UNIQUE SMOKE...that smoke control requirements have been continually changing. With all of the effort in code writing and research, it is fair to ask, “Has the effort been worth

FIRE PROTECTIONFIRE PROTECTION

PROVIDING PRACTICE-ORIENTED INFORMATION TO FPEs AND ALL IED PROFESSIONALS

Summer 2000 Issue No.7

SMOKE MANAGE-MENT SYSTEMS –DO THEY WORK?

USING MODELS TOSUPPORT SMOKEMANAGEMENTSYSTEM DESIGN

AN OVERVIEW OFATRIUM SMOKEMANAGEMENT

ALSO:

4

17

24

SFPE

Cele

brat

ingour 50th Anniversary

MANAGEMENT SMOKE

UNIQUE

DESIGNSMANAGEMENT SMOKE

UNIQUE

DESIGNS

Fire Protection Engineering 1

3 VIEWPOINTSmoke control methods differ according to the structure, but their effectiveness depends on one common factor.April Berkol

4 SMOKE MANAGEMENT SYSTEMS – DO THEY WORK?More and more sophisticated smoke control management requirementshave developed as codes have changed, but their in-service performancemay depend on factors that have not yet received sufficient attention.William A. Webb, P.E.

17 USING MODELS TO SUPPORT SMOKE MANAGEMENTSYSTEM DESIGNComputer-based and physical models can be used as an aid in testingsmoke management systems.James A. Milke, Ph.D., P.E.

24 AN OVERVIEW OF ATRIUM SMOKE MANAGEMENTThe zone fire model concept and other aspects of atrium smoke management, including new information proposed for addtion to NFPA 92B, are reviewed.John H. Klote, Ph.D., P.E.

36 MODELING SPOT-TYPE SMOKE DETECTOR RESPONSECurrently available methods and their merits for everyday use.William E. Pucci, P.E.

44 CAREER CENTER

46 SFPE RESOURCES

52 FROM THE TECHNICAL DIRECTORRecent legislative changes in Florida prompt an examination of the engineer’s role in sprinker design and layout.Morgan J. Hurley, P.E.

contents S U M M E R 2 0 0 0

9COVER STORYUNIQUE SMOKE MANAGEMENT DESIGNSThe unusual architectural designs of the Luxor Hotel and Casino and the MGM Mansionpresent the need for mechanical smoke management sytems that are just as unique.Douglas Evans, P.E.

Subscription and address change correspondence should be sent to: Society of Fire Protection Engineers,Suite 1225 West, 7315 Wisconsin Avenue, Bethesda, MD 20814 USA. Tel: 301.718.2910. Fax: 301.718.2242.E-mail: [email protected].

Copyright © 2000, Society of Fire Protection Engineers. All rights reserved.

FIRE PROTECTIONFIRE PROTECTION

SFPE

Cele

brat

ingour 50th Anniversary

Fire Protection Engineering (ISSN 1524-900X) is published quarterly by the Society of Fire ProtectionEngineers (SFPE). The mission of Fire ProtectionEngineering is to advance the practice of fire protectionengineering and to raise its visibility by providing information to fire protection engineers and allied professionals. The opinions and positions stated arethe authors’ and do not necessarily reflect those of SFPE.

Editorial Advisory BoardCarl F. Baldassarra, P.E., Schirmer EngineeringCorporation

Don Bathurst, P.E.

Russell P. Fleming, P.E., National Fire SprinklerAssociation

Douglas P. Forsman, Oklahoma State University

Morgan J. Hurley, P.E., Society of Fire ProtectionEngineers

William E. Koffel, P.E., Koffel Associates

Jane I. Lataille, P.E., HSB Industrial Risk Insurers

Margaret Law, M.B.E., Arup Fire

Ronald K. Mengel, Honeywell, Inc.

Warren G. Stocker, Jr., Safeway Inc.

Beth Tubbs, P.E., International Conference of BuildingOfficials

Regional Editors

U.S. HEARTLAND

John W. McCormick, P.E., Code Consultants, Inc.

U.S. MID-ATLANTIC

Robert F. Gagnon, P.E., Gagnon Engineering, Inc.

U.S. NEW ENGLAND

Robert G. Sawyer, III, University of New Haven

U.S. SOUTHEAST

Jeffrey Harrington, P.E., The Harrington Group, Inc.

U.S. WEST COAST

Scott Todd, Gage-Babcock & Associates, Inc.

ASIA

Peter Bressington, P.Eng., Arup Fire

AUSTRALIA

Richard Custer, Custer Powell, Inc.

CANADA

J. Kenneth Richardson, P.Eng., Ken Richardson FireTechnologies, Inc.

NEW ZEALAND

Carol Caldwell, P.E., Caldwell Consulting

UNITED KINGDOM

Dr. Louise Jackman, Loss Prevention Council

Publishing Advisory BoardBruce Larcomb, P.E., BOCA International

Douglas J. Rollman, Gage-Babcock & Associates, Inc.

George E. Toth, Rolf Jensen & Associates

PersonnelPUBLISHER

Kathleen H. Almand, P.E., Executive Director, SFPE

TECHNICAL EDITOR

Morgan J. Hurley, P.E., Technical Director, SFPE

MANAGING EDITOR

Joe Pulizzi, Penton Custom Media

ART DIRECTOR

Pat Lang, Penton Custom Media

COVER DESIGN

Dave Bosak, Penton Custom Media

viewpoint

SUMMER 2000 Fire Protection Engineering 3

To the uninitiated, the idea ofsomehow directing the smokegenerated in a fire seems like a

wonderful solution to the problem: bysomehow keeping smoke away fromthe means of egress, the exiting occu-pants are protected; by exhausting andcontrolling smoke from fire, occupantswho stay in a building during a fire arenot exposed to the toxic products ofcombustion. According to Klote andFothergill in the document entitled“Design of Smoke Control Systems forBuildings” (1983 from American Societyof Heating, Refrigerating and Air-Conditioning Engineers), some form ofairflow management has been done formore than 40 years.

The simplest method for controllingsmoke is found in stair pressurizationsystems. Simply put, such systems workby forcing air into the stair enclosure,thereby creating a positive pressureinside the enclosure. When the doorsto the enclosure are opened by exitingoccupants, the smoke is kept out as theair from the stairs “pushes out” andkeeps smoke from entering the stair.Injecting air via fans into a stairway orelevator shaft can be done from thetop, from the bottom, or from multiplesources depending on the height of thebuilding. The assumption is that, bydoing so, the corridors leading to thestair and the remainder of the buildingwill have a negative pressure in relationto the stair, so instead of going into thestair, the smoke is “pushed away” by arush of air escaping the pressurizedstair enclosure. Of course, for all of thisto work as desired, the systems musthave been designed to compensate forone or more doors being propped openor kept open by the occupants stream-ing into the stairway, have been testedand inspected on a regular basis, andnot be dependent on an unprotectedpower source.

Smoke control for the building prop-er can involve a combination of differ-ent methods depending on what isactually desired. The simplest approach

is to “sandwich” the fire floor by creat-ing positive pressure on the floorsabove and below the fire floor.Most often this is done by usingthe existing heating, ventilating,and air conditioning system com-ponents (HVAC). Fansare activated, dampersare opened andclosed, and essentially more air isforced into the floors above and belowwhile air is exhausted from the firefloor. In some buildings, a dedicatedsmoke management system may berequired. Such systems are there exclu-sively for managing smoke from fire.Automatic activation of the smoke man-agement system components is not themost common method; it is more com-mon for the responding firefighters totake control of the systems and activatethem manually as they see fit.

Smoke control for malls, atria, andlarge areas requires the use of a combi-nation of different smoke control meth-ods. Atria are multistory, large, openspaces which communicate with multi-ple floors of the buildings they arefound in. Malls are large, open spacescreated when individual shops on mul-tiple levels are under one commonroof. Where malls tend to spread outhorizontally, atria tend to be more verti-cal in orientation. Large-volume areasunder a single roof which have notbeen divided up into separate spacespresent a similar challenge to smokecontrol. Examples of such spaces areopen manufacturing spaces, large ware-houses, arenas, and the like.

For such spaces, a combination ofsmoke exhaust and a method for limit-ing the spread of smoke depending onthe type of occupancy is common. Theeffect desired is selected based on theuse of the building. In malls, arenas,convention halls, ballrooms, exhibitionhalls, concert halls, etc., and atria inhotels and office buildings, the desire isto protect the occupants and afford asafe means of egress for them. Whereoccupants are limited in number, the

objective might beto prevent smoke damage tosensitive manufacturing equip-ment.

Nowadays, the design ofsmoke control systems for most struc-tures frequently involves the use ofcomputer modeling of different scenar-ios in order to come up with the mostdesirable method or combination ofmethods. Computer models can fairlyrealistically simulate different types offires in different parts of a structure.Using these programs, the design engi-neer can simulate the effects of variousmethods of smoke control, singly andin combination, to come up with thebest method to manage smoke in thespace.

Ultimately, the success or failure ofthe installed systems depends on thebuilding owners and how they test andmaintain these systems. The compo-nents of smoke control systems aresimple: fans, ducts, vents, dampers, etc.Keeping those devices in top workingorder is something else. Once installed,such systems are often not tested regu-larly. Changes to the structure will alsocreate changes in how smoke moveswithin them. Minor changes don’t usu-ally trigger a review of the smoke con-trol system and whether there is a needto make changes to it as well. Overtime, the building interior may changesubstantially, while the fixed compo-nents of the smoke control system areeither compromised or rendered use-less. Perhaps if building maintenancepeople and building owners were moreknowledgeable of how these systemsare meant to work, they would bemore conscientious about testing, main-taining, and including them in planswhen changes are made to buildings.

Personally, I am skeptical about theeffectiveness of the more complicatedsystems over the life of a building.

April Berkol is with Starwood Hotelsand Resorts worldwide.

SMOKE CONTROL

4 Fire Protection Engineering NUMBER 7

By William A. Webb, P.E.

BACKGROUND

The development of modern smoke management systemsbegan in the 1960s. After a series of high-rise fires, there was anincreased interest in smoke management systems, which led toresearch on the time to evacuate high-rise buildings and smokespread caused by stack effect. A few notable fires that occurredin buildings containing atria caused a reexamination of the pro-visions provided for smoke management in atria and in coveredmalls. All of the model codes have had changes to the require-ment for smoke control or smoke management in each editionsince at least 1970. The International Building Code (IBC)1 hascontinued the trend.

A significant body of research influenced the code changes. Thecombined effect of the code changes and the research has beenthat smoke control requirements have been continually changing.

With all of the effort in code writing and research, it is fair toask, “Has the effort been worth the cost; and do these systemswork?”

This article will present information concerning how wellthese systems perform after they have been in service and toencourage discussion and research to determine their effective-ness and to improve their maintenance.

BRIEF HISTORY

In 1968, the National Research Council of Canada (NRCC) conducted a survey of exit facilities in 20- to 40-story office buildings.2, 3, 4 The results revealed dangerously excessive exit

DO THEY

WORK?

SMOKE MANAGEMENT SYSTEMS

©2000SFPE. All rights reserved.


times. A little earlier, NRCC discovereda serious smoke control problem in tallbuildings, partly due to stack effect andpartly due to deficiencies in HVAC sys-tems and other elements of construction.

On January 24, 1969, Chicago news-papers reported “Four Die in ChicagoApartment Fire.” On Thursday, August 6,1970, the headlines read, “SkyscraperFire Kills Two” with a subheading thatread, “31 Injured in Blaze on WallStreet.” On December 4, 1970, a front-page story told of a fire on the fifthfloor of a 47-story building in NewYork City that killed three workmen.The most disturbing part of these firesis that they all happened in modernfire-resistive, high-rise office buildings;all involved loss of life, poor elevatoroperation, substantial smoke spreadthrough the building, and problemswith exiting.

The fire protection industry reactedto these events with a series of confer-ences. Two notable ones were spon-sored by the Fire Protection EngineeringDepartment of the Illinois Institute ofTechnology in 1970 and by the U.S.General Services Administration in1971.5

While code changes were beingdeveloped using the conclusions fromthe conferences, there were three high-rise fires of national prominence inNovember 1972. The first was a fire onthe 95th floor of Chicago’s John HancockCenter. The photographs were as spec-tacular as the fire loss, though therewere no fatalities. The second and thirdcame back to back in a New Orleanshigh-rise with a restaurant on the 19th

floor and in an Atlanta retirement homewhere the fire started on the 12th floor.

The fires causing concern inChicago were in apartment buildings;those in New York were in officebuildings. Although the structures inthe two cities had much in common,they differed to a major extent in com-partmentation. This drove differentsolutions in the requirement initiallyadopted in each city for high-rise firesafety. New York’s requirements,known as Local Law 5,6 applied tooffice buildings. Either compartmenta-tion and stair pressurization or auto-matic sprinkler protection was required,among other features such as voiceevacuation, firefighters’ communica-tion, and elevator recall.

Chicago’s requirements, adopted in1975, applied to all high-rise buildings.They required either compartmentationand smoke control or sprinkler protec-tion. The smoke control system was tobe designed to pressurize nonfire com-partments on the fire floor and exhaustthe fire compartment. A smokeprooftower was required for nonsprinkleredbuildings. The voice evacuation andfirefighters’ communication require-ments were similar to those in LocalLaw 5.

Fire safety requirements for high-risebuildings were subsequently adopted inthe model codes and those of the majorcities in North America. Among featuresintended to prevent smoke spreadbetween floors or groups of floorsincluded stair pressurization, HVAC sys-tem shutdown, fire-floor venting orexhausting, and automatic sprinklerprotection. The specific features,required or allowed, varied among thecodes. As the popularity of atriumhotels and covered malls increased, sodid the concern for smoke spread inthese occupancies. This caused addi-tional smoke management considera-tions to be adopted in the codes.

Each year, the model codes haverevised smoke management require-ments. As a result, more and moresophisticated methods of smoke con-trol and smoke management havebeen needed to comply with the code.The terms I use in this article for“smoke control system” and “smokemanagement system” are those definedon page 24.

