and will mostlikely benefit from the manual operation of SGEvac

Use in other rail systems due to recurringinfrastructure components and inherent functionallogic related to the laying out of the stationcomponents is also feasible provided that shapeconnections and shape rules are corrected to reflectthat countryrsquos legislative requirements

The potential for station layout improvementswith regard to emergency evacuation is an area thatmerits research especially in scenarios that concernthe safety of the general public and form a vital partof Londonrsquos transportation system Potential layoutimprovements could not only benefit evacuationsituations but also enhance the through-flow ofpeople during peak travelling periods

ACKNOWLEDGEMENTS The authors wish to express their gratitude to JacobsEngineering (UK) Ltd which has sponsored theresearch and to London Underground Ltd forproviding the design guidance that is reproduced inthis paper and for authorising the use of the stationplans London Underground has not reviewed thispaper and the content is solely the authorsrsquo work Mr Robert Hutchinson is also thanked for hiscomputer programming assistance

AUTHOR CONTACT DETAILSGGrraahhaamm SSmmiitthh (corresponding author) Senior Architectural

Technologist Jacobs (UK) Ltd Cardinal Square North Point

10 Nottingham Road Derby DE1 3QT UK Tel +44 (0)1332 234482

fax +44 (0)1332 344558 e-mail grahampsmithjacobscomBBoorriiss CCeerraanniicc Programme Leader for BScBA (Hons) Architectural

Programmes Room Number 105 Faculty of Arts Design and

Technology University of Derby Markeaton Street Derby DE22

3AW UK Tel +44 (0)1332 593136 e-mail bceranicderbyacuk

REFERENCES BS 7974 2001 Application of Fire Safety Engineering Principles to the

Design of Buildings London BSI

Cha MY and Gero JS 1999 Style Learning Inductive Generalisation of

Architectural Shape Patterns Paichai University and Sydney University

co-paper

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 119

Chase SC 1996 lsquoUsing logic to specify shapes and spatial relations in

design grammarsrsquo in Workshop notes Grammatical Design Fourth

International Conference on Artificial Intelligence in Design Stanford

University US June

Department of the Environment 1996 Design Principles of Fire Safety

London The Stationery Office Books

Flemming U 1987 lsquoMore than the sum of its parts the grammar of

Queen Anne housesrsquo in Environment and Planning B Planning and

Design 14 323ndash350

Fruin JJ 1971 Pedestrian Planning and Design New York Metropolitan

Association of Urban Designers and Environmental Planners

Galea ER 1999 lsquoA general approach to validating evacuation models with

an application to EXODUSrsquo in Journal of Fire Sciences 16 414ndash436

Galle P and Kovacs LB 1992 lsquoThe logic of worms a study in

architectural knowledge representationrsquo in Environment and Planning B

Planning and Design 19 5ndash31

Garling T Lindberg E and Mantyla T 1983 lsquoOrientation in buildings ndash

effects of familiarity visual access and orientation aidsrsquo in Journal of

Applied Psychology 68(1) 177ndash186

Grimsdale RL and Chang CW 1996 lsquoThe layout design language A

technique for generating layout plansrsquo in Computer Graphics Forum

15(2) 97ndash106

Heskestad AW 1999 lsquoPerformance in smoke of way guidance systemsrsquo

in Fire and Materials 23(6) 375ndash381

Knight T 1995 lsquoConstructive symmetryrsquo in Environment and Planning B

Planning and Design 22 419ndash450

Knight TW 1999 lsquoShape grammars six typesrsquo in Environment and

Planning B Planning and Design 26 15ndash31

Kuligowski ED and Peacock RD 2005 A Review of Building Evacuation

Models Technical Note 1471 National Institute of Standards and

Technology (NIST) Technology Administration US Department of

Commerce

Li A I-Kang 2003 lsquoThe Yingzao fashi in the information agersquo conference

paper for the Beaux-Arts Paul Philippe Cret and Twentieth Century

Architecture in China The University of Pennsylvania October

Liew H 2003 lsquoSGML ndash A shape grammar meta-languagersquo in Digital

Design 21st eCAADe Conference Proceedings Graz Austria September

639ndash648

Loomis B 2003 lsquoA note on generative design techniques SGGA a user

driven genetic algorithm for evolving non-deterministic shape grammarsrsquo

working paper Massachusetts Institute of Technology (MIT)

Magnusson SE Drysdale DD Fitzgerald RW Mowrer F

Quintiere JG Williamson RB and Zalosh RG 1995

lsquoA proposal for a model curriculum in fire safety engineeringrsquo

in Fire Safety Journal 25 9ndash10

Smith G Ceranic B and Fryer C 2003 lsquoDevelopment of CAD

environment for spatial layout optimisation in sub-surfacesurface rail

station designrsquo in Proceedings of International BIAT Conference The

British Museum London January 2003

SPSG 2007 London Underground Station Planning Standards and

Guidelines SPSG4 (1ndash371) London

Stiny G and Gips J 1971 lsquoShape grammars and the generative

specification of painting and sculpturersquo in Information Processing 71

1460ndash1465

Stiny G 1977 lsquoIce-ray a note on the generation of Chinese lattice

designsrsquo in Environment and Planning B Planning and Design 4 89ndash98

Stiny G 1980 lsquoKindergarten grammars designing with Froebelrsquos building

giftsrsquo in Environment and Planning B 3 409ndash462

Tapia M 1999 lsquoA visual implementation of a shape grammar systemrsquo in

Environment and Planning B Planning and Design 26(1) 59ndash73

Transport for London 2007 Tube Carries One Billion Passengers for First

Time wwwtflgovukcorporatemedianewscentre4770aspx

(accessed 24 March 2008)

Weller M 2007 London Underground private correspondence

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

120 G SMITH AND B CERANIC

Such constraints may be viewed in a positivecontext in terms of services already being providedhelping to justify the business case for a new stationDesign procurement strategies may include decisionssuch as locating toilet areas within x metres of anexisting sewer route or locating electrical plant roomswithin y metres of an existing electrical sub-station

In the real world parameters such as these constrict layout production and the choice ofinfrastructure component placement as opposed toleading to the production of many solutions that arelikely to be discarded when for example projectcosts are reviewed This level of regularity andrestriction typical in station design is very suitable

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 103

SITE

CUT AND COVER STATION TUNNELLED STATION

TICKET HALL

UTS GATELINE (gateline dimensionsdetermine minimumticket hall width)

PASSAGEWAYS ONE- or TWO-WAY

(entrances exits determinea minimum uniform width)

ESCALATOR

STAIRS

RAMPS

TOP RUN-OFF

TRAVELATOR

PAID CONCOURSE SIDE

BOTTOM RUN-OFF

PASSAGEWAYS ONE -WAY or TWO-WAY

NON-PASSENGERSPACE ELEMENTS ATTICKET HALL LEVEL (egStation Managerrsquos Office)

PLATFORM LEVEL

SPACE ELEMENTS ATPLATFORM LEVEL (egplant rooms toilets etc)

SIDE PLATFORM CENTRE PLATFORM STACKED PLATFORM

No OF RAILWAY LINES

MEZZANINEOptional

MEZZANINE

DEEP

SHALLOW

LIFTS

NON-PASSENGER

FIGURE 5 lsquoHigh-levelrsquo layout hierarchy for generic station composition

for the application of a prescriptive shape grammarapproach

In parallel with the shape grammar developmentfunctional logic statements are proposed analogousto the qualitative design review as described in BS7974 Application of Fire Safety EngineeringPrinciples to the Design of Buildings The intention isto develop SGEvac procedures in a manner that isrecognized by fire safety engineers and designers inpractice The Department for the Environment (DOE)(1996) states that lsquothe architectural layout can have asignificant influence on peoplersquos movementComplexity in circulation routes may not be overcomenecessarily by signage these may not have beennoticed or they may have been forgotten as a resultof stressrsquo Hence station spatial requirements can bereviewed alongside a number of functional logicstatements devised in conjunction with the layoutplanning

For example lsquothe probability of successful escapethrough smoke diminishes sharply with (an increasein the) number of decision elements included in theescape pathrsquo (Hesketad 1999)

A functional logic statement response to theabove would be to minimize such decision elementsthus enabling faster egress Other examples offunctional logic statements employed are

no corridor or staircase will reduce in width alongescape routes

length of the passages are to be minimized maximize level areas to facilitate faster travel

speeds for escapees maintain uniformity to aid way-finding and relative

orientation evacuation and normal circulation routes are to

be the same exits will be prominent easily identifiable and

straightforward to reach

Such statements are intuitive by their nature and arepart of the qualitative reviews in the traditional designprocess Due to time constraints only a limitednumber of possible alternatives in terms of the spatiallayout improvements are ever considered if any Incontrast SGEvac offers not only the opportunity toconsider a significant number of possible alternatives

but also the means of quantitative assessment via fireevacuation simulations

Garling et al (1983) contend that newcomers to anenvironment first learn a number of salient locationsthen the paths in between and finally organize theacquired knowledge in a spatial system Indeedavoiding symmetry and lack of differentiation fromthe userrsquos viewpoint is a good practice for stationdesign considering perception of relative orientationin the station This is where interaction between thedesigner and computer program becomes essentialand it is in this context that the research argues therole of SGEvac as a computer-aided design tool notas a substitute for the designer

A comparative methodology overview is given inTable 1 The processes of BS 7974 the procurementof a general logic model (Galle and Kovacs 1992) andthe shape grammar development stages arebenchmarked against SGEvac procedures Thisperhaps has most relevance to fire safetyengineering since it is with this objective that theproposed shape grammar system is to be used inpractice

Research studies such as one by Magnusson et al(1995) note that fire safety engineering has motivateddesigners and regulators to examine alternative codecompliance approaches based on a greater relianceupon quantitative analyses This has led toreassessment of prescriptive approaches andconsequent adjustments as required It is nowaccepted that existing prescriptive practices shouldbe evaluated by quantitative performance-basedmethods as a part of the overall fire safety strategyas encouraged in the Regulatory Reform (Fire Safety)Order (RRFSO) 2005 SGEvac follows this trendaugmenting the array of tools available forperformance-based fire safety engineering design

There is obviously an inevitable need for designerinput and control at all stages of the design processDue to the potential necessity for deviation fromstandards via fire safety engineering judgement theproposed shape grammar approach cannot alonemodel project-specific concessions

Still the movement away from prescription viaBS 7974 and the RRFSO to risk-based judgementplaces computer-aided design tools such as SGEvacin a position of interdependency with other design

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

104 G SMITH AND B CERANIC

criteria neither can be formulated in isolation Thisreinforces the view that computers will in future playan ever-increasing role in aiding and enabling thedesign teams but will never be able to replace them

SYSTEM DEVELOPMENTFigure 6 describes the modular approach proposedfor the system development offering the addedpotential of dual-purpose application in the designprocess It allows the designer to use the systemeither as a whole or to take advantage of itsindependent modules The system offers designerscomplete control over the shape grammarcomposition and layout modifications together withthe constant monitoring facilities and ability todiscard any solutions deemed to be lsquoinferiorrsquo

The proposed system is partitioned into threemain modules ndash lsquoexpert knowledgersquo lsquolayoutrsquo andlsquotestingrsquo modules The expert knowledge modulecaptures the expertise available to facilitate accurateshape grammar generations of layout design Thismodule provides the designer with the ability tointeractively reference Station Planning Standards

and Guidelines (and other relevant fire safetyregulations) allowing himher to make necessary site-specific decisions and to control the lsquohigh levelrsquo shapegrammar composition rules

The designer specifies passenger flow rates anddesignates the overall site boundary dimensions thatdefine the permissible station footprint The final exitpositions are also stipulated at this stage with theseoften being constrained by the existing builtenvironment

With this data supplied the shape grammargenerations commence in the layout module Thearea assignments adjacency requirements and layoutcomposition rules are defined in detail at this stageOnce the designer has made decisions with regard tothe layout composition the computer becomes thecentral tool for generating searching and logicallysorting through the permissible permutations ofpossible layouts The user has the option to auto-generate layouts (usually for sites without manyconstraints) or to manually control the layoutassembly (a decision normally associated with heavilyconstrained sites)

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 105

TABLE 1 Methodology comparison BS 7974 PROCESSES GALLE AND KOVACS SHAPE GRAMMAR SGEvac

LOGIC MODELLING DEVELOPMENT IMPLEMENTATION

Assess benefits Identification Shape grammar Assess constraints

and constraints system identification identify benefitssite

boundariesfinal exit

positions

Qualitative design Conceptualization Initial state Station component

review identification and

interactions

Quantitative analysis Formalization Rules and shape Legislative compliance

language passenger flow figures

area assignments

Satisfactory Implementation Shape grammar Alternative station

unsatisfactory executions layouts generation

Report and review Testing Design analysis Testingvalidation

iterationsevolution of layouts generation

End Experimentation Experimentation

iterationsimprovements

The testing module is used to evaluate potentiallysuitable layouts at reference level by exporting them toAutoCAD via standard drawing data exchange formatand analysing them in the context of project-specificsite constraints and requirements A level of drawingdetail is then added and the layout solution is exportedto Simulex a building evacuation simulation software

package Once the occupant numbers and persongroups are assigned and the escape and verticalcirculation routes defined the simulation software iscapable of the accurate measurement of all travellingdistances creation of distance maps and calculation ofevacuation times for different fire scenarios (egdifferent group affiliations persons exiting via different

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

106 G SMITH AND B CERANIC

EGY D ESIGNDesigner

Feasibility studiesProject brief requirements

Passenger flow data

DDE

DDE

Layout Production

Shape grammar rulesMultiple layout production

Layout Analysis

Layout interrogationsDesign team judgement

No

Yes

DXF

DXF

Yes

No

AutoCAD

Feed lsquoblock reference planrsquoresults into overall project

Improvedsolution identified

EXPERT KNOWLEDGE MODULE

Fire Strategy Design

LU SPSG4FSE legislation amp guidance

Infrastructure size calculationHigh level brief development

LAYOUT MODULE

Layout Composition

Area and adjacency planningLayout option appraisal

Layoutimproved

TESTING MODULE

AutoCAD

SGEvac export into CAD

Simulex

Fire evacuation scenarios

DesignerSGEvac

DDE ndash Database Data Exchange

DXF ndash Standard Drawing Data Exchange

FIGURE 6 Proposed SGEvac algorithm

routes and having delayed movement in response to afire alarm signal) The design team then assesses theresults against the qualitative fire safety design goalsset earlier to establish whether improvements toevacuation times have been made

DETAILED ANALYSIS OF STATIONCOMPONENTSIn conjunction with the site boundary final exitpositions and passenger flow specificationconsideration must also be given to the vertical sitedimension The shape grammar system assumes allfinal exits are at ground level with the ticket hall andplatform(s) being the other two elements that requirethe user to determine their depth relative to the finalexits In so doing SGEvac selects the correctinterchange components by an event-drivenmechanism that evaluates the height change andpassenger volume demand Thresholds for bothcriteria have been defined as shown in Table 2 andEquations 2ndash5

If H (height in mm) = 0 ndash 500 then I (infrastructure) = Ramp (2)

If H = 500 ndash 3000 and P (passengers) le 3000 then I = Stairway (3)

If H = 3000 ndash 5000 And P le 3000 then I = StairwayElse I = Escalator (4)

If H gt 5000 and P le 1000 then I = StairwayElse I = Escalator and Lift (5)

(where lsquoElsersquo denotes the mechanism for the choiceof the alternative infrastructure component if theoriginal condition is not satisfied)

The above constructs also condition the run-offdimensions associated with each of theinterchange components (between 4 m and 12 mas stipulated by SPSG) This being the case run-offs are used to size the ticket hall length Figure 7shows the paid side of the ticket hall with variabledistances shown between the escalator or staircaseand the gateline

The paid side of the ticket hall is always dictatedby run-off distance while the unpaid side is governedby two parameters the run-off and the concoursearea required for the heaviest 15-minute passengerflow If the maximum 15-minute passenger flowfigure multiplied by 1 m2 (say answer A) is larger inplan area than the product of the gateline widthmultiplied by the relevant run-off (say answer B) thenanswer A will be incorporated into the layout Thereverse is true if B is the greater number This ensuresthat the station is always sized appropriately

Figure 8 shows a ticket hall layout with thegateline dictating the width (denoted as Tick_H_wd)itself conditioned by the passenger flow figurescalculated as shown in Table 3 The table onlyevaluates entry flow requirements but SGEvac willrepeat the same process for exit gates to produce acorrectly sized gateline

In this case the maximum 15-minute passengerflow capacity size prevailed over the run-off to createthe unpaid side overall

The divide between the paid and unpaid side ofthe ticket hall is the sized according to Table 4

SGEvac shape rules mandate that entry gates arealways located adjacent to the ticket office queuingspace (designated as Q_TIW) to avoid conflictingpassenger entry and exit flows The LU standard is to

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 107

TABLE 2 Interchange infrastructure selectionINFRASTRUCTURE REASON FOR SELECTION

Ramp Vertical rise between 0ndash05 m preferred to steps due to disabled access issues generally considered

people will travel faster on ramps than up or down steps

Stairs Vertical rise between 05ndash5 m but greater height available for selection if station has a light passenger flow

Escalator Vertical rise is minimum of 3 m for heavy passenger flow stations but used for lightmedium flow stations

above 5 m

Lift Vertical rise above 5 m also required for accessibility of disabled persons Lifts always used with escalators

not as a substitute

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

108 G SMITH AND B CERANIC

Con_Area

Exit

UTS gates

6601900 2450

2400

EFW

Ticketoffice

POMcounter

4000

Tick_H_wd

1000

Run

-off

Entry

Q_space

Q_TIW

FIGURE 8 SGEvac ticket hall arrangement

Heavy flow escalator

Escalator

Staircase

Medium to Heavy flow

Medium to Heavy flow

Light flow escalator

Light flow

Heavy flow 12mMedium flow 8ndash12m

Light flow

Light flow 8m

Light flow 6m

Heavy flow

staircase to gateline

staircase to gateline 10m

Light flow

to gateline12mGateline

Gateline

EscalatorndashGateline

Total Run-off

Heavy flow 10mMedium flow 6ndash10m

StaircasendashGateline

Total Run-off

to gateline 8m

2m 4m 6m 8m 10m

2m 4m 6m

6m

8m 10m

12m

FIGURE 7 Passenger flow rate run-off requirements (SPSG 2007)

allow for 4 m (or a five-person queue) as shown inFigure 8 The overall gateline is defined by summingboth exit and entry gate numbers obtained from user-inputted passenger figures

The length of the ticket office interface with theticket hall is stipulated by the number of ticket issuingwindows (TIWs) plus excess fare windows (EFWs)TIWs always belong on the unpaid side of the tickethall and EFWs always on the paid side both forrevenue protection reasons This situation facilitates aprecise shape grammar labelling for the ticket officeto ticket hallqueue space connection

Passenger operated machine (POM) enclosures arenormally connected to the ticket hall ticket office andits own queue space (noted as Q_space in Figure 8)

To make shape rule specification manageable allstation areas are divided into groups arranged to suitthe most probable connection patterns in the stationTable 5 shows the group station managerrsquos officeexample with all other station components followingthe same concept

London Underground views its stations as largebasements (Weller 2007) and escape distances fromplatforms are based on agreed dispensations with theLondon Fire Brigade SGEvac respects thesedistances by placing restrictive labels on the platformso that they cannot be exceeded as demonstrated bythe example shown in Figure 9

SGEvac escape stair infrastructure may bepositioned up to 20 m from the end of platforms toavoid undesirable lsquodead endrsquo conditions The 45 mmaximum platform travel distance starts from thenearside junction of the two 33 m wide escapestaircases with the 45 m distance rule creating ashortfall in the platform centre of 34 m Generallylocating three interchange components (eg stairs

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 109

TABLE 3 Gateline width calculationMETHOD WORKED EXAMPLE

Obtain peak hourly flow rate through Say 6500 peoplehr

ticket hall

Calculate 5-minute peak passenger flow 650012 = 541

Allowance for ticket rejection 541 + 20 = 561

Allowance for growth in use demand 561 x 110 = 617

Assume 125 people use each gate over

the 5-minute peak passenger flow period 617125 = 493

Round up gates to nearest whole number 5 gates required

TABLE 4 London Underground ticket hall gateline parametersregulations (SPSG 2007)EQUIPMENT DIMENSIONS (W X D) POSITIONARRANGEMENT

Combined manualequipment gate 1750 x 270 mm Height ndash 1300 mm Adjacent to ticket office Can include

1400 mm clear opening luggage facility

Manual gate ndash passenger 1750 x 270 mm Height ndash 1300 mm Adjacent to ticket office With more than 6

gateluggage port 900 mm ndash clearance and walkways in a single gateline or significant

outward opening interchange with mainline stations airport or

tourist traffic provide separate luggage chutes

Manual gate ndash single leaf 1394 mm Height ndash 1300 mm Beside the dedicated excess gates

with luggage port 695 mm clearance

Gateline attendance report 990 x 1240 mm Within the controlled zone but not necessarily

part of the gateline Usually close

Station control unit (SCU) 432 x 430 mm Should be installed within a secure enclosure

(which could be a ticket office) and also with

a clear view of the gates Controls and monitors

the station underground ticketing system

equipment

Manual emergency exit gates 1900 x 125 mm Adjacent to ticket office Add these gates rather

than adding extra UTS gates for evacuation

purposes

or escalators) at these distances will result incompliance with the Underground standards but itmay not always be possible (or practical for aninterchange situation) to procure such a layout

Lengths of station components are controlled bya range of criteria For the platform it is the length ofthe train vehicle for interchange components it isthe vertical height in the terms of the depth to theplatform level for passages it is the designated finalexit positions and the existing station areas alreadygenerated

Vector direction labels are underpinned by thepermissible union of compatible spaces two points ofcontact will be maintained so as not to restrict thewidth of connecting passenger areas This undueconstriction of passenger flow at the head of apassage also satisfies realistic construction practiceNon-passenger and ancillary area vector directions areas per project requirements and the successive areaadditions managed by the SGEvac rule groupings

To create alternative layouts SGEvac uses alsquorandomizersquo rule In this shape grammar rule initialstate infrastructure is generated as per guidelinesbut it can randomly change the location ofcomponent arrangements within the predefined

boundaries so that a number of possible alternativelayouts are produced The designs can then beinterrogated and ranked by their travelling distancesand against the qualitative functional logicstatements

The SGEvac design tool developed in this researchis based on a highly restrictive additive grammar ruleset due to SPSGrsquos prescriptive guidance and thenature of the layout planning problem As Grimsdaleand Chang (1996) argue many published grammarslack high-level reasoning during the iteration processsince they are often concerned with the developmentof shape grammar theories rather than their practicalapplications By contrast SGEvac being a practicalapplication has a complete set of primitive shapesonly possible to arrange in a certain way It thusprovides viable reference design layout compositionsbased on SPSG with further interventions possible bythe designer

SYSTEM DEVELOPMENT TO DATEThe SGEvac prototype is compiled within the Clanguage environment allowing for modular object-oriented programming (suitable for shape grammar)Figure 10 shows the main interface where the stationname and relevant LU fare zone are entered (whichimpacts on sizing infrastructure) as are the ticket halland platform depths the overall site boundaries andfinal exitsrsquo positions The peak passenger flow figuresover the relevant time interval of normally 3 hours arealso specified SGEvac is then capable of accuratelycalculating all passenger area infrastructure based onthe SPSG guidelines as already discussed

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

110 G SMITH AND B CERANIC

TABLE 5 Station manager suite ndash shape rule connectionsELEMENT SHAPE RULE DESIGNATION CONNECTED COMPONENT GROUP ALLOCATION

Group station managerrsquos office SR_GSM_01 Ticket hall paid side GSM

SR_GSM_02 Ticket hall unpaid side

SR_GSM_03 GSM administration

SR_GSM_04 Duty station managerrsquos office

SR_GSM_05 Business managerrsquos office

SR_GSM_06 Administration

SR_GSM_07 Ticket office

SR_GSM_08 Police room

SR_GSM_09 Revenue protection examiner

SR_GSM_10 Interview room

140

Stair Stair

100 2020

4533

34

45

FIGURE 9 Platform escape distance SGEvac label restrictions

The history of auto-generated layouts is accessiblefrom this screen and the user is given the capabilityto delete and copy them (useful when minoralterations are required to produce an alternativelayout)

Shape labelling in the programme is multi-faceted which is necessary since many stationcomponents have various possible connections Anescalator for example may connect to a passagetravelator ramp escalator run-off stair run-off lift orthe ticket hall so all such connections need adding tothe component properties The SGEvac grouping ofcomponent interactions is accessed from menutoolbars that allow insertion of legislative compliantstation area connections as shown in Figure 11

Once the position and the depth of the ticket halland platform are specified the railway track is inserteddependent on the platform configuration The platformis the only possible connection to the track but thetrack can be positioned either side of platforms so thata side or centre (island) platform layout is possible

Non-passenger and ancillary areas are alsoselectable from the shape grammar library via property

windows for each station component eg the ticketoffice properties window as shown in Figure 12

Figure 13 shows examples of the layout modulestation components available in the SGEvacprogramme

Figure 14 shows the layout module interfaceInformation on the components (relative position inthe overall site layout and its absolute sizing) isdisplayed at the bottom of the screen once thecomponent is selected from the list In addition ameasurement bar (in metres) at the base of theinteractive design environment provides a sense ofscale for the designer

A checklist summarizes the station areas includedin the layout but a more detailed analysis is possiblevia a report function if required In addition to thisdetailed list a summary screen accessed from themain interface collates all specified components ineach generated layout This screen lists all passengernon-passenger and ancillary areas in their respectivecategories with infrastructure width data ofpassenger areas also summarized It also has anevacuation summary where the train frequency is

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 111

FIGURE 10 SGEvac main interface

specified and the train evacuation load (taken as peaklsquorush hourrsquo loading) is displayed This section of thesummary screen is useful for input into Simulexwhere fire evacuation test runs of the building layoutcan be implemented

The lsquoshow hierarchyrsquo function in Figure 15 allowsdesign team members to view the progression ofdesigns by superimposing connecting lines betweenstation elements in the order of their compositionThis also serves as a safeguard against deleting

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

112 G SMITH AND B CERANIC

FIGURE 12 SGEvac ticket office properties window

FIGURE 11 SGEvac properties window

components out of sequence that could lead to aninvalid solution

Once auto-generated layouts are complete withcorrectly sized and legitimately connected stationcomponents manual insertion of non-passenger areaswould then commence Upon completion thegenerated layout is transferred to AutoCAD via dxf fileformat as shown in Figure 16 Once in AutoCAD thelayout has a level of detail added by the designer suchas door openings passage connections fire escapeopenings etc so that passenger flows may beassessed It is then exported to Simulex for fireevacuation simulation analysis again via dxf file format

Simulex understands any voids as potentialcirculation routes treating the rest as solid andpermanent obstructions ie walls internal structuralelements permanent fixtures etc Since its solutionoperates in a two-dimensional environment for whatin effect is a three-dimensional problem the means

of vertical circulation (eg staircases) have to beestablished True lengths of staircases must bespecified based on the total length of all flights(sloping length) and the intermediate rests The fireexits need to be defined both in terms of theirlocation and width The links have to be establishedto define the point of the connection between thefloors and staircasesescalatorsramps The travellingdistance maps are then calculated type and numberof occupants specified building populated and theevacuation simulation performed

SYSTEM TESTING AND VALIDATIONKuligowski and Peacock (2005) define and recommendthe following ways of validating evacuation models

validation against code requirements validation against fire drills or other peoplersquos

movement experimentstrials

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 113

Gateline Vertical

Platform Right

Stairs Horizontal Stairs Vertical

Travelator Vertical

Ticket Hall

BMOAdmin Reception

Platform Top Ram Up Bottom Vertical

Lift Passage Horizontal

FIGURE 13 Examples of SGEvac station components

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

114 G SMITH AND B CERANIC

FIGURE 15 SGEvac hierarchy function

FIGURE 14 Layout module graphical user interface

validation against literature on past evacuationexperiments

validation against other models third-party validation

Furthermore Galea (1999) deconstructs the validationprocess into four main areas ndash component testingfunctional testing qualitative validation andquantitative validation This approach is more suitablefor models such as SGEvac than for those based oncompliance with code requirements and hence it isthe one adopted for the validation in this research

Component testing is regarded as testingcomputer code that drives the model checking forelementary errors Functional validation involveschecking that the model possesses the ability to

exhibit the range of capabilities required to performthe desired simulations Component testing andfunctional validation for SGEvac is carried outagainst code requirements insofar that allcalculations for infrastructure sizing are evaluatedmanually so that the method and answer are knownto be correct It uses the Simulex program to aid themodelling of people movement and to test thehypothesis of improvement to evacuation timesthus relying on Simulexrsquos own rigorous validationprocess Simulex has been used extensively inpractice as well as a validating model for otherevacuation softwares (Kuligowski and Peacock2005) SGEvac has gone through its initialcompletion and validation process with finalvalidation proposed for the end of the year

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 115

FIGURE 16 Auto-generated SGEvac station layout export file for fire

evacuation simulation in Simulex

FIRE EVACUATION SIMULATION RESULTSThe considered lsquotrain on fire in stationrsquo scenario takesinto account the service interval of 333 minutes(based on 18 trains per hour) the crush capacity of afour-car train together with the number of peoplewaiting on the platform after cancellation of oneservice (ie after 666 minutes) On the oppositeplatform the evacuation load is taken to be thenormal boardingalighting load assuming nocancellations to services (50 of all boarders andalighters)

Evacuation load predictions were based on TfL2006 projected passenger flows with an averageflow of 36 peopleminute for boarders and 123peopleminute for boarders and alighters combinedThe train crush capacity was averaged at 165peoplecar The simulation also models evacuationfrom the ticket hall level to assess the impact ofdispersing crowds at peak times in this part of thestation From these figures the total evacuation loadwas calculated as follows

Platform 1 Boarders (36 x 666 min) = 240Train crush load (165 x 4) = 660

Platform 2 Boarders and alighters

(123 x 05 x 333 min) = 205Ticket hall = 470Total station evacuation load = 1575 people

The simulation results have been captured as imagesat 25 50 100 and 300 second intervals respectivelyfor both platform and ticket hall levels as shown inFigures 17 to 20

The provided means of escape follow LU SPSG4requirements insofar that the maximum travellingdistance on a platform to an exit is no more than 45 m (ie exits must be no more than 90 m apart) andthat the dead-end travelling distance is no more than20 m to an exit Hence each platform is provided withthree exits The MIPsrsquo (mobility impaired persons)means of escape is provided from both ends ofplatforms via emergency evacuation lifts Totalevacuation time (ie all people reaching an exit onticket hallground level) was 5 minutes and

24 seconds with the platform and train load clearedwithin 3 minutes and 5 seconds This is within therequirements to evacuate the platform and train loadwithin 4 minutes with all people reaching a fire-separated route within 6 minutes (the ticket hall is notclassed as a fire-separated route)

The performed simulation was based on theminimum travelling distance maps resulting in eachperson exiting via the nearest fire exit route Theaverage 10 second lsquotime to start to moversquo wasassigned to all occupants in order to accommodatefor real world people behaviour commonly reportedby fire evacuation research studies The boarders onthe end of both platforms (including most peoplefrom end carriages of the train) evacuated viastaircasesMIP lifts directly to the ground level pointof escape The boarders situated around the middleof platforms together with the people from the twomiddle carriages used the central escalators to reachticket hall fire exits Additionally some people fromthe end carriages and some staffretail unitcustomers used this route as means of escape

In this lsquotrain on fire in stationrsquo scenario it wasassumed that none of the escapes routes areunavailable due to maintenance However furtherevacuation studies will investigate possible scenarioswhere one of the escalator routes may be unavailabledue to maintenance reasons

FUTURE RESEARCHIntended future work will concentrate on the furtherconditioning developing and modelling of shapegrammar design rules based on the outcomes of thefinal system validation and LU expertsrsquo feedbackFurther research is proposed on the development ofa novel layout optimization approach based on theshape grammar rules to investigate the evacuation-related parametric improvements in subsurfacesurface railway station design (eg minimizingtravelling distances)

The fire safety strategy in terms of performance-based fire risk design is an important aspirationalaspect of SGEvac requiring further research andliaison with experts to determine how such variedcriteria could be allied with layout planning This isparticularly important where deviation from standardsoccurs Thus further development of the expert

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

116 G SMITH AND B CERANIC

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 117

FIGURE 17 Station evacuation after 25 seconds

FIGURE 18 Station evacuation after 50 seconds

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

118 G SMITH AND B CERANIC

FIGURE 19 Station evacuation after 100 seconds

FIGURE 20 Station evacuation after 300 seconds

knowledge database would be required particularlyin the light of frequent changes to legislation

On a more practical note it could be beneficial todevelop the system on a stand-alone modular basislinked to each generated station component A list offire safety strategy objectives and functional logicstatements can then be specified for commonlyrecurring items (such as emergency lighting fire alarmdetectors sprinkler heads etc) and spaced accordingto the designerrsquos wishes or via shape rule formulation

Site constraint issues would also benefit fromfuture development as well as a move towardsdevelopment of the 3D shape grammar version Oncedevised other relevant 3D modelling systems such ascomputational fluid dynamics (CFD) smoke movementsoftware could be utilized in conjunction with SGEvac

CONCLUSIONSThe computer-aided modelling environment underdevelopment is intended to act as a layout planning aidfor those who practice in sub-surfacesurface railwaystation design The system under developmentcarefully considers London Underground StationPlanning Standards and Guidelines together withother relevant fire safety regulations that influenceoverall fire strategy design The proposed computer-aided design environment (SGEvac) is based on highlyrestrictive additive shape grammar rules driven by akernel of key passenger flow information to generatestation plan solutions

The major benefit to the design team is the ability toreview multiple layouts of accurately sized and correctlyplaced infrastructure at the reference or preliminarystage of the station layout planning design process Timeis hence saved regarding different option appraisals andquantitative testing using evacuation software shouldlead to a choice of more effective layouts in relation tomeans of escape travelling distances and improvementin the total evacuation times Modifications to layoutsshould be straightforward and there is scope toefficiently assess travelling distances and evacuationtimes for each of the considered alternatives comparedwith a lsquotraditionalrsquo approach

The proposed novel shape grammar approach ismost suited to new-build stations or largerefurbishment projects where more design flexibilityexists Sites with many site-specific constraints have

an array of predetermined parameters and will mostlikely benefit from the manual operation of SGEvac

Use in other rail systems due to recurringinfrastructure components and inherent functionallogic related to the laying out of the stationcomponents is also feasible provided that shapeconnections and shape rules are corrected to reflectthat countryrsquos legislative requirements

The potential for station layout improvementswith regard to emergency evacuation is an area thatmerits research especially in scenarios that concernthe safety of the general public and form a vital partof Londonrsquos transportation system Potential layoutimprovements could not only benefit evacuationsituations but also enhance the through-flow ofpeople during peak travelling periods

ACKNOWLEDGEMENTS The authors wish to express their gratitude to JacobsEngineering (UK) Ltd which has sponsored theresearch and to London Underground Ltd forproviding the design guidance that is reproduced inthis paper and for authorising the use of the stationplans London Underground has not reviewed thispaper and the content is solely the authorsrsquo work Mr Robert Hutchinson is also thanked for hiscomputer programming assistance

AUTHOR CONTACT DETAILSGGrraahhaamm SSmmiitthh (corresponding author) Senior Architectural

Technologist Jacobs (UK) Ltd Cardinal Square North Point

10 Nottingham Road Derby DE1 3QT UK Tel +44 (0)1332 234482

fax +44 (0)1332 344558 e-mail grahampsmithjacobscomBBoorriiss CCeerraanniicc Programme Leader for BScBA (Hons) Architectural

Programmes Room Number 105 Faculty of Arts Design and

Technology University of Derby Markeaton Street Derby DE22

3AW UK Tel +44 (0)1332 593136 e-mail bceranicderbyacuk

REFERENCES BS 7974 2001 Application of Fire Safety Engineering Principles to the

Design of Buildings London BSI

Cha MY and Gero JS 1999 Style Learning Inductive Generalisation of

Architectural Shape Patterns Paichai University and Sydney University

co-paper

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 119

Chase SC 1996 lsquoUsing logic to specify shapes and spatial relations in

design grammarsrsquo in Workshop notes Grammatical Design Fourth

International Conference on Artificial Intelligence in Design Stanford

University US June

Department of the Environment 1996 Design Principles of Fire Safety

London The Stationery Office Books

Flemming U 1987 lsquoMore than the sum of its parts the grammar of

Queen Anne housesrsquo in Environment and Planning B Planning and

Design 14 323ndash350

Fruin JJ 1971 Pedestrian Planning and Design New York Metropolitan

Association of Urban Designers and Environmental Planners

Galea ER 1999 lsquoA general approach to validating evacuation models with

an application to EXODUSrsquo in Journal of Fire Sciences 16 414ndash436

Galle P and Kovacs LB 1992 lsquoThe logic of worms a study in

architectural knowledge representationrsquo in Environment and Planning B

Planning and Design 19 5ndash31

Garling T Lindberg E and Mantyla T 1983 lsquoOrientation in buildings ndash

effects of familiarity visual access and orientation aidsrsquo in Journal of

Applied Psychology 68(1) 177ndash186

Grimsdale RL and Chang CW 1996 lsquoThe layout design language A

technique for generating layout plansrsquo in Computer Graphics Forum

15(2) 97ndash106

Heskestad AW 1999 lsquoPerformance in smoke of way guidance systemsrsquo

in Fire and Materials 23(6) 375ndash381

Knight T 1995 lsquoConstructive symmetryrsquo in Environment and Planning B

Planning and Design 22 419ndash450

Knight TW 1999 lsquoShape grammars six typesrsquo in Environment and

Planning B Planning and Design 26 15ndash31

Kuligowski ED and Peacock RD 2005 A Review of Building Evacuation

Models Technical Note 1471 National Institute of Standards and

Technology (NIST) Technology Administration US Department of

Commerce

Li A I-Kang 2003 lsquoThe Yingzao fashi in the information agersquo conference

paper for the Beaux-Arts Paul Philippe Cret and Twentieth Century

Architecture in China The University of Pennsylvania October

Liew H 2003 lsquoSGML ndash A shape grammar meta-languagersquo in Digital

Design 21st eCAADe Conference Proceedings Graz Austria September

639ndash648

Loomis B 2003 lsquoA note on generative design techniques SGGA a user

driven genetic algorithm for evolving non-deterministic shape grammarsrsquo

working paper Massachusetts Institute of Technology (MIT)

Magnusson SE Drysdale DD Fitzgerald RW Mowrer F

Quintiere JG Williamson RB and Zalosh RG 1995

lsquoA proposal for a model curriculum in fire safety engineeringrsquo

in Fire Safety Journal 25 9ndash10

Smith G Ceranic B and Fryer C 2003 lsquoDevelopment of CAD

environment for spatial layout optimisation in sub-surfacesurface rail

station designrsquo in Proceedings of International BIAT Conference The

British Museum London January 2003

SPSG 2007 London Underground Station Planning Standards and

Guidelines SPSG4 (1ndash371) London

Stiny G and Gips J 1971 lsquoShape grammars and the generative

specification of painting and sculpturersquo in Information Processing 71

1460ndash1465

Stiny G 1977 lsquoIce-ray a note on the generation of Chinese lattice

designsrsquo in Environment and Planning B Planning and Design 4 89ndash98

Stiny G 1980 lsquoKindergarten grammars designing with Froebelrsquos building

giftsrsquo in Environment and Planning B 3 409ndash462

Tapia M 1999 lsquoA visual implementation of a shape grammar systemrsquo in

Environment and Planning B Planning and Design 26(1) 59ndash73

Transport for London 2007 Tube Carries One Billion Passengers for First

Time wwwtflgovukcorporatemedianewscentre4770aspx

(accessed 24 March 2008)

Weller M 2007 London Underground private correspondence

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

120 G SMITH AND B CERANIC

for the application of a prescriptive shape grammarapproach

In parallel with the shape grammar developmentfunctional logic statements are proposed analogousto the qualitative design review as described in BS7974 Application of Fire Safety EngineeringPrinciples to the Design of Buildings The intention isto develop SGEvac procedures in a manner that isrecognized by fire safety engineers and designers inpractice The Department for the Environment (DOE)(1996) states that lsquothe architectural layout can have asignificant influence on peoplersquos movementComplexity in circulation routes may not be overcomenecessarily by signage these may not have beennoticed or they may have been forgotten as a resultof stressrsquo Hence station spatial requirements can bereviewed alongside a number of functional logicstatements devised in conjunction with the layoutplanning

