ML20078R744

Rev 1 to Methodology for Evaluation of Fire Resistance of Structural Steel

ML20078R744
Person / Time
Site:	Limerick
Issue date:	10/03/1983
From:	PROFESSIONAL LOSS CONTROL, INC.
To:
Shared Package
ML20078R738	List: ML20078R735 ML20078R744 ML20078R759 ML20078R775 ML20078R784 ML20078R786 ML20078R791
References
PROC-831003, NUDOCS 8311150278
Download: ML20078R744 (15)

Text

{{#Wiki_filter:-~ . _ _ . - . . . _ . . . s . _.__.. . , _ ._m.__ . _ . . . _ _ . _ . . _ __ _ ,_ __

 'C                                                                                                                                                                          '

, _f[ [ nofaded Lou Wwd, lu. C O METHODOLOGY FOR EVALUATION OF O FIRE RESISTANCE OF STRUCTURAL STEEL g C O O O l October 3,1983 Rev. 1 I* B311150278 831108 l PDR ADOCK 05000352 E PDR P. O. Box 446 e Oak Ridge, Tennessee 37830 e (615) 482-3541 l (.

O INTRODUCTION As part of' the defense-in-depth concept, passive fire barriers are employed .O, as well as active fire protection systems, manual fire fighting capability, and fire prevention activities so that fires will not prevent essential The basic codified require-plant safety functions from being performed. ment is for the fire barriers wt ich provide the separation between differ-O' ent fire areas to have a fire resistance rating of 3 hours. Strict compli-ance with the letter of this requirement entails coating or encasing struc-tural steel members which support ceiling / floor slabs. O The purpose of this methodology is to examine the ability of unprotected structural steel members to withstand the effects of fire involving the combustibles present in the fire areas they enclose. This is a two part problem: first the fire exposure must be determined and second the re-sponse of the structural steel must be assessed. The approach described below successfully treats the problem in a systematic method by assessing simple and conservatively realistic limitations on the combustion process, the resultant room environment and, finally, temeprature histories of the L structural steel. ASSESSING FIRE DEVELOPMENT The types of fixed combustible materials found in a nuclear power plant g which can burn in such a way to present a significant fire exposure to the The prevalent mate-general area in which they are 1ccated are very few. rials encountered in the plant areas analyzed were cable insulation and lubricating oil. The insulation and jacketing on the cabling in cable g trays is susceptible to ignition from internal or external sources and the heat output from a cable tray fire will affect the atmosphere of the room in which it is located. Lubricating oil is present in large pumps and cer-tain other types of plant equipment which can escape and burn. Other types O of combustibles contained within substantial metal enclosures (e.g. cabling in conduit and charcoal in filter units have been assumed not to contribute to fires. L 1 I

;L L

.t/ , The methodology for assessing fire development can be divided into three o different parts: limits on fire development, fire modeling techniques, and local heating effects. Each part will be examined in turn. LIMITS ON FIRE DEVELOPMENT in this section, practical limitations which govern the combustion process in .a room are discussed. This will include physical limitations on the combustion of -any fuel and fire test data regarding the burning character-istics of cable trays and combustible liquids such as lube oil.

  -O

1. -Ventilation Limited Fires One of the best understood and quantifiable limitations on the combus-tion process in a room is the case when the rate of burning is con-trolled by the flow of air into the room. The empirical equation g

given by Coulbert (2) for the rate of burning which can be supported by the fire induced air flow into a room through an opening of given size is: Q = 1580 nA % (. -where; Q- heat output (KW)- Ao -' area of opening (m2 )

                                         -Ho    height of opening (m)
   ' '                     The rate of burning is independent of the type of combustibles which are burning. The fire duration, until. room burnout, is the total heat value (heat of combustion times quantity) of all the combustibles in the room divided by the heat release rate of the fire.

j For the purpose of this analysis, the ventilation rate has been based on the available air flow through openings into the room such as door-ways. Fixed ventilation systems have been assumed not to contribute Q to- the ventilation rate. This is because in-duct smoke detectors are

                           .provided to shut down air hendling systems and/or fire dampers will actuate since the : temperatures achieved in ventilation controlled fires are well above the ratings of their fusible links.

p 2

    . (. -

    .'    . . g. . s ._ , .            -_ .i    _ . _ _ _ _ . , _ _                                       -_ _ . .