EXPERIENCE

Although there have been relativelyfew high-rise fires, those that haveoccurred are often spectacular. Thesame is true of atrium fires. Even inthose cases where protection wasinadequate, a common thread sepa-rates the “failures” from the “successes.”That common thread is inspection,testing, maintenance, and enforcement.

Let us consider a few examples. The January 1982, NFPA Fire

Journal reported on a fire at the MGMGrand. This incident clearly was aninstance of inadequate protection andnumerous deficiencies. Matters such assteel straps bolted across dampers,

HVAC mixing rooms used for storageor as offices, fusible links replaced bysteel wire, inadequate enclosure ofexit passageways, and blocking ofsmokeproof tower smoke vents couldhave been discovered by inspectionand prevented by proper maintenance.While this tragedy may not have beenaverted had the deficiencies been cor-rected, the deficiencies may have beencontributing factors.

In another incident in a high-risebuilding in the Southwest, a fireoccurred in the service elevator lobbyon the 22nd floor of a hotel. The smokedetector in the lobby detected the fireand released the lobby doors.Recently installed carpeting, however,kept the doors from closing. In addi-tion, stair doors had been blockedopen. As a result, smoke infiltratedstairs and the elevator shaft, and fivefloors of the hotel had to be evacuated.

In this instance, the active portion ofthe smoke control system, i.e., thedetection and fans, worked properly.The passive portion, namely the doors,failed because they were blockedopen. This is another instance inwhich inspection and maintenancewould have disclosed the deficiencies.

In another instance in the same city,the HVAC system in a portion of thebuilding continued to operate, circulat-ing smoke from a fire. The duct detec-tors had been disabled during mainte-nance. Inspection by those chargedwith fire safety could have discoveredthe deactivation of the smoke detec-tors so that alternative protectioncould have been provided during theshutdown period.

A Fire Journal10 article reported on anatrium fire which occurred on Novem-ber 19, 1973, at the Hyatt RegencyO’Hare. The article states “ . . . it wasfound that the atrium smoke exhaustsystem had failed to operate. On check-ing, it was found that the switch con-necting the smoke detection system tothe exhaust system had been turned off.The fans were then turned on, and theatrium was cleared of smoke.”

An arson fire in the Blue MaxNightclub of the hotel spread fire andflames into the hotel atrium. The night-club was on the second floor of the10-story high atrium. The atrium was145 feet (44 m) square, topped by arevolving restaurant. The article states,


“Although 1,000 guests were exposedto the fire conditions, only onerequired hospital treatment, because ofa heart condition. One firefighter wastreated for smoke inhalation.” It shouldalso be noted that the nightclub andguestrooms were not protected bysprinklers. Firefighters’ prompt responseand action prevented a tragedy. It isclear that inspection and maintenancewould have disclosed the deficiency sothat it could have been corrected.

The January 1979 edition of FireJournal reported that a fire whichstarted in a small office on the 10th

floor of a 13-story office buildingbroke a window which opened to theatrium and allowed smoke to enter theatrium.11 Upon firefighter arrival, “ . . .thick black smoke was pouring fromthe fire floor and banked down fromthe roof to below the 10th floor.Although the smoke detector operated,only two of the six smoke ventsopened. The other four released; how-ever, maintenance personnel reportedthat the springs had apparently lostsufficient strength to open them fully.”The article goes on to state that main-tenance personnel did open the vents,but smoke continued to bank down.The failure of all of the vents to oper-ate does, however, clearly demonstratethe need for inspection and propermaintenance. In this case, maintenancewould have replaced the springs orthe vents.

CODES

The changes that have occurred tothe model building codes since the1970s included an exception to allowstair pressurization without a vestibulein buildings protected by automaticsprinklers in lieu of smokeproof tow-ers having vestibules with natural ormechanical ventilation. For malls andatria, the requirements have migratedfrom four or six air changes per hour(ACH) of the interconnected volume tofire plume calculations based on NFPA92B. In each of the revisions, most ofthe information has concerned installa-tion for new buildings and for accep-tance testing. Until recently, few, ifany, requirements or guidance for rou-tine inspections and maintenance havebeen adopted in building or firecodes. It is, therefore, no wonder that

regulatory authorities frequently statethat the systems function properlywhen new, but may perform poorlyafter a few months or years in service.

Each of the current editions of themodel building codes and the recentlyadopted International Building Codestate that systems required by the codeare to be maintained in accordance withthe code. There are no specific mainte-nance requirements for smoke controlsystems. This can be expected because,generally, maintenance requirements arecontained in fire codes.

The three model fire codes requiresmoke control and smoke managementsystems to be inspected and operatedor tested. The frequency is quarterlyfor the Uniform Fire Code12 and semi-annually by the BOCA National FirePrevention Code13 and the StandardFire Code14. The International FireCode 15 has the most comprehensiverequirements for maintaining smokecontrol and smoke management sys-tems. It states that required systemsshall be maintained in accordance withthe manufacturers’ instructions and thecode. It requires a written schedule forroutine maintenance and operationaltesting be established. Dedicated sys-tems are to be operated semiannuallyand nondedicated systems operatedannually.

In addition to the model codes, alogical place to look for the frequency

of testing and maintenance of smokecontrol systems would be NFPA docu-ments. Turning first to NFPA 1,16 thereare no specific requirements for testingand maintenance of smoke control sys-tems. There is a general requirementthat required fire safety systems are tobe maintained. In this respect, NFPA 1is similar to the model fire codes.

NFPA 92A suggests that dedicatedsystems should be operated semi-annually and nondedicated systemsshould be operated annually. It states,“Dedicated smoke-control systems areintended for the purpose of smokecontrol only. They are separate sys-tems of air-moving and distributionequipment that do not function undernormal building operating conditions.Upon activation, these systems operatespecifically to perform the smoke-con-trol function. Nondedicated systemsare those that share components withsome other system(s) such as thebuilding HVAC system. Activationcauses the system to change its modeof operation to achieve the smoke-control objectives.” In each case, thesystems should be operated understandby power. These tests are to bedocumented and a log made availablefor inspection. The purpose of the testis to determine that the correct outputis attained for each input.

NFPA 92B suggests semiannual testswith the results documented in a logavailable for inspection. As with 92A,the tests are to determine whethercorrect outputs occur for each input.The purpose is to demonstrate thatthe installed systems will continue tooperate in accordance with theapproved design. It is recommendedthat the tests include both measure-ments of airflow quantities and pres-sure differentials.

NFPA 90A17 recommends thatdampers be examined every two years.It also recommends that fans andmotors should be inspected at leastquarterly and that fan controls be exam-ined and activated at least annually.

NFPA 10118 contains a similar require-ment to those of the model codes tomaintain required systems. It alsorequires that mechanical stair ventilationsystems have their operating parts test-ed semiannually and the results logged.The recently adopted performanceoption chapter of NFPA 101 requires


that systems necessary to achieve thedesign performance be maintained forthe life of the building. No specificsmoke control or smoke managementrequirements are mentioned.

ASHRAE Guideline 519 provides meth-ods for verifying and documenting thatthe performance of smoke managementsystems conforms to the design intent. Itis critical that the designer and owneruse a system such as Guideline 5 if theyexpect the smoke control or smokemanagement system to perform asintended over the life of the building.The Guideline states, “ . . . throughoutthe useful life of the building, there willbe a need to recommission these sys-tems periodically.” A key component torecommissioning is the post-acceptancephase. It requires “as-built” or recorddocuments be reviewed so that theyreflect modifications made to the systemduring construction and throughout thelife of the building. The Guideline spec-ifies how documents are to be main-tained so that they reflect current sys-tem performance. The commissioningdocuments contain records of all theoriginal tests. These are useful in main-tenance tests to see how the systemperformance has changed and to deter-mine what maintenance or replacementis needed to restore the system to itsoriginal state.

It is unnecessary for codes to speci-fy the details of how inspection, test-ing, and maintenance are to be per-formed; however, they should specifywhat and when these tasks are to beperformed.

CONCLUSION

Returning to the question posed atthe beginning of this article, there isinsufficient information on fire experi-ence and cost impact to judge whetherthese systems are worth the cost. It isclear that the systems generally dowork when initially installed. There isanecdotal evidence to question theirin-service performance.

As we embark on performance-baseddesign in the Life Safety Code and theICC Codes, emphasis continues to beplaced on evaluating the initial design.The International Fire Code has, how-ever, taken steps to determine that thedesign continues to meet the objectiveover the life of the building. Perhaps it

is to be expected that it has taken solong to address maintenance, given thelack of attention that testing and main-tenance have been given in the currentcodes. If the performance of smokecontrol and smoke management sys-tems is questionable for those installedunder prescriptive codes, what can weexpect when they are an integral partof a fire protection solution using a per-formance code? Perhaps the reason wehave not recorded more incidents ofcatastrophic failures of these systems isthat the fire suppression systems andfire detection and alarm systems onwhich the smoke management systemsdepend are inspected and maintainedmore thoroughly. It seems obvious thata performance solution must include adetailed maintenance and testing proto-col that regulatory officials must be pre-pared to enforce and owners and oper-ators prepared to implement. If they areunwilling to expend the effort toinspect, test, and maintain the systemsand to enforce the testing and inspec-tion protocol, why should the systemsbe required at all? The cost might betterbe used to enhance the reliability of thesystems, which they are prepared toinspect, test, and maintain.

This article is intended to be a call tocollect information on how well exist-ing smoke management systemsalready installed are being inspected,tested, and maintained. It is encourag-ing that the International Fire Code hasincluded requirements for inspectionand testing of in-service systems. Themaintenance requirements should beinterpreted to be similar to those con-tained in ASHRAE Guideline 5 whichstates that a maintenance programshould include developing and main-taining a standard method of recordingmaintenance tests and their results. Itmakes little difference to fire safety todevelop good requirements based onsound engineering if the systems arenot tested and the requirements are notenforced. Now that we have goodsmoke management requirements inthe codes for installation, let us be surethat the systems are inspected andmaintained properly.

William Webb is with PerformanceTechnology Consulting, LTD.

REFERENCES1 International Building Code,

International Code Council, Inc., FallsChurch, VA, 2000.

2 Galbreath, M. “Fire in High Buildings,”Fire Study No. 21., National ResearchCouncil of Canada, Ottawa, ON, 1968.

3 McGuire, J.H. “Smoke Movement inBuilding Fires.” Fire Technology 3 (3):1968, pp. 163-174.

4 McGuire, J.H. “Control of Smoke inBuilding Fires.” Fire Technology 3 (4):1967, pp. 281-290.

5 Jensen, R.H. “High-Rise Fire Protection.Where Do We Stand? Where Do We Go?”Proceedings, Chicago Committee onHigh-Rise Buildings, September ReportNo. 2: 1972, pp. 5-17.

6 New York City, Local Law No. 5-1973.

7 NFPA 92A, Recommended Practice forSmoke Control Systems. National FireProtection Association. Quincy, MA, 1999.

8 NFPA 92B, Guide for SmokeManagement Systems in Malls, Atria,and Large Areas. National Fire ProtectionAssociation. Quincy, MA, 1995.

9 Fire at the MGM Grand. Fire Journal 76(1): 1982, pp. 19-37.

10 Sharry, J.A. 1973. An Atrium Fire. FireJournal 67 (6): pp. 34-41.

11 Lathrop, J.K. Atrium Fire Proves Difficultto Ventilate. Fire Journal 73 (1): 1979,pp. 30-31.

12 Uniform Fire Code. International FireCode Institute. Whittier, CA, 1997.

13 The BOCA National Fire PreventionCode. Building Officials and CodeAdministrators International, Inc.Country Club Hills, IL, 1999.

14 Standard Fire Prevention Code. StandardBuilding Code Congress International,Inc. Birmingham, AL, 1997.

15 International Fire Code. InternationalCode Council. Falls Church, VA, 2000.

16 NFPA 1. Fire Prevention Code. NationalFire Protection Association. Quincy, MA,1997.

17 NFPA 90A, Standard for the Installationof Air Conditioning and VentilatingSystems. National Fire ProtectionAssociation. Quincy, MA, 1999.

18 NFPA 101, Life Safety Code. National FireProtection Association. Quincy, MA, 1999.

19 ASHRAE Guideline 5, CommissioningSmoke Management Systems. AmericanSociety of Heating, Refrigerating and Air-Conditioning Engineers, Inc. Atlanta, GA,1994.


By Douglas Evans, P.E.

INTRODUCTION

Several of the most unique buildings in the world arelocated in Las Vegas, Nevada. Because of the unusualarchitectural designs incorporated into these buildings, themechanical smoke management systems must also be justas unusual. This article provides an overview of twounique smoke management designs and demonstrates thatstructures need not be completely unique to warrant dif-ferent ways of thinking about limiting smoke movement.

THE LUXOR PYRAMID

The 30-story Luxor Hotel and Casino is certainly one ofthe most unique structures in the world. Its principal fea-ture is its pyramidal shape. The interior contains an atriumexceeding 595,000 m3 (21,000,000 ft3) in volume. Interiordimensions are approximately 150 m by 150 m (500 ft. by500 ft.) at the base and 37 m by 37 m (120 ft. by 120 ft.)at the uppermost level, which is 61 m (200 ft.) up.

The casino level is actually the ground floor and islocated directly below the lowest level of the atrium(Attractions Level). The Attractions Level contains severalinterior structures, including restaurants and three theaters,that are occupiable. Several additional structures are strictlyfacades and essentially unoccupiable.

Balconies, which are open to the atrium on 27 floors,provide access to more than 2,500 guest rooms. A room atthe apex of the pyramid contains mechanical equipment.

SMOKE MANAGEMENT APPROACH

At the time the facility was being designed, the 1988Uniform Building Code required a minimum mechanicalexhaust capacity of four air changes per hour (ACH). In

Attractions Level. Notice slots in “obelisk” for beamdetectors.

UNIQUEUNIQUE SMOKE MANAGEMENTDESIGNS



atria, the code also required one-halfof the air being exhausted to bemechanically injected upward at thebase of the atrium. This translated toan exhaust capacity of approximately635 m3/s (1,350,000 cfm) and 320 m3/s(675,000 cfm) of supply.

