FIRE protection engineering Winter 2002

FIRE PROTECTIONFIRE PROTECTION

PROVIDING PRACTICE-ORIENTED INFORMATION TO FPEs AND ALL IED PROFESSIONALS

WINTER 2002 Issue No. 13

THE FUTURE OF FIRESIMULATION AT NIST

BRIEFING ON THE WORLD TRADE CENTERATTACKS

THE RIGHT TOOL FORTHE JOB

24

46

39

EVALUATING COMPUTERFIRE MODELS

19

ALSO:

page 9page 9

FireModeling

FireModeling

Offices in major cities throughout theUnited States and strategic businessalliances around the world.

Open yourmind to aworld of careerpossibilities.

No one in the life safety consulting profession offers morediverse and exciting career challenges than The RJA Group.With our reputation as a leader and project involvement allover the world, we’ll challenge your abilities every day whilewe give you the opportunity to grow and develop.

Our disciplines range from fire engineering (Rolf Jensen &Associates, Inc.) and security (Sako & Associates, Inc.) toconstruction management (Fire Protection Management,Inc.) and training (Protection Knowledge Concepts, Inc.). Soyour career path can take any direction you desire, in any ofour prime urban locations throughout the country.

If you have an open mind about how far you’d like to takeyour career, contact us at 888-831-4RJA or visit ourweb site: www.rjagroup.com.

www.sfpe.org 1

3 VIEWPOINT

4 LETTERS TO THE EDITOR

6 FLASHPOINTS

19 EVALUATING COMPUTER FIRE MODELSThere are a growing number of computer-based models being used in support offire safety engineering design and analysis. The author examines various types,summarizes guides that help assess uncertainties, and discusses applications.Marc L. Janssens, Ph.D.

24 THE FUTURE OF FIRE SIMULATION AT NISTHow far will fire modeling have advanced in 10 years? NIST researchers forecastparticular challenges developers will face over the next decade and beyond.Kevin McGrattan, Ph.D., Howard Baum, Ph.D., Ron Rehm, Ph.D., Glenn Forney,Ph.D., and Kuldeep Prasad, Ph.D.

39 THE RIGHT TOOL FOR THE JOBSometimes it makes sense to use sophisticated calculation tools to perform quan-titative fire hazard analyses, and sometimes it doesn’t. What should you consider?Frederick W. Mowrer, Ph.D., P.E.

46 BRIEFING ON THE WORLD TRADE CENTER ATTACKSBased on a briefing paper that was prepared two weeks after the World TradeCenter attacks, this article describes the many factors that need to be consideredwhen evaluating how a building responds to such an extreme event. Preventionmeasures are also highlighted.Extreme Events Mitigation Task Force, Arup

54 PRODUCTS/LITERATURE

56 SFPE RESOURCES

58 BRAINTEASER/CORPORATE 100/AD INDEX

60 FROM THE TECHNICAL DIRECTORLearning from TragedyMorgan Hurley, P.E.

Cover illustration by Rick Fischer/Masterfile

Online versions of all articles can be accessed at www.sfpe.org.

Invitation to Submit Articles: For information on article submission to Fire Protection Engineering, go to http://www.sfpe.org/publications/invitation.html.

contents W I N T E R 2 0 0 2

9COVER STORYFROM PHLOGISTON TO COMPUTATIONAL FLUID DYNAMICSAll of combustion science rests at least in part on Lavoisier’s discovery in the late18th century that combustion involves reaction with the element oxygen. This articlereviews the steps that brought fire dynamics from there to the models of today.Harold E. Nelson, P.E.

Subscription and address change correspondence should be sent to: Fire Protection Engineering, Penton Media, Inc., 1300 East 9th Street, Cleveland, OH 44114 USA. Tel: 216.931.9159. Fax: 216.696.7668.e-Mail: [email protected].

Copyright © 2002, Society of Fire Protection Engineers. All rights reserved.

FIRE PROTECTIONFIRE PROTECTION

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 Engineering Corporation

Don Bathurst, P.E.

Russell P. Fleming, P.E., National Fire Sprinkler Association

Douglas P. Forsman, Firescope Mid-America

Morgan J. Hurley, P.E., Society of Fire Protection Engineers

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

Jane I. Lataille, P.E., Los Alamos National Laboratory

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 Building Officials

Regional EditorsU.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

Thomas L. Caisse, P.E., C.S.P., Robert M. Currey &Associates, Inc.

U.S. SOUTHEAST

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

U.S. WEST COAST

Marsha Savin, Gage-Babcock & Associates, Inc.

ASIA

Peter Bressington, P.Eng., Arup Fire

AUSTRALIA

Richard Custer, Arup Fire

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

Personnel

EXECUTIVE DIRECTOR, SFPEKathleen H. Almand, P.E.,

TECHNICAL EDITOR

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

PUBLISHER

Terry Tanker, Penton Media, Inc.

ASSOCIATE PUBLISHER

Joe Pulizzi, Custom Media Group, Penton Media, Inc.

MANAGING EDITOR

Angela Vannucci, Custom Media Group, Penton Media, Inc.

ART DIRECTOR

Pat Lang, Custom Media Group, Penton Media, Inc.

COVER DESIGN

Dave Bosak, Custom Media Group, Penton Media, Inc.

Naturally...Nontoxic. Noncorrosive. Unregulated.Unrestricted. Chemical-free.Safe to breathe.

Call for your FREE brochure:

800-346-3626715-735-7411

www.ansul.com

By Joseph M. Fleming

Performance-based designs appear torely on three sources of information:

1. A Set of Objectives. (This could bea Performance-Based Code.)

2. A Design Guide. (These are gener-al rules on documentation andmethodology.)

3. Reference Material. (For example,an engineering handbook.)

The problem with the use of thismaterial is that it allows too much flexi-bility in the selection of critical “input”items such as performance criteria andassumptions about occupant behavior. Acode official trying to use these docu-ments to ensure the safe design ofbuildings is analogous to a police officertrying to enforce a safe society by usingbooks on philosophy and theology.These books may contain valuableinformation as to how one should con-duct the affairs but are almost useless asa set of enforceable rules.

These input items could substantiallyalter the output of computer modelsand algorithms. How is a code officialsupposed to ensure that the output of adesign process is safe when they haveno method to measure the validity ofthe inputs? This is not a problem withprescriptive codes because the output is“prescribed.” With performance-baseddesigns, the output is not “prescribed.”To ensure the safety of future occu-pants in performance-base-designedbuildings, I believe that the inputs mustbe “prescribed.” Failure to prescribe theinputs could lead to either unsafe oroverly expensive designs.

As Fire Marshal of the City of Boston,

I have reviewed many “deterministic”timed egress analyses. A key componentof these analyses is the selection of ten-ability criteria. This selection couldincrease or decrease the time availablefor egress by several minutes dependingon the fire scenario. Unfortunately, manyguidelines, such as the SFPE Handbook,provide multiple choices for acceptabletenability criteria or “factors to consider”when deciding what tenability limit toutilize. Even the SFPE Engineering Guideto Performance-Based Fire ProtectionAnalysis and Design of Buildings onlystates, “Visibility through smoke mightaffect the occupants’ ability to safely exita building. The factors that affect visibili-ty include the amount of particulate inthe path of vision and the physiologicaleffect on the eyes. Low light levels mightalso affect occupants’ ability to egress.”Utilizing these references would providethe designer has a lot of leeway in thismatter. If a code official wants 0.25OD/m to be used, as the criteria for visi-bility and a designer wants to use 0.5OD/m, how do we “referee” the dis-agreement?

The proper use of safety factors isanother area where there seems to beonly general guidelines as opposed torules. It does not appear that any algo-rithm or model that is currently utilizedis 100% accurate. As a consequence, itwould seem prudent to utilize safety fac-tors to offset the uncertainty. In actuali-ty, many designs that the Boston FireDepartment has seen have not utilizedany Safety Factor. Others have used aSafety Factor of 1.5. All of these selec-tions were based on “engineering judg-ment.” In the SFPE Handbook, JakePauls recommends that “...in relation to

the ‘Life Safety Evaluation,’ there shouldbe a factor of safety, especially in viewof the incomplete technical grasp ofboth egress and fire issues at the pre-sent. For example, in a conservativeapproach, the ‘time available’ should beat least twice as long as the ‘timerequired.’” Despite this well-documentedrecommendation, one of the engineerswho participated in a design wereviewed stated, “Jake Paul’s method(SFPE Handbook) of doubling the occu-pant egress time is not commonlyaccepted or used for fire engineeringanalysis. For almost any engineeringanalysis, you could find someone withan opposite analysis or result.” Do Ihave the right as a code official to rejectthis design based on this comment? Theengineer felt that I did not since his“engineering judgment” justified theopinion. But isn’t this type of thinkingcircular reasoning? An engineer’s opin-ion regarding the proper choice of safety factors has to be justified by more than just the engineer’s opinion.

SOLUTION

It seems reasonable to give designersthe freedom to choose differentdesigns, i.e., different outputs, toachieve the same goal. At the sametime, it does not seem reasonable toallow designers the freedom to pickany criteria and make any assumptionthat they can find a reference for in atechnical handbook or peer-reviewedjournal. Until a set of prescriptive inputscan be published, it seems prudent toalways require the following:

1. The logic justifying the selection ofcriteria or assumptions should beexplicit. Whenever possible itshould be more than “engineeringjudgment.”

2. An independent peer reviewshould be provided to the codeofficial.

3. All references used to justifyassumptions or criteria should bemade available to the code official.

Joseph M. Fleming is with the BostonFire Department.

WINTER 2002 www.sfpe.org 3

viewpoint

PERFORMANCE-BASED DESIGNS

THE LACK OF PRESCRIPTIVE INPUTS PROBLEM IN

In Response to “Lessons Learnedfrom a Carbon Dioxide SystemAccident,” Issue No. 12.

Having been at the INEEL for abouttwo years, talked and worked with per-sonnel involved or familiar with theincident, and toured the area where thetragic event occurred, I have learned afew “lessons” that were not focused onin the article. Here they are:1. Carbon dioxide (CO2) at minimum

design concentrations required tosuppress fires is lethal to humans.Given its inherent hazard, it shouldnot be used in areas subject to occu-pancy – EXCEPT when the risk offire is documented to be greater thanthe risk to personnel AND there areno viable suppression alternatives.When considering the recentadvances our industry has made infire suppression agents (notablywater mist and Halon alternatives),no situations which meet the excep-tion condition come to mind.

2. When it is evident that adherence toa national consensus standard is notsufficient to preclude a fatal accident,that standard must be evaluated todetermine its adequacy. Immediatelyfollowing the tragic event, NFPA 12,Standard on Carbon DioxideExtinguishing Systems (2000 edition)1,was changed to require the additionof an isolation valve to enable thesystem to be physically isolated.Additionally, the appendix guidancein NFPA 72 (1999 edition)2 was modi-fied to suggest the use of deviceswhich detect the flow of agent and

initiate alarm notification appliances.Yet the question remains: With theincorporation of these changes, arethe standards adequate to permit theuse of carbon dioxide in areas sub-ject to occupancy? If they are not,then more evaluation is required withthe intention of revising the standardor restricting the use of CO2 in lethalconcentrations to areas not subject tooccupancy.

3. The specific mechanism that causedthe CO2 system to discharge withoutwarning has been attributed to adesign defect in the manufacturer’sfire panel [Lockheed Martin, 1999].3

When significant defects are identi-fied, many manufacturers issue recallnotices to ensure that not only is theconsumer made aware of the defect,but also that the defect is “correct-ed.” Standards dealing with fire sup-pression agent design concentra-tions, which, if exceeded, could behazardous to occupants, shouldinclude provisions that address prod-uct defects. These provisions shouldrequire manufacturers to immediate-ly notify users of the defective prod-uct and initiate recalls when it isdetermined that the defect is suffi-ciently hazardous to endangerhuman life.

4. Standards dealing with fire suppres-sion agent design concentrations,which, if exceeded, could be haz-ardous to occupants, should includeprovisions that address system config-urations. System configurations wherethe failure of a single componentcould result in hazardous concentra-tions to occupants should be prohib-ited (some “approved” gaseous agentreserve manifolds would not meetthis provision).A final thought. As fire protection

engineers, we need to be continuallyscrutinizing our engineering approachto the fire problem. To quote one of myengineers, just because “we can” dosomething doesn’t mean “we should.”

Stephen Thorne, P.E.INEEL Fire MarshalThe opinions expressed above are my

own and do not necessarily reflectthose of my employer.

References:

1 NFPA 12, Standard on Carbon DioxideExtinguishing Systems, 2000 Edition. NationalFire Protection Association, Quincy, MA,2000.

2 NFPA 72, National Fire Alarm Code, 1999Edition. National Fire Protection Association,Quincy, MA, 1999.

3 Lockheed Martin, 1999. Identification of theSpecific Mechanism by which the CO2

System in Building TRA-648 AccidentallyDischarged.

Ithink the article “Lessons Learnedfrom a Carbon Dioxide System

Accident” (Issue No. 12) could havegone even further and stressed theimportance of simplicity over complex-ity and of mechanical over electronicschemes where robustness is of toppriority. Designers of water-basedextinguishment systems have knownthe value of indicating valves for overa century, and it is still the mostrobust way of shutting off a systemand having its status be unambiguous-ly indicated. Designers, OSHA, and theNFPA 12 committee should considerthe wisdom that a heat-seeking missilemay prove not to be more reliablethan a fly swatter in swatting flies. Getmechanical indicating valves in therenow, and don’t accept non-robust“modern” substitutions!

Vytenis Babrauskas, Ph.D.President, Fire Science and Technology, Inc.

In the Fall 2001 issue (No. 12) ofFire Protection Engineering, the arti-

cle “UL 2360, A New Test for WetBench Plastics”1 by Jane Lataille con-tains a number of inaccuracies andomissions that could misinform read-ers about the state of flammabilitytesting for plastics used in semicon-ductor clean rooms.

Ms. Lataille notes in the article thatFactory Mutual Research introducedfire tests to determine the acceptabilityof plastics for use in the semiconduc-tor industry. These tests, known as theFactory Mutual Research Test Standard49102 (FM4910), were released in 1997and included both small-scale testsand an intermediate-scale (parallel

4 Fire Protection Engineering NUMBER 13

letters to the editor

panel) test adopted in concept severalyears later by UnderwritersLaboratories (UL 2360)3 as discussed inthe article. However, Ms. Lataille inac-curately characterizes Factory MutualResearch’s small-scale test as onewhich plastics manufacturers are notable to run “because the nonstandardtest apparatus is not readily availableand the test results were not repro-ducible.” She adds, “In addition, thetest is complex and expensive.”

As I will explain below, in sharp con-trast, FM4910 provides a reproducible,standardized, technically sound, anddocumented method of selecting non-fire propagating materials for use insemiconductor clean rooms.

The small-scale test apparatus gen-erating the results alleged by Ms.Lataille to be nonreproducible is theapparatus known to many as theFMRC Flammability Apparatus. Thisapparatus has been used to generatethe data in the SFPE Handbook of FireProtection Engineering, Section3/Chapter 4, “Generation of Heat andChemical Compounds in Fires.”4 It isfair to say that this is one of the mostwidely used sources for flammabilityproperties of materials. Moreover, theapparatus is considered of sufficienttechnical soundness to be recognizedas the ASTM E-2058 Fire PropagationApparatus5 and recognized in theNFPA 287, Standard Test Methods forMeasurement of Flammability ofMaterials in Cleanrooms Using a FirePropagation Apparatus (FPA).6 Theapparatus also is available in a com-mercial version from Fire TestingTechnology Ltd. (FTT) and is used byFactory Mutual Research’s Approvalsdivision for its certification testing.

Although the article asserts to thecontrary, it is also important to pointout that test complexity and expenseare not an issue compared to UL 2360.3

The parallel panel test in UL 2360,3

adapted from the Factory MutualResearch test, is virtually the same asthat in FM4910,2 and the tests conduct-ed with the cone calorimeter used inthe UL 23601 are similar in manyrespects to the small-scale ASTM E-20585 tests in FM4910.2 There are, how-ever, important technical differences

between the small-scale test method-ologies.

The FM49102 small-scale test uses a“vertical” sample of plastic material tocharacterize the heat release rate of amaterial under conditions in which firepropagation is supported by the heatflux generated by the material’s ownflame. Forty-percent oxygen is used tosimulate the high flame heat fluxesthat are developed in large-scale tests.8

In contrast, in the UL 23603 small-scaletest, the heat release rate is obtained ina test with a “horizontal” sample ofplastic material in normal air with afixed imposed external heat flux. Thisfixed heat flux is not necessarily char-acteristic of the heat flux that wouldbe generated by the material’s ownflame in a large-scale fire. Thus, thegenerated heat release rate may not beappropriate.

This may be seen in the data present-ed by Ms. Lataille. Table 11 indicatesthat clear PVC, using the UL classifica-tion scheme of Table 4,1 is a limitedpropagating material (i.e., flame propa-gation in the parallel panel is less than8 ft.). Indeed, clear PVC just misses by0.2 ft being classified by the UL parallelpanel as a nonpropagating material.However, the UL 2360 small-scale testclassifies this material as a propagatingmaterial, with the UL index being morethan double the requirements to beconsidered as nonpropagating.

Testing of this material at FactoryMutual Research has indicated that thisinconsistency would not occur underFM4910, as both the parallel panel testand the FM4910 small-scale test wouldclassify the material as nonpropagat-ing. Similar results have been observedfor other materials and reported byFactory Mutual Research.9

Another correction to note in thearticle is the explanation for the equa-tion for the Thermal ResponseParameter (TRP). It is not derived fromequations for the flame height andflame propagation rate. Rather the TRPis derived from the temperatureresponse of a material under thermallythick heating conditions. For furtherdetails, please refer to reference 4.

Of final clarification, the parallelpanel test mentioned frequently in the

article, without explanation, wasdeveloped by Factory Mutual Researchas an intermediate-scale test, whichwould provide an appropriate simula-tion of the fire hazard in semiconduc-tor clean rooms. This was demonstrat-ed in full-scale wet bench tests atFactory Mutual Research.10

Robert G. Bill, Jr., Ph.D.Assistant Vice President and Director,Materials ResearchFactory Mutual Research

References:

1. Lataillle, J., “UL 2360, A New Test for WetBench Plastics,” Fire Protection Engineering,No. 12, 30-33, Fall 2001

2. Test Standard, “FMRC Clean Room MaterialsFlammability Test Protocol,” Class NumberFMRC 4910, Factory Mutual ResearchCorporation, Norwood, MA, 1997.

3. Standard for Safety, “Test Methods forDetermining the Combustibility of PlasticsUsed in Semi-Conductor Tool Construction,”UL 2360, 1st Edition, UnderwritersLaboratories Inc., Northbrook, IL, May 2000.

4. Tewarson, A.,”Generation of Heat andChemical Compounds in Fire,” SFPEHandbook of Fire Protection Engineering,Section 3/Chapter 4, 2nd edition, NFPAPublications, Quincy, MA, 1995.

5. ASTM E2058-00, Standard Methods of Testfor Measurement of Synthetic PolymericMaterial Flammability Using a FirePropagation Apparatus (FPA). TheAmerican Soc. For Testing and Materials,West Conshohocken, PA, 2000.

6. NFPA 287, Standard Test Methods forMeasurement of Flammability of Materials inCleanrooms Using a Fire PropagationApparatus (FPA), National Fire ProtectionAssociation, Quincy, MA, 2001.

7. ASTM E-1354, Standard Test Method forHeat and Visible Smoke Release Rates forMaterials and Products Using an OxygenConsumption Calorimeter, The AmericanSoc. For Testing and Materials, WestConshohocken, PA, 1999.

8. Tewarson, A., Khan, M., Wu, P., and Bill, R.,“Flammability Evaluation of Clean RoomPolymeric Materials for the SemiconductorIndustry,” Fire & Materials, 25, 31-42, 2001.

9. Bill, R., Wu, P., Tewarson, A., and Khan, M.,“Characterizing the Vertical Fire PropagationPropensity of Materials Through Small-ScaleFlammability Testing,” Proc. Fire & Materials2001, 209-219, 7th International Conf.,Interscience, London, January 22-24, 2001.

10.Wu, P., “Parallel Panel Fire Tests forFlammability Assessment,” Proc. 8thInternational Fire Science & EngineeringConf., Edinburgh, Scotland, 605-614,Interscience, London, June 29 – July 1, 1999.


Spivak Receives W.T. Cavanaugh Award

W. CONSHOHOCKEN, PA – The2001 William T. Cavanaugh MemorialAward was recently presented to Dr.Steven M. Spivak, professor emeritusand immediate past-chairperson of theDepartment of Fire ProtectionEngineering at the University ofMaryland in College Park, MD.Established in 1987 in memory of thelate William T. Cavanaugh, ASTM chiefexecutive officer from 1970 until hisdeath in 1985, the award is the highestrecognition ASTM gives to a person orpersons of widely recognized eminencein the voluntary standards system.

Dr. Spivak has been on the faculty ofthe University of Maryland since 1970;he has been with the Department ofFire Protection Engineering since 1992.The prior 20 years, he worked in textilefire research, fiber science, and poly-mer/chemical engineering. He has tak-en three sabbaticals involving standardswork, one at the General ServicesAdministration and two at the NationalInstitute of Standards and Technology.