f'p

2. Fuel controlled Fires l.

C When excess air is available for combustion of the available fuel in a room, the fire will be fuel controlled. The predominant variables controlling the burning rate of such a fire are the surface area of combustibles which are involved and the heat outputs per unit area ,O which are characteristic of the types of combustibles involved. Test l- data' which provides realistic fuel surface area combustion rates for cable trays and lube oil will be discussed in the following sections. i

!O                  A. Cable Tray Fires The best available data on free burning cable trays containing hypolan and neoprene jacketed cables appear in the FMRC/EPRI (3) test reports. For these cables, a mass burning rate of 6.7 Kg/

C- ' min was measured for an array of 12 cable trays, ,each 8' long and 18" w'de. This reduces to a surface controlled burning rate of 0.1 lb/ min ft2 of cable tray or a heat release rate of 1000 BTV/ min ft2 (190 KW/m2). These results are realistic and compare C ' favorably with other test data on solid fuels such as wood cribs. B. Lube Oil Fires Lubricating oil escaping (spilling) from a pfece of plant equip-C ment at a rate sufficient 'to produce a signif' cant fire exposure to the room will form a pool on the room floor. Pool . fires, both large and small, involving liquid hydrocarbon fuels have been studied extensively. Hydrocarbon pool fires have been demon-strated to burn with a regression (surface recession) rate of 4.5 lO to 5 mm/ min (4). For, the purposes of this analysis, the more conservative figure has been used. C For a given room, the quantity of lube oil avai,lable to burn will be fixed. The size of a pool fire is determined by the fact that an equilibrium is reached where the oil leakage rate equals the mass burning rate (pool surface area times the surface recession C rate). Since the quantity of oil available to burn is known, the size of the pool fire (and, consequently, the pool fire duration) 3 (

3 . will be driven by the assumed rate of oil leakage from the pie ; of equipment. Since there is no way to predict the oil leak C. rate, a variety of ' cases are examined whenever a free burning pool fire is assumed to occur. The heat ' output from a pool fire is taken to be the mass burning g rate times the heat of combustion for lubricating oil (149,940 BTU / gal).

3. Flame Spread Rate

_o-The overall extent of a fuel surface controlled fire will be dependent upon the flame spread rate. For the purposes of this analysis, every fuel controlled fire has been assumed to achieve its maximum burning rate instantaneously and maintain that intensity throughout the dura-e- la this section, the methodology for determining tion of the fire. the maximum extent of a fuel controlled fire is explained. The spread of flames across liquid hydrocarbon pools from an ignition c point is very rapid and, for this analysis, has been assumed to be in-stantaneous. The heat output f rom a pool fire under consideration is taken to be constant throughout the duration of the fire. G By contrast, flame spread along horizontal cable trays is very slow. Available test data found in the FMRC/EPRI test reports (3) for hori-znntal fire spread in deep stacks of horizontal cable trays agrees with observations of the rate of fire spread in similar tray arrays in C The horizontal spread the Reactor Building at the Browns Ferry fire. rate for cable tray fires is 6 to 7 ft/hr. For this analysis, a more conservative figure of 10 ft/hr has been assumed. C Since the fire models employed to predict the temperature history of a j fire require a coastant heat output, it is necessary to determine what To maximum heat output will be attained in a spreading cable fire. L this end it is necessary to assume that the cable fire originate in the heaviest concentration of cable trays encountered in the area 4

 ,L l-

under consideration. The fire is assumed to spread out along any hor-g izontal cable trays that intersect the point of fire origin at a rate of 10 ft/hr and instantaneously up any vertical cable trays. The max-imum size of a spreading cable fire is determined by the extent of fire spread which has occurred when the original source of the fire g dies out. The time period required for the original source of the fire to -die out is determined by taking the average figure for cable tray combustible loading (lbs. of insulation per ft2 of tray surface)

             -in the area and dividing by- the mass burning rate for surface con-trolled cable tray fires (0.1 lbs/ min ft2 of tray surface).

o The resultant figure for maximum burning rate is applied as the con-stant heat output throughout the duration of the fire. This is a very conservative assumption since the fire would in actuality take time to

  .C build up to this level.           After the maximum level was reached and passed, the heat output during the remainder of the fire would be less since the total quantity of cabling that would be involved at any one time would be less. The assumption of a spreading cable fire is valid c

only as long as the room gas temperature remains below the auto-ignition temperature of the electric cabling (approximately 1100*F for cross-linked polyethylene insulated and Hypalon cabling).