For example lsquothe probability of successful escapethrough smoke diminishes sharply with (an increasein the) number of decision elements included in theescape pathrsquo (Hesketad 1999)

A functional logic statement response to theabove would be to minimize such decision elementsthus enabling faster egress Other examples offunctional logic statements employed are

no corridor or staircase will reduce in width alongescape routes

length of the passages are to be minimized maximize level areas to facilitate faster travel

speeds for escapees maintain uniformity to aid way-finding and relative

orientation evacuation and normal circulation routes are to

be the same exits will be prominent easily identifiable and

straightforward to reach

Such statements are intuitive by their nature and arepart of the qualitative reviews in the traditional designprocess Due to time constraints only a limitednumber of possible alternatives in terms of the spatiallayout improvements are ever considered if any Incontrast SGEvac offers not only the opportunity toconsider a significant number of possible alternatives

but also the means of quantitative assessment via fireevacuation simulations

Garling et al (1983) contend that newcomers to anenvironment first learn a number of salient locationsthen the paths in between and finally organize theacquired knowledge in a spatial system Indeedavoiding symmetry and lack of differentiation fromthe userrsquos viewpoint is a good practice for stationdesign considering perception of relative orientationin the station This is where interaction between thedesigner and computer program becomes essentialand it is in this context that the research argues therole of SGEvac as a computer-aided design tool notas a substitute for the designer

A comparative methodology overview is given inTable 1 The processes of BS 7974 the procurementof a general logic model (Galle and Kovacs 1992) andthe shape grammar development stages arebenchmarked against SGEvac procedures Thisperhaps has most relevance to fire safetyengineering since it is with this objective that theproposed shape grammar system is to be used inpractice

Research studies such as one by Magnusson et al(1995) note that fire safety engineering has motivateddesigners and regulators to examine alternative codecompliance approaches based on a greater relianceupon quantitative analyses This has led toreassessment of prescriptive approaches andconsequent adjustments as required It is nowaccepted that existing prescriptive practices shouldbe evaluated by quantitative performance-basedmethods as a part of the overall fire safety strategyas encouraged in the Regulatory Reform (Fire Safety)Order (RRFSO) 2005 SGEvac follows this trendaugmenting the array of tools available forperformance-based fire safety engineering design

There is obviously an inevitable need for designerinput and control at all stages of the design processDue to the potential necessity for deviation fromstandards via fire safety engineering judgement theproposed shape grammar approach cannot alonemodel project-specific concessions

Still the movement away from prescription viaBS 7974 and the RRFSO to risk-based judgementplaces computer-aided design tools such as SGEvacin a position of interdependency with other design

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

104 G SMITH AND B CERANIC

criteria neither can be formulated in isolation Thisreinforces the view that computers will in future playan ever-increasing role in aiding and enabling thedesign teams but will never be able to replace them

SYSTEM DEVELOPMENTFigure 6 describes the modular approach proposedfor the system development offering the addedpotential of dual-purpose application in the designprocess It allows the designer to use the systemeither as a whole or to take advantage of itsindependent modules The system offers designerscomplete control over the shape grammarcomposition and layout modifications together withthe constant monitoring facilities and ability todiscard any solutions deemed to be lsquoinferiorrsquo

The proposed system is partitioned into threemain modules ndash lsquoexpert knowledgersquo lsquolayoutrsquo andlsquotestingrsquo modules The expert knowledge modulecaptures the expertise available to facilitate accurateshape grammar generations of layout design Thismodule provides the designer with the ability tointeractively reference Station Planning Standards

and Guidelines (and other relevant fire safetyregulations) allowing himher to make necessary site-specific decisions and to control the lsquohigh levelrsquo shapegrammar composition rules

The designer specifies passenger flow rates anddesignates the overall site boundary dimensions thatdefine the permissible station footprint The final exitpositions are also stipulated at this stage with theseoften being constrained by the existing builtenvironment

With this data supplied the shape grammargenerations commence in the layout module Thearea assignments adjacency requirements and layoutcomposition rules are defined in detail at this stageOnce the designer has made decisions with regard tothe layout composition the computer becomes thecentral tool for generating searching and logicallysorting through the permissible permutations ofpossible layouts The user has the option to auto-generate layouts (usually for sites without manyconstraints) or to manually control the layoutassembly (a decision normally associated with heavilyconstrained sites)

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 105

TABLE 1 Methodology comparison BS 7974 PROCESSES GALLE AND KOVACS SHAPE GRAMMAR SGEvac

LOGIC MODELLING DEVELOPMENT IMPLEMENTATION

Assess benefits Identification Shape grammar Assess constraints

and constraints system identification identify benefitssite

boundariesfinal exit

positions

Qualitative design Conceptualization Initial state Station component

review identification and

interactions

Quantitative analysis Formalization Rules and shape Legislative compliance

language passenger flow figures

area assignments

Satisfactory Implementation Shape grammar Alternative station

unsatisfactory executions layouts generation

Report and review Testing Design analysis Testingvalidation

iterationsevolution of layouts generation

End Experimentation Experimentation

iterationsimprovements

The testing module is used to evaluate potentiallysuitable layouts at reference level by exporting them toAutoCAD via standard drawing data exchange formatand analysing them in the context of project-specificsite constraints and requirements A level of drawingdetail is then added and the layout solution is exportedto Simulex a building evacuation simulation software

package Once the occupant numbers and persongroups are assigned and the escape and verticalcirculation routes defined the simulation software iscapable of the accurate measurement of all travellingdistances creation of distance maps and calculation ofevacuation times for different fire scenarios (egdifferent group affiliations persons exiting via different

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

106 G SMITH AND B CERANIC

EGY D ESIGNDesigner

Feasibility studiesProject brief requirements

Passenger flow data

DDE

DDE

Layout Production

Shape grammar rulesMultiple layout production

Layout Analysis

Layout interrogationsDesign team judgement

No

Yes

DXF

DXF

Yes

No

AutoCAD

Feed lsquoblock reference planrsquoresults into overall project

Improvedsolution identified

EXPERT KNOWLEDGE MODULE

Fire Strategy Design

LU SPSG4FSE legislation amp guidance

Infrastructure size calculationHigh level brief development

LAYOUT MODULE

Layout Composition

Area and adjacency planningLayout option appraisal

Layoutimproved

TESTING MODULE

AutoCAD

SGEvac export into CAD

Simulex

Fire evacuation scenarios

DesignerSGEvac

DDE ndash Database Data Exchange

DXF ndash Standard Drawing Data Exchange

FIGURE 6 Proposed SGEvac algorithm

routes and having delayed movement in response to afire alarm signal) The design team then assesses theresults against the qualitative fire safety design goalsset earlier to establish whether improvements toevacuation times have been made

DETAILED ANALYSIS OF STATIONCOMPONENTSIn conjunction with the site boundary final exitpositions and passenger flow specificationconsideration must also be given to the vertical sitedimension The shape grammar system assumes allfinal exits are at ground level with the ticket hall andplatform(s) being the other two elements that requirethe user to determine their depth relative to the finalexits In so doing SGEvac selects the correctinterchange components by an event-drivenmechanism that evaluates the height change andpassenger volume demand Thresholds for bothcriteria have been defined as shown in Table 2 andEquations 2ndash5

If H (height in mm) = 0 ndash 500 then I (infrastructure) = Ramp (2)

If H = 500 ndash 3000 and P (passengers) le 3000 then I = Stairway (3)

If H = 3000 ndash 5000 And P le 3000 then I = StairwayElse I = Escalator (4)

If H gt 5000 and P le 1000 then I = StairwayElse I = Escalator and Lift (5)

(where lsquoElsersquo denotes the mechanism for the choiceof the alternative infrastructure component if theoriginal condition is not satisfied)

The above constructs also condition the run-offdimensions associated with each of theinterchange components (between 4 m and 12 mas stipulated by SPSG) This being the case run-offs are used to size the ticket hall length Figure 7shows the paid side of the ticket hall with variabledistances shown between the escalator or staircaseand the gateline

The paid side of the ticket hall is always dictatedby run-off distance while the unpaid side is governedby two parameters the run-off and the concoursearea required for the heaviest 15-minute passengerflow If the maximum 15-minute passenger flowfigure multiplied by 1 m2 (say answer A) is larger inplan area than the product of the gateline widthmultiplied by the relevant run-off (say answer B) thenanswer A will be incorporated into the layout Thereverse is true if B is the greater number This ensuresthat the station is always sized appropriately

Figure 8 shows a ticket hall layout with thegateline dictating the width (denoted as Tick_H_wd)itself conditioned by the passenger flow figurescalculated as shown in Table 3 The table onlyevaluates entry flow requirements but SGEvac willrepeat the same process for exit gates to produce acorrectly sized gateline

In this case the maximum 15-minute passengerflow capacity size prevailed over the run-off to createthe unpaid side overall

The divide between the paid and unpaid side ofthe ticket hall is the sized according to Table 4

SGEvac shape rules mandate that entry gates arealways located adjacent to the ticket office queuingspace (designated as Q_TIW) to avoid conflictingpassenger entry and exit flows The LU standard is to

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 107

TABLE 2 Interchange infrastructure selectionINFRASTRUCTURE REASON FOR SELECTION

Ramp Vertical rise between 0ndash05 m preferred to steps due to disabled access issues generally considered

people will travel faster on ramps than up or down steps

Stairs Vertical rise between 05ndash5 m but greater height available for selection if station has a light passenger flow

Escalator Vertical rise is minimum of 3 m for heavy passenger flow stations but used for lightmedium flow stations

above 5 m

Lift Vertical rise above 5 m also required for accessibility of disabled persons Lifts always used with escalators

not as a substitute

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

108 G SMITH AND B CERANIC

Con_Area

Exit

UTS gates

6601900 2450

2400

EFW

Ticketoffice

POMcounter

4000

Tick_H_wd

1000

Run

-off

Entry

Q_space

Q_TIW

FIGURE 8 SGEvac ticket hall arrangement

Heavy flow escalator

Escalator

Staircase

Medium to Heavy flow

Medium to Heavy flow

Light flow escalator

Light flow

Heavy flow 12mMedium flow 8ndash12m

Light flow

Light flow 8m

Light flow 6m

Heavy flow

staircase to gateline

staircase to gateline 10m

Light flow

to gateline12mGateline

Gateline

EscalatorndashGateline

Total Run-off

Heavy flow 10mMedium flow 6ndash10m

StaircasendashGateline

Total Run-off

to gateline 8m

2m 4m 6m 8m 10m

2m 4m 6m

6m

8m 10m

12m

FIGURE 7 Passenger flow rate run-off requirements (SPSG 2007)

allow for 4 m (or a five-person queue) as shown inFigure 8 The overall gateline is defined by summingboth exit and entry gate numbers obtained from user-inputted passenger figures

The length of the ticket office interface with theticket hall is stipulated by the number of ticket issuingwindows (TIWs) plus excess fare windows (EFWs)TIWs always belong on the unpaid side of the tickethall and EFWs always on the paid side both forrevenue protection reasons This situation facilitates aprecise shape grammar labelling for the ticket officeto ticket hallqueue space connection

Passenger operated machine (POM) enclosures arenormally connected to the ticket hall ticket office andits own queue space (noted as Q_space in Figure 8)

To make shape rule specification manageable allstation areas are divided into groups arranged to suitthe most probable connection patterns in the stationTable 5 shows the group station managerrsquos officeexample with all other station components followingthe same concept

London Underground views its stations as largebasements (Weller 2007) and escape distances fromplatforms are based on agreed dispensations with theLondon Fire Brigade SGEvac respects thesedistances by placing restrictive labels on the platformso that they cannot be exceeded as demonstrated bythe example shown in Figure 9

SGEvac escape stair infrastructure may bepositioned up to 20 m from the end of platforms toavoid undesirable lsquodead endrsquo conditions The 45 mmaximum platform travel distance starts from thenearside junction of the two 33 m wide escapestaircases with the 45 m distance rule creating ashortfall in the platform centre of 34 m Generallylocating three interchange components (eg stairs

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 109

TABLE 3 Gateline width calculationMETHOD WORKED EXAMPLE

Obtain peak hourly flow rate through Say 6500 peoplehr

ticket hall

Calculate 5-minute peak passenger flow 650012 = 541

Allowance for ticket rejection 541 + 20 = 561

Allowance for growth in use demand 561 x 110 = 617

Assume 125 people use each gate over

the 5-minute peak passenger flow period 617125 = 493

Round up gates to nearest whole number 5 gates required

TABLE 4 London Underground ticket hall gateline parametersregulations (SPSG 2007)EQUIPMENT DIMENSIONS (W X D) POSITIONARRANGEMENT

Combined manualequipment gate 1750 x 270 mm Height ndash 1300 mm Adjacent to ticket office Can include

1400 mm clear opening luggage facility

Manual gate ndash passenger 1750 x 270 mm Height ndash 1300 mm Adjacent to ticket office With more than 6

gateluggage port 900 mm ndash clearance and walkways in a single gateline or significant

outward opening interchange with mainline stations airport or

tourist traffic provide separate luggage chutes

Manual gate ndash single leaf 1394 mm Height ndash 1300 mm Beside the dedicated excess gates

with luggage port 695 mm clearance

Gateline attendance report 990 x 1240 mm Within the controlled zone but not necessarily

part of the gateline Usually close

Station control unit (SCU) 432 x 430 mm Should be installed within a secure enclosure

(which could be a ticket office) and also with

a clear view of the gates Controls and monitors

the station underground ticketing system

equipment

Manual emergency exit gates 1900 x 125 mm Adjacent to ticket office Add these gates rather

than adding extra UTS gates for evacuation

purposes

or escalators) at these distances will result incompliance with the Underground standards but itmay not always be possible (or practical for aninterchange situation) to procure such a layout

Lengths of station components are controlled bya range of criteria For the platform it is the length ofthe train vehicle for interchange components it isthe vertical height in the terms of the depth to theplatform level for passages it is the designated finalexit positions and the existing station areas alreadygenerated

Vector direction labels are underpinned by thepermissible union of compatible spaces two points ofcontact will be maintained so as not to restrict thewidth of connecting passenger areas This undueconstriction of passenger flow at the head of apassage also satisfies realistic construction practiceNon-passenger and ancillary area vector directions areas per project requirements and the successive areaadditions managed by the SGEvac rule groupings

To create alternative layouts SGEvac uses alsquorandomizersquo rule In this shape grammar rule initialstate infrastructure is generated as per guidelinesbut it can randomly change the location ofcomponent arrangements within the predefined

boundaries so that a number of possible alternativelayouts are produced The designs can then beinterrogated and ranked by their travelling distancesand against the qualitative functional logicstatements

The SGEvac design tool developed in this researchis based on a highly restrictive additive grammar ruleset due to SPSGrsquos prescriptive guidance and thenature of the layout planning problem As Grimsdaleand Chang (1996) argue many published grammarslack high-level reasoning during the iteration processsince they are often concerned with the developmentof shape grammar theories rather than their practicalapplications By contrast SGEvac being a practicalapplication has a complete set of primitive shapesonly possible to arrange in a certain way It thusprovides viable reference design layout compositionsbased on SPSG with further interventions possible bythe designer

SYSTEM DEVELOPMENT TO DATEThe SGEvac prototype is compiled within the Clanguage environment allowing for modular object-oriented programming (suitable for shape grammar)Figure 10 shows the main interface where the stationname and relevant LU fare zone are entered (whichimpacts on sizing infrastructure) as are the ticket halland platform depths the overall site boundaries andfinal exitsrsquo positions The peak passenger flow figuresover the relevant time interval of normally 3 hours arealso specified SGEvac is then capable of accuratelycalculating all passenger area infrastructure based onthe SPSG guidelines as already discussed

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

110 G SMITH AND B CERANIC

TABLE 5 Station manager suite ndash shape rule connectionsELEMENT SHAPE RULE DESIGNATION CONNECTED COMPONENT GROUP ALLOCATION

Group station managerrsquos office SR_GSM_01 Ticket hall paid side GSM

SR_GSM_02 Ticket hall unpaid side

SR_GSM_03 GSM administration

SR_GSM_04 Duty station managerrsquos office

SR_GSM_05 Business managerrsquos office

SR_GSM_06 Administration

SR_GSM_07 Ticket office

SR_GSM_08 Police room

SR_GSM_09 Revenue protection examiner

SR_GSM_10 Interview room

140

Stair Stair

100 2020

4533

34

45

FIGURE 9 Platform escape distance SGEvac label restrictions

The history of auto-generated layouts is accessiblefrom this screen and the user is given the capabilityto delete and copy them (useful when minoralterations are required to produce an alternativelayout)

Shape labelling in the programme is multi-faceted which is necessary since many stationcomponents have various possible connections Anescalator for example may connect to a passagetravelator ramp escalator run-off stair run-off lift orthe ticket hall so all such connections need adding tothe component properties The SGEvac grouping ofcomponent interactions is accessed from menutoolbars that allow insertion of legislative compliantstation area connections as shown in Figure 11

Once the position and the depth of the ticket halland platform are specified the railway track is inserteddependent on the platform configuration The platformis the only possible connection to the track but thetrack can be positioned either side of platforms so thata side or centre (island) platform layout is possible

Non-passenger and ancillary areas are alsoselectable from the shape grammar library via property

windows for each station component eg the ticketoffice properties window as shown in Figure 12

Figure 13 shows examples of the layout modulestation components available in the SGEvacprogramme

Figure 14 shows the layout module interfaceInformation on the components (relative position inthe overall site layout and its absolute sizing) isdisplayed at the bottom of the screen once thecomponent is selected from the list In addition ameasurement bar (in metres) at the base of theinteractive design environment provides a sense ofscale for the designer

A checklist summarizes the station areas includedin the layout but a more detailed analysis is possiblevia a report function if required In addition to thisdetailed list a summary screen accessed from themain interface collates all specified components ineach generated layout This screen lists all passengernon-passenger and ancillary areas in their respectivecategories with infrastructure width data ofpassenger areas also summarized It also has anevacuation summary where the train frequency is

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 111

FIGURE 10 SGEvac main interface

specified and the train evacuation load (taken as peaklsquorush hourrsquo loading) is displayed This section of thesummary screen is useful for input into Simulexwhere fire evacuation test runs of the building layoutcan be implemented

The lsquoshow hierarchyrsquo function in Figure 15 allowsdesign team members to view the progression ofdesigns by superimposing connecting lines betweenstation elements in the order of their compositionThis also serves as a safeguard against deleting

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

112 G SMITH AND B CERANIC

FIGURE 12 SGEvac ticket office properties window

FIGURE 11 SGEvac properties window

components out of sequence that could lead to aninvalid solution

Once auto-generated layouts are complete withcorrectly sized and legitimately connected stationcomponents manual insertion of non-passenger areaswould then commence Upon completion thegenerated layout is transferred to AutoCAD via dxf fileformat as shown in Figure 16 Once in AutoCAD thelayout has a level of detail added by the designer suchas door openings passage connections fire escapeopenings etc so that passenger flows may beassessed It is then exported to Simulex for fireevacuation simulation analysis again via dxf file format

Simulex understands any voids as potentialcirculation routes treating the rest as solid andpermanent obstructions ie walls internal structuralelements permanent fixtures etc Since its solutionoperates in a two-dimensional environment for whatin effect is a three-dimensional problem the means

of vertical circulation (eg staircases) have to beestablished True lengths of staircases must bespecified based on the total length of all flights(sloping length) and the intermediate rests The fireexits need to be defined both in terms of theirlocation and width The links have to be establishedto define the point of the connection between thefloors and staircasesescalatorsramps The travellingdistance maps are then calculated type and numberof occupants specified building populated and theevacuation simulation performed

SYSTEM TESTING AND VALIDATIONKuligowski and Peacock (2005) define and recommendthe following ways of validating evacuation models

validation against code requirements validation against fire drills or other peoplersquos

movement experimentstrials

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 113

Gateline Vertical

Platform Right

Stairs Horizontal Stairs Vertical

Travelator Vertical

Ticket Hall

BMOAdmin Reception

Platform Top Ram Up Bottom Vertical

Lift Passage Horizontal

FIGURE 13 Examples of SGEvac station components

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

114 G SMITH AND B CERANIC

FIGURE 15 SGEvac hierarchy function

FIGURE 14 Layout module graphical user interface

validation against literature on past evacuationexperiments

validation against other models third-party validation

Furthermore Galea (1999) deconstructs the validationprocess into four main areas ndash component testingfunctional testing qualitative validation andquantitative validation This approach is more suitablefor models such as SGEvac than for those based oncompliance with code requirements and hence it isthe one adopted for the validation in this research

Component testing is regarded as testingcomputer code that drives the model checking forelementary errors Functional validation involveschecking that the model possesses the ability to

exhibit the range of capabilities required to performthe desired simulations Component testing andfunctional validation for SGEvac is carried outagainst code requirements insofar that allcalculations for infrastructure sizing are evaluatedmanually so that the method and answer are knownto be correct It uses the Simulex program to aid themodelling of people movement and to test thehypothesis of improvement to evacuation timesthus relying on Simulexrsquos own rigorous validationprocess Simulex has been used extensively inpractice as well as a validating model for otherevacuation softwares (Kuligowski and Peacock2005) SGEvac has gone through its initialcompletion and validation process with finalvalidation proposed for the end of the year

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 115

FIGURE 16 Auto-generated SGEvac station layout export file for fire

evacuation simulation in Simulex

FIRE EVACUATION SIMULATION RESULTSThe considered lsquotrain on fire in stationrsquo scenario takesinto account the service interval of 333 minutes(based on 18 trains per hour) the crush capacity of afour-car train together with the number of peoplewaiting on the platform after cancellation of oneservice (ie after 666 minutes) On the oppositeplatform the evacuation load is taken to be thenormal boardingalighting load assuming nocancellations to services (50 of all boarders andalighters)

Evacuation load predictions were based on TfL2006 projected passenger flows with an averageflow of 36 peopleminute for boarders and 123peopleminute for boarders and alighters combinedThe train crush capacity was averaged at 165peoplecar The simulation also models evacuationfrom the ticket hall level to assess the impact ofdispersing crowds at peak times in this part of thestation From these figures the total evacuation loadwas calculated as follows

Platform 1 Boarders (36 x 666 min) = 240Train crush load (165 x 4) = 660

Platform 2 Boarders and alighters

(123 x 05 x 333 min) = 205Ticket hall = 470Total station evacuation load = 1575 people

The simulation results have been captured as imagesat 25 50 100 and 300 second intervals respectivelyfor both platform and ticket hall levels as shown inFigures 17 to 20

The provided means of escape follow LU SPSG4requirements insofar that the maximum travellingdistance on a platform to an exit is no more than 45 m (ie exits must be no more than 90 m apart) andthat the dead-end travelling distance is no more than20 m to an exit Hence each platform is provided withthree exits The MIPsrsquo (mobility impaired persons)means of escape is provided from both ends ofplatforms via emergency evacuation lifts Totalevacuation time (ie all people reaching an exit onticket hallground level) was 5 minutes and

24 seconds with the platform and train load clearedwithin 3 minutes and 5 seconds This is within therequirements to evacuate the platform and train loadwithin 4 minutes with all people reaching a fire-separated route within 6 minutes (the ticket hall is notclassed as a fire-separated route)

The performed simulation was based on theminimum travelling distance maps resulting in eachperson exiting via the nearest fire exit route Theaverage 10 second lsquotime to start to moversquo wasassigned to all occupants in order to accommodatefor real world people behaviour commonly reportedby fire evacuation research studies The boarders onthe end of both platforms (including most peoplefrom end carriages of the train) evacuated viastaircasesMIP lifts directly to the ground level pointof escape The boarders situated around the middleof platforms together with the people from the twomiddle carriages used the central escalators to reachticket hall fire exits Additionally some people fromthe end carriages and some staffretail unitcustomers used this route as means of escape

In this lsquotrain on fire in stationrsquo scenario it wasassumed that none of the escapes routes areunavailable due to maintenance However furtherevacuation studies will investigate possible scenarioswhere one of the escalator routes may be unavailabledue to maintenance reasons

FUTURE RESEARCHIntended future work will concentrate on the furtherconditioning developing and modelling of shapegrammar design rules based on the outcomes of thefinal system validation and LU expertsrsquo feedbackFurther research is proposed on the development ofa novel layout optimization approach based on theshape grammar rules to investigate the evacuation-related parametric improvements in subsurfacesurface railway station design (eg minimizingtravelling distances)

The fire safety strategy in terms of performance-based fire risk design is an important aspirationalaspect of SGEvac requiring further research andliaison with experts to determine how such variedcriteria could be allied with layout planning This isparticularly important where deviation from standardsoccurs Thus further development of the expert

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

116 G SMITH AND B CERANIC

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 117

FIGURE 17 Station evacuation after 25 seconds

FIGURE 18 Station evacuation after 50 seconds

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

118 G SMITH AND B CERANIC

FIGURE 19 Station evacuation after 100 seconds

FIGURE 20 Station evacuation after 300 seconds

knowledge database would be required particularlyin the light of frequent changes to legislation

On a more practical note it could be beneficial todevelop the system on a stand-alone modular basislinked to each generated station component A list offire safety strategy objectives and functional logicstatements can then be specified for commonlyrecurring items (such as emergency lighting fire alarmdetectors sprinkler heads etc) and spaced accordingto the designerrsquos wishes or via shape rule formulation

Site constraint issues would also benefit fromfuture development as well as a move towardsdevelopment of the 3D shape grammar version Oncedevised other relevant 3D modelling systems such ascomputational fluid dynamics (CFD) smoke movementsoftware could be utilized in conjunction with SGEvac

CONCLUSIONSThe computer-aided modelling environment underdevelopment is intended to act as a layout planning aidfor those who practice in sub-surfacesurface railwaystation design The system under developmentcarefully considers London Underground StationPlanning Standards and Guidelines together withother relevant fire safety regulations that influenceoverall fire strategy design The proposed computer-aided design environment (SGEvac) is based on highlyrestrictive additive shape grammar rules driven by akernel of key passenger flow information to generatestation plan solutions

The major benefit to the design team is the ability toreview multiple layouts of accurately sized and correctlyplaced infrastructure at the reference or preliminarystage of the station layout planning design process Timeis hence saved regarding different option appraisals andquantitative testing using evacuation software shouldlead to a choice of more effective layouts in relation tomeans of escape travelling distances and improvementin the total evacuation times Modifications to layoutsshould be straightforward and there is scope toefficiently assess travelling distances and evacuationtimes for each of the considered alternatives comparedwith a lsquotraditionalrsquo approach

The proposed novel shape grammar approach ismost suited to new-build stations or largerefurbishment projects where more design flexibilityexists Sites with many site-specific constraints have

an array of predetermined parameters and will mostlikely benefit from the manual operation of SGEvac

Use in other rail systems due to recurringinfrastructure components and inherent functionallogic related to the laying out of the stationcomponents is also feasible provided that shapeconnections and shape rules are corrected to reflectthat countryrsquos legislative requirements

The potential for station layout improvementswith regard to emergency evacuation is an area thatmerits research especially in scenarios that concernthe safety of the general public and form a vital partof Londonrsquos transportation system Potential layoutimprovements could not only benefit evacuationsituations but also enhance the through-flow ofpeople during peak travelling periods

ACKNOWLEDGEMENTS The authors wish to express their gratitude to JacobsEngineering (UK) Ltd which has sponsored theresearch and to London Underground Ltd forproviding the design guidance that is reproduced inthis paper and for authorising the use of the stationplans London Underground has not reviewed thispaper and the content is solely the authorsrsquo work Mr Robert Hutchinson is also thanked for hiscomputer programming assistance

AUTHOR CONTACT DETAILSGGrraahhaamm SSmmiitthh (corresponding author) Senior Architectural

Technologist Jacobs (UK) Ltd Cardinal Square North Point

10 Nottingham Road Derby DE1 3QT UK Tel +44 (0)1332 234482

fax +44 (0)1332 344558 e-mail grahampsmithjacobscomBBoorriiss CCeerraanniicc Programme Leader for BScBA (Hons) Architectural

Programmes Room Number 105 Faculty of Arts Design and

Technology University of Derby Markeaton Street Derby DE22

3AW UK Tel +44 (0)1332 593136 e-mail bceranicderbyacuk

REFERENCES BS 7974 2001 Application of Fire Safety Engineering Principles to the

Design of Buildings London BSI

Cha MY and Gero JS 1999 Style Learning Inductive Generalisation of

Architectural Shape Patterns Paichai University and Sydney University

co-paper

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 119

Chase SC 1996 lsquoUsing logic to specify shapes and spatial relations in

design grammarsrsquo in Workshop notes Grammatical Design Fourth

International Conference on Artificial Intelligence in Design Stanford

University US June

Department of the Environment 1996 Design Principles of Fire Safety

London The Stationery Office Books

Flemming U 1987 lsquoMore than the sum of its parts the grammar of

Queen Anne housesrsquo in Environment and Planning B Planning and

Design 14 323ndash350

Fruin JJ 1971 Pedestrian Planning and Design New York Metropolitan

Association of Urban Designers and Environmental Planners

Galea ER 1999 lsquoA general approach to validating evacuation models with

an application to EXODUSrsquo in Journal of Fire Sciences 16 414ndash436

Galle P and Kovacs LB 1992 lsquoThe logic of worms a study in

architectural knowledge representationrsquo in Environment and Planning B

Planning and Design 19 5ndash31

Garling T Lindberg E and Mantyla T 1983 lsquoOrientation in buildings ndash

effects of familiarity visual access and orientation aidsrsquo in Journal of

Applied Psychology 68(1) 177ndash186

Grimsdale RL and Chang CW 1996 lsquoThe layout design language A

technique for generating layout plansrsquo in Computer Graphics Forum

15(2) 97ndash106

Heskestad AW 1999 lsquoPerformance in smoke of way guidance systemsrsquo

in Fire and Materials 23(6) 375ndash381

Knight T 1995 lsquoConstructive symmetryrsquo in Environment and Planning B

Planning and Design 22 419ndash450

Knight TW 1999 lsquoShape grammars six typesrsquo in Environment and

Planning B Planning and Design 26 15ndash31

Kuligowski ED and Peacock RD 2005 A Review of Building Evacuation

Models Technical Note 1471 National Institute of Standards and

Technology (NIST) Technology Administration US Department of

Commerce

Li A I-Kang 2003 lsquoThe Yingzao fashi in the information agersquo conference

paper for the Beaux-Arts Paul Philippe Cret and Twentieth Century

Architecture in China The University of Pennsylvania October

Liew H 2003 lsquoSGML ndash A shape grammar meta-languagersquo in Digital

Design 21st eCAADe Conference Proceedings Graz Austria September

639ndash648

Loomis B 2003 lsquoA note on generative design techniques SGGA a user

driven genetic algorithm for evolving non-deterministic shape grammarsrsquo

working paper Massachusetts Institute of Technology (MIT)

Magnusson SE Drysdale DD Fitzgerald RW Mowrer F

Quintiere JG Williamson RB and Zalosh RG 1995

lsquoA proposal for a model curriculum in fire safety engineeringrsquo

in Fire Safety Journal 25 9ndash10

Smith G Ceranic B and Fryer C 2003 lsquoDevelopment of CAD

environment for spatial layout optimisation in sub-surfacesurface rail

station designrsquo in Proceedings of International BIAT Conference The

British Museum London January 2003

SPSG 2007 London Underground Station Planning Standards and

Guidelines SPSG4 (1ndash371) London

Stiny G and Gips J 1971 lsquoShape grammars and the generative

specification of painting and sculpturersquo in Information Processing 71

1460ndash1465

Stiny G 1977 lsquoIce-ray a note on the generation of Chinese lattice

designsrsquo in Environment and Planning B Planning and Design 4 89ndash98

Stiny G 1980 lsquoKindergarten grammars designing with Froebelrsquos building

giftsrsquo in Environment and Planning B 3 409ndash462

Tapia M 1999 lsquoA visual implementation of a shape grammar systemrsquo in

Environment and Planning B Planning and Design 26(1) 59ndash73

Transport for London 2007 Tube Carries One Billion Passengers for First

Time wwwtflgovukcorporatemedianewscentre4770aspx

(accessed 24 March 2008)

Weller M 2007 London Underground private correspondence

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

120 G SMITH AND B CERANIC

criteria neither can be formulated in isolation Thisreinforces the view that computers will in future playan ever-increasing role in aiding and enabling thedesign teams but will never be able to replace them

SYSTEM DEVELOPMENTFigure 6 describes the modular approach proposedfor the system development offering the addedpotential of dual-purpose application in the designprocess It allows the designer to use the systemeither as a whole or to take advantage of itsindependent modules The system offers designerscomplete control over the shape grammarcomposition and layout modifications together withthe constant monitoring facilities and ability todiscard any solutions deemed to be lsquoinferiorrsquo

The proposed system is partitioned into threemain modules ndash lsquoexpert knowledgersquo lsquolayoutrsquo andlsquotestingrsquo modules The expert knowledge modulecaptures the expertise available to facilitate accurateshape grammar generations of layout design Thismodule provides the designer with the ability tointeractively reference Station Planning Standards

and Guidelines (and other relevant fire safetyregulations) allowing himher to make necessary site-specific decisions and to control the lsquohigh levelrsquo shapegrammar composition rules

The designer specifies passenger flow rates anddesignates the overall site boundary dimensions thatdefine the permissible station footprint The final exitpositions are also stipulated at this stage with theseoften being constrained by the existing builtenvironment

With this data supplied the shape grammargenerations commence in the layout module Thearea assignments adjacency requirements and layoutcomposition rules are defined in detail at this stageOnce the designer has made decisions with regard tothe layout composition the computer becomes thecentral tool for generating searching and logicallysorting through the permissible permutations ofpossible layouts The user has the option to auto-generate layouts (usually for sites without manyconstraints) or to manually control the layoutassembly (a decision normally associated with heavilyconstrained sites)

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 105

TABLE 1 Methodology comparison BS 7974 PROCESSES GALLE AND KOVACS SHAPE GRAMMAR SGEvac

LOGIC MODELLING DEVELOPMENT IMPLEMENTATION

Assess benefits Identification Shape grammar Assess constraints

and constraints system identification identify benefitssite

boundariesfinal exit

positions

Qualitative design Conceptualization Initial state Station component

review identification and

interactions

Quantitative analysis Formalization Rules and shape Legislative compliance

language passenger flow figures

area assignments

Satisfactory Implementation Shape grammar Alternative station

unsatisfactory executions layouts generation

Report and review Testing Design analysis Testingvalidation

iterationsevolution of layouts generation

End Experimentation Experimentation

iterationsimprovements

The testing module is used to evaluate potentiallysuitable layouts at reference level by exporting them toAutoCAD via standard drawing data exchange formatand analysing them in the context of project-specificsite constraints and requirements A level of drawingdetail is then added and the layout solution is exportedto Simulex a building evacuation simulation software

package Once the occupant numbers and persongroups are assigned and the escape and verticalcirculation routes defined the simulation software iscapable of the accurate measurement of all travellingdistances creation of distance maps and calculation ofevacuation times for different fire scenarios (egdifferent group affiliations persons exiting via different

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

106 G SMITH AND B CERANIC

EGY D ESIGNDesigner

Feasibility studiesProject brief requirements

Passenger flow data

DDE

DDE

Layout Production

Shape grammar rulesMultiple layout production

Layout Analysis

Layout interrogationsDesign team judgement

No

Yes

DXF

DXF

Yes

No

AutoCAD

Feed lsquoblock reference planrsquoresults into overall project

Improvedsolution identified

EXPERT KNOWLEDGE MODULE

Fire Strategy Design

LU SPSG4FSE legislation amp guidance

Infrastructure size calculationHigh level brief development

LAYOUT MODULE

Layout Composition

Area and adjacency planningLayout option appraisal

Layoutimproved

TESTING MODULE

AutoCAD

SGEvac export into CAD

Simulex

Fire evacuation scenarios

DesignerSGEvac

DDE ndash Database Data Exchange

DXF ndash Standard Drawing Data Exchange

FIGURE 6 Proposed SGEvac algorithm

routes and having delayed movement in response to afire alarm signal) The design team then assesses theresults against the qualitative fire safety design goalsset earlier to establish whether improvements toevacuation times have been made

DETAILED ANALYSIS OF STATIONCOMPONENTSIn conjunction with the site boundary final exitpositions and passenger flow specificationconsideration must also be given to the vertical sitedimension The shape grammar system assumes allfinal exits are at ground level with the ticket hall andplatform(s) being the other two elements that requirethe user to determine their depth relative to the finalexits In so doing SGEvac selects the correctinterchange components by an event-drivenmechanism that evaluates the height change andpassenger volume demand Thresholds for bothcriteria have been defined as shown in Table 2 andEquations 2ndash5

If H (height in mm) = 0 ndash 500 then I (infrastructure) = Ramp (2)

If H = 500 ndash 3000 and P (passengers) le 3000 then I = Stairway (3)

If H = 3000 ndash 5000 And P le 3000 then I = StairwayElse I = Escalator (4)

If H gt 5000 and P le 1000 then I = StairwayElse I = Escalator and Lift (5)

(where lsquoElsersquo denotes the mechanism for the choiceof the alternative infrastructure component if theoriginal condition is not satisfied)

The above constructs also condition the run-offdimensions associated with each of theinterchange components (between 4 m and 12 mas stipulated by SPSG) This being the case run-offs are used to size the ticket hall length Figure 7shows the paid side of the ticket hall with variabledistances shown between the escalator or staircaseand the gateline

The paid side of the ticket hall is always dictatedby run-off distance while the unpaid side is governedby two parameters the run-off and the concoursearea required for the heaviest 15-minute passengerflow If the maximum 15-minute passenger flowfigure multiplied by 1 m2 (say answer A) is larger inplan area than the product of the gateline widthmultiplied by the relevant run-off (say answer B) thenanswer A will be incorporated into the layout Thereverse is true if B is the greater number This ensuresthat the station is always sized appropriately

Figure 8 shows a ticket hall layout with thegateline dictating the width (denoted as Tick_H_wd)itself conditioned by the passenger flow figurescalculated as shown in Table 3 The table onlyevaluates entry flow requirements but SGEvac willrepeat the same process for exit gates to produce acorrectly sized gateline

In this case the maximum 15-minute passengerflow capacity size prevailed over the run-off to createthe unpaid side overall

The divide between the paid and unpaid side ofthe ticket hall is the sized according to Table 4

SGEvac shape rules mandate that entry gates arealways located adjacent to the ticket office queuingspace (designated as Q_TIW) to avoid conflictingpassenger entry and exit flows The LU standard is to

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 107

TABLE 2 Interchange infrastructure selectionINFRASTRUCTURE REASON FOR SELECTION

Ramp Vertical rise between 0ndash05 m preferred to steps due to disabled access issues generally considered

people will travel faster on ramps than up or down steps

Stairs Vertical rise between 05ndash5 m but greater height available for selection if station has a light passenger flow

Escalator Vertical rise is minimum of 3 m for heavy passenger flow stations but used for lightmedium flow stations

above 5 m

Lift Vertical rise above 5 m also required for accessibility of disabled persons Lifts always used with escalators

not as a substitute

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

108 G SMITH AND B CERANIC

Con_Area

Exit

UTS gates

6601900 2450

2400

EFW

Ticketoffice

POMcounter

4000

Tick_H_wd

1000

Run

-off

Entry

Q_space

Q_TIW

FIGURE 8 SGEvac ticket hall arrangement

Heavy flow escalator

Escalator

Staircase

Medium to Heavy flow

Medium to Heavy flow

Light flow escalator

Light flow

Heavy flow 12mMedium flow 8ndash12m

Light flow

Light flow 8m

Light flow 6m

Heavy flow

staircase to gateline

staircase to gateline 10m

Light flow

to gateline12mGateline

Gateline

EscalatorndashGateline

Total Run-off

Heavy flow 10mMedium flow 6ndash10m

StaircasendashGateline

Total Run-off

to gateline 8m

2m 4m 6m 8m 10m

2m 4m 6m

6m

8m 10m

12m

FIGURE 7 Passenger flow rate run-off requirements (SPSG 2007)

allow for 4 m (or a five-person queue) as shown inFigure 8 The overall gateline is defined by summingboth exit and entry gate numbers obtained from user-inputted passenger figures

The length of the ticket office interface with theticket hall is stipulated by the number of ticket issuingwindows (TIWs) plus excess fare windows (EFWs)TIWs always belong on the unpaid side of the tickethall and EFWs always on the paid side both forrevenue protection reasons This situation facilitates aprecise shape grammar labelling for the ticket officeto ticket hallqueue space connection