Free, Revised ESFRSprinkler Datasheet

JOHNSTON, RI – FM Global, thedeveloper of Early Suppression, Fast-Response (ESFR) sprinkler technologyfor warehouse fire protection – particu-larly high rack storage – has released anewly revised, easier-to-understand ver-sion of its Property Loss PreventionData Sheet 2-2 Installation Guidelinesfor Early Suppression, Fast-ResponseSprinklers. FM Global is making thisdata sheet, written for fire protectionsystem designers and installers, avail-able for free to address all-too-commonmisunderstandings about ESFR technol-ogy and the increased international useof such sprinklers.

The revised and reorganized instruc-tions contain more in-depth explana-tions and illustrations for proper instal-lation of ESFR sprinklers. The datasheet also emphasizes the challengesthat can impact the effectiveness ofESFR sprinklers including roof height,structural members, lighting, HVAC,storage heights, and certain types ofcommodities.

For a free copy, visit www.fmglobal.com/esfr, or call FM Global CustomerServices at 781.255.6681.

New Award forEnvironmental

Awareness in FireProtection

ASHLAND, MA –Kidde plc, in associ-ation with the U.S.EnvironmentalProtection Agency(EPA), has createdthe annual DavidBall Award for Fireand the Environment. The award willrecognize the significant contributionmade by an individual, team, or organi-zation to furthering the understandingof the impact of fire or developingimproved methods of fire control thatbenefit or protect the environment. Thefirst award will be presented at theFourth Earth Technologies Forum inWashington, DC, March 25-27, 2002.

The award is named in honor of Dr.David Ball, a 25-year employee ofKidde, who died in July 2001. Davidwas Kidde’s research manager and waswell known in the global fire protectionindustry as a passionate advocate ofenvironmental awareness.

More more information, contactStephen Summerill at 304.728.3489([email protected]).


flashpoints

fire protection industry news

HEADQUARTERSTERRY TANKER Publisher

1300 East 9th StreetCleveland, OH 44114-1503216.696.7000, ext. 9721fax [email protected]

NORTHEASTMARK D. FORKER District Manager

485 Devon Park Drive, Suite 107Wayne, PA 19087610.995.0202fax [email protected]

NORTH CENTRALJOE DAHLHEIMER District Manager


CENTRALAMY COLLINS District Manager

11900 Wayzata Blvd., Suite 216KMinnetonka, MN 55305952.545.2054fax [email protected]

SOUTHEASTDEBBIE ISGRO District Manager

3312 Piedmont Road, Suite 250Atlanta, GA 30305404.237.5731fax [email protected]

WEST COASTLORA ATHERTON ALBAINY Senior Regional Manager


Sales OfficesSales Offices FIRE PROTECTION

Disability Income Plan

Term Life Plan

High-Limit Accident Plan

Member Assistance Plan

Catastrophe Major Medical Plan

Medicare Supplement Plans

Comprehensive HealthCare Plan

Dental Discount Plan

Cancer Expense Plan

for What Keeps YOU Up at NightCustom-Tailored Protection from SFPE:

Thomas Edison, who invented the

light bulb, was afraid of the dark.

For FREE information about any of the SFPE-sponsored Plans, call toll free 1-800-424-9883.

You’re an individual, with your ownhopes, dreams ... and worries: Wouldyour spouse be able to afford your family’s current lifestyle if something happened to you? Will you be around toput your kids through college? If not,what then?

Unfortunately, it’s not always possible to—like Thomas Edison—invent your own instant solutions to what keeps you up at night.

SFPE can help. As an SFPE member, you can choose from nine qualitySFPE-sponsored insurance plans to helpsafeguard you and your family.

Whether it’s solid Member Assistanceprotection for you, Term Life Insurancefor yourself and your spouse, orComprehensive HealthCare coverage foryour whole family, the SFPE-sponsoredProgram is your one-stop source for custom-tailored protection for what matters most in your life. And each planis available to members like you at SFPE group-negotiated rates that fit your budget.

Rest easy.... The SFPE-sponsoredInsurance Program looks out for you.

Sponsored by:

Call 1-800-424-9883 for FREE information featuring costs, eligibility, renewability, limitationsand exclusions of the following SFPE-sponsored Plans:

The Term Life, High-Limit Accident and Comprehensive HealthCare Insurance Plans are underwritten by New York Life Insurance Company, 51 Madison Avenue, New York, NY 10010. The CatastropheMajor Medical and Dental Discount Plans are underwritten by The United States Life Insurance Company in the City of New York, Member American General Financial Group. The Member AssistancePlan is underwritten by Unum Life Insurance Company of America. The Medicare Supplement and Cancer Expense Insurance Plans are underwritten by Monumental Life Insurance Company.

677-01


By Harold E. Nelson, P.E.

During the last decade, firemodels have had a majorinfluence on applied fire pro-

tection engineering. Modern compart-ment fire models solve fire dynamicsequations that have been only veryrecently seriously investigated. Theinformation developed as a result ofthese investigations has helped transferfire dynamics methods from theresearch realm to field application. Fireprotection engineering applicationsnow range from straightforward alge-braic equations, such as those fordescribing fire plumes, to complexcomputer models that solve complexsimultaneous fire dynamics equationsiteratively.

Over the past 50 years, fire dynamicsinformation was derived from scientificprinciples abstracted from other disci-plines. The basis for the fire dynamicsin compartment fire models lies in thephysics relationships inherent in com-bustion, such as heat balance and con-servation equations.

All of combustion science rests atleast in part on Lavoisier’s discovery inthe late 18th century that combustioninvolves reaction with the element oxy-gen. This article reviews the steps thatbrought fire dynamics from there to themodels of today.

THE DISCOVERY OF OXYGEN –THE FALL OF THE PHLOGISTONTHEORY

In ancient times, fire was consideredone of the four elements, the otherthree being earth, air, and water. From

PHLOGISTON toCOMPUTATIONAL

FLUIDDYNAMICS

From


the time of the early Greek philoso-phers until the late 18th century, it washeld that the essence of fire is a sub-stance called phlogiston. The phlogis-ton theory explained most combustionreactions, at least to the degree of mea-surement available to the experimentersof that day.

At the time of the French Revolution,Antoine Lavoisier, French philosopherand chemist, demonstrated the phlogis-ton theory as inaccurate. By carefulexperiment and measurement, Lavoisiershowed for the phlogiston theory to becorrect, phlogiston had to have a nega-tive mass, an untenable position. At thesame time, the English scientist JosephPriestly isolated what he called“dephlogisticaded air.” Lavoisierdemonstrated that Priestly’s finding wasa substance, which he labeled oxygen.Lavoisier further demonstrated that oxy-gen was part of a combustion reactionand added weight to the product.

Also in the last decade of the 18thcentury, another English scientist,Benjamin Thompson, demonstrated bystudying experiments that no masschanges are associated with tempera-ture change and that heat is a manifes-tation of motion. These discoveriesconstituted the final criticism anddestruction of the phlogiston theory.

FARADAY’S CHRISTMASLECTURES

In the 1840s, the famous scientistMichael Faraday in his Christmas lec-tures at the Royal Institution of GreatBritain presented one of the earliestdemonstrations of the interaction of ele-ments in fire dynamics. Faraday usedthe burning of a candle to demonstratemany of the phenomena involved incombustion. The candle lecture includ-ed demonstrations of radiant energy,liquidation of the candle wax, and cap-illary action.

THE START OF MODERN FIREDYNAMICS

Early Investigations of FirePlumes. Modern fire dynamics, howev-er, finds its initiation in and followingthe World War II period in the early

1940s. One of Britain’s leading physi-cists, Sir Geoffrey Taylor of CambridgeUniversity, used his knowledge of fluiddynamics to develop a system for clear-ing fog at wartime airfields. Gasolinefire plumes rising out of ditches besiderunways were used to entrain andremove fog sufficiently for returning air-craft to find the fields and land.

Critical to quantifying the fire plumeis an analytical understanding of theamount of air (or other surroundinggases) entrained into the rising fireplume. The initial work of the defini-tion of fire plumes, including theentrainment coefficient, appears to havebeen published by W. Schmidt inGermany in 1941. Taylor analyzedSchmidt’s work and a description of athermally driven plume was preparedfor the U.S. Government under the titleof “Dynamics of a Hot Gas Rising inAir” in 1945. The first major publicationof plume theory impacting the analysisof accidental fires, however, was thepaper titled “Turbulent GravitationalConvection from Maintained andInstantaneous Sources.” by B. R.Morton, Taylor, and J. S. Turner, all ofCambridge University, in 1956. Thisdescribed the first heat-driven plumeequation that included entrainment,plume velocity, and plume temperature.

While most current fire models andcalculations do not use the earlyMorton, Taylor, and Turner equations,all of the current methodologies for cal-culating plume properties are, at leastto some extent, descendants of theMorton, Taylor, and Turner work. Thework of Morton, Taylor, and Turnerwas a turning point that has becomethe basis for the modern zone model.

A progression of investigators carriedthe work forward increasing the levelof understanding. In 1960, Sizuo Yokoiof the Building Research Institute ofJapan examined both flame heights andthe hot gas currents of a plume flowingfrom a window, exposing the floorabove.

Bernard McCaffery of the Center forFire Research, in his 1979 report onexperimental results, examined entrain-ment in the area of the flame as well asthe rate of entrainment in the portion ofthe plume above the flame.

In the mid 1970s, Edward Zukoski,along with his colleague T. Kubota andhis doctoral student B. Cetegen at theCalifornia Institute of Technology, pur-sued the entrainment phenomena fur-ther and developed one of the earliestplume equations. Their plume equationwas formally published in 1981 and hasbeen used in fire models such as ASET,FIRE SIMULATOR, and CCFM.

Gunner Heskestad of Factory MutualResearch Corporation developed theconcept of virtual source to account forthe fact that most accidental fires arenot point source in nature. Heskestadthen developed a coordinated set ofalgebraic equations on fire plumes forengineering use. His work, first pub-lished in 1984, is accessible to engi-neers as the chapter on fire plumes inthe SFPE Handbook.

Others, including Philip Thomas atthe UK Fire Research Station and KunioKawagoe at the Building ResearchInstitute of Japan, also examined andimproved the fire plume equations.Craig Beyler, then a visiting scientist atthe UK Fire Research Station, examined14 fire plume and ceiling jet equationsin 1986.

IMPACT OF RESEARCHINSTITUTIONS

The development of the understand-ing of the dynamics of fire has strongties to a number of institutionalizedprograms. It’s worthy to briefly discusssome of the major participants.

The UK Fire Research Station. Inthe UK, both the fire impact ofwartime attacks and the use of alterna-tive materials, some of which werevery combustible (such as low-densityfiberboard linings), produced fires thatexceeded expectations. The Britishcame to realize that there was a widerange of speeds of involvement andextent of fire development, spread offire between buildings, and otheraspects of fire danger permitted by thethen-current bylaws and regulations.During the war, the British sciencecommunity had worked diligently onfire prevention and control, as well ason using fire as a weapon. In 1947,


one of the most important modern fireresearch and development groups wasassembled: The Fire Research Station.Also, the Joint Fire ResearchOrganization, created in 1946, enabledfunding from both the governmentand private interests, particularly thosein the fire insurance industry.

The British fire research programshave long enjoyed a well-deservedleadership position as developers offundamental and applied fire scienceand engineering. Examples include theanalysis of radiation between buildingsas a function of openings, separations,and fire condition by Margaret Law;the venting of hot gases from largespace fires by Peter Hinkley andThomas; and the study of various heatrelease and heat transfer means byRichard Chitty and Geoffrey Cox. Firedynamics developments at the FireResearch Station in recent years haveconcentrated on the advancement ofCFD modeling.

Building Research Institute ofJapan – Japanese Fire ResearchInstitute. In Japan, the immediateproblem in the first years following thewar was the reusability of structuresthat had been damaged during the aircampaign. The program initiallyaddressed structural problems, bothwith and without earthquake consider-ations, but soon extended to the ques-tion of what fire would do in thesebuildings, and which buildings wouldbe safe for continued use. TheBuilding Research Institute of Japancreated a fire research section, andabout the same time, a separate FireResearch Institute was established inThe Ministry of Home Affairs.

There are many contributions to firesafety developed in the Japanese fireresearch programs. In terms of firedynamics and modeling, examples ofimportant work at the BuildingResearch Institute includes that ofYokoi and Kawagoe.

Yokoi was the first director of fireresearch at the Building ResearchInstitute. Under his direction in the1950s, important work was developedon flame height and fire plumes, par-ticularly those projected from build-

ings. Yokoi’s plume research has beenused to develop the spandrel panelrequirements for high-rise buildings inJapan.

Also in the late 1950s, Kawagoe andhis colleagues observed that early inthe development of the room fire, areasonably clean-cut smoke layer – orzone – is established. Kawagoe furtherestablished that in the simple situationwhere there is a compartment with asingle opening and no forced air, theavailable air for combustion is limitedto that which can be drawn in theroom through the opening to replacethe mass expelled through the sameopening. The rate of air intake wasestablished to be a function of the areaof the opening times the square rootof its height. Kawagoe mathematicallyexpressed these phenomena in what isfrequently referred to as Kawagoe’sequation. Kawagoe’s observations andcalculations are key elements in zonemodeling.

There is an anecdote of Kawagoevisiting Zukoski at Cal Tech and tellinghim about the layer interface phenom-ena, which was then contrary to manycommonly held beliefs. Prior toKawagoe’s work, many researchersbelieved that the turbulence in theroom would cause a mixing of all thegases into one zone referred to as awell-mixed reactor. According to theanecdote, Zukoski challengedKawagoe to prove his point and theywent to a nearby mechanics laboratorywhere they initiated the phenomena ina liquid tank. They pumped a lightdensity liquid into a higher density liq-uid as an analogue to fire and clearlysaw the layer development.Understanding these two zone phe-nomena was critical to the concept ofzone models.

RESEARCH ON FIRE DYNAMICSAND MODELING IN THE UNITEDSTATES

Office of Civil Defense FundsMajor Fire Research Program. Inthe 1950s and 1960s in the UnitedStates, the impetus for the OCD fireresearch program, directed by JamesKerr, was the Cold War. The OCD pro-

gram was primarily focused on civildefense, particularly the potentialimpact of nuclear attack. The OCDprograms involved an expenditure ofapproximately 1 million dollars annu-ally on fire research, the vast majorityof which was directed at large-scalenuclear attack-initiated fires. However,work relative to fire dynamics was partof this effort. Extensive work wasdone at the Naval RadiologicalDefense Laboratory on understandingthe ignitability of materials from radi-ant energy sources. Much of this infor-mation was published in the book TheEffects of Nuclear Weapons by SamuelGlasstone. Extensive work onflashover was undertaken by theArmor Research Foundation (nowIllinois Institute of TechnologyResearch Institute). The principleinvestigator was Thomas Waterman.Professor Willis Labes of the IITDepartment of Fire ProtectionEngineering performed investigationsof radiant fire spread as part of theOCD effort.

OCD Sponsors National Academyof Science Committee on FireResearch. In the early 1950s, OCDsponsored the establishment of theNational Science FoundationCommittee on Fire Research. Thiscommittee was initially headed byHoyt Hottel of Massachusetts Instituteof Technology and subsequently byHoward Emmons of Harvard. In theearly 1970s, the committee sponsoredthe publication of Fire Research –Abstracts and Review, essentially thefirst U.S. science-based fire journal.The Fire Research – Abstracts andReview was published 3 times a yearfor approximately 12 years from theearly 1960s to the 1970s. RobertFristrom of Johns Hopkins AppliedPhysics Laboratory was editor. Theexchange of information provided bythis journal was very important to thespread of knowledge among the vari-ous groups interested in fire research.

Woods Hole ConferenceAppraises Application of Science toFire Problem. In 1962, the NASCommittee on Fire Research autho-

Screen shown actual size

s

Fire Safety

Introducing the world’s first BIG-SCREEN fire detection system.FireFinder, the fastest intelligent fire detection system

with the biggest, easiest display screen.

It makes all the difference in an emergency. The big 6” display

on the FireFinder panel is the easiest for firefighters to under-

stand when trying to locate a fire. Because it has an unrivaled

capacity for hundreds of easy-to-read, large-font text characters,

Hazmat icons, NFPA fire safety symbols and graphic maps. All of

these features work to help fire personnel identify fire location,

hazards and occupants faster than ever. Unlike small screens with

room only for vague, abbreviated messages which can delay and

confuse fire personnel.

The FireFinder system also responds with unrivaled speed,

detecting fire and notifying occupants in under 3 seconds. And it’s the simplest intelligent system

to install and maintain. The display lets operators view the entire system and locate specific alarms

with a simple “user prompting” sequence of lighted buttons. Field programmable from a touch

screen, it makes system control and changes simple for all users. For installers, revolutionary wiring

technology can reduce wiring errors and cut installation time by up to 25%*.

What’s more, the FireFinder offers a very-low life-cycle cost. Learn how this big-screen system can

bring you big savings and big peace-of-mind. Call Siemens Building Technologies, Fire Safety Division

at 973-593-2600 or visit cerbpyro.com

* Depending on type and size of facility. ©2002 SIEMENS


rized the Woods Hole Study in WoodsHole, MA, which brought together arepresentation of a cross-section of sci-ences that potentially could contributeto solutions for fire safety problems.Leaders in the applied fire protectionarena were also included. The confer-ence met for four weeks with the firstweek’s agenda consisting of externalspeakers on the sciences as they couldbe applied to fire problems. The majorrecommendation of the conferencewas the need to establish a nationalfire research program within theFederal government. This concept,however, was not widely acceptedoutside of the science community.Leaders of fire protection and preven-tion organizations, such as the NFPAand the fire insurance industry, strong-ly opposed a Federal research pro-gram. Their position was that therewas no need for new information,rather a need for applying the knowl-edge we currently had and that theFederal government should notimpose itself in this arena. Fireresearch and safety bills proposed in1963 and 1964 were never enacted.

The Woods Hole Conference, how-ever, had an important impact in thatit generated a consensus in the scien-tific community that there was apotential to mathematically describe,i.e., model, fire phenomena and theimpact of the fire on the environmentwithin a building.

THE RANN PROGRAM TAKESPLACE OF OCD FIRE RESEARCHPROGRAM

In the mid-1960s, the U.S. govern-ment established an administrative pol-icy that the military should shed itselfof research that did not directly impactmilitary objectives. A research programentitled “Research Applied to NationalNeeds” (RANN) was established in theNational Science Foundation, and thefire safety aspects undertaken by theOffice of Civil Defense were trans-ferred to the RANN organization. Dr.Ralph Long was placed in charge ofthe fire aspects. The RANN programwas focused on research at academicinstitutions. In addition to individual

efforts by eminent scientists, Longencouraged and eventually sponsoredcenters of excellence. Four such cen-ters were established at Harvard;University of California, Berkley; JohnsHopkins Applied Physics Laboratory;and the University of Utah. Of these,the work at Harvard most directlyapplies to fire dynamics and fire mod-eling.

Harvard Program. The Harvardprogram, under the direction ofHoward Emmons, was undertaken incooperation with the basic researchprogram at Factory Mutual ResearchCorporation. This center concentratedon the fire environment in a room andthe various phenomena needing exam-ination to accurately predict the roomfire conditions. Their project was titled“The Home Fire Project.” The Harvardfire model came from this program.

In the late 1970s, the National FireFoundation transferred the RANN pro-gram to NBS. The RANN projects weremerged with the established grant pro-gram of the Center for Fire Research.

FEDERAL FIRE PREVENTION ANDCONTROL ACT OF 1974 SETS UPCENTER FOR FIRE RESEARCH

The National Bureau of Standards(NBS), now known as the NationalInstitute of Standards and Technology(NIST), has had a fire research programsince 1914. Prior to the creation of theCenter for Fire Research (CFR), the pro-gram emphasis was generally notdirected at the dynamics of fire in aspace. It is currently an internationalleader in research and development ofthe understanding of fire dynamics andthe creation of fire models and otherfire protection tools and systems.

PLASTICS INDUSTRY SUPPORTSFIRE RESEARCH THROUGHPRODUCT RESEARCHCOMMITTEE

At approximately the same time asthe enactment of the Federal FirePrevention and Control Act of 1974, theProducts Research Committee wasformed. This committee administered a

trust fund established in the consentorder of 1974 between the FederalTrade Commission and 25 respondersinvolved in the manufacture of cellularplastics. The work of the ProductResearch Committee was undertaken inthe form of grants totaling $5,000,000over a five-year period.

OTHER CONTRIBUTORS

It is impossible to review the contri-butions by all the other organizationsand individual researchers to the under-standing, quantification, and engineer-ing modeling of the individual fire phe-nomena and their interactions thatmake up fire dynamics. A sampling ofthese contributors include:

• Factory Mutual ResearchCorporation

• National Research Council ofCanada

• University of Lund, Sweden• University of Edinburgh, Scotland • University of Maryland

FIRE MODELS

Impact of Livonia and Jaguar Fireson Modeling. In 1953, there was amajor fire at the General Motorshydraulic transmission plant at Livonia,MI. A few years later there was a fire,which got out of control and exceededthe ability of hose streams, at theJaguar Automobile Production Plant inthe UK. In both of these cases, therewas an interest in determining if roofventing could have made a differenceand provide protection in a mannerthat would eliminate the need to erectfirewalls in these spaces.