  'O

4. Combustion Air Availability For buildings with minimal air leakage (infiltration) rates, such as l~

reactor buildings, the total amount of combustion which can occur in L 0 an area may be limited by the total air available in the building. Once the oxygen content in the building atmosphere is reduced below , critical levels,. flaming combustion can no longer be supported. The  ; fire duration,t , can be determined from the enclosure volume and heat g~ output of the fire using the following relationship' developed by Coul-bert (2): T = 29 Ve(m 3 ), in minutes Q(KW) iC This relationship is conservative since it assumes the 02 concentra-tion by volume drops to below 11%, when flaming ccmbustion generally ceases below 15%. 5 (

                                                                                           )

] 1 FIRE MODELING TECHNIQUES The methodology for analyzing the fire resistance of structural steel in a 2 given plant area requires that assessments be made of fire duration and heat output for the fire scenarios to be evaluated using the considerations discussed previously. This information is used along with data on the materials, geometry and ventilation openings for the compartment under consideration as input for simple and conservatively realistic fire models which predict the time-temperature history of the fire. Two fire models have been employed to determine the time-temperature course of the fire. These models should not be confused with sophisticated models such as the Harvard Fire Code which attempt to predict temperature profiles and gas concentrations throughout the room from ignition of the fire through its decay period. Since the heat output of the fire has been assumed constant throughout its duration and the only parameter of interest is room gas temperature, the problem has been greatly simplified.

1. National Bureau of Standards Methodology For fuel surface controlled fires, a realistic empirical method has been developed by the National Bureau of Standards (5) fcr estimating room temperature. This method calculates an upper level gas layer temperature based on correlation of more than 100 sets of test data to basic plume theory. Plume theory relates the ceiling layer tempera-ture rise, T, to the heat release rate Q and the height above the fire H as follows:

AT a Q 23/ . 3-1 73 g Using the premise that in a developing, fuel controlled fire, the fire behavior was similar to a plume the authors sought a power law rela-tionship relating aT to heat released and heat loss in the following form: AT -C Q hN hk Aw M gCp p To Ao Cp Acho 6 L

3. ,

Correlating this form to 112 separate experiments on a variety of fuels, the authors propose the convenient form: AT_= 480 X 1/23 . X2-1/3 O and X hk Aw Where X 1= WC pp To Aoho 2* U poC Aoip Ho The value of hk will be determined by the wall materials and thick-ness, and the fire development time. For a long slow developing fire in a thin walled enclosure, hk is best approximated by steady state O- g conduction through the wall and ceiling, hk*T. For a fast develop-ing short duration fi re in a thick walled enclosure, hk is best approximated using the thermal inertia (assumption of semi-infinite solid), hk=ioCK/t. p The semi-infinite solid approach is most appro-priate for nuclear power plant enclosures because of the massive bar-liers. Q and hk are inputed with the ventilation opening data to cal-culate XI and X2 and subsequently aT. C Since this methodology was empirically developed, its use was confined to scenarios which correlate with the range of fire situations on which it is based. This imposes two limitations. Fi rst, the fire situations on which it is based do not include very large rooms. Use g of this method was confined to the smaller rooms which were evalu-ated. Second, the authors state that the plume correlation on which the formulas are based does not hold for gas temperatures in excess of 1150*F. If a gas temperature in excess of this level is predicted g then the results using this model can not be used.

2. Heat Balance Method Writing a heat balance for the compartment is one of the most straightforward methods of determining the temperature course of a fire, especially when the heat output of the fire is assumed con-stant. A simplification of the method proposed by Babrauskas and Williamson (6) as modified by Berry (7) is used. Two conservative g

assumptions are made which allow this simplification: 7 (

                                                           .. . . - . ,                             ~

1. Radiative and convective heat losses through openings in the enclosure are negligible (see Berry 7).

7 ^

2. Heat loss through the walls will be dominated by the thermal inertia of the barriers, pCp K (assumption of semi-infinite slab approximation).

The heat balance equation can be described as follows: Q=1580Aok = heat release rate (kw) 4 Q=oAnt (Tg 4-T w) = radiant heat transferred to boundary

.O Q=      iupCok At (Tw-To) = conductive heat loss through boundary G

To get Tg as a funcion of t these equations can be solved to yield the following expression: Tg= Q + To+Q6 4 1/4 C aAtn AX t / where Q = 1580o A k K = 1/2 -{ w k p Cp'.