Passenger operated machine (POM) enclosures arenormally connected to the ticket hall ticket office andits own queue space (noted as Q_space in Figure 8)

To make shape rule specification manageable allstation areas are divided into groups arranged to suitthe most probable connection patterns in the stationTable 5 shows the group station managerrsquos officeexample with all other station components followingthe same concept

London Underground views its stations as largebasements (Weller 2007) and escape distances fromplatforms are based on agreed dispensations with theLondon Fire Brigade SGEvac respects thesedistances by placing restrictive labels on the platformso that they cannot be exceeded as demonstrated bythe example shown in Figure 9

SGEvac escape stair infrastructure may bepositioned up to 20 m from the end of platforms toavoid undesirable lsquodead endrsquo conditions The 45 mmaximum platform travel distance starts from thenearside junction of the two 33 m wide escapestaircases with the 45 m distance rule creating ashortfall in the platform centre of 34 m Generallylocating three interchange components (eg stairs

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 109

TABLE 3 Gateline width calculationMETHOD WORKED EXAMPLE

Obtain peak hourly flow rate through Say 6500 peoplehr

ticket hall

Calculate 5-minute peak passenger flow 650012 = 541

Allowance for ticket rejection 541 + 20 = 561

Allowance for growth in use demand 561 x 110 = 617

Assume 125 people use each gate over

the 5-minute peak passenger flow period 617125 = 493

Round up gates to nearest whole number 5 gates required

TABLE 4 London Underground ticket hall gateline parametersregulations (SPSG 2007)EQUIPMENT DIMENSIONS (W X D) POSITIONARRANGEMENT

Combined manualequipment gate 1750 x 270 mm Height ndash 1300 mm Adjacent to ticket office Can include

1400 mm clear opening luggage facility

Manual gate ndash passenger 1750 x 270 mm Height ndash 1300 mm Adjacent to ticket office With more than 6

gateluggage port 900 mm ndash clearance and walkways in a single gateline or significant

outward opening interchange with mainline stations airport or

tourist traffic provide separate luggage chutes

Manual gate ndash single leaf 1394 mm Height ndash 1300 mm Beside the dedicated excess gates

with luggage port 695 mm clearance

Gateline attendance report 990 x 1240 mm Within the controlled zone but not necessarily

part of the gateline Usually close

Station control unit (SCU) 432 x 430 mm Should be installed within a secure enclosure

(which could be a ticket office) and also with

a clear view of the gates Controls and monitors

the station underground ticketing system

equipment

Manual emergency exit gates 1900 x 125 mm Adjacent to ticket office Add these gates rather

than adding extra UTS gates for evacuation

purposes

or escalators) at these distances will result incompliance with the Underground standards but itmay not always be possible (or practical for aninterchange situation) to procure such a layout

Lengths of station components are controlled bya range of criteria For the platform it is the length ofthe train vehicle for interchange components it isthe vertical height in the terms of the depth to theplatform level for passages it is the designated finalexit positions and the existing station areas alreadygenerated

Vector direction labels are underpinned by thepermissible union of compatible spaces two points ofcontact will be maintained so as not to restrict thewidth of connecting passenger areas This undueconstriction of passenger flow at the head of apassage also satisfies realistic construction practiceNon-passenger and ancillary area vector directions areas per project requirements and the successive areaadditions managed by the SGEvac rule groupings

To create alternative layouts SGEvac uses alsquorandomizersquo rule In this shape grammar rule initialstate infrastructure is generated as per guidelinesbut it can randomly change the location ofcomponent arrangements within the predefined

boundaries so that a number of possible alternativelayouts are produced The designs can then beinterrogated and ranked by their travelling distancesand against the qualitative functional logicstatements

The SGEvac design tool developed in this researchis based on a highly restrictive additive grammar ruleset due to SPSGrsquos prescriptive guidance and thenature of the layout planning problem As Grimsdaleand Chang (1996) argue many published grammarslack high-level reasoning during the iteration processsince they are often concerned with the developmentof shape grammar theories rather than their practicalapplications By contrast SGEvac being a practicalapplication has a complete set of primitive shapesonly possible to arrange in a certain way It thusprovides viable reference design layout compositionsbased on SPSG with further interventions possible bythe designer

SYSTEM DEVELOPMENT TO DATEThe SGEvac prototype is compiled within the Clanguage environment allowing for modular object-oriented programming (suitable for shape grammar)Figure 10 shows the main interface where the stationname and relevant LU fare zone are entered (whichimpacts on sizing infrastructure) as are the ticket halland platform depths the overall site boundaries andfinal exitsrsquo positions The peak passenger flow figuresover the relevant time interval of normally 3 hours arealso specified SGEvac is then capable of accuratelycalculating all passenger area infrastructure based onthe SPSG guidelines as already discussed

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

110 G SMITH AND B CERANIC

TABLE 5 Station manager suite ndash shape rule connectionsELEMENT SHAPE RULE DESIGNATION CONNECTED COMPONENT GROUP ALLOCATION

Group station managerrsquos office SR_GSM_01 Ticket hall paid side GSM

SR_GSM_02 Ticket hall unpaid side

SR_GSM_03 GSM administration

SR_GSM_04 Duty station managerrsquos office

SR_GSM_05 Business managerrsquos office

SR_GSM_06 Administration

SR_GSM_07 Ticket office

SR_GSM_08 Police room

SR_GSM_09 Revenue protection examiner

SR_GSM_10 Interview room

140

Stair Stair

100 2020

4533

34

45

FIGURE 9 Platform escape distance SGEvac label restrictions

The history of auto-generated layouts is accessiblefrom this screen and the user is given the capabilityto delete and copy them (useful when minoralterations are required to produce an alternativelayout)

Shape labelling in the programme is multi-faceted which is necessary since many stationcomponents have various possible connections Anescalator for example may connect to a passagetravelator ramp escalator run-off stair run-off lift orthe ticket hall so all such connections need adding tothe component properties The SGEvac grouping ofcomponent interactions is accessed from menutoolbars that allow insertion of legislative compliantstation area connections as shown in Figure 11

Once the position and the depth of the ticket halland platform are specified the railway track is inserteddependent on the platform configuration The platformis the only possible connection to the track but thetrack can be positioned either side of platforms so thata side or centre (island) platform layout is possible

Non-passenger and ancillary areas are alsoselectable from the shape grammar library via property

windows for each station component eg the ticketoffice properties window as shown in Figure 12

Figure 13 shows examples of the layout modulestation components available in the SGEvacprogramme

Figure 14 shows the layout module interfaceInformation on the components (relative position inthe overall site layout and its absolute sizing) isdisplayed at the bottom of the screen once thecomponent is selected from the list In addition ameasurement bar (in metres) at the base of theinteractive design environment provides a sense ofscale for the designer

A checklist summarizes the station areas includedin the layout but a more detailed analysis is possiblevia a report function if required In addition to thisdetailed list a summary screen accessed from themain interface collates all specified components ineach generated layout This screen lists all passengernon-passenger and ancillary areas in their respectivecategories with infrastructure width data ofpassenger areas also summarized It also has anevacuation summary where the train frequency is

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 111

FIGURE 10 SGEvac main interface

specified and the train evacuation load (taken as peaklsquorush hourrsquo loading) is displayed This section of thesummary screen is useful for input into Simulexwhere fire evacuation test runs of the building layoutcan be implemented

The lsquoshow hierarchyrsquo function in Figure 15 allowsdesign team members to view the progression ofdesigns by superimposing connecting lines betweenstation elements in the order of their compositionThis also serves as a safeguard against deleting

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

112 G SMITH AND B CERANIC

FIGURE 12 SGEvac ticket office properties window

FIGURE 11 SGEvac properties window

components out of sequence that could lead to aninvalid solution

Once auto-generated layouts are complete withcorrectly sized and legitimately connected stationcomponents manual insertion of non-passenger areaswould then commence Upon completion thegenerated layout is transferred to AutoCAD via dxf fileformat as shown in Figure 16 Once in AutoCAD thelayout has a level of detail added by the designer suchas door openings passage connections fire escapeopenings etc so that passenger flows may beassessed It is then exported to Simulex for fireevacuation simulation analysis again via dxf file format

Simulex understands any voids as potentialcirculation routes treating the rest as solid andpermanent obstructions ie walls internal structuralelements permanent fixtures etc Since its solutionoperates in a two-dimensional environment for whatin effect is a three-dimensional problem the means

of vertical circulation (eg staircases) have to beestablished True lengths of staircases must bespecified based on the total length of all flights(sloping length) and the intermediate rests The fireexits need to be defined both in terms of theirlocation and width The links have to be establishedto define the point of the connection between thefloors and staircasesescalatorsramps The travellingdistance maps are then calculated type and numberof occupants specified building populated and theevacuation simulation performed

SYSTEM TESTING AND VALIDATIONKuligowski and Peacock (2005) define and recommendthe following ways of validating evacuation models

validation against code requirements validation against fire drills or other peoplersquos

movement experimentstrials

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 113

Gateline Vertical

Platform Right

Stairs Horizontal Stairs Vertical

Travelator Vertical

Ticket Hall

BMOAdmin Reception

Platform Top Ram Up Bottom Vertical

Lift Passage Horizontal

FIGURE 13 Examples of SGEvac station components

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

114 G SMITH AND B CERANIC

FIGURE 15 SGEvac hierarchy function

FIGURE 14 Layout module graphical user interface

validation against literature on past evacuationexperiments

validation against other models third-party validation

Furthermore Galea (1999) deconstructs the validationprocess into four main areas ndash component testingfunctional testing qualitative validation andquantitative validation This approach is more suitablefor models such as SGEvac than for those based oncompliance with code requirements and hence it isthe one adopted for the validation in this research

Component testing is regarded as testingcomputer code that drives the model checking forelementary errors Functional validation involveschecking that the model possesses the ability to

exhibit the range of capabilities required to performthe desired simulations Component testing andfunctional validation for SGEvac is carried outagainst code requirements insofar that allcalculations for infrastructure sizing are evaluatedmanually so that the method and answer are knownto be correct It uses the Simulex program to aid themodelling of people movement and to test thehypothesis of improvement to evacuation timesthus relying on Simulexrsquos own rigorous validationprocess Simulex has been used extensively inpractice as well as a validating model for otherevacuation softwares (Kuligowski and Peacock2005) SGEvac has gone through its initialcompletion and validation process with finalvalidation proposed for the end of the year

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 115

FIGURE 16 Auto-generated SGEvac station layout export file for fire

evacuation simulation in Simulex

FIRE EVACUATION SIMULATION RESULTSThe considered lsquotrain on fire in stationrsquo scenario takesinto account the service interval of 333 minutes(based on 18 trains per hour) the crush capacity of afour-car train together with the number of peoplewaiting on the platform after cancellation of oneservice (ie after 666 minutes) On the oppositeplatform the evacuation load is taken to be thenormal boardingalighting load assuming nocancellations to services (50 of all boarders andalighters)

Evacuation load predictions were based on TfL2006 projected passenger flows with an averageflow of 36 peopleminute for boarders and 123peopleminute for boarders and alighters combinedThe train crush capacity was averaged at 165peoplecar The simulation also models evacuationfrom the ticket hall level to assess the impact ofdispersing crowds at peak times in this part of thestation From these figures the total evacuation loadwas calculated as follows

Platform 1 Boarders (36 x 666 min) = 240Train crush load (165 x 4) = 660

Platform 2 Boarders and alighters

(123 x 05 x 333 min) = 205Ticket hall = 470Total station evacuation load = 1575 people

The simulation results have been captured as imagesat 25 50 100 and 300 second intervals respectivelyfor both platform and ticket hall levels as shown inFigures 17 to 20

The provided means of escape follow LU SPSG4requirements insofar that the maximum travellingdistance on a platform to an exit is no more than 45 m (ie exits must be no more than 90 m apart) andthat the dead-end travelling distance is no more than20 m to an exit Hence each platform is provided withthree exits The MIPsrsquo (mobility impaired persons)means of escape is provided from both ends ofplatforms via emergency evacuation lifts Totalevacuation time (ie all people reaching an exit onticket hallground level) was 5 minutes and

24 seconds with the platform and train load clearedwithin 3 minutes and 5 seconds This is within therequirements to evacuate the platform and train loadwithin 4 minutes with all people reaching a fire-separated route within 6 minutes (the ticket hall is notclassed as a fire-separated route)

The performed simulation was based on theminimum travelling distance maps resulting in eachperson exiting via the nearest fire exit route Theaverage 10 second lsquotime to start to moversquo wasassigned to all occupants in order to accommodatefor real world people behaviour commonly reportedby fire evacuation research studies The boarders onthe end of both platforms (including most peoplefrom end carriages of the train) evacuated viastaircasesMIP lifts directly to the ground level pointof escape The boarders situated around the middleof platforms together with the people from the twomiddle carriages used the central escalators to reachticket hall fire exits Additionally some people fromthe end carriages and some staffretail unitcustomers used this route as means of escape

In this lsquotrain on fire in stationrsquo scenario it wasassumed that none of the escapes routes areunavailable due to maintenance However furtherevacuation studies will investigate possible scenarioswhere one of the escalator routes may be unavailabledue to maintenance reasons

FUTURE RESEARCHIntended future work will concentrate on the furtherconditioning developing and modelling of shapegrammar design rules based on the outcomes of thefinal system validation and LU expertsrsquo feedbackFurther research is proposed on the development ofa novel layout optimization approach based on theshape grammar rules to investigate the evacuation-related parametric improvements in subsurfacesurface railway station design (eg minimizingtravelling distances)

The fire safety strategy in terms of performance-based fire risk design is an important aspirationalaspect of SGEvac requiring further research andliaison with experts to determine how such variedcriteria could be allied with layout planning This isparticularly important where deviation from standardsoccurs Thus further development of the expert

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

116 G SMITH AND B CERANIC

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 117

FIGURE 17 Station evacuation after 25 seconds

FIGURE 18 Station evacuation after 50 seconds

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

118 G SMITH AND B CERANIC

FIGURE 19 Station evacuation after 100 seconds

FIGURE 20 Station evacuation after 300 seconds

knowledge database would be required particularlyin the light of frequent changes to legislation

On a more practical note it could be beneficial todevelop the system on a stand-alone modular basislinked to each generated station component A list offire safety strategy objectives and functional logicstatements can then be specified for commonlyrecurring items (such as emergency lighting fire alarmdetectors sprinkler heads etc) and spaced accordingto the designerrsquos wishes or via shape rule formulation

Site constraint issues would also benefit fromfuture development as well as a move towardsdevelopment of the 3D shape grammar version Oncedevised other relevant 3D modelling systems such ascomputational fluid dynamics (CFD) smoke movementsoftware could be utilized in conjunction with SGEvac

CONCLUSIONSThe computer-aided modelling environment underdevelopment is intended to act as a layout planning aidfor those who practice in sub-surfacesurface railwaystation design The system under developmentcarefully considers London Underground StationPlanning Standards and Guidelines together withother relevant fire safety regulations that influenceoverall fire strategy design The proposed computer-aided design environment (SGEvac) is based on highlyrestrictive additive shape grammar rules driven by akernel of key passenger flow information to generatestation plan solutions

The major benefit to the design team is the ability toreview multiple layouts of accurately sized and correctlyplaced infrastructure at the reference or preliminarystage of the station layout planning design process Timeis hence saved regarding different option appraisals andquantitative testing using evacuation software shouldlead to a choice of more effective layouts in relation tomeans of escape travelling distances and improvementin the total evacuation times Modifications to layoutsshould be straightforward and there is scope toefficiently assess travelling distances and evacuationtimes for each of the considered alternatives comparedwith a lsquotraditionalrsquo approach

The proposed novel shape grammar approach ismost suited to new-build stations or largerefurbishment projects where more design flexibilityexists Sites with many site-specific constraints have

an array of predetermined parameters and will mostlikely benefit from the manual operation of SGEvac

Use in other rail systems due to recurringinfrastructure components and inherent functionallogic related to the laying out of the stationcomponents is also feasible provided that shapeconnections and shape rules are corrected to reflectthat countryrsquos legislative requirements

The potential for station layout improvementswith regard to emergency evacuation is an area thatmerits research especially in scenarios that concernthe safety of the general public and form a vital partof Londonrsquos transportation system Potential layoutimprovements could not only benefit evacuationsituations but also enhance the through-flow ofpeople during peak travelling periods

ACKNOWLEDGEMENTS The authors wish to express their gratitude to JacobsEngineering (UK) Ltd which has sponsored theresearch and to London Underground Ltd forproviding the design guidance that is reproduced inthis paper and for authorising the use of the stationplans London Underground has not reviewed thispaper and the content is solely the authorsrsquo work Mr Robert Hutchinson is also thanked for hiscomputer programming assistance

AUTHOR CONTACT DETAILSGGrraahhaamm SSmmiitthh (corresponding author) Senior Architectural

Technologist Jacobs (UK) Ltd Cardinal Square North Point

10 Nottingham Road Derby DE1 3QT UK Tel +44 (0)1332 234482

fax +44 (0)1332 344558 e-mail grahampsmithjacobscomBBoorriiss CCeerraanniicc Programme Leader for BScBA (Hons) Architectural

Programmes Room Number 105 Faculty of Arts Design and

Technology University of Derby Markeaton Street Derby DE22

3AW UK Tel +44 (0)1332 593136 e-mail bceranicderbyacuk

REFERENCES BS 7974 2001 Application of Fire Safety Engineering Principles to the

Design of Buildings London BSI

Cha MY and Gero JS 1999 Style Learning Inductive Generalisation of

Architectural Shape Patterns Paichai University and Sydney University

co-paper

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 119

Chase SC 1996 lsquoUsing logic to specify shapes and spatial relations in

design grammarsrsquo in Workshop notes Grammatical Design Fourth

International Conference on Artificial Intelligence in Design Stanford

University US June

Department of the Environment 1996 Design Principles of Fire Safety

London The Stationery Office Books

Flemming U 1987 lsquoMore than the sum of its parts the grammar of

Queen Anne housesrsquo in Environment and Planning B Planning and

Design 14 323ndash350

Fruin JJ 1971 Pedestrian Planning and Design New York Metropolitan

Association of Urban Designers and Environmental Planners

Galea ER 1999 lsquoA general approach to validating evacuation models with

an application to EXODUSrsquo in Journal of Fire Sciences 16 414ndash436

Galle P and Kovacs LB 1992 lsquoThe logic of worms a study in

architectural knowledge representationrsquo in Environment and Planning B

Planning and Design 19 5ndash31

Garling T Lindberg E and Mantyla T 1983 lsquoOrientation in buildings ndash

effects of familiarity visual access and orientation aidsrsquo in Journal of

Applied Psychology 68(1) 177ndash186

Grimsdale RL and Chang CW 1996 lsquoThe layout design language A

technique for generating layout plansrsquo in Computer Graphics Forum

15(2) 97ndash106

Heskestad AW 1999 lsquoPerformance in smoke of way guidance systemsrsquo

in Fire and Materials 23(6) 375ndash381

Knight T 1995 lsquoConstructive symmetryrsquo in Environment and Planning B

Planning and Design 22 419ndash450

Knight TW 1999 lsquoShape grammars six typesrsquo in Environment and

Planning B Planning and Design 26 15ndash31

Kuligowski ED and Peacock RD 2005 A Review of Building Evacuation

Models Technical Note 1471 National Institute of Standards and

Technology (NIST) Technology Administration US Department of

Commerce

Li A I-Kang 2003 lsquoThe Yingzao fashi in the information agersquo conference

paper for the Beaux-Arts Paul Philippe Cret and Twentieth Century

Architecture in China The University of Pennsylvania October

Liew H 2003 lsquoSGML ndash A shape grammar meta-languagersquo in Digital

Design 21st eCAADe Conference Proceedings Graz Austria September

639ndash648

Loomis B 2003 lsquoA note on generative design techniques SGGA a user

driven genetic algorithm for evolving non-deterministic shape grammarsrsquo

working paper Massachusetts Institute of Technology (MIT)

Magnusson SE Drysdale DD Fitzgerald RW Mowrer F

Quintiere JG Williamson RB and Zalosh RG 1995

lsquoA proposal for a model curriculum in fire safety engineeringrsquo

in Fire Safety Journal 25 9ndash10

Smith G Ceranic B and Fryer C 2003 lsquoDevelopment of CAD

environment for spatial layout optimisation in sub-surfacesurface rail

station designrsquo in Proceedings of International BIAT Conference The

British Museum London January 2003

SPSG 2007 London Underground Station Planning Standards and

Guidelines SPSG4 (1ndash371) London

Stiny G and Gips J 1971 lsquoShape grammars and the generative

specification of painting and sculpturersquo in Information Processing 71

1460ndash1465

Stiny G 1977 lsquoIce-ray a note on the generation of Chinese lattice

designsrsquo in Environment and Planning B Planning and Design 4 89ndash98

Stiny G 1980 lsquoKindergarten grammars designing with Froebelrsquos building

giftsrsquo in Environment and Planning B 3 409ndash462

Tapia M 1999 lsquoA visual implementation of a shape grammar systemrsquo in

Environment and Planning B Planning and Design 26(1) 59ndash73

Transport for London 2007 Tube Carries One Billion Passengers for First

Time wwwtflgovukcorporatemedianewscentre4770aspx

(accessed 24 March 2008)

Weller M 2007 London Underground private correspondence

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

120 G SMITH AND B CERANIC

The testing module is used to evaluate potentiallysuitable layouts at reference level by exporting them toAutoCAD via standard drawing data exchange formatand analysing them in the context of project-specificsite constraints and requirements A level of drawingdetail is then added and the layout solution is exportedto Simulex a building evacuation simulation software

package Once the occupant numbers and persongroups are assigned and the escape and verticalcirculation routes defined the simulation software iscapable of the accurate measurement of all travellingdistances creation of distance maps and calculation ofevacuation times for different fire scenarios (egdifferent group affiliations persons exiting via different

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

106 G SMITH AND B CERANIC

EGY D ESIGNDesigner

Feasibility studiesProject brief requirements

Passenger flow data

DDE

DDE

Layout Production

Shape grammar rulesMultiple layout production

Layout Analysis

Layout interrogationsDesign team judgement

No

Yes

DXF

DXF

Yes

No

AutoCAD

Feed lsquoblock reference planrsquoresults into overall project

Improvedsolution identified

EXPERT KNOWLEDGE MODULE

Fire Strategy Design

LU SPSG4FSE legislation amp guidance

Infrastructure size calculationHigh level brief development

LAYOUT MODULE

Layout Composition

Area and adjacency planningLayout option appraisal

Layoutimproved

TESTING MODULE

AutoCAD

SGEvac export into CAD

Simulex

Fire evacuation scenarios

DesignerSGEvac

DDE ndash Database Data Exchange

DXF ndash Standard Drawing Data Exchange

FIGURE 6 Proposed SGEvac algorithm

routes and having delayed movement in response to afire alarm signal) The design team then assesses theresults against the qualitative fire safety design goalsset earlier to establish whether improvements toevacuation times have been made

DETAILED ANALYSIS OF STATIONCOMPONENTSIn conjunction with the site boundary final exitpositions and passenger flow specificationconsideration must also be given to the vertical sitedimension The shape grammar system assumes allfinal exits are at ground level with the ticket hall andplatform(s) being the other two elements that requirethe user to determine their depth relative to the finalexits In so doing SGEvac selects the correctinterchange components by an event-drivenmechanism that evaluates the height change andpassenger volume demand Thresholds for bothcriteria have been defined as shown in Table 2 andEquations 2ndash5

If H (height in mm) = 0 ndash 500 then I (infrastructure) = Ramp (2)

If H = 500 ndash 3000 and P (passengers) le 3000 then I = Stairway (3)

If H = 3000 ndash 5000 And P le 3000 then I = StairwayElse I = Escalator (4)

If H gt 5000 and P le 1000 then I = StairwayElse I = Escalator and Lift (5)

(where lsquoElsersquo denotes the mechanism for the choiceof the alternative infrastructure component if theoriginal condition is not satisfied)

The above constructs also condition the run-offdimensions associated with each of theinterchange components (between 4 m and 12 mas stipulated by SPSG) This being the case run-offs are used to size the ticket hall length Figure 7shows the paid side of the ticket hall with variabledistances shown between the escalator or staircaseand the gateline

The paid side of the ticket hall is always dictatedby run-off distance while the unpaid side is governedby two parameters the run-off and the concoursearea required for the heaviest 15-minute passengerflow If the maximum 15-minute passenger flowfigure multiplied by 1 m2 (say answer A) is larger inplan area than the product of the gateline widthmultiplied by the relevant run-off (say answer B) thenanswer A will be incorporated into the layout Thereverse is true if B is the greater number This ensuresthat the station is always sized appropriately

Figure 8 shows a ticket hall layout with thegateline dictating the width (denoted as Tick_H_wd)itself conditioned by the passenger flow figurescalculated as shown in Table 3 The table onlyevaluates entry flow requirements but SGEvac willrepeat the same process for exit gates to produce acorrectly sized gateline

In this case the maximum 15-minute passengerflow capacity size prevailed over the run-off to createthe unpaid side overall

The divide between the paid and unpaid side ofthe ticket hall is the sized according to Table 4

SGEvac shape rules mandate that entry gates arealways located adjacent to the ticket office queuingspace (designated as Q_TIW) to avoid conflictingpassenger entry and exit flows The LU standard is to

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 107

TABLE 2 Interchange infrastructure selectionINFRASTRUCTURE REASON FOR SELECTION

Ramp Vertical rise between 0ndash05 m preferred to steps due to disabled access issues generally considered

people will travel faster on ramps than up or down steps

Stairs Vertical rise between 05ndash5 m but greater height available for selection if station has a light passenger flow

Escalator Vertical rise is minimum of 3 m for heavy passenger flow stations but used for lightmedium flow stations

above 5 m

Lift Vertical rise above 5 m also required for accessibility of disabled persons Lifts always used with escalators

not as a substitute

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

108 G SMITH AND B CERANIC

Con_Area

Exit

UTS gates

6601900 2450

2400

EFW

Ticketoffice

POMcounter

4000

Tick_H_wd

1000

Run

-off

Entry

Q_space

Q_TIW

FIGURE 8 SGEvac ticket hall arrangement

Heavy flow escalator

Escalator

Staircase

Medium to Heavy flow

Medium to Heavy flow

Light flow escalator

Light flow

Heavy flow 12mMedium flow 8ndash12m

Light flow

Light flow 8m

Light flow 6m

Heavy flow

staircase to gateline

staircase to gateline 10m

Light flow

to gateline12mGateline

Gateline

EscalatorndashGateline

Total Run-off

Heavy flow 10mMedium flow 6ndash10m

StaircasendashGateline

Total Run-off

to gateline 8m

2m 4m 6m 8m 10m

2m 4m 6m

6m

8m 10m

12m

FIGURE 7 Passenger flow rate run-off requirements (SPSG 2007)

allow for 4 m (or a five-person queue) as shown inFigure 8 The overall gateline is defined by summingboth exit and entry gate numbers obtained from user-inputted passenger figures

The length of the ticket office interface with theticket hall is stipulated by the number of ticket issuingwindows (TIWs) plus excess fare windows (EFWs)TIWs always belong on the unpaid side of the tickethall and EFWs always on the paid side both forrevenue protection reasons This situation facilitates aprecise shape grammar labelling for the ticket officeto ticket hallqueue space connection

Passenger operated machine (POM) enclosures arenormally connected to the ticket hall ticket office andits own queue space (noted as Q_space in Figure 8)

To make shape rule specification manageable allstation areas are divided into groups arranged to suitthe most probable connection patterns in the stationTable 5 shows the group station managerrsquos officeexample with all other station components followingthe same concept

London Underground views its stations as largebasements (Weller 2007) and escape distances fromplatforms are based on agreed dispensations with theLondon Fire Brigade SGEvac respects thesedistances by placing restrictive labels on the platformso that they cannot be exceeded as demonstrated bythe example shown in Figure 9

SGEvac escape stair infrastructure may bepositioned up to 20 m from the end of platforms toavoid undesirable lsquodead endrsquo conditions The 45 mmaximum platform travel distance starts from thenearside junction of the two 33 m wide escapestaircases with the 45 m distance rule creating ashortfall in the platform centre of 34 m Generallylocating three interchange components (eg stairs

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 109

TABLE 3 Gateline width calculationMETHOD WORKED EXAMPLE

Obtain peak hourly flow rate through Say 6500 peoplehr

ticket hall

Calculate 5-minute peak passenger flow 650012 = 541

Allowance for ticket rejection 541 + 20 = 561

Allowance for growth in use demand 561 x 110 = 617

Assume 125 people use each gate over

the 5-minute peak passenger flow period 617125 = 493

Round up gates to nearest whole number 5 gates required

TABLE 4 London Underground ticket hall gateline parametersregulations (SPSG 2007)EQUIPMENT DIMENSIONS (W X D) POSITIONARRANGEMENT

Combined manualequipment gate 1750 x 270 mm Height ndash 1300 mm Adjacent to ticket office Can include

1400 mm clear opening luggage facility

Manual gate ndash passenger 1750 x 270 mm Height ndash 1300 mm Adjacent to ticket office With more than 6

gateluggage port 900 mm ndash clearance and walkways in a single gateline or significant

outward opening interchange with mainline stations airport or

tourist traffic provide separate luggage chutes

Manual gate ndash single leaf 1394 mm Height ndash 1300 mm Beside the dedicated excess gates

with luggage port 695 mm clearance

Gateline attendance report 990 x 1240 mm Within the controlled zone but not necessarily

part of the gateline Usually close

Station control unit (SCU) 432 x 430 mm Should be installed within a secure enclosure

(which could be a ticket office) and also with

a clear view of the gates Controls and monitors

the station underground ticketing system

equipment

Manual emergency exit gates 1900 x 125 mm Adjacent to ticket office Add these gates rather

than adding extra UTS gates for evacuation

purposes

or escalators) at these distances will result incompliance with the Underground standards but itmay not always be possible (or practical for aninterchange situation) to procure such a layout

Lengths of station components are controlled bya range of criteria For the platform it is the length ofthe train vehicle for interchange components it isthe vertical height in the terms of the depth to theplatform level for passages it is the designated finalexit positions and the existing station areas alreadygenerated

Vector direction labels are underpinned by thepermissible union of compatible spaces two points ofcontact will be maintained so as not to restrict thewidth of connecting passenger areas This undueconstriction of passenger flow at the head of apassage also satisfies realistic construction practiceNon-passenger and ancillary area vector directions areas per project requirements and the successive areaadditions managed by the SGEvac rule groupings

To create alternative layouts SGEvac uses alsquorandomizersquo rule In this shape grammar rule initialstate infrastructure is generated as per guidelinesbut it can randomly change the location ofcomponent arrangements within the predefined

boundaries so that a number of possible alternativelayouts are produced The designs can then beinterrogated and ranked by their travelling distancesand against the qualitative functional logicstatements

The SGEvac design tool developed in this researchis based on a highly restrictive additive grammar ruleset due to SPSGrsquos prescriptive guidance and thenature of the layout planning problem As Grimsdaleand Chang (1996) argue many published grammarslack high-level reasoning during the iteration processsince they are often concerned with the developmentof shape grammar theories rather than their practicalapplications By contrast SGEvac being a practicalapplication has a complete set of primitive shapesonly possible to arrange in a certain way It thusprovides viable reference design layout compositionsbased on SPSG with further interventions possible bythe designer

SYSTEM DEVELOPMENT TO DATEThe SGEvac prototype is compiled within the Clanguage environment allowing for modular object-oriented programming (suitable for shape grammar)Figure 10 shows the main interface where the stationname and relevant LU fare zone are entered (whichimpacts on sizing infrastructure) as are the ticket halland platform depths the overall site boundaries andfinal exitsrsquo positions The peak passenger flow figuresover the relevant time interval of normally 3 hours arealso specified SGEvac is then capable of accuratelycalculating all passenger area infrastructure based onthe SPSG guidelines as already discussed

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

110 G SMITH AND B CERANIC

TABLE 5 Station manager suite ndash shape rule connectionsELEMENT SHAPE RULE DESIGNATION CONNECTED COMPONENT GROUP ALLOCATION

Group station managerrsquos office SR_GSM_01 Ticket hall paid side GSM

SR_GSM_02 Ticket hall unpaid side

SR_GSM_03 GSM administration

SR_GSM_04 Duty station managerrsquos office

SR_GSM_05 Business managerrsquos office

SR_GSM_06 Administration

SR_GSM_07 Ticket office

SR_GSM_08 Police room

SR_GSM_09 Revenue protection examiner

SR_GSM_10 Interview room

140

Stair Stair

100 2020

4533

34

45

FIGURE 9 Platform escape distance SGEvac label restrictions

The history of auto-generated layouts is accessiblefrom this screen and the user is given the capabilityto delete and copy them (useful when minoralterations are required to produce an alternativelayout)

Shape labelling in the programme is multi-faceted which is necessary since many stationcomponents have various possible connections Anescalator for example may connect to a passagetravelator ramp escalator run-off stair run-off lift orthe ticket hall so all such connections need adding tothe component properties The SGEvac grouping ofcomponent interactions is accessed from menutoolbars that allow insertion of legislative compliantstation area connections as shown in Figure 11

Once the position and the depth of the ticket halland platform are specified the railway track is inserteddependent on the platform configuration The platformis the only possible connection to the track but thetrack can be positioned either side of platforms so thata side or centre (island) platform layout is possible

Non-passenger and ancillary areas are alsoselectable from the shape grammar library via property

windows for each station component eg the ticketoffice properties window as shown in Figure 12

Figure 13 shows examples of the layout modulestation components available in the SGEvacprogramme

Figure 14 shows the layout module interfaceInformation on the components (relative position inthe overall site layout and its absolute sizing) isdisplayed at the bottom of the screen once thecomponent is selected from the list In addition ameasurement bar (in metres) at the base of theinteractive design environment provides a sense ofscale for the designer

A checklist summarizes the station areas includedin the layout but a more detailed analysis is possiblevia a report function if required In addition to thisdetailed list a summary screen accessed from themain interface collates all specified components ineach generated layout This screen lists all passengernon-passenger and ancillary areas in their respectivecategories with infrastructure width data ofpassenger areas also summarized It also has anevacuation summary where the train frequency is

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 111

FIGURE 10 SGEvac main interface

specified and the train evacuation load (taken as peaklsquorush hourrsquo loading) is displayed This section of thesummary screen is useful for input into Simulexwhere fire evacuation test runs of the building layoutcan be implemented

The lsquoshow hierarchyrsquo function in Figure 15 allowsdesign team members to view the progression ofdesigns by superimposing connecting lines betweenstation elements in the order of their compositionThis also serves as a safeguard against deleting

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

112 G SMITH AND B CERANIC

FIGURE 12 SGEvac ticket office properties window

FIGURE 11 SGEvac properties window

components out of sequence that could lead to aninvalid solution

Once auto-generated layouts are complete withcorrectly sized and legitimately connected stationcomponents manual insertion of non-passenger areaswould then commence Upon completion thegenerated layout is transferred to AutoCAD via dxf fileformat as shown in Figure 16 Once in AutoCAD thelayout has a level of detail added by the designer suchas door openings passage connections fire escapeopenings etc so that passenger flows may beassessed It is then exported to Simulex for fireevacuation simulation analysis again via dxf file format

Simulex understands any voids as potentialcirculation routes treating the rest as solid andpermanent obstructions ie walls internal structuralelements permanent fixtures etc Since its solutionoperates in a two-dimensional environment for whatin effect is a three-dimensional problem the means

of vertical circulation (eg staircases) have to beestablished True lengths of staircases must bespecified based on the total length of all flights(sloping length) and the intermediate rests The fireexits need to be defined both in terms of theirlocation and width The links have to be establishedto define the point of the connection between thefloors and staircasesescalatorsramps The travellingdistance maps are then calculated type and numberof occupants specified building populated and theevacuation simulation performed

SYSTEM TESTING AND VALIDATIONKuligowski and Peacock (2005) define and recommendthe following ways of validating evacuation models

validation against code requirements validation against fire drills or other peoplersquos

movement experimentstrials

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 113

Gateline Vertical

Platform Right

Stairs Horizontal Stairs Vertical

Travelator Vertical

Ticket Hall

BMOAdmin Reception

Platform Top Ram Up Bottom Vertical

Lift Passage Horizontal

FIGURE 13 Examples of SGEvac station components

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

114 G SMITH AND B CERANIC

FIGURE 15 SGEvac hierarchy function

FIGURE 14 Layout module graphical user interface

validation against literature on past evacuationexperiments

validation against other models third-party validation

Furthermore Galea (1999) deconstructs the validationprocess into four main areas ndash component testingfunctional testing qualitative validation andquantitative validation This approach is more suitablefor models such as SGEvac than for those based oncompliance with code requirements and hence it isthe one adopted for the validation in this research

Component testing is regarded as testingcomputer code that drives the model checking forelementary errors Functional validation involveschecking that the model possesses the ability to

exhibit the range of capabilities required to performthe desired simulations Component testing andfunctional validation for SGEvac is carried outagainst code requirements insofar that allcalculations for infrastructure sizing are evaluatedmanually so that the method and answer are knownto be correct It uses the Simulex program to aid themodelling of people movement and to test thehypothesis of improvement to evacuation timesthus relying on Simulexrsquos own rigorous validationprocess Simulex has been used extensively inpractice as well as a validating model for otherevacuation softwares (Kuligowski and Peacock2005) SGEvac has gone through its initialcompletion and validation process with finalvalidation proposed for the end of the year

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 115

FIGURE 16 Auto-generated SGEvac station layout export file for fire

evacuation simulation in Simulex

FIRE EVACUATION SIMULATION RESULTSThe considered lsquotrain on fire in stationrsquo scenario takesinto account the service interval of 333 minutes(based on 18 trains per hour) the crush capacity of afour-car train together with the number of peoplewaiting on the platform after cancellation of oneservice (ie after 666 minutes) On the oppositeplatform the evacuation load is taken to be thenormal boardingalighting load assuming nocancellations to services (50 of all boarders andalighters)

Evacuation load predictions were based on TfL2006 projected passenger flows with an averageflow of 36 peopleminute for boarders and 123peopleminute for boarders and alighters combinedThe train crush capacity was averaged at 165peoplecar The simulation also models evacuationfrom the ticket hall level to assess the impact ofdispersing crowds at peak times in this part of thestation From these figures the total evacuation loadwas calculated as follows

Platform 1 Boarders (36 x 666 min) = 240Train crush load (165 x 4) = 660

Platform 2 Boarders and alighters

(123 x 05 x 333 min) = 205Ticket hall = 470Total station evacuation load = 1575 people

The simulation results have been captured as imagesat 25 50 100 and 300 second intervals respectivelyfor both platform and ticket hall levels as shown inFigures 17 to 20

The provided means of escape follow LU SPSG4requirements insofar that the maximum travellingdistance on a platform to an exit is no more than 45 m (ie exits must be no more than 90 m apart) andthat the dead-end travelling distance is no more than20 m to an exit Hence each platform is provided withthree exits The MIPsrsquo (mobility impaired persons)means of escape is provided from both ends ofplatforms via emergency evacuation lifts Totalevacuation time (ie all people reaching an exit onticket hallground level) was 5 minutes and

24 seconds with the platform and train load clearedwithin 3 minutes and 5 seconds This is within therequirements to evacuate the platform and train loadwithin 4 minutes with all people reaching a fire-separated route within 6 minutes (the ticket hall is notclassed as a fire-separated route)

The performed simulation was based on theminimum travelling distance maps resulting in eachperson exiting via the nearest fire exit route Theaverage 10 second lsquotime to start to moversquo wasassigned to all occupants in order to accommodatefor real world people behaviour commonly reportedby fire evacuation research studies The boarders onthe end of both platforms (including most peoplefrom end carriages of the train) evacuated viastaircasesMIP lifts directly to the ground level pointof escape The boarders situated around the middleof platforms together with the people from the twomiddle carriages used the central escalators to reachticket hall fire exits Additionally some people fromthe end carriages and some staffretail unitcustomers used this route as means of escape

In this lsquotrain on fire in stationrsquo scenario it wasassumed that none of the escapes routes areunavailable due to maintenance However furtherevacuation studies will investigate possible scenarioswhere one of the escalator routes may be unavailabledue to maintenance reasons