These prescriptive requirements created several problem areas. Themechanical room at the top of the pyr-amid would have to be designed toaccommodate this number and weightof fans. A significant wall area of exte-rior grills would be necessary. Sincethe atrium narrows as it approachesthe top, the upper several levels ofexit balconies could have excessive airvelocity upon activation of the smokemanagement system. Stratification ofsmoke could limit the ability toexhaust smoke and cause intermediatelevels to become untenable.

One of the main concerns wassmoke obscuring exit balconies whileit was being drawn upward. Smoke isexpected to rise until it contacts oneof the exit balconies. That balcony andbalconies above can be expected tobecome untenable.

It appeared that strict code compli-ance might not solve these concernsand could actually increase the hazardto occupants on upper levels. It wasagreed that a performance-basedapproach may be better able toachieve the desired goals than the prescriptive requirements. The basicfire protection goals agreed to were:

• Maintain a tenable environmentfor occupants not intimate withthe fire.

• Limit temperature and smoke generated by a fire to maintainexit balconies tenable for evacuation purposes.

• Reduce the impact of stratificationof smoke at an intermediate level.

The engineer working with thedesign team proposed a radical depar-ture from the prescriptive code. Hesuggested a series of fans and ductssupplying air to the lowest level opento the atrium, oriented such that theentire volume of air would rotate (asviewed from above). This rotation wasto work in conjunction with exhaustfans located in the mechanical room atthe top of the pyramid. He theorizedthat this counterclockwise rotation

would cause smoke to be drawn intothe atrium void, away from exit bal-conies, while it was being exhaustedout the apex of the atrium.

The Clark County review team (con-sisting of Clark County Building andFire Departments and third-party peerreviewers) was skeptical that this pro-posal would achieve the desired goalsbut agreed that the concept would beconsidered if adequate documentationcould be submitted to substantiate theproposed design.

The architectural design of theAttractions Level also needed to betaken into consideration. Due to theheight of the space, the AttractionsLevel is essentially unprotected byautomatic sprinklers. Agreements werereached that significantly restricted thecombustible load in nonsprinkleredportions of the Attractions Level. It wasagreed that the fire size expected within the atrium would not exceed2110 kW (2000 BTU/s) maximum heatrelease rate.

For estimating smoke quantity, car-bon monoxide levels, direction of airmovement, and temperature, the

design team used several recognizedreferences.1, 2, 3, 4, 5, 6

The FloVENT computer model1 indi-cated that this counterclockwise rota-tion in conjunction with the mechani-cal exhaust, would draw the air masstoward the center of the atrium andup. Alternate calculations were per-formed to provide a comfort level thatsmoke would be diluted to tenableconcentrations, as well as reduce heatbuildup on exit balconies. Calculationsindicated that temperature and carbonmonoxide levels would be within ten-able limits 23 m (75 ft.) above the noz-zle of the vortex supply fans.

The Clark County review team thor-oughly analyzed the proposed design.Several meetings and iterations of thedesign were necessary to achieve con-currence, which allowed this uniquesmoke control design to be condition-ally approved. Final acceptance wasbased on the system’s performanceduring commissioning.

Eight supply fans were installed out-side the building at the base of thepyramid. Their associated ductworkwas routed to the interior perimeter

Attractions Level. East interior view taken shortly after Luxor’s grand opening inthe mid-’90s.


portions at the base of the atrium.Each of these fans is capable of pro-viding up to 14 m3/s (30,000 cfm) at16.0 m/s (3,150 ft. per minute). Thedischarge was oriented approximately22 degrees from the horizontal and 11degrees from guestroom balconies. Atotal atrium exhaust rate of 188 m3/s(400,000 cfm) was provided at theapex of the atrium. This combinationof exhaust and injection ports createsa high-velocity air stream parallel toguest room balconies, which causes alow-pressure area that draws smokeinto the atrium to be exhausted outthrough the apex. This design actuallyprovides less than one air change perhour.

The atrium smoke management sys-tem activates whenever any of the fol-lowing events occur: automatic sprin-kler water flow anywhere in the atri-um (including any exit balcony); oper-ation of any two of the more than2,000 area smoke detectors installedon the exit balconies; or activation ofany one of 24 beam smoke detectorslocated in a structure in the center ofthe Attractions Level. Manual overridesare also provided in a protected roomspecifically designated for fire depart-ment emergency response (FCC).

ACCEPTANCE TESTING/COMMISSIONING

After the contractors and designershave confirmed fire protection systemsfunction as intended and prior to grant-ing occupancy for any major facility,Clark County Building and FireDepartments witness “all systems” tests.These series of tests are intended tosimulate reasonable fire scenarios. Asignificant portion of the facility ismethodically stepped through to con-firm proper functioning, as well ascoordination of all fire protection sys-tems.

During testing of the atrium smoke-control system, a 2,110 kW (2,000BTU/s), 3 m (10 ft.) diameter, propaneburner was moved onto the AttractionsLevel. Theatrical smoke was injectedinto the heat plume generated by thepropane burner to visually verify airmovement. Through visual verification,as well as review of the computer out-put from carbon dioxide monitors andthermocouples placed at twelve loca-tions on the exit balconies, it wasdetermined that the atrium systemfunctioned as indicated by the models.Proper configuring of all dampers, fans,and other operating equipment was

also confirmed. Status and manualoverrides were confirmed from the FCC.

To simulate a guestroom fire withthe door blocked open, the engineerof record injected theatrical smokeonto one intermediate-level balcony.While the air mass was rotating, the“floor of origin” was relatively clear.The theatrical smoke did not adverselyimpact alternate floors. Smoke wasdrawn into the atrium and exhaustedout through the apex. Without the airmass rotating, the theatrical smoke notonly impacted the floor of origin, butalso the balconies above and below.

During an alternate test, theatricalsmoke was injected onto one of thecorners of an exit balcony where theexit stairs and elevator lobbies create apartial enclosure. Initially, the smokewas so thick that visibility was reducedto approximately 1 m (3 ft.). Within acouple of minutes, the rotational airmass drew smoke from the enclosedportion of the balcony into the atriumand toward the exhaust fans to theextent that visibility was increased significantly.

THE MGM MANSION

The MGM Hotel and Casino boastsmore guestrooms than any other hotelin the world (more than 5,000). A fewyears after opening, the owners decid-ed to add a four-story atrium with adomed skylight above a garden-likecentral court. This portion of the facili-ty was to contain 25 guest suites forthe world’s elite, each ranging in sizefrom 740 m2 (8,000 ft3) to 1100 m2

(12,000 ft3). Each suite was to be a sin-gle level with windows and privatebalconies opening onto the centralcourt.

To make these “villas” as comfort-able as possible, the designers decidedon operable windows and doors fromthe suites into the atrium on all fourlevels. Furthermore, these doors andwindows were proposed to be neitherfire-rated nor self- or automatic-clos-ing. The prescriptive codes adopted atthat time were the 1994 UniformCodes, which specifically required theinterface between guestrooms and theatrium to act as a smoke barrier and,therefore, did not allow this proposedarrangement.

First Floor. Main entrance into the casino complex.


SMOKE MANAGEMENTAPPROACH

To justify this arrangement, the fireprotection engineer of record pro-posed mitigating measures to providethe level of protection intended by theprescriptive requirements. Since allexits above the atrium floor were inde-pendent of the atrium, it was pro-posed to use the atrium void as asmoke reservoir. Approximately one-half of the mechanically supplied airwould be injected into the guest suiteswith the remainder injected at lowvelocity near the atrium floor. Allexhaust fans were located at the top ofthe atrium.

This design concept was dependenton the area of fire origin. The smokemanagement scenario would be differ-ent if the fire were in the atrium or ina guest suite. To compensate for this,the automatic sprinklers protecting thesuites were zoned independently fromthose protecting the atrium space.Beam-type smoke detectors wereinstalled at two levels within the atri-um to help compensate for stratifica-tion. In addition, area-type smokedetectors were installed in the suites atopenings into the atrium, as well aswithin each sleeping room. Using

these initiating devices, the supply aircould be deactivated within the suiteson the floor of origin if it were deter-mined that the fire originated withinone of the suites.

Automatic sprinklers protecting theatrium were installed just below theatrium skylight, approximately 34 m(110 ft.) high. At this height, the auto-matic sprinklers are not expected toadequately control a fire on the floor.Therefore, a thorough analysis was

conducted of reasonable fire scenarioson the atrium floor. Due to the limitedcombustible load, it was determinedthat the maximum expected fire sizewould not exceed the minimum firesize required by the Uniform BuildingCode of 5,275 kW (5,000 BTU/s).Doubling this fire size for a factor ofsafety, the smoke plume dynamicswere estimated, and the exhaust fanswere sized to provide at least 141 m3/s(300,000 cfm). This design is expected

Three-dimensional view of Luxor interior, showing approximate locations ofvortex injection fans.

Luxor exterior, southeast exposure.


to contain smoke from the maximumdesign fire at a reasonable level abovethe atrium floor, but it is not sufficientto maintain smoke below the upperguestroom levels.

The quantity of supply air into thesuites was proposed to be limited to amaximum, based on all the doors andwindows open. To restrict smokemigration from the atrium into thesuites, the minimum velocity of supplyair needed from the suites throughthese openings was estimated to be0.66 m/s (130 ft./min), which was alsoless than the 1 m/s (200 ft./min) limitbelieved to impact plume dynamics. Ifall the doors and windows wereclosed, the pressure in the atriumwould be negative relative to the guestsuites, which would also restrictsmoke migration into an uninvolvedsuite.

This arrangement is expected toprotect occupants whether on the atri-um floor or in an uninvolved suite bymaintaining smoke above the atriumfloor and restricting smoke migrationinto an uninvolved suite. Thisapproach allows occupants to remainin uninvolved suites or safely evacuatethe building.

ACCEPTANCE TESTING/COMMISSIONING

Airflow and pressure differenceswere measured at various locationsthroughout the addition. Initial testinguncovered aspects of the system thatwere not performing as expected.Upon examination, the problem areaswere determined, and correcting mea-sures were implemented. In this case,it was not only necessary to modifythe system, but it was actually neces-sary to revise the design concept inorder to determine if the system actu-ally was able to provide the level ofprotection intended by code.

The system as described above isthe final design that was accepted forthis facility; but this example not onlyillustrates a unique approach for man-aging smoke, it is also a reminder thatwe must not lose sight of our initialgoal. In this case, our principal goalwas to reduce the potential for smokemigration into an uninvolved suite andmaintain the atrium floor safe for evac-uation. If it appears that our firstattempt to meet our goal has failed,maybe we only need to change theway we think about achieving that

goal in order to realize the simplicityof the solution.

Douglas Evans is with the ClarkCounty, Nevada, Department ofBuilding.

REFERENCES

1 FloVENT, Building Services Research andInformation Association and Flomerics,U.K.

2 AHSRAE, Handbook of Fundamentals,1989, Chapter 31, American Society forHeating, Refrigeration and AirConditioning Engineers, Atlanta, GA.

3 Purser, D.A. “Toxicity Assessment ofCombustion Products,” Section 1,Chapter 14, SFPE Handbook of FireProtection Engineering, NFPA, Quincy,MA, May 1988.

4 Klote, J.H., “Smoke Control,” Chapter 3,Section 9, SFPE Handbook of FireProtection Engineering, NFPA, Quincy,MA, May 1988.

5 NFPA 92A, Recommended Practice forSmoke Control Systems, NFPA 1988.

6 NFPA 92B, Guide for SmokeManagement Systems in Malls, Atria,and Large Areas, NFPA, 1991.

Luxor facility model showing new twin towers along with existing pyramid and sphinx. Viewed from northeast side.


An engineering analysis is needed to assess the ability ofa smoke management system

to satisfy stipulated performance crite-ria. This analysis can be conducted toverify acceptability, numerically test, ortroubleshoot problems associated withthe system.

The pressure difference betweenspaces is the focus of analyzing stairpressurization and zoned smoke con-trol systems. For smoke managementsystems in atria, the analysis consistsof assessing the residual hazard posedby smoke in terms of the extent ofsmoke spread, smoke layer depth, orsmoke layer properties. In many cases,a simplified analysis involving theapplication of algebraic equations1, 2 issuitable to assess the performance ofsmoke management systems.

However, in some cases, theassumptions associated with the alge-braic equations are unacceptable. Forstairwell pressurization systems, thealgebraic equations neglect verticalleakage and wind, and require symme-try if more than one pressurized stairis provided. Limitations of algebraicequation methods for atrium smokemanagement are:

• steady or t2 fires only;• uniform horizontal cross-sectional

area for all levels of the atrium;• uniform conditions throughout the

upper layer/zone, even in spaceswith large horizontal areas; and

• analysis of the pre-venting, smoke-filling period or steady, equilibriumconditions during venting.

Algebraic equation methods cannotaddress the interaction between multi-ple smoke management systems, suchas stair pressurization and atriumsmoke exhaust, in the same building.

Several types of models are availableto assist design professionals either inlieu of or as a supplement to the alge-braic equations. These models includesmall-scale physical models, computer-based zone models, network flowmodels, and CFD models. While thisarticle provides an overview of all ofthese models, the emphasis is onsmall-scale models and network mod-els, given the extensive treatment ofzone and CFD models elsewhere.

SMALL-SCALE MODELS

Small-scale models provide physicalrepresentations of a space, though in areduced scale. Scale models are espe-cially useful in examining atria withnumerous projections or irregularshapes. Milke and Klote review theapplication of scale models as a designaid for smoke management systems3.Quintiere and Dillon developed a scalemodel to assess the performance of asmoke management system in a fireincident in a covered mall.4

A small-scale model may bedesigned following the principles of

Froude modeling. Quintiere5 provideda review of scaling relationships topreserve the Froude number. The scal-ing relationships seek to preserve thefollowing ratios:

• fire energy/flow energy;• fan flow/buoyant flow; and• convection heat transfer/wall heat

transfer.