 ,0                   n   = function of emissivity of fire gases and boundary walls At = total heat loss surf ace area of boundary 0         This relationship is similar in form to that developed by Harmathy (8) except the heat release rate is defined by the ventilation factors. This is not an iterative process.           The formula can be used to determine the gas temperature at any time during the course of the fire.

These heat balance equations can also be used for fuel sur-face controlled fires by inputing the surface limited values 4 of Q. 8

Y . A conservative assumption that has been made in the application of both models is that no heat will be transferred through the ficor. r LOCAL HEATING EFFECTS The fira models discussed in the previous section are used to determine the overall gas temperature in the room. This is necessary in order to assess the effect of generalized heating of the structural steel. However, the heat evolved from burning cable trays and lube oil pool fires will not be instantly dissipated throughout the hot gas layer. In fact, there could be zones ' located above burning objects that are much hotter than the general atmosphere of the room. The conditions in the hot zones are assumed to be as they occur in free burning conditions in clean atmospheres. This is a conservative approach because cable trays immersed in hot smoky layers g~ , would receive much less oxygen than those in clean air.

1. Pool Fires Realistic relations for determining temperature distributions for fire C

plee have been developed by Heskested of FMRC (1). These relation-shi;s am aased on large scale fire tests involving a variety of solid and liquid fuels. These- relations can be used to predict plume tem-peratures at ceiling level assuming a pool fire occurring at floor g level. If this temperature proves to be above the critical tempera-ture for the steel, then the heating of the structural steel would have to be assessed. 3 Fire plumes are considered to have a virtual ori gin (point source) f rom which the plume can be. considered to emanate. A virtual origin has no physical meaning for fires involving most types of solid and liquid fuels. For liquid pool fires the virtual origin height, Zo, (relative to floor level) can be theoretically predicted using the following relation: . Zo = -1.02D + 0.083 Q.4 Where D = pool diameter (m) s 9 1

                                               --=    ~                    -,_ _ _ , _ _,_ , _.

C-Heskested gives a relation which can be used to determine the tempera-ture rise in the plume (above ambient) at any height in the plume.

  ~

This equation is used to determine the temperature to which the struc-tural steel located above the pool will be subjected. A To = 9.1 [T. /(gcp2 p.2)] 333 Oc.667 (z-zo ) -1.67 Where To = temperature rise in plume (*K) T. = ambient temt.erature (*K) g = acceleration of gravity (m/s2) g Cp = specific heat of air (Kj/Kg K)

p. = ambient density of air (Kg/m3)

Qc = convective heat flux in plume, Qc = .65Q, (KW) Z = height above pool surface

   .)    ,

2. Cable Tray Fires Cable tray fire test data was examined to establish temperature pro-files above burning cable trays. Tests performed by Sandia Laborator-j ies (9) and FMRC/EPRI (3) show that temperatures in the vicinity of 1500*F are reached in the flame region immediately above the surface of a burning cable tray. This temperature drops rapidly with increas-ing height above the surface of the cable tray. Temperatures in 9

                    . excess of the critical temperature for restrained composite steel beams are only encountered within one foot above the surface of a burning cable tray.

3 These figures have been used to assess the impact of localized heating of structural steel. Cable. trays located within one foot of the bot-tom of steel beams have been assumed to subject the beam to a constant temperature of 1500*F for the period of time it takes the tray to burn to completion (cable tray combustible loading divided by mass burning rate for tray fires). This approach incorporates several conserva-tisms. In general, the cable tray encountered within one foot of steel beams ran perpendicular to the beams rather than along their length. 'Thus, in reality, only a short (approximately two foot) sec-10

       . _ - _ -            ~

1 Y tion of the bea.n would be heated rather than tha entire length of the 1 beam. The cable tray fire test data shows that the peak fire tempera- t l C ture was maintained for a period of approximately 10 to 20 minutes. j The fire durations calculated using the cable tray combustible load-  : ings were generally in excess of 20 minutes. The temperature data used came from free burning fires in relatively clean atmospheres. In actuality, since cable trays must be located near ceiling level to affect beams, they will be immersed in a smoky layer with reduced oxy-gen content that would be less likely to support flaming combustion. O STRUCTURAL STEEL RESPONSE Once the time-temperature histories have been determined for the types of fires that could occur in the area under consideration, an assessment must be made of the response of the structural steel. The critical temperature 3 of a structural member is defined as the temperature at which it loses its load bearing capabilities. The critical temperature of a structural steel member depends on its physical properties, loading restraint, configura-tion, etc. Determination of the critical temperature for structural steel g members will be discussed in a section which follows. Heating of steel members will be examined first. Heating of Structural Steel Members 3 Since the mass of a steel member must be heated to its critical temperature before failure occurs there is a " thermal lag" behind the fire tempera-ture. If localized heating or room gas temperature results in the heating of structural members to temperatures in excess of their critical tempera-3 tures, then a calculation may be required to determine the heating history of the steel. If the. fire temperature greatly exceeds the critical temper-ature of the structural steel or the critical temperature is reached early g in the duration of the fire, then failure may be assumed by inspection. C 11 K.