Thomas and Hinkley conductedextensive experiments at the UK FireResearch Station simulating conditionsin these fires. While conducting theseexperiments, they noted a large-scaleversion of the same type of layerreported by Kawagoe. A plume devel-oped feeding smoke into that layer fol-lowing the general descriptions ofMorton, Taylor, and Turner, andKawagoe. They then devised means ofdetermining roof vent requirements,based on an equilibrium between themass flow into the upper smoke layer,


and the mass flow out the roof vents.As this work was done before thecommon availability of computers topracticing engineers, Hinkley andThomas expressed their results in theform of nomographs. The nomographsprovided the engineer with a reason-able approximation of the ventingrequirements under a range of build-ing dimensions and exposure fire size.In a way, Hinkley and Thomas devel-oped the first compartment zone firemodel.

While both the US and UK werefamiliar with and have undertaken pro-jects related to both zone models andComputational Fluid Dynamics (CFD)models, a tendency developed in theUS to use zone models while activity inthe UK was more concentrated on CFDmodels. The rest of the world has beensimilarly divided.

ZONE MODELS

A zone model usually divides eachroom into two spaces or zones: anupper zone containing the hot gasesproduced by the fire and a lower zonecontaining all space beneath the upperzone. The lower zone is a source ofair for combustion and is usually thelocation of the fire source. During thecourse of the fire, the upper zone canexpand to occupy virtually all of thespace in the room.

In a zone model, the upper zone isconsidered a control volume thatreceives both mass and energy fromthe fire and loses energy to the sur-faces in contact with the upper zoneby conduction and radiation, by radia-tion to the floor, and by convection ormass movement of gases throughopenings. Some models evaluate con-ditions in the lower layer, othersassume that the lower layer remains atambient conditions. Mass is conserved,accounting for mass entering or ventedfrom the control volume.

Many zone models superimpose thecorrelations developed initially byRonald Alpert at Factory MutualResearch Corporation to describe thehot ceiling jet. In most of these mod-els, the correlation by Alpert had beenadjusted using techniques developed

by either David Evans of NBS orLeonard Copper, then with NBS, toreflect the impact of the hot upperzone on the hot ceiling jet.

Quintiere’s Paper “Growth of Firein Building Compartments.” In 1976,James Quintiere, then at the Center forFire Research, presented a paper at anASTM symposium marking the 75thanniversary of the National Bureau ofStandards. In this paper, he discussesthe possibility of creating a fire modeland outlines the variables the modelwould have to solve and the algorithmsneeded to be solved. He subsequentlygenerated a simple model, which hecalled RUNF. That model has neverbeen in common use, but it andQuintiere’s paper influenced thinkingon the part of other researchers includ-ing Ronald Pape at IIT ResearchInstitute and Zukoski at Cal Tech.

First Published Model, the IITRIModel. Pape developed the IITRIModel, the first published compart-ment fire model. It was a simple sin-gle-room compartment fire modelbased on the approaches proposed byQuintiere which used Zukoski’s plumeequations. Pape also used correlationsof burning rate data from prior workat IITRI for the OCD fire threat pro-gram. The IITRI Model was publishedslightly ahead of the Harvard model.Its importance lies in being the first todemonstrate the potential of a mathe-matical model as a fire protection engi-neering design tool.

First (The Harvard Model). One ofthe main elements of the Home FireProject at Harvard and Factory MutualResearch Corporation was a detailedstudy scientifically measuring and ana-lyzing the development of a fire fromits early ignition through flashover in abedroom situation. For three years, theproject conducted one well-instrument-ed and carefully directed test eachyear. The ensuing year was used toanalyze the data, often invitingresearchers from other fire researchprograms to join the analysis. Muchattention was directed in establishingthe algorithms, equations, and relation-

ships that would define the individualphenomena involved and the interac-tion between phenomena. In 1976, itreached to the point where ProfessorEmmons brought in help, particularlyHenry Mitler, to write a predictive sin-gle-compartment fire model. From this,in 1978, emerged the Harvard ModelMark I. The Mark I model was crudeby current standards; however, itdemonstrated the potential of zonecompartment fire models. With theobject of making the Harvard Model ascomprehensive as possible, the modelwas expanded and enlarged through aseries of improved versions. The lastversion developed at Harvard wasMark V issued in 1983, the year ofEmmons’ retirement. Emmons hadhoped that the Harvard Model couldbecome so complete and accurate thatit could be used to judge the validityof simpler and special purpose mod-els. This objective was not reached.After Emmons retired, the HarvardModel was turned over to the NationalBureau of Standards and renamedFIRST. A small cadre of scientists stilluse FIRST, but it is not generally usedin engineering applications at thistime. Jonathan Barnett and his stu-dents at WPI have modified FIRST intoan extensively expanded model theycall the WPI Model.

Ad Hoc Committee on FireModeling Passes Informationamong Modelers. During most of the1980s, scientists and engineers devel-oping zone fire models met andexchanged modeling knowledge anddiscussed related algorithms at a seriesof open meetings called by the AdHoc Committee on Fire Modeling. Thecommittee meetings were called andprovided with secretarial support byRobert Levine of the Center for FireResearch. This committee had no offi-cial standing but met several times ayear and circulated minutes with suchattachments as the attendees submit-ted. Unfortunately neither the minutesnor any other permanent record of thiscommittee was ever published. Evenso, it was an important factor in theexchange of knowledge and theadvancement of fire modeling.


Zukoski’s Equations Underpinthe Model ASET. Also in the timeperiod of the late 1970s into the 1980s,Zukoski produced a set of equationssuitable for modeling impact of fire ina room. These equations were thenused by Cooper at the Center for FireResearch to produce the compartmentfire model we now know as ASET,which stands for “Available Safe EgressTime.” The ASET model originallyproduced by Cooper was designed formainframe computer and had, by com-parison to the ASETB approach, atedious input program. An importantevent in the transfer of modeling tech-nology from the research communityto the practicing fire protection engi-neer was the development by W. D.(Doug) Walton of ASETB. Walton tookthe basic equations and reduced themto 200 lines of simple BASIC program-ming that could be run at very highspeed on the relatively slow portablecomputers of that day, improvingaccessibility to the practicing engineer.ASETB was released to the public in1985.

DETACT Opens Up the Ability toPredict Sprinkler and DetectorResponse. Within the same period ofthe mid-1980s, Evans combined theequations for ceiling jet temperatureand velocity developed by Alpert inthe late ’60s with the response timeindex and related heat transfer equa-tions from Heskestad’s work to devel-op the DETACT models. These modelsare used to predict the response ofsprinklers and heat detectors.

FPETOOL Is Developed by Nelson.Harold Nelson gathered a series ofhand equations to which he added theASETB Model and the DETACT Model.He assembled this collection in a com-puter program title FIREFORM, anabbreviation for “fire formulas.”

The FIREFORM program was subse-quently expanded by Nelson toinclude the room compartment modelFIRE SIMULATOR, which added meth-ods for estimating the burning rate his-tory of exposure fires. This collectionwas titled FPETOOL. The FPETOOLpackage was first released to the engi-

neering public in 1990. The FPETOOLcollection was eventually expanded toinclude a model of smoke flow in acorridor, and was later extended toadd a model related to smoke filling ofa room remote from the room of fireorigin as a result of smoke transferredto that room from smoke in the corri-dor.

The FPETOOL package gained greatpopularity in the fire protection engi-neering field. Similar packages fol-lowed in other places in the worldsuch as, ASK FRS, published by theUK Fire Research Station, and FIRE-CAL, published in Australia. A surveyin 1997 indicated that 60% of the prac-ticing fire protection engineers usedthe FPETOOL package as their princi-pal computerized fire dynamics calcu-lator.

FAST, CFAST, CCFM, AND BRI2. In1983, Takeyoshi Tanaka of theBuilding Reserch Institute of Japanpublished his paper “A Model of Multi-Room Fire Spread.” He spent the twopreceding years as a visiting scientistat the Center for Fire Researchexpanding on his previous fire model-ing work in Japan. After he returned toJapan, work continued on his model inboth Japan and the US. His efforts pro-duced the model BRI-2.

A Center for Fire Research team leadby Walter Jones extensively revisedand extended Tanaka’s initial programto a point of development of a distinct,separate model. The model was titledFire and Smoke Transfer (FAST) andused the core model of Hazard Mark I.The first version of FAST was releasedin 1985. Cooper concurrently devel-oped the competing model,Consolidated Compartment Fire Model(CCFM). In a consolidation action, thetwo models were compared functionby function, and the approach judgedthe best and most suitable was adopt-ed. To a large extent, the physics solu-tions from FAST and the source codesstructure and solver procedures fromCCFM were adopted into a singlemodel given the name ConsolidatedCompartment Fire and SmokeTransport Model (CFAST). Firstreleased in 1990, CFAST is a multiroom

zone model that includes proceduresfor predicting fire growth and smoketransport from a compartment of fireorigin to connecting compartments. Itmaintains conservation of energy,mass, momentum, and species. Thebasic conditions predicted by CFASTinclude the depth of smoke layer ineach compartment, smoke layer tem-perature, and species concentration(oxygen, carbon dioxide, carbonmonoxide) in the smoke layers. CFASThas been enhanced over its life toinclude additional capabilities such asthe impact of forced airflow or ventingof smoke and the prediction of othertoxic species concentrations such ashydrogen chloride or hydrogencyanide.

SFPE Computer CommitteeSpread the Word on Models. Forabout 10 years starting in 1984, theSFPE had a committee on computers.This committee served as an openexchange, usually meeting at NFPAmeetings. The committee chairmanwas Jack Watts, with major contribu-tions by Walton. This committee holdsan important position in the technolo-gy transfer of modeling and relatedfire science to the practicing fire pro-tection engineer. These meetings hadthe distinct value of presentations bypracticing fire protection engineerswho were using models relating theirexperiences, including problems aswell as successes, to their colleaguesconsidering their use.

CFD MODELS

Computational Fluid Dynamics(CFD) models divide the space beingmodeled into many small cells (on theorder of hundreds of thousands to mil-lions). The basic laws of mass,momentum, and energy conservationare applied in each cell and balancedwith all adjacent cells. The computa-tional modeling is a complex fluidmechanics solution of both turbulentand laminar flow derived from classicfluid dynamics theory. The governingequations are the Navier-Stokes equa-tions. These equations involve a set ofthree-dimensional, nonlinear partial


differential equations expressing con-servation of mass, momentum, andenergy. Important sub-models relatedto individual cells address turbulence,radiation, soot, pyrolysis, flame spread,and combustion. Approaches to themodeling differ among the variousCFD models.

CFD models can examine the fireenvironment in much greater detailthan zone models. In general, CFDmodels are significantly more expen-sive to obtain and use. Important engi-neering decisions are required in set-ting up the problem and interpretingthe output produced by the model.However, the use of CFD models infire protection problems is increasing.CFD models are particularly well suit-ed for situations where the space isirregular, turbulence is a critical ele-ment, or very fine details are sought.CFD models usually require large-capacity computer workstations ormainframe computers. Advancementsin personal computers and theimprovement in the solver routines inthe CFD models are, however, allow-ing some cases to be run on high-endpersonal computers.

The most popular approach for sim-ulation of turbulent flows in most CFDcodes has been some variant of the k-epsilon turbulence model. In thisapproach, additional transport equa-tions are solved for time-averaged tur-bulence variables (i.e., turbulentkinetic energy, k, and its dissipationrate, epsilon). Examples of thisapproach are the Fire Research Stationmodel, JASMINE, and most commer-cially available CFD codes. While thismodel is computationally economical,it is only moderately accurate due tosome simplifying assumptions.

CFD models are by no meansrestricted to fire problems. They arewidely used in the fluid dynamics andcombustion fields. One of the mostfamous cases involving CFD modelswas the analysis of the Kings Crossunderground station fire in 1989. Inthis particular case, the model usedwas the CFD model FLOW 3D devel-oped by the Harwell unit of the Britishatomic energy authority. The details ofthe model showed how the fire in the

escalators bent over to a point wherethe flame was virtually parallel withthe escalator steps, igniting the woodrails, resulting in a fierce and fast firespread up the escalators to the ticket-ing space near street level.

The development of faster comput-ers has made it possible to simulatecomplex flows using other higher-levelturbulence models. An alternativemodel to the time averaged k- epsilonturbulence model is known as LargeEddy Simulation (LES), which allowsfor the prediction of the instantaneousvalues of turbulence as opposed totime-averaged valves. Drs. HowardBaum, Ronald Rehm, and KevinMcGratten have developed andreleased an LES-based model for useon fire problems. This model, titled“Fire Dynamics Simulator” (FDS), hasdemonstrated its ability to identify con-ditions and phenomena not recognizedby other types of analysis, for exam-ple, how fire overcame three firefight-ers in a simple row house fire in theDistrict of Columbia.

The fact that FDS is free and avail-able on the NIST Web site while mostof the other CFD models have annualfee often in the tens of thousands ofdollars will result in widespread use ofFDS as a tool in actual fire protectiondesign and analysis problems.Fortunately, it is a good, soundlybased model with good support.

CFD MODELING HAS BEEN LONGESTABLISHED IN THE UK

During the early 1970s Professor D.Bryan Spalding and his team at theImperial College of Science andTechnology in London entered into aclose relationship with the FireResearch Station with the object ofdeveloping a new generation of math-ematical models with the treatment offires in enclosures.

The model developed, titledPHOENIX, was the first to use compu-tational fluid dynamics as a compart-ment fire model. As noted above, thisapproach to modeling divides thespace into mass numbers of cells andrelates the transportation of fluids andenergy between cells. The efforts at

the Fire Research Station including thework by Professor Spalding wereunder the direction of Geoff Cox. Aspecial fire model known as JASMINEwas developed based on PHOENIX.JASMINE is now being replaced withan improved version called JOSEPINE.

SUMMARY

The application of fire dynamics tothe solution of fire problems is quiterecent. Twenty years ago, Nelson gavea lecture at the SFPE TechnicalSessions at an NFPA meeting and atseveral SFPE chapter meetings. Thelecture consisted of a series of simplealgebraic equations designed to quan-tify fire dynamics applications. Forexample one equation was Thomas’equation for determining the size offire required to flashover a space. Atthe time, there was virtually no interestin these types of computations bypracticing fire protection engineers. In1985, Randy Lawson and Quintierepublished the report “Slide RuleEstimates of Fire Growth.” The reac-tion was about the same as that toNelson’s presentation of equations.Then came the proliferation of theportable computer and the generationof user-friendly fire models. Now fireprotection engineering demands theapplication of fire modeling to theproblem. The trend to performance-based design expects the engineer tomodel a spectrum of fire scenarios.The SFPE has established a committeeexamining the validity of models. Todate, the only model they have beenable to appraise is DETACT. They arecurrently examining ASET, by far theleast complex of all compartment firemodels. This does not mean that it isinherently improper to use other mod-els; indeed, fire protection engineeringpractice has progressed to almost com-mon use of either much more sophisti-cated zone models or CFD models.The using engineer is, however, asalways responsible for the quality ofhis or her product and the appropri-ateness of the solution means chosen.

Harold Nelson is with HughesAssociates, Inc.

The K- 22-Magnum22-MagnumESFR SprinklerESFR Sprinkler

Minimum Water FlowsMinimum Water Flows

forfor Maximum

Impact...

Maximum

Impact...

Performance CharacteristicsPerformance Characteristics for Factury Mutual Requirements:for Factury Mutual Requirements:

Single Sprinkler Pressure & Flow RequirementsSingle Sprinkler Pressure & Flow Requirements

K-14K-14 K-25K-25 K-22K-22PressurePressure FlowFlow PressurePressure FlowFlow PressurePressure FlowFlow

30' 50 psi 99 gpm 20 psi 113 gpm 25 psi 112 gpm

35' 75psi 121 gpm 30 psi 138 gpm 35 psi 133 gpm

40' 75 psi 121 gpm 40 psi 159 gpm 45 psi 150 gpm

45' w/ 35' stor. 90 psi* 133 gpm* 50 psi 178 gpm 50 psi 158 gpm

45' w/ 40' stor. 90 psi* 133 gpm* 50 psi 178 gpm 50 psi 158 gpm

Bldg. Bldg.

Height (Ft.)Height (Ft.)

Lower flows offers opportunities to reduce:

• Interior Piping • Underground Pipe

• Fire Pumps Sizes • Tank Sizes

Maximum deflector distance from ceiling (roof) = 18 inches.

Available in 155°F and 200°F

* K.14 Must Include one row of in-rack sprinklers.

The Reliable Automatic Sprinkler Co., Inc. Manufacturer & Distributor of Fire Protection Equipment

1.800.431.1588www.reliablesprinkler.com

Corporate Offices

Tel: 9.14.668.3470

United Kingdom Office

Tel: 441.372.728899

China Office

Tel: 010.6489.3998

This Sprinkler has been optimized for 45

feet high buildings. The minimum system

design savings in gpm required is 252 gpm —

below the K-25!

FM ApprovedFM Approved!


By Marc L. Janssens, Ph.D.

Evolutions in fire science and technology and computinghave resulted in a growing number of powerful mathe-matical models that are used in support of fire safety engi-

neering design and analysis. Good engineering practice requiresthat a statement of uncertainty accompany all computer firemodel calculations. This article summarizes a series of standardguides that help assess uncertainties in computer fire models. Italso discusses application of the guides by SFPE and others.

COMPUTER FIRE MODELS

What Is a Computer Fire Model?A fire model is a physical or mathematical representation

of burning or other processes associated with fires.1

Mathematical models range from relatively simple formulaethat can be solved analytically to extensive hybrid sets of dif-ferential and algebraic equations that must be solved numeri-cally on a computer. Software to accomplish the latter isreferred to as a computer fire model.

EVALUAT ING


Types of Computer Fire ModelsThe most commonly used computer

fire models simulate the consequencesof a fire in an enclosure. Zone modelsas well as field or CFD models areused for this purpose. Enclosure firemodels have been extended to simu-late the spread of fire and smokethrough multiroom structures.

A second category of computer firemodels predicts how materials, sys-tems, or people respond whenexposed to specific fire conditions.Sprinkler and detection activationmodels fall in this category. Otherexamples are calculation methods toassess the load-bearing capacity ofstructural elements and assembliesexposed to fire, and models that simu-late human behavior and egress in theevent of fire.

Uses of Computer Fire ModelsComputer fire models are used pri-

marily for two purposes: reconstruc-tion and analysis of a fire, and fire-safedesign of a structure. The former isusually an easier task, because there isalways other information availablesuch as forensic evidence, eyewitnessaccounts, fire department reports, etc.A computer fire model in this case ismost often used to supplement theother information in demonstratingthat a particular hypothesis is or is notplausible.

ASTM STANDARD GUIDES

ASTM subcommittee E05.39 on FireModeling (later merged intoSubcommittee E05.33 on Fire SafetyEngineering in 1996) initiated thedevelopment of a set of standardguides in the late 1980s. Four guidesare now available, which cover specif-ic issues pertinent to computer firemodeling:

• ASTM E 1355 addresses evaluatingthe predictive capability of firemodels;

• ASTM E 1472 provides guidelinesfor documenting fire models;

• ASTM E 1591 describes proce-dures to obtain input data for firemodels; and

• ASTM E 1895 addresses uses andlimitations of computer fire mod-els.

Although many of the provisions in

the guides are applicable to othertypes of fire models, the focus is oncompartment zone models.

Evaluating the Predictive Capabilityof Fire Models

ASTM E 1355 was first approved in1990 and slightly revised in 1992. Amajor revision based on work per-formed at the National Institute ofStandards and Technology (NIST)resulted in the current edition, whichwas approved in 1997. The modelevaluation process, according to ASTME 1355, consists of the following foursteps:

1. Define the scenarios for whichthe evaluation is to be conducted.

2. Validate the theoretical basis andassumptions used in the model.

3. Verify the mathematical andnumerical robustness of themodel.

4. Evaluate the model, i.e., quantifyits uncertainty and accuracy.

The first step of the process consistsof a description of the fire scenariosfor which the evaluation is to be con-ducted. Sufficient documentation is

necessary to determine whether themodel is suitable for the intended use,i.e., the simulation of fire scenarios ofinterest. Model documentation pre-pared according to the guidelines inASTM E 1472 contains all the elementsneeded for a proper evaluation.

The second step consists of adetailed review of the theoretical basisof the model, and an assessment ofthe correctness of the assumptions thatare made and the approaches that areused. An independent expert who hasnot been associated with the develop-ment of the model should perform thistask. In practice, often only the modeldeveloper has enough incentive toconduct such a tedious and time-con-suming task.

A model is then verified by assess-ing its mathematical and numericalrobustness. Verification can be per-formed by comparing model output toanalytical solutions of simple problemsfor which such solutions exist (e.g.,steady-state problems), by checkingthe computer source code for irregu-larities and inconsistencies, and/or byinvestigating the accuracy and conver-gence of the numerical solutions ofthe model equations.

Step four is usually based on a com-parison between model output andexperimental data, and provides anindirect method for validation (steptwo) and verification (step three) of amodel for the scenarios of interest(described in step one). It is generallyassumed that the model equations aresolved correctly, and the terms valida-tion and evaluation are therefore oftenused interchangeably. Experimentaldata for model evaluation can beobtained from standard fire tests, ad-hoc fire tests conducted as part of themodel development and evaluationprocess, the literature, and/or experi-ence.

