Research Review 010312lr

7/27/2019 Research Review 010312lr

1/37

Research Review

ISSUE III


2/37

4 Modelling the behaviour of reinforced concrete shear walls under severe earthquake loads

8 Life cycle assessment of Project Onyx

12 Veri cation of CFD for air ventilation analysis in a dense urban environment

16 StructuralComponents: a parametric and associative toolbox for conceptual design

20 Arups 4see model: combining socio-economic and energy aspects of a whole economy

24 Transitions: moving from rhetoric to priority in 21st century Perth

28 Extending the life of critical road system infrastructure in the Netherlands

32 Material ef ciency in construction

36 Revisiting the case for structural steel reuse

38 Church View climate change adaptation

40 3D model simulation and interface between vehicle and pedestrian simulation models

42 Bioreactive faade with photobioreactors for cultivating microalgae as an energy resource

44 Critical success factors of road space reallocation: impacts on local businesses

46 City infrastructure: The SURF-Arup framework for systemic action

50 Sustainable management and reuse of wastewater in masterplanning: Introducing QWIC MENU

54 Full-scale benchmarking of tall building response subjected to wind loads during typhoons

58 Acoustic performance of naturally-ventilated faades using the Ventilated Acoustic Cavity Solver

62 Development of advanced constitutive soil models for nite element applications

66 Shaping a better world: How this challenge differs across two modes of engineering practice

68 Doctoral Module Programme

70 Contact us

Contents Global ResearchI am very pleased to introduce the Arup 2012 Research Review, featuring the bestof the rms collaborative research from around the globe.

The Global Research team is proud to work alongside the worlds best universitiesand companies, fostering links which lead to academic impact and business bene t.Research coupled with design and engineering excellence plays a key role inkeeping Arup at the forefront of its sector.

This often happens at the touch-points between subject areas. The rich variety of

disciplines at Arup means there are many opportunities to innovate.

Research is a global exercise for Arup, and there are Research Champions in eachof the ve regions. They are responsible locally for internal and externalrelationships and link colleagues with university partners.

Within Arup, research is done in a distributed way, mainly within project teamsglobally, with a third of staff being research-active. We capture needs andopportunities, and coordinate global collaboration with research partners.

Using horizon-scanning and road mapping, time-ordered priorities are establishedfor research, balancing what businesses need with what universities are evolving.As well as providing this internally, frequently we are asked to apply these testedmethods when clients want to create a coherent picture of their research needs.

The Global team leads key initiatives to achieve this. In addition to creatingresearch agendas, capabilities include: undertaking management of research

programmes, development of staff through doctoral training in research methods,and work with funding agencies to establish research priorities.

However, early impact is often achieved, so for the rst time this Research Reviewalso includes some Impact Stories, intended to demonstrate the value generated byresearch.

We hope you will nd these articles interesting, and that you will want tocollaborate with Arup.

Jeremy Watson : Director, Global Research

Innovation supported by research is a vital factor in business success.

Value from research comes in a variety of forms as wellas revenue growth, and often over an extended period.

Research at Arup is a tool for creating nancial and knowledge value by building capability and evidence; these enhance business advantage.

The Global Research team acts as a catalyst and strategic hub.

2 3


3/37

Modelling the behaviour of reinforced concreteshear walls under severe earthquake loadsArup: Ibrahim Almufti, Gregory Nielsen, Michael Willford, Richard Sturt

AbstractReinforced concrete walls have beenwidely used as the seismic resistingsystem, especially in high-rise buildings, in the last decade. The behaviour of such buildings in designlevel events is relatively unknown

because most major urban centers have been spared large magnitudeearthquakes. However, data from therecent 2010 earthquake in Chilecon rmed that reinforced concretewalls are vulnerable to signi cantstructural damage, and in some casescollapse (see Fig. 1 and 2).

It has been clear for some time thatmost commercially available analysis programs are not able to predictaccurately the nonlinear behaviour of reinforced concrete walls in strongearthquakes. To this end, this studyfocused on validating newlyimplemented material models inLS-DYNA. This is a nonlinear dynamic

nite element program from LivermoreSoftware Technology Corporation.Arups Advanced Technology andResearch team has developed numerousnew material models and features over

the past 25 years for the LS-DYNA programme.

The results of this research show thatArups capabilities are at the forefrontof engineering practice, and canconvince clients (and ourselves) thatour methods are more accurate andreliable. In the context of performance- based seismic design, this is aninvaluable advantage that enables Arupto meet clients seismic performanceobjectives with greater con dence andeconomy.

IntroductionMost national building codes prescribeminimum seismic design and detailingrequirements for buildings. In the US andmost other jurisdictions the intent of thecodes is for buildings to remain structurallyundamaged in relatively frequent earthquakeevents, and to provide life-safety performance in much larger events; lifesafety performance, whilst preventingcollapse, accepts considerable, and possiblyirreparable, damage to the structure andnon-structural components.

The inherent limitations in the traditionalcode methods, and the desire to understand better, (and control) the likely levels ofdamage, have brought about alternative performance-based approaches which predictthe seismic behavior of buildings throughuse of non-linear computer analysis. Non-linear response history analysis subjectsa 3D numerical representation of the buildingto actual (or modi ed) ground motionsrecorded from past earthquakes. The performance-based approach can be used tomeet project-speci c or enhanced seismic performance objectives which the buildingowner may desire. Whilst it enablesstructural engineers to have greater con dence in the performance of theirdesigns, the method is clearly reliant on thecapability and reliability of the computersoftware used to predict the dynamicresponse of buildings to severe groundshaking.

In the absence of much measured data fromconcrete wall buildings in real earthquakeevents, structural engineers and researchersmust validate their computer predictions bygleaning information from physical testscarried out in laboratory facilities. Most testshave been conducted using quasi-statictesting protocols in which a concrete wallspecimen is loaded cyclically very slowly ineach direction with each cycle increasing in

amplitude until the specimen reaches failure. Such quasi-static tests havesigni cant limitations in that the loading protocols are xed, single-mode behaviour is enforced and the effect of damping andstiffness degradation on the actual demandscannot be accounted for. Large-scaledynamic shake table tests are required toovercome these limitations.

In 2006, a full-scale 7-storey reinforcedconcrete shear wall assembly was built onthe large outdoor shake table at theUniversity of California, San Diego (UCSD). It was subjected to a suite ofground motions with increasing intensity,including one of the strongest motionsrecorded from the 1994 Northridge,California earthquake. This test was the rstof its kind in the world and providedcomprehensive experimental data on theactual nonlinear behaviour of a reinforcedconcrete wall structure. In order to assess thestate of the practice, a blind analysiscompetition was held allowing practitionersand academics to predict the response of thestructure through the numerical simulationmethod of their choosing. Arup did not participate, however, the results of thiscompetition, in which no team successfully predicted the result adequately, revealed thatexisting commercially available softwarewas not adequate to predict the actualseismic behaviour of a concrete wallstructure.

At about the same time as t his test wasconducted, Arup developed a new shellelement formulation for LS-DYNA. Itincorporates two complementary non-linearmaterial models for concrete and reinforcingsteel. This was veri ed against a number ofquasi-static shear wall tests and was rstused for a seismic design project on theTaichung Metropolitan Opera House. TheLS-DYNA modeling approach is rigorousand has become Arups standard of practice

for the seismic design of importantreinforced concrete wall structures on theWest Coast of the United States and China.

The research study compared the results ofthis new material model with the full scaleresults from the 7-storey shear wall shaketable test at UCSD.

MethodologyAll of the reinforced concrete elements of the walls and oors of the test structure weremodelled using the 2D sandwich shellelements in LS-DYNA. It has become clearthat the nonlinear distribution of strain alongthe plane of a reinforced concrete wallsection is an important factor in certainseismic collapse mechanisms of walls. Manyof these behaviours, including shear/ exureinteraction and localised crushing andspalling of concrete cannot be captured byplane sections remain plane beam/ berelements. The 2D sandwich shell elementin LS-DYNA can capture these importanteffects provided that the wall is discretisedinto many shell elements along its length.

The wall was divided into discrete regionstaking account of the steel reinforcing layout,and con nement of boundary elements.Longitudinal and transverse reinforcing barswere modelled in separate layers in the

Fig 1 Collapse o a rein orced concrete shear wall apartment building observed ollowing the 2010 M8.8 Chile earthquake

Fig 2 Buckling ailure o a concrete shear wall boundaryelement observed ollowing the 2010 M8.8 Chile earthquake

The seismic behaviour of reinforced concrete wall buildings involves complexinteraction not easilycaptured by existingnumerical simulationmethods. A new tool isrequired to con dently predict the dynamic responseof these types of buildings.

54 5


4/37

sandwich and oriented in orthogonaldirections, creating a composite element withorthotropic material response.

The LS-DYNA concrete material modelMAT_CONCRETE_EC2 (based onEurocode 2, Part 1.2) uses the well-documented Mander equations to de ne thecon ned and uncon ned concrete stress-strain characteristics with additional parameters to account for concrete cracking,tension stiffening, shear friction, aggregateinterlock, dowel action, spalling and theeffect of orthogonal cracking on thecompressive strength of concrete. Concretestrength was based on experimentaluncon ned concrete cylinder tests whichvaried from between 36MPa to 43MPa over the height of the wall.

