ML20078R791

Speech Entitled Engineering Relations for Fire Plumes, Presented at Soc of Fire Protection Engineers Fire Protection Engineering 820517-20 Seminars in San Francisco, Ca.Addendum & Correction Page Encl

ML20078R791
Person / Time
Site:	Limerick
Issue date:	11/08/1983
From:	Heskestad G FACTORY MUTUAL RESEARCH CORP.
To:
Shared Package
ML20078R738	List: ML20078R735 ML20078R744 ML20078R759 ML20078R775 ML20078R784 ML20078R786 ML20078R791
References
TR-82-8, NUDOCS 8311150299
Download: ML20078R791 (73)

Text

{{#Wiki_filter:s.. . y. n a .- ~ "=e=* *;. ' - -.* O g ea=W, + - -- % q,, ~,,., ~__,,p... s* .y

• 4,

=, t ' +,. - ,t ...s, ;,, ,. ~,. .,i : _',,'e-. ; ', i s y,_

pm.,..r-

c....,y p.. - p,. k., f,.y; t ge, a:.,

.g.sj. .,e (s. '. n.: ,a, M o,, f.",e ! ' J.f l,.s,, 's

• .~

,,,,,.... - ;f;h- ~,q, y [r,g - }*jp,.,'[. g g_ (** L ';u ~ ~. ~,[ - * * *, j Td. a ,,", j f ;r _ ' ;, g y g. ,a ~.,, . ~.-**-... -...~ n n.a ;. z. -,,., n.?. ',,: ~ r: ~.. .x, , "so h. *r.hy.N,.~,.e l *4 I -i . ', ~ ' _ rf -. .'. 7,* ~,. ...- Q h L,, 5 s Q, 4,*., ;.7,,. .js u . r.,., ;..,.,. s..y g...r.. v w +# ',...,,.... t.- .w#*- a-W^ .~ Y. l p -74,'~,,.f'^. -}l.Q, p. j*)'=Ol ' ' ^ *,. '.7 ',' , &;.'T* \\, ' V : .'.h t p a

• i.

.3 ,.9 _Y-p %., ;.t,.f s y. f:. a m.u.3,,; ...,,, y r= { + ,..,,.. y. y. y<.,a.. t ce, ... a. m,. 41.. p? l +, f* g). [ *. ., :: + l - y 1,- .s ?* l ~ SFPE Technology Report 32-8 3: i ENGINEERING RELATIONS FOR FIRE PLUMES FOR OJL FtRirs ~ ^ Gunnar Heskestad, Ph.D. 'h' Principal Research Scientist Factory Mutual Research Corporation .c Norwood, Massachusetts, e s u. v SOCIETY OF FIRE PROTECTION ENGINEERS 60 Batterymarch Street Boston, Massachusetts 02110 Price 54.65 8311150299 831108 {DRADOCK 05000352 PDR U =

e ADDENDUM TO TR 82-8 g NOMENCLATURE PumeradiustopointwheretemperatureriseisfAT,(m) l

bAT b,.

plume radius to point where velocity is f u, (m) 3 mean volumetric concentration (volume fraction) of gas species 1 (m 7,3) Cg 'c specific heat of air (kJ/kg K) p ,D diameter of fire source, or effective diameter such that FD /4 = area of fire source (m) 2 g acceleration of gravity (m/s ) H heat of combustion (kJ/kg) I intermittency (-) L mean flame height (m) M molecular weight of air (-) M molecular weight of gas species 1 (-) g A mass flow rate in plume (kg/s'; s mass burning rate (kg/s) g s mass generation rate of gas species i (kg/s) g N nondimensional parameter defined in eq (2) (-) Q AH, total heat-release rate (kW) Q convective heat flux in plume (kW) R ratio of temperature rises (-) mass stoichiometric ratio, air to volatiles (kg/kg) r T, mean centerline temperature in plume (K) T, ambient temperature (K) AT mean temperature rise above ambient (K) g AT, value of AT on plume centerline (K) i f l

X ADDENDUM TO TR 82-8 t time (s) t, virtual ignition time (s) u, mean velocity. on plume centerline (m/s) u,, maximum value of u,, near flame tip (m/s) z height above top of combustible (m) g height z where AT, = 500 K (m) .z z, . height of virtual origin above top of combustible (m) a fire growth coefficient for parabolica11y growing fires (kW/s ) ( nondimensional parameter defined in eq (9) (-) 3 p, ambient density (kg/m ) e 9 O

,_. :~.. .1. Correction to TR 82-8 Page 6 Ex. 9 should read T /5(e p ) 1/5 2 p u, E g /5 (AT Q )1/5 = 2 o c 9 Page 9 Ex. 18 should read i s = 0.0'054 Q z/(0.166 Q 2/5 4,o) e c (Z < Z g) A e e t

. g. ...w ENGINEERING RELATIONS FOR FIRE PLUMES (Originally presented at the 1982 SFPE Fire Protection Engineering Seminars, Moscone Convention Center, San Francisco, California, May 17-20, 1982) Gunnar Heskestad, Ph.D. Principal Research Scientist Factory Mutual Research Corporation Norwood, Massachusetts ~ Int roduction portant in the design of fire sup pre s-sion systems. Practically all fires go through a ma-jer, initial stage where a coherent, buoy-This paper deals with axisymmetric, ont gas st ream rises above a localized turbulent fire plumes and reviews a number craa undergoing combustion into surround-of relations for predicting the properties ing space of essentially uncont erinated of such plumes. It is assumed th roughou t cir. This stage begins at ignition, con-that the surrounding air is uncontaminated tinues through a possible smold ering in-by fire products and that it is uniform in tarval, into a ' flaming interval and may be t empera ture.. The relations cease to be said to end when the sur rounding space valid beyond the elevation where the plume flashes over. The buoyant gas st ream is enters a smoke layer. hclways turbulent, except when the source is very small and smoldering. If the to urce is a flaming one, the flames too Fire Plume Features will be substantially turbulent, provided thm source is suf ficiently large (perhaps Figure 1 is a schematic representation 0.3 m in diameter, or larger). The buoy-of a fire plume originating at a flaming ant flo w, including any flame s, is re-s ourc e. Volatiles driven off from the ferred to as a fire plume, combustible, by heat fed back from the fire, six with the surrounding air and The properties of fire plumes are im-form a "dif fusion flame." The mean height portant to know in dealing with problems of the flame is denoted L. Surrounding r21sted to fire detection, fire heating of the flame and extending upward is a bound-building st ruc t ure s, smoke-filli ng rates, ary which confines the entire buoyant flow and ' fire ve nti ng. They can also.be in-of combustion products and entrained air. AB STRACT - This paper presents a number of engineering relations drawn from the literature for calculating properties of fire plumes. Plume properties considered include flame ~ heights, - temperatures, velocities, concentra tions of combustion products, and entrain-i cant rates of air f rom the surroundings. In addition, a brief discussion is presented on ' the effect of fire growth to demonstrate the validity of the relations set forth. A note cn virtual origin is also included. --.-w a- -vv v-e=' ---'-Fw^= t e e'--'t- 't -T T--*

• --v e

+ =-r -r ,<c..- i-_r, e ie y

~ Z Z + u g I -w-- h g l 1 i i f I \\ \\ \\ + \\ l l + \\ \\ I \\ Entrained g AT tu o Io Flow j + \\ I \\ \\ Flow I / Profile \\ g + \\ I \\ + \\ l \\ \\ I \\ \\ / I n \\n/i ~ ) ~ Flame a L ( y o = AT ;u g g FIGURE 1 Features of a fire plume, including axial variations on the centerline of mean excess temperature (AT ) and mean velocity (u ). G G 2 SFPE TR82-8

The air is entrained acro ss a very sharp is carried away by the fire plume above g and highly convoluted bound ary, easily the flames, while the remainder of the LF discernible in smoky f i re s. The flow total heat liberated is radiated away in profile could be that of time-averaged all directions, values of temperat ure rise above the am-bient tempera ture, or of the concentration The total heat-release rate (Q) is of-cf a gas (such as CO ) generated by the ten assumed to be equal to the theoretical 2 fire, or of the axial velocity in the fire heat. release rate, which is based on com-plum e. plete combustion of the burning material. The right side of Figure 1 suggests, The theoretical heat release rate (kW) -qualitatively, how the temperature rise on is evaluated as the mass burning rate the centerline, T, and the velocity on (kg/s) times the (lower) heat of complete o the centerline, u, might behave along combustion ( kJ /kg). The ratio of the to-o the plume axis, based on experimental tal heat-release rate to the the ore tical observations [1, 2, 3]. In this example heat-release rate, which is the combustion with a relatively tall flame,. the tempera-efficiency, is close to unity for some t ure s are nearly constant in the lower fire sources [4] (e.g., methanol and hep-portion of the flame, then begin to decay tane pools), but may deviate significantly in the int e rmi ttent, upper portion of the f rom unity for others (e.g., a polystyrene flame as the combastion reactions trail fire, for which a combustion ef ficiency of cff and. air entrained f rom the surround-a bout 45 percent has been measure d [4 ], ings cools the flow. The centerline ve-a nd a fully involved stack of wood pal-

lecities, u,

tend to have the ir maxima lets, for which a combustion ef ficiency of o naar the mean flame height, always decay-63 percent has been measured (4]. ing toward higher elevations and usually decaying toward the top of the combust-

gible, if the combustible is porous and Flame Heights cupports internal combustion, there may not be as pronounced a f allof f in the gas The visible flames above a fire source v21ocity toward the top of the combustible trace out the space where combustion reac-en suggested.

tions are occur ri ng. Typically, the lower part of the flaming region appevrs fairly The total heat generated (Q in this steady in luminosity, while the upper part p;per) is either, convected or radiated appears to be intermittent. The int erm it-eway f rom the combustion regio n. If un-tent flaming region is associated with burning material s urrounds the combustion shedding of vortex st ruc tures [6, 7, 8 ]. z:ne, as in the case of a fire deep in a porous combustible pile (e.g., a stack of Figure 2 helps to define the mean wrod pallets), some of the total heat gen-flame height, L, [8]. The figure shows srated is trapped by, and st ored in, the schematically the variation of int erm it-unburning material; the re st escapes f rom t enc y, I, versus dist anc e above the fire the combustible array as either convective s ourc e, z, where 1(z) is defined as the er radiat ive energy flux. If a major f raction of time that at least part of the f raction of the volatiles released under-flame lies a bove the elevation, z. The gces combustion above the fuel array, as intermittency decreases from unity deep in in pool fires of liquids and other hori-the flame to smaller values in the inter-z ntal-surface fire s, even well-developed mittent-flame region, eventually to reach p:rous pile fire s, then the convect ive zero. The mean flame height, L, is the f rcction of the total heat-release rate is d istanc e above the fire sourc e where the rarely measured at less than 60-70 percent intermittency has* declined to 1/2. Obj ec-of the total heat-release rate [4, 5]. tive determinations of mean flame height The convective flux (Qe in this paper) according to intermittency measure ments L SFPE TR82-8 3 -,,-.nm --w

I l.0 I ( 0.5 = L 0 z(Arbitrary Units) FIGURE 2 Definition by Zukoski et a1 [8] of mean flame height, L, from a =easurements of intermittency, I. 4 SFPE TR82-8 -,---,,.-.-v v-, -,, -, -,, --e v ,-a-- y- - --,~ v4 m-.m n. -s-w g-,n ww ,w

cre fairly. consistent

with, although The ratio (H /r) in the expre ssion e

') ' heights averaged by eye (8], tendi ng to be slightly lower than, flame for N in Eq. 2 is the heat liberated per unit mass of air ent eri ng the combustion reactions and does not vary appreciably Experimentally det ermined mean flame among a number combustibles, especially heights have been correlated [9] 5through when raised to the power 3/5 in Eq. 1. the equation (N in range 10-5 to 10 ): Assuming the value H /r = 3100 kJ/kg, as e for methanol, a value for N at normal at-mospheric conditions can be calculated in Eq. 1. L/D = -1.02 + 15.6 N1/5 term s of 2Q /05 from Eq. 2, which in Eq.1 gives the following approximate rela-tion for flame height under normal atmos-h P eric conditions (293K temperature, 760 where D is the diameter of the fire source (or effective diameter for noncircular mm Hg pressure): 2 fire sourc es such that D /4 = area of fire sourc e) and N is the nondimensional Eq. 4 L = -1.02 D + 0.230 Q2/5 parameter: ~ 2 The range of N for which Eq. I has been c T Q verified (10-5 to 10 ) cor re spond s to 5 p Eq. 2. N= a range in Q2/5/D for which Eq. 4 is 2 3 5 known to be approximately valid: 7 to 700 kW /5/m. (the flame-height relations 2 90 (H /r) D e have not been tested outside these ranges.) In this expression, c is the speci-p fic heat of air; T. a nd 0, are the am-Temperatures and Velocities ()bient f-temperature and density, re spect ive-ly; g is the acceleration of gravity; H Above the flames, the plume radius and e is the heat of combustion; r is the mass centerline values of excess temperature stoichiometric ratio of air to volatiles; and velocity have been found to obey the and Q is the total heat-release rate: f ollowing rei.itions: Eq. 3. Q=iHfe Eq. 5. bdT = 0.12 (T /T.)1/2 (z - z ) o o where if is the mass burning ra t e. This flame-height relationship does not include fire sources with substantial in-depth combustion, but does include liquid pool fires and other horizontal-surf ac e f ire s. A fire source may not have substantial in-Eq. 6. 4T,= 9.1 [T.,,/(g c 2 2))l/3 q 2/3 p depth combustion if a major fraction of the volatiles released (perhaps 2/3 or (z z )-5/3 o greater) undergoes combustion above the f fuel array, i.e., if only a minor fraction (perhaps 1/3 or smaller) of the volatiles is oxidized within the fuel array by air antering the array. Fire sources with _cubstantial in-depth combustion include Eq. 7. ua= 3.4 [g/(cp p,T)]l/3 Qc /3 1 very openly constructed, or well ve nt i-lated, wood cribs [2]. (z z )-1/3 o SFPE -TR82-8 5 I

j Where b4T the plume radius to the point rise falls of f with the -5/3 power of the where the temperat ure rise has declined abscissa, in accord ance with the plume to 1/2 4 T ; T is the centerline tem-law for tempera ture, Eq. 6. From Eq. 4 o o pera ture; Qc is the convective heat-and Eq. 8.one may conclude that abscissa release rate; z is the elevation above values in the range 0.15 to 0.20 corre-the fire sotirc e; and z is the eleva-spond to the mean flame heights of o tion of the so-called virtual origin fires. At smaller abscissa values, the a bove the fire sourc e. (If z is nega-experimentally observed tempera ture rise o tive, the virtual origin lies below the increases more slowly, approaching a val-top of the fire source.) These are known ue deep in the flame of AT = 900 K, ap-as the strong plume. relations - [10]. The proximately. Empirically. [1,3], this numerical coefficients are based on ex-portion of the curve appears to be prac-periments [11, 3]. A plume. radius to the tically universal, just like the portion point where the gas velocity has declined of the curve at higher values of the to 1/2 u-can also be defined, b; abscissa [12]. o u experiment s [11] indica te that b is u perhaps ten percent larger than bdT. Not all fires will produce the high mean temperatures associated with low Only recently has it become possible values of the abscissa. The smallest to predict, approximately the location of possible value for the abscissa is that the virtual origin, z. For normal a t-associated with o z = 0, i.e., -z /Qe /5, 2 mospheric conditions and fire so urce s o which do not have substantial in-depth which can be evaluated with the aid of combustion, the prediction is [12]: Eq. 8 for fire sources without subst a n-tial in-depth combustion. For one par-o = -1.02 D + 0.083 Q2/5 ticular fire with very low flame height Eq. 8. z [3], in which a p rop rietary silicone Equations 5 through 7 cease to be tra nsf ormer fluid was burned in a 2.44-m l valid near the mean flame height and be-diameter pool a value low, as defined by Eq. 4 for fire sources -z /Qc = 0.16 (m kW2/5) o without substantial in-depth combustion. is o bt ai ned. The maximum mean tempera-

However, it is possible to represent ture rise, AT, indicated near the fuel o

AT in a way which produces a general surface was (by slight extrapolation) 440 o plot of experimentally observed tempera-K, which is not much dif f ere nt from the ture variations throughout the length of value ATo= 520 K indicated in Figure 3 the plume, including the flames. The at the abscissa 0.16 (m kW-2/5), method is based on the observation that Fires with very low flame heights (L/D) Qc (z z )l5/3 in Eq. 6 can may generally be expected to produce o be wri tten [(z-z )/Qc2/5}-5/3, lower maximum mean temperatures. o which sug plot ting AT versus (z - z )/Qc /5. gest o 2 o Figure 3 shows the re-The plume law for velocity, Eq. 7, sult in logarithmic coordinate s for nor-may be combined with the plume law for mal atmo spheric conditions. For ' values temperature, Eq. 6, to produce the fol-of the abscissa greater than 0.15 to 0.20 lowing useful nondimensional parameter (m/kW /5), the centerline temperature [2]: 2 Eq. 9. T /5(cpQ 2 1/5 u o C g /5 (dT Qe)l/5 g 2 x 6 SFPE TR82-8 ~ ~

i s D 1000 i iiiii.. 10 0 y ) O F <3 10 : %s ~ i i i i i i i i, 0.01 0.1 1 10 (Z-Z VQ (m kW *) ~ o c FIGURE 3 Temperature rise on the plume centerline for normal atmospheric conditions in a form attributable to'McCaffrey [1] and Kung and Stavrianidis [3]. SFFE TR82-8 7

a In the plume region where Eq. 6 and Eq. 7 the plume can be obtained f rom knowledge are valid, the ir numerical coefficients of the local temperature rise, 4T. l correspond to a constant value C =

2. 2.