The scaling relationships are:

Temperature: Tm = TF (1)

Position: (2)

Pressure: (3)

Velocity: (4)

Time: (5)

Convective Heat Release:

(6)

Volumetric Flow Rate:

(7)

The subscripts m and F correspond tomodel and full-scale, respectively.

x xl

lm Fm

F

=

∆ ∆p pl

lm Fm

F

=

v vl

lm Fm

F

=

1 2/

t tl

lm Fm

F

=

1 2/

Q Ql

lc m c Fm

F, ,

/

=

5 2

˙ ˙, ,

/

V Vl

lfan m fan Fm

F

=

5 2

USING MODELS TO SUPPORT

Smoke ManagementSystem Design

By James A. Milke, Ph.D., P.E.



Many of the parameters in equations(1) to (7) are functions of time. Thusthe scaled parameters should also befunctions of time. Froude modelinghas the advantage of conductingexperiments in air at atmospheric pres-sure. While Froude modeling does notpreserve the Reynolds number, achiev-ing fully developed flow by makingthe critical dimensions of the model atleast 0.3 m minimizes this shortcom-ing. Fully developed flow only needsto be achieved in those areas wherethe smoke behavior is of interest. Thecritical dimension for a model of ashopping center and atria could be thedistance from the floor to the under-side of a balcony.

In addition, Froude modeling doesnot preserve the dimensionless heattransfer parameters. Often, this limita-tion has little effect because the tem-perature is the same for the scalemodel and the full-scale facility. WhileFroude modeling is inapplicable inhigh-temperature locations, e.g., nearthe flame, Froude modeling still pro-vides useful information about smoketransport away from the fire.

Some surface effects can be pre-served by scaling the thermal proper-ties of the construction materials forthe model. The thermal properties canbe scaled by:

Thermal properties:

(8)

However, selection of enclosurematerials may be acceptably based onflow visualization needs, rather thanscaling of thermal properties, given thesecondary effect of the thermal proper-ties on fluid flow.

EXAMPLE 1

A scale model is proposed to deter-mine the equilibrium smoke layerposition for the atrium depicted inFigure 1. Because the horizontal cross-sectional area varies with height, alge-braic equation and computer-basedzone models are of limited value. Theatrium height is 30.5 m, and the designfire is a 5 MW steady fire. The pro-posed exhaust fan capacity is 142 m3/s.By applying the scaling relationships,the basic parameters for the scalemodel are:

• Height: 3.8 m tall model (1/8 scale)• Fire size: 28 kW• Fan capacity: 0.78 m3/sGiven the 1/8 scale, the width of the

scaled spill plume would need to be1/8 of that at full scale.

COMPUTER-BASED ZONE MODELS

Overviews of numerous zone mod-els are available.6, 7 Quintiere summa-rizes the assumptions of zone models8.The principal advantage of computer-based zone models is their ability toaddress transient effects involvingsmoke spread, delays in fan startup,effects of environmental conditions,and a variety of fire-growth profiles.Some computer-based zone modelsare applicable to spaces where theceiling is sloped or the horizontalcross-sectional area varies with height.In addition, some of the computer-based zone models simulate conditionsin multiple rooms or levels, where thealgebraic equations are limited to asingle compartment.

Limitations of these models resultfrom their assumptions. For example:

• the smoke layer forms immediate-ly, neglecting transport lag;

• the plume is unaffected by windor mechanical ventilation;

• the upper layer/zone is uniform,independent of the area involved;and

• as smoke enters a tall room froma short room, entrainment isdetermined based on a newaxisymmetric plume rather thanfrom a balcony spill or lineplume.

FIELD MODELS

Computational fluid dynamics (CFD)models simulate fluid flow at a level ofdetail impossible with other methodsof computer modeling.9, 10, 11 CFD mod-els divide the fluid flow field intonumerous small cells and numericallysolve the conservation equations ofmass, momentum, and energy for eachcell. Boundary conditions are estab-lished at the room boundaries, open-ings to the outside, and exhaust inletsby specifying velocities.

While generalizations concerning thenumber and size of cells are difficultto make, given the wide range of fea-tures and capabilities of CFD models,generally the smallest cells are nearthe fire and at the ceiling. The govern-ing equations cannot account for tur-

k c k cl

lp w m p w F

m

F

ρ ρ( ) = ( )

, ,

.0 9

Figure 1. Small-Scale Model of Atrium


bulence on a scale smaller than thecells. Further, it is important that thecell size and time step be coordinatedso that cells are not “skipped” fromone time step to the next by the mov-ing fluid.

NETWORK FLOW MODELS

Network flow models such as CONTAM9612 can be applied to evalu-ate pressure differences between com-partments, direction of mass flows,and movement of contaminants. CONTAM96 is often described as thesuccessor to ASCOS, an early, widelyused network flow model13. Networkmodels simulate a building as a net-work of airflow paths comprised ofdoorways, windows, vents, and leaksin building assemblies.

The principle advantage of networkflow models is their ability to consider:

• mechanical and natural ventilation;• environmental conditions (includ-

ing wind);

• interacting smoke managementsystems;

• buildings with complex geome-tries; and

• leakage paths between buildingspaces.

Limitations of network flow modelssuch as CONTAM96 are:

• uniform conditions (temperatureand concentration of contami-nants) are assumed throughouteach “zone”, where a “zone” is atleast one room.

• transient fire conditions (e.g., tem-perature or mass flow in thesmoke plume or temperature inthe fire zone) resulting from agrowing fire are not considered.

However, fire conditions can beincorporated into the model by analo-gy. The volumetric flow in a smokeplume can be simulated as a shaft, witha fan supplying air at each level of theshaft. The air entrained at each levelcan be estimated using Heskestad’splume entrainment equation.14

(9)

where:= volumetric entrainment rate (m3/s)

ρ = density of smoke (kg/m3)= convective portion of heat release

rate (kW) z = clear height (m)

The entrainment for a particularlevel needs to be determined based onthe amount of air entrained only with-in that increment of height. As such,the amount of air entrained for a par-ticular level of the building is the dif-ference in the amount of air entrainedup to the top of the level with thatentrained up to the bottom of thatlevel. Buoyancy effects can be includ-ed by setting the temperature at eachlevel of the “shaft” using Heskestad’splume centerline correlation.14

(10)

where:Tc = plume centerline temperature (°C)

= convective portion of heat releaserate of fire (kW)

z = height above top of fuel (m)

While the centerline temperature ofthe plume overestimates the buoyancyof the overall plume, generally thisapproach is adequate for design pur-poses.

The mass release rate of contami-nant can be estimated as:

(11)

where:= generation rate of contaminant

(kg/s)fc = yield fraction (kg contaminant/kg

fuel) [the yield fraction dependson the fuel, burning mode (flam-ing, smoldering) and the avail-able oxygen concentration]

= heat release rate of fire (kW)= heat of combustion (kJ/kg)

CONTAM96 can conduct a steady orunsteady analysis of the flow of conta-minants (smoke). Input for CONTAM96includes:

• Area and height of spaces• Shaft characteristics • HVAC system

Q

Hotel rooms

Atrium

Figure 2. Typical floor of five-story building.

˙˙

m fQ

Hc cc

=∆

Q

∆Hc

∆T Q zc c= −25 2 3 5 3˙ / /

Q

ρ ˙ . ˙ . ˙/ /V Q z Qc c= +071 0 00131 3 5 3

V

mc


• Fans: constant volume, mass, orcurve

• Environmental conditions: wind,temperature

• Connection of spaces via leakagepaths

• Release rate of contaminant(kg/s): unsteady or steady

Output from CONTAM96 includes:• Pressure difference between zones• Airflow between zones• Contaminant concentration in

zone

An example application of CONTAM96 is provided for a five-storybuilding (see Figure 2). Using CONTAM96, the interaction betweenthe atrium smoke management andstairwell pressurization systems isinvestigated. The capacity of the stair-well pressurization fans is 2.83 m3/s.The exhaust fan capacity in the atriumis 160 m3/s, with two simulations con-ducted with the capacity of the make-up air fans being either 76 or 94 m3/s.The resulting pressure differences areprovided in Table 1. The pressure dif-ferences for the stairwell pressurizationsystems acting alone or together withthe atrium smoke management systemare appreciably different.

Computer-based and physical mod-els are applicable as aids for smokemanagement design. Because each ofthe models provides simplifications ofactual behavior, models can be used

as an aid in establishing or testing thedesign of a smoke management sys-tem. The appropriateness of assump-tions should be confirmed, either bycomparing predictions to data or con-ducting a sensitivity analysis.

James Milke is with the University ofMaryland.

REFERENCES

1 Klote, J.H., and Milke, J.A., Design ofSmoke Management Systems, AmericanSociety of Heating, Refrigerating and Air-Conditioning Engineers, Atlanta, GA.1992.

2 NFPA 92B, Guide for Smoke ManagementSystems in Malls, Atria, and Large Areas,Quincy: NFPA, 1995.

3 Milke, J.A., and Klote, J.H., “SmokeManagement in Large Spaces inBuildings,” Building Control Commissionof Victoria, Victoria, Australia, 1998.

4 Quintiere, J.G., and Dillon, M. E., “ScaleModel Reconstruction of Fire in anAtrium,” Proceedings of the Intl. Conf.On Fire Research and Engineering, BFRLand SFPE, 1995, pp. 397-402.

5 Quintiere, J.G., “Scaling Applications inFire Research,” Fire Safety Journal, 15,1989, pp. 3-29.

6 Friedman, R., “An International Survey ofComputer Models for Fire and Smoke,”Journal of Fire Protection Engineering, 4,3, 1992.

7 Walton, W.D., and Budnick, E.K.,“Deterministic Computer Fire Models,”Fire Protection Handbook, 18th Ed., J.L.Linville (ed.), NFPA, Quincy, MA, 1997.

8 Quintiere, J.G., “Compartment FireModeling,” SFPE Handbook of FireProtection Engineering, 2nd Ed., P.J.DiNenno (ed.), NFPA, Quincy, MA, 1995.

9 Chow, W.K., “A Comparison of the Useof Fire Zone and Field Models forSimulating Atrium Smoke-FillingProcesses,” Fire Safety Journal, 25, 4,1995, pp. 337-354.

10 McGrattan, K.B., Baum, H.R., and Rehm,R.G., “Large Eddy Simulations of SmokeMovement,” Fire Safety Journal, 30, 2,1998, pp. 161-178.

11 Rho, J.S., and Ryou, H.S., “A NumericalStudy of Atrium Fries UsingDeterministic Models,” Fire SafetyJournal, 33, 3, 1999, pp. 213-230.

12 Walton, G., CONTAM96, NationalInstitute of Standards and Technology,Gaithersburg, MD, 1997.

13 Klote, J.H., “A Computer Program forAnalysis of Pressurized Stairwells andPressurized Elevator Shafts,” NBSIR 82-2512, National Bureau of Standards,Gaithersburg, 1982.

14 Heskestad, G., “Engineering Relationsfor Fire Plumes,” SFPE TR 82-8, SFPE,Boston, MA, 1982.

Table 1. Pressure Difference at Stairwell (Pa) from CONTAM96 Analysis

No Atrium 160 m3/s exhaust 160 m3/s exhaust Floor Exhaust 76 m3/s supply 94 m3/s supply

1 54.8 65.2 61.0

2 55.0 65.7 61.5

3 55.3 66.2 62.0

4 55.8 67.0 62.7

5 56.8 68.0 64.0

6 58.0 69.5 65.2


By John H. Klote, Ph.D., P.E.

Afew years ago, most codes inthe United States mandated theair change method that based

the smoke exhaust flow rate of an atrium on the volume of the atrium.While this method is simple to apply,it almost always provides the wronganswer. Today, most codes prescribeatrium smoke protection that is basedon the zone fire model concept that isdiscussed later in this paper.

This paper is an overview of smokecontrol technology, including newinformation that has been proposedfor addition to NFPA 92B. Moredetailed design information on thissubject can be found in the followingpublications: NFPA 92B,1 theASHRAE/SFPE smoke control book byKlote and Milke2, and NISTIR 5516 byKlote.3

TERMINOLOGY

Readers of the above publicationsare cautioned about the meaning ofthe terms smoke control and smokemanagement. The term smoke controlis reserved for systems that providesmoke protection by use of pressuriza-tion, such as a pressurized stairwell.Systems that use any technique includ-ing compartmentation, pressurization,airflow, and buoyancy of hot smokeare referred to as smoke managementsystems. Using this terminology, atriumexhaust systems are smoke manage-ment systems because they rely uponthe buoyancy of hot smoke. This alsoholds for the other types of atriumsmoke protection discussed below.

For smoke management purposes,smoke is considered to consist of theairborne products of combustion plusthe air that is mixed with them. Theairborne products are combustiongases, solid particulates, and liquidparticulates. Inclusion of air in the def-inition of smoke allows us to considersmoke protection systems where thesmoke being generated, exhausted, orvented is actually air mixed with rela-tively small quantities of particulatesand combustion gases. Because theconcentrations of these other quantitiesare relatively small, engineering designanalysis for these smoke managementsystems considers the specific heat, gasconstant, and other properties ofsmoke to be the same as those of air.

An atrium can be considered a largespace of two or more stories. Otherlarge open spaces include enclosedshopping malls, arcades, sports arenas,exhibition halls, and airplane hangars.The methods of this paper also apply

to these spaces. For simplicity, theterm atrium is used in this paper in ageneric sense to mean any of theselarge spaces.

ZONE FIRE MODEL CONCEPT

All conventional approaches to atri-um smoke management are based onthe zone fire model concept. This con-cept is also the basis of several com-puter models.4, 5, 6, 7, 8 This section is abrief synopsis of zone fire modeling,but for more information about thissubject, readers are referred to othersources.9, 10, 11, 12

Because zone fire models were orig-inally developed for room fires, thisdiscussion will start with a room fire.In a room fire, hot gases rise abovethe fire, and these gases form a plume.Since the plume entrains air, the diam-eter and mass flow rate of the plumeincrease with elevation. Accordingly,the plume temperature decreases with

An Overview of

Atrium Smoke Management

Smoke layer

Doorjet Doorjet

Plume Plume

Fire Fire

Smoke layer

Figure 1. Room fire (a) sketch and (b) zone model idealization

(a) Sketch of an atrium fire (b) Zone model idealization of atrium fire



elevation. The fire gases of the plumeflow upward and form a hot stratifiedlayer under the ceiling. Hot gases inthis smoke layer can flow throughopenings in walls to other spaces, andsuch flow is referred to as a doorjet.The doorjet is similar to a plume inthat air is entrained with similar effectson mass flow and temperature. Figure1(a) is a sketch of a room fire.