The steel reinforcing material model MAT_ HYSTERETIC_REINFORCEMENT usedthe stress-strain curves from tension tests ofthe actual materials to de ne the plasticresponse of the reinforcement and includedadditional parameters to account for theBauschinger effect, potential bar buckling,and dowel action.

Fig. 3 compares the 3D LS-DYNA modelwith a photo of the actual test specimen.Some features of the test specimen providedmodelling challenges, particularly theinteraction of the oor slabs which connectthe web wall being tested to closely spacedgravity columns and a ange wall (connected by a slotted slab connection providing only partial moment transfer). In addition, a post-tensioned pre-cast segmental pier with

diaphragm bracing was used to preventtorsional responses of the test structure. The experimental measurements indicatedthat these interactions approximately doubledthe moment capacity of the system comparedto that of the web wall alone (Panagiotou et al. 2007).

The shake test program comprised fourseparate motions of increasing intensity(from 0.2g to 0.8g peak ground acceleration)applied in the direction of the web wall.Simulation of the 120 seconds of groundmotion comprising all four strong motionrecordings took approximately eight hours ofcomputing time on an 8-core CPU.

Results and discussionThe LS-DYNA predicted response was anexcellent match to the measured wall behaviour. The global hysteretic response(base moment versus roof drift) of the wallassembly during the last (largest) groundmotion in the sequence is shown in Fig. 4The response shows that the LS-DYNAsimulation resulted in very close agreementof the positive and negative moment anddrift demands associated with all theresponse cycles. This indicates that thenumerical model accurately captures notonly the peak plastic response of the wall butalso the degraded post-yield behaviour onsubsequent cycles.

1.5

1.0

0.5

0

-0.5

-1.0

-1.5-3 -2 -1 0

Roof drift (% height)

1 2 3

B a s e m o m e n

t ( k N / m ) x 1 0 4

EQ4 recorded

EQ4 simulated (LS-DYNA)

Fig 3 Comparison o the 3D LS-DYNA model with the experimental test specimen

Fig 4 Comparison o global hysteretic response o the analytical model with the experimental test

In addition, Fig. 4 does not include themodelled response of the preceding threemotions but the agreement is similarlyaccurate. When the simulation was re-rununder the largest motion alone (without theaccumulated damage due to the initial threemotions), the peak moment was the same, but the peak drift was under-predicted by25%. This highlights the signi cant effect ofaccumulated damage and cracking on theglobal behaviour of the system. This effectcould be particularly important in theassessment of existing buildings that havealready experienced signi cant shaking. Thisstudy offers an analytical model that could be used to account for possible pre-existingdamage and its effect on seismic performance, particularly if an estimate ofthe previous ground motions is known.

Fig. 5 shows a comparison of the plasticcompressive strain in the cover concretefrom the analysis with pictures from theactual test. Regions in pink indicate wherethe concrete is predicted to have spalled, andreasonably matches the locations of spallingand splitting observed in the test specimen.Plastic behaviour (concrete cracking,reinforcement yielding) was largely limitedto the rst and second level web walls andthe slabs, and was consistent with theobservations from the experimental test.

Parameter studies showed that the analysismodel was generally insensitive to typicalvariations in the material strengths and parameters governing the cyclic response ofthe material models, with the exception ofthe tensile strength assumed for the slabconcrete.

Conclusion and next stepsThis research shows that the advanced methodsArup has developed for LS-DYNA can predictthe extremely complex behaviour of reinforcedconcrete walls during strong earthquakemotions. Predictions are more accurate andreliable than other available software.

We have utilised the methods presented inthis article for several projects that arecurrently in design, including some of thetallest buildings in the world. Nowadays, thecapability of Arups advanced materialmodels and the speed of modern computersenable sophisticated performance-basedanalysis of this type to be used as part of theiterative design procedure, rather than as a

nal performance check. The ability to tune building designs to achieve better seismic performance as an integral part of the design provides a signi cant competitive advantage.

As a result of this research, the AdvancedTechnology Council and the Consortium ofUniversities for Research into EarthquakeEngineering have recognised that ArupsLS-DYNA modelling approach is the leadingmethodology for the analytical assessment of reinforced concrete wall buildings. As such,Arup has been approached to provide thewhole building simulations for a number oftall concrete buildings which were severelydamaged in the 2010 Chile earthquake Thesesimulations are bring done for a current NIST funded research project in the USA.

As these methods gain traction throughoutthe industry, Arup will be increasinglyrecognised as a major leader in the globalseismic engineering community, leading tomore opportunities on complex projects.

Value

Improves capability in modellingthe nonlinear seismic behavior ofconcrete shear walls

Allows us to con dently use thesecomputational models to iterativelytest and improve the design of newstructures

Reinforces Arup as a leading presencein the seismic engineering community

References

LS-DYNA Keyword Users Manual Version971 / Release 4 (2009), Livermore So tware Technology Corpora tion (L STC).

Panagiotou, M., Restrepo, J. I., and Conte, J. P.(2007) Shake Table Test o a 7-Story Full ScaleRein orced Concrete Structural Wall BuildingSlice Phase I: Rectangular Wall Section,Structural Systems Research Project ReportNo. SSRP-07-07, Department o StructuralEngineering, University o Cali ornia, San Diego.

Fig 5 Comparison o the plastic compressive strain in the cover concrete rom the analysis with pictures rom the actual test

6 7


5/37

Life cycle assessment of Project OnyxArup: Aaron Yuen, Rob Turk, Melanie Koerner and Tania Smith

AbstractAs new virgin oil sources becomescarce, prices are increasing at rapid

rates. In response, manufacturers aresearching for alternative fuel sources.

Orica Specialty Chemicals hasresponded to this challenge bydeveloping a unique process andtechnology to purify used oil known as Onyx Process Oil (OPO) to replacevirgin oil in the production ofammonium nitrate explosives (ANE).

Whilst there are economic bene tsassociated with this transition, Oricaalso recognises the existence of

environmental impacts from such a process, and engaged Arup to undertake

a process of quantifying theenvironmental bene ts from a life cycle perspective.

Utilising a methodology based onestablished International Standards,Arup undertook a Life CycleAssessment of OPO, with the project being independently reviewed by LifeCycle Strategies. The assessmentconsidered a range of environmentalimpacts in climate change, energyconsumption, non-renewable resourcedepletion, water use, eutrophication,

air acidi cation, photochemical oxidantformation, human health toxicity and

ecotoxicity.

The ndings indicated that the production of OPO has less inherentlife cycle impacts when compared withthe production of virgin oil, particularlyregarding carbon emissions andnon-renewable resource depletion.Additionally, under a global long-term perspective, the use of OPO has lessaggregated impacts in comparison tovirgin oils.

IntroductionIn future decades the scarcity of virgin oildue to global factors such as t he industrialand economic modernisation of India andChina will place considerable pressure on theviability of businesses dependent on virginoil for their operations.

One strategy to minimise an organisationsexposure to this virgin oil demand/supplyimbalance is to actively seek alternative fuelsources which may be utilised in oildependent processes. In this instance, onealternative to virgin oil is the use of wasteor used oil from industrial processes.

Waste or used oil is broadly de ned as petroleum-based or synthetic oil thatcontains impurities or has lost its original properties after use.

Traditionally, waste oil has been disposed of by industry or stored and burnt as fuels for energy. In Australia, the waste oil market isnow reaching the point of being constrained.That is, an end use is able to be found for themajority of waste oil in the Australianmarket.

As an alternative waste oil end use, OricaSpecialty Chemicals (OSC) has developed aunique process through Project Onyx. OSCis a producer of explosive emulsi ers andundertakes research for Orica Chemicals, aleading global manufacturer of chemicalsand services in industrial and miningmarkets.

The Onyx Process has been developed to purify lower quality used oil to produce areplacement for virgin oils in Oricas production of ANE. These are usedextensively in the mining industry as ameans of exposing ore bodies. As a result ofreplacing virgin oils with OPO in the production of ANE, environmental bene tscan be demonstrated in addition to theeconomic cost savings that may be realised.

In order to understand and quantify theenvironmental bene ts or impacts associatedwith the Onyx Process, a Life CycleAssessment (LCA) was undertaken by Arup.The LCA was part funded by OSC, thePlastics and Chemicals Industry Associationof Australia and the EnvironmentalProtection Authority of Victoria.

LCAs have been used by industry as asystematic and quantitatively based approachin comparing a range of environmentaleffects to the delivery of products orservices. The assessment aims to assess theimpacts associated with a product or aservice across all stages of its life; fromextraction of raw material, manufacturing processes, distribution, use and disposal. A LCA approach differs from otherenvironmental management approaches in that it measures and calculates impactsnormalised per unit of output.

For OSC, this LCA quanti ed theenvironmental bene ts of the Onyx Processfor two of its technologies: a xed plant process within metropolitan Melbourne,Victoria; and a remote mine site facilitylocated at Boddington, Western Australia.