C . Where the heating history of the structural steel must be assessed the un-steading state heat transfer calculation outlined by Stanzak (10) is used: 'G AT = 231 IU (Ta - Tj) at where AT = temperature rise in steel member during interval (*C) U = surface of steel member exposed to fire (m 2/m) O G = weight of steel member (Kg/m) Ta = average fire temperature during interval ('C) Ti = temperature of steel member at beginning of interval (*C) At = time interval in minutes O Determination of Critical Temperature The fire endurance of steel beam supported floor slab assemblies depends on the degree of composite action and end restraint of the t'eam. Composite 0 action between the beam and floor slab increase the fire resistance of the steel beam. End restraint of the beam likewise increases the fire resis-tance of steel beams. The combined effects of both composite action and end restraint greatly increases the fire resistance of the steel beam. C This can be expressed as an increase in the critical temperature of the steel. For the purposes of these evaluations, literative values for critical tem-perature of beams were reviewed and realistic criteria were estab-lished(11). The cases were as follows: Noncomposite action with end restraint - for noncomposite action un-restrainted beams the ASTM E-119 failure criteria of 1100*F average temperature and 1300 F local temperature are realistic and accepted. The effects of end restraint also have not been as well quantified by test as composite action has. For this reason, the conservative cri-teria of unrestrainted beams,1100 F average,1300*F local heating, 3 are used. Partial composite action with end restraint - test data indicate a range of critical temperatures from about 1300 F to 1600*F. For this

   '              evaluation an average value of 1477*F is used.
                                                            < 12

    ~

t

r. .

Full composite action --with' end restraint - test data indicate criti-cal temperatures above 1600'F. For. this evaluation, a value of 1600*F -

  ~

is~used. S O y [' 3 .' 13 [Q

7 REFERENCES

1. Gunnar Heskestad, " Engineering Relations for Fire Plumes," Society of C Fire Protection Engineers Technical Report 82-8.

2. Clifford D. Coulbert, " Energy Release Criteria for Enclosure Fire Haz-ards Analysis - Part I," Fire Technology, Vol.13, No. 3, August 1977.

3. FMRC, " Categorization of Cable Flammability, Intermediate Scale Fire Tests of Cable Tray Installations," Electric Power Research Institute, EPRI NP-1881, August 1982.

4. V. I. Blinov and G. H. Khudinkov, " Diffusion Burning of Liquids,"

U.S. Army Corps of Engineers Translation T1490, Moscow,1961.

 '3 ' 5. B.J. McCaffrey et.al., " Estimating Room Temperature and the Liklihood of Flashover Using Fire Test Data Correlations," Fire Technology, Vol .17, No. 2, May 1981.

6. V. Babrauskas and R. B. Williamson, " Post Flashover Compartment Fires," University of California, Berkeley, Report No. UCB FRG 75-1, C December, 1975.

7. D. L. Berry and E. E. Minor, " Nuclear Power Plant Fire Protection -

Fire Barriers (Subsystem Study Task 3)," SAND 78-1990, NUREG/CR-0468, Sandia National Laboratories, September 1979. C 8. T. Z. Harmathy, "A New Look at Compartment Fires," Fire Technology, Vol . 8, No. 4, November 1972.

9. W. H. Schmidt and F. R. Krause, " Burn Mode Analysis of Horizontal Cable Tray Fires," SAND 81-0079, NUREG/CR-2431, Sandia National Labor-atories, February,1982.

   )

10. W. W. Stanzak (translator), "The Calculation of the Fire Resistance of Steel Construction," National Research of Canada, Technical Transla-tion 1425, March, 1971.

11. ASTM STP 422, Symposium on Fire Test Methods - Restraint and Smoke, g American Society of Testing Materials,1967.

O f u. 14

                               -     -   -,                       - . . . . - , . a}}