The concept of zone modeling is anidealization of the room fire conditionsas illustrated in Figure 1(b). For thisidealization, the temperature of the hotupper layer of the room is uniform,and the temperature of the lower layeris also uniform. The height of the dis-continuity between the two layers isthe same everywhere in the room. Thedynamic effect on pressure is consid-ered negligible, so that pressure istreated as hydrostatic. Other propertiesare considered uniform for each layer.Algebraic equations are used to calcu-late the mass flows due to plumes anddoorjets.

Some computer zone fire modelsallow exhaust from the upper layer,and this capability is essential for sim-ulation of atrium smoke exhaust sys-tems. Many zone fire models estimateheat transfer by methods ranging froma simple allowance as a fraction of theheat release of the fire to more com-plicated simulations of conduction,convection, and radiation. Zone modelapplication to atrium smoke exhaust isillustrated in Figures 2(a) and (b).

AXISYMMETRIC PLUME MODELS

Morton, Taylor, and Turner13 devel-oped the classic analysis of the time-averaged flow of plumes. For a height

above the plume source, they consid-ered the air entrained at the plumeedge to be proportional to some char-acteristic velocity of the plume at thatheight. They considered the plume tobe coming from a point source thatmay be either above or below the sur-face of the fuel. Figure 3 is an illustra-tion of a plume next to the idealizedmodel of an axisymmetric plume.

Researchers have extended the workof Morton, Taylor, and Turner to devel-op models of turbulent plumes due tofires in building spaces.14, 15, 16 Noexhaustive study has been conductedto evaluate these plume models forvarious applications. Based on the lim-ited information available, it seems thatthe Heskestad model may be the mostappropriate for atrium applications.

The axisymmetric plume equationsof the design publications mentionedabove are those of Heskestad exceptthat the virtual origin correction hasbeen neglected. The justification forthis simplification is that the virtual ori-gin correction, zo, is small compared tothe plume heights of interest in atriumapplications. Further, for a generaldesign fire of unknown fuel, the virtu-al origin correction, zo, can be eitherpositive or negative. Additional infor-mation about the virtual origin correc-tion is provided by Klote.3

The mass flow of a plume dependson the height, z, and heat release rate,Q . Because of an improved under-standing of plume physics, the pro-posed NFPA 92B has changed z tomean the height above the base of thefuel rather than the height above thetop of the fuel. This results in some-what larger values of z and correspond-ingly higher smoke exhaust flow rates.

The temperature drop due to eleva-tion is dramatic as can be seen fromFigure 4, and the concern with sprin-kler performance for high ceilingspaces becomes apparent.

Heskestad developed his equationsfor strongly buoyant plumes, and itfollows that these equations are notapplicable for small plume tempera-ture rises above ambient. While littleresearch has been conducted on thesetemperature limits, this author suggeststhat the axisymmetric plume equationsof NFPA 92B not be used for tempera-ture rises less than 2 °C (4 °F).

It should be noted that computerzone fire models do not use the plumeequations of NFPA 92B. This isbecause of numerical difficulties associ-ated with a discontinuity in these equa-tions at the mean flame height. Readersshould expect some differencesbetween calculations made with theequations of NFPA 92B and computerzone fire models.

BALCONY SPILL PLUMES

NFPA 92B refers to balcony spillplumes and window plumes. Thesetwo types of plume are very differentand should not be confused. A bal-cony spill plume can be from any sizefire such that smoke flows under abalcony ceiling and into an atrium,and a window plume is from a post-flashover fire such that the smokeflows from the fire room through anopening to an atrium.

Clea

r hei

ght

Exhaust fan

Smokelayer

Plume

Fire

Exhaust

Smokelayer

Plume

Fire

Flame

SmokeAir Air

Z0Virtualorigin

Velocitydistributor

Figure 2. Atrium fire (a) sketch and (b) zone model idealization

(a) Sketch of an atrium fire (b) Zone model idealization of atrium fire

(b) Idealization of plumemodel

Figure 3. Axisymmetric, point source plume(a) sketch and (b) idealized model

(a) Sketch of plume

Q


The balcony spill plume equations inNFPA 92B are for a fire in a room thatopens to a balcony (see Figure 5). Itshould be noted that the balcony spillplume figure in the 1995 edition ofNFPA 92B incorrectly shows the fire onthe balcony, but this will be corrected.

The balcony spill plume equationsof NFPA 92B only apply when there isa doorway lintel between the fireroom and the balcony. This lintel mustbe enough below the ceiling so thatthe momentum of the ceiling jet in thefire room does not directly contributeto the flow out the opening. A ceilingjet consisting of smoke flowing radiallyunder the ceiling forms where a fireplume impacts the ceiling. The depthof the ceiling jet is about 10 percent to20 percent of the height from the baseof the fire to the ceiling.

The balcony spill plume equationsof NFPA 92B also are used in theUnited Kingdom.17 Morgan et al.18 pre-sent a number of approaches that gobeyond the constraints of the NFPA92B equations. In general, theseapproaches are based on the perimeterof the fire rather than the heat releaserate, and they require somewhat morecumbersome calculations. When facedwith situations that are different fromthe spill plume of NFPA 92B, designersmay want to consider the approachesof Morgan, et al.

For narrow balconies, smoke cancurl inward toward the structure andmove into portions of any balconiesabove. Morgan et al. indicate thatexperiments have shown that suchinward curling smoke can occur forbalconies less than 2 m (6.6 feet) wide.

WINDOW PLUMES

A window plume can occur from apost-flashover fire. A post-flashover fireis one in which every object in the fireroom that can burn is burning. Theheat release rate of a post-flashover firedepends on the amount air availablefor combustion, and the airflow intothe room depends on the size andshape of the opening. Because of thecommon use of sprinklers, considera-tion of window plumes generally isreserved for unusual designs.

SMOKE MANAGEMENT SYSTEMS

The following are three types ofsmoke management systems: (1) fan-powered smoke exhaust, (2) smokefilling, and (3) gravity venting. Thesesystems can be designed for a steadyfire or an unsteady fire. For informa-tion about design fires, readers arereferred to NFPA 92B and theASHRAE/SFPE smoke control book.

The three system types are based onthe objective of preventing smokefrom contacting occupants. An alterna-tive objective is a system designed toprovide tenable conditions even withsmoke coming into contact with peo-ple. Such tenability systems are recog-nized by NFPA 92B, the ASHRAE/SFPEsmoke control book, and CIBSE17.Information about tenability design isprovided by Klote19.

SMOKE EXHAUST

Smoke can be exhausted near thetop of an atrium to prevent smoke

050

100

150

200

250 150

100

50

050 100 150 200 250 300 0 0.25 0.50 0.75 1.00 1.25 1.50

Aver

age

tem

pera

ture

( F

)

Elevation above fuel (ft) Atrium exhaust (millions of cfm)Note: Average temperature calculated from equations of NFPA 92B (1995). Note: Atrium exhaust calculated from equations of NFPA 92B.

Clea

r hei

ght (

ft)

Heat release rate (kW):10,000 5,000 2,000

Heat release rate (kW):10,000 5,000 2,000

H

Section

Front view withdraft curtains

b

Draftcurtain

Front view withoutdraft curtains

Figure 4. Average temperature of axisymmetric plume Figure 6. Atrium exhaust needed to maintain a constantclear height with a fire in the atrium

Figure 5. Balcony spill plume

°C = (°F-32)/1.8 1 m = 3.3 ft. 1 m3/s = 2100 cfm


contact with occupants during evacua-tion. There are two approaches todesign of smoke exhaust systems: (1)size the smoke exhaust to maintain aconstant clear height and (2) size thesmoke exhaust so that a descendingsmoke layer does not contact occu-pants during evacuation.

The first approach has the advan-tage that the calculations are relativelysimple, and Figure 6 shows theexhaust needed to maintain a constantclear height with a fire in the atrium.Calculations for the second approachare more complex, and these calcula-tions can be done by computer zonemodels.

SMOKE FILLING

While smoke exhaust is probablythe most common form of atriumsmoke management in the UnitedStates, some atriums are of such sizethat no smoke exhaust is needed tokeep the occupants from contactingsmoke throughout a fire evacuation.This form of smoke management iscalled smoke filling and requires noexhaust capabilities.

Without smoke exhaust, the smokelayer that forms under the ceilinggrows thicker, and the bottom of thatsmoke layer drops downward.Equations can be used to calculate thetime that it takes for the smoke todrop to a level that is above the headsof all the people in the atrium duringa fire evacuation. If this smoke fillingtime is greater than the evacuationtime, smoke filling is a viable form ofsmoke protection.

People movement calculations areused to determine the evacuation time.As we all know, when a fire alarmsounds, most people have a tendencyto wait to see if there really is a fire orto see if conditions are threatening.This decision time needs to be allowedfor in any calculation of evacuation time.Readers interested in people movementcalculations should see The SFPE Hand-book of Fire Protection Engineering.20

GRAVITY VENTING

In the United Kingdom andAustralia, gravity venting is often usedwhere we in the U.S. would use fan-

powered smoke exhaust. This naturalmethod consists of opening vents inthe atrium ceiling or high on the atriumwalls to let the smoke flow out withoutthe aid of fans. The proposed NFPA92B addresses gravity venting systems.

The flow rate through a gravity ventcan be calculated, and it depends onthe (1) size of the vent, (2) depth ofthe smoke layer, and (3) temperatureof the smoke. When smoke is detected,all the vents need to be opened at onetime. Thermally activated vents, likethose often used for industrial heat andsmoke venting, are inappropriate forgravity venting of atria because of thetime delay in opening the vents.

The applicability of gravity ventingdepends primarily on the (1) size ofthe atrium, (2) outside design tempera-tures, and (3) wind conditions. Gravityventing is simpler and less costly thanfan-powered exhausting. Because lossof power can occur during fire situa-tions, there is a significant advantageto a smoke management system thatrequires no power for fans.

Some people are uncomfortablewith gravity venting, probably becauseof the lack of positive assurance ofobtaining the desired flow. However,the reliability and economic benefits ofgravity venting are such that gravityventing will likely find a place in U.S.buildings in the future.

MAKE-UP AIR

For smoke exhaust systems and grav-ity venting systems, air must be sup-plied to the atrium to make up for thesmoke exhaust. A few months ago, anengineer erroneously indicated to theauthor that he thought that make-up airwould not be needed for large atria,because they already have such a largevolume of air. This is not so. Make-upair is essential for all smoke exhaustsystems and gravity venting systems.Make-up air can be either fan poweredor nonpowered. The network computerprogram, CONTAM96,21 can be used toanalyze nonpowered make-up airflows.

VELOCITY LIMIT

There is a concern that air velocitycould disrupt the structure of theplume resulting in failure of the smoke

management system. The best currentinformation available is that a velocityof 1 m/s (200 fpm) or less will notcause such disruption. For this reason,NFPA 92B indicates that velocitiesshould not exceed 1 m/s (200 fpm) inthe atrium where there could be aplume. This applies to any air velocitywhether it is for make-up air or forsome other purpose.

STRATIFICATION AND DETECTION

The issue of smoke stratification isincluded in the proposed NFPA 92B.Often a hot layer of air forms underthe ceiling of an atrium, the result ofsolar radiation on the atrium roof.While studies have not been made ofthis hot air layer, many professionalsbelieve that such layers are often inexcess of 50 °C (120 °F) (Figure 7).When the temperature of the smokeplume is less than that of the prestrati-fied layer, the smoke cannot reach theceiling. In this situation, the smokecannot activate ceiling-mounted smokedetectors (Figure 8).

Beam smoke detectors can be usedto overcome this detection difficulty,and the proposed NFPA 92B describesthree methods of using beam detectorsfor this purpose. Two of the methodsemploy horizontal beams with theintent of (1) detecting the smoke layeror (2) detecting the plume.

The third method uses upward-angled beams with the intent of detect-ing the development of a smoke layerat whatever stratification conditionsexist. In this third approach, one ormore beams are aimed at an upwardangle to intersect the smoke layerregardless of the level of smoke stratifi-cation. For redundancy when usingthis approach, more than one beamsmoke detector is recommended.

NUMBER OF EXHAUST INLETS

When the smoke layer depth belowan exhaust inlet is relatively shallow, ahigh exhaust rate can lead to entrain-ment of cold air from the clear layer(Figure 9). This phenomenon is called“plugholing,” and it is addressed in theproposed NFPA 92B. To preventplugholing, more than one exhaustinlet point may be needed. The maxi-


mum mass flow rate, which can beefficiently extracted using a singleexhaust inlet, is given as:

where:= maximum volumetric flow rate

at Ts, m3/s (cfm)Ts = absolute temperature of the

smoke layer, K (R)To = absolute ambient temperature,

K (R)d = depth of smoke layer below

exhaust inlet, m (ft)β = exhaust location factor (dimen-

sionless)C1 = 0.00887 (0.537)

The above equation is consistentwith the approach of CIBSE (1995).Based on limited information, suggest-ed values of β are 2.0 for a ceilingexhaust inlet near a wall, 2.0 for a wallexhaust inlet near the ceiling, and 2.8for a ceiling exhaust inlet far from anywalls. It is suggested that d/D begreater than 2, where D is the diameterof the inlet. For exhaust inlets, use D =2ab/(a + b), where a and b are thelength and width of the inlet. Theresults of experiments conducted at theNational Research Council of Canadaare consistent with this approach toavoiding plugholing.22, 23, 24

SEPARATION BETWEEN INLETS

When the exhaust at an inlet is nearthis maximum flow rate, adequate sep-aration between exhaust inlets needsto be maintained to minimize interac-tion between the flows near the inlets.This separation is also addressed inthe proposed NFPA 92B. One criterionfor the separation between inlets isthat it be at least the distance from asingle inlet that would result in anarbitrarily small velocity based on sinkflow. Using 0.2 m/s (40 fpm) as thearbitrary velocity, the minimum separa-tion distance for inlets located in awall near the ceiling (or in the ceilingnear the wall) is:

Exhaust fan

Hot air layer

Temperature

Elev

atio

n

Exhaust fan

Plume

Hot air layer

Note: Beam smoke detectorscan be used to overcomethe detection problemcaused by a layer of hot airunder the ceiling.