For each technology the LCA sought toquantify the:

environmental impacts associated with themanufacture of OPO, including Scope 1, 2and 3 greenhouse gas emissions

environmental bene ts in comparison withthe use of virgin oil to Orica customers for:- replacement of diesel oil and SN150 (a

base oil)- replacement of diesel oil and SN150 in

ammonium nitrate explosive products- in Oricas annual operations

strategies and actions to enable OSC toimprove its performance

MethodologyThe LCA was undertaken by Arup and peer-reviewed by Life Cycle Strategies inaccordance with International Standards;ISO 14040 and ISO 14044. As per theinternational standards, Fig. 2 outlines thekey phases of the LCA as: de ning the goaland scope of the assessment; compiling aninventory of relevant inputs and outputs of a product system; evaluating the potentialenvironmental impacts; and interpreting the results in relation to the objectives of this study.

This methodology was applied to thecomparison of the use of OPO or virgin oils in the production process of ANE. Thespeci c oils compared were, for the xed plant; OPO and SN150, and for the remotefacility; OPO and diesel oil.

Goal and scope defnitionAll unit processes which contribute to thesupply of raw materials and main processingenergy demands and those which treat or process waste produced from the main processing system, were included within thesystem boundaries. Furthermore, the impact

Interpretationof results

Life cycleinventoryanalysis

Goal andscope

definition

Lifecycleimpact

assessment

Fig 1 Part o the Onyx process xed plant acility

Fig 2 Li e Cycle Assessment methodology summary

Oricas Project Onyx seeksto de ne a process to replacethe use of virgin oils in the production of ammoniumnitrate explosives with puri ed waste oil. Animportant part of thistransition is understandingthe associated quanti ableenvironmental bene ts froma life cycle perspective.

8 99


6/37

of blasting ammonium nitrate was takeninto account. As used oil contains certainamounts of heavy metals, the release of theseinto the atmosphere during blasting wasconsidered as a worst-case scenario in this study.

The goal and scope of the study was todetermine the life cycle impacts associatedwith a broad range of environmental impactcategories. The speci c impact categoriesselected were; climate change, energyconsumption, non-renewable resourcedepletion, water use, eutrophication, airacidi cation, photochemical oxidantformation, human health toxicity andecotoxicity.

In addition to individual impact categories aReCiPe Analysis was undertaken. ReCiPe isa life cycle impact assessment method whichaggregates multiple impact categories into asingle score. The method considers three perspectives of a global system that in uencethe weightings of impact categories in providing an aggregate score:

Individualist: based on short-terminterests, considering impact types that areundisputed and human adaptation relianton technological advancement

Hierarchical: based on the medium-term,considering common policy principlesadopted by countries

Egalitarian: the most precautionaryperspective, considering the longesttime-frame and impact types where someindication is available

Life cycle inventory analysisA data and processes inventory of the puri cation of waste oil was thendetermined. Data were sourced directly fromOSC Onyx Process engineers and scientists.Where data could not be sourced directly(such as the emission impacts of Australiangrid electricity production), processes anddata were adopted from Ecoinvent andAustralian Life Cycle Assessment peer-reviewed databases.

Life cycle impact assessmentThe impact assessment was modelled usingSimaPro 7.2 LCA software which assists practitioners in building complex modelssystematically and with transparency. Each of the impact categories that weremodelled are based on characterisationmodels that are publicly available andfrequently used in LCA.

As for any quantitatively based assessment,the validity of the study data used underpinsthe accuracy and robustness of theassessment. For the data collected, values ofuncertainty were estimated using a pedigreematrix against a range of criteria. Uncertaintyanalyses were carried out using a MonteCarlo Analysis over one thousand modellingiterations per impact category.

Results and discussionThe LCA indicates that the production ofOPO has an overall result of less lif ecycleimpacts in comparison to the production ofvirgin oil. The LCA has also been used toidentify the manufacturing process or

processes that contribute to the greatestimpacts and assisting in redesign forimproved environmental outcomes. Thespeci c bene ts for the two technologyscenarios are considered below.

Fixed plant processCompared to the production of virgin oil,OPO reduces the production of carbonemissions by approximately 78%. Energyconsumption across the lifecycle is alsosigni cantly reduced by 81%, and 85% of the non-renewable resources normallydepleted from virgin oil production are avoided.

However, the production of OPO has been determined as a more water intensive process compared to SN150 production,mainly due to the washing and coolingrequirements of the waste oil puri cation process. Nevertheless, in absolute terms, the additional water consumption for the Onyx Process is relatively small.

Facility located at a remote mine siteThe production of OPO at a remote mine sitefacility in Boddington, Western Australia,when compared to the production of dieseloil, reduces the production of carbonemissions by approximately 77%. Energyconsumption across the lifecycle is alsosigni cantly reduced by 81%, and 82% ofthe non-renewable resources normallydepleted from diesel oil production areavoided.

1E3kgOnyx Process Oil

72.5kgCO 2e

1.05E3kgWashed oil

(settled)

12.7kgCO 2e

947kgSteam

8.65kgCO 2e

1.2E3kgCooled water

(from cooling tower)

26.8kgCO 2e

1.1E3kgWaste oil and water

mix (for settling)

11kgCO 2e

40.7kgSteam (boiler)

8.65kgCO 2e

183MJElectricity,

high voltage, Victoria/AU U

64.2kgCO 2e

154MJEnergy, from

natural gas/AU U

8.85kgCO 2e

167MJElectricity brown

coal VIC, sentout/AU U

61.8kgCO 2e

1.05kg Volatiles

(carbon scrubber)

7.55kgCO 2e

Fig 3 Climate change impact fows (CO2e) o Onyx Process Oil xed plant acility Fig 5 Onyx Process Oil

Fig 4 Carbon scrubbers

However, the production of OPO has beendetermined as having more impacts to photochemical oxidant formation. Themajority of these impacts are associated withvolatiles that have been vented from thevacuum pump and are released to theatmosphere.

Contextualisation of impactsWhen compared to the production of virginoils, the avoided impact of the annual production of OPO at a remote mine sitefacility, and xed plant respectively, isequivalent to:

Removing 99 to 232 average Australiancars off the road each year in terms ofcarbon emissions

Reducing annual energy consumptionby that consumed by 292 to 684 average

Australians, when considering totalembodied energy

Avoiding the burning of 46 to 112 tonnesof brown coal, when considering non-renewable resource depletion

Additionally, the ReCiPe analysis was usedas a basis for contextualisation of impacts ina global and time-based perspective.

The ReCiPe analysis undertakendemonstrated that under all the considered perspectives of a global system, the use ofOPO led to less aggregated impactscompared to virgin oils (see Fig. 3). A majority of the aggregated impacts forvirgin oils were related to human healthtoxicity in a long-term horizon.

Conclusion and next stepsThere are signi cant environmental bene tsfrom the production of OPO compared to the production of virgin oils. Overall, OPO isless carbon intensive and reduces the relianceon fossil fuels compared to virgin oil. Whenconsidering the total volume of explosivesconsumed across Australasia and globally theimpact of the OPO environmental bene t becomes more signi cant.

There are some negative impacts associatedwith OPO, with regard to human healthtoxicity. These impacts can be attributed tothe worst-case scenario consideration ofheavy metal release in the blasting of ammonium nitrate explosives. However,these emission concentrations appear to bewithin Australian and relevant state air quality limits, which were established toconsider human health impacts.

Overall, the puri cation and use of waste oil presents an economically viable andenvironmentally bene cial alternative to the extraction and use of virgin oils for manufacturers. From a life cycle perspective,waste oil treatment and use has the potentialfor signi cant environmental bene t ifadopted at a global scale. Additionally,considering long-term and global perspectives, the use of OPO presents lessaggregated impacts when compared to virgin oils.

Acknowledgements

We would like to thank Orica SpecialtyChemicals (Tony Palmer, Andrew Blake, and Albert Chemali ) or part- un ding a nd thetechnical support required or the study. Wewould also like to thank Li e Cycle Strategies(Tim Grant) or the third-party peer-review. Theauthors acknowledge its unders, the Plasticsand Chemicals Industry Association (PACIA)and the Environmental Protection Authority o Victori a (EPA VIC) through the Re wards undingprogramme.

References

Goedkoop M.J., Heijungs R, Huijbregts M., DeSchryver A., Struijs J., Van Zelm R. Ruitme enMilieu (2009).Lundie. S., Huijbregts. M., Rowley. H., Mohr.N., Feitz. A. Journal o Cleaner Production 15(2007) 819. AS/NZS ISO 14040:19 98, Int ernati onalOrganisation or Standardisation.

Value

Build and enhance existing clientrelationships with OSC and with thePlastics and Chemicals Association ofAustralia

Link new client relationships betweenOSC and BHP Billiton

I ncreased understanding in thequanti cation of environmentalbene ts from waste oil

Contribute life cycle assessment datainto the newly developed AusLCIdatabase

Enhance the strength of LCAconsulting services to the marketplace

Fig 7 Comparison o environmental impacts o Onyx Process Oil with virgin oils ReCiPe analysis

818

1022

13

19

29

43592

29372

21

I n d i v i d u a l i s t

H i e r a r c h i s t

R e

C i P e p e r s p e c

t i v e s

E g a l i t a r i a n

Onyx Process Oil Fixed plant model

Onyx Process Oil Facility at remote iron site

SN150

Diesel Oil

0 100 200 300 400ReCiPe single-score result

500 600 700

Fig 6 Oil samples

10 11


7/37

Veri cation of CFD for air ventilation analysis in a dense urban environmentArup: Raymond M H Yau, Rumin Yin, Sui-Hang Yan, Jason Yang

AbstractWith the rapid development of theurban environment in cities such asHong Kong, air ventilation has gainedincreasing public attention over the past decade.