FUTURE RESEARCHIntended future work will concentrate on the furtherconditioning developing and modelling of shapegrammar design rules based on the outcomes of thefinal system validation and LU expertsrsquo feedbackFurther research is proposed on the development ofa novel layout optimization approach based on theshape grammar rules to investigate the evacuation-related parametric improvements in subsurfacesurface railway station design (eg minimizingtravelling distances)

The fire safety strategy in terms of performance-based fire risk design is an important aspirationalaspect of SGEvac requiring further research andliaison with experts to determine how such variedcriteria could be allied with layout planning This isparticularly important where deviation from standardsoccurs Thus further development of the expert

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

116 G SMITH AND B CERANIC

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 117

FIGURE 17 Station evacuation after 25 seconds

FIGURE 18 Station evacuation after 50 seconds

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

118 G SMITH AND B CERANIC

FIGURE 19 Station evacuation after 100 seconds

FIGURE 20 Station evacuation after 300 seconds

knowledge database would be required particularlyin the light of frequent changes to legislation

On a more practical note it could be beneficial todevelop the system on a stand-alone modular basislinked to each generated station component A list offire safety strategy objectives and functional logicstatements can then be specified for commonlyrecurring items (such as emergency lighting fire alarmdetectors sprinkler heads etc) and spaced accordingto the designerrsquos wishes or via shape rule formulation

Site constraint issues would also benefit fromfuture development as well as a move towardsdevelopment of the 3D shape grammar version Oncedevised other relevant 3D modelling systems such ascomputational fluid dynamics (CFD) smoke movementsoftware could be utilized in conjunction with SGEvac

CONCLUSIONSThe computer-aided modelling environment underdevelopment is intended to act as a layout planning aidfor those who practice in sub-surfacesurface railwaystation design The system under developmentcarefully considers London Underground StationPlanning Standards and Guidelines together withother relevant fire safety regulations that influenceoverall fire strategy design The proposed computer-aided design environment (SGEvac) is based on highlyrestrictive additive shape grammar rules driven by akernel of key passenger flow information to generatestation plan solutions

The major benefit to the design team is the ability toreview multiple layouts of accurately sized and correctlyplaced infrastructure at the reference or preliminarystage of the station layout planning design process Timeis hence saved regarding different option appraisals andquantitative testing using evacuation software shouldlead to a choice of more effective layouts in relation tomeans of escape travelling distances and improvementin the total evacuation times Modifications to layoutsshould be straightforward and there is scope toefficiently assess travelling distances and evacuationtimes for each of the considered alternatives comparedwith a lsquotraditionalrsquo approach

The proposed novel shape grammar approach ismost suited to new-build stations or largerefurbishment projects where more design flexibilityexists Sites with many site-specific constraints have

an array of predetermined parameters and will mostlikely benefit from the manual operation of SGEvac

Use in other rail systems due to recurringinfrastructure components and inherent functionallogic related to the laying out of the stationcomponents is also feasible provided that shapeconnections and shape rules are corrected to reflectthat countryrsquos legislative requirements

The potential for station layout improvementswith regard to emergency evacuation is an area thatmerits research especially in scenarios that concernthe safety of the general public and form a vital partof Londonrsquos transportation system Potential layoutimprovements could not only benefit evacuationsituations but also enhance the through-flow ofpeople during peak travelling periods

ACKNOWLEDGEMENTS The authors wish to express their gratitude to JacobsEngineering (UK) Ltd which has sponsored theresearch and to London Underground Ltd forproviding the design guidance that is reproduced inthis paper and for authorising the use of the stationplans London Underground has not reviewed thispaper and the content is solely the authorsrsquo work Mr Robert Hutchinson is also thanked for hiscomputer programming assistance

AUTHOR CONTACT DETAILSGGrraahhaamm SSmmiitthh (corresponding author) Senior Architectural

Technologist Jacobs (UK) Ltd Cardinal Square North Point

10 Nottingham Road Derby DE1 3QT UK Tel +44 (0)1332 234482

fax +44 (0)1332 344558 e-mail grahampsmithjacobscomBBoorriiss CCeerraanniicc Programme Leader for BScBA (Hons) Architectural

Programmes Room Number 105 Faculty of Arts Design and

Technology University of Derby Markeaton Street Derby DE22

3AW UK Tel +44 (0)1332 593136 e-mail bceranicderbyacuk

REFERENCES BS 7974 2001 Application of Fire Safety Engineering Principles to the

Design of Buildings London BSI

Cha MY and Gero JS 1999 Style Learning Inductive Generalisation of

Architectural Shape Patterns Paichai University and Sydney University

co-paper

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 119

Chase SC 1996 lsquoUsing logic to specify shapes and spatial relations in

design grammarsrsquo in Workshop notes Grammatical Design Fourth

International Conference on Artificial Intelligence in Design Stanford

University US June

Department of the Environment 1996 Design Principles of Fire Safety

London The Stationery Office Books

Flemming U 1987 lsquoMore than the sum of its parts the grammar of

Queen Anne housesrsquo in Environment and Planning B Planning and

Design 14 323ndash350

Fruin JJ 1971 Pedestrian Planning and Design New York Metropolitan

Association of Urban Designers and Environmental Planners

Galea ER 1999 lsquoA general approach to validating evacuation models with

an application to EXODUSrsquo in Journal of Fire Sciences 16 414ndash436

Galle P and Kovacs LB 1992 lsquoThe logic of worms a study in

architectural knowledge representationrsquo in Environment and Planning B

Planning and Design 19 5ndash31

Garling T Lindberg E and Mantyla T 1983 lsquoOrientation in buildings ndash

effects of familiarity visual access and orientation aidsrsquo in Journal of

Applied Psychology 68(1) 177ndash186

Grimsdale RL and Chang CW 1996 lsquoThe layout design language A

technique for generating layout plansrsquo in Computer Graphics Forum

15(2) 97ndash106

Heskestad AW 1999 lsquoPerformance in smoke of way guidance systemsrsquo

in Fire and Materials 23(6) 375ndash381

Knight T 1995 lsquoConstructive symmetryrsquo in Environment and Planning B

Planning and Design 22 419ndash450

Knight TW 1999 lsquoShape grammars six typesrsquo in Environment and

Planning B Planning and Design 26 15ndash31

Kuligowski ED and Peacock RD 2005 A Review of Building Evacuation

Models Technical Note 1471 National Institute of Standards and

Technology (NIST) Technology Administration US Department of

Commerce

Li A I-Kang 2003 lsquoThe Yingzao fashi in the information agersquo conference

paper for the Beaux-Arts Paul Philippe Cret and Twentieth Century

Architecture in China The University of Pennsylvania October

Liew H 2003 lsquoSGML ndash A shape grammar meta-languagersquo in Digital

Design 21st eCAADe Conference Proceedings Graz Austria September

639ndash648

Loomis B 2003 lsquoA note on generative design techniques SGGA a user

driven genetic algorithm for evolving non-deterministic shape grammarsrsquo

working paper Massachusetts Institute of Technology (MIT)

Magnusson SE Drysdale DD Fitzgerald RW Mowrer F

Quintiere JG Williamson RB and Zalosh RG 1995

lsquoA proposal for a model curriculum in fire safety engineeringrsquo

in Fire Safety Journal 25 9ndash10

Smith G Ceranic B and Fryer C 2003 lsquoDevelopment of CAD

environment for spatial layout optimisation in sub-surfacesurface rail

station designrsquo in Proceedings of International BIAT Conference The

British Museum London January 2003

SPSG 2007 London Underground Station Planning Standards and

Guidelines SPSG4 (1ndash371) London

Stiny G and Gips J 1971 lsquoShape grammars and the generative

specification of painting and sculpturersquo in Information Processing 71

1460ndash1465

Stiny G 1977 lsquoIce-ray a note on the generation of Chinese lattice

designsrsquo in Environment and Planning B Planning and Design 4 89ndash98

Stiny G 1980 lsquoKindergarten grammars designing with Froebelrsquos building

giftsrsquo in Environment and Planning B 3 409ndash462

Tapia M 1999 lsquoA visual implementation of a shape grammar systemrsquo in

Environment and Planning B Planning and Design 26(1) 59ndash73

Transport for London 2007 Tube Carries One Billion Passengers for First

Time wwwtflgovukcorporatemedianewscentre4770aspx

(accessed 24 March 2008)

Weller M 2007 London Underground private correspondence

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

120 G SMITH AND B CERANIC

routes and having delayed movement in response to afire alarm signal) The design team then assesses theresults against the qualitative fire safety design goalsset earlier to establish whether improvements toevacuation times have been made

DETAILED ANALYSIS OF STATIONCOMPONENTSIn conjunction with the site boundary final exitpositions and passenger flow specificationconsideration must also be given to the vertical sitedimension The shape grammar system assumes allfinal exits are at ground level with the ticket hall andplatform(s) being the other two elements that requirethe user to determine their depth relative to the finalexits In so doing SGEvac selects the correctinterchange components by an event-drivenmechanism that evaluates the height change andpassenger volume demand Thresholds for bothcriteria have been defined as shown in Table 2 andEquations 2ndash5

If H (height in mm) = 0 ndash 500 then I (infrastructure) = Ramp (2)

If H = 500 ndash 3000 and P (passengers) le 3000 then I = Stairway (3)

If H = 3000 ndash 5000 And P le 3000 then I = StairwayElse I = Escalator (4)

If H gt 5000 and P le 1000 then I = StairwayElse I = Escalator and Lift (5)

(where lsquoElsersquo denotes the mechanism for the choiceof the alternative infrastructure component if theoriginal condition is not satisfied)

The above constructs also condition the run-offdimensions associated with each of theinterchange components (between 4 m and 12 mas stipulated by SPSG) This being the case run-offs are used to size the ticket hall length Figure 7shows the paid side of the ticket hall with variabledistances shown between the escalator or staircaseand the gateline

The paid side of the ticket hall is always dictatedby run-off distance while the unpaid side is governedby two parameters the run-off and the concoursearea required for the heaviest 15-minute passengerflow If the maximum 15-minute passenger flowfigure multiplied by 1 m2 (say answer A) is larger inplan area than the product of the gateline widthmultiplied by the relevant run-off (say answer B) thenanswer A will be incorporated into the layout Thereverse is true if B is the greater number This ensuresthat the station is always sized appropriately

Figure 8 shows a ticket hall layout with thegateline dictating the width (denoted as Tick_H_wd)itself conditioned by the passenger flow figurescalculated as shown in Table 3 The table onlyevaluates entry flow requirements but SGEvac willrepeat the same process for exit gates to produce acorrectly sized gateline

In this case the maximum 15-minute passengerflow capacity size prevailed over the run-off to createthe unpaid side overall

The divide between the paid and unpaid side ofthe ticket hall is the sized according to Table 4

SGEvac shape rules mandate that entry gates arealways located adjacent to the ticket office queuingspace (designated as Q_TIW) to avoid conflictingpassenger entry and exit flows The LU standard is to

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 107

TABLE 2 Interchange infrastructure selectionINFRASTRUCTURE REASON FOR SELECTION

Ramp Vertical rise between 0ndash05 m preferred to steps due to disabled access issues generally considered

people will travel faster on ramps than up or down steps

Stairs Vertical rise between 05ndash5 m but greater height available for selection if station has a light passenger flow

Escalator Vertical rise is minimum of 3 m for heavy passenger flow stations but used for lightmedium flow stations

above 5 m

Lift Vertical rise above 5 m also required for accessibility of disabled persons Lifts always used with escalators

not as a substitute

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

108 G SMITH AND B CERANIC

Con_Area

Exit

UTS gates

6601900 2450

2400

EFW

Ticketoffice

POMcounter

4000

Tick_H_wd

1000

Run

-off

Entry

Q_space

Q_TIW

FIGURE 8 SGEvac ticket hall arrangement

Heavy flow escalator

Escalator

Staircase

Medium to Heavy flow

Medium to Heavy flow

Light flow escalator

Light flow

Heavy flow 12mMedium flow 8ndash12m

Light flow

Light flow 8m

Light flow 6m

Heavy flow

staircase to gateline

staircase to gateline 10m

Light flow

to gateline12mGateline

Gateline

EscalatorndashGateline

Total Run-off

Heavy flow 10mMedium flow 6ndash10m

StaircasendashGateline

Total Run-off

to gateline 8m

2m 4m 6m 8m 10m

2m 4m 6m

6m

8m 10m

12m

FIGURE 7 Passenger flow rate run-off requirements (SPSG 2007)

allow for 4 m (or a five-person queue) as shown inFigure 8 The overall gateline is defined by summingboth exit and entry gate numbers obtained from user-inputted passenger figures

The length of the ticket office interface with theticket hall is stipulated by the number of ticket issuingwindows (TIWs) plus excess fare windows (EFWs)TIWs always belong on the unpaid side of the tickethall and EFWs always on the paid side both forrevenue protection reasons This situation facilitates aprecise shape grammar labelling for the ticket officeto ticket hallqueue space connection

Passenger operated machine (POM) enclosures arenormally connected to the ticket hall ticket office andits own queue space (noted as Q_space in Figure 8)

To make shape rule specification manageable allstation areas are divided into groups arranged to suitthe most probable connection patterns in the stationTable 5 shows the group station managerrsquos officeexample with all other station components followingthe same concept

London Underground views its stations as largebasements (Weller 2007) and escape distances fromplatforms are based on agreed dispensations with theLondon Fire Brigade SGEvac respects thesedistances by placing restrictive labels on the platformso that they cannot be exceeded as demonstrated bythe example shown in Figure 9

SGEvac escape stair infrastructure may bepositioned up to 20 m from the end of platforms toavoid undesirable lsquodead endrsquo conditions The 45 mmaximum platform travel distance starts from thenearside junction of the two 33 m wide escapestaircases with the 45 m distance rule creating ashortfall in the platform centre of 34 m Generallylocating three interchange components (eg stairs

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 109

TABLE 3 Gateline width calculationMETHOD WORKED EXAMPLE

Obtain peak hourly flow rate through Say 6500 peoplehr

ticket hall

Calculate 5-minute peak passenger flow 650012 = 541

Allowance for ticket rejection 541 + 20 = 561

Allowance for growth in use demand 561 x 110 = 617

Assume 125 people use each gate over

the 5-minute peak passenger flow period 617125 = 493

Round up gates to nearest whole number 5 gates required

TABLE 4 London Underground ticket hall gateline parametersregulations (SPSG 2007)EQUIPMENT DIMENSIONS (W X D) POSITIONARRANGEMENT

Combined manualequipment gate 1750 x 270 mm Height ndash 1300 mm Adjacent to ticket office Can include

1400 mm clear opening luggage facility

Manual gate ndash passenger 1750 x 270 mm Height ndash 1300 mm Adjacent to ticket office With more than 6

gateluggage port 900 mm ndash clearance and walkways in a single gateline or significant

outward opening interchange with mainline stations airport or

tourist traffic provide separate luggage chutes

Manual gate ndash single leaf 1394 mm Height ndash 1300 mm Beside the dedicated excess gates

with luggage port 695 mm clearance

Gateline attendance report 990 x 1240 mm Within the controlled zone but not necessarily

part of the gateline Usually close

Station control unit (SCU) 432 x 430 mm Should be installed within a secure enclosure

(which could be a ticket office) and also with

a clear view of the gates Controls and monitors

the station underground ticketing system

equipment

Manual emergency exit gates 1900 x 125 mm Adjacent to ticket office Add these gates rather

than adding extra UTS gates for evacuation

purposes

or escalators) at these distances will result incompliance with the Underground standards but itmay not always be possible (or practical for aninterchange situation) to procure such a layout

Lengths of station components are controlled bya range of criteria For the platform it is the length ofthe train vehicle for interchange components it isthe vertical height in the terms of the depth to theplatform level for passages it is the designated finalexit positions and the existing station areas alreadygenerated

Vector direction labels are underpinned by thepermissible union of compatible spaces two points ofcontact will be maintained so as not to restrict thewidth of connecting passenger areas This undueconstriction of passenger flow at the head of apassage also satisfies realistic construction practiceNon-passenger and ancillary area vector directions areas per project requirements and the successive areaadditions managed by the SGEvac rule groupings

To create alternative layouts SGEvac uses alsquorandomizersquo rule In this shape grammar rule initialstate infrastructure is generated as per guidelinesbut it can randomly change the location ofcomponent arrangements within the predefined

boundaries so that a number of possible alternativelayouts are produced The designs can then beinterrogated and ranked by their travelling distancesand against the qualitative functional logicstatements

The SGEvac design tool developed in this researchis based on a highly restrictive additive grammar ruleset due to SPSGrsquos prescriptive guidance and thenature of the layout planning problem As Grimsdaleand Chang (1996) argue many published grammarslack high-level reasoning during the iteration processsince they are often concerned with the developmentof shape grammar theories rather than their practicalapplications By contrast SGEvac being a practicalapplication has a complete set of primitive shapesonly possible to arrange in a certain way It thusprovides viable reference design layout compositionsbased on SPSG with further interventions possible bythe designer

SYSTEM DEVELOPMENT TO DATEThe SGEvac prototype is compiled within the Clanguage environment allowing for modular object-oriented programming (suitable for shape grammar)Figure 10 shows the main interface where the stationname and relevant LU fare zone are entered (whichimpacts on sizing infrastructure) as are the ticket halland platform depths the overall site boundaries andfinal exitsrsquo positions The peak passenger flow figuresover the relevant time interval of normally 3 hours arealso specified SGEvac is then capable of accuratelycalculating all passenger area infrastructure based onthe SPSG guidelines as already discussed

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

110 G SMITH AND B CERANIC

TABLE 5 Station manager suite ndash shape rule connectionsELEMENT SHAPE RULE DESIGNATION CONNECTED COMPONENT GROUP ALLOCATION

Group station managerrsquos office SR_GSM_01 Ticket hall paid side GSM

SR_GSM_02 Ticket hall unpaid side

SR_GSM_03 GSM administration

SR_GSM_04 Duty station managerrsquos office

SR_GSM_05 Business managerrsquos office

SR_GSM_06 Administration

SR_GSM_07 Ticket office

SR_GSM_08 Police room

SR_GSM_09 Revenue protection examiner

SR_GSM_10 Interview room

140

Stair Stair

100 2020

4533

34

45

FIGURE 9 Platform escape distance SGEvac label restrictions

The history of auto-generated layouts is accessiblefrom this screen and the user is given the capabilityto delete and copy them (useful when minoralterations are required to produce an alternativelayout)

Shape labelling in the programme is multi-faceted which is necessary since many stationcomponents have various possible connections Anescalator for example may connect to a passagetravelator ramp escalator run-off stair run-off lift orthe ticket hall so all such connections need adding tothe component properties The SGEvac grouping ofcomponent interactions is accessed from menutoolbars that allow insertion of legislative compliantstation area connections as shown in Figure 11

Once the position and the depth of the ticket halland platform are specified the railway track is inserteddependent on the platform configuration The platformis the only possible connection to the track but thetrack can be positioned either side of platforms so thata side or centre (island) platform layout is possible

Non-passenger and ancillary areas are alsoselectable from the shape grammar library via property

windows for each station component eg the ticketoffice properties window as shown in Figure 12

Figure 13 shows examples of the layout modulestation components available in the SGEvacprogramme

Figure 14 shows the layout module interfaceInformation on the components (relative position inthe overall site layout and its absolute sizing) isdisplayed at the bottom of the screen once thecomponent is selected from the list In addition ameasurement bar (in metres) at the base of theinteractive design environment provides a sense ofscale for the designer

A checklist summarizes the station areas includedin the layout but a more detailed analysis is possiblevia a report function if required In addition to thisdetailed list a summary screen accessed from themain interface collates all specified components ineach generated layout This screen lists all passengernon-passenger and ancillary areas in their respectivecategories with infrastructure width data ofpassenger areas also summarized It also has anevacuation summary where the train frequency is

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 111

FIGURE 10 SGEvac main interface

specified and the train evacuation load (taken as peaklsquorush hourrsquo loading) is displayed This section of thesummary screen is useful for input into Simulexwhere fire evacuation test runs of the building layoutcan be implemented

The lsquoshow hierarchyrsquo function in Figure 15 allowsdesign team members to view the progression ofdesigns by superimposing connecting lines betweenstation elements in the order of their compositionThis also serves as a safeguard against deleting

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

112 G SMITH AND B CERANIC

FIGURE 12 SGEvac ticket office properties window

FIGURE 11 SGEvac properties window

components out of sequence that could lead to aninvalid solution

Once auto-generated layouts are complete withcorrectly sized and legitimately connected stationcomponents manual insertion of non-passenger areaswould then commence Upon completion thegenerated layout is transferred to AutoCAD via dxf fileformat as shown in Figure 16 Once in AutoCAD thelayout has a level of detail added by the designer suchas door openings passage connections fire escapeopenings etc so that passenger flows may beassessed It is then exported to Simulex for fireevacuation simulation analysis again via dxf file format

Simulex understands any voids as potentialcirculation routes treating the rest as solid andpermanent obstructions ie walls internal structuralelements permanent fixtures etc Since its solutionoperates in a two-dimensional environment for whatin effect is a three-dimensional problem the means

of vertical circulation (eg staircases) have to beestablished True lengths of staircases must bespecified based on the total length of all flights(sloping length) and the intermediate rests The fireexits need to be defined both in terms of theirlocation and width The links have to be establishedto define the point of the connection between thefloors and staircasesescalatorsramps The travellingdistance maps are then calculated type and numberof occupants specified building populated and theevacuation simulation performed

SYSTEM TESTING AND VALIDATIONKuligowski and Peacock (2005) define and recommendthe following ways of validating evacuation models

validation against code requirements validation against fire drills or other peoplersquos

movement experimentstrials

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 113

Gateline Vertical

Platform Right

Stairs Horizontal Stairs Vertical

Travelator Vertical

Ticket Hall

BMOAdmin Reception

Platform Top Ram Up Bottom Vertical

Lift Passage Horizontal

FIGURE 13 Examples of SGEvac station components

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

114 G SMITH AND B CERANIC

FIGURE 15 SGEvac hierarchy function

FIGURE 14 Layout module graphical user interface

validation against literature on past evacuationexperiments

validation against other models third-party validation

Furthermore Galea (1999) deconstructs the validationprocess into four main areas ndash component testingfunctional testing qualitative validation andquantitative validation This approach is more suitablefor models such as SGEvac than for those based oncompliance with code requirements and hence it isthe one adopted for the validation in this research

Component testing is regarded as testingcomputer code that drives the model checking forelementary errors Functional validation involveschecking that the model possesses the ability to

exhibit the range of capabilities required to performthe desired simulations Component testing andfunctional validation for SGEvac is carried outagainst code requirements insofar that allcalculations for infrastructure sizing are evaluatedmanually so that the method and answer are knownto be correct It uses the Simulex program to aid themodelling of people movement and to test thehypothesis of improvement to evacuation timesthus relying on Simulexrsquos own rigorous validationprocess Simulex has been used extensively inpractice as well as a validating model for otherevacuation softwares (Kuligowski and Peacock2005) SGEvac has gone through its initialcompletion and validation process with finalvalidation proposed for the end of the year

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 115

FIGURE 16 Auto-generated SGEvac station layout export file for fire

evacuation simulation in Simulex

FIRE EVACUATION SIMULATION RESULTSThe considered lsquotrain on fire in stationrsquo scenario takesinto account the service interval of 333 minutes(based on 18 trains per hour) the crush capacity of afour-car train together with the number of peoplewaiting on the platform after cancellation of oneservice (ie after 666 minutes) On the oppositeplatform the evacuation load is taken to be thenormal boardingalighting load assuming nocancellations to services (50 of all boarders andalighters)

Evacuation load predictions were based on TfL2006 projected passenger flows with an averageflow of 36 peopleminute for boarders and 123peopleminute for boarders and alighters combinedThe train crush capacity was averaged at 165peoplecar The simulation also models evacuationfrom the ticket hall level to assess the impact ofdispersing crowds at peak times in this part of thestation From these figures the total evacuation loadwas calculated as follows

Platform 1 Boarders (36 x 666 min) = 240Train crush load (165 x 4) = 660

Platform 2 Boarders and alighters

(123 x 05 x 333 min) = 205Ticket hall = 470Total station evacuation load = 1575 people

The simulation results have been captured as imagesat 25 50 100 and 300 second intervals respectivelyfor both platform and ticket hall levels as shown inFigures 17 to 20

The provided means of escape follow LU SPSG4requirements insofar that the maximum travellingdistance on a platform to an exit is no more than 45 m (ie exits must be no more than 90 m apart) andthat the dead-end travelling distance is no more than20 m to an exit Hence each platform is provided withthree exits The MIPsrsquo (mobility impaired persons)means of escape is provided from both ends ofplatforms via emergency evacuation lifts Totalevacuation time (ie all people reaching an exit onticket hallground level) was 5 minutes and

24 seconds with the platform and train load clearedwithin 3 minutes and 5 seconds This is within therequirements to evacuate the platform and train loadwithin 4 minutes with all people reaching a fire-separated route within 6 minutes (the ticket hall is notclassed as a fire-separated route)

The performed simulation was based on theminimum travelling distance maps resulting in eachperson exiting via the nearest fire exit route Theaverage 10 second lsquotime to start to moversquo wasassigned to all occupants in order to accommodatefor real world people behaviour commonly reportedby fire evacuation research studies The boarders onthe end of both platforms (including most peoplefrom end carriages of the train) evacuated viastaircasesMIP lifts directly to the ground level pointof escape The boarders situated around the middleof platforms together with the people from the twomiddle carriages used the central escalators to reachticket hall fire exits Additionally some people fromthe end carriages and some staffretail unitcustomers used this route as means of escape

In this lsquotrain on fire in stationrsquo scenario it wasassumed that none of the escapes routes areunavailable due to maintenance However furtherevacuation studies will investigate possible scenarioswhere one of the escalator routes may be unavailabledue to maintenance reasons

FUTURE RESEARCHIntended future work will concentrate on the furtherconditioning developing and modelling of shapegrammar design rules based on the outcomes of thefinal system validation and LU expertsrsquo feedbackFurther research is proposed on the development ofa novel layout optimization approach based on theshape grammar rules to investigate the evacuation-related parametric improvements in subsurfacesurface railway station design (eg minimizingtravelling distances)

The fire safety strategy in terms of performance-based fire risk design is an important aspirationalaspect of SGEvac requiring further research andliaison with experts to determine how such variedcriteria could be allied with layout planning This isparticularly important where deviation from standardsoccurs Thus further development of the expert

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

116 G SMITH AND B CERANIC

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 117

FIGURE 17 Station evacuation after 25 seconds

FIGURE 18 Station evacuation after 50 seconds

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

118 G SMITH AND B CERANIC

FIGURE 19 Station evacuation after 100 seconds

FIGURE 20 Station evacuation after 300 seconds

knowledge database would be required particularlyin the light of frequent changes to legislation

On a more practical note it could be beneficial todevelop the system on a stand-alone modular basislinked to each generated station component A list offire safety strategy objectives and functional logicstatements can then be specified for commonlyrecurring items (such as emergency lighting fire alarmdetectors sprinkler heads etc) and spaced accordingto the designerrsquos wishes or via shape rule formulation

Site constraint issues would also benefit fromfuture development as well as a move towardsdevelopment of the 3D shape grammar version Oncedevised other relevant 3D modelling systems such ascomputational fluid dynamics (CFD) smoke movementsoftware could be utilized in conjunction with SGEvac

CONCLUSIONSThe computer-aided modelling environment underdevelopment is intended to act as a layout planning aidfor those who practice in sub-surfacesurface railwaystation design The system under developmentcarefully considers London Underground StationPlanning Standards and Guidelines together withother relevant fire safety regulations that influenceoverall fire strategy design The proposed computer-aided design environment (SGEvac) is based on highlyrestrictive additive shape grammar rules driven by akernel of key passenger flow information to generatestation plan solutions

The major benefit to the design team is the ability toreview multiple layouts of accurately sized and correctlyplaced infrastructure at the reference or preliminarystage of the station layout planning design process Timeis hence saved regarding different option appraisals andquantitative testing using evacuation software shouldlead to a choice of more effective layouts in relation tomeans of escape travelling distances and improvementin the total evacuation times Modifications to layoutsshould be straightforward and there is scope toefficiently assess travelling distances and evacuationtimes for each of the considered alternatives comparedwith a lsquotraditionalrsquo approach

The proposed novel shape grammar approach ismost suited to new-build stations or largerefurbishment projects where more design flexibilityexists Sites with many site-specific constraints have

an array of predetermined parameters and will mostlikely benefit from the manual operation of SGEvac

Use in other rail systems due to recurringinfrastructure components and inherent functionallogic related to the laying out of the stationcomponents is also feasible provided that shapeconnections and shape rules are corrected to reflectthat countryrsquos legislative requirements

The potential for station layout improvementswith regard to emergency evacuation is an area thatmerits research especially in scenarios that concernthe safety of the general public and form a vital partof Londonrsquos transportation system Potential layoutimprovements could not only benefit evacuationsituations but also enhance the through-flow ofpeople during peak travelling periods

ACKNOWLEDGEMENTS The authors wish to express their gratitude to JacobsEngineering (UK) Ltd which has sponsored theresearch and to London Underground Ltd forproviding the design guidance that is reproduced inthis paper and for authorising the use of the stationplans London Underground has not reviewed thispaper and the content is solely the authorsrsquo work Mr Robert Hutchinson is also thanked for hiscomputer programming assistance

AUTHOR CONTACT DETAILSGGrraahhaamm SSmmiitthh (corresponding author) Senior Architectural

Technologist Jacobs (UK) Ltd Cardinal Square North Point

10 Nottingham Road Derby DE1 3QT UK Tel +44 (0)1332 234482

fax +44 (0)1332 344558 e-mail grahampsmithjacobscomBBoorriiss CCeerraanniicc Programme Leader for BScBA (Hons) Architectural

Programmes Room Number 105 Faculty of Arts Design and

Technology University of Derby Markeaton Street Derby DE22

3AW UK Tel +44 (0)1332 593136 e-mail bceranicderbyacuk

REFERENCES BS 7974 2001 Application of Fire Safety Engineering Principles to the

Design of Buildings London BSI

Cha MY and Gero JS 1999 Style Learning Inductive Generalisation of

Architectural Shape Patterns Paichai University and Sydney University

co-paper

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 119

Chase SC 1996 lsquoUsing logic to specify shapes and spatial relations in

design grammarsrsquo in Workshop notes Grammatical Design Fourth

International Conference on Artificial Intelligence in Design Stanford

University US June

Department of the Environment 1996 Design Principles of Fire Safety

London The Stationery Office Books

Flemming U 1987 lsquoMore than the sum of its parts the grammar of

Queen Anne housesrsquo in Environment and Planning B Planning and

Design 14 323ndash350

Fruin JJ 1971 Pedestrian Planning and Design New York Metropolitan

Association of Urban Designers and Environmental Planners

Galea ER 1999 lsquoA general approach to validating evacuation models with

an application to EXODUSrsquo in Journal of Fire Sciences 16 414ndash436

Galle P and Kovacs LB 1992 lsquoThe logic of worms a study in

architectural knowledge representationrsquo in Environment and Planning B

Planning and Design 19 5ndash31

Garling T Lindberg E and Mantyla T 1983 lsquoOrientation in buildings ndash

effects of familiarity visual access and orientation aidsrsquo in Journal of

Applied Psychology 68(1) 177ndash186

Grimsdale RL and Chang CW 1996 lsquoThe layout design language A

technique for generating layout plansrsquo in Computer Graphics Forum

15(2) 97ndash106

Heskestad AW 1999 lsquoPerformance in smoke of way guidance systemsrsquo

in Fire and Materials 23(6) 375ndash381

Knight T 1995 lsquoConstructive symmetryrsquo in Environment and Planning B

Planning and Design 22 419ndash450

Knight TW 1999 lsquoShape grammars six typesrsquo in Environment and

Planning B Planning and Design 26 15ndash31

Kuligowski ED and Peacock RD 2005 A Review of Building Evacuation

Models Technical Note 1471 National Institute of Standards and

Technology (NIST) Technology Administration US Department of

Commerce

Li A I-Kang 2003 lsquoThe Yingzao fashi in the information agersquo conference

paper for the Beaux-Arts Paul Philippe Cret and Twentieth Century

Architecture in China The University of Pennsylvania October

Liew H 2003 lsquoSGML ndash A shape grammar meta-languagersquo in Digital

Design 21st eCAADe Conference Proceedings Graz Austria September

639ndash648

Loomis B 2003 lsquoA note on generative design techniques SGGA a user

driven genetic algorithm for evolving non-deterministic shape grammarsrsquo

working paper Massachusetts Institute of Technology (MIT)

Magnusson SE Drysdale DD Fitzgerald RW Mowrer F

Quintiere JG Williamson RB and Zalosh RG 1995

lsquoA proposal for a model curriculum in fire safety engineeringrsquo

in Fire Safety Journal 25 9ndash10

Smith G Ceranic B and Fryer C 2003 lsquoDevelopment of CAD

environment for spatial layout optimisation in sub-surfacesurface rail

station designrsquo in Proceedings of International BIAT Conference The

British Museum London January 2003

SPSG 2007 London Underground Station Planning Standards and

Guidelines SPSG4 (1ndash371) London

Stiny G and Gips J 1971 lsquoShape grammars and the generative

specification of painting and sculpturersquo in Information Processing 71

1460ndash1465

Stiny G 1977 lsquoIce-ray a note on the generation of Chinese lattice

designsrsquo in Environment and Planning B Planning and Design 4 89ndash98

Stiny G 1980 lsquoKindergarten grammars designing with Froebelrsquos building

giftsrsquo in Environment and Planning B 3 409ndash462

Tapia M 1999 lsquoA visual implementation of a shape grammar systemrsquo in

Environment and Planning B Planning and Design 26(1) 59ndash73

Transport for London 2007 Tube Carries One Billion Passengers for First

Time wwwtflgovukcorporatemedianewscentre4770aspx

(accessed 24 March 2008)

Weller M 2007 London Underground private correspondence

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

120 G SMITH AND B CERANIC

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

108 G SMITH AND B CERANIC

Con_Area

Exit

UTS gates

6601900 2450

2400

EFW

Ticketoffice

POMcounter

4000

Tick_H_wd

1000

Run

-off

Entry

Q_space

Q_TIW

FIGURE 8 SGEvac ticket hall arrangement

Heavy flow escalator

Escalator

Staircase

Medium to Heavy flow

Medium to Heavy flow

Light flow escalator

Light flow

Heavy flow 12mMedium flow 8ndash12m

Light flow

Light flow 8m

Light flow 6m

Heavy flow

staircase to gateline

staircase to gateline 10m

Light flow

to gateline12mGateline

Gateline

EscalatorndashGateline

Total Run-off

Heavy flow 10mMedium flow 6ndash10m

StaircasendashGateline

Total Run-off

to gateline 8m

2m 4m 6m 8m 10m

2m 4m 6m

6m

8m 10m

12m

FIGURE 7 Passenger flow rate run-off requirements (SPSG 2007)

allow for 4 m (or a five-person queue) as shown inFigure 8 The overall gateline is defined by summingboth exit and entry gate numbers obtained from user-inputted passenger figures

The length of the ticket office interface with theticket hall is stipulated by the number of ticket issuingwindows (TIWs) plus excess fare windows (EFWs)TIWs always belong on the unpaid side of the tickethall and EFWs always on the paid side both forrevenue protection reasons This situation facilitates aprecise shape grammar labelling for the ticket officeto ticket hallqueue space connection

Passenger operated machine (POM) enclosures arenormally connected to the ticket hall ticket office andits own queue space (noted as Q_space in Figure 8)

To make shape rule specification manageable allstation areas are divided into groups arranged to suitthe most probable connection patterns in the stationTable 5 shows the group station managerrsquos officeexample with all other station components followingthe same concept

London Underground views its stations as largebasements (Weller 2007) and escape distances fromplatforms are based on agreed dispensations with theLondon Fire Brigade SGEvac respects thesedistances by placing restrictive labels on the platformso that they cannot be exceeded as demonstrated bythe example shown in Figure 9

SGEvac escape stair infrastructure may bepositioned up to 20 m from the end of platforms toavoid undesirable lsquodead endrsquo conditions The 45 mmaximum platform travel distance starts from thenearside junction of the two 33 m wide escapestaircases with the 45 m distance rule creating ashortfall in the platform centre of 34 m Generallylocating three interchange components (eg stairs

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 109

TABLE 3 Gateline width calculationMETHOD WORKED EXAMPLE

Obtain peak hourly flow rate through Say 6500 peoplehr

ticket hall

Calculate 5-minute peak passenger flow 650012 = 541

Allowance for ticket rejection 541 + 20 = 561

Allowance for growth in use demand 561 x 110 = 617

Assume 125 people use each gate over

the 5-minute peak passenger flow period 617125 = 493

Round up gates to nearest whole number 5 gates required

TABLE 4 London Underground ticket hall gateline parametersregulations (SPSG 2007)EQUIPMENT DIMENSIONS (W X D) POSITIONARRANGEMENT

Combined manualequipment gate 1750 x 270 mm Height ndash 1300 mm Adjacent to ticket office Can include

1400 mm clear opening luggage facility

Manual gate ndash passenger 1750 x 270 mm Height ndash 1300 mm Adjacent to ticket office With more than 6

gateluggage port 900 mm ndash clearance and walkways in a single gateline or significant

outward opening interchange with mainline stations airport or

tourist traffic provide separate luggage chutes

Manual gate ndash single leaf 1394 mm Height ndash 1300 mm Beside the dedicated excess gates

with luggage port 695 mm clearance

Gateline attendance report 990 x 1240 mm Within the controlled zone but not necessarily

part of the gateline Usually close

Station control unit (SCU) 432 x 430 mm Should be installed within a secure enclosure

(which could be a ticket office) and also with

a clear view of the gates Controls and monitors

the station underground ticketing system

equipment

Manual emergency exit gates 1900 x 125 mm Adjacent to ticket office Add these gates rather

than adding extra UTS gates for evacuation

purposes

or escalators) at these distances will result incompliance with the Underground standards but itmay not always be possible (or practical for aninterchange situation) to procure such a layout

Lengths of station components are controlled bya range of criteria For the platform it is the length ofthe train vehicle for interchange components it isthe vertical height in the terms of the depth to theplatform level for passages it is the designated finalexit positions and the existing station areas alreadygenerated

Vector direction labels are underpinned by thepermissible union of compatible spaces two points ofcontact will be maintained so as not to restrict thewidth of connecting passenger areas This undueconstriction of passenger flow at the head of apassage also satisfies realistic construction practiceNon-passenger and ancillary area vector directions areas per project requirements and the successive areaadditions managed by the SGEvac rule groupings

To create alternative layouts SGEvac uses alsquorandomizersquo rule In this shape grammar rule initialstate infrastructure is generated as per guidelinesbut it can randomly change the location ofcomponent arrangements within the predefined

boundaries so that a number of possible alternativelayouts are produced The designs can then beinterrogated and ranked by their travelling distancesand against the qualitative functional logicstatements

The SGEvac design tool developed in this researchis based on a highly restrictive additive grammar ruleset due to SPSGrsquos prescriptive guidance and thenature of the layout planning problem As Grimsdaleand Chang (1996) argue many published grammarslack high-level reasoning during the iteration processsince they are often concerned with the developmentof shape grammar theories rather than their practicalapplications By contrast SGEvac being a practicalapplication has a complete set of primitive shapesonly possible to arrange in a certain way It thusprovides viable reference design layout compositionsbased on SPSG with further interventions possible bythe designer

SYSTEM DEVELOPMENT TO DATEThe SGEvac prototype is compiled within the Clanguage environment allowing for modular object-oriented programming (suitable for shape grammar)Figure 10 shows the main interface where the stationname and relevant LU fare zone are entered (whichimpacts on sizing infrastructure) as are the ticket halland platform depths the overall site boundaries andfinal exitsrsquo positions The peak passenger flow figuresover the relevant time interval of normally 3 hours arealso specified SGEvac is then capable of accuratelycalculating all passenger area infrastructure based onthe SPSG guidelines as already discussed

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

110 G SMITH AND B CERANIC

TABLE 5 Station manager suite ndash shape rule connectionsELEMENT SHAPE RULE DESIGNATION CONNECTED COMPONENT GROUP ALLOCATION