Exhaust fan

Plume

Fire

Plugholing ofair into smokeexhaust

Figure 7. Hot layer of air under atriumceiling (a) sketch and (b) temperatureprofile

Figure 8. Stratified smoke in an atrium with a hot layer of air under the ceiling

Figure 9. Plugholing of air into smoke exhaust inlets can result in failure of asmoke exhaust system

(a) Sketch of hot air layer (b) Temperature profile

maxV C d T T To s o= −( )1 5/ 2β

maxV

(1)


where:Smin = minimum separation between

inlets, m (ft)= volumetric flow rate, m3/s (cfm)

β = exhaust location factor (dimen-sionless)

C2 = 0.32 (0.023)

SMOKE LAYER DEPTH

The proposed NFPA 92B clearlyindicates that the smoke layer must bedesigned to be deep enough to allowfor a ceiling jet.

WHEN EQUATIONS DON’T APPLY

As a general rule, it can be statedthat equations don’t apply when theassumptions behind those equationsare not appropriate. For example, theequation for the mass flow of anaxisymmetric plume does not apply ifobstructions break up the plume flow.Also the basic zone model approachdoes not apply if the smoke plumecools so much that there is no well-defined smoke layer under the ceiling.Another example would be that thebalcony spill equations of NFPA 92Bdo not apply when the fire is on thebalcony. It is not possible to catalog allpossible situations for which equationsdon’t apply, because of the variety andcomplexity of buildings. Practitionersneed to understand the assumptionsbehind the equations they use to besure that the equations apply.

Techniques that can be used whenthe equations don’t apply are (1) scalemodeling and (2) computational fluiddynamics (CFD). Probably the mostcommon kind of scale modeling isFroude modeling which preserves theFroude number. This number can bethought of as the ratio of the inertiaforces to gravity forces, and this num-ber is important to smoke modelingbecause the buoyancy of hot smoke isa gravity force. Froude modeling con-sists of building a scale model andburning a scaled-down fire in thatmodel in air at atmospheric pressure.

CFD modeling consists of dividing aspace into a large number of spacesand obtaining approximate solutions

to the fundamental equations of fluiddynamics for each space. Many com-puter CFD programs have been devel-oped that are capable of simulation offire-induced flows (Friedman10).

Several of these are general purposecodes that are commercially available.Readers should be aware that a thor-ough knowledge of CFD requires exten-sive understanding of graduate-levelfluid dynamics. NISTIR 5516 providesintroductory information about bothFroude modeling and CFD modeling.

John H. Klote is a fire and smokeconsultant in Virginia.

REFERENCES

1 NFPA 92B. Guide for SmokeManagement Systems in Malls, Atria,and Large Areas, National Fire ProtectionAssociation, Quincy, MA, 1995.

2 Klote, J.H., and Milke, J.A. Design ofSmoke Management Systems, AmericanSociety of Heating, Refrigerating and Air-conditioning Engineers, Atlanta, GA, 1992.

3 Klote, J.H. Method of Predicting SmokeMovement in Atria With Application toSmoke Management, National Institute ofStandards and Technology, NISTIR 5516,1994.

4 Mitler, H.E., and Emmons, H.W.Documentation for CFC V, the fifthHarvard computer code. Home FireProject Tech. Rep. #45, HarvardUniversity, 1981.

5 Cooper, L.Y. “ASET – A ComputerProgram for Calculating Available SafeEgress Time,” Fire Safety Journal, Vol. 9,pp. 29-45, 1985.

6 Tanaka, T. A Model of Multiroom FireSpread. National Bureau of Standards,NBSIR 83-2718, 1983.

7 Cooper, L.Y., and Forney, G.P. TheConsolidated Compartment Fire Model(CCFM) Computer Code ApplicationCCFM.VENTS - Part I: Physical Basis, 1990.

8 Peacock, R.D., Forney, G.P., Reneke, P.,Portier, R., and Jones, W.W. “CFAST, theConsolidated Model of Fire Growth andSmoke Transport,” National Institute ofStandards and Technology, NISTTechnical Note 1299, 1993.

9 Bukowski, R.W. “Fire Models, the Futureis Now!,” NFPA Journal, No. 85, Vol. 2,pp. 60-69, March/April, 1991.

10 Friedman, R. “An International Survey ofComputer Models for Fire and Smoke,”

Journal of Fire Protection Engineering,Vol. 4, No. 3, pp. 81-92, 1991.

11 Mitler, H.E., and Rockett, J.A. “HowAccurate is Mathematical FireModeling?,” National Bureau ofStandards, NBSIR 86-3459, 1986.

12 Mitler, H.E. Comparison of SeveralCompartment Fire Models: An InterimReport, National Bureau of Standards,NBSIR 85-3233, 1985.

13 Morton, B.R., Taylor, G., and Turner, J.S.“Turbulent Gravitational Convectionfrom Maintained and InstantaneousSources,” Proceedings of the RoyalSociety of London, Vol. 234, pp. 1-23,1956.

14 McCaffrey, B.J. “Momentum Implicationsfor Buoyant Diffusion Flames,Combustion, and Flame,” Vol. 52, No. 2,pp. 149-167, 1983.

15 Cetegan, B.M., Zukoski, E.E., andKubota, T. Entrainment and FlameGeometry of Fire Plumes, Ph.D. Thesisof Cetegen, California Institute ofTechnology, Pasadena, 1982.

16 Heskestad, G. “Engineering Relations forFire Plumes,” Fire Safety Journal, Vol. 7,No. 1, pp. 25-32, 1984.

17 Relationships for Smoke ControlCalculations. TM19: Chartered Instituteof Building Service Engineers, London,1995.

18 Morgan, H.P. et al. Design Methodologiesfor Smoke and Heat Exhaust Ventilation,CRC Ltd, London, 1999.

19 Klote, J.H. An Engineering Approach toTenability Systems for Atrium SmokeManagement, ASHRAE Transactions, Vol.105, Part 1, 1999.

20 DiNenno, P. (ed) SFPE Handbook of FireProtection Engineering, National FireProtection Association, Quincy, MA, 1995.

21 Walton, G.N. CONTAM96 User Manual,National Institute of Standards andTechnology, NISTIR 6056, 1997.

22 Lougheed, G.D. and Hadjisophocleous,G.V. “Investigation of Atrium SmokeExhaust Effectiveness,” ASHRAETransactions, Vol. 103, Part 2, 1997.

23 Lougheed, G.D, Hadjisophocleous, G.V.,McCartney, C., and Taber, B.C. “Large-Scale Physical Model Studies For AnAtrium Smoke Exhaust System,”ASHRAE Transactions, Vol. 104, 1999.

24 Hadjisophocleous, G.V., Lougheed, G.D.,and Cao, S. “Numerical Study of theEffectiveness of Atrium Smoke ExhaustSystems,” ASHRAE Transactions, Vol.104, 1999.

S C Vmin e= 2β ˙(2)

Ve


Currently Available Methods and Their Merits for Everyday Use

MODELING SPOT-TYPE SMOKE DETECTOR RESPONSE MODELING SPOT-TYPE SMOKE DETECTOR RESPONSE

By William E. Pucci, P.E.

INTRODUCTION

Members of the fire protection com-munity are often faced with predictingthe response time of spot-type smokedetectors. Many studies have beendone and many papers written, butthe fact still remains that such predic-tions are, at best, approximate. Thisarticle summarizes the methods cur-rently available with a focus on whateveryone should know about smokedetector use.

MOTIVATION

Understanding how smoke detectorsreact to fire is of utmost importance inmeeting fire safety objectives. Earlydetection of smoke plays a key role inthe life safety of building occupants.Many times, smoke detector responseis the first indication of a fire in abuilding. As a result, it would be idealfor design engineers to be able todemonstrate that this response allowsoccupants sufficient time to safelyevacuate before untenable conditionsare reached. Similarly, report review-ers, such as Authorities HavingJurisdiction (AHJs), need assurancethat smoke detector response calcula-tions are soundly based. They alsoneed to be able to determine whetheror not the input to the prediction

model was accurate and whether ornot a sufficient safety margin wasincluded based on the limitations ofthe prediction method used. Finally,whether or not spot-type smoke detec-tors responded properly and promptlyis also important to those involved inpost-fire reconstruction.

Many professionals oppose the theo-retical modeling of smoke detectorresponse altogether. In some respectsthis opposition is warranted, as themethods commonly used to predictresponse are not even consistent withthe detection mechanism. However,numerous studies have been done forspecific situations which, when used aspart of an overall analysis, have somemerit. This article focuses on conven-tional, spot-type smoke detectors com-monly found in commercial, nuclear,industrial, and residential applications.

REASONS FOR LACK OF ACCURATE MODELS

Despite a fair amount of research onthe subject, modeling of smoke detec-tor response is still far from refined.The reason for this lies in the largenumber of interrelated variables thatcome into play. These variablesinclude those that affect conditions inthe room (which in turn affect opera-tion of detectors) and those that affectthe response of a particular detector tothose conditions.

Examples of variables that affect con-

ditions within the room include but arenot limited to the size and geometry ofthe room and building, the ambientconditions, the type and arrangementof combustibles, and the nature of firespread and growth. Since detectorresponse is contingent on conditionswithin a room, even the slightestchange in a single variable can signifi-cantly affect detector response.

As far as variables that affect theresponse of a particular detector, differ-ent detectors respond differently to dif-ferent types of fires. First, photoelectric(light-scattering) smoke detectors use adifferent operating principle than ion-ization detectors. Second, small changesin design features between two detec-tors of the same type can result in dif-ferent response characteristics. The datareported by manufacturers and testinglaboratories are extremely limited interms of characterizing the response ofan individual detector to the many dif-ferent types of fires that it may beexposed to during its lifetime.

CURRENT METHODS

In short, there are no practical meth-ods for directly modeling the responseof photoelectric or ionization spot-typesmoke detectors. It isn’t that theoriesdon’t exist, but the uncertainly in thecalculations and the lack of pertinentdata make these methods unusable inmost cases. What remains then aremethods to estimate the response of


spot-type detectors. “Estimate” isdefined in Webster’s Ninth NewCollegiate Dictionary as “to judge ten-tatively or approximately the value,worth, or significance of; to determineroughly the size, extent, or nature of.”Thus, anyone who writes a report ormakes a statement that a smoke detec-tor “WILL go off at x seconds” is mis-representing the facts.

Photoelectric detectors operate byprojecting a light source into a sensingchamber and positioning a light receiv-er at some angle to that source. Ifsmoke is present, light is reflected andrefracted (bent) by the smoke onto thereceiver to produce a signal. Ionizationdetectors rely on a small radioactivesource to produce an electrical currentbetween two oppositely chargedplates. When smoke enters the cham-ber of an ionization detector, someions attach to the smoke particles. Theincreased mass of the ion slows itdown and may allow the small air cur-rents to carry the particle/ion out ofthe chamber before it reaches the plateon the other side. This reduces the netcurrent flow between the electricalplates. Also, the particle/ion combina-tion is a larger target for collision withan oppositely charged particle. When apositive ion collides with a negativeion, they recombine and further reducethe current flow in the chamber.1

The purpose of this article is not togo into detail on the operating princi-ples of the different types of spot-typedetectors. However, it is important torecognize that the most widespread

method for estimating the response ofspot-type detectors is by comparing theamount of smoke (obscuration, or opti-cal density per unit length) at a detec-tor location to the amount of smokerequired for that detector to go intoalarm. While this concept seemsstraightforward, neither photoelectricnor ionization detectors use obscura-tion as a detection method.Nevertheless, optical density is the onlypractical parameter for estimating theresponse of spot-type detectors. Twomethods on how to predict opticaldensity are discussed below.

TEMPERATURE APPROXIMATION

The most common way that opticaldensity is predicted is through what isknow as the “temperature approxima-tion method”. In this theory, the opticaldensity in the vicinity of a detector ispostulated to be directly proportional tothe temperature rise at that location.While the operation of smoke detectorsis independent of temperature rise, atemperature rise indicates that combus-tion products necessary for activatingthe smoke detectors are present.Because models for estimating tempera-ture rise are fairly well developed, usersneed only know the temperature riserequired for smoke detector activation.

Unfortunately, the principle bywhich smoke detectors sense fire in aroom is considerably more complexthan simply monitoring the temperaturerise in the vicinity. The temperatureapproximation method completelyignores several phenomena that affectthe response of the detector and there-fore cannot be expected to give accu-rate response times. That being said,due to a lack of other available meth-ods, there is some merit to using thismethod, if done properly.

First and foremost, the type of fuelthat is being burned must be known.As an example, the study that firstdocumented the temperature approxi-mation approach2, 3 demonstrated that acertain ionization detector required atemperature rise of 14 °C (25 °F) for awood fire but only 1.7 °C (3 °F) for acotton fire. Similarly, there are manydifferent kinds of ionization detectors,all of which can alarm at different levels.

In 1985, one study4 showed how a

particular combination of ionizationdetector and wood crib fuel wouldresult in a temperature rise at responseof 13 °C (23 °F). This value hasbecome a common benchmark fordetector response since that time.More recent work5 has proposed atemperature rise of 5 °C (9 °F) forhydrocarbon fires and ionizationdetectors responding to an obscurationlevel of 2.5% per meter (i.e., theirreported sensitivity). The choice of acertain temperature rise criteria shouldbe carefully selected and justified (seereferences for limited available data).