As an advanced tool to study air ventilation in the urban environment,there are various international windcomputational uid dynamics (CFD)guidelines across different regions,such as COST Action C14 for Europeancountries. However, the urbanmorphology for these countries isrelatively low-rise and low density. Forcities like Hong Kong, there is not yet acommonly accepted guideline that is

suited for ventilation assessment in thehigh density context. The current studyattempts to extend the application ofDetached Eddy Simulation (DES) to adense urban environment. To carefullyaddress this topic, the study was carriedout with close collaboration with theChinese University of Hong Kong, theHong Kong University of Science &Technology, Tohoku University andTokyo Polytechnic University.

A series of simulations were carried outto assess the suitability of the proposedmethodology of CFD for the denseurban condition in Hong Kong. The methodology was put through acomparative study with a wind tunnel

test to nd out the capability of the DES model in general. Then, a blind test through a third party wasconducted as a detailed validation on

both the methodology and outcome of the CFD study to ensure the robustnessof application on similar urbanenvironment.

The study found that the proposed CFDmethodology was suitable for studyingthe pedestrian wind environment in thedense building context. The DESturbulence model could better predictvelocity ratios at individual test pointswhen compared to the velocity ratios provided by standard k- turbulencemodel.

IntroductionFor urban ventilation study, both CFD andwind tunnel tests have the ability toapproximate the large-scale, complexunsteady ow. For CFD, it has beengenerally accepted to be accurate under awide range of applications, but it has not

been fully validated for the urbanenvironment, especially a dense buildingcontext like Hong Kong. Unfortunately, it isin such big cities that ventilation becomes a big concern and requires careful study. Tocarry out and validate a methodology for robust CFD application such as AirVentilation Assessment, a solid reference isrequired to be compared with, such as on-sitemeasurements. However, site windconditions are time-varying in nature andcontain many other uncertainties, e.g. thecomplexity of street conditions near pedestrian level. The best readily availablesource of comparison is the scaled windtunnel test. Although a wind tunnel test isunable to achieve complete dynamicsimilarity to real site conditions and hasinaccuracies in taking measurements withinthe boundary layer, it is an acceptedmethodology for studying the urban windcondition. Therefore, this study aimed toestablish the reliability test of CFD throughcomparison to wind tunnel data and therelative accuracy between the two methodswas assessed.

It is well known that the problem of thestandard k- turbulence model is that itcannot reproduce the separation and reverse

ow at the leeward side of a building due toits overestimation of turbulence energy k atthe impinging region of the building wall. To propose a more advanced and suitablemethodology for urban environments such asHong Kong, an international standard could be referred to: the COST Action C14 forEuropean countries and The Architect

Institute of Japan (AIJ) Guidelines. Based onthese standards, a similar but tailor-madeCFD methodology using DES is proposedfor Hong Kongs unique built environment.To investigate a typical building context inHong Kong, the famous Tsim Sha Tsui areawas selected as the study area for an initialtest of the proposed methodology. This areacovers more than 100 hectares and comprisesa large number of closely packed low-riseand high-rise buildings. The 3D model wasconstructed and simulated by using both k-and DES turbulence models. The reliabilityof CFD methodology was assessed byreferring to the wind tunnel test results andcorresponding accuracy and issues of eachmethod were discussed. A follow up studywould be conducted at another 80 hectaressite in Tuen Mun area, located in south west New Territories. The comparison betweenCFD results and wind tunnel test results wasconducted independently (a blind test) by athird party to investigate the correlation ofdifferent sets of data. This impartial anddirect comparison process would beconsidered as a validation on both themethodology and outcome of CFD study.Both standard k- turbulence model and DESturbulence model were simulated.

MethodologyTo achieve accurate prediction of wind, thesetting of CFD study needs to be carefullydesigned, including the 3D model andcomputational method for the simulation.This methodology took into considerationavailable wind CFD standards and how thewind tunnel test was conducted so that themethodology could best reveal urban windconditions while at the same time allowcomparison to the wind tunnel result.

Building the 3D modelThe CFD models of two sites are built withreference to the wind tunnel model andavailable site plan. The buffer area providedfor all lateral, in ow and out ow area was5H (H = the height of the highest building insite area) in Tsim Sha Tsui. Minor edge effectwas noted in the simulation of Tsim Sha Tsuiarea, hence this length was increased to 6Hfor the Tuen Mun model.

Setting up turbulence modelThe most important factor of CFD simulationis the selection of the turbulence model.More sophisticated models means nertreatment of wind turbulence but morecomputational effort. Therefore, whenchoosing the appropriate model for thisstudy, both factors required carefulconsideration. To accomplish this, theadvanced CFD technique, namely the DESwas employed. The DES turbulence model is an hybrid model of the Large EddySimulation (LES) turbulence model to thearea which classical LES is not affordablein terms of computational resources (e.g.high-Re external aerodynamics simulations)without sacri cing the advantage of the LESmodel. This model combines thecomputational ef ciency of ReynoldsAveraged Navier Stokes (RANS) models,and the accuracy of LES in the separatedregions, a key requirement for accuratemodelling of external air ow.

Fig 1 CFD simulation o Tuen Mun area

Fig 2 Comparison o turbulence models with wind tunnel test

An advanced CFD tool withimproved turbulence modelwas employed to study the airventilation in a compact andhigh-density urbanenvironment in Hong Kong.A robust methodology wasestablished throughinternational academiccollaboration. Similar methods can be employed tostudy the microclimate at thescale of cities.

0.7

0.6

0.5

0.4

0.3

0.2

0.1

0

C F D s i m u

l a t i o n

Wind tunnel

0 0 .1 0 .2 0 .3 0 .4 0 .5 0 .6 0 .7

DESk-

12 1313


8/37

In order to resolve the detailed ow structurewithin the area of interest, a ne scale grid isneeded in the assessment area to capture the

ner scale eddies within the ow. Due to thecomputational resource for DES simulations,the grid would act as a lter to distinguishlarge and small eddies that are explicitlysimulated or passed on into the sub-gridmodel. Hence an adaptive meshing approachwith varying size mesh was proposed. Forthis study, the vertical grid expansion ratiowas set to 1.5 from ground level asrecommended in COST Action C14 and AIJguideline. The main street with measurement points (or test points) was meshed with 6-10cells across its width. A prism layer will becreated for the rst 5m above ground, whichwas also consistent with COST Action C14and AIJ guideline, with at least 0.5mresolution to capture the boundary layer ow.

Given the high Reynolds number of the ow,near-wall turbulent ow structures areassumed to have a relatively small effect onthe bulk ow. These structures can only berepresented if the grid resolution is extremely

ne. Such spatial resolution would lead to asmaller time step due to an increased CFLnumber, making wall-resolved CFD of this

ow unfeasible. Therefore, an approximatewall boundary condition is used in this work.The logarithmic law boundary condition isapplied at the ground level.

Other settings of the CFD model such assolving algorithms and convergence criteriaare listed in Table 1.

Results and discussionThe velocity ratios (VR) between a windtunnel test and a CFD simulation, which isde ned as pedestrian wind velocity overwind velocity at the top of wind boundarylayer (usually at 500m or above ground),were compared for both of the selected wind

directions. Results of average velocity ratiofor both k- and DES turbulence modellingare presented in Table 2.

In comparing the results obtained from awind tunnel test against the computationalmodels, the DES model gave lower deviationin average VR values for both East andSouth East wind conditions k (see Fig. 4).This indicated that the DES model is moreaccurate than the k model in predictinggeneral wind conditions of the study area.On the other hand, in terms of the individualVR value, the R-square (represented byaccumulated squared deviation from thewind tunnel result) was also larger for k,which meant a larger error (higher R-Squarevalue). This also implied that the DES modelis better in predicting the magnitude of the

average VR values and offers better quantitative predictions as compared to k-turbulence model. The initially proposedmethodology for DES modelling waseffective in providing comparative predictionwhen referenced against wind tunnel results.

With some further improvements on CFDwith the purpose of eliminating edge effectcaused by the insuf cient CFD domain size,the study case at Tuen Mun further veri edthe value of applying a DES turbulencemodel to the urban wind environment. TheDES model offered an average velocity ratiocloser to the wind tunnel test result as shownin Table 2. Moreover, Fig. 6 showed that theresults simulated by DES model were closerto the diagonal line, and along with CFD andwind tunnel, yielded identical results. Withreference to the wind tunnel test results, Fig. 4 showed that the DES model offeredmuch smaller cumulative deviation than thestandard k model, indicating better prediction of wind condition in speci clocations by DES, whereas k model tendedto have a relatively larger error.

