Modern Steel Construction / December 2000

By Barbara Lane, Ph.D.

Performance-Based Design for

The standard fire resistance test

Fire resistance requirements in theUS building codes are based on thepresumed temperature profile andduration of a standard fire, asdescribed in ASTM E119. The testdetermines load bearing capacity (theability of a building element to con-tinue its function for a period of timewithout collapse), integrity (the pas-sage of flames or gases hot enough toignite cotton waste) and insulation(assuring the temperature on theunheated side of the element does notexceed 250°F).

In the test, a beam, column, wallor floor under its calculated designload is exposed to a standard firedefined by a prescribed tempera-ture/time curve. Programming thetemperature of a test furnace throughcontrolling the rate of fuel supplyachieves this curve. The fire resis-tance of the element is taken as thetime to the nearest minute, betweencommencement of heating and fail-ure under one or all of the criteriaoutlined above (load bearing capaci-ty, integrity or insulation). Periods offire resistance are normally specifiedas half hour, one hour and/or twohours up to four hours.

This test is based on methods firstdeveloped in the early 1900s whenthere was very little knowledge ofhow fires behave and their effect onstructural performance (AISI 1981).The standard fire test has been wide-ly criticized. The difference betweenthe standard test temperature-timecurve and temperature–time curvesmeasured in real compartment fires isconsiderable (see graph).

The graph shows the tempera-ture/time curve for the standard testcompared to real fire temperaturesfrom compartment fires with variouswindow areas. The differences areclear. The duration and severity of areal fire is not defined well as thestandard fire test curve. This figurealso shows that in many cases periodsof fire resistance are over-specifiedwhere the standard test results are

Effect of window area on fire temperatures during burnout tests with naturalventilation (SCI 1991).

applied, specifically where the decayphase of the real fires has begun butthe standard fire test curve stillincreases. This figure also illustratesthat the maximum fire temperatureswill vary as a function of the windowarea, or ventilation conditions of thefire compartment. This is not consid-ered when the standard fire test curveis assumed.

Real fire behavior

Extensive research has beendevoted to determine the factors thatcontribute to the behavior of fires inenclosures and examine the resultingstructural mechanisms. Real fires area function of the area and height ofthe compartment, ventilation provi-sion, type, configuration and quantityof fuel in the compartment andwhether or not sprinklers are provid-ed (Drysdale 1990). When levels offire resistance are derived from a firetest using a furnace with a definedtemperature-time curve these factorsare ignored.

Another major criticism of thestandard fire resistance test is thatsingle elements are tested althoughthey are normally designed as compo-nent parts of complex two or three-dimensional structures. This meansthe beneficial effects of load transferwhere a relatively hot member cantransfer load to a cooler memberthereby maintaining the overallframe stability is ignored.

The beginnings of change – thepractical evidence

No major structural frame failuresas a result of fire have been recordedin recent times. The standard fireresistance test methods are consid-ered effective for design require-ments. However, a growing belief hasarisen that the recommendations inBuilding Regulations throughout theworld are too conservative, lack cost-effectiveness and overlook many realfire behaviors that may affect the nec-essary levels of fire protection.

In Broadgate, UK, a large fire in amulti-story, steel framed building thatwas under construction and only par-tially fire protected caused the steelindustry to reconsider its approach tothe fire protection of structures (SCI1991). The structure was only partial-ly protected. Despite some largedeflections, the structure behavedwell and did not experience the fail-ure of members expected from stan-dard codes at the high temperaturesexperienced.

As a result of this real fire evi-dence, a major initiative to betterunderstand structural behavior inrealistic fires was established to refinecurrent fire protection levels.

A series of full-scale fire testsoccurred at Cardington, UK byBritish Steel and the BuildingResearch Establishment. The projectsteering committee also includedmembers from the Steel ConstructionInstitute and the University ofSheffield. Results from these testsshow that designing structural framesbased on single element behavior

does not give a realistic idea of whatframe behavior will occur in fire con-ditions (British Steel 1998). Normaloffice fuel loads were used in thecompartment tests and though tem-peratures in excess of the traditional1000°F were experienced by steelelements, structural collapse and/orbreaching of compartments did notoccur; a new design guide due thisyear by the Steel ConstructionInstitute will take account of thesenewly understood behaviors whendesigning for steel structures in fire.

There will always need to be arelationship between the standardfire test and real fire analysis becauseof the wealth of component knowl-edge available to designers from yearsof use of the test. However, fire engi-neers now investigate the relationshipwith a view to improving efficiencyand, even in the case of steel struc-tures, proving that added fire protec-tion material can sometimes beunnecessary.

Exposed steelwork supporting the North façade atrium enclosure at the AlfredLerner Student Center, Colombia University, New York City.

(photo by Esto P

hotographics, Peter M

auss)

Performance-based design solutions

A more rational approach to fireresistance is based on a “t-equiva-lent” analysis. This determines theheating effects of the actual fire loadon a given compartment’s construc-tion. The term “t-equivalent” meansthe exposure time in the standard fireresistance test that gives the sameheating effect on a structure as agiven real compartment fire (Law1991). Such features as fire load,ventilation and compartment dimen-sions characterize the compartmentfire.