This value,has been confirmed for a num-As an example, consider conc entra-ber of test fires [2], at least down to tions of carbon dioxide in nonre ac ting the mean flame height. Equation 9 with ( plumes above methanol f ire s. According = 2.2 is a useful relation for detennin-to measurements [4], the convective heat ing the velocity f rom knowledge of the flux over methanol fires can be taken as temperature rise (e.g., f rom Figure 3 or 80 percent of the the ore tical heat-f rom measurements) and the convective release rate, i.e., heat flux. It is also useful for deter-mining the maximum velocity in the Eq. 13. Qe = 0.80 ($fH ) e plume, which occurs _ near the mean flame height where the temperature rise may be where 6f is the mass burning rate of 500 K. For fuel (methanol) and H is the heat of taken approximately aT o e normal atmo spheric conditions and the c omple t e combustion of methanol. Ac-value ( = 2.2, Eq. 9 become s: c ording to the combustion reactions for

methanol, 1.38 mass units of CO2 are Eq. 10.

u /(4T Qe)l/5 = 0.54 produced per unit mass of fuel consumed. o o Henc e the ratio b;/Qc in Eq. 12 can The maximum velocity near the mean be written (i = CO ): 2 flame height, u, is obtained by set-o d /Qc = 1.38 mf (0.80 i / ting ATo = 500 K: Eq. 14 ~ om = 1. 87 Qc / 5 '~ l C Eq. 11. u which, substituted in Eq. 12 gives 19,960 kJ/kg, M For fires varying in convective outputs, ( t aking H 29, Mi g = = e Qc, in the ra nge 10 to 10,000 kW, the = 44, and cp = 1.01 kJ /kg K ): 4 maximum velocity near the mean flame C /4T = 6.0 10-5 height, u will range from 3.0 m/s to 'Eq. 15. i om, 11.8 m/s. Consequently, near the flame tip where Concentrations of Combustion Products 4T 500 K, the volumetric CO2 c o n- = o ce ntration on the centerline is calcu-C /aT is expected to be lated as 500 6.0 10-5 0.030, or The ratio = i constant eve rywhere in the non reacti ng 3.0 percent, plume, where Ci is the volumetric c o n-ce ntra tion (volume fraction) of gas species i at a point and AT is the tem-Entrainment parat ure rise at the same point [13]. 6, of After ignition,, the fire plume c a r-Given mass generation

rate, 1

species i, the co nce ntration/ tempera,ture-ries fire products diluted in entrained rise ratio can be calculated f rom [13]: air to the ceiling. A layer of diluted fire products, or " smoke," forms under c eiling, which thickens and generally be-C /dT = (M/M )c 5 /Qc comes hotter with time. The fire envi-Eq. 12. i 1 p1 ronment is intimately tied to the be-havior of this ' layer which, in turn, de-pends to a major extent on the mass flow where M and Mi are the molecular rate of plume (luid into the layer. Co n-weights of air and species i, re spe c-sequently, it is important to be able to tively. Clearly, with the knowledge of predict the mass flow rate which may C /4T, the volumetric conc entra tions in occur in a fire plume. i 8 SFPE TR82-8

v 9 The. bass : flow at a particular eleva-are limited to fire sources which do not g cttributabletion in a fire plume is nearly completely have substantial in-depth combustion. to air entrained.by the plume at lower. elevations. The mass flow It is instructive to evaluate Eq. 16, contributed by the fire source itself is Eq. 17, Eq. 18 for a sample case. Co n-insignificant in comparison. sider a 1-m diameter pool of heptane, which can be expected [4] to have a total -Extensive measurements of mass flow heat output of - Q = 2040 kW and a convec- . rates in pluses [7,8] have been found to tive output of Qc 1330 kW, approxi- = fit 'theorectical predicions based on the mately. From Eq. 8, the virtual origin is et rong-plume relations. [14]. Two p redi c-located at zo= 0.73 m (above the pool) tion relations are used, one pertaining and, f rom Eq. 16, the limiting elevation - to the esentially nonreacting plume ex-is zg = 3.68 m. With this inf ormation, tending above a limiting elevation, z the mass flow rate in the plume can be reacting,g calculated from Eq. 17 for z 2-3.68 m and end one pertaining to the flaming region at elevations lower than f rom Eq. 18 for z< 3.68 m. The linear a g. The limitig elevation, z is an increase in the flow rate up to g z=zg is . olevation in the plume which corresponds replaced by a more rapid increase at closely to the mean flame height; speci-higher elevations. If, at a given time fically it is defined as the elevation in from ignition, the smoke layer had de- 'the. plume where AT 500 K accordig scended to an elevation of 10 m above the = o to Eq. 6. For normal atmospheric condi-pool, the fire would be pumping 35 kg/s of tions, the limiting elevation can be ex-air from the clear layer underneath into pressed [4]: the smoke layer, including combustion air converted chemically in the fire. o + 0.166 Qe /5 2 Eq. 16. .z =z Unsteady Fires f For z2 z g and normal atmospheric c on-

ditions, the preditions for mass flow All the preceding relations implicitly rate in the plume, m, is:

assume that the fire size does not vary m = 0.071 Qe /3(z - z )S/3[1 + 0.026 Qe /3(z - z )-5/3) l 2 Eq. 17 o o ( z2,. z ) g For z - <. z g a nd normal atmospheric cond i- !~ tions,. the prediction ! s: Eq. 18, 5 = 0'.0054 Qe /(0.166 Qc + z )(z ( z ).. z o g Equation 18 is limited to pool fires or appreciably with time. Intuitively, one horizontal-surface fires, becoming inaccu-expects that the properties of a fire ~ rate for fire sources with substantial in-plume 'can be related to the instantaneous depth combustion. Furthermore, the flames fire size, hence enabling the results to a cust be predominantly turbulent (flame di-applied to growing fire s. Previous ex-caeter at base : greater than perhaps 0.3 to periments and. analysis [17]

indicate,

. 0. 5 m [15, 16]. Equation 17.is not lim-approximately, under what circumstances ited with respect to fire source; however, the plume properties can be related to the present capabilities of predicting the lo-instantaneous burning rate; i.e., the cation of the virtual origin entering both plume can be considered to be in a quasi-Eq. 17 ' and Eq. 18, as embodied in Eq. 8 steady state. SFPE TR82 9

As _ illustrated in Figure 4, it is as-rise is a fraction R, or better, of the g cused that, af ter an incubation period of quasi-steady temperature rise. The solu-law-level fire activity, a growing fire tion is: can be approximated by parabolic growth: Eq. 19. Q = n(t-t )2 275={/522/(1 - R3/4)]4j3 Eq. 21. dT /2 o a i where a is a. fire growth coefficient; t Equation 21 has been plotted in Figure 6 is time; a nd t is a virtual ignition for R = 0.75. The local minimum tempera-o t ic e.' Most fires are believed to fall ture rise of a quasi-steady fire to within a range of fire-growth coefficients achieve R J_ 0.75, AT, is shown as a i o ranging from 0.001 kW/s2 I (very slow fire function of height above the fire source, growth to I kW/s2 (very fast fire z, for foup different values of the fire-grcwth); fire growth behavior in this growth coef ficient,'a. For example, for a 0.01 kW/s2 and z= 10 m, the tempera-rsnge is illustrated in Figure 5. ture rise of a quasi-steady fire must be At a fixed elevation in the plume, the 22K or greater if the temperature rise of Iccal flow at. a particular time may be the actual, growing fire is to be within consid ered to belong to a smaller fire 25 percent of the prediction for the size of an earlier time. Consequently, quasi-steady fire. ana. expects that the actual temperature rise will be smaller than the t empera ture In summary, it appears possible to use ~ rise of the corre sponding quasi-steady the steady-state relations for (para bol-fire - (steady fire matching the instanta-ically) growing fire s in a, quasi-steady naous size of the growing fire). The sense, provided - the fires have developed ratio, _ R, between_ the centerline tempera-beyond some minimum size. ( ture rise of a paravolically growing fire cnd that of the corresponding quasi-steady Note On Virtual Origin fire is [17] (normal atmospheric conditions): Several of the relationships presented here incorporate the location of the vir-tual origin, z. The expre ssion pro-o Eq. 20. vided for calculating this location,- Eq. 8, is limited to fire sources which do not 22 have substantial in-depth combustion. In R = [1 - ]4/3 many c c.se s, e specially involving high storage, it will be difficult to determine (aT z /5j 2/5)3/4 whether the in-d ep th combustion i s-sub-3 o

stantial, i.e.,

whether Eq. 8 is appli-where AT is the centerline ~ tempera ture cable. It is recommended that, in such o rise belonging to the quasi-steady fire, cases, the virtual origin be chosen coin-Nste that R is always smaller than unity. cident with the top of the combustible, Equation _20 may be solved for oT, which,

i. e., setting z

= 0 The choice z = o o o ray be thought. of as the local (in z) O has been an acceptable approximation in cinimum temperature rise of a quasi-steady applying the plume relations to fire s in fire beyond which the actual temperature high storage. 4 10 SFPE TR82-8

E 9 a Established Parabolic Fire Growth O W O Z W W i O W ..n 'tJ Z .e-- O* --incubation Period I s ,s' L ,s ~ f. t Time', t / -~ o FIGURE 4 Parabo'lically growing firet,following incubation' period of low-level' ~ ~ fire activity. ^ ^ / s 't 9 + + ~ f s s - ~. SFPE TR82 - 11 l i l

• ~

,o,

m . ~ -m.. -- -. - - -, -., " - - -. -. -, ~.... - - - < ~ - - - -

e xp 8. i c 8000 & =l kW/s2 6000 /- 3 ~ 2 = 0. l kW/s -o 4000 T = 0.01 kW/s* 2000 l 2 E = 0.001 kW/s 0 O 200 400 600 800 t-t(s) 0 FIGURE 5 Examples of parabolically growing fires, depending on fire-growth 12 coefficient, c. SFPE TR82-8 -. ~.. ~

4 3 14 0 i 12 0 10 0 2 E =1 kW/s 80 P k .t v Fo so <3 40 2 & = 0.1 kW/s 2 &=0.01 kW/s 20 & =0.001 kW/s* O O 5 10 'l5 20 Height Above Fire Source,,Z (m) FIGURE 6 Minimum temperature rise on plume axis in quasi-steady fires beyond h which the actual, growing fires (according to Q = a(t-t ) ) will SFPE TR82-8 ha te a temperature rise within 25 percent of the prediction for the 13 quasi-steady fires.

T .I BIBLIOGRAPHY - I 4 1 McCaf f rey, B.J.," Purely Buoyant Diffusion Flames: Some Experimental Re sult s," Cent er for Fire Re search, National Bureau of S t a nda rd s, Wa shi ng ton, D.C. NBSIR 79-1910, 1979. 2 Heskestad, G., " Peak Gas Velocities and Flame Heights of Buoyancy-Controlled Tubu-lent Diffusion Flame s," Eighteenth Symposium (International) on Combustion, The Combustion Institute, Pittsburgh,1981, p 951 3 Kung, H.C. and Stavrianidis, P., " Buoyant Plumes of Large-Scale Pool Fires," To be presented at Nineteenth Symposium (Internatonal) on Combustion 4 Heskestad, G., "A Fire Products Collector for Calorimetry into t he MW Ra nge," Fac-tory Mutual Research Corp. Norwood, MA, Report OC2El.RA, 1981 5.

Tawarson, A.,

"Physico-Chemical and Combustion / Pyrolysis Properties of Polymeric Ma t eri als," Factory Mutual Research Corp., Norwood, MA, Report OEON6.RC,1980 6

Portscht, R., "Uber das Zeitverhalten der Temperaturstrahlung also Kenngrosse eines B ra nd e s,"

Dokt or-Ing enieurs Thesis, Fakultat fur Elekt ro tech nik der Rheinisch-Westf alischen Technishen Hochschule Aachen,1970 7

Zukoski, E.E., Kubota, T.

and Cetegen, B., "Entrainment in Fire Plumes," Fire i Safety Journal, 3 (1980/81), p 107 8 Zukoski, E.E., Kubots, T. and Cetegen, B., "Entrainment in the Near Field of Fire Plumes," California Institute of Technology, Daniel and Florence Guggenheim Jet Propulsion Center, August 1981 9. Heskestad, G., " Luminous Heights of Turbulent Diffusion Flames," submitted to Fire Safety Journal

10. Mort on, B.R., "Modeling of Fire Plumes," Tenth Sympoium (International) on Combus-tion, The Combustion Institute, Pittsburgh,1965, p 973

~

11. George, W.K.,

Alpert, R.L.

and Tamanini, F., "Tur bulence Me asurements in an Axisymmetric buoyant Plume," Int. J. Heat Mass Transfer, 20 (1977), p 1145

12. Heskestad, G., " Virtual Origins of Fire Plumes, " submitted to Fire Safety Journal

13. Heskestad, G., " Pressure Profiles Generated by Fire Plumes Impacting on Horizontal Ceilings," Factory Mutual Research Corp. Norwood, MA, Report OF0El.RU, 1980

14. Heskestad, G., " Air Entrainment Into Fire Plumes," sumbitted to Combustion and Flame

15. Blinov, V.I.

and Khudiakov, G.N., "Certain Laws Governing Diffusive Burning of Liquids,".Academilia Nauk, SSSR Doklady, 113 (1975), p 1094

16. Hottel, H.C., Review of "Certain Laws Governing Diffusive Burning of Liquids" by Blinov, V.I. and Khudiakov, G.M., Fire Research Abstracts and Reviews, 1 (1959), p41 14 SFPE TR82-8 m__.

17. Heskestad G., "The -Initial Convective Flow in Fire," Seventeenth Symposium (Inter-

'y national) on Combuston,' The Combustion Institute, Pittsburgh,1979, p 1113 ' s y-e 4 4 SFPE TR82-8 15

~ 91.

n. s. rearce' rmd '

f'. 5IarcaU w$::qv } r o l.oad and Fire Test Data on Steel-L Supported Floor Assemblies etf

g,'

hi It F IT it l~.N CI'.: N. S pearce and W. W. Stanrak, *I.oarl and l' ire Test ht Data on Stecl supp.rted I lant Aucmidies," Symis<nmm on f are Teir 4 y L Alcrhools--Ilcuruuti & 5 mole 1966, A Albl 511' 422, Asn. Soc. Testing l0 o r " Mats,19fJ, p. 5. h[ks, < 't AlWi lt A CI : Information rained from standard load and fire tests on +. eleven steel surported constructions is presented An clastic analysis of M

• ~

an idealired compmite steel concrete beam is dcscloped, and a definition for the degree of composite action mhcrent in Icst specimens is propmed. E b l he restraint allordett by the test frame and its importance to continued compmite action during the fire test are esamined 'Ibc des'rce of com-i pmite action is calculaicil for a r umber of assemblics, and the mecha-a k...jw nism of load failure durinc Gre test is discuncil briefly. thing this infor-e t i ma t ion, some aspects of present Grc test practice are esamined and '/'4: l pmsible imprmements are surcesled j-p I{I,Y M OllDS: Grc tests steel construction, reinfor(cd concrete, failure. .i s thermal esp.msion restraint ,,, q [ Nomenclature l ; h A, Cross sectional area of steci beam q- [j O I A, Transformett cross sectional area of composite beam = I b rquisaient ciTectise width of concrcle slab Q w b C Degree of composite action ?f d lificctise depth of concrete slab d'f.. I A d. Distance from top of slab to centroid of steel be im 14 D Depth of steel beam i F. h Oscrall depth of structural system I; ;p .,4 c{:%, /r lleiglit of neutral axis from lower Dange of steel beam i Map fr. licight of neutral axis from lower Dange of beam. floor assembly ( , g'l I Moment of inertia of beam Door assembly ( 1, Monient of iiicttia of Hoor t ,r ..V T

• Auistant chief cneincer. Underwriters' I ahoratories of Canada, Scarborough, I-fPd SE I " *

[ Ont., Canada. 'StecI indust ries fellow, Div;,;on of I1nihting Research, National Research g Council, Onawa, Ont., Can.nla. 5

I" "" *"Y represcutative of the tested tssembly. It should be^ted that < I, Mon ofinertia of,tcel heam and floor without shear coonection tithough assemblics may tppetr to h've basic struct:ral sim. .ics and 1, Moment ofinertia of steel beam 9 similar protection, their behavior under identical temperature conditions I, Moment of inertia of composite beam with full shear connection in the standard fire test varies considerably. One reason for such varia '- I Effectise span of beam tion is thought to be the degree of composite action inherent in the con-Ratio of modulus of elasticity of steel to concrete n struction of such assemblics. A second reason may be the variation in

ir Web thickness of steel beam restraint afforded by the attachment of the Goor to the support beam.

i Y ' Distance from top of slab to neutral axis of composite beani es Extreme liber stress at bottom of steel 6 cam Vailous groups participating in the work of Committec E 5 have been active in acquiring information intended to provide for greater under-i e', Extreme fiber stress at top of steel beam standing of these probicms. This paper presents the results of a study e, Stress at top of concrete slab based on a series of fire tests on noor and ceiling assemblics conducted M Live load bending moment in accordance with the present requirements of the standard. The purpose - This paper is conectned with loading and restraint as it alTects the of the investigation was: structural performance of stect-supported Roor' constructions during a

1. To study the degree of composite action inherent in beam-floor fire test. The applied loading on these structures during the fire test has assemblics commonly submitted for fire test.

been a matter for concern since fire testing began in North America.

2. To investigate the degree of restraint provided by the test frame l

The 19I8 edition of the " ASTM Standard for Fire Tests of Building to steel beams incorporated with various types of floor assemblies during Construction and Materi ds." which in those days went under the desig-fire tests. nition ASTM C 19 - 18, contained the following paragraph:

3. To examine the inRuence of (1) and (2) above on the structural performance of an assembly during fire tests.

The Moor or roof shall be landed in a manner to develop in each member of To this end, measurements were made of: the construction. stresses equal to the maximum safe working stress allowed m ive load strams m the cr.tical fibers of the beam dur.ing and sub-i the material of the members. (a) 1 sequent to the application of the load. By the year 1926. the paragraph had been modified by the addition of (h) The strain on the frame during the fire test. wording which was intended to take into account the apparent imprac-(c) DeRections and beam temperatures throughout the test. ticability of fulfilling these loading requirements. The applicable para-graph of the stands:d, then num'ocred ASTM C 19 -26 T, read as fol. Apparatus I ** Floor furnace-The equipment used by Undcrwriters' Laboratories of Canada for " Fire Tests of Floor or Roof and Ceiling Constructions" i During the fire endurance and fire stream icsts. the construction shall be f loaded in a manner calculated to develop theoretically, as nearly as practicable, consists of a test furnacc designed to evaluate the performance of these the working stress in each nxmber contemplated by the design. constructions umler fire conditions in accordance with the requirements

1. • • "" " "E " "'**

• The current edition of the standard (E I19-61) contains wording sub-stantially similar to this, except that reference to the fire stream test has ura sicIauen

.ng ud. dmensny'M appmaWy y i been deleted, since this is no longer a mandatory requirement of the '**","""E"** "EE

• standard, and the load is now specifically referred to as a superimposed mately 900 ft' is enclosed by a dry wall concrete block construction pro-load b"' # ^" *"

"E Discussion within ASTM Committee E 5 which is charged with the re-

• • I# " "" *" #".

sponsibility for the standard has indicated that there is again a need to re- "' 8"' "'"*" *"""E* 4 vise the test method or interpretation of the standard, bearing in mind the supplied with combustion air through ports located below this level. manner in which th.. formation is applied. There is an mercasing tend. Loading System-The hydraulically operated loading equipment is is m ency to employ mformation denved from fire tests out of the context of the designed to simulate the bending moments resulting from a uniformly specific assemblics tested. Tius,s particularly true of the practice whereby distributed load applied over the floor and a linear uniform load over i beam classifications obtamed from constructions which are tested as the support member (s). The load is applied by hydraulic jacks, mounted floor and ceihng assemblics arc applied to structural members which are a

on fise separate loading units, which span the test assembly, with each Assembly 7-Structur$1 steel beam, supporting a blM system-of ^ loadir it controlled through a separate channel. The jacking errange-Ruted and ccIlul:r (Hi-Bond) steel floor rnits,1 % in. det, arranged ::: ment provides for a total deflection at the center of the assembly of equal spans, with 2%-in. concrete topping. Protected by sprayed fiber. approximately 18 in. The system is designed to provide for floor load. Assembly 8-Structural steel beam, supporting equal spans of blendeti, ings di up.to 450 lb/ft2 with a maximum superimposed linear load 3 along the support beam of 3000 lb/ft. nuted, and cellular steel Hoor units,1% in. deep with 2%-m,. concrete topping. Protected by suspended membrane ceiling. Dc/lection Gages-Floor deflections are indicated by vernier measur-Assembly 9-Structural secci beam supporting a blend system of noted ing tapes connected through a piano wire and pulley system to st all - and cellular steel Hoor units,1 % in. deep arranged in spans of 10 ft and counterweights located at significant locations on the surface of the 6% ft, with 2%-in. concrete topping. Protected by sprayed fiber. assembly. The measuring tapes cre calibrated to read in increments of M o. in. I I Temperature Recording-Furnace temperatures are measured -by chromel-alumel thermocouples, unexposed surface temperature by iron-l constantan thermocouples, arranged in accordance with the requirements . s m u s of the standard and displayed on a strip chart recorder. _%.g - -auump- -- I j Strain Gages-The strains are measured by metal film, epoxy-backed, m w n temperature compensated, strain gages having a gage length of % in. and j j a resistance of 120 ohms. The gages are connected in a quarter bridge, i external dummy, three-wire configuration. pig, i_ty,,g,,,f,,,,g,,,,,,,,,pporting sreci members. a 4 Strain Indication-The read-out devices used with the strain gages are I of two types: (1) a bridge amplifier and meter and (2) a digital strain indi-cator. Test Assemblies h h ~_ _ ~ _ ~ _ _ ~ _ " - " - ~ - - " " " ~ Assembly 1-Structural steci beam, supporting equal spans of ce!!ular 1 steel floor units,3 in. deep, with 2%-in. concrete topping. Protected by l- -E ME