Group station managerrsquos office SR_GSM_01 Ticket hall paid side GSM

SR_GSM_02 Ticket hall unpaid side

SR_GSM_03 GSM administration

SR_GSM_04 Duty station managerrsquos office

SR_GSM_05 Business managerrsquos office

SR_GSM_06 Administration

SR_GSM_07 Ticket office

SR_GSM_08 Police room

SR_GSM_09 Revenue protection examiner

SR_GSM_10 Interview room

140

Stair Stair

100 2020

4533

34

45

FIGURE 9 Platform escape distance SGEvac label restrictions

The history of auto-generated layouts is accessiblefrom this screen and the user is given the capabilityto delete and copy them (useful when minoralterations are required to produce an alternativelayout)

Shape labelling in the programme is multi-faceted which is necessary since many stationcomponents have various possible connections Anescalator for example may connect to a passagetravelator ramp escalator run-off stair run-off lift orthe ticket hall so all such connections need adding tothe component properties The SGEvac grouping ofcomponent interactions is accessed from menutoolbars that allow insertion of legislative compliantstation area connections as shown in Figure 11

Once the position and the depth of the ticket halland platform are specified the railway track is inserteddependent on the platform configuration The platformis the only possible connection to the track but thetrack can be positioned either side of platforms so thata side or centre (island) platform layout is possible

Non-passenger and ancillary areas are alsoselectable from the shape grammar library via property

windows for each station component eg the ticketoffice properties window as shown in Figure 12

Figure 13 shows examples of the layout modulestation components available in the SGEvacprogramme

Figure 14 shows the layout module interfaceInformation on the components (relative position inthe overall site layout and its absolute sizing) isdisplayed at the bottom of the screen once thecomponent is selected from the list In addition ameasurement bar (in metres) at the base of theinteractive design environment provides a sense ofscale for the designer

A checklist summarizes the station areas includedin the layout but a more detailed analysis is possiblevia a report function if required In addition to thisdetailed list a summary screen accessed from themain interface collates all specified components ineach generated layout This screen lists all passengernon-passenger and ancillary areas in their respectivecategories with infrastructure width data ofpassenger areas also summarized It also has anevacuation summary where the train frequency is

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 111

FIGURE 10 SGEvac main interface

specified and the train evacuation load (taken as peaklsquorush hourrsquo loading) is displayed This section of thesummary screen is useful for input into Simulexwhere fire evacuation test runs of the building layoutcan be implemented

The lsquoshow hierarchyrsquo function in Figure 15 allowsdesign team members to view the progression ofdesigns by superimposing connecting lines betweenstation elements in the order of their compositionThis also serves as a safeguard against deleting

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

112 G SMITH AND B CERANIC

FIGURE 12 SGEvac ticket office properties window

FIGURE 11 SGEvac properties window

components out of sequence that could lead to aninvalid solution

Once auto-generated layouts are complete withcorrectly sized and legitimately connected stationcomponents manual insertion of non-passenger areaswould then commence Upon completion thegenerated layout is transferred to AutoCAD via dxf fileformat as shown in Figure 16 Once in AutoCAD thelayout has a level of detail added by the designer suchas door openings passage connections fire escapeopenings etc so that passenger flows may beassessed It is then exported to Simulex for fireevacuation simulation analysis again via dxf file format

Simulex understands any voids as potentialcirculation routes treating the rest as solid andpermanent obstructions ie walls internal structuralelements permanent fixtures etc Since its solutionoperates in a two-dimensional environment for whatin effect is a three-dimensional problem the means

of vertical circulation (eg staircases) have to beestablished True lengths of staircases must bespecified based on the total length of all flights(sloping length) and the intermediate rests The fireexits need to be defined both in terms of theirlocation and width The links have to be establishedto define the point of the connection between thefloors and staircasesescalatorsramps The travellingdistance maps are then calculated type and numberof occupants specified building populated and theevacuation simulation performed

SYSTEM TESTING AND VALIDATIONKuligowski and Peacock (2005) define and recommendthe following ways of validating evacuation models

validation against code requirements validation against fire drills or other peoplersquos

movement experimentstrials

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 113

Gateline Vertical

Platform Right

Stairs Horizontal Stairs Vertical

Travelator Vertical

Ticket Hall

BMOAdmin Reception

Platform Top Ram Up Bottom Vertical

Lift Passage Horizontal

FIGURE 13 Examples of SGEvac station components

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

114 G SMITH AND B CERANIC

FIGURE 15 SGEvac hierarchy function

FIGURE 14 Layout module graphical user interface

validation against literature on past evacuationexperiments

validation against other models third-party validation

Furthermore Galea (1999) deconstructs the validationprocess into four main areas ndash component testingfunctional testing qualitative validation andquantitative validation This approach is more suitablefor models such as SGEvac than for those based oncompliance with code requirements and hence it isthe one adopted for the validation in this research

Component testing is regarded as testingcomputer code that drives the model checking forelementary errors Functional validation involveschecking that the model possesses the ability to

exhibit the range of capabilities required to performthe desired simulations Component testing andfunctional validation for SGEvac is carried outagainst code requirements insofar that allcalculations for infrastructure sizing are evaluatedmanually so that the method and answer are knownto be correct It uses the Simulex program to aid themodelling of people movement and to test thehypothesis of improvement to evacuation timesthus relying on Simulexrsquos own rigorous validationprocess Simulex has been used extensively inpractice as well as a validating model for otherevacuation softwares (Kuligowski and Peacock2005) SGEvac has gone through its initialcompletion and validation process with finalvalidation proposed for the end of the year

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 115

FIGURE 16 Auto-generated SGEvac station layout export file for fire

evacuation simulation in Simulex

FIRE EVACUATION SIMULATION RESULTSThe considered lsquotrain on fire in stationrsquo scenario takesinto account the service interval of 333 minutes(based on 18 trains per hour) the crush capacity of afour-car train together with the number of peoplewaiting on the platform after cancellation of oneservice (ie after 666 minutes) On the oppositeplatform the evacuation load is taken to be thenormal boardingalighting load assuming nocancellations to services (50 of all boarders andalighters)

Evacuation load predictions were based on TfL2006 projected passenger flows with an averageflow of 36 peopleminute for boarders and 123peopleminute for boarders and alighters combinedThe train crush capacity was averaged at 165peoplecar The simulation also models evacuationfrom the ticket hall level to assess the impact ofdispersing crowds at peak times in this part of thestation From these figures the total evacuation loadwas calculated as follows

Platform 1 Boarders (36 x 666 min) = 240Train crush load (165 x 4) = 660

Platform 2 Boarders and alighters

(123 x 05 x 333 min) = 205Ticket hall = 470Total station evacuation load = 1575 people

The simulation results have been captured as imagesat 25 50 100 and 300 second intervals respectivelyfor both platform and ticket hall levels as shown inFigures 17 to 20

The provided means of escape follow LU SPSG4requirements insofar that the maximum travellingdistance on a platform to an exit is no more than 45 m (ie exits must be no more than 90 m apart) andthat the dead-end travelling distance is no more than20 m to an exit Hence each platform is provided withthree exits The MIPsrsquo (mobility impaired persons)means of escape is provided from both ends ofplatforms via emergency evacuation lifts Totalevacuation time (ie all people reaching an exit onticket hallground level) was 5 minutes and

24 seconds with the platform and train load clearedwithin 3 minutes and 5 seconds This is within therequirements to evacuate the platform and train loadwithin 4 minutes with all people reaching a fire-separated route within 6 minutes (the ticket hall is notclassed as a fire-separated route)

The performed simulation was based on theminimum travelling distance maps resulting in eachperson exiting via the nearest fire exit route Theaverage 10 second lsquotime to start to moversquo wasassigned to all occupants in order to accommodatefor real world people behaviour commonly reportedby fire evacuation research studies The boarders onthe end of both platforms (including most peoplefrom end carriages of the train) evacuated viastaircasesMIP lifts directly to the ground level pointof escape The boarders situated around the middleof platforms together with the people from the twomiddle carriages used the central escalators to reachticket hall fire exits Additionally some people fromthe end carriages and some staffretail unitcustomers used this route as means of escape

In this lsquotrain on fire in stationrsquo scenario it wasassumed that none of the escapes routes areunavailable due to maintenance However furtherevacuation studies will investigate possible scenarioswhere one of the escalator routes may be unavailabledue to maintenance reasons

FUTURE RESEARCHIntended future work will concentrate on the furtherconditioning developing and modelling of shapegrammar design rules based on the outcomes of thefinal system validation and LU expertsrsquo feedbackFurther research is proposed on the development ofa novel layout optimization approach based on theshape grammar rules to investigate the evacuation-related parametric improvements in subsurfacesurface railway station design (eg minimizingtravelling distances)

The fire safety strategy in terms of performance-based fire risk design is an important aspirationalaspect of SGEvac requiring further research andliaison with experts to determine how such variedcriteria could be allied with layout planning This isparticularly important where deviation from standardsoccurs Thus further development of the expert

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

116 G SMITH AND B CERANIC

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 117

FIGURE 17 Station evacuation after 25 seconds

FIGURE 18 Station evacuation after 50 seconds

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

118 G SMITH AND B CERANIC

FIGURE 19 Station evacuation after 100 seconds

FIGURE 20 Station evacuation after 300 seconds

knowledge database would be required particularlyin the light of frequent changes to legislation

On a more practical note it could be beneficial todevelop the system on a stand-alone modular basislinked to each generated station component A list offire safety strategy objectives and functional logicstatements can then be specified for commonlyrecurring items (such as emergency lighting fire alarmdetectors sprinkler heads etc) and spaced accordingto the designerrsquos wishes or via shape rule formulation

Site constraint issues would also benefit fromfuture development as well as a move towardsdevelopment of the 3D shape grammar version Oncedevised other relevant 3D modelling systems such ascomputational fluid dynamics (CFD) smoke movementsoftware could be utilized in conjunction with SGEvac

CONCLUSIONSThe computer-aided modelling environment underdevelopment is intended to act as a layout planning aidfor those who practice in sub-surfacesurface railwaystation design The system under developmentcarefully considers London Underground StationPlanning Standards and Guidelines together withother relevant fire safety regulations that influenceoverall fire strategy design The proposed computer-aided design environment (SGEvac) is based on highlyrestrictive additive shape grammar rules driven by akernel of key passenger flow information to generatestation plan solutions

The major benefit to the design team is the ability toreview multiple layouts of accurately sized and correctlyplaced infrastructure at the reference or preliminarystage of the station layout planning design process Timeis hence saved regarding different option appraisals andquantitative testing using evacuation software shouldlead to a choice of more effective layouts in relation tomeans of escape travelling distances and improvementin the total evacuation times Modifications to layoutsshould be straightforward and there is scope toefficiently assess travelling distances and evacuationtimes for each of the considered alternatives comparedwith a lsquotraditionalrsquo approach

The proposed novel shape grammar approach ismost suited to new-build stations or largerefurbishment projects where more design flexibilityexists Sites with many site-specific constraints have

an array of predetermined parameters and will mostlikely benefit from the manual operation of SGEvac

Use in other rail systems due to recurringinfrastructure components and inherent functionallogic related to the laying out of the stationcomponents is also feasible provided that shapeconnections and shape rules are corrected to reflectthat countryrsquos legislative requirements

The potential for station layout improvementswith regard to emergency evacuation is an area thatmerits research especially in scenarios that concernthe safety of the general public and form a vital partof Londonrsquos transportation system Potential layoutimprovements could not only benefit evacuationsituations but also enhance the through-flow ofpeople during peak travelling periods

ACKNOWLEDGEMENTS The authors wish to express their gratitude to JacobsEngineering (UK) Ltd which has sponsored theresearch and to London Underground Ltd forproviding the design guidance that is reproduced inthis paper and for authorising the use of the stationplans London Underground has not reviewed thispaper and the content is solely the authorsrsquo work Mr Robert Hutchinson is also thanked for hiscomputer programming assistance

AUTHOR CONTACT DETAILSGGrraahhaamm SSmmiitthh (corresponding author) Senior Architectural

Technologist Jacobs (UK) Ltd Cardinal Square North Point

10 Nottingham Road Derby DE1 3QT UK Tel +44 (0)1332 234482

fax +44 (0)1332 344558 e-mail grahampsmithjacobscomBBoorriiss CCeerraanniicc Programme Leader for BScBA (Hons) Architectural

Programmes Room Number 105 Faculty of Arts Design and

Technology University of Derby Markeaton Street Derby DE22

3AW UK Tel +44 (0)1332 593136 e-mail bceranicderbyacuk

REFERENCES BS 7974 2001 Application of Fire Safety Engineering Principles to the

Design of Buildings London BSI

Cha MY and Gero JS 1999 Style Learning Inductive Generalisation of

Architectural Shape Patterns Paichai University and Sydney University

co-paper

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 119

Chase SC 1996 lsquoUsing logic to specify shapes and spatial relations in

design grammarsrsquo in Workshop notes Grammatical Design Fourth

International Conference on Artificial Intelligence in Design Stanford

University US June

Department of the Environment 1996 Design Principles of Fire Safety

London The Stationery Office Books

Flemming U 1987 lsquoMore than the sum of its parts the grammar of

Queen Anne housesrsquo in Environment and Planning B Planning and

Design 14 323ndash350

Fruin JJ 1971 Pedestrian Planning and Design New York Metropolitan

Association of Urban Designers and Environmental Planners

Galea ER 1999 lsquoA general approach to validating evacuation models with

an application to EXODUSrsquo in Journal of Fire Sciences 16 414ndash436

Galle P and Kovacs LB 1992 lsquoThe logic of worms a study in

architectural knowledge representationrsquo in Environment and Planning B

Planning and Design 19 5ndash31

Garling T Lindberg E and Mantyla T 1983 lsquoOrientation in buildings ndash

effects of familiarity visual access and orientation aidsrsquo in Journal of

Applied Psychology 68(1) 177ndash186

Grimsdale RL and Chang CW 1996 lsquoThe layout design language A

technique for generating layout plansrsquo in Computer Graphics Forum

15(2) 97ndash106

Heskestad AW 1999 lsquoPerformance in smoke of way guidance systemsrsquo

in Fire and Materials 23(6) 375ndash381

Knight T 1995 lsquoConstructive symmetryrsquo in Environment and Planning B

Planning and Design 22 419ndash450

Knight TW 1999 lsquoShape grammars six typesrsquo in Environment and

Planning B Planning and Design 26 15ndash31

Kuligowski ED and Peacock RD 2005 A Review of Building Evacuation

Models Technical Note 1471 National Institute of Standards and

Technology (NIST) Technology Administration US Department of

Commerce

Li A I-Kang 2003 lsquoThe Yingzao fashi in the information agersquo conference

paper for the Beaux-Arts Paul Philippe Cret and Twentieth Century

Architecture in China The University of Pennsylvania October

Liew H 2003 lsquoSGML ndash A shape grammar meta-languagersquo in Digital

Design 21st eCAADe Conference Proceedings Graz Austria September

639ndash648

Loomis B 2003 lsquoA note on generative design techniques SGGA a user

driven genetic algorithm for evolving non-deterministic shape grammarsrsquo

working paper Massachusetts Institute of Technology (MIT)

Magnusson SE Drysdale DD Fitzgerald RW Mowrer F

Quintiere JG Williamson RB and Zalosh RG 1995

lsquoA proposal for a model curriculum in fire safety engineeringrsquo

in Fire Safety Journal 25 9ndash10

Smith G Ceranic B and Fryer C 2003 lsquoDevelopment of CAD

environment for spatial layout optimisation in sub-surfacesurface rail

station designrsquo in Proceedings of International BIAT Conference The

British Museum London January 2003

SPSG 2007 London Underground Station Planning Standards and

Guidelines SPSG4 (1ndash371) London

Stiny G and Gips J 1971 lsquoShape grammars and the generative

specification of painting and sculpturersquo in Information Processing 71

1460ndash1465

Stiny G 1977 lsquoIce-ray a note on the generation of Chinese lattice

designsrsquo in Environment and Planning B Planning and Design 4 89ndash98

Stiny G 1980 lsquoKindergarten grammars designing with Froebelrsquos building

giftsrsquo in Environment and Planning B 3 409ndash462

Tapia M 1999 lsquoA visual implementation of a shape grammar systemrsquo in

Environment and Planning B Planning and Design 26(1) 59ndash73

Transport for London 2007 Tube Carries One Billion Passengers for First

Time wwwtflgovukcorporatemedianewscentre4770aspx

(accessed 24 March 2008)

Weller M 2007 London Underground private correspondence

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

120 G SMITH AND B CERANIC

allow for 4 m (or a five-person queue) as shown inFigure 8 The overall gateline is defined by summingboth exit and entry gate numbers obtained from user-inputted passenger figures

The length of the ticket office interface with theticket hall is stipulated by the number of ticket issuingwindows (TIWs) plus excess fare windows (EFWs)TIWs always belong on the unpaid side of the tickethall and EFWs always on the paid side both forrevenue protection reasons This situation facilitates aprecise shape grammar labelling for the ticket officeto ticket hallqueue space connection

Passenger operated machine (POM) enclosures arenormally connected to the ticket hall ticket office andits own queue space (noted as Q_space in Figure 8)

To make shape rule specification manageable allstation areas are divided into groups arranged to suitthe most probable connection patterns in the stationTable 5 shows the group station managerrsquos officeexample with all other station components followingthe same concept

London Underground views its stations as largebasements (Weller 2007) and escape distances fromplatforms are based on agreed dispensations with theLondon Fire Brigade SGEvac respects thesedistances by placing restrictive labels on the platformso that they cannot be exceeded as demonstrated bythe example shown in Figure 9

SGEvac escape stair infrastructure may bepositioned up to 20 m from the end of platforms toavoid undesirable lsquodead endrsquo conditions The 45 mmaximum platform travel distance starts from thenearside junction of the two 33 m wide escapestaircases with the 45 m distance rule creating ashortfall in the platform centre of 34 m Generallylocating three interchange components (eg stairs

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 109

TABLE 3 Gateline width calculationMETHOD WORKED EXAMPLE

Obtain peak hourly flow rate through Say 6500 peoplehr

ticket hall

Calculate 5-minute peak passenger flow 650012 = 541

Allowance for ticket rejection 541 + 20 = 561

Allowance for growth in use demand 561 x 110 = 617

Assume 125 people use each gate over

the 5-minute peak passenger flow period 617125 = 493

Round up gates to nearest whole number 5 gates required

TABLE 4 London Underground ticket hall gateline parametersregulations (SPSG 2007)EQUIPMENT DIMENSIONS (W X D) POSITIONARRANGEMENT

Combined manualequipment gate 1750 x 270 mm Height ndash 1300 mm Adjacent to ticket office Can include

1400 mm clear opening luggage facility

Manual gate ndash passenger 1750 x 270 mm Height ndash 1300 mm Adjacent to ticket office With more than 6

gateluggage port 900 mm ndash clearance and walkways in a single gateline or significant

outward opening interchange with mainline stations airport or

tourist traffic provide separate luggage chutes

Manual gate ndash single leaf 1394 mm Height ndash 1300 mm Beside the dedicated excess gates

with luggage port 695 mm clearance

Gateline attendance report 990 x 1240 mm Within the controlled zone but not necessarily

part of the gateline Usually close

Station control unit (SCU) 432 x 430 mm Should be installed within a secure enclosure

(which could be a ticket office) and also with

a clear view of the gates Controls and monitors

the station underground ticketing system

equipment

Manual emergency exit gates 1900 x 125 mm Adjacent to ticket office Add these gates rather

than adding extra UTS gates for evacuation

purposes

or escalators) at these distances will result incompliance with the Underground standards but itmay not always be possible (or practical for aninterchange situation) to procure such a layout

Lengths of station components are controlled bya range of criteria For the platform it is the length ofthe train vehicle for interchange components it isthe vertical height in the terms of the depth to theplatform level for passages it is the designated finalexit positions and the existing station areas alreadygenerated

Vector direction labels are underpinned by thepermissible union of compatible spaces two points ofcontact will be maintained so as not to restrict thewidth of connecting passenger areas This undueconstriction of passenger flow at the head of apassage also satisfies realistic construction practiceNon-passenger and ancillary area vector directions areas per project requirements and the successive areaadditions managed by the SGEvac rule groupings

To create alternative layouts SGEvac uses alsquorandomizersquo rule In this shape grammar rule initialstate infrastructure is generated as per guidelinesbut it can randomly change the location ofcomponent arrangements within the predefined

boundaries so that a number of possible alternativelayouts are produced The designs can then beinterrogated and ranked by their travelling distancesand against the qualitative functional logicstatements

The SGEvac design tool developed in this researchis based on a highly restrictive additive grammar ruleset due to SPSGrsquos prescriptive guidance and thenature of the layout planning problem As Grimsdaleand Chang (1996) argue many published grammarslack high-level reasoning during the iteration processsince they are often concerned with the developmentof shape grammar theories rather than their practicalapplications By contrast SGEvac being a practicalapplication has a complete set of primitive shapesonly possible to arrange in a certain way It thusprovides viable reference design layout compositionsbased on SPSG with further interventions possible bythe designer

SYSTEM DEVELOPMENT TO DATEThe SGEvac prototype is compiled within the Clanguage environment allowing for modular object-oriented programming (suitable for shape grammar)Figure 10 shows the main interface where the stationname and relevant LU fare zone are entered (whichimpacts on sizing infrastructure) as are the ticket halland platform depths the overall site boundaries andfinal exitsrsquo positions The peak passenger flow figuresover the relevant time interval of normally 3 hours arealso specified SGEvac is then capable of accuratelycalculating all passenger area infrastructure based onthe SPSG guidelines as already discussed

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

110 G SMITH AND B CERANIC

TABLE 5 Station manager suite ndash shape rule connectionsELEMENT SHAPE RULE DESIGNATION CONNECTED COMPONENT GROUP ALLOCATION

Group station managerrsquos office SR_GSM_01 Ticket hall paid side GSM

SR_GSM_02 Ticket hall unpaid side

SR_GSM_03 GSM administration

SR_GSM_04 Duty station managerrsquos office

SR_GSM_05 Business managerrsquos office

SR_GSM_06 Administration

SR_GSM_07 Ticket office

SR_GSM_08 Police room

SR_GSM_09 Revenue protection examiner

SR_GSM_10 Interview room

140

Stair Stair

100 2020

4533

34

45

FIGURE 9 Platform escape distance SGEvac label restrictions

The history of auto-generated layouts is accessiblefrom this screen and the user is given the capabilityto delete and copy them (useful when minoralterations are required to produce an alternativelayout)

Shape labelling in the programme is multi-faceted which is necessary since many stationcomponents have various possible connections Anescalator for example may connect to a passagetravelator ramp escalator run-off stair run-off lift orthe ticket hall so all such connections need adding tothe component properties The SGEvac grouping ofcomponent interactions is accessed from menutoolbars that allow insertion of legislative compliantstation area connections as shown in Figure 11

Once the position and the depth of the ticket halland platform are specified the railway track is inserteddependent on the platform configuration The platformis the only possible connection to the track but thetrack can be positioned either side of platforms so thata side or centre (island) platform layout is possible

Non-passenger and ancillary areas are alsoselectable from the shape grammar library via property

windows for each station component eg the ticketoffice properties window as shown in Figure 12

Figure 13 shows examples of the layout modulestation components available in the SGEvacprogramme

Figure 14 shows the layout module interfaceInformation on the components (relative position inthe overall site layout and its absolute sizing) isdisplayed at the bottom of the screen once thecomponent is selected from the list In addition ameasurement bar (in metres) at the base of theinteractive design environment provides a sense ofscale for the designer

A checklist summarizes the station areas includedin the layout but a more detailed analysis is possiblevia a report function if required In addition to thisdetailed list a summary screen accessed from themain interface collates all specified components ineach generated layout This screen lists all passengernon-passenger and ancillary areas in their respectivecategories with infrastructure width data ofpassenger areas also summarized It also has anevacuation summary where the train frequency is

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 111

FIGURE 10 SGEvac main interface

specified and the train evacuation load (taken as peaklsquorush hourrsquo loading) is displayed This section of thesummary screen is useful for input into Simulexwhere fire evacuation test runs of the building layoutcan be implemented

The lsquoshow hierarchyrsquo function in Figure 15 allowsdesign team members to view the progression ofdesigns by superimposing connecting lines betweenstation elements in the order of their compositionThis also serves as a safeguard against deleting

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

112 G SMITH AND B CERANIC

FIGURE 12 SGEvac ticket office properties window

FIGURE 11 SGEvac properties window

components out of sequence that could lead to aninvalid solution

Once auto-generated layouts are complete withcorrectly sized and legitimately connected stationcomponents manual insertion of non-passenger areaswould then commence Upon completion thegenerated layout is transferred to AutoCAD via dxf fileformat as shown in Figure 16 Once in AutoCAD thelayout has a level of detail added by the designer suchas door openings passage connections fire escapeopenings etc so that passenger flows may beassessed It is then exported to Simulex for fireevacuation simulation analysis again via dxf file format

Simulex understands any voids as potentialcirculation routes treating the rest as solid andpermanent obstructions ie walls internal structuralelements permanent fixtures etc Since its solutionoperates in a two-dimensional environment for whatin effect is a three-dimensional problem the means

of vertical circulation (eg staircases) have to beestablished True lengths of staircases must bespecified based on the total length of all flights(sloping length) and the intermediate rests The fireexits need to be defined both in terms of theirlocation and width The links have to be establishedto define the point of the connection between thefloors and staircasesescalatorsramps The travellingdistance maps are then calculated type and numberof occupants specified building populated and theevacuation simulation performed

SYSTEM TESTING AND VALIDATIONKuligowski and Peacock (2005) define and recommendthe following ways of validating evacuation models

validation against code requirements validation against fire drills or other peoplersquos

movement experimentstrials

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 113

Gateline Vertical

Platform Right

Stairs Horizontal Stairs Vertical

Travelator Vertical

Ticket Hall

BMOAdmin Reception

Platform Top Ram Up Bottom Vertical

Lift Passage Horizontal

FIGURE 13 Examples of SGEvac station components

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

114 G SMITH AND B CERANIC

FIGURE 15 SGEvac hierarchy function

FIGURE 14 Layout module graphical user interface

validation against literature on past evacuationexperiments

validation against other models third-party validation

Furthermore Galea (1999) deconstructs the validationprocess into four main areas ndash component testingfunctional testing qualitative validation andquantitative validation This approach is more suitablefor models such as SGEvac than for those based oncompliance with code requirements and hence it isthe one adopted for the validation in this research

Component testing is regarded as testingcomputer code that drives the model checking forelementary errors Functional validation involveschecking that the model possesses the ability to

exhibit the range of capabilities required to performthe desired simulations Component testing andfunctional validation for SGEvac is carried outagainst code requirements insofar that allcalculations for infrastructure sizing are evaluatedmanually so that the method and answer are knownto be correct It uses the Simulex program to aid themodelling of people movement and to test thehypothesis of improvement to evacuation timesthus relying on Simulexrsquos own rigorous validationprocess Simulex has been used extensively inpractice as well as a validating model for otherevacuation softwares (Kuligowski and Peacock2005) SGEvac has gone through its initialcompletion and validation process with finalvalidation proposed for the end of the year

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 115

FIGURE 16 Auto-generated SGEvac station layout export file for fire

evacuation simulation in Simulex

FIRE EVACUATION SIMULATION RESULTSThe considered lsquotrain on fire in stationrsquo scenario takesinto account the service interval of 333 minutes(based on 18 trains per hour) the crush capacity of afour-car train together with the number of peoplewaiting on the platform after cancellation of oneservice (ie after 666 minutes) On the oppositeplatform the evacuation load is taken to be thenormal boardingalighting load assuming nocancellations to services (50 of all boarders andalighters)

Evacuation load predictions were based on TfL2006 projected passenger flows with an averageflow of 36 peopleminute for boarders and 123peopleminute for boarders and alighters combinedThe train crush capacity was averaged at 165peoplecar The simulation also models evacuationfrom the ticket hall level to assess the impact ofdispersing crowds at peak times in this part of thestation From these figures the total evacuation loadwas calculated as follows

Platform 1 Boarders (36 x 666 min) = 240Train crush load (165 x 4) = 660

Platform 2 Boarders and alighters

(123 x 05 x 333 min) = 205Ticket hall = 470Total station evacuation load = 1575 people

The simulation results have been captured as imagesat 25 50 100 and 300 second intervals respectivelyfor both platform and ticket hall levels as shown inFigures 17 to 20

The provided means of escape follow LU SPSG4requirements insofar that the maximum travellingdistance on a platform to an exit is no more than 45 m (ie exits must be no more than 90 m apart) andthat the dead-end travelling distance is no more than20 m to an exit Hence each platform is provided withthree exits The MIPsrsquo (mobility impaired persons)means of escape is provided from both ends ofplatforms via emergency evacuation lifts Totalevacuation time (ie all people reaching an exit onticket hallground level) was 5 minutes and

24 seconds with the platform and train load clearedwithin 3 minutes and 5 seconds This is within therequirements to evacuate the platform and train loadwithin 4 minutes with all people reaching a fire-separated route within 6 minutes (the ticket hall is notclassed as a fire-separated route)

The performed simulation was based on theminimum travelling distance maps resulting in eachperson exiting via the nearest fire exit route Theaverage 10 second lsquotime to start to moversquo wasassigned to all occupants in order to accommodatefor real world people behaviour commonly reportedby fire evacuation research studies The boarders onthe end of both platforms (including most peoplefrom end carriages of the train) evacuated viastaircasesMIP lifts directly to the ground level pointof escape The boarders situated around the middleof platforms together with the people from the twomiddle carriages used the central escalators to reachticket hall fire exits Additionally some people fromthe end carriages and some staffretail unitcustomers used this route as means of escape

In this lsquotrain on fire in stationrsquo scenario it wasassumed that none of the escapes routes areunavailable due to maintenance However furtherevacuation studies will investigate possible scenarioswhere one of the escalator routes may be unavailabledue to maintenance reasons

FUTURE RESEARCHIntended future work will concentrate on the furtherconditioning developing and modelling of shapegrammar design rules based on the outcomes of thefinal system validation and LU expertsrsquo feedbackFurther research is proposed on the development ofa novel layout optimization approach based on theshape grammar rules to investigate the evacuation-related parametric improvements in subsurfacesurface railway station design (eg minimizingtravelling distances)

The fire safety strategy in terms of performance-based fire risk design is an important aspirationalaspect of SGEvac requiring further research andliaison with experts to determine how such variedcriteria could be allied with layout planning This isparticularly important where deviation from standardsoccurs Thus further development of the expert

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

116 G SMITH AND B CERANIC

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 117

FIGURE 17 Station evacuation after 25 seconds

FIGURE 18 Station evacuation after 50 seconds

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

118 G SMITH AND B CERANIC

FIGURE 19 Station evacuation after 100 seconds

FIGURE 20 Station evacuation after 300 seconds

knowledge database would be required particularlyin the light of frequent changes to legislation

On a more practical note it could be beneficial todevelop the system on a stand-alone modular basislinked to each generated station component A list offire safety strategy objectives and functional logicstatements can then be specified for commonlyrecurring items (such as emergency lighting fire alarmdetectors sprinkler heads etc) and spaced accordingto the designerrsquos wishes or via shape rule formulation

Site constraint issues would also benefit fromfuture development as well as a move towardsdevelopment of the 3D shape grammar version Oncedevised other relevant 3D modelling systems such ascomputational fluid dynamics (CFD) smoke movementsoftware could be utilized in conjunction with SGEvac

CONCLUSIONSThe computer-aided modelling environment underdevelopment is intended to act as a layout planning aidfor those who practice in sub-surfacesurface railwaystation design The system under developmentcarefully considers London Underground StationPlanning Standards and Guidelines together withother relevant fire safety regulations that influenceoverall fire strategy design The proposed computer-aided design environment (SGEvac) is based on highlyrestrictive additive shape grammar rules driven by akernel of key passenger flow information to generatestation plan solutions

The major benefit to the design team is the ability toreview multiple layouts of accurately sized and correctlyplaced infrastructure at the reference or preliminarystage of the station layout planning design process Timeis hence saved regarding different option appraisals andquantitative testing using evacuation software shouldlead to a choice of more effective layouts in relation tomeans of escape travelling distances and improvementin the total evacuation times Modifications to layoutsshould be straightforward and there is scope toefficiently assess travelling distances and evacuationtimes for each of the considered alternatives comparedwith a lsquotraditionalrsquo approach

The proposed novel shape grammar approach ismost suited to new-build stations or largerefurbishment projects where more design flexibilityexists Sites with many site-specific constraints have

an array of predetermined parameters and will mostlikely benefit from the manual operation of SGEvac

Use in other rail systems due to recurringinfrastructure components and inherent functionallogic related to the laying out of the stationcomponents is also feasible provided that shapeconnections and shape rules are corrected to reflectthat countryrsquos legislative requirements

The potential for station layout improvementswith regard to emergency evacuation is an area thatmerits research especially in scenarios that concernthe safety of the general public and form a vital partof Londonrsquos transportation system Potential layoutimprovements could not only benefit evacuationsituations but also enhance the through-flow ofpeople during peak travelling periods

ACKNOWLEDGEMENTS The authors wish to express their gratitude to JacobsEngineering (UK) Ltd which has sponsored theresearch and to London Underground Ltd forproviding the design guidance that is reproduced inthis paper and for authorising the use of the stationplans London Underground has not reviewed thispaper and the content is solely the authorsrsquo work Mr Robert Hutchinson is also thanked for hiscomputer programming assistance

AUTHOR CONTACT DETAILSGGrraahhaamm SSmmiitthh (corresponding author) Senior Architectural

Technologist Jacobs (UK) Ltd Cardinal Square North Point

10 Nottingham Road Derby DE1 3QT UK Tel +44 (0)1332 234482

fax +44 (0)1332 344558 e-mail grahampsmithjacobscomBBoorriiss CCeerraanniicc Programme Leader for BScBA (Hons) Architectural

Programmes Room Number 105 Faculty of Arts Design and

Technology University of Derby Markeaton Street Derby DE22

3AW UK Tel +44 (0)1332 593136 e-mail bceranicderbyacuk

REFERENCES BS 7974 2001 Application of Fire Safety Engineering Principles to the

Design of Buildings London BSI

Cha MY and Gero JS 1999 Style Learning Inductive Generalisation of

Architectural Shape Patterns Paichai University and Sydney University

co-paper

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 119

Chase SC 1996 lsquoUsing logic to specify shapes and spatial relations in

design grammarsrsquo in Workshop notes Grammatical Design Fourth

International Conference on Artificial Intelligence in Design Stanford

University US June

Department of the Environment 1996 Design Principles of Fire Safety

London The Stationery Office Books

Flemming U 1987 lsquoMore than the sum of its parts the grammar of

Queen Anne housesrsquo in Environment and Planning B Planning and

Design 14 323ndash350

Fruin JJ 1971 Pedestrian Planning and Design New York Metropolitan

Association of Urban Designers and Environmental Planners

Galea ER 1999 lsquoA general approach to validating evacuation models with

an application to EXODUSrsquo in Journal of Fire Sciences 16 414ndash436

Galle P and Kovacs LB 1992 lsquoThe logic of worms a study in

architectural knowledge representationrsquo in Environment and Planning B

Planning and Design 19 5ndash31

Garling T Lindberg E and Mantyla T 1983 lsquoOrientation in buildings ndash

effects of familiarity visual access and orientation aidsrsquo in Journal of

Applied Psychology 68(1) 177ndash186

Grimsdale RL and Chang CW 1996 lsquoThe layout design language A

technique for generating layout plansrsquo in Computer Graphics Forum

15(2) 97ndash106

Heskestad AW 1999 lsquoPerformance in smoke of way guidance systemsrsquo

in Fire and Materials 23(6) 375ndash381

Knight T 1995 lsquoConstructive symmetryrsquo in Environment and Planning B

Planning and Design 22 419ndash450

Knight TW 1999 lsquoShape grammars six typesrsquo in Environment and

Planning B Planning and Design 26 15ndash31

Kuligowski ED and Peacock RD 2005 A Review of Building Evacuation

Models Technical Note 1471 National Institute of Standards and

Technology (NIST) Technology Administration US Department of

Commerce

Li A I-Kang 2003 lsquoThe Yingzao fashi in the information agersquo conference

paper for the Beaux-Arts Paul Philippe Cret and Twentieth Century

Architecture in China The University of Pennsylvania October

Liew H 2003 lsquoSGML ndash A shape grammar meta-languagersquo in Digital

Design 21st eCAADe Conference Proceedings Graz Austria September

639ndash648

Loomis B 2003 lsquoA note on generative design techniques SGGA a user

driven genetic algorithm for evolving non-deterministic shape grammarsrsquo

working paper Massachusetts Institute of Technology (MIT)

Magnusson SE Drysdale DD Fitzgerald RW Mowrer F

Quintiere JG Williamson RB and Zalosh RG 1995

lsquoA proposal for a model curriculum in fire safety engineeringrsquo

in Fire Safety Journal 25 9ndash10

Smith G Ceranic B and Fryer C 2003 lsquoDevelopment of CAD

environment for spatial layout optimisation in sub-surfacesurface rail

station designrsquo in Proceedings of International BIAT Conference The

British Museum London January 2003

SPSG 2007 London Underground Station Planning Standards and

Guidelines SPSG4 (1ndash371) London

Stiny G and Gips J 1971 lsquoShape grammars and the generative

specification of painting and sculpturersquo in Information Processing 71

1460ndash1465

Stiny G 1977 lsquoIce-ray a note on the generation of Chinese lattice

designsrsquo in Environment and Planning B Planning and Design 4 89ndash98

Stiny G 1980 lsquoKindergarten grammars designing with Froebelrsquos building

giftsrsquo in Environment and Planning B 3 409ndash462

Tapia M 1999 lsquoA visual implementation of a shape grammar systemrsquo in

Environment and Planning B Planning and Design 26(1) 59ndash73

Transport for London 2007 Tube Carries One Billion Passengers for First

Time wwwtflgovukcorporatemedianewscentre4770aspx

(accessed 24 March 2008)

Weller M 2007 London Underground private correspondence

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

120 G SMITH AND B CERANIC

or escalators) at these distances will result incompliance with the Underground standards but itmay not always be possible (or practical for aninterchange situation) to procure such a layout

Lengths of station components are controlled bya range of criteria For the platform it is the length ofthe train vehicle for interchange components it isthe vertical height in the terms of the depth to theplatform level for passages it is the designated finalexit positions and the existing station areas alreadygenerated

Vector direction labels are underpinned by thepermissible union of compatible spaces two points ofcontact will be maintained so as not to restrict thewidth of connecting passenger areas This undueconstriction of passenger flow at the head of apassage also satisfies realistic construction practiceNon-passenger and ancillary area vector directions areas per project requirements and the successive areaadditions managed by the SGEvac rule groupings

To create alternative layouts SGEvac uses alsquorandomizersquo rule In this shape grammar rule initialstate infrastructure is generated as per guidelinesbut it can randomly change the location ofcomponent arrangements within the predefined

boundaries so that a number of possible alternativelayouts are produced The designs can then beinterrogated and ranked by their travelling distancesand against the qualitative functional logicstatements

The SGEvac design tool developed in this researchis based on a highly restrictive additive grammar ruleset due to SPSGrsquos prescriptive guidance and thenature of the layout planning problem As Grimsdaleand Chang (1996) argue many published grammarslack high-level reasoning during the iteration processsince they are often concerned with the developmentof shape grammar theories rather than their practicalapplications By contrast SGEvac being a practicalapplication has a complete set of primitive shapesonly possible to arrange in a certain way It thusprovides viable reference design layout compositionsbased on SPSG with further interventions possible bythe designer

SYSTEM DEVELOPMENT TO DATEThe SGEvac prototype is compiled within the Clanguage environment allowing for modular object-oriented programming (suitable for shape grammar)Figure 10 shows the main interface where the stationname and relevant LU fare zone are entered (whichimpacts on sizing infrastructure) as are the ticket halland platform depths the overall site boundaries andfinal exitsrsquo positions The peak passenger flow figuresover the relevant time interval of normally 3 hours arealso specified SGEvac is then capable of accuratelycalculating all passenger area infrastructure based onthe SPSG guidelines as already discussed

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

110 G SMITH AND B CERANIC

TABLE 5 Station manager suite ndash shape rule connectionsELEMENT SHAPE RULE DESIGNATION CONNECTED COMPONENT GROUP ALLOCATION