The typical fuel load is considered,and therefore, the temperatures froman agreed design fire rather than thestandard furnace test can be used.The ventilation provision is calculat-ed along with the volume of the com-partment. The construction type ofthe compartment is determined and afactor incorporated based on its insu-lating properties.

This method is now part of theEuropean Building Codes. Other fac-tors can also be determined to takeaccount of the consequence of anuncontrolled fire, such as the proba-bility of fire occurrence and the ben-efits of sprinklers.

A factor was determined to addressthe consequence of an uncontrolledfire based on the following:

◆ Ease of evacuation (stair andperimeter evacuation, stairs only,evacuation time and simultaneousor phased evacuation) and

◆ Ease of fire fighting (internal andperimeter, internal only and firespread to adjacent compartments)

The comparative ease of evacua-tion and firefighting has traditionallybeen related to depth below gradeand building height.

Probability of occurrence factorshave been derived based on the num-ber of fires reported to the fire service(BSI 1997).

The benefits of sprinklers areincorporated by applying a factor of0.6 to the final fire resistance calcu-lated (Magnusson, Pettersson, andThor 1976).

An “equivalent” fire resistance canthen be calculated taking these para-meters into account. The t-equivalentmethod has a recognized scientificbasis and correlates well with com-partment fire test results (Eurocode1995)

This design method ensures realfire behaviour and its effect on struc-tures are addressed as well as provid-ing a useful link back to the familiarbenchmark of the standard fire test.

Benefits of performance-baseddesign solutions

By using performance-baseddesign methods, real fire effects areaddressed based on credible ‘worstcase’ design scenarios. This can leadto increased design freedom fromprescriptive code restrictions whilstmaintaining safety. Appropriate andcost-effective fire safety measures arederived. Practically, it can mean thatintumescent coatings can be specifiedrather than a cementitious spray orboard protection due to the reduced

fire resistance required. Other fireprotection systems can be utilized; forexample, using a sprinkler system togive an integrated sprinkler/watercooling system that keeps steel tem-peratures below the temperaturerequired to cause failure. It can alsomean that steel elements can be leftcompletely unprotected in large openspaces with low fire load areas: open-sided car parks, stadia, transportationterminals, high atrium spaces and soforth.

These design solutions have beenproposed, accepted and integratedthroughout Europe and are increas-ingly being proposed for the US.

Completed performance-baseddesigns

Arup Fire has used a performance-based design approach on projectsthroughout the world, and localauthorities throughout Europe,Australia and the US have approvedit. The following provide a briefdescription of some of the projectswhere Arup Fire has implementedthese techniques.

Modern Steel Construction / December 2000

(photo by Esto P

hotographics, Jeff Goldberg)

The Gathering Space at the Mashantucket Pequot Museum and ResearchCenter, Ledyard, Connecticut. This museum is a significant public building andhas been designed to reflect the culture of the Pequot Indians. The buildingcontains one very large sprinklered gathering space, 48 feet.

Modern Steel Construction / December 2000

Alfred Lerner Student Center,Columbia University, NY

This 240,000 sq. ft, $68 millionproject includes: a dual theater andcinema, a black box theater, studentclubs, 6000 student mailboxes, com-puter labs, a bookstore, the campusradio station and administrativeoffices.

Arup Fire examined the need forfire protection to the exposed steel-work supporting the north façadeatrium enclosure. By adopting a per-formance-based design approach,combining hazard analysis with realfire behavior modeling and thenusing this to analyze the structuralresponse in a fire, additional passivefire protection was not required forthe design fires analyzed.

This approach was presented tothe New York City BuildingDepartment who granted a waiverfrom the requirements of the NYCBuilding Code.

Mashantucket Pequot Museumand Research Center, Ledyard, CT

This museum, a significant publicbuilding, has been designed to reflectthe culture of the Pequot Indians.The building contains one very largesprinklered gathering space, 48’ high,which can be used as an auditorium,exhibition space or for receptions.Arup Fire examined the need for fireprotection to the exposed steelworksupporting the glazed wall and theroof. Two hours of fire resistance wasrequired by code for this steelwork.

A flashover fire was not a concerndue to the height of the space and theprovision of a smoke control system.The study concluded that the fire sce-nario of interest was a fire local to theglazed wall. Of particular interestwere the members, which tie the wallto the horizontal arch. For the analy-sis, it was assumed the membersforming this truss were engulfed inflame. It was also assumed that theinclined steel ties were fully engulfed.

(photo courtesy Ove A

rup & P

artners)

Hong Kong Air Cargo Terminals Ltd. (HACTL). Butterfly wing truss concept forthe steel roof system over the container storage system. Sprinklers were pro-vided in the steel tubular structure ensuring that surrounding steelworkremained well below the calculated failure temperature.