Fig 3 Mesh o Tsim Sha Tsui Area

Table 2 Local velocity ratio (LVR) comparison between windtunnel and CFD

Table 1 CFD model setting and processing

Fig 4 R-square values o DES and k- turbulence models

2.0

1.5

1.0

0.5

0

R - s q u a r e v a

l u e

E wind N windSE wind SSW wind

DES

k-

Site WindDirection

WindTunnel

TestDES k-

Tsim Sha Tsui

EWind 0.18 0.17 0.12

SEWind 0.17 0.16 0.13

TuenMun

NWind 0.22 0.20 0.25

SSWWind 0.23 0.24 0.25

Model Scale 1:1 Model

Model details Include topography, buildings, streets, ootbridges

DomainIncludes whole building in ormation within the calculation boundary:Stage 1: lateral: 5H; Infow: 5H; Outfow: 5HStage 2: lateral: 6H; Infow: 6H; Outfow: 6H

Surrounding building area 2H area

Grid expansion ratioIn vertical direction: 1.5 rom low levels;In horizontal directions, the grid should satis y the grid resolutionrequirement and with maximum expansion ratio = 2

Prismatic layer Prism layer must cover pedestrian level, spacing recommend 0.5m orthe rst 2m above ground level

Grid Resolution Minimal 6-10 cells or street width with measurement points

In ow boundary condition Incoming wind pro le and turbulent intensity as measured rom wind tunnel

Out ow boundary Pressure boundary condition with dynamic pressure equal to zero

Wall boundary condition Logarithmic law boundary

Solving algorithms QUICK or momentum equation;MARS or all other equations

Convergence criteria Below 1x10-4

Blockage ratio < 5%

Conclusion and next stepsIn this research, the two 80-100 hectares siteswere simulated by CFD using both standardk- turbulence model and DES turbulencemodel. The results of CFD and wind tunneltests were assessed, and the correspondingaccuracy and issues were discussed.

By adopting the method through the CFDmodel of Tsim Sha Tsui area, the DES resultsof the CFD model of Tuen Mun area werecomparable to the wind tunnel test data. Forthe DES model, analogous to the ndings forTsim Sha Tsui, the velocity ratios deviatedless from the wind tunnel result as comparedto the k- model. The simulation results alsoshowed that the DES model was able to predict velocity ratios at individual test points more accurately compared to thestandard k- model.

On the choice of turbulence model, COSTAction C14 and AIJ guidelines haveconcluded that the standard k modelshould not be used in simulation for windengineering problems, as demonstrated bythis study.

Based on the comparison results andliterature review, the followingrecommendations were made.

The DES turbulence model could beconsidered as good tool for predictingurban wind conditions, such as inHong Kong.

A standard k- turbulence model should beavoided when conducting Air VentilationAssessment, whereas a DES model couldful l the requirement.

Acknowledgements

Pro essor A. Mochida, Pro essor R. Yoshie,Pro essor Y. Tominaga, and Dr H Kataoka , The Architec tural I nstitu te o Ja pan (AI J) CFD G roupPro essor Edward Ng, School o Architecture, The Chi nese U niversi ty o H ong KongPro essor Jimmy Fung, Department o Mathematics, The Hong Kong University o Science & TechnologyPro essor S C Kot, Senior Lecturer Emeritus,Department o Mechanical Engineering, TheHong Kong University o Science & Technology

or their invaluable comments. Thanks are also due to Dr Trevor N g, Aru p or h isreview o the paper.

References

Ghosal, S., Moin, P., (1995), The BasicEquations or Large Eddy Simulation o TurbulentFlows in Complex Geometries, Journal o Computational Physics,1995, 118, pp24-37Spalart, P., Jou, W.-H., Strelets, M., Allmaras,S., (1997) Comments on the easibilty o LES orwings and on the hybrid RANS/LES approach, Advances in DNS/L ES, 1st AFOSR In ternat ionalCon erence. on DNS/LES, Greden Press, 1997. Tominaga, Y., Mochi da, A. , Yoshie, R., Ka taoka,H., Nozu, T., Yoshikawa , M. and Shirasawac, T.,(2008) AIJ guidelines or practical applicationso CFD to pedestrian wind environment aroundbuildings, Journal o Wind Engineering andIndustrial Aerodynamics, 96(10-11):1749-1761 Yau, R.M .H., Yan, S.H. , and Yin, R ., (200 8),Air Ventilation Assessment by External WindModelling Using Di erent Turbulence Models Arup Re search Review 2008.

Value

Systematic setup and veri edmethodology of using advancedDetached Eddy Simulation (DES)turbulence model for a dense urbancontext such as Hong Kong

First piece of research undertakenfor dense urban built environmentapplication

Accurate computational predictionof complex external air ow byDES with much less requirement ofcomputational resource compared toLarge Eddy Simulation (LES)

Alternative ef cient tool to studyair ventilation with good accuracycompared to scaled wind tunnel

Built good relationship with leadingresearch collaboration teams in HongKong and Japan

Form the basis of future air

ventilation assessment in Hong Kongfor detailed study by computationalCFD method

Fig 5 CFD simulation o Tsim Sha Tsui area

Fig 6 Wind tunnel model o Tsim Sha Tsui area Fig 7 Contour plot o velocity ratio (VR)

14 15


9/37

StructuralComponents: a parametric and associative toolbox for conceptual designArup: Anke Rolvink, Bastiaan van de Weerd, Jeroen Coenders

AbstractStructuralComponents is a parametricand associative design and estimationtoolbox for conceptual structuraldesign. The current implementation isaimed at tall buildings. The tool allowsthe structural engineer to quicklycompose design reasoning and justi cation which forms the conceptualstructural design in a matter of minutes,while simultaneously running analysesto explore the resulting performance ofthe building.

The analysis results are immediatelydisplayed to the engineer and serve as a basis for further conceptual designdecisions, creating a continuous loop ofdesign justi cation, exploration andadjustment. StructuralComponents aimsto describe this process as computablelogic which can be used, reused andadjusted with instant feedback. Thiswill document the argumentation in a

exible form and provide a starting point for further design.

The philosophy behind the tool is toallow the engineer to easily and quicklycompose the conceptual justifying storyfor the structural design of a buildingand analyse its performance. This storynow often only exists in the brain of thedesigning engineer and describes thedifferent responses the building willgive to different scenarios. Therequirement for ease and speed stemsfrom the dynamic nature of early stagedesign, where architects make quickchanges and developers want solidestimates.

IntroductionDesign itself, whether it is structural designor any other form, evolves through theiterative cycle of proposals, testing andmodi cations. Important in early designstages is to reason through and justify thedesign concept step by step to get anunderstanding of the relationship betweenthe scenarios the design has to face and itsresponses to those scenarios. The aim of thisdesign justi cation is to build con dence inthe solution, used methodology, methods,tools, etc. A designing engineer makes use ofdifferent methods to achieve this goal. Priorknowledge and experience are combinedwith forecast methods to predict the performance of a design.

Justi cation often takes the form of anexplanatory story that sometimes only existsin the mind of the designing engineer. Thisstory of justi cation explains how a structureworks, how it deals with different scenariosand with unforeseen situations. This justi cation story is based on gatheredevidence and application of logic, makinguse of reasoning, thought, schematisation,calculations, modelling, analysis and thedevelopment of scenarios. These justi cations are based on a set of designassumptions and the de nition of therequirements, design constraints and boundary conditions. Forming the conceptualstory concludes the conceptual design stageand allows the engineer to present thegenerated solution to the other parties in thedesign process or to the client.

In the current design practice, computationaldesign tools are not commonly used tocompose, de ne, explore, communicate andvisualise structural design alternatives and tosupport the engineer with the formation ofthe conceptual story during early designstages. At this design stage highly accuratesolutions are not yet required. It is more

important to quickly compose and assessdifferent alternatives and gain insight in theimpact of decisions. With the current toolsavailable to the engineer, it is dif cult tocompose and justify conceptual structuralsolutions, which contrasts with the often highimpact of decisions made during these earlystages. Towards later stages it becomes verycostly and sometimes even impossible tocompensate for design choices made at the beginning of the design process.

Another problem which can be identi edwhen considering the characteristics of computation and structural design is the black box problem. Software applicationsoften do not provide insight into the internalworking methods and it is therefore hard for the engineer to have con dence in thesoftware. If an engineer does notcomprehend the software and havecon dence in its working methods, he willnot use it. Hand in hand with this problemgoes a problem of inadaptability of mostsoftware. It often is not possible to adapt thesoftware to t certain purposes in a simplemanner.

The objective of this research is the ongoingdevelopment of a design toolbox whichallows for a structural design approach whichcombines parametric and associativemodelling with direct analysis. This supportsthe appropriate speed and accuracy for theearly design stages. The toolbox must assistthe user in setting up the conceptual story ascomputable logic which can be used, re-usedand changed in near real-time and todocument this argumentation in a exibleform which can serve as a starting point forfurther design. This design story wouldalways be able to refer back to the conceptsdocumented to check, control or verify thedevelopments in the design against thedocumented benchmark formed by earlier reasoning.