2 E..___ E _ __h.--[ E--- ~2 suspended membrane ceiling.

~ " - - ~" Assembly 2-Structural steel beam, supporting open web steci joists on 2-ft centers and arranged in spans of 10 ft and 6M It. Concrete top-support sua ping, average depth 3% in., placed on ribbed metal lath reinforced by wire fabric mesh. Protected by suspended membrane ceiling. Flo. 2-twarion of s rain awars on restrainiar frame. Assembly J--Structural steel beam, supporting equal spans of cellular Assembly 10--Structural steel beam, supporting open-web steel joists stect floor units,3 in. deep, with 2%-irg. concrete topping. Protected by on 2-ft centers and arranged in equal spans. Concrete topping, average sprayed fiber. depth 3% in., placed on ribbed metal lath reinforced by wire fabric Assembly AThree structural steel beams, each supporting 5-ft wide sections of cellular steel floor units 1% in. deep, with 2%-in. concrete mesh. Protected by suspended membrane ceiling. topping. Protected by sprayed fiber. Assembly ll'-A simple beam of rectangular cross section, with one end resting on a bearing plate and the other on rollers protected with %- Assembly S-Structural steel beam, supporting equal spans of steci Hoor units 1% in. deep, arranged in a blend of cellular and fluted units in. asbestos fiber board, and the furnace oper.ing was closed with an unattached insulating topping consisting of % in. of asbestos liber board, with 2% in. of concrete topping. Protected by suspended membrane 4 %-in. asbestos millboard and 6 in. of rockwool approximately 3 ft wide. ceiling. 1 The intermediate support beams in each of the assemblies bricHy de-Assembly 6__ Structural steci beam, supporting a b!cnd system of fluted and cellular (Ili-flond) steel Hoor units 3 in. deep, arranged in scribed above (except Assembly II) had an effective span of approxi-mately 160 in. and were restrained against longitudinal expansion. The 1 spans of 10 ft and 6M ft, with 2%-in. concrete topping. Protected by sprayed fiber. j 1 eTest conducted at NRC Division of Building Research. I ) 1 m. ,tt

• • '8

TABLE l-Composite Narstre of the Assemblies. 1 ensund 5tnis poi Aversse Uve Load Desism g Assembly neeni ts. Concrete tieneting Uve Lead 68 , f. ,f.. fe.

l..

f. in. an.* in.* In.* in. .No. Type in.* Strength. t eat. 48 s. g,,,, g,,,,, g _ f. Flange Flange l. 10 WF 23 133.2 2 808 11 464 000 17 550 9 960 13 140 40 201 138 489 5.7 0.179 A.36 2 8 W F 31 109.7 2 863 10 502 000 18 300 14 460 15.100 40 135 115 365 4.12 0.080 A 36 3. 10 S25.4 122.1 3 570 8 375 000 15 350 5 490 9 720 40 247 129 557 6.4 0.276 A.7 4. 8 WF 24 82.5 ~3 140 10 389 000 18 720 5 500 13 170 40 167 88 276 5.65 0.421 A.36 5.. 8 WF 17 56.4 2 950 10 185 000 14 000 900 6 255 40 207 62 215 7.0 0.947 A.36

6..

10 WF 25 133.2 3 000 10 466 000 17 650 3 000 10 050 40 356 138 499 7.7 0.604 A.36 7. 10 WF 25 133.2 3 100 10 428 000 16 240 0 5 160 40 138 410 1.00 A.7 8. 10 WF 21 106.3 2 600 12 359 000 16 700 4 230 10 140 40 250 111 325 7.05 0.650 A 36 9. 10 WF 33 170.9 4 346 7 646 000 18 470 5 360 10 820 40 389 176 516 6.52 0.627 1 A.36 10.. 8 WF 35 126.5 2 357 13 502 000 18 750 13 620 14 OJO 40 146 114 333 4.10 0.146 A.36 0 11. 2 by 4 Rect 10.7 20 000 .I l' l A-7 t

• Taken as //4.

l l ~ua ta.

s.. _

=ne x. cp r s.- 9 = r. . R = a. s nsmuutm a a2 gy a2._=. n =,,, e

s. 2., s

.a i, =,g.T ..= x.. =.- = o.T 3=. e ocs-2 =. s = c }' E3=l? 25!3 !b - i' - bd b3 E

-Y N

H.it-! 8;esa2 ia i!!!; i ia! N*s -q

==TsE G

g. s a n i a"_qw es 1

3? h.\\. N. a _~80G _~_ c.g - a - 2 "s a :. i.,f SG Fe-g.3 .. I a l- ~t ~~ .=,aA -a c e. a as e s = !! a c. Y =- = ""E s*-r,==. \\.\\. Sf r a i!. i a r ara s i t i.,,a.R sii 5 -~a a a =a N. x g www_ i g g a g =- 5g =a \\ =N. _ ssgau. 3 m n o. a e = = =. =: a 8-l

s. S.

=a g. a. g -~~ N x.., -r ~ s r g_ w, n=~gn e o sa= I ~- =2e \\.N., f ,!2 ~ ,ok f. s

2. g hh. aU

• n m

e. a

-sa m,_ $I 13 ! _n_ a _ _e =p

==. =. s e

c. =.

s s.~ 5 I Sid i i ! !d_ "y is i_: -i 831151 vi ~ A,. -n z ng a- . ag5 g I "n,- naa e 2.a 1 * :.- _v p4 ! s. a, a s a_. - a a 2.X 5 .n -o =aa

a. =. x e.

. a ~e.d ~s .ci a e= s 3 sw = 8 .,v_ h., ,ag ,, = s usaa s a :1 C-. h = I' y agT5 , _ i,c$- 4 o a !.4 3'. 3s3333 Y a *h a k F E $ F f_d l .A

et cacl rement of thrust. Strain readings were recorded for the dura-Calculatime of Degree of Composite Action r tion el . fire tests of those t.ssemblics designated as Nos.1,3,5,8,9 The following is concerned with the development of ca. Apression. and 10. The dcHection and temperature of the support member were also defining the degree of composite action present in each of the tested recorded at segmficant intervals during the test. assemblics. The simpic composite beam model shown in Fig. 4 may be employed MH for the purpose of analysis. The equivalent width of the slab is taken as General I/4 in accordance with the design requirement of liv: current CSA Standard.5 The truc effective width of concrete slab varies with the ade-Information pertaining to the composite nature of the assemblics is quacy of the shear connection between the slab and the steel beam. The contained in Table 1. Table 2 gives information on end thrust measured general calculation method is as follows (referring to Fig. 4); during the fire tests of Assemblics I,3,5,8,9, and 10. Figure 3 shows the A4 t Y = n 2 + AA, i .......(1) i-i s.,3 3 4. y. where f j

s..s..*.g ; G,

[ compression J .b, .vg.t q A g + A...... ..........(la) 4 I Then the moment of inertia of the composite beam is: = nemir. asi' I, bd' + M Y d * + I, + A,(4, - Y)'. ....(2) 2 12n n As is seen, this calculation procedure is based on transformed concretc f area. f in attempting to analyze the actual beam-Hoor used in the fire test I (which is not normally fully composite), clastic analysis may again be i 6 conveniently employed. The distance of the lower flange of the beam FIG. 4--Simpic composite 6com. from the neutral axis (h.) may be determined from the measured stresses by means of the relation: stresses based on strains

• which were developed during load application eD

] in the beams of Assemblics 2,5, and 9. The linear character of these J,,. ,,.,,, {3) curves indicates that at room temperature, clastic analysis may be ap-

• + *
• plied to the structures under consideration. (llcam strains were not The moment of inertia of the beam is then obtained from the expression i

me isured during the fire test.) Each of these structures displays a certain degree of composite action, as indicated by the difference in the stresses f Afh., ,,,,,,,,,{4) measured in the upper and lower Ranges of the stect beam. As used in a this paper, the term " composite action" means the load sharing of the which, combined with Eq 3 yicids support beam and the Door it supports by virtue of the shcar connection AfD existing between these two elements of the construction. Shear strength y,, .....(5) of M-in. fusion (plug) welds used to secure steci deck section to sup-

• + 'e.........

j porting structural steel members varies with gage of the metal and is it should he noted that this method of analysis is base'd on the assump-j approximately 5000 lb for 16 gage deck. i '"Stect Structures for Iluitdings," CSA Stenelsrel S161965, Canadian Standards

• An clastic modulus 30 x IO* psi was assumed.

Assn., Ottawa, Ont., Canada.

is e.. t na u..nvu, of en asscmbly influence its structural fire endurance? An camminition. 3-I ri I I j of the deflection and rtte of deflection' curves illustrated in 1 : Se end b, respectively, simw that the simply supported noncomposite beam (No. / -{ teritt, no.2

11) approached failure at an average temperature of i 185 F. a beam with w g

Idos a very low degree of composite action (No. 2) approached failure at an r ii 7 16u .- na g (II'"'" n' r,- 5i. - finni.,ie. n-i. '(hcrits, ne. 2 .y -Jioni.,ie.3.1., n. i -, u

1. 88 e.,

E e 3 6ee u ten ( fictiti, no.it g i a l g 3 ~ 490 1

see 5

\\ l ra I I I l l 400 l 0 10 20 30 de 50 60 l }xto' i n. tin. 200 FIG. Sa-De#cefirm enrres. i lion that the neutral axis of the beam is located in the web of the steel I I I I beam; it is not valid for other cases. Finally, the degree of composite I' I action (as calculated and entered in Tabic 1) is given by the expression: ({f x te* In. fin tmin FIG. 5h-Rote of depeerion. l - I. .(6) i C,_, average temperature of 1270 F, and a fuity composite beam (No. 7) h,ad in which not approached failure at an average temperature of 1600 F. j g .(7) j

t. - I, + 12n..

.In order that fire tests might be terminated prior to, but reasonatily close to ultinnate conapse, Rohntson ami Ryan (" proposed Criteria, for Defining I.oad l AccordinE to this definition for deEree of composite action, C = 0 when Faiiure of Reams. Floors, and Roof Condructions During Fire Tess," Amrnal of Rc3c,rch, National nureau standards. 63C,1959, pp.12t-12 4 propa ed that the I = I., and C = 1 when I = 1,, point at which tmth a = r/8uo d and o' = F/ ISO d can be regarded as an indica. tian of load failure. In the$e exrre$sions a = maximum detiection, in s' = rate of Effect of Composite Action upon Structural Fire Endurance delicetion, indhr. I = eicar span of principal structural element. in.. J = distance between the upper and lower extreme tihers of the principal structural element,in. From the practical point of view, the question arising from the results These values have been marked by arrows in Fig. 6 to give an indiention of the l Of the foregoing analysis is: To what degree does the composite nature structural condition of the assembly at the time the test was terminated.

~. 67 tr ARcr AHo STANZA 9( ON silll4Urrositu stoo2 Ai*.tMutta I'6 rice Tfst METHODS Relallanship of APP ed Restraint and Degree of Cosnposite A .and The fia cst on the simpic noncomposite beam, as well as more basic li considerations,' mdicate that such members will become incapable of Their Combined Elfect on Fire Endurance of an Assembly supportmg their design loads at temperatures between 1100 and 1200 F. Turning now to the question of restraint Fig. 6 compares frame load Review of the fire test results show that the degree of compcsite action (as obtained from the strain measurements) with average beam, tem-emisiing in an assembly has the effect of prolonging the structural fire perature for Assemblics 9 and 10 which incorporated stect beams of endurance period beyond that of a simpic noncomposite beam, by an almost equal cross-sectional area. It is interesting to note that the maxi-tmount which is proportional to the degree of composite action. mum reaction was almost the same for both assemblics, but in the case Variation in the behavior of individual assemblies-(for esampic, in of No. 9 the maximum point occurred at a much higher average beam the case of Assemblics 3 and 9, the former showed no sign of rapid de-temperature. This development is thought to result from the difference in flection at an average beam temperature of 1400 F, while the latter was degree of composite action inherent in these two assemblics. Tests at the tpproaching failure at the same temperature, although it had the greater Ohio State University

• indicate that there is an " optimum" degree of external restraint which can be' applied. The fact that development of 82 "

composite action between the beam and the floor reduces the magnitude e ing of applied restraint at a given beam temperature may have the effect of optimizing the restraint of the steel beam. The important practical effect of the applied or external restraint is to preserve the shear connection between the floor and the support

n i,

3 beam-and, hence, the composite action. Therefore, the degree of com-posite action and applied restraint are factors which complement each j j.a other and, acting together, have the result of notably increasing the fire l r l 0 ~ endurance of an assembly beyond that of a simple, noncomposite beam. -

su l

j Cnnetusions A review of the test data developed from the representative assemblics ) too described in this paper indicates that the actual live load stresses devel-oped in the steci support beams of such assemblics may be expected to be below the design working stresses to a degree which varies greatly e too.no roa,,a from one assembly to another. In addition, the development of a Varying APP AR(161 rR AMt 1040. LOs degree of applied restraint against thctmal expansion to both the steci FIG. 6--End rhnnt, beam and the concrete during the fire test has tended to obscure the significance of the structural behavior of these assemblics. degric of composite action) may be explained by a difference in the It is suggested that the existing standard might he modified to advan-shear forces resulting from the applied load experienced during the fire tage by requiring more representative loading to assemblics propor-I test. (The live load on the beam in Assembly 9 was 1.73 times that on tioned by clastic design methods, thus re-establishing the intent of the Assembly 3, although the strength of the wclds was approx;mately the original standard. With assemblies loaded in such a manner as to de-same in cach casc.) During the course of the fire test there is an audibic 'cIOP n each member the actual stresses contemplated by the design, i indication of the failure of shear connections between the steci deck there would be greater justification for a wider application of the data 4 and the suport beam. On assemblics where this has been observed, there derived from such tests. is a marked increase in deflection rate after failure of the shear connec-l tions has been initiated. This indicates that significant reduction of the

• • P 63-degree cf composite action precipitates load failure for this type of as-sembly.

' See p. 40. s e

n n.. u a m, m ^ resting directly on the beam flange. The dificrence in pciform-e be- '. DISCU ,0N tween Test 2, the bar joist construction, tnd Tests 3,7, cad which have the slab in contact with the top of the beam flange, could well be attributed to the difTerence in heat sink. Figuic 6 shows that the average end thrust occurs at a lower tem-perature for the har joist than for the steel deck Hoor construction. This

1. A. Benjamin 8 (written discussion)-The authors are to be com-situation is co.isistent with the concept that a better heat sink will pro-mended on having taken the time and cliott over many years to compile vide a greater dilTerential between top and hottom Hange, therefore allow-actual stress and strain data on steel beams from test constructions.

ing a higher temperature to occur in the stect before the maximum thrust The data which they have tabulated indicated quite clearly that over a is reached. The graph does not necessarily prove that the temperature fairly wide range of stect beam Hoor constructions. composite action is difTerence at maximum restraint is the result of a variation in degree of developed regardless of the ti corctical stress calculations. Although the canposite actmn. I beams shown in Table 'I are presumed to have all been initially stressed in sunmmy, I agree w% k auhd statend h k WM mm to a calculated dead and live load stress of 20,000 psi, the highest stress developed in a typical lhor construction are less than the theoretical de-measured in any floor construction consisting of steci deck and rein-sign stresses-this is the nature of the design assumptions which arc 1 l forced concrete was 15.600 psi or 78 per cent of the theoretical value. used. Any other loading requirement would be a forced condition and, The authors feel that the degree of composite action alTects the critical particularly on certain types of reinforced concrcic constructmns, would i temperature and rate of deflection of the beam, as they have shown require loading above design load to crack the concrete in order to de-i graphically in Figs. Sa and h. They point out that the simply supported velop the dcugn stress in reinforcing bars. noncomposite beam No. Il approached failure at an average tempera-The degree of restraint and the amount of heat sink d,rectly, contact i m ture of 1130 F. lioncver, this beam is not a " simply supported" beam with the top flange of the beam may be the governing factors in beam in terms of structural design concept; but is a rolicr-supported beam-perrmmance. ttypical for design considerations. The elTective dilTerence in restraint E S. hec aml li'. W. Stan ak (anthors' closure)-The authors between roller supports and standard end connections has been markedly wish to thank Mr. Benjamin for his interest in their paper. 4 brought out in Professor Blctzacker's paper.: Mr. Bcnjamin has commented upon the data included in Figs. Su and l As indicated in Professor Bletracker's paper, roller-supported beams b which were included in the paper to illustrate the dilierence in per-with no end restraint, hasine a rance of measured fiber stresses from fonnance exhibited by assemblics having varying degrees of composite i 16.900 to 22.000 psi, all failed within a few minutes of each other. This action. From further review of the data, the authors would be most would indicate that the stress on the bottom fiber is not the only critical reluctant to conclude that the dilTerence in the heat sink alTorded by the condition when the beam is on a theoretical roller-supported condition. assembly of bar joist construction. Assembly 2, contributed in a more The authors' data would indicate that the stress in the extreme liber significant way to performance of the assembly than did the dilTerence does not tell the whole story, even for restrained beams. For exampic, in degree of composite action. The authors would refer to flarmathy's i 1 beams No. 3 and 7 are both approximately 25 lb/ft sections. Although Paper' which clearly shows that the creep resistance of a beam at an aver-one has twice the stress in the bottom flange of the other and a range of age temperature of, say,1600 F is so small that the beam would have j composite action from 0.26 to 1.0, they exhibited similar performance. dilliculty in suprodng its own weight. The efore, in assemblics such as j [Of considerabic interest in the authors' paper is the observation that No. 7, as the r. ult of composite action,it is conceivable that no appre-1 all the floor sections tested, composed of steel deck with concrete placed ciabic bending stress was present in the support beam during the ad-i on top, showed average temperatures in excess of 1200 F before reaching I limiting rates of deHection} vanced s! ages of the fire test. As the aserage beam temperatures plotted in Figs. Su and h included upper Range thermocouple readings. the heat Although the data are not complete, we might infer from Fics. Sa an I sink cITect does not necessarily explain the dilTerence in structural per-h that the critical factor is not the stress in the section but the ' mount of a I nuance of the various assemblies, although the relative coolness of 1 heat sink on top of the beam Range. The bar joist tyle of constructions, upper beam nange and the Hoor deck is most necessary in preserving which are most sensitive to temperature, do not have the concrete slab the composite nature of the assembly.

Director of research. Granco Steel Products Co., St. truk. hto.