Group station managerrsquos office SR_GSM_01 Ticket hall paid side GSM

SR_GSM_02 Ticket hall unpaid side

SR_GSM_03 GSM administration

SR_GSM_04 Duty station managerrsquos office

SR_GSM_05 Business managerrsquos office

SR_GSM_06 Administration

SR_GSM_07 Ticket office

SR_GSM_08 Police room

SR_GSM_09 Revenue protection examiner

SR_GSM_10 Interview room

140

Stair Stair

100 2020

4533

34

45

FIGURE 9 Platform escape distance SGEvac label restrictions

The history of auto-generated layouts is accessiblefrom this screen and the user is given the capabilityto delete and copy them (useful when minoralterations are required to produce an alternativelayout)

Shape labelling in the programme is multi-faceted which is necessary since many stationcomponents have various possible connections Anescalator for example may connect to a passagetravelator ramp escalator run-off stair run-off lift orthe ticket hall so all such connections need adding tothe component properties The SGEvac grouping ofcomponent interactions is accessed from menutoolbars that allow insertion of legislative compliantstation area connections as shown in Figure 11

Once the position and the depth of the ticket halland platform are specified the railway track is inserteddependent on the platform configuration The platformis the only possible connection to the track but thetrack can be positioned either side of platforms so thata side or centre (island) platform layout is possible

Non-passenger and ancillary areas are alsoselectable from the shape grammar library via property

windows for each station component eg the ticketoffice properties window as shown in Figure 12

Figure 13 shows examples of the layout modulestation components available in the SGEvacprogramme

Figure 14 shows the layout module interfaceInformation on the components (relative position inthe overall site layout and its absolute sizing) isdisplayed at the bottom of the screen once thecomponent is selected from the list In addition ameasurement bar (in metres) at the base of theinteractive design environment provides a sense ofscale for the designer

A checklist summarizes the station areas includedin the layout but a more detailed analysis is possiblevia a report function if required In addition to thisdetailed list a summary screen accessed from themain interface collates all specified components ineach generated layout This screen lists all passengernon-passenger and ancillary areas in their respectivecategories with infrastructure width data ofpassenger areas also summarized It also has anevacuation summary where the train frequency is

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 111

FIGURE 10 SGEvac main interface

specified and the train evacuation load (taken as peaklsquorush hourrsquo loading) is displayed This section of thesummary screen is useful for input into Simulexwhere fire evacuation test runs of the building layoutcan be implemented

The lsquoshow hierarchyrsquo function in Figure 15 allowsdesign team members to view the progression ofdesigns by superimposing connecting lines betweenstation elements in the order of their compositionThis also serves as a safeguard against deleting

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

112 G SMITH AND B CERANIC

FIGURE 12 SGEvac ticket office properties window

FIGURE 11 SGEvac properties window

components out of sequence that could lead to aninvalid solution

Once auto-generated layouts are complete withcorrectly sized and legitimately connected stationcomponents manual insertion of non-passenger areaswould then commence Upon completion thegenerated layout is transferred to AutoCAD via dxf fileformat as shown in Figure 16 Once in AutoCAD thelayout has a level of detail added by the designer suchas door openings passage connections fire escapeopenings etc so that passenger flows may beassessed It is then exported to Simulex for fireevacuation simulation analysis again via dxf file format

Simulex understands any voids as potentialcirculation routes treating the rest as solid andpermanent obstructions ie walls internal structuralelements permanent fixtures etc Since its solutionoperates in a two-dimensional environment for whatin effect is a three-dimensional problem the means

of vertical circulation (eg staircases) have to beestablished True lengths of staircases must bespecified based on the total length of all flights(sloping length) and the intermediate rests The fireexits need to be defined both in terms of theirlocation and width The links have to be establishedto define the point of the connection between thefloors and staircasesescalatorsramps The travellingdistance maps are then calculated type and numberof occupants specified building populated and theevacuation simulation performed

SYSTEM TESTING AND VALIDATIONKuligowski and Peacock (2005) define and recommendthe following ways of validating evacuation models

validation against code requirements validation against fire drills or other peoplersquos

movement experimentstrials

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 113

Gateline Vertical

Platform Right

Stairs Horizontal Stairs Vertical

Travelator Vertical

Ticket Hall

BMOAdmin Reception

Platform Top Ram Up Bottom Vertical

Lift Passage Horizontal

FIGURE 13 Examples of SGEvac station components

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

114 G SMITH AND B CERANIC

FIGURE 15 SGEvac hierarchy function

FIGURE 14 Layout module graphical user interface

validation against literature on past evacuationexperiments

validation against other models third-party validation

Furthermore Galea (1999) deconstructs the validationprocess into four main areas ndash component testingfunctional testing qualitative validation andquantitative validation This approach is more suitablefor models such as SGEvac than for those based oncompliance with code requirements and hence it isthe one adopted for the validation in this research

Component testing is regarded as testingcomputer code that drives the model checking forelementary errors Functional validation involveschecking that the model possesses the ability to

exhibit the range of capabilities required to performthe desired simulations Component testing andfunctional validation for SGEvac is carried outagainst code requirements insofar that allcalculations for infrastructure sizing are evaluatedmanually so that the method and answer are knownto be correct It uses the Simulex program to aid themodelling of people movement and to test thehypothesis of improvement to evacuation timesthus relying on Simulexrsquos own rigorous validationprocess Simulex has been used extensively inpractice as well as a validating model for otherevacuation softwares (Kuligowski and Peacock2005) SGEvac has gone through its initialcompletion and validation process with finalvalidation proposed for the end of the year

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 115

FIGURE 16 Auto-generated SGEvac station layout export file for fire

evacuation simulation in Simulex

FIRE EVACUATION SIMULATION RESULTSThe considered lsquotrain on fire in stationrsquo scenario takesinto account the service interval of 333 minutes(based on 18 trains per hour) the crush capacity of afour-car train together with the number of peoplewaiting on the platform after cancellation of oneservice (ie after 666 minutes) On the oppositeplatform the evacuation load is taken to be thenormal boardingalighting load assuming nocancellations to services (50 of all boarders andalighters)

Evacuation load predictions were based on TfL2006 projected passenger flows with an averageflow of 36 peopleminute for boarders and 123peopleminute for boarders and alighters combinedThe train crush capacity was averaged at 165peoplecar The simulation also models evacuationfrom the ticket hall level to assess the impact ofdispersing crowds at peak times in this part of thestation From these figures the total evacuation loadwas calculated as follows

Platform 1 Boarders (36 x 666 min) = 240Train crush load (165 x 4) = 660

Platform 2 Boarders and alighters

(123 x 05 x 333 min) = 205Ticket hall = 470Total station evacuation load = 1575 people

The simulation results have been captured as imagesat 25 50 100 and 300 second intervals respectivelyfor both platform and ticket hall levels as shown inFigures 17 to 20

The provided means of escape follow LU SPSG4requirements insofar that the maximum travellingdistance on a platform to an exit is no more than 45 m (ie exits must be no more than 90 m apart) andthat the dead-end travelling distance is no more than20 m to an exit Hence each platform is provided withthree exits The MIPsrsquo (mobility impaired persons)means of escape is provided from both ends ofplatforms via emergency evacuation lifts Totalevacuation time (ie all people reaching an exit onticket hallground level) was 5 minutes and

24 seconds with the platform and train load clearedwithin 3 minutes and 5 seconds This is within therequirements to evacuate the platform and train loadwithin 4 minutes with all people reaching a fire-separated route within 6 minutes (the ticket hall is notclassed as a fire-separated route)

The performed simulation was based on theminimum travelling distance maps resulting in eachperson exiting via the nearest fire exit route Theaverage 10 second lsquotime to start to moversquo wasassigned to all occupants in order to accommodatefor real world people behaviour commonly reportedby fire evacuation research studies The boarders onthe end of both platforms (including most peoplefrom end carriages of the train) evacuated viastaircasesMIP lifts directly to the ground level pointof escape The boarders situated around the middleof platforms together with the people from the twomiddle carriages used the central escalators to reachticket hall fire exits Additionally some people fromthe end carriages and some staffretail unitcustomers used this route as means of escape

In this lsquotrain on fire in stationrsquo scenario it wasassumed that none of the escapes routes areunavailable due to maintenance However furtherevacuation studies will investigate possible scenarioswhere one of the escalator routes may be unavailabledue to maintenance reasons

FUTURE RESEARCHIntended future work will concentrate on the furtherconditioning developing and modelling of shapegrammar design rules based on the outcomes of thefinal system validation and LU expertsrsquo feedbackFurther research is proposed on the development ofa novel layout optimization approach based on theshape grammar rules to investigate the evacuation-related parametric improvements in subsurfacesurface railway station design (eg minimizingtravelling distances)

The fire safety strategy in terms of performance-based fire risk design is an important aspirationalaspect of SGEvac requiring further research andliaison with experts to determine how such variedcriteria could be allied with layout planning This isparticularly important where deviation from standardsoccurs Thus further development of the expert

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

116 G SMITH AND B CERANIC

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 117

FIGURE 17 Station evacuation after 25 seconds

FIGURE 18 Station evacuation after 50 seconds

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

118 G SMITH AND B CERANIC

FIGURE 19 Station evacuation after 100 seconds

FIGURE 20 Station evacuation after 300 seconds

knowledge database would be required particularlyin the light of frequent changes to legislation

On a more practical note it could be beneficial todevelop the system on a stand-alone modular basislinked to each generated station component A list offire safety strategy objectives and functional logicstatements can then be specified for commonlyrecurring items (such as emergency lighting fire alarmdetectors sprinkler heads etc) and spaced accordingto the designerrsquos wishes or via shape rule formulation

Site constraint issues would also benefit fromfuture development as well as a move towardsdevelopment of the 3D shape grammar version Oncedevised other relevant 3D modelling systems such ascomputational fluid dynamics (CFD) smoke movementsoftware could be utilized in conjunction with SGEvac

CONCLUSIONSThe computer-aided modelling environment underdevelopment is intended to act as a layout planning aidfor those who practice in sub-surfacesurface railwaystation design The system under developmentcarefully considers London Underground StationPlanning Standards and Guidelines together withother relevant fire safety regulations that influenceoverall fire strategy design The proposed computer-aided design environment (SGEvac) is based on highlyrestrictive additive shape grammar rules driven by akernel of key passenger flow information to generatestation plan solutions

The major benefit to the design team is the ability toreview multiple layouts of accurately sized and correctlyplaced infrastructure at the reference or preliminarystage of the station layout planning design process Timeis hence saved regarding different option appraisals andquantitative testing using evacuation software shouldlead to a choice of more effective layouts in relation tomeans of escape travelling distances and improvementin the total evacuation times Modifications to layoutsshould be straightforward and there is scope toefficiently assess travelling distances and evacuationtimes for each of the considered alternatives comparedwith a lsquotraditionalrsquo approach

The proposed novel shape grammar approach ismost suited to new-build stations or largerefurbishment projects where more design flexibilityexists Sites with many site-specific constraints have

an array of predetermined parameters and will mostlikely benefit from the manual operation of SGEvac

Use in other rail systems due to recurringinfrastructure components and inherent functionallogic related to the laying out of the stationcomponents is also feasible provided that shapeconnections and shape rules are corrected to reflectthat countryrsquos legislative requirements

The potential for station layout improvementswith regard to emergency evacuation is an area thatmerits research especially in scenarios that concernthe safety of the general public and form a vital partof Londonrsquos transportation system Potential layoutimprovements could not only benefit evacuationsituations but also enhance the through-flow ofpeople during peak travelling periods

ACKNOWLEDGEMENTS The authors wish to express their gratitude to JacobsEngineering (UK) Ltd which has sponsored theresearch and to London Underground Ltd forproviding the design guidance that is reproduced inthis paper and for authorising the use of the stationplans London Underground has not reviewed thispaper and the content is solely the authorsrsquo work Mr Robert Hutchinson is also thanked for hiscomputer programming assistance

AUTHOR CONTACT DETAILSGGrraahhaamm SSmmiitthh (corresponding author) Senior Architectural

Technologist Jacobs (UK) Ltd Cardinal Square North Point

10 Nottingham Road Derby DE1 3QT UK Tel +44 (0)1332 234482

fax +44 (0)1332 344558 e-mail grahampsmithjacobscomBBoorriiss CCeerraanniicc Programme Leader for BScBA (Hons) Architectural

Programmes Room Number 105 Faculty of Arts Design and

Technology University of Derby Markeaton Street Derby DE22

3AW UK Tel +44 (0)1332 593136 e-mail bceranicderbyacuk

REFERENCES BS 7974 2001 Application of Fire Safety Engineering Principles to the

Design of Buildings London BSI

Cha MY and Gero JS 1999 Style Learning Inductive Generalisation of

Architectural Shape Patterns Paichai University and Sydney University

co-paper

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 119

Chase SC 1996 lsquoUsing logic to specify shapes and spatial relations in

design grammarsrsquo in Workshop notes Grammatical Design Fourth

International Conference on Artificial Intelligence in Design Stanford

University US June

Department of the Environment 1996 Design Principles of Fire Safety

London The Stationery Office Books

Flemming U 1987 lsquoMore than the sum of its parts the grammar of

Queen Anne housesrsquo in Environment and Planning B Planning and

Design 14 323ndash350

Fruin JJ 1971 Pedestrian Planning and Design New York Metropolitan

Association of Urban Designers and Environmental Planners

Galea ER 1999 lsquoA general approach to validating evacuation models with

an application to EXODUSrsquo in Journal of Fire Sciences 16 414ndash436

Galle P and Kovacs LB 1992 lsquoThe logic of worms a study in

architectural knowledge representationrsquo in Environment and Planning B

Planning and Design 19 5ndash31

Garling T Lindberg E and Mantyla T 1983 lsquoOrientation in buildings ndash

effects of familiarity visual access and orientation aidsrsquo in Journal of

Applied Psychology 68(1) 177ndash186

Grimsdale RL and Chang CW 1996 lsquoThe layout design language A

technique for generating layout plansrsquo in Computer Graphics Forum

15(2) 97ndash106

Heskestad AW 1999 lsquoPerformance in smoke of way guidance systemsrsquo

in Fire and Materials 23(6) 375ndash381

Knight T 1995 lsquoConstructive symmetryrsquo in Environment and Planning B

Planning and Design 22 419ndash450

Knight TW 1999 lsquoShape grammars six typesrsquo in Environment and

Planning B Planning and Design 26 15ndash31

Kuligowski ED and Peacock RD 2005 A Review of Building Evacuation

Models Technical Note 1471 National Institute of Standards and

Technology (NIST) Technology Administration US Department of

Commerce

Li A I-Kang 2003 lsquoThe Yingzao fashi in the information agersquo conference

paper for the Beaux-Arts Paul Philippe Cret and Twentieth Century

Architecture in China The University of Pennsylvania October

Liew H 2003 lsquoSGML ndash A shape grammar meta-languagersquo in Digital

Design 21st eCAADe Conference Proceedings Graz Austria September

639ndash648

Loomis B 2003 lsquoA note on generative design techniques SGGA a user

driven genetic algorithm for evolving non-deterministic shape grammarsrsquo

working paper Massachusetts Institute of Technology (MIT)

Magnusson SE Drysdale DD Fitzgerald RW Mowrer F

Quintiere JG Williamson RB and Zalosh RG 1995

lsquoA proposal for a model curriculum in fire safety engineeringrsquo

in Fire Safety Journal 25 9ndash10

Smith G Ceranic B and Fryer C 2003 lsquoDevelopment of CAD

environment for spatial layout optimisation in sub-surfacesurface rail

station designrsquo in Proceedings of International BIAT Conference The

British Museum London January 2003

SPSG 2007 London Underground Station Planning Standards and

Guidelines SPSG4 (1ndash371) London

Stiny G and Gips J 1971 lsquoShape grammars and the generative

specification of painting and sculpturersquo in Information Processing 71

1460ndash1465

Stiny G 1977 lsquoIce-ray a note on the generation of Chinese lattice

designsrsquo in Environment and Planning B Planning and Design 4 89ndash98

Stiny G 1980 lsquoKindergarten grammars designing with Froebelrsquos building

giftsrsquo in Environment and Planning B 3 409ndash462

Tapia M 1999 lsquoA visual implementation of a shape grammar systemrsquo in

Environment and Planning B Planning and Design 26(1) 59ndash73

Transport for London 2007 Tube Carries One Billion Passengers for First

Time wwwtflgovukcorporatemedianewscentre4770aspx

(accessed 24 March 2008)

Weller M 2007 London Underground private correspondence

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

120 G SMITH AND B CERANIC

The history of auto-generated layouts is accessiblefrom this screen and the user is given the capabilityto delete and copy them (useful when minoralterations are required to produce an alternativelayout)

Shape labelling in the programme is multi-faceted which is necessary since many stationcomponents have various possible connections Anescalator for example may connect to a passagetravelator ramp escalator run-off stair run-off lift orthe ticket hall so all such connections need adding tothe component properties The SGEvac grouping ofcomponent interactions is accessed from menutoolbars that allow insertion of legislative compliantstation area connections as shown in Figure 11

Once the position and the depth of the ticket halland platform are specified the railway track is inserteddependent on the platform configuration The platformis the only possible connection to the track but thetrack can be positioned either side of platforms so thata side or centre (island) platform layout is possible

Non-passenger and ancillary areas are alsoselectable from the shape grammar library via property

windows for each station component eg the ticketoffice properties window as shown in Figure 12

Figure 13 shows examples of the layout modulestation components available in the SGEvacprogramme

Figure 14 shows the layout module interfaceInformation on the components (relative position inthe overall site layout and its absolute sizing) isdisplayed at the bottom of the screen once thecomponent is selected from the list In addition ameasurement bar (in metres) at the base of theinteractive design environment provides a sense ofscale for the designer

A checklist summarizes the station areas includedin the layout but a more detailed analysis is possiblevia a report function if required In addition to thisdetailed list a summary screen accessed from themain interface collates all specified components ineach generated layout This screen lists all passengernon-passenger and ancillary areas in their respectivecategories with infrastructure width data ofpassenger areas also summarized It also has anevacuation summary where the train frequency is

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 111

FIGURE 10 SGEvac main interface

specified and the train evacuation load (taken as peaklsquorush hourrsquo loading) is displayed This section of thesummary screen is useful for input into Simulexwhere fire evacuation test runs of the building layoutcan be implemented

The lsquoshow hierarchyrsquo function in Figure 15 allowsdesign team members to view the progression ofdesigns by superimposing connecting lines betweenstation elements in the order of their compositionThis also serves as a safeguard against deleting

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

112 G SMITH AND B CERANIC

FIGURE 12 SGEvac ticket office properties window

FIGURE 11 SGEvac properties window

components out of sequence that could lead to aninvalid solution

Once auto-generated layouts are complete withcorrectly sized and legitimately connected stationcomponents manual insertion of non-passenger areaswould then commence Upon completion thegenerated layout is transferred to AutoCAD via dxf fileformat as shown in Figure 16 Once in AutoCAD thelayout has a level of detail added by the designer suchas door openings passage connections fire escapeopenings etc so that passenger flows may beassessed It is then exported to Simulex for fireevacuation simulation analysis again via dxf file format

Simulex understands any voids as potentialcirculation routes treating the rest as solid andpermanent obstructions ie walls internal structuralelements permanent fixtures etc Since its solutionoperates in a two-dimensional environment for whatin effect is a three-dimensional problem the means

of vertical circulation (eg staircases) have to beestablished True lengths of staircases must bespecified based on the total length of all flights(sloping length) and the intermediate rests The fireexits need to be defined both in terms of theirlocation and width The links have to be establishedto define the point of the connection between thefloors and staircasesescalatorsramps The travellingdistance maps are then calculated type and numberof occupants specified building populated and theevacuation simulation performed

SYSTEM TESTING AND VALIDATIONKuligowski and Peacock (2005) define and recommendthe following ways of validating evacuation models

validation against code requirements validation against fire drills or other peoplersquos

movement experimentstrials

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 113

Gateline Vertical

Platform Right

Stairs Horizontal Stairs Vertical

Travelator Vertical

Ticket Hall

BMOAdmin Reception

Platform Top Ram Up Bottom Vertical

Lift Passage Horizontal

FIGURE 13 Examples of SGEvac station components

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

114 G SMITH AND B CERANIC

FIGURE 15 SGEvac hierarchy function

FIGURE 14 Layout module graphical user interface

validation against literature on past evacuationexperiments

validation against other models third-party validation

Furthermore Galea (1999) deconstructs the validationprocess into four main areas ndash component testingfunctional testing qualitative validation andquantitative validation This approach is more suitablefor models such as SGEvac than for those based oncompliance with code requirements and hence it isthe one adopted for the validation in this research

Component testing is regarded as testingcomputer code that drives the model checking forelementary errors Functional validation involveschecking that the model possesses the ability to

exhibit the range of capabilities required to performthe desired simulations Component testing andfunctional validation for SGEvac is carried outagainst code requirements insofar that allcalculations for infrastructure sizing are evaluatedmanually so that the method and answer are knownto be correct It uses the Simulex program to aid themodelling of people movement and to test thehypothesis of improvement to evacuation timesthus relying on Simulexrsquos own rigorous validationprocess Simulex has been used extensively inpractice as well as a validating model for otherevacuation softwares (Kuligowski and Peacock2005) SGEvac has gone through its initialcompletion and validation process with finalvalidation proposed for the end of the year

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 115

FIGURE 16 Auto-generated SGEvac station layout export file for fire

evacuation simulation in Simulex

FIRE EVACUATION SIMULATION RESULTSThe considered lsquotrain on fire in stationrsquo scenario takesinto account the service interval of 333 minutes(based on 18 trains per hour) the crush capacity of afour-car train together with the number of peoplewaiting on the platform after cancellation of oneservice (ie after 666 minutes) On the oppositeplatform the evacuation load is taken to be thenormal boardingalighting load assuming nocancellations to services (50 of all boarders andalighters)

Evacuation load predictions were based on TfL2006 projected passenger flows with an averageflow of 36 peopleminute for boarders and 123peopleminute for boarders and alighters combinedThe train crush capacity was averaged at 165peoplecar The simulation also models evacuationfrom the ticket hall level to assess the impact ofdispersing crowds at peak times in this part of thestation From these figures the total evacuation loadwas calculated as follows

Platform 1 Boarders (36 x 666 min) = 240Train crush load (165 x 4) = 660

Platform 2 Boarders and alighters

(123 x 05 x 333 min) = 205Ticket hall = 470Total station evacuation load = 1575 people

The simulation results have been captured as imagesat 25 50 100 and 300 second intervals respectivelyfor both platform and ticket hall levels as shown inFigures 17 to 20

The provided means of escape follow LU SPSG4requirements insofar that the maximum travellingdistance on a platform to an exit is no more than 45 m (ie exits must be no more than 90 m apart) andthat the dead-end travelling distance is no more than20 m to an exit Hence each platform is provided withthree exits The MIPsrsquo (mobility impaired persons)means of escape is provided from both ends ofplatforms via emergency evacuation lifts Totalevacuation time (ie all people reaching an exit onticket hallground level) was 5 minutes and

24 seconds with the platform and train load clearedwithin 3 minutes and 5 seconds This is within therequirements to evacuate the platform and train loadwithin 4 minutes with all people reaching a fire-separated route within 6 minutes (the ticket hall is notclassed as a fire-separated route)

The performed simulation was based on theminimum travelling distance maps resulting in eachperson exiting via the nearest fire exit route Theaverage 10 second lsquotime to start to moversquo wasassigned to all occupants in order to accommodatefor real world people behaviour commonly reportedby fire evacuation research studies The boarders onthe end of both platforms (including most peoplefrom end carriages of the train) evacuated viastaircasesMIP lifts directly to the ground level pointof escape The boarders situated around the middleof platforms together with the people from the twomiddle carriages used the central escalators to reachticket hall fire exits Additionally some people fromthe end carriages and some staffretail unitcustomers used this route as means of escape

In this lsquotrain on fire in stationrsquo scenario it wasassumed that none of the escapes routes areunavailable due to maintenance However furtherevacuation studies will investigate possible scenarioswhere one of the escalator routes may be unavailabledue to maintenance reasons

FUTURE RESEARCHIntended future work will concentrate on the furtherconditioning developing and modelling of shapegrammar design rules based on the outcomes of thefinal system validation and LU expertsrsquo feedbackFurther research is proposed on the development ofa novel layout optimization approach based on theshape grammar rules to investigate the evacuation-related parametric improvements in subsurfacesurface railway station design (eg minimizingtravelling distances)

The fire safety strategy in terms of performance-based fire risk design is an important aspirationalaspect of SGEvac requiring further research andliaison with experts to determine how such variedcriteria could be allied with layout planning This isparticularly important where deviation from standardsoccurs Thus further development of the expert

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

116 G SMITH AND B CERANIC

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 117

FIGURE 17 Station evacuation after 25 seconds

FIGURE 18 Station evacuation after 50 seconds

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

118 G SMITH AND B CERANIC

FIGURE 19 Station evacuation after 100 seconds

FIGURE 20 Station evacuation after 300 seconds

knowledge database would be required particularlyin the light of frequent changes to legislation

On a more practical note it could be beneficial todevelop the system on a stand-alone modular basislinked to each generated station component A list offire safety strategy objectives and functional logicstatements can then be specified for commonlyrecurring items (such as emergency lighting fire alarmdetectors sprinkler heads etc) and spaced accordingto the designerrsquos wishes or via shape rule formulation

Site constraint issues would also benefit fromfuture development as well as a move towardsdevelopment of the 3D shape grammar version Oncedevised other relevant 3D modelling systems such ascomputational fluid dynamics (CFD) smoke movementsoftware could be utilized in conjunction with SGEvac

CONCLUSIONSThe computer-aided modelling environment underdevelopment is intended to act as a layout planning aidfor those who practice in sub-surfacesurface railwaystation design The system under developmentcarefully considers London Underground StationPlanning Standards and Guidelines together withother relevant fire safety regulations that influenceoverall fire strategy design The proposed computer-aided design environment (SGEvac) is based on highlyrestrictive additive shape grammar rules driven by akernel of key passenger flow information to generatestation plan solutions

The major benefit to the design team is the ability toreview multiple layouts of accurately sized and correctlyplaced infrastructure at the reference or preliminarystage of the station layout planning design process Timeis hence saved regarding different option appraisals andquantitative testing using evacuation software shouldlead to a choice of more effective layouts in relation tomeans of escape travelling distances and improvementin the total evacuation times Modifications to layoutsshould be straightforward and there is scope toefficiently assess travelling distances and evacuationtimes for each of the considered alternatives comparedwith a lsquotraditionalrsquo approach

The proposed novel shape grammar approach ismost suited to new-build stations or largerefurbishment projects where more design flexibilityexists Sites with many site-specific constraints have

an array of predetermined parameters and will mostlikely benefit from the manual operation of SGEvac

Use in other rail systems due to recurringinfrastructure components and inherent functionallogic related to the laying out of the stationcomponents is also feasible provided that shapeconnections and shape rules are corrected to reflectthat countryrsquos legislative requirements

The potential for station layout improvementswith regard to emergency evacuation is an area thatmerits research especially in scenarios that concernthe safety of the general public and form a vital partof Londonrsquos transportation system Potential layoutimprovements could not only benefit evacuationsituations but also enhance the through-flow ofpeople during peak travelling periods

ACKNOWLEDGEMENTS The authors wish to express their gratitude to JacobsEngineering (UK) Ltd which has sponsored theresearch and to London Underground Ltd forproviding the design guidance that is reproduced inthis paper and for authorising the use of the stationplans London Underground has not reviewed thispaper and the content is solely the authorsrsquo work Mr Robert Hutchinson is also thanked for hiscomputer programming assistance

AUTHOR CONTACT DETAILSGGrraahhaamm SSmmiitthh (corresponding author) Senior Architectural

Technologist Jacobs (UK) Ltd Cardinal Square North Point

10 Nottingham Road Derby DE1 3QT UK Tel +44 (0)1332 234482

fax +44 (0)1332 344558 e-mail grahampsmithjacobscomBBoorriiss CCeerraanniicc Programme Leader for BScBA (Hons) Architectural

Programmes Room Number 105 Faculty of Arts Design and

Technology University of Derby Markeaton Street Derby DE22

3AW UK Tel +44 (0)1332 593136 e-mail bceranicderbyacuk

REFERENCES BS 7974 2001 Application of Fire Safety Engineering Principles to the

Design of Buildings London BSI

Cha MY and Gero JS 1999 Style Learning Inductive Generalisation of

Architectural Shape Patterns Paichai University and Sydney University

co-paper

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 119

Chase SC 1996 lsquoUsing logic to specify shapes and spatial relations in

design grammarsrsquo in Workshop notes Grammatical Design Fourth

International Conference on Artificial Intelligence in Design Stanford

University US June

Department of the Environment 1996 Design Principles of Fire Safety

London The Stationery Office Books

Flemming U 1987 lsquoMore than the sum of its parts the grammar of

Queen Anne housesrsquo in Environment and Planning B Planning and

Design 14 323ndash350

Fruin JJ 1971 Pedestrian Planning and Design New York Metropolitan

Association of Urban Designers and Environmental Planners

Galea ER 1999 lsquoA general approach to validating evacuation models with

an application to EXODUSrsquo in Journal of Fire Sciences 16 414ndash436

Galle P and Kovacs LB 1992 lsquoThe logic of worms a study in

architectural knowledge representationrsquo in Environment and Planning B

Planning and Design 19 5ndash31

Garling T Lindberg E and Mantyla T 1983 lsquoOrientation in buildings ndash

effects of familiarity visual access and orientation aidsrsquo in Journal of

Applied Psychology 68(1) 177ndash186

Grimsdale RL and Chang CW 1996 lsquoThe layout design language A

technique for generating layout plansrsquo in Computer Graphics Forum

15(2) 97ndash106

Heskestad AW 1999 lsquoPerformance in smoke of way guidance systemsrsquo

in Fire and Materials 23(6) 375ndash381

Knight T 1995 lsquoConstructive symmetryrsquo in Environment and Planning B

Planning and Design 22 419ndash450

Knight TW 1999 lsquoShape grammars six typesrsquo in Environment and

Planning B Planning and Design 26 15ndash31

Kuligowski ED and Peacock RD 2005 A Review of Building Evacuation

Models Technical Note 1471 National Institute of Standards and

Technology (NIST) Technology Administration US Department of

Commerce

Li A I-Kang 2003 lsquoThe Yingzao fashi in the information agersquo conference

paper for the Beaux-Arts Paul Philippe Cret and Twentieth Century

Architecture in China The University of Pennsylvania October

Liew H 2003 lsquoSGML ndash A shape grammar meta-languagersquo in Digital

Design 21st eCAADe Conference Proceedings Graz Austria September

639ndash648

Loomis B 2003 lsquoA note on generative design techniques SGGA a user

driven genetic algorithm for evolving non-deterministic shape grammarsrsquo

working paper Massachusetts Institute of Technology (MIT)

Magnusson SE Drysdale DD Fitzgerald RW Mowrer F

Quintiere JG Williamson RB and Zalosh RG 1995

lsquoA proposal for a model curriculum in fire safety engineeringrsquo

in Fire Safety Journal 25 9ndash10

Smith G Ceranic B and Fryer C 2003 lsquoDevelopment of CAD

environment for spatial layout optimisation in sub-surfacesurface rail

station designrsquo in Proceedings of International BIAT Conference The

British Museum London January 2003

SPSG 2007 London Underground Station Planning Standards and

Guidelines SPSG4 (1ndash371) London

Stiny G and Gips J 1971 lsquoShape grammars and the generative

specification of painting and sculpturersquo in Information Processing 71

1460ndash1465

Stiny G 1977 lsquoIce-ray a note on the generation of Chinese lattice

designsrsquo in Environment and Planning B Planning and Design 4 89ndash98

Stiny G 1980 lsquoKindergarten grammars designing with Froebelrsquos building

giftsrsquo in Environment and Planning B 3 409ndash462

Tapia M 1999 lsquoA visual implementation of a shape grammar systemrsquo in

Environment and Planning B Planning and Design 26(1) 59ndash73

Transport for London 2007 Tube Carries One Billion Passengers for First

Time wwwtflgovukcorporatemedianewscentre4770aspx

(accessed 24 March 2008)

Weller M 2007 London Underground private correspondence

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

120 G SMITH AND B CERANIC

specified and the train evacuation load (taken as peaklsquorush hourrsquo loading) is displayed This section of thesummary screen is useful for input into Simulexwhere fire evacuation test runs of the building layoutcan be implemented

The lsquoshow hierarchyrsquo function in Figure 15 allowsdesign team members to view the progression ofdesigns by superimposing connecting lines betweenstation elements in the order of their compositionThis also serves as a safeguard against deleting

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

112 G SMITH AND B CERANIC

FIGURE 12 SGEvac ticket office properties window

FIGURE 11 SGEvac properties window

components out of sequence that could lead to aninvalid solution

Once auto-generated layouts are complete withcorrectly sized and legitimately connected stationcomponents manual insertion of non-passenger areaswould then commence Upon completion thegenerated layout is transferred to AutoCAD via dxf fileformat as shown in Figure 16 Once in AutoCAD thelayout has a level of detail added by the designer suchas door openings passage connections fire escapeopenings etc so that passenger flows may beassessed It is then exported to Simulex for fireevacuation simulation analysis again via dxf file format

Simulex understands any voids as potentialcirculation routes treating the rest as solid andpermanent obstructions ie walls internal structuralelements permanent fixtures etc Since its solutionoperates in a two-dimensional environment for whatin effect is a three-dimensional problem the means

of vertical circulation (eg staircases) have to beestablished True lengths of staircases must bespecified based on the total length of all flights(sloping length) and the intermediate rests The fireexits need to be defined both in terms of theirlocation and width The links have to be establishedto define the point of the connection between thefloors and staircasesescalatorsramps The travellingdistance maps are then calculated type and numberof occupants specified building populated and theevacuation simulation performed

SYSTEM TESTING AND VALIDATIONKuligowski and Peacock (2005) define and recommendthe following ways of validating evacuation models

validation against code requirements validation against fire drills or other peoplersquos

movement experimentstrials

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 113

Gateline Vertical

Platform Right

Stairs Horizontal Stairs Vertical

Travelator Vertical

Ticket Hall

BMOAdmin Reception

Platform Top Ram Up Bottom Vertical

Lift Passage Horizontal

FIGURE 13 Examples of SGEvac station components

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

114 G SMITH AND B CERANIC

FIGURE 15 SGEvac hierarchy function

FIGURE 14 Layout module graphical user interface

validation against literature on past evacuationexperiments

validation against other models third-party validation

Furthermore Galea (1999) deconstructs the validationprocess into four main areas ndash component testingfunctional testing qualitative validation andquantitative validation This approach is more suitablefor models such as SGEvac than for those based oncompliance with code requirements and hence it isthe one adopted for the validation in this research

Component testing is regarded as testingcomputer code that drives the model checking forelementary errors Functional validation involveschecking that the model possesses the ability to

exhibit the range of capabilities required to performthe desired simulations Component testing andfunctional validation for SGEvac is carried outagainst code requirements insofar that allcalculations for infrastructure sizing are evaluatedmanually so that the method and answer are knownto be correct It uses the Simulex program to aid themodelling of people movement and to test thehypothesis of improvement to evacuation timesthus relying on Simulexrsquos own rigorous validationprocess Simulex has been used extensively inpractice as well as a validating model for otherevacuation softwares (Kuligowski and Peacock2005) SGEvac has gone through its initialcompletion and validation process with finalvalidation proposed for the end of the year

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 115

FIGURE 16 Auto-generated SGEvac station layout export file for fire

evacuation simulation in Simulex

FIRE EVACUATION SIMULATION RESULTSThe considered lsquotrain on fire in stationrsquo scenario takesinto account the service interval of 333 minutes(based on 18 trains per hour) the crush capacity of afour-car train together with the number of peoplewaiting on the platform after cancellation of oneservice (ie after 666 minutes) On the oppositeplatform the evacuation load is taken to be thenormal boardingalighting load assuming nocancellations to services (50 of all boarders andalighters)

Evacuation load predictions were based on TfL2006 projected passenger flows with an averageflow of 36 peopleminute for boarders and 123peopleminute for boarders and alighters combinedThe train crush capacity was averaged at 165peoplecar The simulation also models evacuationfrom the ticket hall level to assess the impact ofdispersing crowds at peak times in this part of thestation From these figures the total evacuation loadwas calculated as follows

Platform 1 Boarders (36 x 666 min) = 240Train crush load (165 x 4) = 660

Platform 2 Boarders and alighters

(123 x 05 x 333 min) = 205Ticket hall = 470Total station evacuation load = 1575 people

The simulation results have been captured as imagesat 25 50 100 and 300 second intervals respectivelyfor both platform and ticket hall levels as shown inFigures 17 to 20

The provided means of escape follow LU SPSG4requirements insofar that the maximum travellingdistance on a platform to an exit is no more than 45 m (ie exits must be no more than 90 m apart) andthat the dead-end travelling distance is no more than20 m to an exit Hence each platform is provided withthree exits The MIPsrsquo (mobility impaired persons)means of escape is provided from both ends ofplatforms via emergency evacuation lifts Totalevacuation time (ie all people reaching an exit onticket hallground level) was 5 minutes and

24 seconds with the platform and train load clearedwithin 3 minutes and 5 seconds This is within therequirements to evacuate the platform and train loadwithin 4 minutes with all people reaching a fire-separated route within 6 minutes (the ticket hall is notclassed as a fire-separated route)

The performed simulation was based on theminimum travelling distance maps resulting in eachperson exiting via the nearest fire exit route Theaverage 10 second lsquotime to start to moversquo wasassigned to all occupants in order to accommodatefor real world people behaviour commonly reportedby fire evacuation research studies The boarders onthe end of both platforms (including most peoplefrom end carriages of the train) evacuated viastaircasesMIP lifts directly to the ground level pointof escape The boarders situated around the middleof platforms together with the people from the twomiddle carriages used the central escalators to reachticket hall fire exits Additionally some people fromthe end carriages and some staffretail unitcustomers used this route as means of escape

In this lsquotrain on fire in stationrsquo scenario it wasassumed that none of the escapes routes areunavailable due to maintenance However furtherevacuation studies will investigate possible scenarioswhere one of the escalator routes may be unavailabledue to maintenance reasons

FUTURE RESEARCHIntended future work will concentrate on the furtherconditioning developing and modelling of shapegrammar design rules based on the outcomes of thefinal system validation and LU expertsrsquo feedbackFurther research is proposed on the development ofa novel layout optimization approach based on theshape grammar rules to investigate the evacuation-related parametric improvements in subsurfacesurface railway station design (eg minimizingtravelling distances)

The fire safety strategy in terms of performance-based fire risk design is an important aspirationalaspect of SGEvac requiring further research andliaison with experts to determine how such variedcriteria could be allied with layout planning This isparticularly important where deviation from standardsoccurs Thus further development of the expert

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

116 G SMITH AND B CERANIC

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 117

FIGURE 17 Station evacuation after 25 seconds

FIGURE 18 Station evacuation after 50 seconds

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

118 G SMITH AND B CERANIC

FIGURE 19 Station evacuation after 100 seconds

FIGURE 20 Station evacuation after 300 seconds

knowledge database would be required particularlyin the light of frequent changes to legislation

On a more practical note it could be beneficial todevelop the system on a stand-alone modular basislinked to each generated station component A list offire safety strategy objectives and functional logicstatements can then be specified for commonlyrecurring items (such as emergency lighting fire alarmdetectors sprinkler heads etc) and spaced accordingto the designerrsquos wishes or via shape rule formulation

Site constraint issues would also benefit fromfuture development as well as a move towardsdevelopment of the 3D shape grammar version Oncedevised other relevant 3D modelling systems such ascomputational fluid dynamics (CFD) smoke movementsoftware could be utilized in conjunction with SGEvac

CONCLUSIONSThe computer-aided modelling environment underdevelopment is intended to act as a layout planning aidfor those who practice in sub-surfacesurface railwaystation design The system under developmentcarefully considers London Underground StationPlanning Standards and Guidelines together withother relevant fire safety regulations that influenceoverall fire strategy design The proposed computer-aided design environment (SGEvac) is based on highlyrestrictive additive shape grammar rules driven by akernel of key passenger flow information to generatestation plan solutions

The major benefit to the design team is the ability toreview multiple layouts of accurately sized and correctlyplaced infrastructure at the reference or preliminarystage of the station layout planning design process Timeis hence saved regarding different option appraisals andquantitative testing using evacuation software shouldlead to a choice of more effective layouts in relation tomeans of escape travelling distances and improvementin the total evacuation times Modifications to layoutsshould be straightforward and there is scope toefficiently assess travelling distances and evacuationtimes for each of the considered alternatives comparedwith a lsquotraditionalrsquo approach