Three types of uncertainties con-tribute toward the accuracy of firemodels when quantified by comparingmodel predictions with experimentaldata. The input uncertainty is primarilydue to the errors and assumptions forthe input data. Sensitivity analyses areused to identify the critical input para-meters, which must be specified withmuch greater care than the parametersto which the model is relatively insen-sitive. A sensitivity analysis of a com-


plex model might involve a very largenumber of runs to assess the effect ofall input parameters individually andof possible interactions between differ-ent parameters. Special mathematicaltechniques can be used to drasticallyreduce the number of computer modelruns without losing much information.2

The model uncertainty is primarily dueto the assumptions made by themodel, and can be quantified as aresult of the validation process (steptwo of the evaluation). Full-scale firetest data are subject to experimentaluncertainty. Therefore, discrepanciesbetween model predictions and exper-imental data might be, at least partly,due to measurement errors. There areprocedures to determine the precisionof standard test methods on the basisof interlaboratory trials or round robins(e.g., see ASTM E 691 or ISO 5725).Custom, nonstandard, full-scale fireexperiments are usually not repeatedfor cost reasons. However, the uncer-tainty of custom test data is probablycomparable to that of standard full-scale fire tests. Round robins of stan-dard full-scale fire test methods haveshown that the uncertainty of somemeasurements may be as high as ± 30 %.3

There are many problems in com-paring the results from fire model sim-ulations to data from full-scale fireexperiments. Some of the problemsare due to the differences betweenthe form of the recorded experimentaldata and the form needed for compar-ison with model predictions. Forexample, contrary to the assumptionof pre-flashover compartment firezone models, there often is not a clearand sharp change distinguishing thelower and upper gas layers.4

Quantifying the agreement between acalculated and measured curve ofvariable expressed as a function oftime also presents a major challenge.Researchers at NIST have made somerecommendations, but more work isneeded to address this problem.5

Documenting Computer FireModels

ASTM E 1472 was first published in1992, and reapproved in 1998. Thisguide requires that computer fire mod-els be documented with a technicalreference, a user’s manual, and a pro-

grammer’s guide. The technical docu-mentation describes the theoretical andmathematical foundations of themodel. The user’s manual providesinstructions for installing and operatingthe software. Sample runs should beincluded to allow the user to verifycorrect operation of the program. Theprogrammer’s guide includes thesource code and instructions for userswho want to customize the program.Model documentation preparedaccording to ASTM E 1472 contains allthe elements that are needed for anevaluation according to ASTM E 1355.

Data for Computer Fire ModelsComputer fire models typically

require physical, chemical, and flam-mability properties of materialsinvolved in the fire. ASTM E 1591describes procedures to measure manyof these properties, and includesnumerous references to the open liter-ature where property values can befound. ASTM E 1591 was firstapproved in 1994, and was slightlyrevised in 2000. A major revision is inprogress to extend the scope to CFDcodes and other types of computer firemodels.

Use and Limitations of ComputerFire Models

ASTM E 1895 was approved andpublished in 1997. Several surveyshave been published and should beconsulted to determine which modelsare available.6, 7, 8 ASTM E 1895 pro-vides guidance to model users on usesand limitations of computer fire mod-els, and thus facilitates the selection ofthe model that is most suitable for aparticular task. The document alsoprovides guidance to model develop-ers and to authorities having jurisdic-tion that review designs based onmodel calculations.

APPLICATIONS OF THE ASTMGUIDES

NIST ModelsNIST has been at the forefront of

computer fire model development formore than two decades. Many differentfire models can presently be down-loaded freely from the NIST BFRL Website (http://www.fire.nist.gov).

Although NIST has not formally

applied the ASTM guides to any of itsmodels, the contents of the ASTMguides are largely based on the pio-neering work done by NIST. Forexample, the documentation for theCFAST and FDS models covers most, ifnot all, of the elements that aredescribed in ASTM E 1472.9, 10 Severalextensive studies have been performedto evaluate the predictive capability ofCFAST and its predecessors.11

FIRMIn 2000, the author of this article

published the second edition of anintroductory text on mathematical firemodeling.12 The book describes thedevelopment of a zone fire model thatpredicts the consequences of a user-specified fire in a single compartment.The model is an extension of ASETdeveloped by Cooper at NIST. Itincludes algorithms to determine flowsthrough a ventilation opening in oneof the vertical walls. The model isreferred to as FIRM (Fire Investigationand Reconstruction Model). The revi-sion presented a unique opportunityto use the ASTM guides. This book isthe first formal application of theASTM guides to a fire model.13

SFPE Task Group on ModelEvaluation

In June of 1995, SFPE formed a TaskGroup to evaluate the scope, applica-tions, and limitations of computer firemodels. The approach used by theSFPE Task Group is largely based onthe ASTM guides, in particular E 1355.The DETACT-QS model was selectedas a test case. DETACT-QS is a modelfor predicting the response of detec-tors and sprinklers to a user specifiedfire. Despite its simplicity, it took fiveyears to complete a draft and initiate aballot. Efforts are currently under wayto add more comparative data and toenhance or clarify some of the docu-mentation. This illustrates how difficultand time-consuming it is to properlyevaluate the predictive capability of afire model. The SFPE Task Group iscurrently also evaluating the ASETmodel.

RECOMMENDATIONS

Over the past decade, ASTM pub-lished a series of standard guides that


facilitate the assessment of computerfire model uncertainties. So far, theapplication of these guides has beenvery limited. Model users in the fireprotection engineering community areencouraged to apply the guides andshare their experience with the SFPEFire Model Evaluation Task Group andthe ASTM E05.33 Subcommittee. Thisfeedback is essential to improve theguides and increase the acceptance ofcomputer fire modeling.

Marc Janssens is with the Universityof North Carolina, Charlotte.

REFERENCES

1. ASTM E 176, Standard Terminology ofFire Standards, Annual Book ofStandards, Vol. 04.07, ASTM, WestConshohocken, PA.

2. Peacock, R., Reneke, P., Forney, L., andKostreva, M., “Issues in Evaluation of

Complex Fire Models,” Fire SafetyJournal, Vol. 30, pp. 103-136, 1998.

3. Beitel, J., “International Fire StandardsProject Report, Interlaboratory TestProgram, Proposed ASTM StandardMethod for Room Fire Test of Wall andCeiling Materials and Assemblies,” PCN33-000012-31, ASTM, Institute forStandards Research, Philadelphia, PA,1994.

4. Janssens, M., and Tran, H., “DataReduction of Room Tests for ZoneModel Validation,” Journal of FireSciences, Vol. 10, 1992, pp. 528-555.

5. Peacock, R., Reneke, P., Davis, W., andJones, W., “Quantifying Fire ModelEvaluation Using Functional Analysis,”Fire Safety Journal, Vol. 33, pp. 167-184,1999.

6. Friedman, R., “An International Surveyof Computer Fire Models for Fire andSmoke,” Journal of Fire ProtectionEngineering, Vol. 4, 1992, pp. 81-92.

7. Janssens, M., “Room Fire Models, Part A:General,” Chapter 6 in Heat Release in Fires,Elsevier, New York, 1992, pp. 113-157.

8. Sullivan, P., Terro, M., and Morris, W.,“Critical Review of Fire-DedicatedThermal and Structural ComputerPrograms,” Journal of Applied FireScience, Vol. 3, 1993-94, pp. 113-135.

9. Peacock, R., Reneke, P., Jones, W.,Bukowski, R., and Forney, G., User’sGuide for FAST: Engineering Tools forEstimating Fire Growth and SmokeTransport, NIST SP 921, NationalInstitute of Standards and Technology,Gaithersburg, MD; March 2000.

10. McGrattan, K., Baum, H., Rehm, R.,Hamins, A., and Forney, G., FireDynamics Simulator: TechnicalReference Guide, NISTIR 6467, NationalInstitute of Standards and Technology,Gaithersburg, MD, January 2000.

11. Peacock, R., Jones, W., and Bukowski,R., “Verification of a Model of Fire andSmoke Transport,” Fire Safety Journal,Vol. 21, pp. 89-129, 1993.

12. Birk, D., An Introduction toMathematical Fire Modeling, TechnomicPublishing Co., Lancaster, PA, 1990.

13. Janssens, M., An Introduction toMathematical Fire Modeling (SecondEdition), Technomic Publishing Co.,Lancaster, PA, 2000.

*The SFPE Task Group and ASTM E05.33Subcommittee are chaired by Mr. DanMadrzykowski of NIST ([email protected]) and Dr. Ron Alpert of FM Global([email protected]), respectively.

Your job’s complicated enough. Fortunately,specifying the right candela doesn’t have to be.

Helping People Take Action™

INC.

273 Branchport Ave. • Long Branch, NJ 07740TEL: (732) 222-6880 • FAX: (732) 222-2588

http://www.wheelockinc.comE-Mail: [email protected]

• Field Selectable Candela Settings of: 15, 30, 75 or 110cd • Satisfies Virtually ALLRequirements for Indoor, Wall Mount Applications • Both Sync and Non Sync In ONEAppliance • Approvals: UL 1971; New York City (MEA), California State Fire Marshall (CSFM),Factory Mutual (FM), Chicago (BFP) and ULC • ADA / NFPA / ANSI Compliant • Synchronizewith Wheelock SM, DSM Sync Modules or Wheelock’s PS-12/24-8 Power Supply with Built-InSync Protocol • Backward Compatible With Other Wheelock Appliances

FEATURES

Specifying the right Fire Alarm Notification Appliance has never been easier withthe introduction of Wheelock’s revolutionary Multi-Candela product line.

In fact, whatever the candela intensity required, there’s now a single product tofulfill all requirements. Each audible and visual appliance now includes strobe settingsof 15, 30, 75 or 110 candela intensities.

It’s called SPEC1FIRE™, as in ONE spec for all your Fire Alarm NotificationAppliance needs. It will make your job less complicated, and far more cost effective inthe process.

Introducing

SPEC1FIREFIRE™

A Family of Multi�Candela AppliancesSM


By Kevin McGrattan, Ph.D.,Howard Baum, Ph.D., Ron Rehm, Ph.D., Glenn Forney, Ph.D., and Kuldeep Prasad, Ph.D.

INTRODUCTION

Scientists and engineers are oftenasked to make predictions of the stateof technology in the future and areusually laughably wrong. The bestprognosticators get the trends right,but cannot possibly fill in the details.Think of Jules Verne predicting a tripto the moon, albeit in a projectiledecked out in lavish red velvet,manned by champagne-sipping adven-turers, and shot out of a giant cannon.Unlike Jules Verne, we dare only look10 years into the future, rather than100. Also, we focus on the work goingon in the Building and Fire ResearchLab at NIST because, even thoughmost agree that modeling will play anincreasing role in fire research, thenature of the models is a subject ofintense debate. We do not presume tospeak for the entire community, andwe welcome the opinions of otherresearchers as to the direction of mod-eling in the future.

In the Spring 2000 issue of FireProtection Engineering,1 Howard Baumwrote a brief history of fire simulationin which he listed three major chal-

lenges facing fire modelers:First, there are an enormous number

of possible fire scenarios to consider.Second, we do not have either thephysical insight or the computingpower (even if we had the insight) toperform all the necessary calculationsfor most fire scenarios. Finally, sincethe “fuel” in most fires was neverintended as such, the data needed tocharacterize both the fuel and the fireenvironment may not be available.

Ten years from now, these issueswill remain. Certainly the wide rangeof fire scenarios will persist, even wid-ening due to the constant emergenceof new materials and new architecturalforms. Computing power will certainlyincrease, but not to the point of allow-ing for direct numerical solutions ofthe governing equations. As modelsfocus in on the small-scale combustionprocesses in a fire, ever-more complexchallenges will emerge that are, fornow, neglected. Fortunately, there ishope. The reason is that models basedon fundamental principles willimprove automatically as computersget faster and the temporal and spatialresolution improves. In lookingtowards the future, we need to adoptfundamentally sound physical mecha-nisms that will retain an elegance andsimplicity over time, that will shift usfrom empirical to deterministic descrip-tions of fire behavior, and that will be

useful to fire protection engineers andresearchers alike.

BLOWING SMOKE

Even as we develop more sophisticat-ed numerical algorithms to describe thegrowth and suppression of fires, themajority of design calculations will con-tinue to address a subject for whichzone and field fire models were firstdeveloped – smoke movement. Becausesmoke inhalation and carbon monoxidepoisoning are and will remain the mostdangerous actors in a fire, code officialswill continue to enforce regulationsdesigned to ensure safe evacuation of abuilding. Originally, two-zone fire mod-els were developed to predict thedescent of the smoke layer in a fairlysimple building, but as building geome-tries become more complex, fire protec-tion engineers are turning to field mod-els to track the smoke in open-plan,multilevel buildings.

Ten years from now, engineers willstill be interested in smoke movementfrom fires whose size and growth ratewill be predefined. Current field modelscan handle these problems in theory,but computation times are often toolong or the grid resolution is too coarseto capture important features of the flow.The solution to this problem is fastercomputers, better allocation of gridcells and parallelization, all of which

The Future of

SimulationFire


are subjects of active research by com-puter scientists because the applicationof these ideas goes way beyond fire.

What we can do now is adopt models that improve automatically asnumerical grids become more refined.The best example of this idea is LargeEddy Simulation (LES). We have foundover roughly twenty years that goodsimulations result from solving theNavier-Stokes equations with as fewempirical parameters as possible ongrids with as many cells as possible.While we always want more, we havefound that very good results areobtainable with modest calculations,allowing engineers running our mod-els to investigate a wide variety ofproblems without having to worryabout numerical parameters for whichthey have little training.

It is our belief that the LES conceptwill emerge as the prevailing method-ology for smoke and heat transportbecause of its ability to render realis-tic, time-resolved animations of theflow of gases throughout a building. Itis inevitable that as computers getfaster, users of CFD models willdemand more lifelike simulationsrather than time-averaged or steady-state images. This is particularly true ofthe fire community since the audiencefor many of the simulations are theauthorities having jurisdiction whooften have little training in fire model-

ing. The design engineer must demon-strate to the official what is being cal-culated with something more than stat-ic images or time-temperature plots.Animations of smoke flow provide avisual check of the building geometry,grid resolution, and other features ofthe calculation that are difficult to con-vey any other way.

WHAT ABOUT THE FIRE?

Skeptics of fire models have com-plained from the outset that the fire isnot really modeled in a fire model. Toa large extent, this criticism is valid,and here are two reasons why. First,most engineers are usually interestedin smoke movement, so there’s no rea-son to model the fire other than as apoint source of smoke and heat.Second, the combustion processesoccur at length and time scales belowthe resolution limits of most practicalcalculations, so much so that informa-tion obtained from the resolvablescale, like an average cell temperature,is useless in even the most simplisticof combustion models.

Much of the future research in firemodeling will focus on improvementsin the way small “subgrid-scale” physi-cal processes are modeled. Examplesof such processes include (but are notlimited to) soot formation and growth,combustion in vitiated atmospheres,

heat transfer to pyrolyzing surfaces,and radiation from gaseous combus-tion products. All of these phenomenaoccur in both laboratory-scale experi-ments and material test apparatus.These processes will serve as the start-

at NIST

FIGURE 1. Simulation of a sample ofwood burning in the cone calorimeterperformed with FDS. Courtesy SimoHostikka, VTT Building and Transport,Finland.


ing point for developing better firesubmodels because they are well-con-trolled, relatively simple, and, mostimportantly, small. Because they aresmall, the calculations can be per-formed at sufficiently high resolutionto capture the important phenomenadirectly, and then the same calcula-tions can be performed at lower reso-lution to see how well the new algo-rithms perform on larger-scale simula-tions. The objective of this effort is notto produce the most detailed descrip-tion of the phenomena. Very detailedsubmodels of most fire phenomenaexist now; the challenge is to designan overall fire model that balances theaccuracy of each submodel. Balancemeans that the level of detail incorpo-rated into each is roughly the same.Everyone learns in high school thatadding a measurement accurate to thenearest millimeter and one accurate tothe nearest centimeter yields a resultthat is only accurate to the nearestcentimeter. Similarly, a fire model willonly be as accurate as the least accu-rate of its components.

A good starting point for a betterfire model is a well-controlled testapparatus, like the cone calorimeter(Figure 1). A set of solid and gasphase models should be developedthat would hopefully provide a reason-able, balanced description of the inter-action of the fire with the test appara-tus. In essence, this is the procedurethat was followed in the developmentof the Fire Dynamics Simulator (FDS),a general-purpose fire field modelreleased into the public domain in theyear 2000. The approach had been tomodel the large-scale gas phase trans-port as faithfully as possible for agiven numerical grid, and then intro-duce extra features that were consis-tent with the detail (or lack thereof)afforded by the smoke transport algo-rithm. In some sense, the fire itselfwas just another one of these extrafeatures. At first, the fire was aGaussian-distributed blob of heatsuperimposed on the numerical grid,then the fire was a set of hot particlesejected from the burning object, andfor the time being the fire is a surfaceon which fuel and oxygen meet andreact infinitely fast. The emerging firemodel may move beyond this, but atthe moment it should be possible to

work with this description of the com-bustion while the solid phase mecha-nisms are brought up to par.

There are two advantages to thisevolutionary strategy. First, the varioussubmodels, even in their primitivestates, have been useful to FPEs forsmoke movement and simple heattransfer calculations, and to introducethe next generation to the technology.Second, all aspects of the simulationimprove at the same pace – sprinklers,radiation, burning objects – so that nopart of the calculation looks out ofplace. A good analogy is classicalsculpture. The artist transforms a blockof marble into a human form bypainstakingly chipping away stone firstto reveal the gross outlines of thehead, arms, torso, etc., and only thenfocuses in on finer details. Considerthat the most beautifully sculpted handwould look ridiculous if one arm werelonger than the other.

INDUSTRIAL-STRENGTH FIRES

A few years ago, in parallel withlarge-scale tests, the development ofFDS turned towards the problem offire suppression in large warehousesand warehouse retail stores2 (Figure 2).As simplistic as the combustion andheat transfer algorithms were at the

time, they were not nearly as primitiveas the sprinkler spray and suppressionmodels. With the exception of thethermal activation equation, which bythat time had become widely adopted,the water droplets emerging from thepipe, landing on the commodity, andeventually interacting with the firewere by far the greatest source ofuncertainty in the simulation.

A series of bench-scale experimentswas conducted at NIST to developnecessary input data for the model.These experiments generated datadescribing the burning rate and flamespread behavior of the cartoned plasticcommodity, thermal response parame-ters and spray pattern of the sprinkler,and the effect of the water spray onthe commodity selected for the tests.The missing link in the analysis wasthe spray characterization of the sprin-kler itself; that is, the water wasassumed to leave the sprinkler in asimple umbrella pattern quantifiedonly by visual observation. What madethe model work reasonably well wasthe fact that the spray parameters weretweaked until a match between com-puted and observed water density pat-terns on the floor was obtained.Hundreds of hours were needed toroughly characterize one fuel and onesprinkler because the characterization

FIGURE 2. Simulation of a rack storage commodity fire. The simulation predictsthe growth and suppression of a fire that originates at the floor. These types ofsimulations are by far the most challenging attempted to date, and it remains tobe seen how much the relatively simple solid phase pyrolysis and suppressionalgorithms can be improved.

One Company. One Call.

e are the one company in theindustry that takes total responsibilityfor meeting all of your needs. Fromportable fire extinguishers and sprin-klers to fire alarm and suppressionsystems, we’ll inspect, test, maintainand repair any and all of your fire andsecurity equipment.Think of it thisway:Anything. Anytime. Anywhere .We’ll also give you an ironclad guar-antee from SimplexGrinnell: We won’tstop until you’re satisfied that you’vereceived the best possible service.

It’s all part of our total services, totalexcellence approach to protecting your people and property. WithSimplexGrinnell as your protectionpartner, you can rest assured that inan emergency situation, your safetysystems will operate properly. Andwhen all is said and done, what could be more important than that?

Call us today for your free servicesevaluation.800-746-7539 in the U.S.or 800-565-5400 in Canada. Or visit our web site for more information.www.simplexgrinnell.com/safe

�

Total protection. SimplexGrinnell.We are the number one choice in comprehensive single-sourceprotection for fire and life safety.

© Copyright 2001, SimplexGrinnell LP

TOTALSERVICES

TotalExcellence

Inspection

Testing

PreventiveMaintenance

CentralMonitoring

OngoingSupport

24/7Emergency

Service

TOTALSERVICES

TotalExcellence

Inspection

Testing

PreventiveMaintenance

CentralMonitoring

OngoingSupport

24/7Emergency

Service

Protecting People, Property and Peace of Mind

With mounting flexibility, and a NEMA 6P rating, our outside switches can take

the heat (and cold) and still perform heroically. Day in and day out, even in the

harshest conditions, underwater or above ground. For over a century, our strength

has been our proven record of performance, protecting businesses just like yours.

PThe Symbol of Protection

1 . 8 0 0 . 3 2 5 . 3 9 3 6 w w w. p o t t e r s i g n a l . c o m

®

Remember, if it doesn’t have The Symbol of Protection, It isn’t Potter.P


was almost all empirical – little of itwas based on fundamental physicalmodels because the phenomena wasso very complex. As a result, users ofthe FDS model were not able to applyit easily to other commodities andsprinklers; a problem that persists tothis day.