.See p. M See p. M. i

.......mn..+ With reference to Fig. 6 the authors agree that t rester tempera-ture dilTerential between the upper and lower portions m the beam r iight, in part. explain the temperature dilTerence between the points of maxi-mum thrust. The authors intend to pursue this point in their fut'ure in-M vestigations. E. G.11urcher' (written discussion)-The authors of this most inter-esting paper make the comment that the development of a varying degree of applied restraint against thermal expansion has tended to obscure the significance of the composite assemblics tested. The importance of this factor is now becoming generally realized and is probably one of the causes of variation in fire test results. The measured values of the end restraint which obtained in two of the tests reported in this paper indi-cate that these forces are comiderable 0.p to 180.000 lb) (Fig. 6 and i Table 2). The authors ascribe the difference in the end thrust shown by the two curves of Fig. 6 as being due to the difference in degree of composite action inherent in the two assemblics. Ilut is it not po*.sibic that small dilTerences in the assembling into the test frame could have produced the same cITect? The difference between the two curves in the O to 200 F region would seem to indicate that this might have been the case. In discussing the effect of composite action upon the fire endurance, the authors make the point that a simpl1 supported noncomposite beam approached failure at 1885 F (640 C),[_a bram with a small amount of. composite action failed at 1270 F (680 CLbut a fully composite beam had not approached failure at 1600 F (870 C). It is not clear, however, whether these results are due to any difTerence in the fire behavior of the composite beam or whether they are due to the fact that the loads used developed lower stresses in the composite beams than in the simple beam. Ficure 3 and Table I would seem to indicate the latter. N. S. I'carce and W. It'. Stan ak (author's dosure)-The authors also wish to thank Mr. Ilutcher for his interest. We would agree that a varia-tion in the end thrust shown by the initial portion of the two curves of Fig. 6 could be produced by small difTerences in the method of assembly i into the test frame. Ilowc5cr, after reviewing the test data and recalling k'Ef. '.. ? the care with which the structural stect beam supporting members were ,' l. placed into the restraining frame, we would prefer to remain with the

Tljij -}

observations contained in the paper. As to the clicct of composite action upon the fire cmlurance period. it ws the intention of the authors to attribute the variation of fire endurance between assemblics incorporating support beams having no compmite action, a small amount of composite action and fully composite action, to the variation in the initial stresses resulting from cach assembly having been loaded as though the beams were simply supported.

• )oint Iire Research Organiration. Fire Research Station. Iloreham Wood.

Iterts. Fngland. E l

-._____.m_ m /j;f .u. 'W aU ~ 3 .) 4 Span. ~ ' ' T. Z. Harmathy. f m

1. 2,

.j n NumF j P E ver j Deflection and Failure of Steel-Supported y,7 y ij porce Floors and Beams in Fire 1 (L; w h:; 8: I Tm. J r# - T Tenr. 1 H U raf.

• T Equm a

Dms f) REFERTNCE: T Z Harmathv. "Deflectmn and Failure of Steel-Nup- .i De Pe. far Tru it r6.'- ported Floors and Beams in Fire." 5s mvow,m,m ): DeRe. yg

• Rcuruerir & Smoke 1966. ASTM STP 4::. Am Soc Testing Mats.196" Dime-P*

2.

Zene, ABSTR ACT: Some bauc in f orma hon conce n:n; the deformatio 1 of

^ sands is dacuwed nrie'h in add:. en to pr :n.m; some crac:#cai dat.: (ireck Lee is ake outhned This theo. mas ~.Nrscreraea 45 an e ran-a tbcors acul.e. n .,f sien of Dorn s creer theor) inio a f orm any. r.e c: a i rcre-R e ~ ous 3 de f ormation processes at sariat le ter y..:res anc w ses 4 Str, I. the caicuwwn of the Jerle. and umph6ed te;r rocues are descrit*ed ice 3.d 'emo-rature cri'eria of Tm tion of ionis and 5e.sms during 6te earnst.'e : e For an A h steel beara. for exampie. me E structural f adve are developed criterion of f auure is I'* 0 nun I ' ~ w T, = 45 62 - 4 23 /.3 ~ ti R..c j St:- An obvious shortcoming of AST41 Methods E 119 3s pnied out a KEY WORD %- 6re tests. steel construcoon, floors. be.' m s structural hd't thermal steen. creer. deflecaon. f anure. building construction, deformanon Aihm "P""" G. - d of tre l Nomenclature l IM A Cross sectional area. m : A W )? c d Distance see Fy oa.. in / tm Modulus of elasticity twithout subsenpt. that of steeih psi l 8 E Ofi g f Function N "" j y' h Depth of steel support nc element. m A'/ aH Actnation enercy of creep. Btu tb mole Rem / Moment of inertia s without suescript that of stee! supporting 3" (L element:. in.' T ""e I Length of joist memeer, in. / lha At 11 ' Research O Mcer, Fire Researc h Section. Dis.sion of Building Research. a M L-Nanonal Research Councd. Ouawa, Canada- "O t i A y.. m ~

[ l .am -- w ' N s JMM - G=1FER'_ M'engnum., y m, m ~~ : j gA \\ r, e, .r 41 l

• (

namarHy ON otetEcTiON AND FAILutE Of sTEtt SUPPORTER / PLOOas L l 7 L Span, in. b -Q, m I, 2. n f Number of joist members, dimensionless n 7-P External force, Ib D Radius of gyration, in. 'h r Upported R Gas constant, Btu /lb mole deg R i Force in the joist member due to 1-Ib force applied at the joint at _] which is sought, Ib'!b 3 i Time, hr T Temperature, deg R (unless otherwise stated) Uniform load on steel supporting element, ib>in. ^ w Equisalent uniform load on steel supporting element, Ibfin. E ? Dimension along the length of beam, in. 1 I x re cf Steel Sup-y Dellection, in. j ,,',"j,'[,'[, $7, y,* Deflection defined by Eq 18,in. T Dimension along the depth of beam, in. Z Zener Hollomon parameter, hr-' C ,g, e practical data. Greek Lerrers Y

ed ss an enran.

ne calculation of a increment iresses. Rigorous Strain (without subscript: total load strain), dimensionless on of the defice. rature entena of i Total strain. dimensionless I ' **** P'h* s Se a V' j 6 Temperature. compensated time, hr G. t Variable defined by Eq S, dimensionless ? i g m Radius of cursature, in. I s, p => l e Stress, psi nted out. t beams. structural l f ormation, thermal j Allowable 2,' i a Central, at the midspan g c l d Of the deck [ e Elastic l l A Of the key member i l I Limiting s.! I 5***U' E*' Of the mth member 5 m Nonrecoserable n As t - 0 or e - 0; as y, - 0 o g Recoverable r .t of steel supporting in the secondary period of creep ] s t Time dependent T Thermal At the lower " flange" a i of Building Research, At the upper " flange" d 1 _e x -3

.g M ~~ '" 6 .m.., .j 5^ ~- ') y% cgg,W

• yrg,,,

- -.v: '-Air z., .&=-. g hh, .- - ~ - 2_ u w.,.y -w. m _-=, g .i i ) 42 Fif E TEST mETHCD5 fa i b ' c f otner i j g i 't e t - E,, I u-5 g w 7 s

l. IJ % f it s

U D. S Ptn g II w a t 6'u s 's-( I b Ig Is I I. F [o 3.- y j n.,, l bis - w st^ f, .C l t w i L, w I' O [ on:. = and ~ 8

s., ~

len : l r e s,, - wi: s ~ Flo t L ari wi n ~s ' u rc o 11 has ! ore meer recoc ized that the simph5ed co" cept that the de fornutio" d solics is a auw irstartare sus resporse o ;o ice g as s e' p' ce5r:te nrmtations Aiinoug9 it is i row n 1h.1 a t suic.e-1l-nigh te-' pr..u :, peratures the actormauan is posernej 5s the.reep pr.me ties o' c th. so<id. the lack 0 krowieJge concernir; 'ne se pr Te'i t es arj of we d;ir T estaNshed cesign praejures nas preser:ed cryce's :ron to r er, into accourt in sarious he as of entineern';i ess Une of these Be;Js ;s ec._ the destyr of bulidin@ C:CTCnts f4ir [IrC C:1d drm ~5e p;< igj in this paper information wiU 5e rser on the creer char,;;enst: s s r ; ;, a structurai stet!. and a caicuiattor procedure wil, he yeser;e; wr.;- I although ceseioped primanly for preaicung the deRection anJ po n: a r. '.c bI)kdhNkfsh[h!bbk m-

messow-g 's inania&W'-Cussm *Nr984MIF& k' 4 HARAAATMY ON DEFLECTION AND FAJLURE OF sTIEL SUPPORTED FLOORS 43 I failure of steel supported construction:,8 in fire, may find application in l. other fields of engineering, where the creep deformation of metals at l7 sariable temperatures and loads is of interest. To apply this calculation technique, one must know the temperature 1 history of the steel components of the construction (or of the tretal form t to be examined, in generalt Since the technique of carrying out numeri. cal heat flow analyses has already been described fl.J!,2 in this paper the temperature history of the steel components will be regardec as asailable has:c information. { I. l Llastic Strain and Creep Strain 1 l. During the past four decades numerous theories base been esobed by noted research workers to explain the time dependent deformation of metals, caused by the application of stress Unfortunately, the termmoi. -f opy is still far from being uniform, it is necessary. therefore, to include a e,'t' [ brief note about the terminolo;> to be used in this paper 8 I in Fig. Ih the stram history of a steel specimen is shown at constant { r, temperature, following the stepwise apphcation of a constant tensde stress (more accurately : a constant load), as shown in Fig la To asor' many apparent contradictions now frequently found in the hierature, in this paper any time-dependent deformation process whi be referred to as j creep, and the resultmg deformation.,,, as " creep strain " Accorcmgh,

8. '."

the.f BC and DE sections of the OABCDEFcurse are creep strain curses, M { or brielh, " creep curses " 2~ { The strain history curse shown in Fig. Ib is ty pical of. =.m curses 9 obtainaDie with consentional creep tests, that is, with tests at T = const and cr = const for r > 0. Its mitial sertical section. 0A, represents the quasi-instantaneous response of the material to loadmg. Followmg this (2,i response, creep sets otT at a high, but steadily decreasing, creep rate, e,i s de, dt i, os er the period 0 < t 5 i,. Later e, attains a salue which will remam approumately unchanged oser the reiatisch long perioJ of i,, < t 5 t, iprouded the load is left on ince6mte!> n Final!), owmg to v' some localized reduction in the cross-sectional area of the specimen. the creep rate will steaddy increase again loser the period t~ <t 5t up 'q y t that the de-to the point of rupture. These three periods are commonis referred to as dmg has sery primary, secondary, and tertiary periods of creep respectisely. From a l tly high tem. practical point of siew the creep during the secondar, period is os far 'erties of the the most important. The approumately constant ulue of the creep rate and of wen. during this period is cal!cd " secondary creep rate." and is denoted by e, i taking creep The total load strain ' strain caused by the apphcation of load.. e. mas g~' ,y ) these heids is be regarded at any time as consisting of three terms These terms are e plotted separately in Figs ic, d. and c The quasninstantaneous response racteristics of ,w, tronise is noi arrhcaNe to presircued concreic ented which, 'The it4hc nurntyr> in tradets refer to the hst of reierences appended io inn and point of N per .- 4 _ - ~' ] \\ y s - s% W y ) Y'- .O e as <.-. y x. k .,. a u W 3- -8,94 e Y a ~~

l w a r a m) 5'm e n t m s_a_w n_____ m e - - --- ~ m, - m= - p f' " - 'T'e u m ce w w* \\ ~m m_ qw w fy ;~ = <1 ] d4 FIRE TIST METHODS 40 3 3 3 3 1, 2 I '4 30 -i ~ 3 8 j &&as o o ) 5 b 20 -

• o 4

~ A. o i \\ i 1 w E } e I 10 Source 4 ~

• Leo and Croether (3)

~ A Verse" ( 4 ) o Garofalo et oi (5) ) l f i i i I i e i i j f 0 200 400 600 800 1000 1200 1400 Temperature, F 1 O!E FlG. -Modulus of elastscary of carbore steels. 1M to loading is the familiar clastic strain, e,, which is known to obey i } Hooke's law,' e g e

e. = a/E (1)

E { hf and to be completely recoserable upon the removal of the load. l The modulus of clasticity, E, is a material property known to be rela. = . j ) W i i tisely insensitive to the microstructure of the material. The dependence b i 5-on temperature of E for structural steels is plotted in Fig. 2. $ l' i ([- j

• See nomenclature p. 40.

i 3 W

_h y HARn4ATHY ON DEFLECTION AND FAltLRE OF STEEL-5UFPORTED FLOORS 45 y a s t I I I i l I 1 \\ \\ \\ 1' +: ,I 4 2; ~ { i - 3 =: 3 = 5 - 3I \\ JC o c. o I C CI C g e ~ - ~; \\ \\cs. 6 ; i N \\ c \\ ~~i { 2 l Nl E \\ {l 3 { -- e ; N C N \\ \\ (gj,l _ ? ~ N a\\ t ~ j o.in. m N v; \\ j ! ~" ? N g\\\\ i_ \\ im SN %x \\ \\ I r \\ g ,L, j] ~ f %N \\ \\ N \\ \\t E 7 N Nx \\ \\,y\\ uoa im \\ x x s i N x x \\n .~a x x o 'N is known to otry G G G E ~ c e i (ll af the load {* known to be rela Jal. The dependence in Fig... i ( P e

7' e -miWa*Pg A s- ,,.m mam m j h j ~j s J 46 MRE TIsf METHOes 's I0 f The creep strain consists of a recoserable and a nonrecoserable part. j The recoserable part, e,., is called "anelastic strain." The nonrecover-j able time-dependent strain, e.,,, is generally referred to as " plastic creep d, 1 strain." With the types of problems to be discussed in this paper all time dependent strains may be regarded as essentra!!y nonrecoserable. } t h a t is. t... a n d e,, :: 0, therefore for consenience the shorter term " creep strain" will be used hereafter in place of "piastic creep strain " l From among the most wide!y accepted deformation theor. that pro-pounded by Dorn M.,! has been sciected for use in this paper, not only c,7 4 because of its merits, but also because of the adsantages it oTers in numerical creep analysis The basis statement of Dorn's theory is as follow s in the deformatmn processes that dese;op at some constant , 01 stress lesel, the creep strain is a unique function of the stress and of a so- { called " temperature compensated time " e, ~

0. 5 e,t e.ai for de di = 0 t w hen t > 0; Q

= t e, t where e is de5ned as o.4 i i e = f c-' '" ' " dt '3:

3. 3

~ I or as 9 = c-a er ,for T = cong 3ar The actisation energy for creep, a#, is insensitise to the microstructure 9I"- of the material, and is approximately equal to tne actnation energy for ~ self-ditTusion iabout 140,000 Blu Ib mole in the case of iron; l p in Fig. 3 a family of e, sersus 9 curses for an A36 steei is shown at const saiues withm the domain of mam practica! interest g _ seseral a = g a j fy These curses represent some of the resuits of a comprehensise creep 0.3: study still in progress in the DBR Fire Research laboratory F These I I' *, curses also exhibit straight-kne sections which, if the temperature is held constant, extena oser the intersa s corresponding to the secondary j ' [_ penod of creep Dorn's theory imphes that the slope of the straignt hne A!thou; i I g I j section of any e, sersus e curse twhich is commonly referred to as the creep prEe 1 [ Zener Hollomon para.neter '8', and is denoted by Z; and the intercept. temperatur obtained by extending the straight kne section to the e, axis, are within the lll l uniquely determined by the apphed stress !/On that is. tem pera t u r e,. i To make e.,c""" f, a - <4, n: story oi s; Z= ,( = = ducing two i i wnen da dr = 0 following ec i and t .. = fd a r when do;dt =0 de g_ I i s } l

e t ejo _ -w 42 O m n a e a {Q 'i? T' MY g,'3[*bh#rb" D M MA'# Y M il ML A T=D O+-M*. } G MARMATHY ON DULECTION AND FAILURE OF STEEL 5UPPORTED FLOORS 47 1.0 werab!c part. 8 e nonrecover- " plastic creep g,, (g]

his paper all inrecoserab!c.

g,

shorter term c:

eep strain M ries that pro .per. not oniy O'

s it otters in n,

theory is as = ame conpant 06 ,s and o' a so-Q 4: h ~

0. 5

'O ,i>0 [.- d 04 a ~ ~ 03 .. j. .O y,$5 C005 J 02 Et microgructor .on energy for ,y 1? I

is snow n at i

neal interest 0 .,0 ;; ..ey 0 SC1 0.31 I 10

nensne creep

.,r y lv These

e / 4:

]:g .emperatare i8 FIG s--Generai.:cJ crect cun c ^ '[ the secandar y 9 e straign kne l Altnough the appiscability of Dorn's theory is generally restr: ted to ?" 9 ' crcer processes occurring at temperatures acose one half of the me! ting rred to as tt,e l . the intercept. temperature. in the case of carbon stee!s it seemed to be appheanie j mitnin the 750 to 1.100 F intersal. wnich actuaHy represents the entire ne e, am, are temperature range of practical interest y), .t To m pe Dorn's theory applicacie to the study of the deformation hatcry of stee! structures in fire, it was necessary to espand it by into Jm:ng two piausihie assumptions With the and of these assumptions the - lf[ owmg quan ns were obtamed A M'\\ .h dr = 0 'J ' ,,, In, cosh-': 2" '" b-y, dr = 0 (5i i when de dr = 01 ? W~ e r # E W W N i n t 4h' 919m Jg e n nemunnas

.4 e ._e# ww. mir ev eN' smr ^4NYMM% M hw. i 48 ettE TEST anETHODS 7 %, y 7 10 s e i i ..ei , i i r 2 4 i i 1 0.010 i, 10

• g 2

i i / 2 10 : e _ ^ 0 00. 22 10 ( h N 2 10 ' C.CCt /* j : ict: 8 7 g 2 E N A 10 S a ,j a 004 " 10 j [ ~ ue:s.re ea' :: o Ic - / o 7cc r

• 8CC r i

g ;;7 10 ~ 4 900 F a !OO; r g~

trCC r 10'5 -

' '200 ICQ 5Caie r e t ,o a 1 0 4 6 8 10 15 20 30 4C 50 60 7 1000

5' Fit FIG $-Zener.Hallomon parameter sersus stress corec:atuart f or an.4 3^ ster-h.is been phitte l

and E in t h' foh*m. s

• 1 de, _,

rj d9 L e Equation 6 can t3e used to calculate creen deformation occurring at constant stresses To facihtate such calcu!ations. in Fig. 4 the group 'an be utihted From the re 2# g=h-eg, in:p.irtat h-The actnatm l T

5 .wN NMT-h ~ J MARMATHY ON DEFLECTION AND FAILURE OF STEEL SUPPORTEC FLOORS 49 6 i i 0.010 o ~ e o - 0.008 ? o ~ 0.006 l ~ l 0.004 F \\

1. l<S.

~ ^

• z,Y,::s L, c

0.002 - f - I i r 0 200 4C0 600 800 1000 1200 1400 .n $c ec temperature. F o FlG 6-Thermal ezranswn of an A 36 steel tanntatedt

n,4 in strei N.

has been platted against the 29 e,, group, with the aid of Eq 6. Equation 7,in the following finite dirTerence form, a, = Z coth in 2 i as e9 y occurring at G. f can be utilized wheneser the do, di condition is not fulfilled. ne group From the point of view of the problems on hand, tertsary creep is not unp riant, therefore, it will not be discussed here any furthe-(8) I i The activation energy for creep,aH, and theexphcit arms of Eqs 4 and f f l 1 r

A {[. ; !. '.'$ ' ?, ',',

• T.? ?