The proposed novel shape grammar approach ismost suited to new-build stations or largerefurbishment projects where more design flexibilityexists Sites with many site-specific constraints have

an array of predetermined parameters and will mostlikely benefit from the manual operation of SGEvac

Use in other rail systems due to recurringinfrastructure components and inherent functionallogic related to the laying out of the stationcomponents is also feasible provided that shapeconnections and shape rules are corrected to reflectthat countryrsquos legislative requirements

The potential for station layout improvementswith regard to emergency evacuation is an area thatmerits research especially in scenarios that concernthe safety of the general public and form a vital partof Londonrsquos transportation system Potential layoutimprovements could not only benefit evacuationsituations but also enhance the through-flow ofpeople during peak travelling periods

ACKNOWLEDGEMENTS The authors wish to express their gratitude to JacobsEngineering (UK) Ltd which has sponsored theresearch and to London Underground Ltd forproviding the design guidance that is reproduced inthis paper and for authorising the use of the stationplans London Underground has not reviewed thispaper and the content is solely the authorsrsquo work Mr Robert Hutchinson is also thanked for hiscomputer programming assistance

AUTHOR CONTACT DETAILSGGrraahhaamm SSmmiitthh (corresponding author) Senior Architectural

Technologist Jacobs (UK) Ltd Cardinal Square North Point

10 Nottingham Road Derby DE1 3QT UK Tel +44 (0)1332 234482

fax +44 (0)1332 344558 e-mail grahampsmithjacobscomBBoorriiss CCeerraanniicc Programme Leader for BScBA (Hons) Architectural

Programmes Room Number 105 Faculty of Arts Design and

Technology University of Derby Markeaton Street Derby DE22

3AW UK Tel +44 (0)1332 593136 e-mail bceranicderbyacuk

REFERENCES BS 7974 2001 Application of Fire Safety Engineering Principles to the

Design of Buildings London BSI

Cha MY and Gero JS 1999 Style Learning Inductive Generalisation of

Architectural Shape Patterns Paichai University and Sydney University

co-paper

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 119

Chase SC 1996 lsquoUsing logic to specify shapes and spatial relations in

design grammarsrsquo in Workshop notes Grammatical Design Fourth

International Conference on Artificial Intelligence in Design Stanford

University US June

Department of the Environment 1996 Design Principles of Fire Safety

London The Stationery Office Books

Flemming U 1987 lsquoMore than the sum of its parts the grammar of

Queen Anne housesrsquo in Environment and Planning B Planning and

Design 14 323ndash350

Fruin JJ 1971 Pedestrian Planning and Design New York Metropolitan

Association of Urban Designers and Environmental Planners

Galea ER 1999 lsquoA general approach to validating evacuation models with

an application to EXODUSrsquo in Journal of Fire Sciences 16 414ndash436

Galle P and Kovacs LB 1992 lsquoThe logic of worms a study in

architectural knowledge representationrsquo in Environment and Planning B

Planning and Design 19 5ndash31

Garling T Lindberg E and Mantyla T 1983 lsquoOrientation in buildings ndash

effects of familiarity visual access and orientation aidsrsquo in Journal of

Applied Psychology 68(1) 177ndash186

Grimsdale RL and Chang CW 1996 lsquoThe layout design language A

technique for generating layout plansrsquo in Computer Graphics Forum

15(2) 97ndash106

Heskestad AW 1999 lsquoPerformance in smoke of way guidance systemsrsquo

in Fire and Materials 23(6) 375ndash381

Knight T 1995 lsquoConstructive symmetryrsquo in Environment and Planning B

Planning and Design 22 419ndash450

Knight TW 1999 lsquoShape grammars six typesrsquo in Environment and

Planning B Planning and Design 26 15ndash31

Kuligowski ED and Peacock RD 2005 A Review of Building Evacuation

Models Technical Note 1471 National Institute of Standards and

Technology (NIST) Technology Administration US Department of

Commerce

Li A I-Kang 2003 lsquoThe Yingzao fashi in the information agersquo conference

paper for the Beaux-Arts Paul Philippe Cret and Twentieth Century

Architecture in China The University of Pennsylvania October

Liew H 2003 lsquoSGML ndash A shape grammar meta-languagersquo in Digital

Design 21st eCAADe Conference Proceedings Graz Austria September

639ndash648

Loomis B 2003 lsquoA note on generative design techniques SGGA a user

driven genetic algorithm for evolving non-deterministic shape grammarsrsquo

working paper Massachusetts Institute of Technology (MIT)

Magnusson SE Drysdale DD Fitzgerald RW Mowrer F

Quintiere JG Williamson RB and Zalosh RG 1995

lsquoA proposal for a model curriculum in fire safety engineeringrsquo

in Fire Safety Journal 25 9ndash10

Smith G Ceranic B and Fryer C 2003 lsquoDevelopment of CAD

environment for spatial layout optimisation in sub-surfacesurface rail

station designrsquo in Proceedings of International BIAT Conference The

British Museum London January 2003

SPSG 2007 London Underground Station Planning Standards and

Guidelines SPSG4 (1ndash371) London

Stiny G and Gips J 1971 lsquoShape grammars and the generative

specification of painting and sculpturersquo in Information Processing 71

1460ndash1465

Stiny G 1977 lsquoIce-ray a note on the generation of Chinese lattice

designsrsquo in Environment and Planning B Planning and Design 4 89ndash98

Stiny G 1980 lsquoKindergarten grammars designing with Froebelrsquos building

giftsrsquo in Environment and Planning B 3 409ndash462

Tapia M 1999 lsquoA visual implementation of a shape grammar systemrsquo in

Environment and Planning B Planning and Design 26(1) 59ndash73

Transport for London 2007 Tube Carries One Billion Passengers for First

Time wwwtflgovukcorporatemedianewscentre4770aspx

(accessed 24 March 2008)

Weller M 2007 London Underground private correspondence

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

120 G SMITH AND B CERANIC

components out of sequence that could lead to aninvalid solution

Once auto-generated layouts are complete withcorrectly sized and legitimately connected stationcomponents manual insertion of non-passenger areaswould then commence Upon completion thegenerated layout is transferred to AutoCAD via dxf fileformat as shown in Figure 16 Once in AutoCAD thelayout has a level of detail added by the designer suchas door openings passage connections fire escapeopenings etc so that passenger flows may beassessed It is then exported to Simulex for fireevacuation simulation analysis again via dxf file format

Simulex understands any voids as potentialcirculation routes treating the rest as solid andpermanent obstructions ie walls internal structuralelements permanent fixtures etc Since its solutionoperates in a two-dimensional environment for whatin effect is a three-dimensional problem the means

of vertical circulation (eg staircases) have to beestablished True lengths of staircases must bespecified based on the total length of all flights(sloping length) and the intermediate rests The fireexits need to be defined both in terms of theirlocation and width The links have to be establishedto define the point of the connection between thefloors and staircasesescalatorsramps The travellingdistance maps are then calculated type and numberof occupants specified building populated and theevacuation simulation performed

SYSTEM TESTING AND VALIDATIONKuligowski and Peacock (2005) define and recommendthe following ways of validating evacuation models

validation against code requirements validation against fire drills or other peoplersquos

movement experimentstrials

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 113

Gateline Vertical

Platform Right

Stairs Horizontal Stairs Vertical

Travelator Vertical

Ticket Hall

BMOAdmin Reception

Platform Top Ram Up Bottom Vertical

Lift Passage Horizontal

FIGURE 13 Examples of SGEvac station components

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

114 G SMITH AND B CERANIC

FIGURE 15 SGEvac hierarchy function

FIGURE 14 Layout module graphical user interface

validation against literature on past evacuationexperiments

validation against other models third-party validation

Furthermore Galea (1999) deconstructs the validationprocess into four main areas ndash component testingfunctional testing qualitative validation andquantitative validation This approach is more suitablefor models such as SGEvac than for those based oncompliance with code requirements and hence it isthe one adopted for the validation in this research

Component testing is regarded as testingcomputer code that drives the model checking forelementary errors Functional validation involveschecking that the model possesses the ability to

exhibit the range of capabilities required to performthe desired simulations Component testing andfunctional validation for SGEvac is carried outagainst code requirements insofar that allcalculations for infrastructure sizing are evaluatedmanually so that the method and answer are knownto be correct It uses the Simulex program to aid themodelling of people movement and to test thehypothesis of improvement to evacuation timesthus relying on Simulexrsquos own rigorous validationprocess Simulex has been used extensively inpractice as well as a validating model for otherevacuation softwares (Kuligowski and Peacock2005) SGEvac has gone through its initialcompletion and validation process with finalvalidation proposed for the end of the year

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 115

FIGURE 16 Auto-generated SGEvac station layout export file for fire

evacuation simulation in Simulex

FIRE EVACUATION SIMULATION RESULTSThe considered lsquotrain on fire in stationrsquo scenario takesinto account the service interval of 333 minutes(based on 18 trains per hour) the crush capacity of afour-car train together with the number of peoplewaiting on the platform after cancellation of oneservice (ie after 666 minutes) On the oppositeplatform the evacuation load is taken to be thenormal boardingalighting load assuming nocancellations to services (50 of all boarders andalighters)

Evacuation load predictions were based on TfL2006 projected passenger flows with an averageflow of 36 peopleminute for boarders and 123peopleminute for boarders and alighters combinedThe train crush capacity was averaged at 165peoplecar The simulation also models evacuationfrom the ticket hall level to assess the impact ofdispersing crowds at peak times in this part of thestation From these figures the total evacuation loadwas calculated as follows

Platform 1 Boarders (36 x 666 min) = 240Train crush load (165 x 4) = 660

Platform 2 Boarders and alighters

(123 x 05 x 333 min) = 205Ticket hall = 470Total station evacuation load = 1575 people

The simulation results have been captured as imagesat 25 50 100 and 300 second intervals respectivelyfor both platform and ticket hall levels as shown inFigures 17 to 20

The provided means of escape follow LU SPSG4requirements insofar that the maximum travellingdistance on a platform to an exit is no more than 45 m (ie exits must be no more than 90 m apart) andthat the dead-end travelling distance is no more than20 m to an exit Hence each platform is provided withthree exits The MIPsrsquo (mobility impaired persons)means of escape is provided from both ends ofplatforms via emergency evacuation lifts Totalevacuation time (ie all people reaching an exit onticket hallground level) was 5 minutes and

24 seconds with the platform and train load clearedwithin 3 minutes and 5 seconds This is within therequirements to evacuate the platform and train loadwithin 4 minutes with all people reaching a fire-separated route within 6 minutes (the ticket hall is notclassed as a fire-separated route)

The performed simulation was based on theminimum travelling distance maps resulting in eachperson exiting via the nearest fire exit route Theaverage 10 second lsquotime to start to moversquo wasassigned to all occupants in order to accommodatefor real world people behaviour commonly reportedby fire evacuation research studies The boarders onthe end of both platforms (including most peoplefrom end carriages of the train) evacuated viastaircasesMIP lifts directly to the ground level pointof escape The boarders situated around the middleof platforms together with the people from the twomiddle carriages used the central escalators to reachticket hall fire exits Additionally some people fromthe end carriages and some staffretail unitcustomers used this route as means of escape

In this lsquotrain on fire in stationrsquo scenario it wasassumed that none of the escapes routes areunavailable due to maintenance However furtherevacuation studies will investigate possible scenarioswhere one of the escalator routes may be unavailabledue to maintenance reasons

FUTURE RESEARCHIntended future work will concentrate on the furtherconditioning developing and modelling of shapegrammar design rules based on the outcomes of thefinal system validation and LU expertsrsquo feedbackFurther research is proposed on the development ofa novel layout optimization approach based on theshape grammar rules to investigate the evacuation-related parametric improvements in subsurfacesurface railway station design (eg minimizingtravelling distances)

The fire safety strategy in terms of performance-based fire risk design is an important aspirationalaspect of SGEvac requiring further research andliaison with experts to determine how such variedcriteria could be allied with layout planning This isparticularly important where deviation from standardsoccurs Thus further development of the expert

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

116 G SMITH AND B CERANIC

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 117

FIGURE 17 Station evacuation after 25 seconds

FIGURE 18 Station evacuation after 50 seconds

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

118 G SMITH AND B CERANIC

FIGURE 19 Station evacuation after 100 seconds

FIGURE 20 Station evacuation after 300 seconds

knowledge database would be required particularlyin the light of frequent changes to legislation

On a more practical note it could be beneficial todevelop the system on a stand-alone modular basislinked to each generated station component A list offire safety strategy objectives and functional logicstatements can then be specified for commonlyrecurring items (such as emergency lighting fire alarmdetectors sprinkler heads etc) and spaced accordingto the designerrsquos wishes or via shape rule formulation

Site constraint issues would also benefit fromfuture development as well as a move towardsdevelopment of the 3D shape grammar version Oncedevised other relevant 3D modelling systems such ascomputational fluid dynamics (CFD) smoke movementsoftware could be utilized in conjunction with SGEvac

CONCLUSIONSThe computer-aided modelling environment underdevelopment is intended to act as a layout planning aidfor those who practice in sub-surfacesurface railwaystation design The system under developmentcarefully considers London Underground StationPlanning Standards and Guidelines together withother relevant fire safety regulations that influenceoverall fire strategy design The proposed computer-aided design environment (SGEvac) is based on highlyrestrictive additive shape grammar rules driven by akernel of key passenger flow information to generatestation plan solutions

The major benefit to the design team is the ability toreview multiple layouts of accurately sized and correctlyplaced infrastructure at the reference or preliminarystage of the station layout planning design process Timeis hence saved regarding different option appraisals andquantitative testing using evacuation software shouldlead to a choice of more effective layouts in relation tomeans of escape travelling distances and improvementin the total evacuation times Modifications to layoutsshould be straightforward and there is scope toefficiently assess travelling distances and evacuationtimes for each of the considered alternatives comparedwith a lsquotraditionalrsquo approach

The proposed novel shape grammar approach ismost suited to new-build stations or largerefurbishment projects where more design flexibilityexists Sites with many site-specific constraints have

an array of predetermined parameters and will mostlikely benefit from the manual operation of SGEvac

Use in other rail systems due to recurringinfrastructure components and inherent functionallogic related to the laying out of the stationcomponents is also feasible provided that shapeconnections and shape rules are corrected to reflectthat countryrsquos legislative requirements

The potential for station layout improvementswith regard to emergency evacuation is an area thatmerits research especially in scenarios that concernthe safety of the general public and form a vital partof Londonrsquos transportation system Potential layoutimprovements could not only benefit evacuationsituations but also enhance the through-flow ofpeople during peak travelling periods

ACKNOWLEDGEMENTS The authors wish to express their gratitude to JacobsEngineering (UK) Ltd which has sponsored theresearch and to London Underground Ltd forproviding the design guidance that is reproduced inthis paper and for authorising the use of the stationplans London Underground has not reviewed thispaper and the content is solely the authorsrsquo work Mr Robert Hutchinson is also thanked for hiscomputer programming assistance

AUTHOR CONTACT DETAILSGGrraahhaamm SSmmiitthh (corresponding author) Senior Architectural

Technologist Jacobs (UK) Ltd Cardinal Square North Point

10 Nottingham Road Derby DE1 3QT UK Tel +44 (0)1332 234482

fax +44 (0)1332 344558 e-mail grahampsmithjacobscomBBoorriiss CCeerraanniicc Programme Leader for BScBA (Hons) Architectural

Programmes Room Number 105 Faculty of Arts Design and

Technology University of Derby Markeaton Street Derby DE22

3AW UK Tel +44 (0)1332 593136 e-mail bceranicderbyacuk

REFERENCES BS 7974 2001 Application of Fire Safety Engineering Principles to the

Design of Buildings London BSI

Cha MY and Gero JS 1999 Style Learning Inductive Generalisation of

Architectural Shape Patterns Paichai University and Sydney University

co-paper

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 119

Chase SC 1996 lsquoUsing logic to specify shapes and spatial relations in

design grammarsrsquo in Workshop notes Grammatical Design Fourth

International Conference on Artificial Intelligence in Design Stanford

University US June

Department of the Environment 1996 Design Principles of Fire Safety

London The Stationery Office Books

Flemming U 1987 lsquoMore than the sum of its parts the grammar of

Queen Anne housesrsquo in Environment and Planning B Planning and

Design 14 323ndash350

Fruin JJ 1971 Pedestrian Planning and Design New York Metropolitan

Association of Urban Designers and Environmental Planners

Galea ER 1999 lsquoA general approach to validating evacuation models with

an application to EXODUSrsquo in Journal of Fire Sciences 16 414ndash436

Galle P and Kovacs LB 1992 lsquoThe logic of worms a study in

architectural knowledge representationrsquo in Environment and Planning B

Planning and Design 19 5ndash31

Garling T Lindberg E and Mantyla T 1983 lsquoOrientation in buildings ndash

effects of familiarity visual access and orientation aidsrsquo in Journal of

Applied Psychology 68(1) 177ndash186

Grimsdale RL and Chang CW 1996 lsquoThe layout design language A

technique for generating layout plansrsquo in Computer Graphics Forum

15(2) 97ndash106

Heskestad AW 1999 lsquoPerformance in smoke of way guidance systemsrsquo

in Fire and Materials 23(6) 375ndash381

Knight T 1995 lsquoConstructive symmetryrsquo in Environment and Planning B

Planning and Design 22 419ndash450

Knight TW 1999 lsquoShape grammars six typesrsquo in Environment and

Planning B Planning and Design 26 15ndash31

Kuligowski ED and Peacock RD 2005 A Review of Building Evacuation

Models Technical Note 1471 National Institute of Standards and

Technology (NIST) Technology Administration US Department of

Commerce

Li A I-Kang 2003 lsquoThe Yingzao fashi in the information agersquo conference

paper for the Beaux-Arts Paul Philippe Cret and Twentieth Century

Architecture in China The University of Pennsylvania October

Liew H 2003 lsquoSGML ndash A shape grammar meta-languagersquo in Digital

Design 21st eCAADe Conference Proceedings Graz Austria September

639ndash648

Loomis B 2003 lsquoA note on generative design techniques SGGA a user

driven genetic algorithm for evolving non-deterministic shape grammarsrsquo

working paper Massachusetts Institute of Technology (MIT)

Magnusson SE Drysdale DD Fitzgerald RW Mowrer F

Quintiere JG Williamson RB and Zalosh RG 1995

lsquoA proposal for a model curriculum in fire safety engineeringrsquo

in Fire Safety Journal 25 9ndash10

Smith G Ceranic B and Fryer C 2003 lsquoDevelopment of CAD

environment for spatial layout optimisation in sub-surfacesurface rail

station designrsquo in Proceedings of International BIAT Conference The

British Museum London January 2003

SPSG 2007 London Underground Station Planning Standards and

Guidelines SPSG4 (1ndash371) London

Stiny G and Gips J 1971 lsquoShape grammars and the generative

specification of painting and sculpturersquo in Information Processing 71

1460ndash1465

Stiny G 1977 lsquoIce-ray a note on the generation of Chinese lattice

designsrsquo in Environment and Planning B Planning and Design 4 89ndash98

Stiny G 1980 lsquoKindergarten grammars designing with Froebelrsquos building

giftsrsquo in Environment and Planning B 3 409ndash462

Tapia M 1999 lsquoA visual implementation of a shape grammar systemrsquo in

Environment and Planning B Planning and Design 26(1) 59ndash73

Transport for London 2007 Tube Carries One Billion Passengers for First

Time wwwtflgovukcorporatemedianewscentre4770aspx

(accessed 24 March 2008)

Weller M 2007 London Underground private correspondence

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

120 G SMITH AND B CERANIC

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

114 G SMITH AND B CERANIC

FIGURE 15 SGEvac hierarchy function

FIGURE 14 Layout module graphical user interface

validation against literature on past evacuationexperiments

validation against other models third-party validation

Furthermore Galea (1999) deconstructs the validationprocess into four main areas ndash component testingfunctional testing qualitative validation andquantitative validation This approach is more suitablefor models such as SGEvac than for those based oncompliance with code requirements and hence it isthe one adopted for the validation in this research

Component testing is regarded as testingcomputer code that drives the model checking forelementary errors Functional validation involveschecking that the model possesses the ability to

exhibit the range of capabilities required to performthe desired simulations Component testing andfunctional validation for SGEvac is carried outagainst code requirements insofar that allcalculations for infrastructure sizing are evaluatedmanually so that the method and answer are knownto be correct It uses the Simulex program to aid themodelling of people movement and to test thehypothesis of improvement to evacuation timesthus relying on Simulexrsquos own rigorous validationprocess Simulex has been used extensively inpractice as well as a validating model for otherevacuation softwares (Kuligowski and Peacock2005) SGEvac has gone through its initialcompletion and validation process with finalvalidation proposed for the end of the year

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 115

FIGURE 16 Auto-generated SGEvac station layout export file for fire

evacuation simulation in Simulex

FIRE EVACUATION SIMULATION RESULTSThe considered lsquotrain on fire in stationrsquo scenario takesinto account the service interval of 333 minutes(based on 18 trains per hour) the crush capacity of afour-car train together with the number of peoplewaiting on the platform after cancellation of oneservice (ie after 666 minutes) On the oppositeplatform the evacuation load is taken to be thenormal boardingalighting load assuming nocancellations to services (50 of all boarders andalighters)

Evacuation load predictions were based on TfL2006 projected passenger flows with an averageflow of 36 peopleminute for boarders and 123peopleminute for boarders and alighters combinedThe train crush capacity was averaged at 165peoplecar The simulation also models evacuationfrom the ticket hall level to assess the impact ofdispersing crowds at peak times in this part of thestation From these figures the total evacuation loadwas calculated as follows

Platform 1 Boarders (36 x 666 min) = 240Train crush load (165 x 4) = 660

Platform 2 Boarders and alighters

(123 x 05 x 333 min) = 205Ticket hall = 470Total station evacuation load = 1575 people

The simulation results have been captured as imagesat 25 50 100 and 300 second intervals respectivelyfor both platform and ticket hall levels as shown inFigures 17 to 20

The provided means of escape follow LU SPSG4requirements insofar that the maximum travellingdistance on a platform to an exit is no more than 45 m (ie exits must be no more than 90 m apart) andthat the dead-end travelling distance is no more than20 m to an exit Hence each platform is provided withthree exits The MIPsrsquo (mobility impaired persons)means of escape is provided from both ends ofplatforms via emergency evacuation lifts Totalevacuation time (ie all people reaching an exit onticket hallground level) was 5 minutes and

24 seconds with the platform and train load clearedwithin 3 minutes and 5 seconds This is within therequirements to evacuate the platform and train loadwithin 4 minutes with all people reaching a fire-separated route within 6 minutes (the ticket hall is notclassed as a fire-separated route)

The performed simulation was based on theminimum travelling distance maps resulting in eachperson exiting via the nearest fire exit route Theaverage 10 second lsquotime to start to moversquo wasassigned to all occupants in order to accommodatefor real world people behaviour commonly reportedby fire evacuation research studies The boarders onthe end of both platforms (including most peoplefrom end carriages of the train) evacuated viastaircasesMIP lifts directly to the ground level pointof escape The boarders situated around the middleof platforms together with the people from the twomiddle carriages used the central escalators to reachticket hall fire exits Additionally some people fromthe end carriages and some staffretail unitcustomers used this route as means of escape

In this lsquotrain on fire in stationrsquo scenario it wasassumed that none of the escapes routes areunavailable due to maintenance However furtherevacuation studies will investigate possible scenarioswhere one of the escalator routes may be unavailabledue to maintenance reasons

FUTURE RESEARCHIntended future work will concentrate on the furtherconditioning developing and modelling of shapegrammar design rules based on the outcomes of thefinal system validation and LU expertsrsquo feedbackFurther research is proposed on the development ofa novel layout optimization approach based on theshape grammar rules to investigate the evacuation-related parametric improvements in subsurfacesurface railway station design (eg minimizingtravelling distances)

The fire safety strategy in terms of performance-based fire risk design is an important aspirationalaspect of SGEvac requiring further research andliaison with experts to determine how such variedcriteria could be allied with layout planning This isparticularly important where deviation from standardsoccurs Thus further development of the expert

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

116 G SMITH AND B CERANIC

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 117

FIGURE 17 Station evacuation after 25 seconds

FIGURE 18 Station evacuation after 50 seconds

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

118 G SMITH AND B CERANIC

FIGURE 19 Station evacuation after 100 seconds

FIGURE 20 Station evacuation after 300 seconds

knowledge database would be required particularlyin the light of frequent changes to legislation

On a more practical note it could be beneficial todevelop the system on a stand-alone modular basislinked to each generated station component A list offire safety strategy objectives and functional logicstatements can then be specified for commonlyrecurring items (such as emergency lighting fire alarmdetectors sprinkler heads etc) and spaced accordingto the designerrsquos wishes or via shape rule formulation

Site constraint issues would also benefit fromfuture development as well as a move towardsdevelopment of the 3D shape grammar version Oncedevised other relevant 3D modelling systems such ascomputational fluid dynamics (CFD) smoke movementsoftware could be utilized in conjunction with SGEvac

CONCLUSIONSThe computer-aided modelling environment underdevelopment is intended to act as a layout planning aidfor those who practice in sub-surfacesurface railwaystation design The system under developmentcarefully considers London Underground StationPlanning Standards and Guidelines together withother relevant fire safety regulations that influenceoverall fire strategy design The proposed computer-aided design environment (SGEvac) is based on highlyrestrictive additive shape grammar rules driven by akernel of key passenger flow information to generatestation plan solutions

The major benefit to the design team is the ability toreview multiple layouts of accurately sized and correctlyplaced infrastructure at the reference or preliminarystage of the station layout planning design process Timeis hence saved regarding different option appraisals andquantitative testing using evacuation software shouldlead to a choice of more effective layouts in relation tomeans of escape travelling distances and improvementin the total evacuation times Modifications to layoutsshould be straightforward and there is scope toefficiently assess travelling distances and evacuationtimes for each of the considered alternatives comparedwith a lsquotraditionalrsquo approach

The proposed novel shape grammar approach ismost suited to new-build stations or largerefurbishment projects where more design flexibilityexists Sites with many site-specific constraints have

an array of predetermined parameters and will mostlikely benefit from the manual operation of SGEvac

Use in other rail systems due to recurringinfrastructure components and inherent functionallogic related to the laying out of the stationcomponents is also feasible provided that shapeconnections and shape rules are corrected to reflectthat countryrsquos legislative requirements

The potential for station layout improvementswith regard to emergency evacuation is an area thatmerits research especially in scenarios that concernthe safety of the general public and form a vital partof Londonrsquos transportation system Potential layoutimprovements could not only benefit evacuationsituations but also enhance the through-flow ofpeople during peak travelling periods

ACKNOWLEDGEMENTS The authors wish to express their gratitude to JacobsEngineering (UK) Ltd which has sponsored theresearch and to London Underground Ltd forproviding the design guidance that is reproduced inthis paper and for authorising the use of the stationplans London Underground has not reviewed thispaper and the content is solely the authorsrsquo work Mr Robert Hutchinson is also thanked for hiscomputer programming assistance

AUTHOR CONTACT DETAILSGGrraahhaamm SSmmiitthh (corresponding author) Senior Architectural

Technologist Jacobs (UK) Ltd Cardinal Square North Point

10 Nottingham Road Derby DE1 3QT UK Tel +44 (0)1332 234482

fax +44 (0)1332 344558 e-mail grahampsmithjacobscomBBoorriiss CCeerraanniicc Programme Leader for BScBA (Hons) Architectural

Programmes Room Number 105 Faculty of Arts Design and

Technology University of Derby Markeaton Street Derby DE22

3AW UK Tel +44 (0)1332 593136 e-mail bceranicderbyacuk

REFERENCES BS 7974 2001 Application of Fire Safety Engineering Principles to the

Design of Buildings London BSI

Cha MY and Gero JS 1999 Style Learning Inductive Generalisation of

Architectural Shape Patterns Paichai University and Sydney University

co-paper

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 119

Chase SC 1996 lsquoUsing logic to specify shapes and spatial relations in

design grammarsrsquo in Workshop notes Grammatical Design Fourth

International Conference on Artificial Intelligence in Design Stanford

University US June

Department of the Environment 1996 Design Principles of Fire Safety

London The Stationery Office Books

Flemming U 1987 lsquoMore than the sum of its parts the grammar of

Queen Anne housesrsquo in Environment and Planning B Planning and

Design 14 323ndash350

Fruin JJ 1971 Pedestrian Planning and Design New York Metropolitan

Association of Urban Designers and Environmental Planners

Galea ER 1999 lsquoA general approach to validating evacuation models with

an application to EXODUSrsquo in Journal of Fire Sciences 16 414ndash436

Galle P and Kovacs LB 1992 lsquoThe logic of worms a study in

architectural knowledge representationrsquo in Environment and Planning B

Planning and Design 19 5ndash31

Garling T Lindberg E and Mantyla T 1983 lsquoOrientation in buildings ndash

effects of familiarity visual access and orientation aidsrsquo in Journal of

Applied Psychology 68(1) 177ndash186

Grimsdale RL and Chang CW 1996 lsquoThe layout design language A

technique for generating layout plansrsquo in Computer Graphics Forum

15(2) 97ndash106

Heskestad AW 1999 lsquoPerformance in smoke of way guidance systemsrsquo

in Fire and Materials 23(6) 375ndash381

Knight T 1995 lsquoConstructive symmetryrsquo in Environment and Planning B

Planning and Design 22 419ndash450

Knight TW 1999 lsquoShape grammars six typesrsquo in Environment and

Planning B Planning and Design 26 15ndash31

Kuligowski ED and Peacock RD 2005 A Review of Building Evacuation

Models Technical Note 1471 National Institute of Standards and

Technology (NIST) Technology Administration US Department of

Commerce

Li A I-Kang 2003 lsquoThe Yingzao fashi in the information agersquo conference

paper for the Beaux-Arts Paul Philippe Cret and Twentieth Century

Architecture in China The University of Pennsylvania October

Liew H 2003 lsquoSGML ndash A shape grammar meta-languagersquo in Digital

Design 21st eCAADe Conference Proceedings Graz Austria September

639ndash648

Loomis B 2003 lsquoA note on generative design techniques SGGA a user

driven genetic algorithm for evolving non-deterministic shape grammarsrsquo

working paper Massachusetts Institute of Technology (MIT)

Magnusson SE Drysdale DD Fitzgerald RW Mowrer F

Quintiere JG Williamson RB and Zalosh RG 1995

lsquoA proposal for a model curriculum in fire safety engineeringrsquo

in Fire Safety Journal 25 9ndash10

Smith G Ceranic B and Fryer C 2003 lsquoDevelopment of CAD

environment for spatial layout optimisation in sub-surfacesurface rail

station designrsquo in Proceedings of International BIAT Conference The

British Museum London January 2003

SPSG 2007 London Underground Station Planning Standards and

Guidelines SPSG4 (1ndash371) London

Stiny G and Gips J 1971 lsquoShape grammars and the generative

specification of painting and sculpturersquo in Information Processing 71

1460ndash1465

Stiny G 1977 lsquoIce-ray a note on the generation of Chinese lattice

designsrsquo in Environment and Planning B Planning and Design 4 89ndash98

Stiny G 1980 lsquoKindergarten grammars designing with Froebelrsquos building

giftsrsquo in Environment and Planning B 3 409ndash462

Tapia M 1999 lsquoA visual implementation of a shape grammar systemrsquo in

Environment and Planning B Planning and Design 26(1) 59ndash73

Transport for London 2007 Tube Carries One Billion Passengers for First

Time wwwtflgovukcorporatemedianewscentre4770aspx

(accessed 24 March 2008)

Weller M 2007 London Underground private correspondence

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

120 G SMITH AND B CERANIC

validation against literature on past evacuationexperiments

validation against other models third-party validation

Furthermore Galea (1999) deconstructs the validationprocess into four main areas ndash component testingfunctional testing qualitative validation andquantitative validation This approach is more suitablefor models such as SGEvac than for those based oncompliance with code requirements and hence it isthe one adopted for the validation in this research

Component testing is regarded as testingcomputer code that drives the model checking forelementary errors Functional validation involveschecking that the model possesses the ability to

exhibit the range of capabilities required to performthe desired simulations Component testing andfunctional validation for SGEvac is carried outagainst code requirements insofar that allcalculations for infrastructure sizing are evaluatedmanually so that the method and answer are knownto be correct It uses the Simulex program to aid themodelling of people movement and to test thehypothesis of improvement to evacuation timesthus relying on Simulexrsquos own rigorous validationprocess Simulex has been used extensively inpractice as well as a validating model for otherevacuation softwares (Kuligowski and Peacock2005) SGEvac has gone through its initialcompletion and validation process with finalvalidation proposed for the end of the year

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 115

FIGURE 16 Auto-generated SGEvac station layout export file for fire

evacuation simulation in Simulex

FIRE EVACUATION SIMULATION RESULTSThe considered lsquotrain on fire in stationrsquo scenario takesinto account the service interval of 333 minutes(based on 18 trains per hour) the crush capacity of afour-car train together with the number of peoplewaiting on the platform after cancellation of oneservice (ie after 666 minutes) On the oppositeplatform the evacuation load is taken to be thenormal boardingalighting load assuming nocancellations to services (50 of all boarders andalighters)

Evacuation load predictions were based on TfL2006 projected passenger flows with an averageflow of 36 peopleminute for boarders and 123peopleminute for boarders and alighters combinedThe train crush capacity was averaged at 165peoplecar The simulation also models evacuationfrom the ticket hall level to assess the impact ofdispersing crowds at peak times in this part of thestation From these figures the total evacuation loadwas calculated as follows

Platform 1 Boarders (36 x 666 min) = 240Train crush load (165 x 4) = 660

Platform 2 Boarders and alighters

(123 x 05 x 333 min) = 205Ticket hall = 470Total station evacuation load = 1575 people

The simulation results have been captured as imagesat 25 50 100 and 300 second intervals respectivelyfor both platform and ticket hall levels as shown inFigures 17 to 20

The provided means of escape follow LU SPSG4requirements insofar that the maximum travellingdistance on a platform to an exit is no more than 45 m (ie exits must be no more than 90 m apart) andthat the dead-end travelling distance is no more than20 m to an exit Hence each platform is provided withthree exits The MIPsrsquo (mobility impaired persons)means of escape is provided from both ends ofplatforms via emergency evacuation lifts Totalevacuation time (ie all people reaching an exit onticket hallground level) was 5 minutes and

24 seconds with the platform and train load clearedwithin 3 minutes and 5 seconds This is within therequirements to evacuate the platform and train loadwithin 4 minutes with all people reaching a fire-separated route within 6 minutes (the ticket hall is notclassed as a fire-separated route)

The performed simulation was based on theminimum travelling distance maps resulting in eachperson exiting via the nearest fire exit route Theaverage 10 second lsquotime to start to moversquo wasassigned to all occupants in order to accommodatefor real world people behaviour commonly reportedby fire evacuation research studies The boarders onthe end of both platforms (including most peoplefrom end carriages of the train) evacuated viastaircasesMIP lifts directly to the ground level pointof escape The boarders situated around the middleof platforms together with the people from the twomiddle carriages used the central escalators to reachticket hall fire exits Additionally some people fromthe end carriages and some staffretail unitcustomers used this route as means of escape

In this lsquotrain on fire in stationrsquo scenario it wasassumed that none of the escapes routes areunavailable due to maintenance However furtherevacuation studies will investigate possible scenarioswhere one of the escalator routes may be unavailabledue to maintenance reasons

FUTURE RESEARCHIntended future work will concentrate on the furtherconditioning developing and modelling of shapegrammar design rules based on the outcomes of thefinal system validation and LU expertsrsquo feedbackFurther research is proposed on the development ofa novel layout optimization approach based on theshape grammar rules to investigate the evacuation-related parametric improvements in subsurfacesurface railway station design (eg minimizingtravelling distances)

The fire safety strategy in terms of performance-based fire risk design is an important aspirationalaspect of SGEvac requiring further research andliaison with experts to determine how such variedcriteria could be allied with layout planning This isparticularly important where deviation from standardsoccurs Thus further development of the expert

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

116 G SMITH AND B CERANIC

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 117

FIGURE 17 Station evacuation after 25 seconds

FIGURE 18 Station evacuation after 50 seconds

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

118 G SMITH AND B CERANIC

FIGURE 19 Station evacuation after 100 seconds

FIGURE 20 Station evacuation after 300 seconds

knowledge database would be required particularlyin the light of frequent changes to legislation

On a more practical note it could be beneficial todevelop the system on a stand-alone modular basislinked to each generated station component A list offire safety strategy objectives and functional logicstatements can then be specified for commonlyrecurring items (such as emergency lighting fire alarmdetectors sprinkler heads etc) and spaced accordingto the designerrsquos wishes or via shape rule formulation

Site constraint issues would also benefit fromfuture development as well as a move towardsdevelopment of the 3D shape grammar version Oncedevised other relevant 3D modelling systems such ascomputational fluid dynamics (CFD) smoke movementsoftware could be utilized in conjunction with SGEvac

CONCLUSIONSThe computer-aided modelling environment underdevelopment is intended to act as a layout planning aidfor those who practice in sub-surfacesurface railwaystation design The system under developmentcarefully considers London Underground StationPlanning Standards and Guidelines together withother relevant fire safety regulations that influenceoverall fire strategy design The proposed computer-aided design environment (SGEvac) is based on highlyrestrictive additive shape grammar rules driven by akernel of key passenger flow information to generatestation plan solutions

The major benefit to the design team is the ability toreview multiple layouts of accurately sized and correctlyplaced infrastructure at the reference or preliminarystage of the station layout planning design process Timeis hence saved regarding different option appraisals andquantitative testing using evacuation software shouldlead to a choice of more effective layouts in relation tomeans of escape travelling distances and improvementin the total evacuation times Modifications to layoutsshould be straightforward and there is scope toefficiently assess travelling distances and evacuationtimes for each of the considered alternatives comparedwith a lsquotraditionalrsquo approach

The proposed novel shape grammar approach ismost suited to new-build stations or largerefurbishment projects where more design flexibilityexists Sites with many site-specific constraints have

an array of predetermined parameters and will mostlikely benefit from the manual operation of SGEvac

Use in other rail systems due to recurringinfrastructure components and inherent functionallogic related to the laying out of the stationcomponents is also feasible provided that shapeconnections and shape rules are corrected to reflectthat countryrsquos legislative requirements

The potential for station layout improvementswith regard to emergency evacuation is an area thatmerits research especially in scenarios that concernthe safety of the general public and form a vital partof Londonrsquos transportation system Potential layoutimprovements could not only benefit evacuationsituations but also enhance the through-flow ofpeople during peak travelling periods

ACKNOWLEDGEMENTS The authors wish to express their gratitude to JacobsEngineering (UK) Ltd which has sponsored theresearch and to London Underground Ltd forproviding the design guidance that is reproduced inthis paper and for authorising the use of the stationplans London Underground has not reviewed thispaper and the content is solely the authorsrsquo work Mr Robert Hutchinson is also thanked for hiscomputer programming assistance

AUTHOR CONTACT DETAILSGGrraahhaamm SSmmiitthh (corresponding author) Senior Architectural

Technologist Jacobs (UK) Ltd Cardinal Square North Point

10 Nottingham Road Derby DE1 3QT UK Tel +44 (0)1332 234482

fax +44 (0)1332 344558 e-mail grahampsmithjacobscomBBoorriiss CCeerraanniicc Programme Leader for BScBA (Hons) Architectural