MASS OPTICAL DENSITY METHOD

The second method available to pre-dict the optical density in a space iscalled the mass optical density (Dmass)method1. In this method, the followingequation is used to estimate the opti-cal density per unit length:

where:Du = optical density per unit length

(m-1)L = path length (m)Dmass = mass optical density (m2/g)Wf = mass of fuel burned (g)Vc = volume in which the smoke is

dissipated

Cone and furniture calorimeters areused to experimentally determine val-ues of Dmass.

The method requires the user toselect a value of Dmass for the particularfuel and burning mode. The mostcomplete set of data is available in theSFPE Handbook of Fire ProtectionEngineering.6 However, not all fuelsand burning modes have been testedand reported. Also, while only onevalue of Dmass can be specified, manyfuel packages are actually combina-tions of several different fuel types.

In addition to reported values ofDmass, Tewarson7 examined data fromseveral sources and developed the fol-lowing equations for when the smokeyield (Ys) is available:

Dmass = 0.10×ln(Ys)+0.52 (for flaming fires)Dmass = 0.10×ln(Ys)+0.65 (for smoldering

fires)

DD

LD

W

Vu massf

c

= =

KEY POINTS OF SMOKE DETECTOR

MODELING

■ There are no practical

methods for directly mod-

eling the response of spot-

type detectors.

■ The methods that do exist

are merely estimates, and

the results from using

these methods are approxi-

mate at best.

(1)


Regardless of what method is usedto determine Dmass, the mass opticaldensity method assumes that thesmoke produced is uniformly distrib-uted throughout the specified volume.Called a zone model, this, too, intro-duces error into the results, as opticaldensity can vary from point to pointwithin the volume. Also, not all smokeends up in the volume, as some is typ-ically deposited on walls, ceilings, andobjects within a room as soot. Theselimitations of zone models, as well asthe fuel-specific nature of Dmass, shouldbe recognized when making estimatesof detector response using mass opti-cal density.

LIMITATIONS ON THE USE OFOPTICAL DENSITY

Neither photoelectric nor ionizationdetectors use obscuration as a detect-ing mechanism. Some other importantlimitations on the use of optical densi-ty, the most widespread method ofestimating the response of spot-typesmoke detectors, are discussed below.

Sensitivity RatingsSpot-type smoke detectors are

labeled by testing laboratories with asingle calibration point in obscuration– specifically the response to a smol-dering, cotton lamp wick in a chamberwith a forced-air current. Fuels otherthan lamp wicks have different valuesof obscuration at response, as do dif-ferent burning modes (i.e., flamingversus smoldering). For example, adetector labeled with a sensitivity of2.5% per meter may respond at a dif-ferent obscuration level for a flamma-ble liquid pool fire. Those predictingsmoke detector response should con-sider that the labeled sensitivity doesnot represent the obscuration level atresponse for all fuels and burningmodes.

Fuel CharacteristicsThe two most common types of

spot-type detectors, ionization andphotoelectric, respond differently todifferent types of fires. All of the workdone to date on establishing tempera-ture-rise criteria has been done forflaming fires. A smoldering fire mightgenerate considerably more smoke inthe early stages, when the temperature

in the room may not necessarily besignificantly elevated.

While there are values of Dmass

reported for pyrolysis (i.e., smolderingcombustion), values for flaming firesare more widely reported. An addition-al limitation of using mass optical den-sity is that the reported values are forspecific fuel. In an actual fire, the itemfirst ignited may actually be comprisedof several different fuels. The user ofthis model must therefore determinewhat value of Dmass to use – certainly asubjective determination.

Although little work has been donewith smoldering fires, anyone predict-ing smoke detector response should atleast subjectively consider the potentialfor smoldering fires and their effectson predictions. As a rule of thumb,ionization detectors generally respondmore quickly to flaming fires, andphotoelectric detectors respond morequickly to smoldering fires.

Wavelength of LightThe value of optical density mea-

sured in a test depends on the wave-length of light used. For a given set oftest conditions, if the wavelength ofthe measuring light beam is reduced,the measured optical density willincrease. Since the wavelength of thelight source in a detector may not bethe same as the wavelength of themeasuring light beam in a given test,error will be introduced into calcula-tions where the wavelength of light isnot considered. This can have a signif-icant impact on the results and is pre-sent whenever test data are used torepresent potential future fires.

Smoke Entry ResistanceIt is possible to take one step beyond

the conventional optical density meth-ods, as proposed by Heskestad8 andfurther pursued by researchers at theVTT Research Institute in Finland9 andby Marrion10 and Oldweiler11. Thismethod recognizes the fact that spot-type smoke detectors experience a timelag, as the smoke outside of the detec-tor must penetrate to the inside wherethe detection mechanism is located. Inother words, just because the opticaldensity outside a detector has reachedthe alarm threshold does not necessari-ly mean that the optical density insidethe detector has reached that level.

Called “smoke entry resistance”, theconcept is very similar to the RTI ofheat detectors in that it identifies a timeconstant for the lag time associatedwith entry of smoke into the detectionchamber. Items such as insect screensand the geometry of the detector itselfcontribute to smoke entry resistance.Heskestad proposed that the time lagcould be represented by a time con-stant using the following relaton:8

where:L = characteristic length (also known

as “L-number”) (meters)τ = detector time constant (seconds)U = ceiling jet velocity flowing past the

detector (m/sec)

In short, the L-number, which hasunits of length, is interpreted as thedistance the smoke would travel at agiven velocity before the optical densi-ty inside the detector reaches thevalue outside of the detector. The L-number is thought to be a propertyof the detector that is independent ofthe smoke and ceiling jet properties.While this theory holds promise, thedata are sparse, and the effect ofvelocity needs further evaluation.However, the phenomenon has beenobserved, and particularly for smolder-ing fires with low gas velocities, usersof models should give consideration toits possible influence on results.

CONCLUSIONS

Due to the uncertainty involved inpredicting smoke detector response, arange of potential detection timesshould generally be provided.Additionally, the limitations of themodels being used should always bestated, and the applicability of inputdata to the postulated fire scenarioshould be considered.

It is clear that more data are neededto be able to predict smoke detectorresponse more accurately. For photo-electric detectors, more information isneeded on the optical characteristicsproduced by different fuels, specifically,how the smoke produced by certainfuels reflects and refracts light at thewavelength used by detectors, and howmuch reflected/refracted light is

L U= ×τ (2)


required for response. For ionization detectors, data are neededon the number and size of smoke particles produced by differ-ent fuels. Also, data on the velocity of airflow/currents throughthe detection chamber and its effect on response are needed.

Despite the aforementioned limitations, the past perfor-mance of both photoelectric and ionization detectors shouldnot go unnoticed. If installed in accordance with their listingand manufacturer’s installation instructions, both types havedemonstrated the ability to provide sufficient time for occu-pants to safely egress. The merits of modeling their responsequantitatively will see greater attention as performance-basedrequirements replace prescriptive ones in model buildingcodes and as owners begin to establish fire safety objectivesthat include preservation of property and continuity of opera-tions in addition to the life safety of the building occupants.

William Pucci is with Performance Consultants, Inc.

REFERENCES

1 Schifiliti, R. P., and Pucci, W. E., “Fire Detection Modeling: State ofthe Art,” Fire Detection Institute, 1996.

2 Heskestad, G., and Delichatsios, M., “Environments of FireDetectors – Phase I: Effect of Fire Size, Ceiling Height, andMaterial, Volume 1 – Measurements,” NBS-GCR-77-86, NationalBureau of Standards, Gaithersburg, MD, May 1977.

3 Heskestad, G., and Delichatsios, M., “Environments of FireDetectors – Phase I: Effect of Fire Size, Ceiling Height, andMaterial, Volume 2 – Analysis,” NBS-GCR-77-95, National Bureauof Standards. Gaithersburg, MD, June 1977.

4 Evans, D., and Stroup, D., “Methods to Calculate the ResponseTime of Heat and Smoke Detectors Installed Below Large,Unobstructed Ceilings,” U.S. Department of Commerce, NationalBureau of Standards, February 1985.

5 Davis, W. D., and Notarianni, K. A., “NASA Fire Detector Study:NISTIR 5798,” National Institute of Standards and Technology,March 1996.

6 Mulholland, G. W., “Smoke Production and Properties,” SFPEHandbook of Fire Protection Engineering, 2nd Edition. NationalFire Protection Association, Quincy, MA, 1995.

7 Tewarson, A., “Generation of Heat and Chemical Compounds inFires,” SFPE Handbook of Fire Protection Engineering, 2nd Edition.National Fire Protection Association, Quincy, MA, 1995.

8 Heskestad, G., “Generalized Characteristics of Smoke Entry andResponse for Products-of-Combustion Detectors,” Proceedings –7th International Conference on Problems of Automatic FireDetection, Rheinish-Westfalischen Technischen HochschuleAachen, March 1975.

9 Bjorkman, J., Kokkala, M. A., and Aloha, H., “Measurement of theCharacteristic Length of Smoke Detectors,” Fire Technology. Vol.28, No. 2, May 1992, pp. 99-109.

10 Marrion, C., “Lag Time Modeling and the Effects of Ceiling JetVelocity on the Placement of Optical Smoke Detectors,” MasterThesis, Worcester Polytechnic Institute, Worcester, MA, 1989.

11 Oldweiler, A., “Investigation of the Smoke Detector L-Number in theUL Smoke Box,” Master Thesis, Worcester Polytechnic Institute,Worcester, MA, 1995.

Fire protection engineering is a growing profession with many challenging career opportunities. Contact the Society of Fire Protection Engineers at www.SFPE.org or the organizations below for more information.


Explore your full potential! Come join the team at TVA Fire & Life Safety, Inc. – a growing international fire protection engi-

neering/consulting firm headquartered in San Diego, CA, w/officesin CA, MI, GA, NJ, and TX.

Qualified individuals are Registered PEs with 3+ yrs. exp., in-depthknowledge of Model Codes and NFPA stds., and excellent communi-cation/interpersonal skills. Duties include, but are not limited to:• Preparing studies of industrial, commercial, and other properties considering

factors such as fire resistance, usage or contents of buildings, water suppliesand delivery, and egress facilities.

• Designing or recommending materials/equipment such as structural componentsprotection, fire detection equipment, alarm systems, extinguishing devices and systems, and advising on location, handling, installation, and maintenance.

• Consulting with customers to define needs and/or issues and gathering informa-tion to determine the scope of work.

• Conducting meetings with fire and building officials to discuss upcoming and existing projects and answering any questions that may arise.

• Advising customers on alternate methods or recommending specific solutions tosolve problems that may arise.

• Conducting job site inspections, preparing and providing a technical report of findings to customers and/or AHJs.

Enjoy a competitive salary, medical/dental benefits, profit-sharing,401(k), and company stock purchase plan. (EOE) Send your résumé to: HR Department, TVA Fire & Life Safety, Inc.

2820 Camino del Rio South, Suite 200 San Diego, CA 92108Fax: 619.296.5656 E-mail: [email protected]

Arup Fire has immediate openings for Fire Protection Engineersin New York. Successful candidates will play a very active role

in developing the practice in the USA and will work closely withmany of the world’s leading architects and building owners develop-ing innovative design solutions for a wide range of building, indus-trial, and transport projects.

Candidates should possess a Fire Protection Engineering degree,approximately five years of experience, and preferably an FPE – PE.Risk management, industrial fire engineering, and computer model-ing skills will be highly regarded.

Similar opportunities available in London, Leeds, Dublin, HongKong, and Australia, with opportunities available in Boston, SanFrancisco, and Los Angeles in the near future.

Arup Fire offers competitive salaries and benefit packages. Pleasesubmit résumé and salary history to:

Chris Marrion, PEArup Fire Ove Arup & Partners 155 Avenue of the AmericasNew York, NY 10013Telephone: +1.212.896.3269Fax: +1.212.229.1986E-mail: [email protected]

EQUAL OPPORTUNITY EMPLOYER

Harrington Group, Inc., is focused on continuous, profitable growthand currently has openings in Atlanta for Fire Protection Engineers.

For full details on current job openings, please visit our Web site atwww.hgi-fire.com or submit your résumé via e-mail or fax to:

Ms. Patsy Sweeney [email protected] Fax: 770.564.3509 Phone: 770.564.3505

Harrington Group is a full-service fire protection engineering designand consulting firm. Founded in 1986, Harrington Group has con-sistently provided a high level of quality and value to clientsthroughout North America, South America, and Germany.

Should you be interested in us? YES! If:

• You want a career with increasing responsibility and compensation.• You care about quality and service delivered to the client.• You have strong engineering skills and people skills.• You are creative and desire to use your creativity.• You are honest and hard-working.• You want to be trusted and respected by your company.• You want to participate in the financial aspects of your company – like owners do.• You are in a dead-end where you are now.

Check us out, and discover whatmakes Harrington Group such anexcellent career opportunity.

Fire ProtectionEngineers






As the global leader in fire protection, security, and life safety solutions, Rolf Jensen & Associates, Inc., is always looking for

talented, dynamic individuals. Opportunities exist throughout oureleven offices for engineering and design professionals looking forgrowth. We are looking for engineers with experience in fire alarm,sprinkler, and security design; code analysis; and business develop-ment.

Check out our Web site at www.rjagroup.com for more details. Send your résumé to:

Ralph Transue, PEThe RJA Group, Inc.549 W. Randolph St., 5th FloorChicago, IL 60661

RJA EmploymentOpportunities

RJA EmploymentOpportunities


Founded in 1973, Code Consultants, Inc. (CCI), is a nationally recog-nized fire protection engineering firm providing professional con-

sulting and design services to developers, owners, architects, and othersignificant clients throughout the United States. With a staff of 55, CCIis a dynamic, growing firm that has an unmatched reputation fordeveloping innovative fire protection and life safety solutions, codecompliance guidance, and cost-effective designs which are equallywell received by clients and governing officials. CCI’s projects includesome of the nation’s largest shopping malls, retail stores, stadiums andarenas, hospitals, convention centers, detention/correctional facilities,transportation (air and rail) facilities, warehouses, and theaters for theperforming arts, to name a few.