~ a h. ~ fa 4 '* *l ' ',e" - =

• ~

_-__L__-- % ' r '.1LD, _. -:,.; $.', ' ;2, -b[. ~ h 4. - %)?y;'ltj..k ?:..; .?.' e m- _.. m g p % _ _ y g .? .b J l j ? so rier tist mzTwoos i ff 5 hase to be determined experimentally. For the A36 steel examined at y, j the Division of Building Research, a#/R = 70.000 deg R. The Zener-i== g } Hollomon parameter has been plotted against stress in Fig. 5. The i y experimental points may be approximated by the following expressions: 7, 3 I0.026e . if a < 15,000 psi I for 15,000 < c < 85,000 psi,f (10; y Z-{ -]' (1.23 X 10'(c = T e i f.

j The intercept term for the same steel can be calculated by the following a

empirical formula: ? 1 'I } . i. = 1.7 x 10-'re' " .(11) e 4 9 l he " theoretical curses" show n in Fig. : hase been plotted with the aid i of t hese c 7 presso,n s, and of h 6. 7 I ~* O*m; to the rciatncly mmor role that compressae creer may play { in the pronicms on hand. 't will be assumed here that esery re ationship L j introduced so far is appheable to both tensile and compressne deforma.

• 49 l

tions. In agreement with the consentions, compressne stresses and

• ] 4 j

strains indicating a shortening of the fibers will be regarded as negaine I i , y,,, j quanuties <v l t '{ Thermal Strains So far oni) strains associalec with the appl cation of load have been i j discussed. There is another kind of strain additne to the load strain, but 3?

7. i practically indepencent of the load. the strain caused by the therma T'

I; expansion of stee! This thermal strain, er, is an instantaneous 2sponse l to the temperature sariation am.'. if the temperature does not exceed 1 -W 13m F. is completely re.oserabic l 2,k 'j brause of the aJwa ity of the lo G trains and the tqerma! strair., the 3 j total strain is gnen by tai D. l i (il i= a e, =,,-e,er (1: C k=f { %,7:.. In Fig 6 the inctmai strain of carbor Mecis is plotted against the a - 3l icniper.iture, based on the information gnen in Ref //. It is we!l known W S"" - i' ( { that such., versus T relations are relatisery insensrtne to the mi;rostrue. m ['. J ture of the material. (di sur W.>t The thermal strain correspondmg to a change in the temperature from .[$ 'j T to T; can be taken approumately as i i v; i i

• * *tr

,3 e r % e rt er ([3 CaII0d I' t J u rl: [ Obuous!), the thermal strain is positne if T: > T, and negaine if n ', t q ] i Tr < T. " g," i g mem br. i.]i.y. Deflection of Steel Joists and Trusses in Fire are ca < t. .f-As an exampic of the use of the mformation p er,er tec in the presious rons" l s si,0. j hi-sections, the dclormotion history of a stee! cust of a commonly usec, so. J 74 l l 4

• 3 6

Yk? m --~~~~-

~ _. am e - ey e.7 N .e EergsspwM.D f48 E&C T 4 n. n'.hiw : g W' '"*.+m& ~^* K_Q HARMATHY ON DEFLECTION AND FAILURE OP sTIEL.$UPPotifD FLOOas 51 O p neel examined at __\\ f.

g R. The Zener-

{fpV,NjN / g. p-{-{ in Fi. 5. The o L 6..... .ing expressions: a) (10) i 45,000 psi: i n by the following h. h'(.. ., !"'e kI,,, b./ } u 1 (11) tied with the aid /* /- 3 .i n ,x /n

i. \\/.

• Nf. Nf. N /* 'N A 3+

creep may play a serv relationship 5 l ressise deforma. ~ j.- ,s ae stresses and N'

7

r;

. a; ', ;' j,*, '; rded as negatise \\ s'j'N s' r ...i .n i...., 1 j 9..,...4 ..:.j.>.. l

)

' load have been e . load strain, but - - - - ~ - - ~ '. ' by the thermal p [ \\,/ N taneous response does not exceed V. Q) ~ ,ermal strain, the tal Diagrammatic picture of the floor assembly. (1) Steel jom (g;) (2) Aletal lath / (3) Spr.lyed asbes!Os. itted acainst the g 3j' [, ( ) lt is well known

pressivc,

.o the micrestrue. (ci Siress diagram of ihe steel joist. l (da Sampl:6cd model of the steel joist l .cmperature from FIG. 7-The Moor anembly examsa.ed u,i E2amt>le 1. l (yy) called membrane protected, floor and ceiling assembly during a fire test I will be discussed (Example 1). The geometry of the joist is shown dia-l and negalise if grammatically in Fig. 7o. Since the load bearing capacity of the deck and l membrane is obviously neghgible, the stresses in the members of the joist are calculatic from the imposed load, and can be taken as approximately constant throughout the fire test. The stress distribution in the structure

d in the previous is shown in Fig. 7b.

nmonly used, so-l

.- y_ .s

• O%

m _.._w -- _ - - n-m u--^ m T .,_e o 1 52 REE TEST *ETHOD5 7 I I., j f Ltirts.i

I % s il t '.

In !!'c i ,l j_ T., e t an,-m 7 j n: Fns j g sw J. J l gg,., \\ mp II T.s - 1 i j j rrew is i ll-w Z 2 [ E is,

;,,: : : :: : : ; i s

~ -

~ ~ i l in.~ F 'g y,, ? i g, ,' I

• ~~

Y 5 lo., 1 i j 3 ^' ' N lh j i m.. - } sir a 2_ 5 3 5 in F .ir e is. 3 n ) 3 .... - -..... - ~ -, - - b w:: e. ~

,ir 4

~ k, l s I:. 0 .s" St. N f 3i ,7, 's l ~ a m. ~ 'a F.> 3 e = ( m ;w.~.= iis E : ; " 1 I i

a[

1;; , e.e ln F g lwe. As,,,, n:..- i

2-w-w eg m3 ~ w m -_ & % ,y n ^' ^ -^a.a.- 9 K -=%; L __ l ( HAaM ATHY ON DEFLEC710N AND F AILURE OF STEEL sUPPOaTED FLOC As 53 l t Thejoist is assumed to hase been built from A 36 steel The tempera-tures of the lower chord, the web members, and the upper chord are e' assumed to vary in the way shown in Fig. Sa. The various steps involsed in the deSection analysis can best be followed by inspecting Figs. Sb to i. j To plot Figs. 8b to g. in addition to the basic information 'the stress j; i distribuuon rFig 7bi and the temperature history (Fig. Sud the graphs 9.. .] in Figs 2, 4. and 6. and Eq 10.11, and 13 were unhzed The 9 sersus ,ji 3 time plot in Fig Sc was obtained. m accorcance with Ea 3 by measuring Ns the areas under the curses m Fig Sd } i 1~. [ To plot the deSection history of the joist in Fig si the foilowmg ex. Q 3 pression was used: b [ y= s.,!. i.

la-a which,in a form restricted to the calculation of clastic dedecuons of steel e

j trusses, is well known from sarious sources From a pracucal pomt of view the deSection at the center of the joist. -:r., I r,. is of primary interest The salues of J. for calcu:ating p can be deter .g ) { noned graphically from the stress diagram ootamed by applying a 1-ib ~- load to the central iomt, as shown in Fig ~c By urtue of Eas 12 and 14. deSecuans caused oy various types of -[ 5 trams are adeinse, a y =c y. ~ y, -.- y r t15' [ In Fig Si these three components of the centra l deSecuan of the joist 2 are also plotted separately

• ~ [, ' * ~ '

I It is seen from Fig. Si that as the temperature of the lower chord es y cecas WO F, quickly cesef oping creep primarily in members E 12. and W :"S :' t it as shown n Fig Sh. results in rapid increase m ine deSecuan Tne [ [ } joist would faii shortly after : nr. 30 min W 5 4 Accordmc to Fic sc. at the ome when fadure is due to occur. memoer i f 12 of he joist is already in the secondary period oicreep t .i l Since the 6nal stage of 6re exposure is goserned entireis by the creep i of the sentral member m the lower chord, member 12 the time of (l [ f.niure can be consernently studied with the aid of a simph6ed modei of the joist. based on a "' e: member" concept, such as th t show n in Fig 'a l 7d For this mocel the central ceSection can be espressed as g I

-ig j y, = ! j[] i.

r 16 8,j l ( in Fig Si the deSecuon history of this modei is also shown by a dasned 3 l hoe A somewhat unusual feature of the deSecuan curses shawn in Fig. Si is their sery steep rise after the temperature of the lower cnord exceeded i d

i l [Q% f p... - a e may e w w = + w.5-1 m mm=_pygqqf g, .4 { l j 3d FIRE TEST METHCOS l k Isne5 ncee t h e.',I j t__

1. _

4 ,g j ._ f ' 55 %. __ _, Me. __ _ _ H u sc e 1 ..e . 's n> a we n,. s. - 0) ,I low -, e,- m.m.-- .-~ - w 4 ~ t. m -1 i em: n.,n, j k ,.,N

• L*

g j y we ,e 4~ j 4 N i 1 w her ~ ^^ as m ..x. h 3 e_. C)

• -g--

o ..e' as b I c-The ~- ] } ofib tion - j .n. i .I ) [ ' i l [ } x 'i 3 w her y 9) WT [l } l l e, tal" a: 'M =% -- 2 hn;i: = p L.e = a wtm l Oi m m m c .u u t wauuuu. - - -- - Q - -- 4 Utiln las Aciual beam assembl> T' i 11# eel beam W F 12 by Y, 27 lb f t. A = 7 97 in. / = 204 i in ' r = 5 on ( I i:1 Concrete deck 1, = 15 0 in ', l. = 7 8 1 ic

• icd' t.li Sieelshee Jcca pia:e in ;

j tJ6 Sprayed insuution ggg u q be Temperariae distrietaion in the beam at one particular time hius v ici Stress dit muiron ai one pariac.a r tir-e a edi Stran unirinunon at one p.irncuLr ome ne; I i ici The i, versus 1 rios at are pather :,me g {.., ifi The m as versus r plot ai cne par:'et.iar tirne l igi DeMection of the beam J! une parti ahr time ihe Simpiirica rodel of the beam anem-b cons s s (ii Idealized crou secison of..' e steel bearn g( ; j,e s FIG 9-The heam assemon exammca ou Esample : l,, lion 10m F T his steen rise occurs because au components except the foist loa g of the Soor assembiy are incapabie of carryinc any load in actua. Socr c a n.,( assembhes the load carryinc capacity of the deck :s far from ocing neghp.

g,g bie Since the temperature of the deck is always th

fowest of iny part of an assemb!), the deck is capable of sustaimng an increasing portion of

' sc e 1 h

Ym w Li ^ t g mq)*Ce,gn.rnana*Q yzrmv y-- " ^ m e <* t mg WMM&s-idtEnnisan r mm ~~ewmmste wmn. t; e MARMATHY ON Q(PLICTICN AND FAilutt CP stEft SUPPORTED FLOCis 55 ,,,,, J. the load and transferring ii directly to the walls, as the dedection of the o steel supporting :lement inueases. Based on a prmri reasomng, the fol-p y. I lowing expression has been derised to describe the relief of load on, or C0 the stresses at a: y location in, the steel supporting element: ...ui I.Y. ~ 17 =c*<*> t) L. s 6 g 5 "' where 5L% C .n. O. 5 W " 59 E4 ,,., ~ I The reciprocal of r,* will be s eierred to hercafter as the %.id resntan,e y of the deck It can be determined esperimentally by measuring the dedec tion of the assemoly under a gisen ioad ._-\\te- , y, = O I r, ' = 1 n - 1 ', ,10: .e.e* ^ i where r, is the measured dedection, anc i. is the dedection that the steel r supporting e!cment alone would etninit uncer the same loac 'to be ob. 7] tained by calculation. % hen the load resistance of the ded is not neghgibie, the dedecuon history of the assemniy nas to oe calcuiatea oy a followup technique a _2 whien wiii now be described 6-i g -~ Dettection of Beams in Fire The buiiding element to ne studied as Exampie : is shown and de M i in '. * = 50" scribeo in Fig. 4a It is a beam assemoly similar to those sue;ected to nre f, tests recentis at Ohio State Unisersity in eeneral. the bending moment in ocams saries along their length in i this way shear stresses are f.roduced, w hich result in dedections acJi-f lionJ to those resuiting from bending These shear stresses are. noweser dg' necog:bie for beams which are long in comparison with the:r derth. therefore. shear dedections will not be taken into account in this paper The 'ciassical" way oi calculating tne aedection hi< tory of stee! Deams consists of numerous caiculations aimed at 5ncing the dedected shape 1 of the beam at sufhc:ently small time intersals The calculation technicue ,y m m.' has neen eescribed by Pepos :/2; and Mordfin '/3: The casic informa iion here. too, is the temperature history of the neam, and the origina E

ts ewe;,t the loist lo.ia distr:bution The operations to oe performed for each time intersa!

.id in actual floor c.in ne understood by esamining Figs 9b to e The cursature of the neam. from being neghgi-1 o, at a few locations along the length of the ocam is obtainabic from

• cst of any part of y

.reasing portion of ' See p 63 .f y

e{f ]g j j l i l ( $>I " g } U E y.W g j e $ag U x: i ,. g 8 " W ' i 4 't 's ,3 I - 3

• i 3

' -a 'i ' I S rr h s 1 7 oO H o. '. ' '* ti ni.. o l h. ". 1 1 e s. e o 'N t l' -. u ',n ' m 1 t n t i o = s s s ' I ti ti o o ot i. o ? I y 2 4 n i x i,, n y in,,. e e t, ,a 3 I 2 n 4 s o f, a t a g r c c p / u p ni y ai 1 8 ( 1 o a 's j c a j A a y f p t o x h, N t 1 r 4 o i i i 4 n, m h I n e s s o i e f as gg 2 , j. :. l g r. w eg /, i n g e . n t i g g u h, o ) ) gg ll [ r p a x l, 7 4 g g 3 h i 3 6 e [g l, 7 a, 1 n 'e 7 n g f g 1 g l n I y f 1 y 9 o k h /. I l, e e H c m x s t l 0 i 6 j 7 i d N if, 1 t9 9 n 4 j 8 7 f l i,, p 6 i i g s I 5 O I 1 x i a 7 n 4 5 9 8 n s N l N f 7 7 f I o 1 7 I N 1 h i e J, A 6 5 7 7 6 1 7 6 4 I 4 d s., 0 l 0 l 6a d a 7 2 t il s 3 7' 7 S H M 7 i i ~ d l 1 1 l 4 1 O 6 1 O 4 ? 2 6 0 4 4 1 2 1 9 1 4 1 1 1 1 j k e I l H S 5 0 5 6 7 7 'p 5 f g I, ,@f l o lp: a ,) !p q p y 4 j D I*A 3 t i e

w a, m'Ye,me m__em m aw wa. ~ + 'M TELMMt BMW 1*hh"O44hr Ma&3ft %&*NgadILura _* w; j MAamafMY ON Of 7LECTION AND F Altuff CP STf fL 5UPPOe!!D Floors 37 l g g l; CE i% 2 7 E 'e $i _ s-:m : , - 2 4.. - 2 +, 3.+t _ pa,, em 1 :; 7 s 5 r. _ ~.-. , ~ ~, - ~ e , ~~ 0 % 1 5 3 e. = 2 I ~ ) = _. ss ?I =_ =_ 5 7,. j 5: = = : = -. = - ? 3 ~- ~ d 3 3 3 Y ~ _= = = y 4J = I 3 h 5 5 [ "r '~ x s. ~. s = 1 wi ta' e ~ e ~ ? ~ -. ~ ~ - e < ~ F4 , 9 2 T. ~5 ; I ~7 3 i ^ ~ ~ ~ 3 E 2_ 2 i = _ _ _ _ = i E i E F i

.4 b 5 *. h, 3

• j_

.m. m.r.arwr v sm. remera-e-et e. s. 4. ,,e, s I l l ) 3s Nat TEST *ETNC05 H A s u A t= .exf (c) " ] ,.ocl 'C jy

1. The load il xj W

/ 1 An "equisaient se j -j duces the san >e 1 j } 'a'o*cl! 1, jil / assembly esame g .3 - ,4 3 g ex si exi / sl j 'l j / j .. The cross s ix, j c two rectanguLr .c x. l ard moment of i e i ou i section 1 x e l oc.ci reoc ' t 6 /

3. The temre 4

e m; ..te ~" Q,ix: j the length of the l I ' s >x \\ .t The share 4 oa. l 3,[ ].j, j sine curse. Uttiir I j g-c .oc. c - I l 2 p ...---j where ir t Or _f % nh the aid m.*=*. .,U interest, the h:st. l l 3 l stresses and strar, f i ), l [ s/ The simpliEe beam assem bi> s q j ? c. ce===- i are presented in a n u, x .a c x u x <m = 18.,.0 r FIG lo-tacaa:ea caicuwiwn or inc JeMecnon or une beam assemtn shomon iEq l8i. The ten - e m r,e 9 isee ie u f or de:, ins, { plotted basec or l in Fig 10a the strain distr:bution piot. Fig 4d.. with the aid of the following expres j The calculatio-l sion j salue of y, by S i then used to ri a t 1 e, es 4 t.0 the next time a:te- = l i As the temper. From the well known relation. j allow ceseiopme. with the elastic s:' d. 1 l 'l' 65 n'in 5re expo' = l dr a 4 the main compor l the t = p Ai dellection curse can be finally obtaineJ by repeated grapni-l by repeated appn. tai unegration of the I a sersus x curve gisen in Colone l l This caiculation procedure may well be adoptable for computer studies, Fig 5 Those of; l but it is much too laborious for engineering purposes. To simplify the 10d, w hich was a { caiculations, the following assumptions will be used 10c The therma; i i 1 1 ___.m

r i l naeuarm ou cestrcnc~ u.e potune os sittt.su ecatro

• Loons 59 q

r

1. The load is uniformly distributed alon-the leneth of the beam.

I e n v': t An "ecunaient uniform lead," 6 may be de5ned as that which pro- 'j q duces the same maximum bending moment as the actual load. For the } assembly esamired j j + l a = STd U

. The cross sectior of the beam can ne satisimeto" 3 erre<ented by two re:tangular piane surf ces. "Sanges

• 14 wnien oFer tne same area a

ard montent of mettia, as the origmal section Fig W For tms ideanzea sec tion L: Ar4 .23, [ 2, =

e.. = - o,,

= 1 1 - The temperatures in the lower and upper Langes do not sar a.on; ( s tr e :eretn of the bean. ses t 4 The shape of the dede teJ bean is sausfa;: ora. descr:nea ry a 4-l sire cur e L tuinng Eq 20. tre centra! dedeet.er can thus ne espressca as F-- I l f: n=. e, -i,. ( w n l wrere h 1 2r W :h the aid af mese assun'rtions the m'o m ttic at en nr.uca. i i irte est the mston of tentra: ordecuen is octain D.e rs euminir; the I' .M.