Sprinkler spray characterization willremain largely empirically basedbecause each sprinkler has its ownunique design that makes predictingwhich way the water will go difficult.To simulate the sprinkler spray, weneed to know the initial distribution ofthe droplet size and velocity.Measuring these quantities has provento be very difficult and still veryexpensive. The most promising tech-nique for measuring droplet size isthrough Phase Doppler Interferometry(PDI) and droplet velocity throughParticle Image Velocimetry (PIV) (seeFigure 3). Both are nonintrusive, laser-based techniques that require veryexpensive equipment and skilled tech-nicians with a high level of training inlaser diagnostics. This is worrisomebecause calculations should be cheap-er than experiments, or else what’s thepoint? If high-level modeling of chal-lenging industrial fire scenariosbecomes more routine and starts toshow potential benefits to sprinklermanufacturers and building owners,

there ought to be more investment inthe measurement techniques requiredfor input data. The Catch-22 is that it’shard to show benefits with little data.

Understanding how various standardcommodities burn and how theyrespond to water ought to be lessempirically based than sprinklersprays, assuming the necessary solidphase models are developed thatretain enough of the fundamentalphysics to accommodate a betterdescription of suppression, yet simpleenough to be used in large-scale simu-lations. We discussed above the needfor more fundamentally based modelsof pyrolysis, starting with relativelysmall-scale calculations of standard testapparatus and eventually moving tolarge-scale. It is unclear how todescribe the burning of real commodi-ties, which are mixtures of cardboard,plastics, woods, etc., other than withthe simple lumped parameter modelsdeveloped to date. It is hoped that at aminimum, we will have a way of relat-ing the burning rate of the fuel to theheat feedback to the surface based onthe thermophysical properties of thefuel rather than simply an exhaustiveseries of experiments that are oftentoo expensive to perform given thewide variety of fuels burning in a sin-gle fire. This is possible now with alimited number of pure fuels, liquidsespecially, but hopefully this list canbe extended in the future.

THE FIRE HAS LEFT THE BUILDING

The fire models with which we arefamiliar were developed to describeresidential and commercial buildingfires. However, there is a differentclass of models developed by the for-est management and agricultural com-munities designed to predict thespread of wildland and forest fires.These models are semiempirical andare built upon very different assump-

FIGURE 3. Image of a sprinkler spraycreated with Particle ImageVelocimetry (PIV). The green arrowsrepresent the velocity vectors of waterdroplets leaving the sprinkler orifice. Thetechnique involves taking twophotographs of the spray in rapidsuccession and backing out the velocityfrom the displacement of the droplets.Courtesy Dave Sheppard, NorthwesternUniversity and UnderwritersLaboratories.

FIGURE 4. Simulation of a brush fireadvancing on a house. Preliminarycalculations such as these are now beingperformed to assess the feasibility ofsimulating community-scale fire spread.Here, the domain is a few hundredmeters on a side, the grid cells about 1m. The trees serve as a drag on theoncoming wind.

EST3 SYNERGY-ENABLED LIFE SAFETY SYSTEM.THE BACKBONE OFA SECURE ENVIRONMENT.

EST3 Life Safety Platform supports up to160,000 points, making it the most flexiblelife safety controller in the business.

The 3-SAC Security Access ControlModule resides on the local rail of the EST3panel, providing synergy between fire, secu-rity and access control equipment.

Fully compatible with Genesis‘and SignatureSeries devices. Shown here are: GenesisHorn Strobe, Signature intelligent multiplesensor detector, SIGA-MD passive infraredmotion detection module and SIGA-SEC2Security Module.

This large LCD display and telephone-style keypad allows for easy managementof security events.

Adding CCTV surveillance capabilityenhances the synergy of the overall system.

Each Card Reader Controller monitors andcontrols a door with both entry and exitreaders, door strikes or magnetic locks.

The Access Control Database programuses familiar Windows-based commandsto grant or deny cardholders access toany combination of doors.

The FireWorks family of software and hard-ware options utilizes a four-quadrantapproach, presenting prioritized informa-tion in an easy-to-manage, graphical andtext format.

For over 130 years, Edwards Systems Technology

has been designing and manufacturing leading

edge fire alarm systems for Fortune 500 compa-

nies around the world. Now, we’re proud to bring

the superiority of world class fire alarm technology

to the security industry.

Unlike fire alarm systems, security and access

control systems are not mandated by code to

meet stringent agency life safety requirements.

But with a Synergy Enabled EST3 platform as the

backbone, all of your building’s systems – from fire

to access control to security and CCTV—work at

a higher level of fault-tolerant interoperability. So

not only can the systems work together to fend off

any type of intrusion, EST3 survivability insures

they won’t die trying.

The End Of The Spineless Solution.

The EST3 System supports all types of security

devices by integrating them into our system through

Signature Series Security Modules. One backbone.

One source of command. One great system.

LowerYour Costs

While RaisingYour Standards.

You may expect to pay more, but the fact is, you

pay less. Less for both installation and mainte-

nance costs. Less to manage it. So don’t settle for

a spineless security, access control and fire solu-

tion. Insist on the Synergy Enabled Life Safety

System from Edwards Systems Technology.

Just call 1-888-EST-2DAY or visit our website.

r

enabledTM

EDWARD S SY STE M S TE CHN OLO GY

World leader in life safety technology.www.est.net/SecurityAccessFire

EST 1125 Bui l di ng Spr ead 9/ 5/ 01 2: 13 PM Page 2


tions than building fire models. Theyare closer to zone models in philoso-phy since they are designed to runfaster than real time so that they canbe used the day of the actual fire. Itwould be hard to believe that fieldmodels for community-scale firespread could be developed in tenyears’ time to be run in real time, butfield models are being developednow, both at NIST and elsewhere, tostudy the behavior of large outdoorfires to aid in planning efforts.

Last year, wildland fires cost an esti-mated one billion dollars for just thefirefighting. However, fires in the builtenvironment are generally even morecostly. The famous 1991 Oakland Hillsfire in Berkeley, CA, alone did an esti-mated $1.5 billion in damage whilekilling 25 people and injuring 150 oth-ers. While large-scale models of firepropagation are useful in wildland set-tings, corresponding models for com-munity-scale (rural, suburban, orurban) fire spread, i.e., fires spreadingbetween structures and natural fuel,are still in their infancy. Developmentof such models suffer from the follow-ing Catch-22 – validated models ofcommunity-scale fire spread are need-ed because experiments on that scaleare almost impossible to carry out; butwithout experiments, how do we vali-date the models?

The main numerical problem tocommunity-scale fire prediction is gridresolution. Consider a square kilome-ter of terrain containing both structuresand dry vegetation. Any field modeltracking the progression of a brush firethrough the area would require severalmillion grid cells, which, even if clev-erly distributed, would provide spatialresolution of at best a meter. Existinglarge-scale models of wildland firesregard the fuel (vegetation) as continu-ous and assume the fire to propagateas a line. Resolvable-length scales forthese models are tens to hundreds ofmeters. The technical challenge for thecommunity-scale fire model is todevelop a mathematical description forthe ignition and burning of individualtrees and shrubs, and to determine firespread between wildland elements andstructures. Such a mathematicaldescription must include fire spread bybrand generation, transport, and sub-sequent ignitions for both wildland

fuels and structures. As with any usefulmodel, these descriptions must be vali-dated using experimental data andmust then be integrated into a CFDflow solver generalized to account foran atmospheric boundary layer flowconditioned by natural topography,upwind structures, and trees.

In addition to the numerical chal-lenges posed by community-scale fireprediction, it is often difficult orimpossible to obtain meteorologicaland topological information in a formthat can be used in the calculation.The meteorological conditions drivingthe fire have to be postulated orderived from a mesoscale weathermodel with a minimum resolutionmeasured in kilometers and the terrainfeatures obtained from a database thatmay or may not exist at the requiredresolution for that particular patch ofthe earth. Fortunately, there are nowefforts within the meteorological andgeographical research communities todevelop and maintain models anddatabases that would be useful to thefire community. For example, digitalelevation data from LIDAR overflightmeasurements is being made increas-ingly available and cost-effective.

PROOF BY PRETTY PICTURE

Modelers are looked upon withskepticism by the rest of the fire com-munity because of the perception thatthey all too often hide behind an eye-catching image or animation withoutunderstanding the physics underpin-ning the model. In fact, some havestarted to refer to CFD as “colorfulfluid dynamics.” This is often a fairassessment, but it is short-sighted.While the rapid improvement in visual-ization techniques has been a boon tomany in the field who use field mod-els on a regular basis, within the next10 years what is now gee-whiz willbecome ho-hum. This is good becauseas field modeling becomes routine, thediscussion will be raised beyond thesuperficial level we are at now to apoint where the quality of a simulationwill be judged by the spatial and tem-poral fidelity of its images and anima-tions. With any field model, the userchooses a numerical grid on which todiscretize the governing equations. Themore grid cells, the better but more

time-consuming the simulation. Thepayoff for investing in faster comput-ers and running bigger calculations isthe proportional gain in realism mani-fested by the images.

As the community at large becomesaccustomed to looking at various pic-tures and animations, model develop-ers will find new ways to dazzle. Upto now, most visualization techniqueshave provided useful ways of analyz-ing the output of a calculation, likecontour and streamline plots, withoutmuch concern for realism. A rainbow-colored contour map slicing downthrough the middle of a room is finefor researchers, but for those who areonly accustomed to looking at realsmoke-filled rooms, it may not have asmuch meaning. Visualization in thenext 10 years will turn towards provid-ing as much information as the rain-bow contour map but in a way thatspeaks to modelers and nonmodelersalike. Take, for example, Figure 5.Presented are two ways of visualizingthe same calculation, each figure madefor a different audience. The trend inscientific visualization is to combinethe features of each into one to reachboth groups of people.

A good example is smoke visibility.Unlike temperature or species concen-tration, smoke visibility is not a localquantity but rather depends on theviewpoint of the eye and the depth offield. Advanced simulators and gamescreate the illusion of smoke or fog inways that are not unlike the tech-niques employed by fire models tohandle thermal radiation (Figure 6). Itis envisioned that eventually graphicshardware and software will play a rolein actually computing results ratherthan just drawing pretty pictures. AHJsoften ask whether or not buildingoccupants will be able to see exitsigns at various stages of a fire. Thefire model can predict the amount ofsoot in each grid cell, but that doesn’tanswer the question. The harder taskis to compute on the fly within thevisualization program what the occu-pant would see and not see.

CAN’T YOU MAKE IT GO FASTER?

Computational Fluid Dynamics wasbuilt around weather prediction andaerospace design. A quick browse


through any CFDtextbook will makethis point clear. Althoughfire modeling borrowsmany of the same physicalassumptions from weather modelers andnumerical algorithms from the aerody-namics community, it is different in oneimportant way – it is practiced by rela-tively small organizations whose engi-neers have limited backgrounds innumerical methods and computing.Although many small FPE firms havebeen absorbed by larger, more diversifieddesign and architecture firms, the typicalfire modeling effort within one of theseorganizations is modest – a few engi-neers working for a few weeks on agiven design problem with computersnot much more powerful than thosefound in the home. The reason for this isthat fire protection is but one of manyfeatures of an overall building design,and one which is typically squeezedwhen other items in the budget run inthe red.

Because of how it is practiced, firemodeling has always emphasized sim-plicity and efficiency. One of the firstquestions that we are asked wheneverwe demonstrate the latest simulation ishow long did it take to set up the caseand how long did it take to run. Theanswer to both of these questions needsto be on the order of a day or less (anddon’t tell me I can’t run it on my lap-top!). If it’s more, then we’ve lost 90% ofour audience. This presents us with a bitof a dilemma – how do we stay at theforefront of CFD but still serve the com-munity of practitioners? One way is todesign the model so that one can easilystart doing simple calculations with sim-ple geometries and then systematicallywork up towards more elaborate appli-

FIGURE 6. This figure is an example of how fog is used to bring more realism intothe scene. Shown is a simulated smoke plume made with the ALOFT (A Large OutdoorFire plume Trajectory) model. The plume is embedded within a fog-shrouded oil fieldvisualized with the commercial software package SGI Performer.

FIGURE 5. Two different ways ofvisualizing a fire simulation. On theleft, contours of gas temperature areshown superimposed on the numer-ical grid. On the right, the fire andsmoke are shown as an orange-colored sheet and black dots.

Clas•sic (klas’ik) adj. 1a. Belonging to the highest rank or class; b. Serving as the established

model or standard; c. Having lasting significance or worth. 2. Formal, refined, and restrained in style.

AnaLASER® II from Fenwal

High sensitivity smoke detection fortelecommunications, data processing and other mission critical facilities where businessinterruptions are simply not tolerated.

Active detection technology that’s thousands of times more sensitive than

traditional smoke detectors. Selectable sensitivity for installation flexibility.

Networking for up to 127 devices in one or multiple locations. Engineered

for reliability from Fenwal, mission-critical specialists for more than 30 years.

www.AnaLASER.com

Protection SystemsFENWAL PROTECTION SYSTEMS 400 Main Street • Ashland, MA 01721 • (508) 881-2000


cations. The trouble with many types ofengineering software is that it is packedwith so many features that the learningcurve for even the simplest of problemsis too steep. Fire modeling, especiallyfield modeling, will advance only ifthere is a large enough core of users tojustify the time and expense of develop-ing and maintaining a very complexcomputer code. Sustaining that core ofusers means making the software acces-sible to a wide audience.

Not only must the software be easy touse, but the calculations must run as fastas possible. Veteran CFD practitioners donot find week-long calculations unusual,but fire protection engineers who onlyhave experience with zone models findit intolerable. Faster computers havesoothed some, but the demands formore-detailed calculations often negategains made in computer speed. To keepup with demand, the fire models willneed to exploit advances in computerscience and numerical methods that gobeyond just faster chips. Parallel process-ing is becoming more of a reality in cer-tain fields, but still is a few years awayfor those using the current generation ofpersonal computers. However, in thenot-too-distant future, relatively inexpen-sive desktop computers will come with2, 4, or 8 processors, plus the necessaryhardware and software to make thesechips work together effectively. Also,techniques to better distribute the gridcells will allow for greater flexibility inthe design of simulations. One techniquethat is used by many CFD packages (butnot yet FDS) is called multiblocking. Anexample of how this would work is ahouse in which every room has its ownnumerical grid. Those rooms requiringmore spatial resolution could have finergrids, those that don’t need it couldremain coarsely gridded. The numericalalgorithms presently used in single-blockcodes will not change except thereneeds to be extra logic built into thealgorithms so that information is proper-ly communicated across block interfaces.Such a technique is perfect for fire mod-els because most simulations investigatebuildings with relatively simple, rectan-gular geometries. Contrast this with theaerospace industry where simulations areperformed on very complicated bodyshapes. These models utilize numericalgrids that are far more sophisticated anddifficult to construct than those neededfor fire models.

FIGURE 7. Several snapshots of a fire spreading through a townhouse. The fireoriginates in the kitchen area (lower left), and eventually spreads throughout the house.The front door (right) is assumed to be open, as are the windows on the second level.


WHO’S GOING TO USE FIRE MODELS?

The discussion thus far has focussed mainly on the applica-tion of fire models to design problems. This is not surprising,since fire models have traditionally worked best when the fireitself was considered merely a model input rather than a modeloutput. However, fire models have been used as forensic toolsin the past, and their use as such will accelerate in the future.In fact, much of the work to improve fire models past the pointof just smoke movement has benefited the fire investigatorsmoreso than the designers who are most often content to dial ina design fire rather than try to predict its growth and suppression.

The fire service in particular has traditionally been skeptical ofany type of model, usually preferring full-scale experimental dataover computer simulations. However, recent work3 with fire mod-els to reconstruct several fire losses has moved some in the fireservice to consider the use of fire models as training tools for fire-fighters. If the present interest in simulation by the fire servicecontinues, a great deal of effort will be placed on understandingthe spread of fire through an entire house, not just a single room.Reconstructing a raging house fire goes way beyond simplesmoke and heat transport because, as the fire spreads from itspoint of ignition to envelop entire rooms in flame, the responseof the wall materials becomes tightly coupled to the progressionof the fire in a way that up to now fire models have largelyneglected. Presently simulations of entire house fires are beingperformed to roughly scope out the grid resolution and physicalmechanisms necessary to at least capture qualitatively thesequence of events from primary ignition to second-object igni-tion to room flashover to room-to-room fire spread (Figure 7). Itis difficult to validate such calculations, but at least we are startingto understand what we’re up against. Validation will come frommore controlled single-room experiments, like the ISO 9705 roomcorner test, and from simulations of test apparatus.

Ultimately, the users of fire models, whether they beresearchers, design engineers, firefighters, or litigators, will drivethe development of new features. The challenge to developersis to create fire models that can be used and understood by allof these groups. Even if some do not need the entire set ofmodel features, their use of the models will speed the develop-ment and acceptance of them because much of the skepticismassociated with modeling will diminish as more people growcomfortable with the capabilities and limitations.

ACKNOWLEDGMENTS

The work described in this article is the contribution ofmany people at NIST and beyond. We are indebted to a num-ber of guest researchers and post-doctoral associates who havecontributed to the modeling effort over the past 10 years,including Dike Ezekoye, Ruddy Mell, Javier Trelles, FrancineBattaglia, Jason Floyd, and Simo Hostikka; in addition, our col-leagues at NIST for input data, validation experiments, andsoftware testing: Jason Averill, Dave Evans, Paul Fuss, BillGrosshandler, Anthony Hamins, Dan Madrzykowski, RickPeacock, Bob Vettori, Doug Walton, and John Widmann.Finally, thanks to the many users of our various fire models,for useful feedback, suggestions, and critiques.

The authors are with the National Institute of Standardsand Technology.

REFERENCES

1 Baum, H.R., “Large Eddy Simulations of Fire.” Fire Protection

Engineering, 6:36-42, Spring 2000.

2 McGrattan, K.B., Hamins, A., and Stroup, D., “Sprinkler, Smoke & Heat

Vent, Draft Curtain Interaction – Large Scale Experiments and Model

Development.” Technical Report NISTIR 6196-1, National Institute of

Standards and Technology, Gaithersburg, MD, September 1998.

3 Madrzykowski, D., and Vettori, R.L., “Simulation of the Dynamics

of the Fire at 3146 Cherry Road NE, Washington, D.C., May 30,

1999.” Technical Report NISTIR 6510, National Institute of

Standards and Technology, Gaithersburg, MD, April 2000.

Engineering Manager/Principal-in-Charge

Interested in career advancement with one of the most respected fire protec-tion and security engineering firms in the United States? If so, Gage-Babcock & Associates may have the opportunity for you. We’re currentlylooking for a dynamic engineering professional to be the Principal-in-Charge of our San Francisco regional office. This Principal-level manage-ment position would have complete oversight to all business and engineer-ing operations of the office, covering all of the Pacific Northwest. As aPrincipal within the GBA management team, you will report directly to thePresident of the firm.

Only candidates with a proven track record of effective management of busi-ness operations and strong business development skills will be considered.If you get satisfaction from developing new business opportunities andbuilding a client network, as well as from the challenge of managing anddeveloping clients and building an office, send your résumé. Positionrequires engineering registration and at least 7-10 years of proven manage-ment experience in the engineering field.

Please submit your résumé and salary history in confidence to:

Thomas W. Jaeger, P.E., PresidentGage-Babcock & Associates, Inc.5175 Parkstone Drive; Suite 130Chantilly, VA 20151-3816Fax: 703-263-1549e-Mail: [email protected]

CITY OF LAS VEGASASSISTANT FIRE PROTECTION ENGINEER

APPROX. MONTHLY SALARY: $3,769 – $5,878FILING DEADLINE: Feb. 4, 2002 @ 4:30 P.M.

ORAL TEST: Feb. 19, 2002

QUALIFICATIONS: Bachelor’s degree in engineering, preferably in fireprotection engineering. Two years of experience in plans review, including oneyear in fire protection engineering work, is desirable. Valid driver’s license onthe date of application. Uniform Fire Code certification and Uniform BuildingCode Plans Examiner certification within 18 months of the date of hire.

To download an application, visit our Web site at www.ci.las-vegas.nv.usand click on employment opportunities. Or call us at 702-229-6315 to requestan application through the mail. The application can be faxed to us at 702-385-1259 or mailed to the address on the application.

EOE/AA M/F/DTDD# 386-9108

ACCLIMATE,™ by System Sensor, is an advanced photoelectric-thermalfire sensor with unique capabilities. ACCLIMATE continuously monitorsits environment and automatically adjusts its sensitivity accordingly.Whether a location is an office environment or a manufacturing facility,occupied, vacant or in transition, ACCLIMATE will always providemaximum protection. Time saving features, quality and innovation,coupled with System Sensor's experience and service, are the heart of our field-proven designs.

Time ACCLIMATE intelligent fire sensors are your best choiceto save you time. By automatically adjusting its sensitivity, ACCLIMATEeliminates the need for technicians to return to a site when propertyuse changes. In addition, on-board drift compensation offsets the effectof dust build up in the sensing chamber and reduces maintenance.

Experience System Sensor’s intelligent fire sensor productsare based on proven expertise from millions of installations around theworld. System Sensor makes research and development a priority, withfocused engineering and well-equipped test facilities. System Sensor’syears of experience ensure your peace of mind whether you arespecifying or installing.

Quality System Sensor quality is an integral part of every firesensor and accessory. Our fire sensors are built to stringent requirements,exceeding those of regulatory agencies throughout the world.