The firm is seeking degreed Fire Protection Engineers and otherdegreed individuals with a high level of experience applying ModelCodes and NFPA standards to service clients and projects throughoutthe country.

These positions offer a unique income opportunity, including partici-pation in CCI’s lucrative performance incentive program. The positionrequires residency in the St. Louis area.

Code Consultants, Inc.1804 Borman Circle Dr.St. Louis, MO 63146314.991.2633

Koffel Associates, Inc., is a fire protection engineering and codeconsulting firm with offices in Connecticut, Maryland, and

Tennessee that provides services internationally. Positions are avail-able at the following levels:

Senior Fire Protection EngineerRegistered Fire Protection EngineerFire Protection Engineers (BS or MS in FPE)Fire Protection Engineering Technician (AutoCAD experience,

NICET, or technology degree desirable)

Responsibilities may include:• Fire protection engineering and life safety surveys• Design and analysis of fire protection systems including automatic sprin-

klers, clean agent, fire alarm and detection, water supply, and smoke man-agement systems

• Code consultation with architects, engineers, developers, and owners during design and construction

• Post-fire analysis and investigation• Computer fire modeling• Fire risk and hazard assessments• Codes and standards development

Koffel Associates, Inc., personnel actively participate in the activitiesof professional engineering organizations and the codes and stan-dards writing organizations. The firm offers a competitive salary andbenefits package including conducting its own in-house professionaldevelopment conference for all employees.





National fire protection consulting firm with career growth opportunities has immediate need for entry level and senior

level Fire Protection Engineers in their Chicago office. Opportunities also available in San Francisco, San Diego, Los Angeles, Dallas, Las Vegas, Miami, and Washington, DC.

Competitive salary/benefits package. EOE/M/F

Mail or fax résumé to:G. JohnsonSchirmer Engineering Corporation707 Lake Cook Rd.Deerfield, IL 60015-4997Fax: 847.272.2365



San Francisco Fire Department has immediate openings for fire protec-tion engineers. Candidates must possess an engineering degree, six

year’s experience in fire protection analysis, a valid FPE license issued byCalifornia Board of Registration for Professional Engineers or a valid regis-tration from another state that is transferable to California, and a valid driver’s license. Working knowledge of job-related codes and standards(e.g., UBC, UFC, NFPA) desirable.

Salary range is $74,150 - $90,123 annually. For information and/or an application packet, please contact Becky Benoza at 415.558.3610



When it was first pub-lished in 1990, theSFPE Journal of Fire

Protection Engineering was asymbol of our discipline’stransition to a mature engineering profession. Incelebration of SFPE’s 50thAnniversary, the first tenyears of the Journal are nowavailable on CD-ROM fromSFPE. All articles are search-able by title, keyword, andauthor.

SFPE Member Price: $50.00Nonmember Price: $195.00

Look what’s new at SFPE!

Update Your Library with RecentFire Protection Engineering Titles

The First TenYears of theSFPE Journal ofFire ProtectionEngineering inCD Form!


ASCE/SFPE Standard Calculation Methods for Structural Fire Protection, 2000.

In 1997, SFPE began a collaborative activity with the American Society ofCivil Engineers and structural material trade associations to develop a series ofstandards on the calculation of structural fire resistance. The first of these standards, developed under ASCE’s ANSI-approved standards development pro-cedures, is a collection of the various methods now available to predict theperformance of steel, concrete, masonry, and timber structures in response tothe standard fire-resistance test.

SFPE/ASCE Member Price: $28.50 Nonmember Price: $38.00

Enclosure Fire Dynamics, by Bjorn Karlsson and James G.Quintiere, 1999. This 300-page text has been developed to serve as a frame-work and reference for how to estimate the environmental consequences of afire in an enclosure. It is based in part on the work in this area developed byProfessor Magnusson, Lund University, and is expanded upon with new topicsand information from the authors. Ten chapters and three appendices coversuch subjects as fire plumes and flame heights, pressure profiles and vent flows,heat transfer, and computer modeling, as well as suggestions for educators.

SFPE Member Price: $59.95 Nonmember Price: $79.95

Engineering Guide to Performance-Based FireProtection Analysis and Design of Buildings, by the SFPE Task Group on Performance-Based Analysis and Design, 2000.This guide outlines a process for carrying out these designs and is essential foranyone who will apply, approve, or be affected by performance-based codesand standards. Chapters cover such topics as: defining your project scope andidentifying goals; specifying stakeholders and design objectives; developing performance criteria; creating design fire scenarios and trial designs; evaluatingtrial designs; and documentation and specifications. Equip yourself for the coming era of performance-based codes with this unique guide!

SFPE/NFPA Member Price: $46.75 Nonmember Price: $52.00

Fire SprinklerSystems VideoSeries,Protection KnowledgeConcepts, Inc., 1999. Thisfour-tape series is designedfor anyone who designs,specifies, inspects, buys,approves, or maintains thesevital systems. This series cov-ers the basics of fire sprinklersystems, hazard classification,water supplies, and care andmaintenance. The setincludes a set of 40 testquestions and certificates forthose who successfully com-plete the series.

SFPE Member Price: $349.00Nonmember Price: $399.00

Skin Burn Guide. The SFPE Engineering Guide to Predicting 1stand 2nd Degree Skin Burns summarizes accepted calculation methods forpredicting pain and first- and superficial second-degree skin burns fromradiant heat transfer. Calculation methods are presented that range fromsimple algorithms to more detailed calculation methods. For each method,the data requirements, possible data sources, inherent assumptions, andlimitations are presented. The Guide also provides an overview of thephysiology of the skin as it relates to thermal injury.



Sprinkler Hydraulics and What It’s All About,2nd Edition, by Harold S. Wass, Jr. Significantly expanded and updated tothe 1999 edition of NFPA 13©, this comprehensive reference on sprinklerhydraulics contains practical information on all aspects of hydraulic designincluding: sprinkler discharge; friction losses; backflow prevention; relation-ships to water supply; examples of dead-end, loop, grid, and in-rack sprin-kler designs; and inspection and reliability. Written in an easy-to-understandformat by one of the industry’s acknowledged experts on sprinklerhydraulics.


The F.P. Connection

An electronic, full-service fire protection resource Web site.

The FP Connection offers posting of employment opportunities andrésumés of fire protection professionals. If, as a fire protection

service provider or equipment manufacturer, your Web site is difficultto locate using search engines and keywords, let us post your bannerand provide a direct link for use by our visitors who may requireyour services.

Please visit us at www.fpconnect.com or call 724.746.8855.

For posting information, e-mail [email protected].

Fax: 724.746.8856

The F.P. Connection

AN INVITATION TO JOINBENEFITS OF MEMBERSHIP

• Recognition of your professional qualifications by your peers •• Chapter membership •

• Fire Protection Engineering magazine •• SFPE Today, a bi-monthly newsletter •

• The peer-reviewed Journal of Fire Protection Engineering •• Professional Liability and Group Insurance Plans •• Short Courses, Symposia, Tutorials & Seminars •

• Representation with international and U.S.A. engineering communities •

Ask for our membership applicationSociety of Fire Protection Engineers

7315 Wisconsin Avenue, Suite 1225W • Bethesda, MD 20814

301-718-2910www.sfpe.org [email protected]

Society of Fire Protection EngineersA growing association of professionals involved in advancing the

science and practice of fire protection engineering

BRAIN TEASER

A grocer purchased 100 kg of potatoes. When they werepurchased, the moisture content of the potatoes was99.0%. Prior to selling the potatoes the grocer checkedthe moisture content of the potatoes and determinedthat it was now 98.0%. How many kilograms of potatoesdid the grocer now have to sell?

Thanks to Merritt Bauman, P.E., for providing this issue’sBrain Teaser.

Solution to last issue’s Brain Teaser

The demand for a fire protection system is 0.0576 m3/s.For this same system, the design area divided by thedensity is 1,000,000 m × s. What is the system densityand design area?

Solving the second equation for A and substituting it intothe first yields

Therefore, the density, D, is 0.00024 m/s = 0.24 mm/s, andthe design area, A, is 240 m2.

• Ansul .......................................................Page 23

• Central Sprinkler ....................................Page 20

• Chemguard ...............................................Page 8

• Edward Systems.............................Page 26 - 27

• Fenwal....................................Inside Back Cover

• Fike Protection Systems...........................Page 2

• Grinnell Fire Protection Systems ..........Page 32

• Grinnell Supply Sales & Marketing .......................................Back Cover

• Koffel Associates ....................................Page 43

• NOTIFIER Fire Systems .........................Page 16

• NESCO ....................................................Page 51

• Potter Electric .........................................Page 49

• Protection Knowledge Concepts..........Page 29

• Protectowire............................................Page 19

• The RJA Group.....................Inside Front Cover

• Reliable Automatic Sprinkler....Page 11 and 31

• Seabury & Smith ....................................Page 47

• Siemens...................................................Page 37

• Simplex ...................................................Page 41

• TVA Fire & Life Safety, Inc. ..................Page 15

• Viking Corporation ................................Page 39

• Wheelock................................................Page 35

Index of Advertisers

ADT, Inc.Arup Fire Automatic Fire Alarm AssociationBourgeois & Associates, Inc.Central Sprinkler Corp.The Code Consortium, Inc.Code Consultants, Inc.Copper Development AssociationDraka USADuke Engineering and ServicesEdwards Systems TechnologyFactory Mutual Research Corp.Fike CorporationFire Consulting Associates, Inc.Fire Suppression Systems AssociationGage-Babcock & Associates, Inc.Grinnell IncorporatedHarrington Group, Inc.HSB Professional Loss ControlHubbell Industrial ControlsHughes Associates, Inc.Industrial Risk InsurersJames W. Nolan Company (emeritus)Joslyn Clark Controls, Inc.Koffel Associates, Inc.

Marsh Risk ConsultingMountainStar EnterprisesNational Electrical Manufacturers AssociationNational Fire Protection AssociationNational Fire Sprinkler AssociationNuclear Energy InstitutePoole Fire Protection Engineering, Inc.The Protectowire Co., Inc.The Reliable Automatic Sprinkler Co., Inc.Reliable Fire Equipment CompanyRisk Technologies, LLCRolf Jensen & AssociatesSafeway, Inc.Schirmer Engineering CorporationSiemens, Cerberus DivisionSimplex Time Recorder CompanyS.S. Dannaway Associates, Inc.Starwood Hotels and ResortsTVA Fire and Lifesafety, Inc.Tyco Laboratories Asia Pacific Pty., Ltd.Underwriters Laboratories, Inc.Van Rickley & AssociatesVision Systems, Inc.Wheelock, Inc.W.R. Grace Company

C O R P O R AT E 1 0 0 The SFPE Corporate 100 Programwas founded in 1976 to strengthen the relationship between industry and thefire protection engineering community. Membership in the program recog-nizes those who support the objectives of SFPE and have a genuine concernfor the safety of life and property from fire.


D A m s× = 0 0576 3. /

AD m s= ×1 000 000, ,

1 000 000 57 6002 2 2, , , /D m s=

Over the last two years, there havebeen a number of legislative proposalsin the State of Florida that were aimedat modifying the statutory relationshipbetween engineers and technicianswithin the state. Some of these legisla-tive proposals, if implemented, wouldhave reduced the role of the fire pro-tection engineering community in thedesign of fire protection systems andwould have been a disservice to thepublic at large.

The State of Florida presentlyrequires that engineers design sprin-kler systems with 50 or more sprin-klers. Sprinkler systems with fewerthan 50 sprinklers could be designedeither by an engineer or by a techni-cian. This 50-sprinkler threshold isconsistent with the SFPE White Paper1

which states that, for “limited projects,”a technician could execute a designconcept; and 50 sprinklers would bethe definition in the State of Florida ofa “limited project.”

The SFPE White Paper provides thefollowing examples of “design con-cepts” for sprinkler systems: determin-ing the density, water flow, and pres-sure requirements; classification ofhazards and commodities to be pro-tected; preparing a preliminaryhydraulic design; and confirmation ofhydraulic data for the water supply.Similarly, the White Paper lists asexamples of “layout” determining thelayout of risers, cross mains, branchlines, and sprinklers heads; sizing ofpipe; determining hanger locationsand; performing detailed hydraulic cal-culations.

In the State of Florida, “design” hastraditionally been defined to includelocating sprinklers and piping, sizingsprinkler piping, and performinghydraulic calculations in addition todeveloping the sprinkler system designconcepts. This arrangement is consis-

tent with the SFPE White Paper if theengineer has the necessary educationand experience to perform these tasks.In some states, these layout tasks aretypically delegated to technicians whowork for contractors.

Recent legislative changes in Florida,which are scheduled to become effec-tive in June 2001, add a definition of“layout” to the statutes. Their definitionof “layout” is consistent with the defin-ition in the SFPE White Paper.

An owner of a structure has a legaland ethical responsibility to ensurethat the structure is designed, con-structed, and operated in a safe man-ner. Building owners typically engagefire protection engineers, among otherprofessionals, to assist the owner inmeeting this responsibility. In additionto having a contractual responsibilityto an owner, engineers also have aresponsibility to the public at large.

While an engineer’s clients expectengineers to develop designs that arefunctional and meet the client’s goals,the public also has expectations ofengineers. The public places a trust inengineers to develop designs that pro-vide an acceptable level of safety. Tothis end, states license engineers, andprofessional societies publish codes ofethics.

Compliance with a code or standardin itself is not sufficient for public safe-ty. It is also necessary to consider anyunique features or conditions presentand analyze if they warrant considera-tion beyond what a code or standardrequires.

REFERENCES

1 “The Engineer and the Technician:Designing Fire Protection Systems,”Society of Fire Protection Engineers, 1998.Available from http://www.sfpe.org/what-isfpe/final_white_paper.html.

Sprinkler Design and Layout

Morgan J. Hurley, P.E.Technical DirectorSociety of Fire Protection Engineers


Documents

UNIQUE SMOKE...that smoke control requirements have been continually changing. With all of the effort in code writing and research, it is fair to ask, “Has the effort been worth