• t stresses ana strains at the midspan alore j

/ The simpbheJ caicu;ahons concerning the dede;tior history o' t he ,) tseam assembiy shown in Fic A and m umpiinea form m F:gs % and i. a, - !4o ;b c Eu :: 4 are prese-tea.n Tar'e 1 For this assen&> U '. h y-1 s 4'o ru Eq :.. anc *ah E. = IV psi = = .E ; IS The temperature mstory of the o*er ard urre-Dr.:es ha neen t n assmm v,,e piutted rased on information presented by Bietza Aer see Footnott 9 m Fig 10a ?, >: lowing espres The caiculatiors m ea;5 row 0: the tan'e are a:med m amanne a 'ew saiue o' t, by tre erc 0' a( mir intena! Colum

Thn,a:ue a

,.s - Eq :3 appa ab. to tren used to hrd tne saiues of a Eq l' anu e. g .:0 the nest ume miena! As the temperature a: the u; per Sange -eser ne;omes n,g-enougn to anoa Jes e:orme-t of creer strairs.,,. Calumn - is et;ual througn ut wah the e.ast;e stram. e,, in the ;ower dan ce. on ine otner nard. a.m. s, Caium-li. sik becomes { f 21' of min 6te esposure the creep strain ire m.m componert of.., Coiumn 4 F o m ' nun or..... n ca.ca.aea " Y. repeated grapni by repeated appheauon of Eq o The detaik of these ca:eu..diens are gnen ir Columns 7 through 11. The saiues of 2., are found ny usirg F ;

• Those of.:., can be cetermmed wah tne aid of the c.me m F:g

( rruter studies, To s;mpmy the 10J. w ni;n was ot'tair.ed by grapht:al integration of tne curse in Fig 10c The thermal component of the deSection, yr, was caiculated 4

_3 3-- _ _ _ _ _ _ _ __~ _ 7 ,? ), - -__-_=_ mWD 2 !b f A W 1*

y 60 nat Test METMCDs

?J' b separately, using Fig.10h and the appropriate form of Eq 24, and is >3 plotted, in addition to the total de6ection, in Fig.10f. In Table 1, yr. 9 is added to the load denection in Column 12. ?kd L The values of the hyperbohc cotangent term (Column 8) indicate that, h q according to Eq 7, the central portion of the lower Sange of the beam g enters the secondary period of creep at about 90 min. Tba g j To illustrate the progress of stress rehef, the calculated l a, l versus 6 of nru 4 t g time curse is shown in Fig.10c. durine o. An experimental de6ection curse obtained by the Ohio State Uni-To$ .] sersity$ for this assembly is shown as a dashed line in Fig.10f The agree- } ment between the theoretical and experimental curses is satisfactory empf m - ^; from an engineering point of view. the strI .-( { in the case of fully restrained beam < the bending moment at the mid-pg7 4 span is only one third of that for simply supported beams. Because # Ms hn this, and the greater stabihty provided by the 6xed ends, restramed the d- ,.Q beams are expected to function in 6re better than simple beams. Since f. I g j the degree of actual restraint is neser known, and the simple support g g represents the most adverse conditions it would be in accordance unh a' j good engmcermg practice to regard simple support as standard m fire and 1. y endurance mucstigations (g gg, )g ? Temperature Criteria of Failure e 4 i T* 4 j* g Based on an examination of 50 6re test reports, Robertson and Ryan '? e l/J) found that the structural failure of Soors and beams couabe satis-

t\\r

~ 3 factor lv de6ned as the point at which

~

J< 2 ( 25 ] 80 I l l I dr i L ( I ..;) f f 2 50 li 1 i Utih f is 2 hr e It will be shown now that, with the and of inese two "deGection criteria, gooda. .[ l i temperature criteria of structural failure can also be found. ment. i in the fmal stage of fire exposure the rate of denection, dy,/dt,is a' ways The I a monotome increasing function of time. Because of this one must be bet w eet .p able to find, correspondmg to any arbitra:ily selected but sufficiently deck. [, high "hmiting" rate of central denection, (dy,/dt):, a time limit within in co-j,"j - j which, after reaching this selected "hmiting" rate of de0cction, failure ?t ture ir J J will occur for any type of steel supporting element. factor i h. 3 By making use of the difTerentiated forms of Eqs 16 and 24, and of do not< pg Eq 4 (by virtue of the fact that at the Sna! stage of 6re exposure the ? }.~@ j deSection is governed by the secondary creep rate in the lower Sange . gem jf.J (or member) at the midspan), and furthermore, by accepting the second and c y Robertson.Ryan criterion as (dy,,di):, the fol lowing two equations can fihe w be derssed: nW ' %l a 5 t

== e

..,N q ;y $,..,.-m y ',9 ',i f7 ;,? (4 t - 'f f s'. 4,.1 acn .w. f Qvp f.;h W 'Y.'b*N $ ;;tL:;j ? [ '$ W?. KC.- -- L _,w-L --- - _ _ ( HAnxATHY on otrtscTicW AND FAILutt OF STitt SUPPORTED Flocas 61 AH/R Eq 24, and is T. in (37.511./L)Z.]. (for joist, truss) (27) n Table 1, he AH/R , indicate that, T., in l(150/r:)2..] .(for beam)' (28)

of the beam

' These equations, in entirely general form, are the temperature criteria ',II**'5"5 of structural failure for steel joists and trusses, and beams, respectively, mo State Um.- during fire exposure. -i To apply these to particular constructions one must know (1) the I0f. The agree-activation energy of creep, aH, and the Z = Z(e) relation for the steel is satisfactory employed, and (2) the value of a. or a., at a time " reasonably close to the structural failure."

nt at the mid-For A 36 steels, AH/R = 70,000 deg R. The Z = Z(a) relationship

= ,s. Because of has been given by Eq 10. (Here, because of the load-relieving effect of ds, restruined the deck, the expression for a < 15,000 psi is applicable.) beams. Since With the aid of the first Robertson Ryan criterion and Eqs.17 and 18, 'S

mple support expressions can be derived for a. and a., at a time reasonably close to the cordance with point of failure, in terms of a. (the allowable stress for steel), E,, I,,

F andard in fre and I. Taking a. - 20,000 psi and E = 3 X 10' psi (for concrete), the d following simple temperature criteria are finally obtained: 14M. 70.000 '8 ,or i Ryan T. = 46.52 - 2.3 log (L//.) - 4.23(L/I) 5 oulu De satts-I -460 (in deg F). . (joist, truss) (29) ( 25) 70.000 '/ T., = 45.62 - 4.23(1,/I) -460 (in deg F). (beam)* (30) p;:. y, .(26) Utilizing these formulas, the time of failure for the joist (Example 1) e Cl is 2 hr 29 min, and for the beam (Example 2) I hr 25 min, which are in b.Q# .a n criteria,,, good agreement with the information obtained by calculations or experi-L ment. .. z. 'dt, is always The limiting temperatures calculated by Eqs 29 and 30 usually range

• • [U.

j ne ust be between 1074 and 1231 F, depending on the moment of inertia of the at sufhciently deck. a limit within in conclusion, attention is called to the fact that, although the tempera-stion, failure lure in the lower chord (or flange) at the midspan is the most decisive factor in the failure of steel supporting elements, ASTM Methods E119 nd 24, and of do not even require the recording of this information. exposure the lower Onge

• Recent DBR test results indicate that a better agreement between exp rimental ng the second and calculated data can be obtained if T.,

is interpreted a the average tempera-sure or the lower half of the beam at the midspan, instead of as the temperature equations can of the lower flange at the midspan.

i l .m w1.t umEL*Wt'J***4f M ^ ^ " '^2: .-. _ - - - -. =. _ - = =m~-~~*~~L1 w 62 mt TIST mETHOC5 -s* R. W I 4 References l i pi Harm. th. T Z. " A Trear se on TFec'ren;al Fire Erdurance Raur: F.- Tru \\f rihods. A ST \\f STP 301. Arnerscan Socnety Tesung and %12 t e r ais, a lei. p le ': H a r m.u m. T Z. "F "e :t of \\1oisiure on the Fire Endurance of Hood n: flfQ P,.w.

t, F rc Icus AST\\1STr 1

Freme"3. \\1. ' s \\ tan vu !**.r'a ? Ar,c ican Souen T es..r; and \\1 te, -*;. nee of t*e M o flOC MF o- 'er O H "T ret 1 l { i3 Lt... F C. anJ C r are of Ot*er Preper es of Me:a.s *nt T e m pe s t.: r e f er - \\... o StrUC i h

• r 4'

M y N erse. G "T he b w % Propernes of %ee: at H.ca Temre atures.' T ec r.3.x t 9 A m a r u.an w e s Nt e,

• a n e.u F r : nec es \\

..19: 5 r!

1. a 9

(5 Garet.oa E. W.erock P R and wr'n*. G \\ "T he irri. e e of Tem f ) per ture or :he i ! <, Core.inis of some < ornrne :ul sieco lic:r,, un. m u T\\, sTe i:x an,e,m n w en Tesar; < e %, i o, sue c ,.o m:. j rv, le: p to He Temre aure C reer io-Dor n. J E. "S,,rne f undament al E ye-em - l ka \\o W44 :' = ' g; g /cor,:a, o :j c sf. r o.ar e a,,a rm e t + Dorn. J L., ' Proc ess n L ode wrd ; H p T e~rerature ( eer Amen- .cm so..e: Tes 7 ..N M.e 2 tme" M e-ust c ! e;mt-l '4 *. P' 7erer. C and Ho mer J H."E.le, or 5:rm Ra e on me Puc. F ie. 'g. j or siee r ' j.- 4-- .2 P s w. \\o ;< tw r ::

G-H a r,;. i r -. T ?

a,a 5wrz

a. W W Toe Deep Prer n es er Si uc 4 g.i a

%een 'o ne re is* e ; ) 'lo H.-@, T Z 'A Came ehenme F err Mede %ML P.M

e,

Mi~ : rresea:ed a: i** Me:ao Er;:reer r; i erte e e Homn Tes. IYb' g. Ar .6

/! Pe rs I h. T *e m.;al Erc. ears H anJ*oen ' 3rd ed. M.Gra* H::

NeA ! 4 4 ') r2 I-a 4 ..ca P. 7 t Pero, t r. s-,e r; ei sear's

• n n creer uur e.2 N.>. ;i. twa+ r 2*<

ll!' Mordhn. L "Anah c a '. Sa uds of C eer Defecnon of 5 ruc urai B e a m.' 3 A d -'. r " i i.'r 4' N AS A T N D ^^.. N men. Aeror.um.nJ 3r se IJ (Lics m:. F A G..nd H id. J Arecteu.:=c r in ctam u 1

• • ce=$

1 5roceer % er. ;. Be :r l u:. r ** l< k. e<:scr A F, cc R a-J V " P -c rw r,t: f.- De - -: !ca' Di '"2bPf I" M "d hy' e Of he.W s f$

• s ar i NM '.rw

. n "' 3 Nat.cra, hereau of Sundam \\ c; hk. W' 4 r... e, ef i<esearco t; w. k r. I J I m l 3 hl t in i ' t Pcm - P. 1 -~

s'4 E feet o Deck on Failure Temperature o Steel Beams W. W STANZAH and T 7.. H ARMATHY Di: nn v.' B. a nc luvarci. Na Re.ca d. G i. r.ci. Canada It :3 g. n r.d . n.d '.;-i t th. heat s:nx .aracter:stics of decb . t\\+ s.J'.:1< ;; nt e !E ' O n * ?,

• , e r;t t i, re

,t t wl 8t ra ct i.;ra)

• . ::3 m rs r tire H ow n er ti rem. t3 of a

+ *+ " -. .+r.! e '. <! < !a i! '.s s :pJn >rt th S g ur;t en t ;< in, e SS W: ".;t : fr'im wel -kn :w'1 d* !}eition and rate O[ de-l o ...m ~~ n,n..rm . end a v.m

..m;...ra mre cr term f.t ; ar-s ev.+u pp"rtee !T r 'rs a:n: ! e.u:u can a;so W developed

! b ' N !' ' Lw Vdrl'tuF It*C};;;1Ca} grr eups Indicated, how ever ' Mat there Wa3 a

r.i<

msu:.,.m:an ling Concem:ng these temperature cruer:a. Some

!P UCII' .it Jdi .! in t < > Sr im t> W p+% :',e te mperature the hea! + d e-g:p iu' n ;i,s., I,e ;r:ciud. d [n m tre real:5;. , r. 1r;t ' e - - ,N i r 6,'u:t- ert T:

• W, +rn d ;3. J.v.

.L-lti;" r W.L. u n d"it.1' e:

1. +

( n 1; .N t-3'. 'W . s; c\\ e: nidl. t rie i. qtt s;nn C!),trJC!ert-1;cS maV f f+' ' lin t' II T+' **nd u ra n ct' 15 C' '?h eTil' r I k;t a re " J n.! 1-I.t T .t.s + s .. ' a: *.' u J: I l u.-I ! ' ht'ir inc.u !O'. LT; 3 ' ru CI U T. i. [;t;alte Cr;'eru TEST S I' E C I M E N S Tnnv -

e.,:, w er carr;+ 2 ou' on 1. cam assemi, a s ' 'u : were identical h aie:

iTp*

n Cit pa C;, ^ /i[ Ihe deCh.

+\\* r'5;* .cNt t ;,1 ti s; + Ji

• d w

8 F.3 ' "U m 'i. o. )[ ' ' u ~e JSc cm !'ile

• I.1 ; - i are 4hnWP in f..; Cure 1 l

s d'. O :.. L es 1 h* . d < >w ;ng Ct mp O ';b IIe m 1 + Iigure is a!' * * ~ 40 i -~ - } l g 5 I s e 7

• t :,

' T M'_ N ue e e ~ w .-...e -a

• g*t

's A e. 9 s' *: 5 .[

• ' h f

t* gA;.63 w p ~^ 'wa ., n a 7.- . y b k .. s..g'.%_;c. .m. ~ Wi

• 44

,f. A -%4,Q.+- r1" '\\ T';,y= v. -2:Fr7 M.m. w mta ?n.m % 5=:n nec.s.w.m =%m;wic.= n+im...,..w ~ c p, I mrMQ

r. -

etM ..mc w ,.s .~.w .mim.w,;a ~.

n..w

,[

266 Fire Technology 8-in. wide flange,15-ft 9-in. long beam made of ASTM A36-61T steel and weighing 17 lbs ft. Four stwl pans (Item 2), 31 in. by 48 in. by 5 in, and H in. thick, were tack welded to the top of the beam. Each contained a 4-in. thick layer of sand (Item 3). The steel beam was protected by two courses of H in. thick asbestos board (Item 4) assembled with their joints staggered. Thtse tx>ards were fastened to the beam by chromel wires < Item 5; a >>ut 6 in. apart. The bottom of each pan was protected by H-in. thick Fiberfrax < Item 6 r, which was cemented to the steel with an inorganic binder. The steel beam was always simply supported with one end resting on a rolkr. No restramt was provided against longitudinal expansion. The temperatuns of the ste-1 beams were measured at that cross sec-tions - at mid. span and at the quarter-spans. Seven thermocouples were attached to the midapan section; the locations of the junctions of these thermocouples are denoted by the letters a, b, c, d, /, g, and h in Figure 1. Only three thermocouphs were fastened to the quarter-span sections. The junctions of these were placed at locations c, e, and g. The assembly was loaded at two cross sections, each 3 ft from mid-span. The force was calculated to yield a total strss of 14,000 psi in the extreme fibers of the beam at mid-span. The load nsistancei of the pans, which, together with the sand. simulated the deck, was disregarded in thew stnss calcula t ions. It appeared later, however, that this practice was not fully justifia ble. In practical floor constructions, the load resistance of the deck in relation to that of the beam is generally very signi6 cant especially in the rinal stage of the fire exposure, and depends to a certain degree on the composite action between the beam and the deck.2 The thne test specimens difTend only in the amount of water contained by the sand formme the deck. In Test No.1, oven dried sand was used. In Tests 2 and 3 rv etively, the water content of the sand was 5 per cent and 10 per cent by weicht. In this way, marked differences in the heat sink chara ' eristics of the deck could be acraesed. THEORETICAL CONSIDERATIONS The temperature criterion of failure for a steel beam can be written in the followmg general form

• aH R 1_,

,7, 4, = in 150 ; Z. where T. is to be mterpreted as the average temperature of the lower half of t he beam. As has been shown m a recent paper.' for ASTM A36 steel aH 70.000 R. R % hs:. ' nmencuture on rage 2t;9 m -w. . _. C kb;Y i fhhhh$h{h$E.u,uba'Q - h

i

• teel Heanin 267 i

2 - 0.026 a * ' if a 515,000 psi, or Z = 1.23 X 10"e ' **

• if a > 15,000 psi.

Since at a = 14,000 psi, 2 - 7.976 X 10", from Equation i the temper-ature of failure, T,, is 1,593' R or 1,133' F. As the decks were designed to offer the same load resistance, it was expected that, in all three tests, the failure of the assembly would occur when the average temperature of the lower half of the steel beam exceeded this value, irrespective of the amount of water in the sand. The experimental points of failure can be conveniently determined from the mid-span deSection versus time records with the aid of the Robertson-Ryan criteria. According to these criteria, the structural failure of doors and beams is imminent when 1 ILd y > S00-- \\ h j 1 (2) t and IO L ds 1 =>-{ l (3, dt ~ 150 ghj It > cases examined. L = 166 m. and h = 5 in., thus the failure criteria are s > 5.4 m. and ds dr > 25 5 in. hour or 0.45 in. min, TEST RESULTS The f re tests were carried out essentially according to Standard N1ethods of Fire Tests of Building Construction and Materials iNFPA h 251. Ame inforn ation concerning the teraperature histeries of the steel h"ams at mid-span is presented in Figure 2. For each test. three curves are shown. One represents the average temperature of the bottom Bange from thermocouple locations a. b, and c the second, tne average temper-ature of the web ifrom locations d and / and the third, the average tem-perature of the top Sange clocations g and h. The temperatures recorded from the quarter-spans were essentially equal to those obtained from the corresponding pomts at mid-span, except in Test No. 2. During this test. , wing to the development of a small gap in the asbestos board insulation. -bghtly higher temperatures were recorded for the bottom Bange at mid- -pan. Because of the occurrence of this localized hot spot at mid-span, the

uarter-span values were also taken into account in the calculation of the iverage bottom Bange temperature for Test No. 2. The arrows in Figure 2

.ndicate the times of structural failure for the three assemblies. The dedection and rate of dedection measurements recorded during the +ts are reproduced in Figure 3. With the aid of the Robertson-Ryan 2 D 4 j y 4 V !-- T "?"

r I 268 Fire Technolo I i 4 i i i i 6 4 ,m /f ~ ~ ,.-,/ f~ e -m ,. -,.- / 7a ,s,: f' - SOC / ',Y / /c. J -Test No 1. 0% moisture / / - fest No 2.5% moestore 250

1. '/

-Test No 3,1C% moisture e O 20 40 60 80

1. 00 12 0 HO 16 0 18 0 200 time, min Figure 2.

Temperature histornes of the steel frames. criteria, the times of failure were determined and are marked with arrows. If the average temperature of the lower half of the beam is defmed as 2 (bottom flange temp) + web temp 3 from the temperature plots presented in Figure 2, the values shown in Table 1 are obtained. It is clearly seen from this table that, even though the moisture content of the sand had a marked effect on the time of failure, it had practically no effect on T.. This proves in turn that the concept of using some specified temperature of the steel supporting elements as a failure criterion is valid. It is appropriate, however, to point out some practical difficulties in connection with the calculation of the temperature of failure. The most 'O i 1e., ., s ' res, ~ 2 \\ 'A { '5's ;;.w,&- -b--- j l j )! ?

• )

- ca '[ e...c., 'l f5 3 v. [ E 5L eet ci.g g C3 j '~j sh, 5 02 3 / .i i, / ~ ~~:.. -~& ..._,g.' /,I.) n I" ,. l C J c z: 40 6: e; ix 12 ; itc 6c iac 2x* ' tee, min Figure 3. Deflections and rates of deflection. b p__ l Y q edy.