MethodologyTo demonstrate the above mentionedconcept, a prototype of the toolbox has beendeveloped for the conceptual structuraldesign of tall buildings. This prototype has been implemented in C# based on the .NETFramework. The toolbox has a modularsoftware architecture, so that pre- programmed blocks of logic can bedeveloped which can be recombined as partof the composition process. Theimplementation consists of two parts:

The framework (a layered structure ofclasses, applications, libraries, etc) thatcan be used, extended or customised forspeci c design questions of the toolbox,contains all the design logic, analysismethods, solvers, etc. It has been setup independently of existing softwareapplications, which offers the possibilityto connect the framework to differentparametric software packages via theuser interface or to build a stand-aloneapplication.

Design, engineering andconstruction process

InformationDesign

freedomDesign

freedom

Fig 1 A design de nition o a tall building set up in StructuralComponents

Fig 2 Oversight o the design reedom during a designproject versus the in ormation available to base decisions on

Using the toolbox, adesigners intuitive reasoningis rapidly recorded, evaluatedand justi ed, providingimmediate deep and veri ableinsight and overview withouthaving to assemble laboriouscalculation models.

16 1717


10/37

The user interface which can be built ontop of the framework and can be loadedinto a parametric and associative softwareapplication. The user interface containsonly visualisation and geometry featuresand the current version has been developedfor Grasshopper, a graphical algorithmeditor tightly integrated with Rhinos 3Dmodelling tools.

Design compositionThe user of the toolbox can compose a modelfrom different building components availablein the components library of the userinterface to build justi cation underneath thedesign concept. This describes differentsituations, scenarios, cases, assumptions, andcombinations of these, as well as checksagainst codes, standards, requirements,experience or common sense. Designcomposition is a process which developswith the engineers insight in the designitself and the performance of the design.

Visualisation on a dashboardAs the user combines various buildingcomponents each embedded with designlogic the design performance issimultaneously analysed and instantaneouslyand comprehensibly displayed on adashboard. The dashboard provides a meansto build the design composition for fastreview, ease-of-use and overview over thecomposition and its underlying logic as wellas the analysis results and checks of thedesign concept being constrained.

Structural design and analysisStructuralComponents provides a largenumber of analysis methods. The main aim isthat these analyses are (near) real-time tostreamline the sensitive and fast changing process of composition. The building

components mentioned in the previous paragraph have structural design logicembedded in the form of simple rules ofthumb, analytical and numerical analysisalgorithms, standards, etc. As the userassembles a design concept from abovementioned components, the structural performance is automatically analysed.

During the rst composition steps, mainlyrules of thumb, codes and standards are usedto provide feedback to the user, such as thewind load on the building, possible structuralcon gurations and required distribution ofstiffness. As the design heads forward theanalysis becomes more advanced. A niteelement solver has been implemented,amongst other solvers, into the framework.This version is based on super-elements,whereby a tall building model is composedfrom a limited amount of elements, such ascores, outriggers and columns. These areanalysed using super-elemental differentialequations to provide fast analysis with highaccuracy.

Results and discussionThe research has resulted in a prototype ofStructuralComponents which provides fastmodelling of abstract concepts for structuraldesign of tall buildings. The developmentefforts around StructuralComponents havefocused on a proof-of-concept and resultedin an extensible toolbox architecture with anindependent computational core framework.The toolbox augments and complements theintelligence and creativity of the engineer during the conceptual design stages withcomputational output. The result is thatstructural insight is provided, even duringearly stages, without having to assemblelaborious models, while enabling theengineer to see more accurately thein uences of changes on the key designdrivers from the beginning of the design process. In this way the engineer is able todescribe the conceptual design process in aform of computable logic which he or shecan use, reuse and adapt in near-real timeand which can serve as a basic for further design.

Fig 4 The latest developments are based on a client-server architecture, which allows the development o the tool or thinclients, such as the iPad

Number of outriggers: 3

Optimum outrigger locations:17,37,60 storey

Core bending stiffness:1E9 (kNm 2 )

Columns bending stiffness:6.4E7 (kNm 2 )

Maximum wind load: 3.66 (kN/m 2 )Building height: 312m

Number of storeys: 80

Storey height: 3.9m

Building width: 40m

Building depth: 40mLateral displacement Bending momentWind load

Output dataInput data Outrigger structure

320

280

240

200

160

120

80

40

0

B u

i l d i n g

h e i g

h t ( m )

-5 0

Displacement (m)

Maximum allowed displacement

Displacement of outrigger structure

5 10

320

280

240

200

160

120

80

40

0

B u

i l d i n g

h e i g

h t ( m )

-10 -5

Moment (millions kN/m 2 )

Moment caused by external load

Bending moment in core

0 5

320

280

240

200

160

120

80

40

0

B u

i l d i n g

h e i g

h t ( m )

-500 0

Wind load (kN/m 2 )

Total wind load along building height

Applied wind load along building height

500 1000

Alternatives synthesis

Server/cloudThin client

Parametric-associative modelling

Realtime performance feedback

Performance index 1Range 0.0 - 1.0

Performance index 2Range 25 - 500

Finite element analysis

101100001

010001010

101110101

011001111

100100000

101010010

011001100

101100001

010001010

101110101

011001111

011001111

100100000

101010010

011001100

Operation

Operation

Operation

Param

Param

Result

Result

Fig 3 Example o some analysis results o the toolbox

This research shows that new design andengineering opportunities arise from buildingtools on new digital technologies. Custom-developed design tools and tool strategiescan focus on more speci c engineering problems than standard software packages,which are more broadly focused and primarily target later design stages. Thesedesign tools tend to have an open structureand allow the user to understand, review,adapt and complement its working methods,avoiding the black box syndrome. This offers

exibility in early design stages and allows achoice of those tools that best suit the uniquenature of the design process.

It is important to note that it is not enough just to develop new and advancedtechnologies. The rate of future success ofearly stage design tools depends on theiradaptability and recognition within theengineering practice. The use of thesetechnologies requires a change in currentworking methodologies and the engineer needs to innovate in order to use andunderstand the new technologies. Theengineer must have con dence in thetoolbox, before he or she will use it fordesign purposes. To gain con dence, theengineer must know what the toolbox canand cannot do and therefore needs some basic knowledge concerning parametric andassociative design. Second, since each design project is unique, the toolbox will inherentlyhave to be adaptable and allow extensionwith new methods and algorithms.

Conclusion and next stepsThis article has presented the development of StructuralComponents. The developed prototype is a proof of concept and a medium to explore ideas. Enhancements arerelatively easily made due to the modular set

up which allows the toolbox to be extendedfor future design needs.

The latest developments, aside fromincremental updates such as improvingvisualisation and summary output,implementing new design and analysismethods and incorporating design methodsfor other disciplines such as MEP or lightingand other structural typologies, such asstadiums or bridges, focus on the following:

To build the next version ofStructuralComponents as a stand aloneapplication. Running the program on anunderlying proprietary software package from athird party developer requires licenses for everymachine, limiting mobility and potentiallycreating future issues of dependency. Astand alone application allows a much greaterdegree of development freedom. Furthermore,the choice was made to continue developmentalong a server-client architecture with webservices interfaces. Computations will be

performed on capable server side hardwarewhile thin clients such as on an iPad, in aweb browser or on an embedded device provide the end user interface.

To focus more on the generation of a rangeof design alternatives by incorporating abandwidth approach to the designcomposition. The toolbox would allow theuser to initially specify design variables aslying within a range instead of as a singlevalue, leaving the nal design speci cationto later stages where more information isavailable, provided that the values staywithin the earlier speci ed range. Thetoolbox analyses the synthesisedalternatives and presents visual feedbackon the result of changing design variablesin speci c directions, continuously guidingthe user throughout the process.

Acknowledgements

This r esearch has be en per ormed as a cl osecollaboration between the BEMNext laboratoryo Del t University o Technology and Arup Amsterda m. It is par t o on going r esearch intothe development o new digital design tools tosupport (structural) design processes.

References

Coenders J., Wagemans L., The next step inmodelling or structural design: structural designtools, Proc. o the International Association orShell and Spatial Structures, Theory, Technique, Valuation , Main tenance , 2005, pp. 85-9 2.Fricke G., Success ul individual approaches indesign engineering, Research in EngineeringDesign, vol. 8, pp. 151-165.Liberty, J. and Xie, D. Programming C# 3.0.OReilly Media, Inc. i th edition, 2007.McNeel (2010). Grasshopper - GenerativeModelling or Rhino. Retrieved June 20, 2010,

rom: http://www.Grasshopper3D.com/ Rolvink, A., van de Weerd, B., Coenders, J.StructuralComponents, Proc. o the IABSE-IASS symposium, Taller, Longer, Lighter,London, UK, 2011.

Sta ord Smith, B. & Coull, A. (1991). Tallbuilding structures - Analysis and design, JohnWiley & sons, Inc.Steenbergen, R. (2007). Super Elements inHigh-Rise Buildings under Stochastic WindLoad, PhD thesis, Del t University o Technology.