Service System Sensor’s dedication to meeting your needsdoesn’t end at the point of sale. We have application engineers readyto help…a customer support department that is on call to supportyour needs…and, technical documentation available 24/7 through automated FAX or CD-ROM.

Find out why ACCLIMATEmakes sense for your application. Call for a free E•DOCSTM CD-ROM, a comprehensive resource of technicalinformation 800/927-6676.

What Fire Sensor Has the Intelligence to Make Sense of Your Changing Environment?

ACCLIMATE™ reduces nuisance alarms. Faced with an increase in “noise,”such as dust or fibers, ACCLIMATE will increase its alarm level. When thedisturbance subsides, the unit becomes more sensitive.

ALARM LEVEL

AMBIENTNOISE

SE

NS

ITIV

ITY

TIME


By Frederick W. Mowrer, Ph. D., P.E.

INTRODUCTION

A master builder wouldn’t use a fram-ing hammer for finish work or a tackhammer for framing. The masterbuilder knows what tools are availablefor different tasks, knows how toproperly use each tool, and under-stands which tool is appropriate forthe task at hand. If not qualified for aparticular task, the master buildereither develops the skills to performthe necessary tasks to complete theproject, or, more likely, hires anotherprofessional with the required skillsand expertise.

The practice of virtually any trade orprofession demands a similarapproach. We all have a limited set ofskills and expertise. Consequently, wemust all understand our limitations aswell as our capabilities. This demandsthat we understand the appropriateuses and limitations of the tools thatare available to us as well as our owncompetencies in applying the availabletools. Where our competencies arelacking, we need to either enhanceour skills or obtain the services ofpeople with the required skills.

In the field of fire protection engi-neering, the “toolbox” of models andcalculations available for quantitativefire hazard analysis has expandedgreatly over the past two decades. Wehave gone from “slide rule estimates offire growth”1 and other hand calcula-tions and correlations embodied inmany chapters of the SFPE Handbookof Fire Protection Engineering,2 throughcomputer-based single- and multiroomtwo-zone fire models,3, 4 to emerginghighly detailed computational fluiddynamics (CFD) models.5, 6

Spurred by the ever-increasingprocessor speeds and memory capaci-ties of desktop computers, it is nowpractical to perform detailed firedynamics calculations and visualizeresults in ways that were virtuallyunimaginable only a few years ago.While these ever-increasing computa-tional capabilities are valuable, they arenot needed for every job. Given theenormous computational needs ofthese sophisticated calculations, itmakes sense to make use of computa-tionally less-intensive tools where pos-sible and save the more-detailed, com-putationally intensive calculations forthose tasks where the simpler tools areinadequate.

One of the first things the fire protec-tion engineer must do before perform-ing a quantitative fire hazard analysis isdecide what is the right tool for the job.

WHY FIRE MODELING?

All of the tools discussed in this articlepermit the calculation of one or moreaspects of enclosure fire dynamics. It isimportant to keep sight of the objectivesof such analyses. We perform calcula-tions and other analyses to evaluate theadequacy of a proposed or existingcomponent, system, or process to fulfillits functions. In other words, we per-form calculations to support design deci-sions. As noted in the SFPE Guide toPerformance-Based Fire ProtectionAnalysis and Design of Buildings,7 thefirst steps in the process involve theestablishment of fire safety goals, objec-tives, and performance criteria.

Once the performance criteria areestablished, different calculations canbe performed to determine whether thespecified performance criteria will beachieved by a proposed design. Someevaluations may involve only a fewsimple calculations, while others, suchas CFD analyses, may require millionsor even billions of calculations. In gen-eral, there are three possible outcomesfor a calculation within this context:

• The calculation clearly demon-strates achievement of the objec-tive performance criteria;

• The calculation does not clearlydemonstrate either achievement ofor failure to achieve the objectiveperformance criteria;

• The calculation clearly demon-strates failure to achieve the objec-tive performance criteria.

It is for this reason that calculationmethods ranging from “back-of-the-envelope” hand calculations throughthe most sophisticated computer-basedCFD models all have a place in the fireprotection engineering toolbox. A rela-tively simple hand calculation mayclearly demonstrate the achievement of,or the failure to achieve, a specifiedperformance criterion, in which caseuse of a more powerful tool would beunnecessary as well as uneconomical.Ideally, we would like to think that themore powerful the tool used for ananalysis, the smaller would be theuncertain middle area of the possibleoutcomes. The perfect analytical toolwould be one that eliminates theuncertain middle area, such that theresults of a calculation would eitherdemonstrate achievement of the objec-tive performance criteria or the failureto achieve it. This concept is illustratedin Figure 1. Unfortunately, the perfect

RIGHT TOOLfor the

JOBThe


analytical tool does not exist, so therewill always be uncertainty that must beconsidered.

ELEMENTS OF ENCLOSURE FIREMODELS

Before considering the analyticaltools available for quantitative fire haz-ard analysis, it is useful to consider theelements of enclosure fires. Based onobservations and measurements duringreal and laboratory enclosure fires, it ispossible to subdivide the complex andinterrelated phenomena occurring dur-ing an enclosure fire into a number ofdiscrete elements. As illustrated inFigure 2, these phenomena include:

• The fire source• The fire plume• The ceiling jet• The smoke layer

• The lower layer• Vent flows and mechanical

ventilation• Boundary heat transfer• Target heating and response

These are the phenomena that areaddressed with varying degrees ofsophistication and detail by differentfire models. To a large extent, the levelof detail with which a model treatsthese elements distinguishes one modelfrom another. In some of the simplermodels, the user must specify some orall of these elements, while in morecomplex models, the model calculatessome of these same phenomena basedon fundamental principles.

For example, most models requirespecification of a fire source; only afew attempt to calculate fire growthbased on fundamental principles.

Some of the simpler models calculateair entrainment into a fire plume basedon a particular correlation while moresophisticated models calculate entrain-ment based on the fundamentaldynamics of the fluid flow. Similarly, insimple models the ceiling jet is calculat-ed in terms of an available correlation,while in more complex models the ceil-ing jet develops as a natural part of thefluid flow calculations. The simplermodels can only be used reliably forconditions and scenarios where the cor-relation is reasonably accurate, whilethe more sophisticated models can beused with more confidence for a widerrange of scenarios.

TOOLS OF THE TRADE

Over the past 25 years, considerableprogress has been made with respect tounderstanding the dynamics of fires inbuildings. While models to simulateconditions in post-flashover fires havebeen around for even longer than this,efforts to understand and model theelements and interactions of pre-flashover fires began in earnest duringthe mid-1970s.

Models for evaluating the dynamicsof enclosure fires can be classified inthree categories, ranging from the sim-plest to the most complex:

• Hand calculations and correlations• Zone models• Field modelsHand calculations generally include

“closed-form” equations that can besolved directly without the need for iter-ation. For example, a number of corre-lations have been developed that permitthe calculation of flame heights, airentrainment into fire plumes, fire plumeand ceiling jet temperatures and gasvelocities, and smoke layer descentrates in closed rooms for fires withknown energy release rates. Whilecalled hand calculations because oftheir closed form, these calculations typically involve the evaluation of expo-nents, so they are usually performedwith the aid of a calculator, a spread-sheet template, or a basic computerprogram to make the process easier.

The best-known compilation of handcalculations in the United States is theFIREFORM (for Fire Formula) suitedeveloped at the Building and FireResearch Laboratory (BFRL) at theNational Institute of Standards andTechnology (NIST). While now morethan ten years old, this suite of calcula-tions is still widely used, either within

Increasing powerof analytical tools

Performance criteriaclearly achieved

Performance criteriaclearly not achieved

Not clear ifperformance

criteria clearlyachieved

Simple toolsadequate

Simple toolsadequate

More powerful tools may beuseful to reduce uncertainties

Targetheating

Ventflows

Lower layer

Fireplume

Mec

hani

cal v

entil

atio

n

Ceiling jet

Firesource

Boundaryheating

Smokelayer

Figure 2. Elements of enclosure fire models.

Figure 1. Ideal representation of calculation results as a function of the increasingpower of the analytical tools employed.


the FPETool program,3 as part of recentversions of the FAST program,4 or sepa-rately. Similar suites have been devel-oped in the United Kingdom, whereASKFRS8 was developed at the FireResearch Station of the BRE, and inAustralia, where FIRECALC9 was devel-oped at CSIRO.

The term “zone model” is generallyapplied to the many two-zone modelsthat have been developed over the past25 years. In a two-zone model, a roomis divided into at least two thermody-namic control volumes or zones, asmoke layer of buoyant gases andproducts of combustion that developsbeneath the ceiling, and a lower layerof relatively fresh air that remains nearthe floor, as illustrated in Figure 2.Conditions within each zone areassumed to be uniform, and the zonesare separated by a distinct interface.Mass, species, and energy conservationequations are applied to each zone todetermine the average temperature andsmoke composition within each zone.While clearly an idealization of whatactually occurs in room fires, the two-zone model has proven to be an effec-tive tool for estimating room fire condi-tions.

A large number of two-zone firemodels with varying levels of sophisti-cation have been developed interna-tionally. The first two-zone model toenjoy widespread use among practicingfire protection engineers was the ASET(for Available Safe Egress Time)model,10 which calculates the descent ofa smoke layer within a single closedroom as a result of gas expansion andair entrainment into an axisymmetricfire plume. Today, the most widelyused two-zone model is probably theFAST (for Fire And Smoke Transport)model,4 which calculates smoke layerconditions in a multiroom building.Both of these models were developedat NIST and can be downloaded fromthe BFRL Web site (www.fire.nist.gov).

The term “field model” is generallyapplied to the types of models moregenerally known as computational fluiddynamics (CFD) models in other disci-plines. In a field model, a space is sub-divided into a very large number ofcells, which may be in the hundreds,thousands, or even tens of thousands,depending on the desired resolution ofthe calculation and the available com-putational resources. The conservationequations for mass, species, energy,and momentum are then applied toeach cell along with appropriate initial

conditions and boundary conditions forthe calculation domain.

Field models generally demand muchmore in the way of computationalresources than zone models do. Whileonce relegated to advanced scientificworkstations, field model calculationscan now be run on many personalcomputers. Some practical applicationsmay take days or even weeks to run,but with the multitasking capabilities ofmodern computers, these calculationscan be performed in the backgroundwhile a computer is used for other tasks.They can also be performed over eve-nings and weekends, providing full em-ployment for otherwise idle computers.

A number of general-purpose com-mercial CFD models have been appliedto fire calculations. These models gener-ally require some modification to dealwith some of the unique aspects of fireproblems, including the large tempera-ture and density variations and gradientsin the fire plume, and the significantrole that thermal radiation plays at hightemperatures. These commercial CFDmodels typically include preprocessorsand postprocessors to simplify datainput and visualize computed results,but the cost for these commercial pack-ages can add up to tens of thousands ofdollars per year for a site license.

At least two CFD models are beingdeveloped specifically for fire dynamicscalculations. The first, known as SOFIE(for Simulation Of Fire In Enclosures),is being developed primarily by a con-

sortium of European fire research labo-ratories. SOFIE is a field model of thetype known as k-epsilon based on theway that turbulence is addressed by themodel. Results of an example SOFIEcalculation are illustrated in Figure 3.More information about the SOFIE pro-gram is available at the SOFIE Web site(www.cranfield.ac.uk/sme/sofie/).

The second model being developedspecifically for fire dynamics calcula-tions is known as the FDS (for FireDynamics Simulator) model. The FDSmodel is being developed at theBuilding and Fire Research Laboratoryat NIST. FDS is based on the concept ofLarge Eddy Simulation (LES). A recentpaper by Baum in Fire ProtectionEngineering11 provides an overview ofthe FDS model. The FDS model israpidly gaining popularity among prac-ticing fire protection engineers. Onereason for the growing popularity ofFDS is inclusion of the SMOKEVIEWprogram that permits the results of anFDS simulation to be visualized andanimated. These free programs providecapabilities normally associated withexpensive CFD and visualization soft-ware packages. Results of an exampleFDS calculation are illustrated in Figure4. Version 1 of the FDS model is cur-rently available for download at theBFRL Web site (www.fire.nist.gov/fds/).Version 2 is currently in beta testing.

Field fire models, including SOFIEand FDS, require considerable knowl-edge and substantial computer power

Figure 3. Results of an example SOFIE calculation.


to apply properly. While casual usersmay have some success running thesemodels, they require a high level ofexpertise to use them properly. Despitethese caveats, CFD models are the pro-grams of choice for situations whereadequate correlations do not exist tomake use of the simpler tools.

FIRE MODELING OR FIRECONSEQUENCE ANALYSIS?

Many, if not most, applications of firemodeling should more accurately becalled fire consequence analysesbecause the fire itself is not being mod-eled, only the expected consequencesof a specified fire are being calculated.Even the most sophisticated fire modelshave only rudimentary ability to modelthe actual development of a fire in anenclosure at the present time. Moretypically, the user specifies a fire ener-gy release rate history, which mayinclude an incipient period of lowpower output, a growth period ofaccelerating power output, a fairlysteady period of peak power output,and a decay period of decreasingpower output, as shown in Figure 5.

This approach has an obvious poten-tial pitfall. A user can specify an incor-rect fire history, resulting in an inaccu-rate analysis regardless of the accuracyof the actual calculation methodemployed. While there is not a simplesolution to this problem, it is importantfor both fire modelers and for thosereviewing fire modeling results tounderstand and accept the bases for thespecified fire histories.

While most current fire models maynot be able to accurately calculate thegrowth and spread of fire, many mod-els can and do address the influence ofoxygen on burning rates within anenclosure. For most typical com-bustibles that burn in fires, there is adirect relationship between the energy

released by the fireand the amount ofoxygen consumedfrom the atmosphere.Many fires become“ventilation limited,”which means that theirenergy release ratebecomes governed bythe rate of airflow tothe fire enclosurerather than by the fuelrelease rate. Modelsthat do not account forthis ventilation limita-

tion on combustion can produce inac-curate results.

MODELING UNCERTAINTIES

Models, by definition, are incompleterepresentations of the component, sys-tem, or process being modeled.Consequently, there will always beuncertainties associated with the resultsof a calculation. In general, theseuncertainties will be associated eitherwith the model itself or with the inputparameters. Modeling uncertainty is pri-marily epistemic because it relates tothe level of knowledge related to thephenomena being modeled. Parameteruncertainty is primarily aleatory as itrelates to random variations in theproperties or attributes of the inputparameters.

Enclosure fire models are a type ofdeterministic model because they pro-duce the same specific set of outputresults for the same input parameters.Only a few enclosure fire models haveattempted to address uncertaintythrough specification of probability dis-tribution functions for the input para-meters and in some cases for the mod-eling equations. The COMPBURNmodel12 that is widely used for nuclearpower plant fire hazard analyses isprobably the best known of these mod-els. Such models produce probabilitydistribution functions for the outputparameters by repeating the same cal-culations hundreds or thousands oftimes with the different input parame-ters in a process known as Monte Carlosimulation.

Whether uncertainty is addressedquantitatively or qualitatively, it isimportant that uncertainty is addressed.Perhaps the worst mistake a fire model-er can make is to run a model onlyonce and then rely on those results asconclusive without considering howvariations in input parameters might

affect the results. One of the benefits offire modeling is the ability to vary inputparameters over expected ranges ofconditions in order to evaluate the sen-sitivities of the results to variations ininput parameters. An exception to thiswould be for situations where an analy-sis is being performed for a “reasonableworst case” scenario. In this case, a sin-gle calculation might suffice providedthat the input parameters are truly rep-resentative of reasonable worst caseconditions.

GARBAGE IN – GOSPEL OUT?

Most of us have heard the phraseassociated with computers: “garbage in– garbage out.” While this phrase hasbecome trite from overuse, it isoverused because it remains so true.While it may be appropriate to trustthat a computer calculation will be per-formed accurately once a program hasbeen debugged and verified, this is noguarantee that the calculation result willaccurately reflect reality. If the scenariobeing modeled is not specified accu-rately, then the model will not producemeaningful results. On the contrary,results of a sophisticated computer-based calculation may provide the per-ception of a high level of accuracydespite producing erroneous results.This has sometimes been referred to as“garbage in – gospel out.” For this rea-son, it is critical that the assumptionsand uncertainties associated with theinput parameters be clearly describedand adequately justified.

Consider, for example, the practice ofusing a design fire heat release rate of5 MW for the analysis and design ofsmoke management systems in largespaces such as atria. Such a specifica-tion is widely used in building codes(e.g., 1997 Uniform Building Code and2000 International Building Code) andreferenced in design standards (e.g.,1995 NFPA 92B). While such a specifi-cation may be reasonably conservativefor the vast majority of applications, itis inadequate for those applicationswhere larger fires might be expected tooccur. Appropriate administrative con-trols are needed to limit fire sizes tothose used for analysis. Otherwise,“garbage in – garbage out.”

WHAT IS THE RIGHT TOOL FORTHE JOB?

It would be nice if there were a sim-ple and universal answer to the ques-

Figure 4. Results of an example FDS calculation.


tion: What is the right tool for the job?Unfortunately, there isn’t one. The righttool depends on the task at hand. It isthe job of the engineer to understandthe uses and limitations of the availabletools and to select a tool that is suitablefor the task. It is equally important thatthe engineer understands and docu-ments the bases and limitations for aparticular analysis.

It is easier to develop a procedurethat can be followed to help the engi-neer in the selection and application ofthe right tool than it would be to pro-vide specific guidance on which specif-ic tools apply to which scenarios. Thisprocedure would be:

1. Identify and document the pur-pose of the analysis in as muchdetail as possible.

Fire models are generally used todemonstrate that fire mitigation strate-gies will be effective before unaccept-ably hazardous conditions develop.For example, they can be used to helpdemonstrate that a large space can beevacuated before occupants becomeimmersed in the descending smokelayer. Or they can be used to evaluatewhen a fire detection device or sprin-kler would be expected to activate indifferent fire scenarios. Ideally, theycould be used to demonstrate theeffectiveness of a sprinkler systemdesign, although such capabilities aregenerally beyond the current state ofthe art.

For this first step, it is important thatall the different purposes for which ananalysis is being conducted are identi-fied. It is equally important that the

purposes are documented along withthe performance criteria that will beused to evaluate the acceptability of aproposed design. Sometimes fire pro-tection engineers take for granted thepurpose of an analysis because theyroutinely perform such analyses.However, the purpose of an analysismay not be apparent to other stake-holders, so this documentation isimportant.

2. Identify modeling tools that cansupport the level of detail demandedby the analysis.

In order to identify appropriatemodeling tools for a particular task, itis important for the fire protectionengineer to understand the uses andlimitations of the models that are avail-able for a task. For example, the wide-ly used model to predict fire detectoractivation, DETACT,13 is based onunobstructed axisymmetric plumes inlarge spaces with smooth horizontalceilings. The engineer needs to recog-nize what the effects of obstructedplumes, small spaces, and obstructedor sloped ceilings will be beforeapplying this model to such plumes orspaces. Where it is determined that theDETACT model does not apply due tocomplicated geometries, other moresophisticated tools may be needed,such as a CFD model, that can addressthese aspects.

Many analyses include a screeningstep during which simple tools are usedto perform conservative analyses.Dungan14 recently provided anoverview of the FIVE Methodology, arisk-based methodology used in nuclear

power plant fire safety analysis thatuses such a screening step. As illustrat-ed in Figure 1, simple screening toolsmay be adequate to clearly demonstratethat performance criteria clearly are, orare not achieved, for many scenarios.While the ultimate objectives of ananalysis may require the use of moresophisticated tools for scenarios that donot screen out with a simple tool,screening steps frequently permit manyscenarios to be screened so that themore sophisticated tools can be appliedonly when needed.

This step should also include consid-eration of how the tools being evaluat-ed have been validated and verified.Ideally, the modeling tools being con-sidered for use will have comparedfavorably with experimental data forscenarios similar to those being consid-ered. If not, some validation studiesmay be needed to develop confidencein the accuracy of the modeling toolsfor the scenarios being considered.

3. Determine what input parame-ters are needed for the models beingconsidered, and identify and docu-ment the data sources to be used asbases for the input parameters.

Most of the input parameters need-ed for different fire models are basi-cally the same. While they may bespecified in different terms, all enclo-sure fire models require specificationof the enclosure and ventilation open-ing dimensions, the boundary thermalproperties, and the fire properties,either in terms of a specified energyrelease rate or in terms of flammabilityproperties. While results of a simula-tion may be relatively insensitive tomany of these input parameters, theywill be relatively sensitive to others,notably the fire history.

For this step, it is important for thedata sources that are used to justify theselection of different input parametersto be identified and documented. Thiswill provide the justification for the val-ues selected as well as provide an audittrail so that others can confirm and ver-ify that appropriate values have beenselected. This documentation of inputparameter selection may also provide abasis for an uncertainty analysis.

4. Consider how model resultswill be used to support design deci-sions.

Finally, before a model is selected foruse, consideration should be given tohow the model results will be used. At

Incipient Growth Fully developed

Totalenergy

released(kJ)

Decay

TimeIgnition

Heat

rele

ase

rate

(kW

)

Figure 5. The stages of fire energy release rate.


this step, model results must be communicated to other, gen-erally nontechnical people. While such people may not beinterested in the details of an analysis, it is important for themto understand the limitations of the analysis so that they canmake better risk-informed decisions.

Fredrick Mowrer is with the Department of Fire ProtectionEngineering at the University of Maryland.