[ cl Beams 269 serious among these is that, because of the presence of the deck, the actual load carried by the steel is not known accurately. In fact, the effective load usually varies with the time of fire exposure. Some test data reported by Bletzackers indicate that, due to the increasing participation of the deck in transferring the load to the walls, the relief of load in the steel beam may be as high as 80 per cent near the point of structural failure. TABLE 1. Temperature Criteria of Structural Failure Moisture Time of Average temperature of the lower Test content failure half of the beam (* F) No. (!'d (min s experimental calculated 1 0 109 1,173 1,133 2 5 135 1,207 1,133 3 10 180 1,177 1,133 Another difficulty is that the two-flange model on which the validity of Equation 1 rests is rarely sufficiently accurate. This model implies that the maximum tensile stresses always occur in the bottom flange of the beam Recent computer calculations

• indicate that. as the temperature of the bot-tom flange increases, the neutral axis and the location of the maximum tensile stresses gradually shift toward the colder regions of the beam The finite load resistance of the steel pans and the inadequacy of the two-flange model are probably the main reasons why the calculated tem-peratures of failure in Table 1 are 40 to 74* F lower than the experimental It is reasonable, however, to draw the following conclusion.

es. During fire tests, the temperature (equal to the temperature of the lower half of the beam) at the time of structural failure of steel beams carrying decks of similar load resistances (or stiffness) and subjected to similar unit l stresses can be expected to be practically independent of the heat sink characteristics of the deck. NOMENCLATURE h = depth oi steel beam, in. .1H = activation energy of creep, Btu /lb mole L = length of span, in. R = gas constant, Btu /lb mole R T, = average temperature of the lower half of the beam at structural failure,

• R t = time, hours unless otherwise specified y = deflection at mid. span, in.

2 = Zener-Hollomon parameter, corresponding to the design stress (maximum tensile stress), hr i e = stress, psi REFERENCES ' Harmathy T. Z.. "Deffection and Faihare of Steel. Supported Floors and Beams in Ftre." special Technical Publication No. 422,1967, American Society for Testing u-w - 6 yggemmensE m, _xa~esu~a md y = f

270 Fire Technology and Matenals. Phdadelphia, p. 40

• Robertson. A. F. and Ryan. J. V. " Proposed Cntens for Denrung Load Fadure of Beams. Floort and Roof Cc istructions Dunng Fire '"

." Jot.rW of Researen of the National Bureau of Standards. 63C. l';59. p.121. 8 Pearce. N S and Stanzak. W. W. " Load and rtre Test Da _a on Steel-Supported Floor Assembhes." Spectal Technical Pubhcation No. 422.1967 Amencan Society for Testing and Matenals. Phdadelphia. p. 5.

• 11armathy, T. Z and Stanzak. W. W, " Elevated Tempera t ure Te. sde and C:eep Properties of Some Structural and Prestrectne Steels " submitted for pubhcation.
• Bletracker. It W " Fire Reststance of Protected Steel Beam and Root Ass.emehes as AtTected by Structural Restrr. int.

Special Technical Pu b btation No. 422. 1967, Amernan Sw&tv for Tatinc and Matenals. Pndadei; hm... 63.

• Harrna.hy, T Z.. " Creep Bendmg of Nonuniformly fieated Beams.' paper in preparation AcKNowlFDGMPsr The hre tests were conaucted Ly E. O Porteous and J. E.

Nrndt The ex;wnmental work on which th:s paper is nased was carned out as part of a t w perative research program under a fellowsrup arrangement bet w een the steel industrv of Canada and the Dinsion of Bmldmg Research of the National Research Couned netween 1964 and 1967 Tne nrst author was the steel mdustry's Fellow This paper is a contnbut on from the Div sion of Buddmg Researcn. Nanonal Research Counal of Canada, and s puo'.sned with the approval of the Director of the Division. 4 .m_=__ i 1 e = 1 Q7* +.e. W. e, j o $'nb Iikg W ho.qw y%% 7. %,'gZWm if ~y P,'< q^ w y p.. f . %& $5% f _v fi ak ' L h; ,.i' M$.N;WS& . M. P f.._ h ~~_w n__ _5_ Mi&&id"6Mh%QBEMOEiGMEM ' $& %T! W,?$.,s;M W2H, pp

.y -. h sh g +,a,,,$'. $.r.Ej @.' 4 Y M, k - >

1. ).,

%,4 [ 5 M ir W mad.- 7.T if .t t N .. ~ $mQ.h. v ,. h h - $WA z;. - .a s- ~ - &M., /,: m-.L % N. h. 4 Yg.g% -=~ M &....W_%-$~ .N A i =C w sh.c:..e. m.y x. ? l ot rz -m..e=$#e. w.. m . w.. 4 .a g.m "Y r: - c # .- d... g, ;.: s. m.-:p,=%. .-w. y2 e. e n W &,r f e.7.A'g$,8,. i. g w.)g1 J .j Q!R&mk f g, 4r.. s e.m.. E:t.p;gy8su.g .e w a. p.e .p - ' 4 r.Sm +. +:4 n, y~. -~~1 m

est o

^Wi$ ?$' : 5$ ~ R 4hdi."$ eL y&y&. e>% ? 'j0'hiL NW _ A= m**--Sp.Ep' '~ I.'SwD .q 4* CE -pr.jg* mt .n.m 4 4 = 1

===c _.a _.---m l 1 l 'E s a 2' = ~ e, 6 4 6;=!s=;ici7s=Es =?,:2,e n e e i =v : o : -2= = -v, .=

=

8 e, e v 3 = ;, =5.e, c - - o : : - y e, _= s - a ; : EO 2;+ -~ = = y =_ - = , =.: ,c e-

=o

c u

=

= =

o.=: l O C - -: = _ = w = = t 3 = =, =, =: ~ -: :=,: o = - .- =, - _= o = s: =

==:a,o

= -==s o ,= mv s -: :

a ? =.= =,-

== = =

== =

e. ; ;

cc o =- =*=y=- =5 = r =:s -= = + E m: 5 e, > 3 i,. o 3s i.- r = 3 i = 4 :' s ;2 :,, s i : ; =s 5.: - = 2 c =J i _; " < E - t-:: r s s' yzi,2ss-0 = O _=7 e4 = L o -:= =,. s. o =.c.~ ,.= : :s >= we o .g,s = e. s - -_ x - - = s,_ =,_== .e - -= v _ -: w: o,,,r= =,o_f = ~,

= >

u, = _ _, =x: =- -,-=, a.- os = = :, === : o== y W o ~ e,=- c c a.. - x n,.3, :.,< < .-=.c e o = s=w: = = = _, :, s-2.2 u se=,- =4;-- o = o om m, e= -= = e-3 .,~=: = -,.,

==, - - s o -o - r, = y :=- *= 3,s =,.,,, =. r -. =:;.=,,, 2. = 2 - 3 _, .a ,.,- oo o- ,. oo = . _ = .e. c .-x,- -==. o==e.- - y e = -= e= s =. - c o== = 's va = ~o i 2, w = -

• 2. v 3
• 2 L e =v 2 -

s= s = -d*E; 5-: o,,, s e s e 2 ::v s : "= 2 - s 5 I

= = =. c7 r: esa o -E = = r =s e,3: s - ; s = y - :5 O -=3. =ts'v c 3: = o ,, s =- ce-u o a coo =

==s e. -= = =, s ;: s o,.a :,.

. -: : s = _= =v _', = 3==.

m .: x a_ w .o, =-is==>

,-=s

/,.=,-==- .o = -, =, o=o.=_ = -u u- _= = - c-- c o 3.'! 'l,i 5 ", j h ' y 5 6

• i s i =! ~ : i j j i f ' ; I 'E 5 31 2 ; 5 O<]

9 2 s r e _,: g = -" 3 : = E i.' ,

• 5 = 2 ',e
• a G :

c'.: 5 y a =- w -E 9 1. o

• eou u

r-u 7

• =

=2 ' =: ce,,=e=-= o- = -~~, 2i;..= =s=s7C=7'09:v =, - = 2 's 5. s s = Y I.? 5 m e: = s m v O c = =- = = =o y-u -.c = o = -: o 7e. =;I 3 t-c%e z.': o - = = = y, ;=.i =E.

• _ 3 :

5 5 - ' u -m., = _,,, y ; = =. j2 O 3 o,- - _, =,ces,,,"-::-o-3.c, ' ;-E.;- 1 2 *,i i c : -4.i s. 3.4 --e3 3 ,s 5 ~0-0 :,2 +: - i..: i n * "

• u av e- -

- = ;-

x C 0 z- =j o - 5; E E oO $ zi!j $$1]i;1stil;jijsjys??f1r s,.,,--

==_=, ff 3:=,' m E i = - - o - _ = . = =. u e= ,=,=-_:= .=e =, u .= i = ~ -y

:sv,- - r v = _.=

u ~v

== o y _;c u a < * - ' ': ',. c - = q :' '= : c== = = * ~: - :, : e5 = =.2,, =: 1 o

,, = ' '

== = O m 2 '~^ ~ ~ - '~- -= = . - _O u u ~ N www . mme. - ? =,T C he4 'e-i. ! ~ $ =" z& g C 4 t$. 2$ tJ "t s,90 3 -=; .= - 3 : -= u. =- = 5;2 % s i t 2-3: e = ,=e = o =. 2 v 2 .. Z =. -= =. -7 ,, = =.; 8 - m s u=2 ~ i. -2e'- =7 ~ E t

• - - 2s I

m. Y ut' -2 0.% 7 - a-5

E ~ ~i j - "" :

s G

==

=

==~ Eo-E_E.u e

• j:

v0s ct .~ s v c-r M o r- ?= =, c. g :/ o-- ? =. = - "" J a: ., /s: .2 = c. o x 3u = y 4 - 4

== 2 =, 2.* b_ [., _U _ - " - = I:

1. =

=7 , u.P. = =

• P

.= 4, =4 d = =[ 2 ., shg E <o - -.= =7 = = * *

• / -.

=., e. c., -,- Q E = ~ ", " " g 'r-,I. " ". _m s ~ .M -, = = =% . s, Ik id"w 9 P=>k= = 25252% 025. =ge

s 1 ^cusa on reoncio sneet seam troon 4, noor Assems es u tiet . mainoos roof, ami ceiling sysicans availat ic for building construction. 'Ihis has temperatures. Considering the variety of load con 6gurations posssble, stemmed primarily from the development of new buihling materials, fab-the applicatie of handimoit strength-temperasure relationships is highly 'l"**'I""dhl*- ricalism processes, and design innovations. i lhe present program was designed to End the nature and efect of As new prmlnces ami assemblics were developed, more and more were restrauung forces generated during the progress of a standard fire ecs: and subjecied to AS'IM Mettuxis of 1 ire Tests of fluilding Construction a (E I 19 -- 61). In assemblics incorgxuating protected sacel beams as struc-l cxplore the character and magnitude of restraint which might be gen-erated m aclual buil ling frames under 6te conditions. tural clements, comparisons between gest results showed inconsistent 4 variations in fire resistive performance, even for essentially identical con-i strucihms. 'Ihree factors were recognized as having an innuence on the struciural performance: (t) variations in construction materials or con-The purpose of the study was to explore the effect of end restraint ap-struction details, (2) design assumptions and levels of load required to be plied 10 protected steel beams and, in some cases, the associated concrete sustained by the ecst assembly during the fire eximsure, and (3) clicces slabs umict the ASTM Method E 119 Gre caposure conditions. of cod restraint generated by the test assembly during the progicss of the The study encompassed twelve test assemblics of independent beam 1 l fire endurance test. and slab configurations loaded for simple span bending. The following Of these factors, emi resiraint was considered the most significant clicct important sinsctural conditions were considered: on the structural performance of test assemblics under fire exposure. In I. The method of framing or supporting the protected secel beam in i this context, cnd resiraint is defined as the combination of reactive mo. the test furnace including simple sliding-anti Amed-hinge bearings, stanJ-ments and forces generated by the dead and live loads plus the forces gen. ard AISC 11 series connection angles with beam ends either Hush or re-I crated by temperature variations in structural members during the fire cessed, and fully welded emi plate connections. endurance test. 'Ihis resiraint is applied at the perimeter of the Icst as.

2. The clicct of she concrete slab with ends bcaring against a restrain-sembly as the result of the method of framing or solerance of fit in the fire ing frame, test furnxc.

3. The effect of design concept and construction techniques, including

'lhe restraint condition developed in the icst should be typical of the noncomposise action between beam and slab, partial composiec action, 3 i restraint generated in actual buihlings imder fire conditions. In this re. amt fully composite acthm. gard, Section 23(b) of AS'IM Method E I19 - 65 cads as follows: p The comparative effect of unrestrained beam expansion and end ro-talion with restramed beam expansion and end rotation through applica-

23. (b) ticams or joists forming part of the anembly shalt he suptorted in ac-tions of various levels of axial thrust and end moment.

cordame with the recommemled fabrication procedures for the type of construe.

5. The eficct of applied vertical load on the resulting worki#E stresses-lion Aucmhtecs reprewming forms of cosnirmuon that sestrain snuctural cle-1 ments and top dal shall he supimeted by a restramma frame inworpswated in mg,ggg,,,

i the furnace structure stimulJimg hifuulatingl suth restraint. For the purpose of this paper, the following definitions apply: 3 Alliyough the intent of th,s requ, cment is quite clear, the melluxi of Inela pen,14 set ifcain-Slab Asseinblics.-A representative element of a i u achievmg ihc desned clicca _s m no way spelled out. ihe magnitude and concrcle slab and protected stect beam floor sy?Iem consisting of a pro-i characect of the desned restraint is dellicult to assess, and consistent ap-tected steel beam and lypical width of concrete slab, designed for simple plication of the requirement is virtually impossihic. span bending and compicicly unrestrained along the longitudinal edges. i Published data l1,2l8 unlicate that the strength temperature relation-yc,tical Design loads-Superimposed loads, determined by calcula-j ships for secci are determined by scsts of small steel specimens uniformly lions to imimse allowable stresses according to accepted engineering de-comhtioned at specific temperatures with the load applied so as to dis-sign practice. 4 tribute the stress uniformly across Ihc section. 'Ihis is a totally dillerent p,.3ign Allemable Stra ssc s-Timse urem piM h 4 g-stress comlition than that of flexural stresses m a beam where Imth leanile cagions and building codes-the maximum allowable stresses for the and compressive stresses in varying degrees are present (3l. I m thermore. purposes of design by accepted engineering practice. as stresses are developed m a beam, itJs wcll recogniicd that they are gg o, Aing Stre m s-The stresses, whether measured or not, actually redistohuicd to other portions of the beam at both ambient and cicvated existing in the structural cicments of a floor system after application of the verucal service loads or the vertical design k> ads. iiatic numbers in bradets ecles so she sist ut setencnces appemica i* End Restraint-The combination of axial forces and end moments

ccenzacaen on reorecica sreet seau rioon 4.m soor asshs g[ sa seetest** moos imposw on a Aoor sysicm es a result of both thermal capansion and o l composite action belween slab and heam by insing a scrip of sheet TcGon thermal gradient in a nee caposure. In a building, restraint is generated 12 in. wide and 0.001 in. thick helween slic steel dcck and the top Aange - as a result of the resistance of the surrounding construction. In AS,1M of the secci beam. No mechanical connection was provided between the l Method E 119 restraint as generated by the surroundmg fmnace frame deck and beam. This assembly was constructed to sinieslase alte structural and may depend on the solcrance of 64 of the assembly m this frame. condition wherein the top slab acts only to transfee applied loads to the Noncomposisc scam-Slo 6-The design concept in which the beam is beam. assumed to provide the entire resistance lo applied bending moments. Test assemblies B-2 and B-4 through B-Il were designed pseudo-without assistance from the concrcle stab. Tins alcalued condition can composite. The assemblics were constructed with the metal deck phag a be approached by climinating all mechamcal fastenings between she beam welded to the top Aange of the secel beam 12 in. on centec along the and slab and by providing an essentially inctionicss interface between beam and slab. n... c e Composite Scom-Slo 6-The design concept and construceion tech. Tenaa saw **=ea a gy a y i, we f."

• • • • "[* *"[*'". g[,,
• ',T, *y 'M'.* """"" " '"*' **""*

nique by which mechanical shear connectors are placed across the inter- ,, q [ face between the beam and slab so that applied bending moments are resisted by the combined beam-slab section. b.3.- A-g g, b3.. _f. . 4 w, (", ', 7 '",',*,,"..*i...p'*"~-- Psemlo-composite Scam-Slab-The constructitm technique represen-tative of normal ficid practice in which composite action is achieved due I,me j ce.cas.i to friction at the beam-slab interface and the connections between deck e t's **ssfe f I and beam.

r. F a N

j / p. El sewer Description of Test Assemblies ,d .of q i A floor or roof construction selected as the basic construction of all N)

• No., N d *."a"4"aw"=e ecst assemblics in this study consisted of a 4-in. concrcic slab,36 in. wide,

,,,,,' 'M*"* ' ' i cast over a 22-gar.e steel deck. The concrete slab was suppmeed by a . l'a 'I'g,, e o. kr-12 WF 27 ASTM A 36 siecl beam, and the firc-cuposed surface was pro-I. l ] f* g' ,I aau*,* *** g h2 esp **e sected with a spray. applied insulation. 'V J l Design l J

• • *b i

'"I d j t's M [e g p, ev er y Conventional engineering practice in the design of Hoor systems as-Lt AcJ a I." j 1 sumes that the funcsion of the Hoor slab is to transfer loads to the sup- "',,*',,',,,,.1 portire 1. cams. The function of the supporting beams is to transfer loads N'd""'*'*'"'""""'*"**8* rw i,,,,,,,,,,,, j to the 1:ain girders, bearing walls, or columns. The supporting beams IIw'(d "" ***

• l may be assumed to function as simple beams, and no composite action FIG t-Comsrwsion dereits of scar enemues, j

between beams and slabs is considered; or the beams may be designed to chiain composite action between the slah and beam in the building center line of the beam. Conventi< mal design of this type of assembly i construction. In Imth multibay and multistory buihlings, frame action is auumes no composite action. 1 often a design consideraiion. The elements of such frames are girders and Test assemblics 11-3 and 11-12 were designated composite. The as. columns, whereas the elements of principal concern in the present study semblics were constnicted with the metal deck plug wchied to lhe top are the floor slab and Roor beams or joists. H.mge of the steel beam as in slic case of assembly 15-2. Ilolcs I % in. in diameter were drilled in the metal deck to c! ar she heads of shear con-The vertical load calculations were based upon design allowable I stresses. All test assemblics were assumed to function in simple bending untors which were welded la the top Range of the beam,2 in. cach side prior to the simulated fire caposure, of the centerline. The shcar connectors were embedded.n the concrete j Comtruction slah. Composiac design assumes that the. slab and beam act as an integral r unit. i The noncomposite test assembly (11-1) was constsucted to climinate The fire caposed underside of each test assembly was treased with a i

mincare oe4 reotecto sitet c:a rioon mo soor Assamatus c, ,,,,,,5, oos tied vermiculite cementitious mixture. Type MK, fire insulation. Restrained Scams--The concrete-filled steel restraining frame, which .a ness of insulation applied to the underside of the metal deck is permanently mounted on top of the Gre test furnace, was utiliecd as 11 e thi cvera ed % in. T he thickness of imulation upphed over the contour of the pumary reaction frame. This frame was reinforced with foest I%-in.. diameter high tensile stect rods. Two were mounted on either side of the ihe se et licam averaged % in. The application of the fire insulation con-formed to the recommended practice of the Vermiculite Institute.1he longitudinal axis of the restrained test assembly. The high tensile bars thickness of insulation was chosen to provide simulated fire engxisures were drawn taut with nuts l'carmg against a heavy fabricated secci rein-in the range of 1% to 2% hr, and solely for the convenience of the test forcmg girder at cach end. The scinforcing girder was designed to apply the added reactive force to the furnace restreming frame in line with the piogram. The construction details of the test assemblics are shown m I ag. g. 6 El O __ lj .E....n,.. , f,=, g.p g Io u , t', ..... _ m y _a l i I =. t, .w n I i ! ..,,,,.,s i ! 'l e;,-',-.,,,. . w.. < ~- L ___ e v, . x .g. m.Q.. e..I g,,~ t n ,+ .. f.ra l \\..y, ~ne. l os,s, --r a d

• _D' l...

.....ra ~e... w.* c L..p.