ValueThe embedded design logic in tools based on a parametric and associativeapproach provides the designingengineer:

A means of setting up design conceptseasily and quickly during early designstages

Better insight in the in uence ofdecisions during the early designstages, avoiding costly changes laterin the process

The possibility to explore a wide

range of design alternatives in aneffective and ef cient way

Fig 5 Conceptual story and design justi cation

Conceptual story

Justification

AssumptionsBoundary conditionsDesign constraints

RequirementsCodes and standards

R e a s o n

i n g

T h o u g

h t m o

d e l

S c

h e m a

t i s a

t i o n

A n a

l y s i s

S c e n a r i o s

C a s e s

M o

d e

l

S i m p

l i f i c a

t i o n

18 19


11/37

Arups 4see model: combining socio-economicand energy aspects of a whole economyArup: Simon Roberts

AbstractIn support of low-carbon transition plans, there is extensive research on theindividual technologies needed for renewable energy and energyef ciency. In contrast, there is muchless work on analysing whole transition plans in a combined physical andeconomic approach for an entireeconomy. Arups 4see modellingapproach has been developed to ll thisgap. Its purpose is to identify andexplore the systemic impact ofinvestment options. A 4see model isdata-driven using of cial statistics,

bringing together economic activity,capital stocks, employment, energy,transportation and balance of payments.As in embodied carbon analysis, 4seemakes the link between end productsand services delivered by the economyand the source fossil fuels consumed. It takes this further, capturing otherembodied concepts en route togetherwith their time dependent trends andsystemic interactions.

Applying the approach in a UK model,a business-as-usual scenario suggests

that increasing economic activity andtravel will cancel ef ciency gainsresulting in constant emissions. Themodel can be used to quantify how theinvestment part of gross domestic product (GDP) must be increased todeliver low-emission policies showingthe systemic consequences of these.4see will be applied in future work toevaluate low-carbon investments, suchas wind turbines and energy ef ciencyretro ts to existing buildings.

IntroductionThe UK economy consumes fossil fuelsgiving rise to CO2 emissions, a keycontributor to climate change. The sector

break down of these emissions and their variation over time are shown in Fig. 2: (a)using historical data, and (b) in a forecast to 2030.

The historic half from 1990 shows only asmall reduction in emissions, despite theincreasingly high pro le and concern thatclimate change now receives. The projectionin the right half is a possible business-as-usual scenario from Arups 4see model. The4see model continues historical trends inenergy ef ciency of vehicles and buildings.It also trends changes in nal demand thatgive rise to economic growth and increasingtravel. The outcome of constant overallemissions on the right shows that ef ciencygains will be cancelled by greater activity.

Reduced emissions could be achieved in ascenario of economic recession since thisreduces general activity, but a recessionexacerbates unemployment so is the leastdesirable option. To bring emissions down,investments are needed at a faster rate thantheir market viability would otherwisewarrant. However at a macro level,irrespective of the detail of theirimplementation instrument, such assubsidies, tax concessions or regulation, they all divert nal output from the economyto investment.

This research is about the novel approach ofa 4see model to understanding this systemic problem and the development of optimalinvestment plans. It examines the question ofhow a nation can adopt a low-carbontransition without increasing unemploymentor bankrupting the economy.

MethodologyThe 4see model was developed to capture thesystemic interactions within a national

economy of: economic activity, physicalassets, employment, energy generation anduse, transportation and balance of payments.

We know from economics that a marketconsists of supply and demand. The 4seemodel has full coverage of the supply sideincluding the way supply might be limited,either by physical constraints, by the needfor investment or by time needed to makechanges. The model then splits demand intotwo parts: intermediate demand at each stepthrough the supply chain on the one hand,and nal demand by end consumers on theother. In simple terms, a 4see model sets outto answer the question: given projectedvalues for nal demand, what necessarysteps in supply and intermediate demand are needed? For various scenario projectionsof nal demand, the model computes theconsequences for supply along withindicators such as unemployment, CO2 emissions and dependency on importedenergy.

A 4see model takes as its starting point a physical perspective of economic assets:goods need factories for their manufacture;services require of ces; even transactions on the web require computers that have a physical location in a building. Physicalassets ( xed capital) consist of buildings,machinery and transport equipment whichtogether make up economic sectors, such asmanufacturing, construction and the servicesector. Data for physical assets comes fromnational accounts details ofcapital stocks ,these being derived from the size of investment needed to make the asset, thendepreciated according to asset life times.

The next step in modelling each sector is thevolume of economic output (domestic production) of goods, buildings and services,also known as gross value added (GVA)from national accounts. Modelling of eachsector is completed by detailing inputsneeded from various data sources.

Consider the example of the service sector in Fig. 1(a) which shows historical data forinputs of gas and jobs, physical assets andeconomic output of services (GVA). The4see model rst compares the economicoutput to the size of the asset, and thencompares each input necessary for theoutput, these ratios shown in Fig. 1(b).

For economic output compared to physicalassets (return on capital), the result isremarkably at over the historical period inthis example. It seems reasonable, at least toa rst approximation, to extend this constantratio into the future. For the amount of gasand jobs each needed per unit of output,these both show falling trends over thehistorical period. These trends areextrapolated into the future using best- tdecay curves, as shown.

Having examined one sector in detail (Fig. 1), lets now consider how the 4seemodel covers the whole UK economy and itssystemic interactions. This aspect of themodel can be represented by the diagram inFig. 3 which extends from inputs on the leftof fossil fuels used in the economy and jobs

Year

800

600

400

200

0

M ( 1 9 9 0 ) 0 0 0 / y

Year

0.06

0.03

01990

Ratio jobsto output

k / ( M ( 1 9 9 0 ) / y )

2000 2010 2020 2030

600

450

300

150

01990

GasdemandInputs

(a)

(b)

Output/asset size Input/outputInput/output

Asset size Economic output

Year

Year

Year

P J / y

1994 1998 2002 2006 2010

40

30

20

10

01990

Jobs

k ( 0 0 0 )

1994 1998 2002 2006 2010

1.00

0.75

0.50

0.25

0

Ratio outputto assetsize

( M ( 1 9 9 0 ) / y )

/ M ( 1 9 9 0 )

Year

0.002

0.001

01990

Ratio gasuse to output

( P J / y )

/ ( M ( 1 9 9 0 ) / y )

2000 2010 2 0 20 2 0 301990 2000 2 0 10 2020 2030

Year

1,2000

900

600

300

01990

M ( 1 9 9 0 ) 0 0 0

1 9 94 199 8 2 0 02 200 6 2 010 1990 19 9 4 1998 20 02 2006 2010

600

450

300

150

01 990 20 00

(a) (b)

2010 Year

Transport

Housing

Manufacturing

Othersectors

Service sector

M t C O

2 / y

20 20 20 30

Fig 1 Service sector: (a) historical data or inputs, asset size and economic output; (b) ratios extrapolated into the uture

Fig 2 CO2 emissions rom ossil uel use in the UK economy:(a) historical data; (b) business-as-usual scenario using the4see model

Renewables and energyef ciency measures reduceemissions but are capitalintensive. Such measuresshould be evaluated against acountrys physical ability to provide the goods,construction capacity andservices that form the basisof investment.

20 2121


12/37

through to nal consumption expenditure byconsumers on the right. The form is a Sankeydiagram in which the line widths are proportion to ow. (Width scales for thefossil energy and electricity lines areequivalent in energy terms.) Note that thearea of the yellow boxes does not correspondto the size of the assets they represent. Forclarity, the diagram misses off balance of payments and assets of the smaller economicsectors.

There are three destinations for nal outputsof the economy. Largest is consumption byconsumers, either as direct purchases or bene tting from government services paidfor indirectly through taxation. At the bottomare exports of goods and services. The last isthe return ow to sector assets that enablesthe economy to grow, known as gross xed

capital formation or the investment proportion of GDP. Historical data for the parts of nal demand and imports are shownin Fig. 4(a).

An expanded Sankey diagram of theeconomic sectors is shown in Fig. 5 detailingthe inter-sector ows, using data fromSupply and Use Tables in national accounts.For example, imported manufactured goodsstarting at the top left are combined withdomestic manufacture or GVA. Somemanufactured goods are used by the servicesector, so embodied in their nal output atthe bottom right.

DiscussionA novel feature of the 4see approach is themixture of units, such as monetary value,energy and jobs. Insight can be gained byusing time series data to examine the

mixed-unit ratios over time (Fig. 1). Wherethere is a clear historical trend, this insightinto the sectors dynamics forms the rst stepin the foresight process of creating futurescenarios. Using the complete data setscompiled for a 4see model, rst-orderextrapolations of historical insight are derivedfor all the variables enabling a f oresightscenario to be created, as in Fig. 2(b). Thoseextrapolated ratios that give rise to majorcomponents of the scenario output mightwarrant revisiting, perhaps consulting withappropriate experts for a more informed viewof future trends. 4see scenarios can either beiterated with revised extrapolations orsubjected to sensitivity analysis.

Another consequence of not being tiedonly to monetary values is to select unitsappropriate to each variable. The electricity

bill might be much less than labour costs, butelectricity is essential since if it goes off business operation is jeopardised. Electricityin energy terms is a necessary input,complementary to other inputs and assets.