REFERENCES

1 Lawson, J. R., and Quintiere, J. G., “Slide-Rule Estimates of FireGrowth,” Fire Technology, Vol. 21, No. 4, 267-292, November 1985.

2 DiNenno, P., Editor-in-Chief, SFPE Handbook of Fire ProtectionEngineering, 2nd edition, National Fire Protection Association,1995.

3 Nelson, H. E., “FPETOOL: Fire Protection Engineering Tools forHazard Estimation,” NISTIR 4380, National Institute of Standardsand Technology, Gaithersburg, MD, October 1990.

4 Peacock, R. D., Forney, G. P., Reneke, P. A., Portier, R. W., Jones,W. W., “CFAST, The Consolidated Model of Fire Growth andSmoke Transport,” NIST TN 1299, National Institute of Standardsand Technology, Gaithersburg, MD, February 1993.

5 McGrattan, K. B., Baum, H. R., Rehm, R. G., Hamins, A., Forney,G. P., “Fire Dynamics Simulator: Technical Reference Guide,” NISTIR 6467, National Institute of Standards and Technology,Gaithersburg, MD, January 2000.

6 Rubini, P.A., “SOFIE – Simulation of Fires in Enclosures,”Proceedings of 5th International Symposium on Fire Safety Science,International Association for Fire Safety Science, Melbourne,Australia, March 1997.

7 SFPE Engineering Guide to Performance-Based Fire ProtectionAnalysis and Design of Buildings, Society of Fire ProtectionEngineers, Bethesda, MD, 2000.

8 Chitty, R., and Cox, G., “ASKFRS: An Interactive Computer Programfor Conducting Engineering Estimations,” HMSO, London, 1988.

9 “FIRECALC Computer Software for the Fire EngineeringProfessional,” CSIRO Division of Building, Construction, andEngineering, North Ryde, NSW, Australia, 1991.

10 Cooper, L. Y., “Mathematical Model for Estimating Available SafeEgress Time in Fires,” Fire and Materials, Vol. 6, No. 3-4, 135-144,September/December 1982.

11 Baum, H. “Large Eddy Simulations of Fires,” Fire ProtectionEngineering, Society of Fire Protection Engineers, No. 6, Spring 2000.

12 Ho, V., Siu, N., Apostolakis, G., “COMPBRN III—A Fire HazardModel for Risk Analysis,” Fire Safety Journal, Vol. 13, No. 2-3, 137-154, 1988.

13 Stroup, D. W., Evans, D. D., “Use of Computer Fire Models forAnalyzing Thermal Detector Spacing,” Fire Safety Journal, Vol. 14,33-45, 1988.

14 Dungan, K. W., “Practical Applications of Risk-Based Technologies,”Fire Protection Engineering, Society of Fire Protection Engineers,No. 10, Spring 2001.


INTRODUCTION

Towers 1 and 2 of the World TradeCenter in New York were both hit bycommercial aircraft on Tuesday,September 11, 2001. The aircraft hadbeen hijacked, and the incidents causedboth towers to collapse. Neighboringbuildings also subsequently collapsed,and there was damage over a wide area.

This was a deliberate act of terrorism,planned to create devastation anddestruction. Such events are not antici-pated, and current codes and designpractice do not address damage on thisscale being inflicted on buildings.Nevertheless, looking at the total col-lapse of both towers and its impact onsociety, we must see if we can dosomething in the way we design build-

FOREWARDArup has established an Extreme Events Mitigation

Task Force to help understand the events ofSeptember 11 in New York and Washington, andtheir consequences for the buildings industry.Results from this work will be made available to theindustry.

The following is based on a “briefing paper” thatwas prepared two weeks after the World Trade Centerdisaster by this task force. The briefing paper is post-ed on the Arup Web site and is being periodicallyupdated as more information becomes known.Members of the task force include:

Tony Fitzpatrick – Chairman, Arup Americas BoardRay Crane – Team Leader, New YorkAshok Rajii – Mechanical, New YorkBob Cather – Materials, LondonBrian Meacham – Risk/Fire, BostonCraig Gibbons – Structures, Hong KongDick Custer – Fire and Forensics, BostonFaith Wainwright – Structures, LondonJim Pinzari – Risk/Business Continuity, BostonJim Quiter – Fire, San FranciscoJohn MacArthur – Structures, New YorkLeo Argiris – Structures, New YorkMichael Willford – Structures, LondonRichard Hough – Structures, Sydney

This brief was prepared for a broad audienceinside and outside of Arup. Thus, some material willbe familiar to readers of this magazine. However, thebrief described many of the factors that need to beconsidered when evaluating how a building respondsto such an extreme event. Accordingly, it forms onepart of a broader risk assessment designed to adviseclients on individual risk, whether in a new buildingor an existing one. Prevention measures designed toreduce the probability of a terrorist attack will betaken into account in such an assessment.

A Briefing on the

WORLDTRADECENTER

Attacks

A Briefing on the

WORLDTRADECENTER

Attacks


ings to prevent this happening again,and we must explore the options andoffer the choices to society. These weremanmade structures – society makeschoices about how we design buildingsand how they should perform, but weare the professionals with the neededexpertise to inform the decisions.

The consequences of the September11 tragedy will be far-reaching, and asgovernments look at security and inter-national political actions, we need tothink how we will respond on our pro-jects. What are the options, and howmight they be pursued?

This report sets out some facts,issues, and questions, and suggestshow we might begin to address these.It was written shortly after the eventand will be further developed andupdated by the Extreme EventsMitigation Team Arup has established.

THE EVENT: WHAT HAPPENED?

Events on the morning ofSeptember 11, 2001, until the col-lapse of the WTC Towers 1 and 2

At 8:45 am, American Airlines Flight11 carrying 92 people was flown byhijackers into the north tower, calledOne WTC of the World Trade Centercomplex, hitting the tower on the northface around the 90th floor. At 9:06 am,United Airlines Flight 175 carrying 65people and also hijacked, hit the southface of Two WTC (the South Tower) ataround the 60th floor.

Both aircraft penetrated the buildingsand burst into flames. Large and wide-spread fires were observed followingimpact, involving both combustion ofthe fuel from the aircraft as well ascombustible materials typically found inoffice buildings.

At 10:00 am, 54 minutes after theimpact, Two WTC collapsed complete-ly. At 10:29 am, one hour, 44 minutesafter the impact on One WTC, it, too,collapsed to the ground. Many peopleescaped in the time that the buildingsremained standing, but 2,442 peopleare currently unaccounted for. Morethan 550 people have been confirmeddead as of December 20, 2001.

The planes were Boeing 767 200-ERs, which have a cruising speed of830 km/h (530 mph) at 10 km (35,000')and a fuel capacity of 90,000 l (23,980

Pictures courtesy of Paul Doherty and Henry Gifford.


US gallons). At this point in time, theimpact speed is not known. At an alti-tude of less than 600 m (2000'), it islikely that the air speed was less than500 km/h (300 mph).

It is reported that normal occupancyof the buildings would be 40,000 to50,000 total in the two towers. It is notyet known how many actually were inthe buildings; but significantly fewerwere likely there at that time in themorning.

On the same morning, the Pentagonwas also struck, causing nearly twohundred fatalities. Analysis of this eventis outside of the scope of this article.

Neighboring buildingsThe twin towers were part of the

World Trade Center complex of sevenbuildings.

In addition to these high-rise build-ings, there was a 47-story high-risebuilding (Seven WTC), a 22-story high-rise building (Marriott hotel), two 9-story buildings, and one 8-story build-ing. Excluding the hotel, most of theoccupied space within the buildingswas dedicated for office use. All thebuildings, except for Seven WTC, wereconstructed over a plaza area that con-tained a shopping mall, four under-ground levels of public parking, andtwo utility levels.

The entire center contained approxi-mately 1.25 million square meters (13.5million square feet) of rentable spaceincluding .18 million square meters (2million square feet) in Seven WTC.

The collapse of these neighboringbuildings was probably due to the

combination of the impact loads fromthe collapse of One and Two WTC,the tremendous fire that ensued, andfoundations being affected by the col-lapsing towers. Specifics of the col-lapse mechanism are being analyzedby many parties.

Falling debris also damaged build-ings beyond the World Trade Centercomplex. An assessment of thesebuildings was undertaken during theweek of September 17, by SEAoNY(Structural Engineers Association ofNY) and is available on the SEAoNYWeb site at www.seaony.org. Theassessment was based on an ATC-20Post-Seismic Building Assessment. Theaffected perimeter buildings containapproximately 1.5 million squaremeters (16 million square feet) ofspace. In all, more than 30% of down-town Manhattan office space has beenaffected.

Preliminary analysis of what led tothe total collapse of One and TwoWTC

The initial impact destroyed manycolumns and more than one floor. Howmuch was damaged may never beknown, but the impact itself did notcause the towers to collapse. The build-ings remained standing long enough toallow many people to escape.

The initial impact probably removedsome columns and significantly weak-ened many others. Deformed, andprobably unbraced over two or morefloors, these columns would also havebeen subjected to additional load fromdamaged floors and debris. The spread-ing fire would have continued to weak-en the structure, most likely leading toa collapse under vertical load.

The aircraft were heavily laden withjet fuel, having just taken off fromBoston en route to Los Angeles. The


resulting hydrocarbon-fueled fire wouldhave been much hotter and more rapidthan the type of fire against whichbuildings are normally designed.

THE BUILDINGS – DESCRIPTIONOF THE DESIGN

StructureThe Towers were steel framed, 110

stories high, and square on plan. Thebuildings were designed by Skilling,Helle, Christianson, Robertson, whichwas the structural engineering firm ofrecord for the World Trade Center com-plex, and completed in phases startingin 1970. One WTC was 417.0 meters(1368') tall, and Two WTC was 415.1meters (1362') tall. Each 110-story towerhad a floor plate 637 x 637 meters (209'by 209'). The central core in eachbuilding was 26.2 x 42.3 meters (86' x139'), constructed with steel columnsand lightweight drywall for infill.

Around the perimeter of the build-ings, 356 mm (14") steel box columnswere spaced at 1 meter (3'-3") on cen-ter, with 1.2 meters (48") deep plategirder spandrels at each floor. At thethird level, the columns transitioned inan arch-like formation to a 3.05 meter(10'-0") spacing for the lower story.Floors were supported by steel trussesspanning 18.3 meters (60'-0") from thecore to the perimeter wall on each sideof the building.

The building perimeter structure wasthe key element in the performance ofthe building. In addition to taking verti-cal gravity loads, it also resisted all hor-izontal loads by framing actionbetween the close-centered columnsand the spandrel beams, such that theperimeter structure acted as a piercedtube in resisting loads.

Vertical structure inside the buildingsupported gravity loads only.

Fire protectionThe World Trade Center was devel-

oped and constructed by the PortAuthority of New York and New Jersey,a self-supporting agency of the twostates. Although the Port Authority hasnot adopted any specific fire safetycode, it considers the requirementsdeveloped by the National FireProtection Association (NFPA), NewYork City, and the Building Officials andCode Administrators International

(BOCA) when developing designs. Thesteel frame had been clad in fire spray.Asbestos-based spray had been used ini-tially in the building, but later changedto a nonasbestos material. Further specif-ics, including which structural memberswere treated and to what level of fireresistance, are still being determined.

Blast resistance One WTC had already survived an

explosion in February 1993 when arental truck packed with explosives wasdetonated in basement level B2. Thatbomb blew out one section of a northtower cross-brace between two of theperimeter columns. The blast rippedout sections of three structural slabs inthe basement levels, but did little dam-age to the columns.

According to one of the originaldesigners, the towers were originallydesigned to take the impact of aBoeing 707. A member of the originaldesign team has clarified recently thatthis addressed impact only and not theconsequences of a jet-fuel fire.

THE QUESTIONS

How would our current buildingsfare under such an attack?

The answer is not simple. Structuresare designed to building and fire codes,and these documents reflect our soci-ety’s tolerance of risk from fire, earth-quake, floods, and a host of other haz-ards, and establish the minimum levelsof safety. Historically, building and firecodes have evolved in response to fireand other events that have affected

buildings, incorporating specific mea-sures intended to address lessonslearned. Because building and firecodes do not typically consider deliber-ate, willful attacks, such as arson or ter-rorism, these events have not played asignificant role in code development.

The current code-based designapproach on its own is inadequate foraddressing the question “how wouldbuildings perform under such anattack?” However, it is possible to lookat any building on a case-by-case basis.

To inform future design, perhaps itmay be better to turn the questionaround – “What are the implications fordesign if the building were required toremain standing for longer after suchan extreme event?” This is addressedbelow.

Is this a question just for tall buildings?

The World Trade Center’s promi-nence and symbolism must have been

THE WORLD TRADE CENTER

TYPICAL FLOOR PLATE


factors in why it was chosen as a tar-get. The Pentagon was also a target.As we face the question “will this affectthe future of tall buildings?”, we haveto recognize that tall buildings willalways be prominent structures on thecityscape. However, avoiding symbol-ism and expression in buildings canonly lead to a bunker mentality – andeven if this were pursued, would webe safe from determined terrorist acts?Wherever large quantities of peoplegather in one building or complex, beit for work or for play watching somesporting event, then the actions of peo-ple determined to kill and themselvesto die in the process are a completenew paradigm for design. The focusmay be on tall buildings initially, but itis not a question only for tall buildings.

How will this affect the approach tobuilding design?

To answer this, we first have to lookat the current approach to design andthen consider how we would address

the design for safety in such anextreme event.

CURRENT BUILDING DESIGNPRACTICE

Performance requiredCurrent building codes in the United

States are predominantly prescriptive innature. In other words, the code pre-scribes specific minimum requirementsfor parameters such as structural loads,fire resistance ratings, and egressrequirements. In some other countries, aperformance- or functional-based systemhas been adopted, wherein the overallperformance or function of the buildingand its systems are defined, but thedesign solutions are not mandated.Frequently these performance codescontain acceptable or deemed-to-satisfysolutions that serve as a prescriptiveoption to the performance approach.

A performance-based approach iscurrently being introduced into theUnited States as well, but for the pre-

sent time, any engineered or perfor-mance-based designs are typically usedonly under the “alternate methods andmaterials” clauses found in most UScodes. It is not unusual, for example, toundertake egress analyses for large orunusual buildings under the “alternatemethods and materials” clauses, takinginto consideration such factors as fireresistance ratings, sprinklers, compart-mentation, travel distances, and variousfire scenarios.

It is worth noting that to operate in aperformance-based environment, per-formance criteria and tools and meth-ods for assessing performance must beavailable.

Evacuation and refugeThe US does not have a single,

nationally enforced building code.Rather, there is a system of model build-ing codes that are adopted by statesand local jurisdictions, sometimes withmodifications. In some jurisdictions,such as New York City, there are also

DecoShield Systems, Inc.The Creative Cover Support System

COVERS IT ALL!!!

FOR MORE INFORMATION

&SAMPLES

CALL800-873-0894

OR VISIT OUR WEBSITE @

www.decoshield.com

CONCEALING sprinkler piping,plumbing, electrical wire and cablinghas never been easier, thanks to theUL-Listed DecoShield® Cover andSupport Systems.

We have been serving the fire sprin-kler retrofit industry for more than tenyears. Our system is clean, attractive,paintable, durable, tamper-resistant,economical and quick to install. Thatis why we are the #1 choice for retro-fitting all types of buildings.

Watts ACV Has It Now!Whether you need UL-Listed Pressure ReducingValves, Factory Mutual System approved and

UL-Listed Fire Pump Relief Valves, Factory MutualSystem approved Fire Pump Suction Control Valvesor UL-Listed Pneumatic, Hydraulic, or ElectronicallyActuated Deluge valves, Watts ACV has them in

sizes from 3” through 10”, in globe or angle pattern.

Whatever You Need in Fire Protection Automatic Control Valves...

For more detailed information, call us at:1-713-943-0688or visit our website at www.wattsacv.com.


locally developed codes. As a result,comments about “standard” US proce-dures are necessarily generalizations.

Having said that, standard practice isto evacuate automatically only thosefloors in the vicinity of the event (usu-ally floor of origin, floor above, andfloor below, although some jurisdic-tions require more). The respondingemergency professionals can thenmake a decision whether further evac-uation is necessary. The voice alarmsystem, which is much like a special-ized public address system, can beused to notify people on a selective orgeneral basis.

There are several reasons to do this:a. By evacuating only those in the

vicinity of the event, it is likelythat they can more rapidly enterthe stairs, rather than queuing atthe entrance while other nonaffect-ed people (from other floors) fillthe stairwell.

b. Crowd management is more effec-tive if only those directly at riskare moved.

c. In tall buildings, people can berelocated to alternate floors, ratherthan out of the building.

This approach allows those directlyaffected to leave their floors and enterareas of safety, such as pressurized fire-resistive stairs or other floors. Peopleremote from the event are assumed tobe safe, at least for some time.

Using this scenario, responding per-sonnel will generally go to the floorbelow (or some number of floorsbelow) the fire, either via elevator orstair, depending on the FireDepartment.

This approach clearly implies that cat-astrophic failure within a short period oftime cannot be allowed to happen.

SprinklersFire protection by automatic sprin-

klers assumes that there will be wateravailable at a sufficient pressure todeliver a spray of water to the fire area.

Sprinkler systems are not typicallydesigned to support water flow fromall sprinklers simultaneously, as thewater demand could be enormous andis typically not needed. The intent isthat the water supply system providesonly enough water for a specificdesign area, with a certain number ofsprinklers at the most hydraulicallyremote location. The density and

water supply volume is determinedaccording to the occupancy type ofthe building. The sprinklers would notbe designed to cope with a wide-spread aviation fuel fire.

The sprinklers themselves aredesigned such that each sprinkler headactuates individually. In typical officeoccupancy, each individual sprinklerhas a fusible link, or other reaction

device, which, when a defined temper-ature is reached, allows water to flow.The aim is that a sprinkler will actuatewhen a fire is relatively small, and theavailable water will control or suppressthe fire.

In the WTC towers, the initial colli-sion and explosion likely damaged aconsiderable part of the sprinkler sys-tem and fire protection water supply

The Soffi-Steel™ system is today’spremier modular steel pipe coveringsystem, designed to provide a durable,maintenance-free facing that is attractiveand easily installed. The Soffi-Steel™ system accommodatesall types of sprinkler devices includinginstitutional, recessed, and concealedsprinklers. In addition, it also functions asa shield for utility lines and HVAC. Soffi-Steel’s clean design and durabilitymake it perfect for the use in commercial/residential properties where the aestheticsof the shield are as important as itsfunction. Soffi-Steel™ is secure-it is virtuallyirremovable once assembled (Patent#5,526,617 and others pending).Installation service, training, and technicalsupport available at competitive pricing.

121 E. Burbank Ave.Janesville, WI 53546800/800-3213; Fax: 608/757-1452E-mail: [email protected]

• Secure• Attractive• Versatile• Economical• Durable


system on the floors of initial impact. Ifso, the sprinkler system would not havebeen able to operate as intended. Forthose parts of the system that remainedintact, the rapid growth and spread ofthe fire resulting from the jet fuel wouldlikely have actuated a large number ofsprinklers, taxing the available watersupply. In addition, if any fire protec-tion water supply risers were taken outon a lower floor, the water supply tofloors above may have been impacted.

Progressive collapse There are no explicit requirements

for design to prevent progressive col-lapse of buildings in the US modelbuilding codes. There is an under-standing that requirements in the codesdevelop a limited resistance to progres-sive collapse although this is not anexplicit design requirement. For zonesof high seismicity, US model buildingcodes, including the UBC, NBC, SBC,and IBC, all have special detailing pro-visions that have the goal of increasingductility and toughness in structuresthat reduce the possibilities of progres-sive collapse during seismic events.

The buildings were constructed bythe Port Authority of New York andNew Jersey, and compliance with theNew York City Building Code was vol-untary. There was no seismic detailingprovision in the code at the time theWTC was designed and constructed.

There are no explicit requirements inthe US model building codes or NewYork City Building Codes for debrisloading.

WHAT ARE THE IMPLICATIONSFOR DESIGN?

From a code perspective, the WTCdisaster would have to be consideredan unexpected event at the time it wasdesigned and constructed. The US has not previously been subjected tosuch a deliberate act of terrorism ofthis magnitude, and therefore thebuilding and fire Codes did not have adirect past history to draw upon. In theaftermath, it will be a societal decisionwhether we want to tolerate the risk ofanother similar event or whether wewant to accept the restraints associatedwith minimizing this type of event.

How do we begin to consider thebasis on which to make such decisions?

Design for robustness under suchan extreme event

The challenge here is that an eventsuch as the WTC collision involvesimpact, explosion, and fire all at once.The “design scenario” cannot be one ofa set of loads, but of a level of damageinflicted by whatever cause. One mightdevelop a set of “what ifs?” to try toenvisage a “worst possible” or “worstcredible” event. Then by posing possi-ble design options, we can ask, “If xxxhad been incorporated, would this pre-vent a collapse?”

For example:• If the steelwork at Towers 1 and 2

had not succumbed to the fire, thebuildings might still be standing. Whatif we incorporated additional fire resis-tance into the structure? How much isenough? Would traditional fire resis-tance have withstood the fire regardlessof the rating? Would fire resistance haveadhered to the steel given the impact ofthe aircraft?

• If the floors had not continuedunchecked in a progressive collapse tothe ground, then the buildings mightstill be standing. What if we incorporat-ed “collapse” stories every ten floors –a level of structure designed to carrythe debris load from the nine collapsingfloors above?