Programmes Room Number 105 Faculty of Arts Design and

Technology University of Derby Markeaton Street Derby DE22

3AW UK Tel +44 (0)1332 593136 e-mail bceranicderbyacuk

REFERENCES BS 7974 2001 Application of Fire Safety Engineering Principles to the

Design of Buildings London BSI

Cha MY and Gero JS 1999 Style Learning Inductive Generalisation of

Architectural Shape Patterns Paichai University and Sydney University

co-paper

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 119

Chase SC 1996 lsquoUsing logic to specify shapes and spatial relations in

design grammarsrsquo in Workshop notes Grammatical Design Fourth

International Conference on Artificial Intelligence in Design Stanford

University US June

Department of the Environment 1996 Design Principles of Fire Safety

London The Stationery Office Books

Flemming U 1987 lsquoMore than the sum of its parts the grammar of

Queen Anne housesrsquo in Environment and Planning B Planning and

Design 14 323ndash350

Fruin JJ 1971 Pedestrian Planning and Design New York Metropolitan

Association of Urban Designers and Environmental Planners

Galea ER 1999 lsquoA general approach to validating evacuation models with

an application to EXODUSrsquo in Journal of Fire Sciences 16 414ndash436

Galle P and Kovacs LB 1992 lsquoThe logic of worms a study in

architectural knowledge representationrsquo in Environment and Planning B

Planning and Design 19 5ndash31

Garling T Lindberg E and Mantyla T 1983 lsquoOrientation in buildings ndash

effects of familiarity visual access and orientation aidsrsquo in Journal of

Applied Psychology 68(1) 177ndash186

Grimsdale RL and Chang CW 1996 lsquoThe layout design language A

technique for generating layout plansrsquo in Computer Graphics Forum

15(2) 97ndash106

Heskestad AW 1999 lsquoPerformance in smoke of way guidance systemsrsquo

in Fire and Materials 23(6) 375ndash381

Knight T 1995 lsquoConstructive symmetryrsquo in Environment and Planning B

Planning and Design 22 419ndash450

Knight TW 1999 lsquoShape grammars six typesrsquo in Environment and

Planning B Planning and Design 26 15ndash31

Kuligowski ED and Peacock RD 2005 A Review of Building Evacuation

Models Technical Note 1471 National Institute of Standards and

Technology (NIST) Technology Administration US Department of

Commerce

Li A I-Kang 2003 lsquoThe Yingzao fashi in the information agersquo conference

paper for the Beaux-Arts Paul Philippe Cret and Twentieth Century

Architecture in China The University of Pennsylvania October

Liew H 2003 lsquoSGML ndash A shape grammar meta-languagersquo in Digital

Design 21st eCAADe Conference Proceedings Graz Austria September

639ndash648

Loomis B 2003 lsquoA note on generative design techniques SGGA a user

driven genetic algorithm for evolving non-deterministic shape grammarsrsquo

working paper Massachusetts Institute of Technology (MIT)

Magnusson SE Drysdale DD Fitzgerald RW Mowrer F

Quintiere JG Williamson RB and Zalosh RG 1995

lsquoA proposal for a model curriculum in fire safety engineeringrsquo

in Fire Safety Journal 25 9ndash10

Smith G Ceranic B and Fryer C 2003 lsquoDevelopment of CAD

environment for spatial layout optimisation in sub-surfacesurface rail

station designrsquo in Proceedings of International BIAT Conference The

British Museum London January 2003

SPSG 2007 London Underground Station Planning Standards and

Guidelines SPSG4 (1ndash371) London

Stiny G and Gips J 1971 lsquoShape grammars and the generative

specification of painting and sculpturersquo in Information Processing 71

1460ndash1465

Stiny G 1977 lsquoIce-ray a note on the generation of Chinese lattice

designsrsquo in Environment and Planning B Planning and Design 4 89ndash98

Stiny G 1980 lsquoKindergarten grammars designing with Froebelrsquos building

giftsrsquo in Environment and Planning B 3 409ndash462

Tapia M 1999 lsquoA visual implementation of a shape grammar systemrsquo in

Environment and Planning B Planning and Design 26(1) 59ndash73

Transport for London 2007 Tube Carries One Billion Passengers for First

Time wwwtflgovukcorporatemedianewscentre4770aspx

(accessed 24 March 2008)

Weller M 2007 London Underground private correspondence

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

120 G SMITH AND B CERANIC

FIRE EVACUATION SIMULATION RESULTSThe considered lsquotrain on fire in stationrsquo scenario takesinto account the service interval of 333 minutes(based on 18 trains per hour) the crush capacity of afour-car train together with the number of peoplewaiting on the platform after cancellation of oneservice (ie after 666 minutes) On the oppositeplatform the evacuation load is taken to be thenormal boardingalighting load assuming nocancellations to services (50 of all boarders andalighters)

Evacuation load predictions were based on TfL2006 projected passenger flows with an averageflow of 36 peopleminute for boarders and 123peopleminute for boarders and alighters combinedThe train crush capacity was averaged at 165peoplecar The simulation also models evacuationfrom the ticket hall level to assess the impact ofdispersing crowds at peak times in this part of thestation From these figures the total evacuation loadwas calculated as follows

Platform 1 Boarders (36 x 666 min) = 240Train crush load (165 x 4) = 660

Platform 2 Boarders and alighters

(123 x 05 x 333 min) = 205Ticket hall = 470Total station evacuation load = 1575 people

The simulation results have been captured as imagesat 25 50 100 and 300 second intervals respectivelyfor both platform and ticket hall levels as shown inFigures 17 to 20

The provided means of escape follow LU SPSG4requirements insofar that the maximum travellingdistance on a platform to an exit is no more than 45 m (ie exits must be no more than 90 m apart) andthat the dead-end travelling distance is no more than20 m to an exit Hence each platform is provided withthree exits The MIPsrsquo (mobility impaired persons)means of escape is provided from both ends ofplatforms via emergency evacuation lifts Totalevacuation time (ie all people reaching an exit onticket hallground level) was 5 minutes and

24 seconds with the platform and train load clearedwithin 3 minutes and 5 seconds This is within therequirements to evacuate the platform and train loadwithin 4 minutes with all people reaching a fire-separated route within 6 minutes (the ticket hall is notclassed as a fire-separated route)

The performed simulation was based on theminimum travelling distance maps resulting in eachperson exiting via the nearest fire exit route Theaverage 10 second lsquotime to start to moversquo wasassigned to all occupants in order to accommodatefor real world people behaviour commonly reportedby fire evacuation research studies The boarders onthe end of both platforms (including most peoplefrom end carriages of the train) evacuated viastaircasesMIP lifts directly to the ground level pointof escape The boarders situated around the middleof platforms together with the people from the twomiddle carriages used the central escalators to reachticket hall fire exits Additionally some people fromthe end carriages and some staffretail unitcustomers used this route as means of escape

In this lsquotrain on fire in stationrsquo scenario it wasassumed that none of the escapes routes areunavailable due to maintenance However furtherevacuation studies will investigate possible scenarioswhere one of the escalator routes may be unavailabledue to maintenance reasons

FUTURE RESEARCHIntended future work will concentrate on the furtherconditioning developing and modelling of shapegrammar design rules based on the outcomes of thefinal system validation and LU expertsrsquo feedbackFurther research is proposed on the development ofa novel layout optimization approach based on theshape grammar rules to investigate the evacuation-related parametric improvements in subsurfacesurface railway station design (eg minimizingtravelling distances)

The fire safety strategy in terms of performance-based fire risk design is an important aspirationalaspect of SGEvac requiring further research andliaison with experts to determine how such variedcriteria could be allied with layout planning This isparticularly important where deviation from standardsoccurs Thus further development of the expert

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

116 G SMITH AND B CERANIC

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 117

FIGURE 17 Station evacuation after 25 seconds

FIGURE 18 Station evacuation after 50 seconds

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

118 G SMITH AND B CERANIC

FIGURE 19 Station evacuation after 100 seconds

FIGURE 20 Station evacuation after 300 seconds

knowledge database would be required particularlyin the light of frequent changes to legislation

On a more practical note it could be beneficial todevelop the system on a stand-alone modular basislinked to each generated station component A list offire safety strategy objectives and functional logicstatements can then be specified for commonlyrecurring items (such as emergency lighting fire alarmdetectors sprinkler heads etc) and spaced accordingto the designerrsquos wishes or via shape rule formulation

Site constraint issues would also benefit fromfuture development as well as a move towardsdevelopment of the 3D shape grammar version Oncedevised other relevant 3D modelling systems such ascomputational fluid dynamics (CFD) smoke movementsoftware could be utilized in conjunction with SGEvac

CONCLUSIONSThe computer-aided modelling environment underdevelopment is intended to act as a layout planning aidfor those who practice in sub-surfacesurface railwaystation design The system under developmentcarefully considers London Underground StationPlanning Standards and Guidelines together withother relevant fire safety regulations that influenceoverall fire strategy design The proposed computer-aided design environment (SGEvac) is based on highlyrestrictive additive shape grammar rules driven by akernel of key passenger flow information to generatestation plan solutions

The major benefit to the design team is the ability toreview multiple layouts of accurately sized and correctlyplaced infrastructure at the reference or preliminarystage of the station layout planning design process Timeis hence saved regarding different option appraisals andquantitative testing using evacuation software shouldlead to a choice of more effective layouts in relation tomeans of escape travelling distances and improvementin the total evacuation times Modifications to layoutsshould be straightforward and there is scope toefficiently assess travelling distances and evacuationtimes for each of the considered alternatives comparedwith a lsquotraditionalrsquo approach

The proposed novel shape grammar approach ismost suited to new-build stations or largerefurbishment projects where more design flexibilityexists Sites with many site-specific constraints have

an array of predetermined parameters and will mostlikely benefit from the manual operation of SGEvac

Use in other rail systems due to recurringinfrastructure components and inherent functionallogic related to the laying out of the stationcomponents is also feasible provided that shapeconnections and shape rules are corrected to reflectthat countryrsquos legislative requirements

The potential for station layout improvementswith regard to emergency evacuation is an area thatmerits research especially in scenarios that concernthe safety of the general public and form a vital partof Londonrsquos transportation system Potential layoutimprovements could not only benefit evacuationsituations but also enhance the through-flow ofpeople during peak travelling periods

ACKNOWLEDGEMENTS The authors wish to express their gratitude to JacobsEngineering (UK) Ltd which has sponsored theresearch and to London Underground Ltd forproviding the design guidance that is reproduced inthis paper and for authorising the use of the stationplans London Underground has not reviewed thispaper and the content is solely the authorsrsquo work Mr Robert Hutchinson is also thanked for hiscomputer programming assistance

AUTHOR CONTACT DETAILSGGrraahhaamm SSmmiitthh (corresponding author) Senior Architectural

Technologist Jacobs (UK) Ltd Cardinal Square North Point

10 Nottingham Road Derby DE1 3QT UK Tel +44 (0)1332 234482

fax +44 (0)1332 344558 e-mail grahampsmithjacobscomBBoorriiss CCeerraanniicc Programme Leader for BScBA (Hons) Architectural

Programmes Room Number 105 Faculty of Arts Design and

Technology University of Derby Markeaton Street Derby DE22

3AW UK Tel +44 (0)1332 593136 e-mail bceranicderbyacuk

REFERENCES BS 7974 2001 Application of Fire Safety Engineering Principles to the

Design of Buildings London BSI

Cha MY and Gero JS 1999 Style Learning Inductive Generalisation of

Architectural Shape Patterns Paichai University and Sydney University

co-paper

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 119

Chase SC 1996 lsquoUsing logic to specify shapes and spatial relations in

design grammarsrsquo in Workshop notes Grammatical Design Fourth

International Conference on Artificial Intelligence in Design Stanford

University US June

Department of the Environment 1996 Design Principles of Fire Safety

London The Stationery Office Books

Flemming U 1987 lsquoMore than the sum of its parts the grammar of

Queen Anne housesrsquo in Environment and Planning B Planning and

Design 14 323ndash350

Fruin JJ 1971 Pedestrian Planning and Design New York Metropolitan

Association of Urban Designers and Environmental Planners

Galea ER 1999 lsquoA general approach to validating evacuation models with

an application to EXODUSrsquo in Journal of Fire Sciences 16 414ndash436

Galle P and Kovacs LB 1992 lsquoThe logic of worms a study in

architectural knowledge representationrsquo in Environment and Planning B

Planning and Design 19 5ndash31

Garling T Lindberg E and Mantyla T 1983 lsquoOrientation in buildings ndash

effects of familiarity visual access and orientation aidsrsquo in Journal of

Applied Psychology 68(1) 177ndash186

Grimsdale RL and Chang CW 1996 lsquoThe layout design language A

technique for generating layout plansrsquo in Computer Graphics Forum

15(2) 97ndash106

Heskestad AW 1999 lsquoPerformance in smoke of way guidance systemsrsquo

in Fire and Materials 23(6) 375ndash381

Knight T 1995 lsquoConstructive symmetryrsquo in Environment and Planning B

Planning and Design 22 419ndash450

Knight TW 1999 lsquoShape grammars six typesrsquo in Environment and

Planning B Planning and Design 26 15ndash31

Kuligowski ED and Peacock RD 2005 A Review of Building Evacuation

Models Technical Note 1471 National Institute of Standards and

Technology (NIST) Technology Administration US Department of

Commerce

Li A I-Kang 2003 lsquoThe Yingzao fashi in the information agersquo conference

paper for the Beaux-Arts Paul Philippe Cret and Twentieth Century

Architecture in China The University of Pennsylvania October

Liew H 2003 lsquoSGML ndash A shape grammar meta-languagersquo in Digital

Design 21st eCAADe Conference Proceedings Graz Austria September

639ndash648

Loomis B 2003 lsquoA note on generative design techniques SGGA a user

driven genetic algorithm for evolving non-deterministic shape grammarsrsquo

working paper Massachusetts Institute of Technology (MIT)

Magnusson SE Drysdale DD Fitzgerald RW Mowrer F

Quintiere JG Williamson RB and Zalosh RG 1995

lsquoA proposal for a model curriculum in fire safety engineeringrsquo

in Fire Safety Journal 25 9ndash10

Smith G Ceranic B and Fryer C 2003 lsquoDevelopment of CAD

environment for spatial layout optimisation in sub-surfacesurface rail

station designrsquo in Proceedings of International BIAT Conference The

British Museum London January 2003

SPSG 2007 London Underground Station Planning Standards and

Guidelines SPSG4 (1ndash371) London

Stiny G and Gips J 1971 lsquoShape grammars and the generative

specification of painting and sculpturersquo in Information Processing 71

1460ndash1465

Stiny G 1977 lsquoIce-ray a note on the generation of Chinese lattice

designsrsquo in Environment and Planning B Planning and Design 4 89ndash98

Stiny G 1980 lsquoKindergarten grammars designing with Froebelrsquos building

giftsrsquo in Environment and Planning B 3 409ndash462

Tapia M 1999 lsquoA visual implementation of a shape grammar systemrsquo in

Environment and Planning B Planning and Design 26(1) 59ndash73

Transport for London 2007 Tube Carries One Billion Passengers for First

Time wwwtflgovukcorporatemedianewscentre4770aspx

(accessed 24 March 2008)

Weller M 2007 London Underground private correspondence

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

120 G SMITH AND B CERANIC

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 117

FIGURE 17 Station evacuation after 25 seconds

FIGURE 18 Station evacuation after 50 seconds

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

118 G SMITH AND B CERANIC

FIGURE 19 Station evacuation after 100 seconds

FIGURE 20 Station evacuation after 300 seconds

knowledge database would be required particularlyin the light of frequent changes to legislation

On a more practical note it could be beneficial todevelop the system on a stand-alone modular basislinked to each generated station component A list offire safety strategy objectives and functional logicstatements can then be specified for commonlyrecurring items (such as emergency lighting fire alarmdetectors sprinkler heads etc) and spaced accordingto the designerrsquos wishes or via shape rule formulation

Site constraint issues would also benefit fromfuture development as well as a move towardsdevelopment of the 3D shape grammar version Oncedevised other relevant 3D modelling systems such ascomputational fluid dynamics (CFD) smoke movementsoftware could be utilized in conjunction with SGEvac

CONCLUSIONSThe computer-aided modelling environment underdevelopment is intended to act as a layout planning aidfor those who practice in sub-surfacesurface railwaystation design The system under developmentcarefully considers London Underground StationPlanning Standards and Guidelines together withother relevant fire safety regulations that influenceoverall fire strategy design The proposed computer-aided design environment (SGEvac) is based on highlyrestrictive additive shape grammar rules driven by akernel of key passenger flow information to generatestation plan solutions

The major benefit to the design team is the ability toreview multiple layouts of accurately sized and correctlyplaced infrastructure at the reference or preliminarystage of the station layout planning design process Timeis hence saved regarding different option appraisals andquantitative testing using evacuation software shouldlead to a choice of more effective layouts in relation tomeans of escape travelling distances and improvementin the total evacuation times Modifications to layoutsshould be straightforward and there is scope toefficiently assess travelling distances and evacuationtimes for each of the considered alternatives comparedwith a lsquotraditionalrsquo approach

The proposed novel shape grammar approach ismost suited to new-build stations or largerefurbishment projects where more design flexibilityexists Sites with many site-specific constraints have

an array of predetermined parameters and will mostlikely benefit from the manual operation of SGEvac

Use in other rail systems due to recurringinfrastructure components and inherent functionallogic related to the laying out of the stationcomponents is also feasible provided that shapeconnections and shape rules are corrected to reflectthat countryrsquos legislative requirements

The potential for station layout improvementswith regard to emergency evacuation is an area thatmerits research especially in scenarios that concernthe safety of the general public and form a vital partof Londonrsquos transportation system Potential layoutimprovements could not only benefit evacuationsituations but also enhance the through-flow ofpeople during peak travelling periods

ACKNOWLEDGEMENTS The authors wish to express their gratitude to JacobsEngineering (UK) Ltd which has sponsored theresearch and to London Underground Ltd forproviding the design guidance that is reproduced inthis paper and for authorising the use of the stationplans London Underground has not reviewed thispaper and the content is solely the authorsrsquo work Mr Robert Hutchinson is also thanked for hiscomputer programming assistance

AUTHOR CONTACT DETAILSGGrraahhaamm SSmmiitthh (corresponding author) Senior Architectural

Technologist Jacobs (UK) Ltd Cardinal Square North Point

10 Nottingham Road Derby DE1 3QT UK Tel +44 (0)1332 234482

fax +44 (0)1332 344558 e-mail grahampsmithjacobscomBBoorriiss CCeerraanniicc Programme Leader for BScBA (Hons) Architectural

Programmes Room Number 105 Faculty of Arts Design and

Technology University of Derby Markeaton Street Derby DE22

3AW UK Tel +44 (0)1332 593136 e-mail bceranicderbyacuk

REFERENCES BS 7974 2001 Application of Fire Safety Engineering Principles to the

Design of Buildings London BSI

Cha MY and Gero JS 1999 Style Learning Inductive Generalisation of

Architectural Shape Patterns Paichai University and Sydney University

co-paper

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 119

Chase SC 1996 lsquoUsing logic to specify shapes and spatial relations in

design grammarsrsquo in Workshop notes Grammatical Design Fourth

International Conference on Artificial Intelligence in Design Stanford

University US June

Department of the Environment 1996 Design Principles of Fire Safety

London The Stationery Office Books

Flemming U 1987 lsquoMore than the sum of its parts the grammar of

Queen Anne housesrsquo in Environment and Planning B Planning and

Design 14 323ndash350

Fruin JJ 1971 Pedestrian Planning and Design New York Metropolitan

Association of Urban Designers and Environmental Planners

Galea ER 1999 lsquoA general approach to validating evacuation models with

an application to EXODUSrsquo in Journal of Fire Sciences 16 414ndash436

Galle P and Kovacs LB 1992 lsquoThe logic of worms a study in

architectural knowledge representationrsquo in Environment and Planning B

Planning and Design 19 5ndash31

Garling T Lindberg E and Mantyla T 1983 lsquoOrientation in buildings ndash

effects of familiarity visual access and orientation aidsrsquo in Journal of

Applied Psychology 68(1) 177ndash186

Grimsdale RL and Chang CW 1996 lsquoThe layout design language A

technique for generating layout plansrsquo in Computer Graphics Forum

15(2) 97ndash106

Heskestad AW 1999 lsquoPerformance in smoke of way guidance systemsrsquo

in Fire and Materials 23(6) 375ndash381

Knight T 1995 lsquoConstructive symmetryrsquo in Environment and Planning B

Planning and Design 22 419ndash450

Knight TW 1999 lsquoShape grammars six typesrsquo in Environment and

Planning B Planning and Design 26 15ndash31

Kuligowski ED and Peacock RD 2005 A Review of Building Evacuation

Models Technical Note 1471 National Institute of Standards and

Technology (NIST) Technology Administration US Department of

Commerce

Li A I-Kang 2003 lsquoThe Yingzao fashi in the information agersquo conference

paper for the Beaux-Arts Paul Philippe Cret and Twentieth Century

Architecture in China The University of Pennsylvania October

Liew H 2003 lsquoSGML ndash A shape grammar meta-languagersquo in Digital

Design 21st eCAADe Conference Proceedings Graz Austria September

639ndash648

Loomis B 2003 lsquoA note on generative design techniques SGGA a user

driven genetic algorithm for evolving non-deterministic shape grammarsrsquo

working paper Massachusetts Institute of Technology (MIT)

Magnusson SE Drysdale DD Fitzgerald RW Mowrer F

Quintiere JG Williamson RB and Zalosh RG 1995

lsquoA proposal for a model curriculum in fire safety engineeringrsquo

in Fire Safety Journal 25 9ndash10

Smith G Ceranic B and Fryer C 2003 lsquoDevelopment of CAD

environment for spatial layout optimisation in sub-surfacesurface rail

station designrsquo in Proceedings of International BIAT Conference The

British Museum London January 2003

SPSG 2007 London Underground Station Planning Standards and

Guidelines SPSG4 (1ndash371) London

Stiny G and Gips J 1971 lsquoShape grammars and the generative

specification of painting and sculpturersquo in Information Processing 71

1460ndash1465

Stiny G 1977 lsquoIce-ray a note on the generation of Chinese lattice

designsrsquo in Environment and Planning B Planning and Design 4 89ndash98

Stiny G 1980 lsquoKindergarten grammars designing with Froebelrsquos building

giftsrsquo in Environment and Planning B 3 409ndash462

Tapia M 1999 lsquoA visual implementation of a shape grammar systemrsquo in

Environment and Planning B Planning and Design 26(1) 59ndash73

Transport for London 2007 Tube Carries One Billion Passengers for First

Time wwwtflgovukcorporatemedianewscentre4770aspx

(accessed 24 March 2008)

Weller M 2007 London Underground private correspondence

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

120 G SMITH AND B CERANIC

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

118 G SMITH AND B CERANIC

FIGURE 19 Station evacuation after 100 seconds

FIGURE 20 Station evacuation after 300 seconds

knowledge database would be required particularlyin the light of frequent changes to legislation

On a more practical note it could be beneficial todevelop the system on a stand-alone modular basislinked to each generated station component A list offire safety strategy objectives and functional logicstatements can then be specified for commonlyrecurring items (such as emergency lighting fire alarmdetectors sprinkler heads etc) and spaced accordingto the designerrsquos wishes or via shape rule formulation

Site constraint issues would also benefit fromfuture development as well as a move towardsdevelopment of the 3D shape grammar version Oncedevised other relevant 3D modelling systems such ascomputational fluid dynamics (CFD) smoke movementsoftware could be utilized in conjunction with SGEvac

CONCLUSIONSThe computer-aided modelling environment underdevelopment is intended to act as a layout planning aidfor those who practice in sub-surfacesurface railwaystation design The system under developmentcarefully considers London Underground StationPlanning Standards and Guidelines together withother relevant fire safety regulations that influenceoverall fire strategy design The proposed computer-aided design environment (SGEvac) is based on highlyrestrictive additive shape grammar rules driven by akernel of key passenger flow information to generatestation plan solutions

The major benefit to the design team is the ability toreview multiple layouts of accurately sized and correctlyplaced infrastructure at the reference or preliminarystage of the station layout planning design process Timeis hence saved regarding different option appraisals andquantitative testing using evacuation software shouldlead to a choice of more effective layouts in relation tomeans of escape travelling distances and improvementin the total evacuation times Modifications to layoutsshould be straightforward and there is scope toefficiently assess travelling distances and evacuationtimes for each of the considered alternatives comparedwith a lsquotraditionalrsquo approach

The proposed novel shape grammar approach ismost suited to new-build stations or largerefurbishment projects where more design flexibilityexists Sites with many site-specific constraints have

an array of predetermined parameters and will mostlikely benefit from the manual operation of SGEvac

Use in other rail systems due to recurringinfrastructure components and inherent functionallogic related to the laying out of the stationcomponents is also feasible provided that shapeconnections and shape rules are corrected to reflectthat countryrsquos legislative requirements

The potential for station layout improvementswith regard to emergency evacuation is an area thatmerits research especially in scenarios that concernthe safety of the general public and form a vital partof Londonrsquos transportation system Potential layoutimprovements could not only benefit evacuationsituations but also enhance the through-flow ofpeople during peak travelling periods

ACKNOWLEDGEMENTS The authors wish to express their gratitude to JacobsEngineering (UK) Ltd which has sponsored theresearch and to London Underground Ltd forproviding the design guidance that is reproduced inthis paper and for authorising the use of the stationplans London Underground has not reviewed thispaper and the content is solely the authorsrsquo work Mr Robert Hutchinson is also thanked for hiscomputer programming assistance

AUTHOR CONTACT DETAILSGGrraahhaamm SSmmiitthh (corresponding author) Senior Architectural

Technologist Jacobs (UK) Ltd Cardinal Square North Point

10 Nottingham Road Derby DE1 3QT UK Tel +44 (0)1332 234482

fax +44 (0)1332 344558 e-mail grahampsmithjacobscomBBoorriiss CCeerraanniicc Programme Leader for BScBA (Hons) Architectural

Programmes Room Number 105 Faculty of Arts Design and

Technology University of Derby Markeaton Street Derby DE22

3AW UK Tel +44 (0)1332 593136 e-mail bceranicderbyacuk

REFERENCES BS 7974 2001 Application of Fire Safety Engineering Principles to the

Design of Buildings London BSI

Cha MY and Gero JS 1999 Style Learning Inductive Generalisation of

Architectural Shape Patterns Paichai University and Sydney University

co-paper

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 119

Chase SC 1996 lsquoUsing logic to specify shapes and spatial relations in

design grammarsrsquo in Workshop notes Grammatical Design Fourth

International Conference on Artificial Intelligence in Design Stanford

University US June

Department of the Environment 1996 Design Principles of Fire Safety

London The Stationery Office Books

Flemming U 1987 lsquoMore than the sum of its parts the grammar of

Queen Anne housesrsquo in Environment and Planning B Planning and

Design 14 323ndash350

Fruin JJ 1971 Pedestrian Planning and Design New York Metropolitan

Association of Urban Designers and Environmental Planners

Galea ER 1999 lsquoA general approach to validating evacuation models with

an application to EXODUSrsquo in Journal of Fire Sciences 16 414ndash436

Galle P and Kovacs LB 1992 lsquoThe logic of worms a study in

architectural knowledge representationrsquo in Environment and Planning B

Planning and Design 19 5ndash31

Garling T Lindberg E and Mantyla T 1983 lsquoOrientation in buildings ndash

effects of familiarity visual access and orientation aidsrsquo in Journal of

Applied Psychology 68(1) 177ndash186

Grimsdale RL and Chang CW 1996 lsquoThe layout design language A

technique for generating layout plansrsquo in Computer Graphics Forum

15(2) 97ndash106

Heskestad AW 1999 lsquoPerformance in smoke of way guidance systemsrsquo

in Fire and Materials 23(6) 375ndash381

Knight T 1995 lsquoConstructive symmetryrsquo in Environment and Planning B

Planning and Design 22 419ndash450

Knight TW 1999 lsquoShape grammars six typesrsquo in Environment and

Planning B Planning and Design 26 15ndash31

Kuligowski ED and Peacock RD 2005 A Review of Building Evacuation

Models Technical Note 1471 National Institute of Standards and

Technology (NIST) Technology Administration US Department of

Commerce

Li A I-Kang 2003 lsquoThe Yingzao fashi in the information agersquo conference

paper for the Beaux-Arts Paul Philippe Cret and Twentieth Century

Architecture in China The University of Pennsylvania October

Liew H 2003 lsquoSGML ndash A shape grammar meta-languagersquo in Digital

Design 21st eCAADe Conference Proceedings Graz Austria September

639ndash648

Loomis B 2003 lsquoA note on generative design techniques SGGA a user

driven genetic algorithm for evolving non-deterministic shape grammarsrsquo

working paper Massachusetts Institute of Technology (MIT)

Magnusson SE Drysdale DD Fitzgerald RW Mowrer F

Quintiere JG Williamson RB and Zalosh RG 1995

lsquoA proposal for a model curriculum in fire safety engineeringrsquo

in Fire Safety Journal 25 9ndash10

Smith G Ceranic B and Fryer C 2003 lsquoDevelopment of CAD

environment for spatial layout optimisation in sub-surfacesurface rail

station designrsquo in Proceedings of International BIAT Conference The

British Museum London January 2003

SPSG 2007 London Underground Station Planning Standards and

Guidelines SPSG4 (1ndash371) London

Stiny G and Gips J 1971 lsquoShape grammars and the generative

specification of painting and sculpturersquo in Information Processing 71

1460ndash1465

Stiny G 1977 lsquoIce-ray a note on the generation of Chinese lattice

designsrsquo in Environment and Planning B Planning and Design 4 89ndash98

Stiny G 1980 lsquoKindergarten grammars designing with Froebelrsquos building

giftsrsquo in Environment and Planning B 3 409ndash462

Tapia M 1999 lsquoA visual implementation of a shape grammar systemrsquo in

Environment and Planning B Planning and Design 26(1) 59ndash73

Transport for London 2007 Tube Carries One Billion Passengers for First

Time wwwtflgovukcorporatemedianewscentre4770aspx

(accessed 24 March 2008)

Weller M 2007 London Underground private correspondence

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

120 G SMITH AND B CERANIC

knowledge database would be required particularlyin the light of frequent changes to legislation

On a more practical note it could be beneficial todevelop the system on a stand-alone modular basislinked to each generated station component A list offire safety strategy objectives and functional logicstatements can then be specified for commonlyrecurring items (such as emergency lighting fire alarmdetectors sprinkler heads etc) and spaced accordingto the designerrsquos wishes or via shape rule formulation

Site constraint issues would also benefit fromfuture development as well as a move towardsdevelopment of the 3D shape grammar version Oncedevised other relevant 3D modelling systems such ascomputational fluid dynamics (CFD) smoke movementsoftware could be utilized in conjunction with SGEvac

CONCLUSIONSThe computer-aided modelling environment underdevelopment is intended to act as a layout planning aidfor those who practice in sub-surfacesurface railwaystation design The system under developmentcarefully considers London Underground StationPlanning Standards and Guidelines together withother relevant fire safety regulations that influenceoverall fire strategy design The proposed computer-aided design environment (SGEvac) is based on highlyrestrictive additive shape grammar rules driven by akernel of key passenger flow information to generatestation plan solutions

The major benefit to the design team is the ability toreview multiple layouts of accurately sized and correctlyplaced infrastructure at the reference or preliminarystage of the station layout planning design process Timeis hence saved regarding different option appraisals andquantitative testing using evacuation software shouldlead to a choice of more effective layouts in relation tomeans of escape travelling distances and improvementin the total evacuation times Modifications to layoutsshould be straightforward and there is scope toefficiently assess travelling distances and evacuationtimes for each of the considered alternatives comparedwith a lsquotraditionalrsquo approach

The proposed novel shape grammar approach ismost suited to new-build stations or largerefurbishment projects where more design flexibilityexists Sites with many site-specific constraints have

an array of predetermined parameters and will mostlikely benefit from the manual operation of SGEvac

Use in other rail systems due to recurringinfrastructure components and inherent functionallogic related to the laying out of the stationcomponents is also feasible provided that shapeconnections and shape rules are corrected to reflectthat countryrsquos legislative requirements

The potential for station layout improvementswith regard to emergency evacuation is an area thatmerits research especially in scenarios that concernthe safety of the general public and form a vital partof Londonrsquos transportation system Potential layoutimprovements could not only benefit evacuationsituations but also enhance the through-flow ofpeople during peak travelling periods

ACKNOWLEDGEMENTS The authors wish to express their gratitude to JacobsEngineering (UK) Ltd which has sponsored theresearch and to London Underground Ltd forproviding the design guidance that is reproduced inthis paper and for authorising the use of the stationplans London Underground has not reviewed thispaper and the content is solely the authorsrsquo work Mr Robert Hutchinson is also thanked for hiscomputer programming assistance

AUTHOR CONTACT DETAILSGGrraahhaamm SSmmiitthh (corresponding author) Senior Architectural

Technologist Jacobs (UK) Ltd Cardinal Square North Point

10 Nottingham Road Derby DE1 3QT UK Tel +44 (0)1332 234482

fax +44 (0)1332 344558 e-mail grahampsmithjacobscomBBoorriiss CCeerraanniicc Programme Leader for BScBA (Hons) Architectural

Programmes Room Number 105 Faculty of Arts Design and

Technology University of Derby Markeaton Street Derby DE22

3AW UK Tel +44 (0)1332 593136 e-mail bceranicderbyacuk

REFERENCES BS 7974 2001 Application of Fire Safety Engineering Principles to the

Design of Buildings London BSI

Cha MY and Gero JS 1999 Style Learning Inductive Generalisation of

Architectural Shape Patterns Paichai University and Sydney University

co-paper

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

Spatial Layout Planning in Sub-Surface Rail Station Design for Effective Fire Evacuation 119

Chase SC 1996 lsquoUsing logic to specify shapes and spatial relations in

design grammarsrsquo in Workshop notes Grammatical Design Fourth

International Conference on Artificial Intelligence in Design Stanford

University US June

Department of the Environment 1996 Design Principles of Fire Safety

London The Stationery Office Books

Flemming U 1987 lsquoMore than the sum of its parts the grammar of

Queen Anne housesrsquo in Environment and Planning B Planning and

Design 14 323ndash350

Fruin JJ 1971 Pedestrian Planning and Design New York Metropolitan

Association of Urban Designers and Environmental Planners

Galea ER 1999 lsquoA general approach to validating evacuation models with

an application to EXODUSrsquo in Journal of Fire Sciences 16 414ndash436

Galle P and Kovacs LB 1992 lsquoThe logic of worms a study in

architectural knowledge representationrsquo in Environment and Planning B

Planning and Design 19 5ndash31

Garling T Lindberg E and Mantyla T 1983 lsquoOrientation in buildings ndash

effects of familiarity visual access and orientation aidsrsquo in Journal of

Applied Psychology 68(1) 177ndash186

Grimsdale RL and Chang CW 1996 lsquoThe layout design language A

technique for generating layout plansrsquo in Computer Graphics Forum

15(2) 97ndash106

Heskestad AW 1999 lsquoPerformance in smoke of way guidance systemsrsquo

in Fire and Materials 23(6) 375ndash381

Knight T 1995 lsquoConstructive symmetryrsquo in Environment and Planning B

Planning and Design 22 419ndash450

Knight TW 1999 lsquoShape grammars six typesrsquo in Environment and

Planning B Planning and Design 26 15ndash31

Kuligowski ED and Peacock RD 2005 A Review of Building Evacuation

Models Technical Note 1471 National Institute of Standards and

Technology (NIST) Technology Administration US Department of

Commerce

Li A I-Kang 2003 lsquoThe Yingzao fashi in the information agersquo conference

paper for the Beaux-Arts Paul Philippe Cret and Twentieth Century

Architecture in China The University of Pennsylvania October

Liew H 2003 lsquoSGML ndash A shape grammar meta-languagersquo in Digital

Design 21st eCAADe Conference Proceedings Graz Austria September

639ndash648

Loomis B 2003 lsquoA note on generative design techniques SGGA a user

driven genetic algorithm for evolving non-deterministic shape grammarsrsquo

working paper Massachusetts Institute of Technology (MIT)

Magnusson SE Drysdale DD Fitzgerald RW Mowrer F

Quintiere JG Williamson RB and Zalosh RG 1995

lsquoA proposal for a model curriculum in fire safety engineeringrsquo

in Fire Safety Journal 25 9ndash10

Smith G Ceranic B and Fryer C 2003 lsquoDevelopment of CAD

environment for spatial layout optimisation in sub-surfacesurface rail

station designrsquo in Proceedings of International BIAT Conference The

British Museum London January 2003

SPSG 2007 London Underground Station Planning Standards and

Guidelines SPSG4 (1ndash371) London

Stiny G and Gips J 1971 lsquoShape grammars and the generative

specification of painting and sculpturersquo in Information Processing 71

1460ndash1465

Stiny G 1977 lsquoIce-ray a note on the generation of Chinese lattice

designsrsquo in Environment and Planning B Planning and Design 4 89ndash98

Stiny G 1980 lsquoKindergarten grammars designing with Froebelrsquos building

giftsrsquo in Environment and Planning B 3 409ndash462

Tapia M 1999 lsquoA visual implementation of a shape grammar systemrsquo in

Environment and Planning B Planning and Design 26(1) 59ndash73

Transport for London 2007 Tube Carries One Billion Passengers for First

Time wwwtflgovukcorporatemedianewscentre4770aspx

(accessed 24 March 2008)

Weller M 2007 London Underground private correspondence

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

120 G SMITH AND B CERANIC

Chase SC 1996 lsquoUsing logic to specify shapes and spatial relations in

design grammarsrsquo in Workshop notes Grammatical Design Fourth

International Conference on Artificial Intelligence in Design Stanford

University US June

Department of the Environment 1996 Design Principles of Fire Safety

London The Stationery Office Books

Flemming U 1987 lsquoMore than the sum of its parts the grammar of

Queen Anne housesrsquo in Environment and Planning B Planning and

Design 14 323ndash350

Fruin JJ 1971 Pedestrian Planning and Design New York Metropolitan

Association of Urban Designers and Environmental Planners

Galea ER 1999 lsquoA general approach to validating evacuation models with

an application to EXODUSrsquo in Journal of Fire Sciences 16 414ndash436

Galle P and Kovacs LB 1992 lsquoThe logic of worms a study in

architectural knowledge representationrsquo in Environment and Planning B

Planning and Design 19 5ndash31

Garling T Lindberg E and Mantyla T 1983 lsquoOrientation in buildings ndash

effects of familiarity visual access and orientation aidsrsquo in Journal of

Applied Psychology 68(1) 177ndash186

Grimsdale RL and Chang CW 1996 lsquoThe layout design language A

technique for generating layout plansrsquo in Computer Graphics Forum

15(2) 97ndash106

Heskestad AW 1999 lsquoPerformance in smoke of way guidance systemsrsquo

in Fire and Materials 23(6) 375ndash381

Knight T 1995 lsquoConstructive symmetryrsquo in Environment and Planning B

Planning and Design 22 419ndash450

Knight TW 1999 lsquoShape grammars six typesrsquo in Environment and

Planning B Planning and Design 26 15ndash31

Kuligowski ED and Peacock RD 2005 A Review of Building Evacuation

Models Technical Note 1471 National Institute of Standards and

Technology (NIST) Technology Administration US Department of

Commerce

Li A I-Kang 2003 lsquoThe Yingzao fashi in the information agersquo conference

paper for the Beaux-Arts Paul Philippe Cret and Twentieth Century

Architecture in China The University of Pennsylvania October

Liew H 2003 lsquoSGML ndash A shape grammar meta-languagersquo in Digital

Design 21st eCAADe Conference Proceedings Graz Austria September

639ndash648

Loomis B 2003 lsquoA note on generative design techniques SGGA a user

driven genetic algorithm for evolving non-deterministic shape grammarsrsquo

working paper Massachusetts Institute of Technology (MIT)

Magnusson SE Drysdale DD Fitzgerald RW Mowrer F

Quintiere JG Williamson RB and Zalosh RG 1995

lsquoA proposal for a model curriculum in fire safety engineeringrsquo

in Fire Safety Journal 25 9ndash10

Smith G Ceranic B and Fryer C 2003 lsquoDevelopment of CAD

environment for spatial layout optimisation in sub-surfacesurface rail

station designrsquo in Proceedings of International BIAT Conference The

British Museum London January 2003

SPSG 2007 London Underground Station Planning Standards and

Guidelines SPSG4 (1ndash371) London

Stiny G and Gips J 1971 lsquoShape grammars and the generative

specification of painting and sculpturersquo in Information Processing 71

1460ndash1465

Stiny G 1977 lsquoIce-ray a note on the generation of Chinese lattice

designsrsquo in Environment and Planning B Planning and Design 4 89ndash98

Stiny G 1980 lsquoKindergarten grammars designing with Froebelrsquos building

giftsrsquo in Environment and Planning B 3 409ndash462

Tapia M 1999 lsquoA visual implementation of a shape grammar systemrsquo in

Environment and Planning B Planning and Design 26(1) 59ndash73

Transport for London 2007 Tube Carries One Billion Passengers for First

Time wwwtflgovukcorporatemedianewscentre4770aspx

(accessed 24 March 2008)

Weller M 2007 London Underground private correspondence

ARCHITECTURAL ENGINEERING AND DESIGN MANAGEMENT 2008 VOLUME 4 PAGES 99ndash120

120 G SMITH AND B CERANIC