The complementary ratios of inputs are notxed but evolve over time. For example, as

investment replaces depreciation and to growthe business, there are opportunities forchange; replacement machines are oftenmore automated, desirable for reducinglabour costs. The behaviour of input ratioschanging over time is referred to here asevolving complementarity , a key aspect of4see; occurring at a micro level, it manifestsat a macro level.

The economic dependency between sectors(Fig. 5) is also complementary and can berelated to the work of the economist Wassily

Lontieff who rst applied input-output (I-O)tables to the USA economy in the 1930s.

He treated the I-O tables as a matrix andused his Lontieff inverse function tocompute the input requirements needed tomeet output objectives. Several countriesused to use this technique to develop their

ve-year economic plans, but its dependencyon xed technical coef cients is a limitationgiven that they cant be assumed to remain

xed over time. In contrast, 4sees evolvingcomplementarity fully embraces continualchange.

The complementary aspect within 4seeextends to the inputs for investment fromthree economic sectors, as illustrated in Fig. 3. Supposing there is increasingconsumer demand for services, this wouldneed increased investment in the servicesector, and following back along the blue linefor investment from this sector shows thesystemic consequences across the economy.

oil

net imports

jobs

unemployed

other

other

coal

gas

net imports

net importsnet exports

petroleum products

otherother

exports

consumerpurchases & governmentservices

totalelectricityincludingnuclear &

renewables

manu-facturing

service

sector

construc-tion

housing

transport(assets

includedwithin

economicsectors)

refinery

FINALDEMAND

INVESTMENT

imports

imports

G D P & i m p o r t s

FINALSUPPLY

construction

goods

services

fossil energy(500PJ/y)

economic(50G[1990]/y)

electricity(100TWh/y)

jobs(5M)

oilpet. prod.coalgas

invest.goodsconstr.services

Fig 3 Sankey diagram o the UK economy using data or 2008, showing major fows o energy, jobs, economic output and trade. For details o fows between economic sectors, see Fig. 5

Year

M ( 1 9 9 0 ) 0 0 0 / y

1990 20102000 2020 2030

1,500

1,000

500

0

-500

Investment

Consumerpurchases& government

services

Exports

Imports

(a) (b)

Fig 4 Components o GDP: (a) historical data, (b) uturescenario o increased investment

ResultsThere is much research into the technologiesneeded to reduce carbon emissions, such asimproved energy ef ciency, more renewableenergy or using less fossil fuel per unit ofelectricity. Many of the solutions are capitalintensive and currently not competitive orneeding long payback times. There are proposals to correct this disadvantage by

nancial incentives, but all such investment plans have to be evaluated against acountrys physical ability to provide thegoods, construction capacity and servicesthat form the basis of investment. So whenconsidering the investment needs for emissions reduction, can current investment

be diverted?

The Sankey diagram in Fig. 3 is useful forshowing key features of the UK economy, in

particular how services dominate. For jobs,the vast majority are in the service sector.The point in the diagram between nalsupply and nal demand corresponds toGDP plus imports. Since most imports aregoods, it is clear from Fig. 5 that services area major proportion of GDP. Finally, themajority of investment goes to the servicesector (Fig. 3).

Referring to Fig. 1(b), this shows that jobs-needed-per-unit-of-service-output has been declining incessantly over the historical period. Employing staff is a major cost for allservice sector organisations, so it is hardlysurprising that there is a continuous pressureto reduce these costs by ef ciency gains tostay in business in a competitive market.However if unemployment overall is not torise as jobs-per-unit-output continues to fall,then the whole service sector would need to

compensate for the loss of jobs. The historicaldata shows that the economy has managed toaccomplish sector growth but with aconsequent large draw on investment. Thisdiscussion of the service sector suggests thatits signi cant draw on investment is anessential part of its growth and provision of jobs, so cant be reduced. Thereforeinvestment for emissions reduction beyondcurrent market trends needs additionalinvestment which would need to be divertedfrom nal consumption by consumers, asshown in Fig. 4(b). The key point of this shiftfrom consumption to investment is not toreduce overall economic activity (the envelope)otherwise unemployment might increase.

Conclusion and next stepsIn summary, a 4see model works withcomplete data sets and generates insights

from hindsight of historical data to createforesight scenarios. It is a tool to testconsequences of trajectories in nal demand,implications of proactive investments andsensitivity to changes in the insightrelationships. Policies that result are thusevidence-based.

While existing data on embodied emissionscan inform and bias purchasing and designdecisions to lower emissions, there are manysystem constraints and the systemicinteractions are lost by shrinking the resultdown to a single embodied carbon value. Incontrast, a 4see model keeps the granularityof variables through the economy together with their relationships and trends fromhistorical data.

Next steps for the 4see project are toexamine the implications of proactive

investments for low-carbon beyond business-as-usual. For each technologyoption, information is needed onimplementation planning, marginalinvestment cost and marginal energy change.The marginal cost simply forms part of theinvestment component in the 4see model.Technologies that can be evaluated include:wind turbines, photovoltaics, concentratedsolar power in North Africa with transmissionacross Europe to the UK, carbon capture andsequestration, building retro t, LED lightingand range-extended electric cars.

In addition to emissions reduction, 4seemodels can also be used to investigate other issues, such as energy shortages andtechnological solutions to mitigate effects of peak oil on an economy. More inputs can beadded to 4see models, such as water, and the

4see framework is readily transferable fromthe example of the UK here to other nationaleconomies.

Acknowledgements

Colin Axon (Brunel University, London)Charlotte Bonneville ( ormerly at colePolytechnique, Paris)Dave Crane (Historic Futures)Stephen Duncan (University o Ox ord)Barney Foran ( ormerly at CSIRO, Canberra) Adam Fr iedber g (Arup)Nigel Goddard (University o Edinburgh)Emilie Hergott ( ormerly at cole Polytechnique,Paris)Jane King ( ormerly at University o Edinburgh)Jeremy Wake ord (University o Cape Town)Benjamin Warr (INSEAD, Paris) Travis Wi nstanl ey (Univ ersity o Camb ridge)

Value

Sets actions on climate change andsustainability in a country-widecontext

Brings rigor to investment planningat a national level

Thought leadership

Brings disciplines together

imports

imports

servicesector

servicesector

constr-uction

constr-uction

FINALSUPPLY

construction

goods

services

tax

tax

tax

distribution & retail

DOMESTICPRODUCTION

SECTORDEPENDENCY

manu-facturing

manu-facturing

fossil energy(500PJ/y)

economic(50G[1990]/y)

electricity(100TWh/y)

jobs(5M)

oilpet. prod.coalgas

invest.goodsconstr.services

Fig 5 Detail o Fig. 3 clari ying economic fows between the economic sectors. Boxes on the le t represent what the sectorsthemselves produce in addition to imports; boxes on the right show how intermediate output rom one sector becomes a parto another sectors nal output

22 23


13/37

Transitions: moving from rhetoric to priority in 21st century PerthArup: Ryan Falconer, Owen Thomas, Kym Hockley, Danya Alexander

AbstractThis research examined the increasingimportance of public transport inWestern Australian transport policy and planning. It also identi ed current philosophies in relation to publictransport service provision. Theseinclude the intent for public transport tooperate in synergy with smart growthinitiatives and as a realistic competitorto private motor vehicle travel alongmajor corridors.

This information underpinned anexploration of international leading

practice to assist with the developmentof A Practitioners Guide to Bus

Movement and Priority (2011) on behalf of the Public Transport Authority(PTA) of Western Australia andDepartment of Transport (DoT).

The Guide is intended to help drivefuture investment decisions inconjunction with the implementation ofthe Activity Centres Policy (Directions2031) and a Moving People NetworkPlan, which is due for release in 2012.The Guide was developed

collaboratively by Arup, theDepartment of Transport and PublicTransport Authority as an update to lessuser friendly guides developed in theearly 2000s. This article also describesthe policy context for the Guide, how priority projects are being planned andimplemented. It outlines some of thechallenges that arise when trying todeliver the best value public transportservices using limited resources.

IntroductionThe last two decades have seen a partial shiftin transport policy away from predict and provide (e.g. forecast traf c growth and provide road infrastructure to cater to it)towards travel demand management. As partof this agenda, the State Government ofWestern Australia has made signi cantinvestments in public transport. Some of thisinvestment has been in major projectsde ning higher frequency and qualityservices on key routes to increase thecompetitiveness of public transportcompared to the private vehicle. However, asizeable proportion of added investment hasalso been required because of growth in theurban footprint of Perth and subsequentdemand for new bus services to green eldsdevelopments.

This increased investment has been paralleled by a surge in public transport patronage over the last 10-15 years. From1997-2007, there were 36% and 42.4%increases in rail and bus passenger journeys,respectively. Over the same period, there wasa 7.9% increase in passenger kilometres byrail and 44.6% by bus. These statisticscompared to growth in public transport patronage in other Australian cities, indicateda 2.7%, 32.6% and 40.9% rise in passenger journeys over the same period in Sydney,Melbourne and Brisbane. Based on 2006Census data, about 10.4% of work trips inPerth are undertaken using pub

Documents

Research Review 010312lr