• If the towers had had a reinforcedconcrete core able to continue to pro-vide stability after the loss of the floorsand perimeter frame, then a shaft pro-viding a safe refuge might haveremained – should we consider dual-stability structures, where one standsafter the other has failed?

Phased or total evacuation Information is slowly being pieced

together about escape from the towers,principally from people who escaped.This will ultimately help in the discus-sion of phased or total evacuation.However, regardless of the intendedevacuation approach, some phasingwill occur naturally based on people’sperception of the threat, the presenceof loved ones in the building, and thelimitations for the flow of peoplethrough exit routes.

Total evacuation is not necessarilyalways the best answer and needs tobe considered at each event assessingwho is most at risk, how safe they arewithout evacuation, and access for fire-

fighters or emergency personnel tryingto enter the building. For example,with a moderate fire at low level thentotal evacuation would bring manypeople through the danger zone andinto the area of the firefighting effort.These are difficult decisions under dif-ficult circumstances requiring accurateinformation.

Way forwardDesign measures to mitigate the

effects of this and other extreme eventsare possible and, therefore, must beresearched and evaluated as a matter ofurgency. This requires an holisticassessment of all the issues discussedabove. The results can then be part ofthe decision process on the level ofrobustness to which a building shouldbe constructed.

In the United States, the next genera-tion regulatory approach, a perfor-mance-based one, is about to be intro-duced. This will be an important venueto facilitate the overall discussion ofrisk tolerance and the related social,technical, and financial impacts. Thisterrible event emphasized our vulnera-bility in dramatic fashion, but policy-makers should avoid a knee-jerk reac-tion in seeking to prevent a recurrence.Obviously, it is not just high-rise build-ings that are vulnerable to terroristattack; any change in codes will impacton many other occupancy types.

• www.sfpe.org • www.asce.org • www.seaony.org • www.ncsea.com• www.aisc.org• www.arup.com• www.m-yamasaki.com• www.newscientist.com• www.greatbuildings.com• www.skyscrapers.com• www.nytimes.com• www.usatoday.com• www.cnn.com• FEMA – Federal Emergency

Management Agency• SEAoNY – Structural Engineers

Association of New York• NCSEA – National Council of

Structural Engineers Associations: • AISC – American Institute of Steel

Construction.

RESOURCES


IBC Analysis Guide

The CADDY Loop Hanger SurgeRestraint (LHSR6) restricts theupward surge movement of activat-ed fire sprinkler systems to meetNFPA 13 and ensure proper spraypatterns. One clip fits 1/2 through2-in. loop hangers. The LHSR6grips the loop hanger and not thenut, allowing adjustment of thehanger on the threaded rod.

www.erico.com—ERICO®, Inc.

The latest microprocessor technology insmoke detection is available in theXP95-M Analog Addressable Multisensor,which is optimal for most generalapplications. The XP95-M includes bothphotoelectric and thermal sensing ele-ments integrated into a single device,designed to enhance performance,reduce false alarms, and provide quick,accurate response.

www.gamewell.com—Gamewell™ Worldwide

Rolf Jensen & Associates, Inc. (RJA) intro-duces From Model Codes to the IBC: ATransitional Guide, a publication aimed athelping architects, engineers, and buildersinterpret the new International BuildingCode (IBC). The book, written by RJA con-sultants and produced and marketed byR.S. Means, presents a convenient way tocompare IBC requirements to other modelbuilding codes.

www.rjagroup.com—The RJA Group, Inc.

Halon Replacement AlternativeA new fire extinguishing material– 3M™ Novec™ 1230 FireProtection Fluid – was devel-oped to be a long-term replace-ment for halon. It is designed tobalance the need for extinguish-ing performance, human safety,and low environmental impact.Its chemical structure is low inacute toxicity, making it ideal for use in occupied spaces.

www.3M.com—3M

The DSTA (Dry System Trouble Alarm) is amicroprocessor-based local supervisoryannunciator designed to monitor low/highair pressure and low room temperature of aDry Pipe Sprinkler System Riser (the pres-sure and room temperature switch is soldseparately). Each unit includes an LED display and an internal buzzer and may beriser- or wall-mounted.

www.pottersignal.com—Potter Electric Signal Co.

Advanced DetectionSystemsNOTIFIER’s new, eight-page AdvancedDetection brochure provides simple solu-tions for environments with complex firedetection needs – from sterile cleanrooms toharsh industrial environments. Written forend-users and specifiers, it highlights state-of-the-art detection and control technologies.It also includes application suggestions forthe full spectrum of NOTIFIER detection systems.

www.notifier.com—NOTIFIER®/Honeywell Home and Building Control

Loop Hanger Surge Restraint

System Sensor has released a MAC orWindows-compatible Applications GuidesCD-ROM, free to qualified individuals.The CD-ROM includes information onthe proper application of various detec-tion products in life safety and propertyprotection environments. It providescomprehensive data on every SystemSensor application, navigates easily, andis interactive.

www.systemsensor.com—System Sensor

Applications Guide on CD-ROMDry System Trouble Alarm

Analog Addressable MultiSensor

With these Internet-based InformationManagement solutions,Emergency ServiceProviders can communicate vehicle status throughout their departmentand simplify paperwork for vehicle and facilities management. Using aWeb browser, personnel can create, review, approve, and distributeinformation on vehicle or equipment repairs. Firefighters can updatethe status of any vehicle at any time.

www.manifolddata.com—Manifold Data Systems

Internet-Based Vehicle Management

Products/Literature


Remote Sensing Fire Alarm

The EasyPac Commercial Riser Manifoldcombines a flow switch, drain valve, sightglass, and pressure gauge into a lightweight,compact unit. It is available in five orificesizes ranging from 3/8 to 5/8 in., in sizes 2-6in., with a standard take-out dimension of 13in. There is a choice of end connections,and installation may be horizontal or vertical.

www.vikingsupplynet.com—Viking SupplyNet

This handbook outlines practical solutions forsecurity and access control applications. Thishandbook provides more than 40 differentcase studies, each including a description,parts list, diagram, and application tips. TheSecurity and Access Control Handbook is avaluable tool for engineers who design orspecify security and access control systems. (Security and Access ControlHandbook, A Practical Guide to Application and System Design, pub-lished by EST Press, an imprint of Edwards Systems Technology, Inc.)

www.est.net—Edwards Systems Technology (EST)

To address the problem of nonworking firealarms due to dead or missing batteries, FirstAlert has developed model SA302, which uses“smart sensing” and remote control test/silencetechnology. The SA302 can be tested orsilenced by anyone, using virtually any homeremote control device. A “smart-sensing”microchip also helps the alarm distinguish non-threatening conditions from real emergencies.

www.firstalert.com—First Alert

Viking has released a new color productbrochure entitled, “Your Fire ProtectionSolutions Partner.” A reference guide forfire protection specialists, it contains draw-ings of sprinkler systems suitable for train-ing. The sprinkler section contains a fullrange of products, with photographs andtechnical reference tables. Other areas cov-ered include foam and foam equipmentand other related fire protection products.

www.vikingcorp.com—Viking Corp.

More Commercial Riser Manifold Options

Fire Protection Reference Guide

The Security and AccessControl Handbook

Pyrotenax Cables Ltd. manu-factures mineral-insulated (MI)cable that has no organic com-pounds, so nothing burns, releases toxic fumes, or propagates flames.The Pyrotenax System 1850™ two-hour fire-rated power cable is usedfor fire protection of critical life safety circuits in commercial buildings;it arrives with a UL fire-resistive classification, so conduit or additionalfireproofing is not required.

www.pyrotenax.com—Pyrotenax Cables Ltd.

Fire-ResistiveCable

Established in 1939, Schirmer Engineering was the first independent fireprotection engineering firm to assist insurance companies in analyzing andminimizing risk to life and property. Schirmer continues to be a leader in theevolution of the industry, using insight from tradition and experiences of ourpast. Today, Schirmer Engineering is synonymous with providing high-qualityengineering and technical services to national and international clients.

Career growth opportunities are available for entry-level and senior-levelfire protection engineers, design professionals, and code consultants.Opportunities available in the Boston, Chicago, Charlotte, Dallas, Denver, LasVegas, Los Angeles, Miami, Phoenix, San Diego, San Francisco, andWashington, DC, areas. There is also a need for experienced loss controlengineers countrywide. We offer a competitive salary/benefits package. EOE

Send résumé to:G. JohnsonSchirmer Engineering Corporation707 Lake Cook RoadDeerfield, IL 60015-4997Fax: 847.272.2365e-mail: [email protected]

SENIOR PLANS EXAMINER(Fire Protection)

City of Pasadena – Fire DepartmentSalary: $59,120 up to $73,900

(effective 3/25/02)

Oversees the plan review and inspection process for fire protection; reports to the Fire Marshall; provides advice and technical assistance to citizens, staff, and fire personnel.

Qualifications: BS in fire protection or mechanical engineering or related field, plus two years’ experience performing plan review or fire protection design. Must possess an ICBO Plans Examiner and Uniform Fire Code Inspection Certification.

Apply immediately to: City of Pasadena, (626)744-4366.www.ci.pasadena.ca.us/humanresources/currentopenings.asp


The 4th InternationalConference on

Performance-BasedCodes and Fire Safety

Design Methods

Resources

Performance-oriented codes, regulations, and designmethods are gaining formal and widespread acceptance inmany countries. This three-day conference will again bringtogether professionals from around the world to discussthe state of the art in performance-based codes and designmethods. As in the prior international conferences held in1996, 1998, and the most recent held in June 2000 in Lund,Sweden, this conference will benchmark the rapidly accel-erating codification and implementation of performance-based design.

The first day of the conference will be dedicated to per-formance codes and will feature updates from Australia,United States, Canada, Japan and the UK on the status of

implementation of these codes and the challenges theypresent. On the second day, the theme of the conferencewill turn to design in the performance environment, anddesign aspects such as structural fire performance, smokemanagement, and reliability assessments will be explored.The third day is a workshop dedicated to an internationalcomparison of a standard building design using perfor-mance design methods in regulatory environments fromaround the world. A mini Trade Fair will be held in conjunction with the Conference.

REGISTRATION: To register, visit www.sfpe.org or contact SFPE at [email protected]

March, 20-22 2002, at the Melbourne Exhibition & Convention Centre in Melbourne, Australia.

February 5-6, 2002

Flame Retardants 2002London, EnglandInfo: www.intercomm.dial.pipex.com/fr2002cfp.htm

March 18-19, 2002

Structures in FireUniversity of Canterbury, Christchurch, New ZealandInfo: www.civil.canterbury.ac.nz

March 20-22, 20024th International Conference on Performance-Based Codes and Fire Safety Design MethodsMelbourne, AustraliaInfo: www.sfpe.org

May 19-23, 2002NFPA World Fire Safety Congress and ExpositionMinneapolis, MNInfo: www.nfpa.org

June 16-21, 2002The 75th International symposium on Fire Safety ScienceWorcester Polytechnic InstituteWorcester, Massachusetts, USAInfo: www.iafss.org UP

COMI

NG EV

ENTSUPCOMING EVENTS

What is The Society of Fire Protection Engineers (SFPE)?SFPE, established in 1950, is a growing association of professionals involved in advancing the scienceand practice of fire protection engineering and fostering fire protection engineering education.

What are the benefits of SFPE membership?The Society will provide you with many new opportunities for professional advancement, education,and networking. The specific benefits members receive are:

Free access to SFPE’s periodicals This includes:▲ Fire Protection Engineering magazine. ▲ The peer-reviewed Journal ▲ SFPE Today - Our bimonthly Society newsletter. of Fire Protection Engineering.

Substantial discounts on continuing educationThis includes:▲ Technical symposia on current fire protection issues.▲ International conferences on state-of-the-art applications of fire protection engineering.▲ Short courses and seminars offering hands-on instruction.▲Discounts on fire-related publications.

Other benefits include:▲ Recognition of your professional qualifications. ▲ Contribute to the profession through technical task ▲ Opportunity to participate in the SFPE Annual Meeting. groups and committees.▲ Opportunity to network in local chapters. ▲ A periodic profile of the fire protection engineer, ▲ Low cost group life, health, and liability insurance. including salary information.

Advancing the Science and Practice of Fire Protection Engineering

I’m interested in learning more about joining SFPE. Please send me additional information.

Name Title

Company/Organization

Address City State/Province Zip/Postal Code

Country

Work Phone Fax E-mail

Fax to 301/718-2242 ▲ Visit the SFPE Web Site: www.sfpe.org

For more information, contact The Society of Fire Protection Engineers:7315 Wisconsin Avenue, Suite 1225 West ▲ Bethesda, MD 20814

Phone: 301/718-2910

Societyof Fire ProtectionEngineers

An Invitation to Join


• Advanced Fire Technology, Inc..............Page 44• Ansul, Inc. ..................................................Page 2• Central Sprinkler ......................................Page 59• Commercial Products Group ..................Page 45• DecoShield Systems, Inc. ........................Page 50• Edwards Systems Technology ...........Page 30-31• Essex Fluid Controls................................Page 44• Fenwal Protection Systems .....................Page 34• Fike Protection Systems ..........................Page 53• Fire Control Instruments, Inc. ...................Page 8• Grice Engineering....................................Page 51• Koffel Associates......................................Page 38

• Marsh Affinity Group Services ..................Page 7• NOTIFIER Fire Systems...........Inside Back Cover• Potter Electric Signal Company ..............Page 28• The Protectowire Co., Inc. ......................Page 22• The RJA Group.......................Inside Front Cover• Reliable Automatic Sprinkler ..................Page 18• Siemens Fire Safety ............................Page 12-13• SimplexGrinnell .......................................Page 27• System Sensor ..........................................Page 37• Tyco Fire Products .............................Back Cover• Watts Regulator (ACV).............................Page 50• Wheelock, Inc..........................................Page 23

Index ofAdvertisers

CORPORATE 100

Solution to last issue’s brainteaser

The difference between any two numbers in the set{2, 3, 4} is equal to their greatest common factor. Thesame is true of any two numbers in the sets {6, 8, 9, 12}and {8, 9, 10, 12}. Find a set of five numbers for whichthis is true.

Answer: Two sets of solutions are {900, 912, 915, 918,920} and {1664, 1665, 1666, 1668, 1680} and the multi-ples of both sets.

Solve the following equation for x:

Thanks to Jane Lataille, P.E., for providing thisissue’s brainteaser.

B R A I N T E A S E RThe SFPE Corporate 100 Program was foundedin 1976 to strengthen the relationship betweenindustry and the fire protection engineeringcommunity. Membership in the programrecognizes those who support the objectives

of SFPE and have a genuine concern for thesafety of life and property from fire.

BENEFACTORSRolf Jensen & Associates, Inc.

PATRONSCode Consultants, Inc.Edwards Systems TechnologyHughes Associates, Inc.The Reliable Automatic Sprinkler CompanySchirmer Engineering Corporation

MEMBERSArup FireAutomatic Fire Alarm AssociationBFPE InternationalFactory Mutual Research CorporationFike CorporationGrinnell Fire Protection SystemsHarrington Group, Inc.HSB Professional Loss ControlHubbell Industrial ControlsJoslyn Clark Controls, Inc.James W. Nolan Company (Emeritus)Industrial Risk InsurersKoffel Associates, Inc.Marsh Risk ConsultingNational Electrical Manufacturers AssociationNational Fire Protection AssociationNational Fire Sprinkler AssociationNuclear Energy InstituteThe Protectowire Co., Inc.Reliable Fire Equipment CompanyRisk Technologies, LLCSiemens Cerberus DivisionSimplexGrinnellTVA Fire and Lifesafety, Inc.Underwriters Laboratories, Inc.Wheelock, Inc.W.R. Grace Company

DONORSGage-Babcock & Associates, Inc.

SMALL BUSINESS MEMBERSBourgeois & Associates, Inc.Demers Associates, Inc.Fire Consulting Associates, Inc.MountainStar EnterprisesPerformance Technology Consulting Ltd. Poole Fire Protection Engineering, Inc.S.S. Dannaway & Associates, Inc.The Code Consortium, Inc.

x xx x2 9 20

5 5 12

− +( ) =− +( )

Max. Storage Ht. Max. Ceiling Ht. K17 ESFR Min. Press. K14 ESFR Min. Press.Ft. (m) Ft. (m) psi (bar) psi (bar)

40 (12.2)* 45 (13.7) 63 (4.3) 90 (5.2)

35 (10.7) 40 (12.2) 52 (3.6) 75 (5.2)

25 (7.6) 32 (9.8) 42 (2.9) 60 (4.1)

25 (7.6) 30 (9.1) 35 (2.4) 50 (3.4)

*Indicates One Level Of In-Rack Sprinklers Required. See Tech Data Sheet 3-1.5 for other ESFR applications.

Central’s new K17 ESFR Pendent Sprinklers usher in a newgeneration of ESFR fire protection. The K17 ESFR offers an increasedlevel of flexibility for designers, while providing the fast, proven andeffective response of ESFR technology for warehouse and high-piledstorage fire protection needs. The K17 ESFR utilizes the same flows astraditional 14 K-Factor ESFR sprinklers, but at much lower pressures.The results are significant installation cost savings due to reducedbranch line pipe sizing and the potential elimination of fire pumps.

451 North Cannon Avenue • Lansdale, PA 19446Phone: (800) 523-6512 • Fax: (215) 362-5385

www.tyco-central.com

Central’s K17 ESFR Pendent Sprinkler for warehouse and high-piled storagefire protection. A generation aheadin ESFR technology.


By now, we have all read numer-ous reports regarding the per-formance of the Twin Towers

and the Pentagon and the occupantsof those buildings on September 11,2001. Some of these reports offerexplanations of what went wrong,what went right, and the changes thatshould be made in the way that build-ings are designed in the future. Someof these reports have a sound basis inscience and engineering, while othersdo not.

A fundamental tenet of engineering islearning from past events to prevent ormitigate the effects of similar events inthe future. SFPE has developed a three-phase plan for learning as much as wecan from the events that occurred onSeptember 11, 2001, at the World TradeCenter and the Pentagon.

The first phase is to document avail-able factual information and data. Thisincludes information such as the struc-tural frame of the buildings, installedfire protection, fuel loading, numberand distribution of occupants, impactlocation of the aircraft, etc.

The second phase involves filling inthe gaps in the information gleanedduring the first phase and developing abetter understanding of what actuallyhappened. Such an analysis will involvethe development of hypotheses andtesting those hypotheses via analyticalmethods such as modeling.

The third phase is be to develop alisting of the costs and benefits of thetypes of changes that might be consid-ered in how buildings are designed andconstructed in the future. While it is notexclusively up to the engineering com-munity to decide what, if any changesshould be made, the engineering com-munity has a responsibility to provideinformation on the costs and benefits ofany changes that might be considered,so that policymakers can makeinformed decisions.

Also, any changes must be consid-ered in a broad context. Prior toSeptember 11, a scenario involving thecollision of a jumbo jet with a buildingwas typically used as an example of thetype of high-consequence/low-probabil-ity scenario that would typically not beconsidered in the design of a building.Since September 11, numerous changeshave been made in the aviation indus-try to reduce the likelihood of similaractions in the future. Additionally, weneed to be careful to ensure thatchanges made to mitigate similar eventsdo not have negative consequences formore frequent events.

Designing to mitigate the effects ofthe types of events that occurred onSeptember 11 in the future could bringwith them significant costs – costs thatcould dissuade developers from build-ing structures of national significance inthe future. Therefore, it is important forthe engineering community to informdecision-makers as they considerwhether any changes in the design andconstruction of buildings are warranted.

While there is only preliminary infor-mation available at this time, it appearsthat the structures of both of the twintowers were damaged by collisions onSeptember 11. The resulting fires furtherreduced the load-carrying capability ofthe structures, which led to the subse-quent progressive collapses. If thesepreliminary findings are correct, itshows that the structural behavior andfire behavior were closely coupled.

It is too early to consider what, ifany, changes should be made to thedesign and construction of buildings asa result of the lessons learned from thistragedy. However, one lesson that canbe learned is the importance of taking acomprehensive, multihazard approachwhen designing buildings in the future.

Learning from Tragedy

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

from the technical director

The Flexible Onyx™ 640 PanelNeed flexibility in a fire detection panel? NOTIFIER introduces the Onyx 640 — the first panel in the newOnyx series of fire detection products.

The new Onyx 640 panel is:• Expandable — one loop to two, 318 points per loop• Scalable — can put multiple two-loop units into a

single enclosure• Networkable — stands alone or can be linked to

other panels• Backward compatible — works with existing

NOTIFIER devices• Voice ready — using the new Onyx XPIQ transponder

Plug into the Onyx 640 panel — first in the new Onyxseries of products from NOTIFIER.

Call (203) 484-7161 or visit www.notifier.com

™

At Tyco Fire Products, innovations such as the ESFR-1, K17 Storage Sprinklers, ESFR-17,ESFR -25 and EC-25 are just a few examples of ourcommitment to meet the dynamic challenges ofthe 21st century.

These applications will continue to change —and we’re committed to develop specific productsfor them.

For the latest in storage protection, visit yourlocal Gem or Star Independent Distributor. For theTFP Distributor nearest you, visit our website atwww.tyco- fire.com or call us at 800-558-5236.

FIRE protection engineering Winter 2002

Documents

Transcript of FIRE protection engineering Winter 2002