The support structure for the Credit Lyonnaise, above the station, on the left.The station structure beneath is strengthened to survive a fire but without intro-ducing cladding.

(Photo courtesy P

hoto Poteau)

the current codes, have been consid-ered, plus data from fire tests and realfire incidents, resulting in a revisedapproach to required periods of fireresistance. Now it is possible to pro-pose a performance-based definitionof fire resistance that relates to fireload, compartment size and ventila-tion, as well as the ease of evacuationand firefighting rather than relyingsolely on the standard fire resistancetest.

Barbara Lane, Ph.D., is a FireStrategist with Arup Fire in New YorkCity

The fire load analysis demonstrat-ed that the fire load in the gatheringspace was enough to support an un-sprinklered fire local to the wall for15 minutes. The heat transfer to theunprotected elements engulfed inflame was calculated for this period.This demonstrated that the tempera-ture increase in the steel elementswas not enough to cause failure.

By using performance-baseddesign techniques, it was shown thatthe intent of the code was being metwithout the need for additional struc-tural fire protection.

HACTL SuperTerminal 1, ChepLap Kok, Hong Kong

The photo of HACTL Super-Terminal 1 shows the delicate butter-fly wing truss concept for the steelroofing systems over the containerstorage system at Chep Lap Kok air-port, Hong Kong. The Hong KongFire Services Department agreed to atwo-hour fire rating for the roofstructure after prolonged negotia-tions. Intumescent paint on the rela-tively slender members clearly couldnot provide the necessary rating, andcladding them with fire-rated boardwould be unattractive; a water-filledtubular system was proposed.However, the thermal capacity wasinadequate unless the water was cir-culating. Sprinklers were provided inthe steel tubular structure itself, andit was connected to the fire servicedrencher system. In the event of afire, any activated sprinkler wouldactivate receiving cold water, ensur-ing that surrounding steelworkremained well below the critical tem-perature. The Loss PreventionCouncil carried out full-scale tests,and Arup Fire and the Hong KongAuthorities witnessed them to provethis concept worked as calculated.Steel temperatures did not exceed40°C even with a full propane gas fireload.12

Lille TGV Station, France

The Lille TGV Station extendsalong Quai Londres at Lille TGV withthe support structure for the CreditLyonnaise on one side. French regu-lations require two hours fire resis-tance for structures adjacent to high-rise buildings to prevent fire spread.Therefore, the roof and structuresupporting the roof at Lille TGV sta-tion lying underneath the WorldTrade Center and Credit Lyonnaisetowers required two hours fire pro-tection.

Rather than use a cladding system,a fire engineering study was carriedout and design fires were developed.An analysis of the effect of these fireson individual steel elements thendetermined which elements wouldfail. An analysis of the total structurein the fire load case then occurred,assuming the failed members. It wasassumed the steel members formingthe arches failed and that the archcarried no load.

The subsequent fire-engineeringanalysis showed the columns support-ing the roof structure required addi-tional fire protection to withstand thedesign fires but for a period of 30minutes. Applied fire protection wasnot desirable to the design so thethickness of the steel in the columnswas increased by 5 to 10mm wherethe additional fire protection wasneeded.

In this way the standards requiredwere achieved in that the structurewould survive the fire but withoutcompromising the architectural uni-formity of the column lines.

Performance-based design tech-niques are increasingly being accept-ed as a rational approach to fire safe-ty. During the past few years a num-ber of unique structures have beenthe subject of fire resistance analysesand have not required added fire pro-tection. For other structures, reducedlevels of fire protection are accept-able. These specific analyses, as wellas background data used to develop

Selected ReferencesAISI, Fire protection through Modern Building Codes,

5th ed., 1981.British Standards Institution (BSI), Fire Tests on

Building Materials and Structures, BS 476 Part20 - 23, 1987

BSI Draft for Development, Fire safety engineering inbuildings, Part 1 – Guide to the application of firesafety engineering principles, DD 240: Part 1:1997.

British Steel. The Behavior of a Multi-storey SteelFramed Building Subjected to Fire Attack -Experimental Data, Swindon Technology Center,1998

The Building Regulations 1991, Fire Safety,Approved Document B, HMSO 1992

CIB W14 Workshop. “Design Guide – Structural FireSafety,” Fire Safety Journal, March 1986.

Drysdale, D. D. An Introduction to Fire Dynamics,Wiley and Sons, 1990

Eurocode 1: Basis of Design and Action onStructures, Part 2.2 – Actions on StructuresExposed to Fire, ENV 1991-2-2:1995, EuropeanCommittee for Standardization

Law, M. “A Review of Formulae for T-Equivalent,”Fire Safety Science, Proceedings of the FifthInternational Symposium, pp. 985–996,International Association for Fire Safety Science,1997.

Magnusson, S. E.; Pettersson, O.; and Thor, J. Fireengineering design of steel structures,Publication 50, Swedish Institute of SteelConstruction, Stockholm, 1976.

Steel Construction Institute (SCI). Structural FireEngineering: Investigation of Broadgate Phase 8Fire, 1991.