..v

.m. L ' 8 t n.i.e i,,... E.,. . c e c. r r.-r ' I* ***? f ' * #' "' # - r. ....., o.. a. .. m. : ; p , g. ,, 3 o n-c ,a-r, - -.. _. -,r. n _-..- l tho.?,rfj ~i

d. " "' a-. -

!,g. ,g-g y i. r~ ~ s__ +3.**~~ m A =.r.. r. < <, r,..... E. g.. i om { j g

e. g.,;.s e _

4 p J 1 i 1/ s.5.. o i 7,,=,-.

• ::::,a n..

8 !G' 3 ~Furn ' t '"'"I')i'"#i'" I'd 8""' E.P.a.'.a B..' '9 HG. 2-scariar Jences fo' *"'nad"rd '""- } longitudinal axis of the test assembly. The details of these modi 6 cations are shown in Fig. 3. Design of Equip.nent 3 Two jacking frames were designed to permit the appropriate framing lhe ihmr.cciling furnace facility designed to pc form the standard' fire connection for the test assembly and inco Imrate the means to apply and endurance tests was the basic tool for the study. la was necessary, how-measure the restraining forces generated as the fire exposure test pro-ever, to mmhly that facility to incorporate equipment that would Imth pressed. Each jacking frame contained a face plate drilled and tappeal to apply and measure the restraining forces generated. J receive the 11 Scrics stect beam connections and had a rocker bearing at y ses immt of vertical suptwrt. The rocLer bearing rested on a roller as-Test Facdifics sembly consisting of a set of two 1.in.-diameter carbon steel rod rollers linrestrained scams-The initial phase of the study involved tes'is of 1 between two hardened secci plates. This assembly provided free hori-threc assemblics which were intemled to be totally face of end restraint. g /ontal movement of the hearing point, to accommodate longitudinal ex-1he test furnace was provided with I. casing devices which would insure p.msion of the test assembly and rotation at the end of the test assembly. simpic-span heam action.1hese hearing devices are shown in Fig. 2. g t he expansion cud jacking frame had a matched pair of 100. ton hy-

n test it. .ttpoos e tizAcate on enoincit) sMas efAM 7800e AND 800f Assimet 7: draulic rams znd a matched pair ef 54-va hydrzulic rams. Each pair of ,,...g,,,, rams was symmetrically placed.acor re s:rtical axis of the test assembly. The 100-ton rams were placed 2- =- me the midheight of the secci ' ' _*{,. - -- - ] **P -! ~ ] 3. -l beam. The 50-ton rams wer: -i. ~ m. below the midhcight of the secci beam. ,4 3 pv The fixed cod jacking frarr.e h.ni 1 pul: of heavy slect p,m hinges 10-y y I ((g yl { cated opposite the I(hton ram:s r r:e expansion jacking frame. The gj'i f l Ud. e i.a.a. 7 g._. i = fixed jacking frame incorporated a ra f :d pair of 50-ton rams equal ,9 l i rnd opposite the 50-ton rams m u.: :t o:hion jacking frame. This pro-j'y " G.,,,,.. g N i nl (

r.,..-.

+- g -- 4 y t.;t'" e.r rr.-- q y 5 v 5 a ,. r s n.-,.. j y --t c=.. - r i ; j f .t *, }j i - e El O ~~---- m T v Tr

ii

~ . t.,,.. : -.. l ~.- h (, { b ( ~ 2rr j y Y j ~i T L i 'f j (. l g g* 7 :3 p __.__t s.~ c - c ~ !-- 4. 1 T.o... e,. n O j i..... t =ts g._ _..t.. .__,__1 c. , a. a l _ p ,.v z.p l _7,.. __,1 =---A '7 7 . q t.... g, .. ~. -., ., 41 I. sl-or l g7 - ol a q g _ -t y.... r l ~ y, s - t see r .u .I t, A .1

*-- o s

p g j n_, .., w - a; _x q L.... g _ i --..,, ::.. - - aw h.c. .u a _..a + -- -u Q' p*q~ m : :,, FIG 4-Jacs: : n-r:

• restrmnr.l sests.

.,.,, y v,- m......: vided a structurally stable test a-r ::. ent. The 100-ton rams and op-Mco -T:.pn ul sess arrangement. posing hinges prov.ded the ma cr ?,: :10.:al thrust with capacity to resist i negative end rotation. The of:.s ; 5 -:an rams piosided the major re-perature Curve. Temgv;ntur measurements were recorded or the sistance to gesitive end rotati.:r. a-f addmonal capacity for horizontal unexposed surface of the : .t ammbly, at various cross sections on the pro-thrust. 'Ihe details of the jackin-; '. n are shown m I1g. 4. Icued steel beam, on the r.. tal deck, in the concrete slab, and at critical A plan and section siew of e :+: it:en. scc with typical unrestrained points in abc test facihty. II;'le ti<m was measured at the center and outer and restrained test assemblics is Kcan in lag. 5. qu.irter guints of the test amcr-bly span. Elongation and end rotation of Descsiption of iturrameenior.oc die test aucmidy was mer. ; red. Electronic load cells were used to meas-Chromel-alumel thermococ n 4::: u+cd to control furnace tempera- .ute the forces generated F. the hsdraulic rams in the end 'ackinE rames ~ I f turesin accordance with the.6INI Nf:that E 119 Standard Time-Tem, \\md E" '

• • P I hed at the two outer qua.-.:. pdnis of the test assembly span. Type SR-4

.....~.~.m~ draulic rams tnd .atched pair of 50-ton hydraulic rams. Isach pair of e .r-o-R w _..] , j rams was symmet. .ly placed about the verticti axis of the test assembly. g '**'*, ^ l 'Ihe 100-ton rams were placed 4 in. above the midheight of the steel t' a g beam. 'Ihc 50-ton rams were placed 7 in. helow the aridheight of the }3 -ll steel beam. g ,,6 = p 'the fixed end jacLing frame had a pair of heavy stect pin hinges 1o-cracd opposite the 100-ton rams in the expansion jacking frame. 'Ihe a [y ; i e g sC O e.='.

O - -

= _I. Si 'a 4d p i o fixed j.acLing fiame incorporated a matched pair of 50 ton rams equal

• • • *adae q

l { k lI and opposite she 50-ton rams in the capansion jacking frame. 'Ihis pro-g r h F.ei.. sent

-g

9,;.

4. p.,

l c rx- ) z p j j j h

g-t o.

.a a s.Y O -~ ~- - W e -f ---s,..-,'---~

l. j, j I(

,[ j ' ' j ' ~~ 4 -- J. 9" 2-n ir 4 l bs-f ( .e i a ,p u::r 4 g... ,~ - -.I l [ i _J I h ' _ t; t 3.c.s I ' Pt.. 4 o e. _ f._.._.c3 t I J 9 it ( a C g.. SU a C - C t 4 l. U I g e. .2 e

i. _ +_1. ' \\.,,,,,,,,,,,,,

d.

l t t n.. er.s - s..- - -.,- n u p . r p.a.s a ,...pd Hea j .q.s l ,,...w g. se' 9 [ { te.e. es g. j .'__,'.g. ,...... ] (-'m & - t 32wr 27 6..e 'L

• g a

+ P,. gResh ...g . ol g g - -- t saca a p.n _ his a_ a { t + g [,[J- ( - - t i.o.

&=.." 5m .X '.ee.[} te. p N .g j- [j e-g i

i. ___.

3 f I * *, 1 -- --- j

---

A g

,. < r e N h ..l' j I ..R. b,.,,... i. l-t e=penen be

• g es at t e..e. te..

s -[ (..,,.. f. 2.. l - - - - IS*- 3 f'd t s - - 3 T' C3 l8 "I'l lp.*. 1 srCHON A ~ inohtstm..1L.D.43 t-tG. 4-Jading fra nes for restrained rests. e vided a structurally stable test arrangement. The 100-ton rams and op-I IG. 5-Tyrv. f oris arrangemens. gmsing hinges provided the major horizontal thrust with capacity to resist 1, I negative end rotation. The opposing 50 ton rams provided the major re-pciainre Curve. Temperature measurements were recorded on the sistance to positive end rotation and additional capacity for horizontal micsiweil surface of the test assesulity, at various cross sections on the pro-e thrust. 'Ihe details of the jacking frames are shown in Fig. 4. S sc.i d >tect beam, on the metal deck. in the concrete slab, and at critical A plan and section view of the test furnace with typical unrestrained (3 pomis in the test facility. Dellection was measured at the center and outer and restrained test assemblics is shown in I ig. 5. g .guaisir points of the test assembly span. Elongation and end rotation of Description of instrasmentation ,M the set assembly was measured. Electronic load cells were used to meas-Chromcl.alumel thermocouples were used to control furnace scenpera

• y mc she forces generated by the hyihaulic rams in the end jacking frames l

,9 an.l.. measure and control the superimposed vertical design loads ap-turcs in accordance with the ASIM Method II i19 S4andard Time-Tem-g ph..I at the two outer quarter points of the test assembly span. Type SR-4 I h

e I e. speot usersp tospssa jo uopositdde Jatje peuistuism consioJ pua = * :puskal pau"!sJ:ssJ spus qsp stassu'o3 w a x x x x a emissed 3 (0) (0) Of) (0) Of) ('h) 01) "un 'uoissedza gemJaqi pamot:y sdig sdpt B . v; oot asisted u; 30, Q Es tuariom 0 0 olis..uom e. e e e pus wasq jo uopvioy uopipuos suisJ:say g 4 3 t1 t El C t! (*tt t'Cl (*tt (*tt t't! C Cl S'It ( t! (*Cl "tuods wesq ayt jo stutod JatJenb Jaino om: ty) f sdi3 *spuol uSis*O pailddy g u air [d pua paptag z di) (fi) (0) (?i) Di) (fi) Of) (!i) x x x x x x x x ( ui *asueJests U pua) uottsauuos sa Jat a pJepue S 5 x s x avoN 3 uopsauuo3 SutweJd u -(spnas Joaqs) airsodwo) g c x x x x x x x x x (swesq = on paptam Sntd qsap totaw 'uoit g . san plag tesidA1) atisodwos-opnasd x qeis pus wesq usamiaq uorisau uos testurysaw ou) attsodwosuo.y usisac l'!u l l s l Of l 4 s 4 9 l : D f f I g l 18 sendussaq tJ 4rquaesy nut

• ssojqwatto gott wesqpassated sof tuouspuos suroestas puo 'Sutwas]'uoparussuo2 utstag-g 3-}gy.1,

. ***Et oo u 5.= m..'.E^g a4 i. E 2 .ms 3 c. 2 U =e if - .E u.2 e y *a E. F a , u.-u .u_ o E .2'.e. u .e 2= ".=o c 's u u.c

s u

m - n. a .,. ? y e - u - = u e n 5 .y *E a., % Q ^=,* u.il a .a ~ '== .S "x 41 %u* 8-90 5.= 2 u c.E E 1

• 2 o

y? u e7 23u e va eUgtar .x = = me e2-s t a ea c

c. a u u,.

..e -: o e u = u e-c-Eo .= . 5 c.>u u- .= u

• o.=g, E E 3 c.c

.s=c o a

o. g.,

2%=o u -a a = -u @a

• aE yo-E

.= c n a u u u 3 g o== m3 o, - s u g c,u - m E = u gaa a : - o a s - a, o. :x 3 u n e n ,a *a u u u u u. - 2 a= .2 a o' = u,3 aeue=a .= = E s c. 3 u c = "5,.s u e e a e sq 8x-N 9 2 e, =n -5 y .= _. co eu a.e u o 3 gcu = = m - 58 = e u

• S 2.3 = 3

( 5 E.o 3o

..E.6.- u s u u.1 n = 5 9 s x

== E :- " 's '% 3 i ce3 y j:t t - 0 E t23u=xj .E!"2 6 5h 5 '=52 3 3 ~ .M$ d$$U$ 03# Y EOkE# s s.g E. a .u e = =.3 0 u. E = u S n 5 g,a 4 :,.8 C .w oc ,ua e: = _c. u 2 a.nE.2 % .u ca a e a - 5.u u a 4uL -, s =v J u.E e e3 M s3 = u o = e o t E 5ow s c.= 2 e Ec. M.5 1 ",- a y = eu M gL a u s a e o

• 3

-~ g*o. ,c-u=- g.E.c.= M a o e m32 =s z o u. e, 3. % *d c-u - u '5 a.g: 2 =

• u E

= c. 'g", 4 9 a N u c. u6 M = u a M .g g3 3 =$ y $.O E2 au E j M ~.g h*"6 e

c. c.y " -s a2 n = = u-a=

-w-o yA N $ E 0 E 8 E 2.M (, hE e.e

• 3 E

"U " ** "s O 3 8 3 72 4% g g M g.w W 2 E *2 w= u o "*8 "u = '3 4 5

• J, s l'

T a, E p2 8. ~3 2 E 'S - ~ 5 ~y a 0 3 g3N g,g2%g. u u IEU a E" ~~ 3 2 R "R e uM -yo4 c e :0 2 c.."3 .;; " g. c "e a "i! 3 s. u*8.3 o .E = = 3 X s 8,3 h.M u R C 8 e 2 s u o u o ~u u -s c .e 3 $ m 'E @ e~$ o,h 3 3.M M C e % =.e U "E. o u u oo

• E "oe c3=

- 8. G t.E R 5 E 8 C' e. E a- "3 8# .e ":'a u u - u c. a s -= 3 me, e u ^an w x - N "h c-o C M E 's tU .E U

• u G

@ !" hn ~e ; 3 '6, =" 5 $ 'E O '5 3 5 :.S k I M S $ $ $ $ "h'* 3. u d C.

• I 8 =, 3 '_E M

o u = u-o u . c co-o e. =r .,..u. .o y 3.. - a y' 1.o N._7 u g.> g, o y s a u .=. u,2 9 'We . C - M 3 8. = q e o g, o o'S.S.q ,3u Cc3 'j u 2.!?,a = yT-

• u i

Me u g3=a330s $<-0 3 'E

• 3 C.2 2* k u E w
• E e 3 *g.3 E*4 3

.l"ii t j 8 N o. y.c 4 he ~3 e O '=8 e u E 2 *,- i l 2 : E - ssE i= !*5a .c E s e s x o u a , c.= ur- > > =, G.= -=.5 =oo c. = x o e -: E 4 a t== o gu CW y a cv-

2. E = u g C.2-u -

9

y -3 a.:.

sr , x._ m - *

• u =u -

c. o .u u

• o a" -

e -u -s c s .E. .y - E"m u u

• u

-Eu .e.-== >uc Eu-u "u a e' u u w,.M u. o u e w Ou g a u u - c. 5 s u e u u o g e '3 c. I

4 74 rest kitHOos CETIAcKee ON PSOstCite sufft ef AM MOOe AND SOOf AssenettfS cicments in simpic hemling with varying magnitudes of cnd restraint. Incessian of Test Itesules Assemblics 15-4 through 119,11-11. and 11-12 were fastened to the lack-ing frames with 15 Scrics bolted connections. Assembly 15-10 was f astenedl'ic-fire load Test Arreof Stress Versus Pesign Stress to the jacLing frames with tmits through a stillened plate fully wchied The measured working stresses actually imposed by line vertical desegn o each end of the stect beam. During the progress of the test, the opposmg load for cach test assembly are shown in Fig. 6 together with the design pairs of 50-ton rams were used primarily to control she pcrnoised degree allowahic seresses used in the calculations of these vertical loads. Cosn-of cnd rotation. *Ihc pair of 100-ton rams and their opposing hmgc reac-paring the theoretical stresses wish those actually nicasured durung the lion were used primarily to control the permitted degree of longitudinal load test cycic,it may be seen that the actual working seresses were Ig to thermal expansion. 36 per cent hclow the design allowable stresses. This was metributed to the The basic test procedure involved maintaining, throughout the fire exposure period, the end rotation attained aller applitation of tlie vertical t. s......y. - ' "n--. _S~ r-f - '~ P sa r o... c.,o, .. m. c.-r- =, - - ~ g3 -/ f DL LL oL LL DL LL g.3 ,, 3 2s egc) 2 2tc) syl g.4 / ,i 2 2{c5 \\ 5 .e.s 2 2tel cl 1 .e of / / / ( g ( h ( so er oor m, m,y,i ,,,,,,,,-g,'4hs,, {g, g g /j j\\ f 2 rn_za a _z ri'Lzistit e j e2 e-4 e-s e3 o-is se /,f j/ / / l ,e 3pl in 3go i3 sJcl 9 1 26sl 40 e-2 f / 5-5 \\ 8 / / i s s \\ { / a a red <w n s a o. i h 3 3 j dsin is 7to er riit_ s ot'L_ _. Nu p g 3o i es e1 8- [* l s3 gcl 10 2. c) so6 cl 7, 5 5 I 4 20 i iL5DL_ @!L_ is si?L_. / j e-9 e to e-n e i2 1 S oit) 3 Sich $ 2ec) 2 4ts) 1 8 l0 K es2tti 13 9t') IS S(4 IS itU _f] g o ao 20 0 e to 30 4o so so 70 80 so co iso sao no eso FIG. 6-Brum.strcurs m,Jcr dnign fouJs. i tm. v ,u i design loads. On Assembly 11-10, the orientation of slic beam ends prior FIG. 7-L..uJ and rimc-Jcifccrime reimeiosisAip. to the application of the vertical design load was maintained during loth the verticalload application and the fire exposure perimi. On Aucmbly discrepancy in the assumptions made in accepted engineering design 11-11, the orientation of the beam cmts prior to the application of the practice as compared to the actual structural mechanisms built into the i vertical design load was maintained during the vertical load application representative beam-slab assernblics. In the present program, great care and the initial stages of the fire caposure pcsital. 'lhereafter, the end was taken to obtain idealized simpic span hemling conditions ehtough j slope was cased. On Assemblics 119 and 11-12, the orientation of the the use of hinges, n41 cts, and sliding cnd hearings. In addition, II.e test beam ends af ter application of the vertical loads wa's i saintained during awcmllics were constructed ami mounted in the furnace as "imicpendent the initial stages of the fire exposure perimi until a net end moment of heans-slah'* clements. 'Ihis meant that there was no h>rm of support along j 408 in.-Lips and 300 in.-Lips, respectivly, was achieved. 'lhescafter, the the transverse edges of the assembly, thus fusther assuring simple span end slope was cased to maintain the desiscal end moments. hending comlitions. These provisions rmt withstanding, the actual work-A summary of construction, Isaming and end restiamt for cach test ing stresses wcic still below the thcosciscally calculased design allowable Stresses. j assembly is shown in 'lable 1.

l l 76 riet - aatIHOos CIIIACZER ON PeOIECit~lb sitti C'Aas itOOe ANO 800f As5eans.stSApp The least discrepancy bceween the measured stresses and those calcz- ~ ~ rest seeees s. through e,e I: led by accepied engi,ccring design practice occurred in the composite pi.,osth,,,,,,,,,,,,,,,,,,, ~ '600 'eodeaet o' she she'=ocovoies test Assembly 11-3. In siiss instance the design awumptions do evaluate located of Ihe mid-height oI the the bending resistance of the concrete slah.m conjunction with the beams. .en ser sections s},q., N{ ,4og 'Ihc greatest discrepancy between the measured and calculated stresses l occurred in the pseudo-composite test assemblies (11-2 and 11-4 through l

e. - e, -

- se - o,e _ k gg , *g ll-II). In this instance the design assumptions do not evaluate the bcnd-mg resistance of the concrete sl.ab nor any composite action, yet the con- .-8000 %gg (Mg '[' situction techniques do develop some degsce of composite actum. The discrepancy between the measured and calculated stresses in the E Me,-o, Es -o: 800 e h , 600 f est Serses B. Iheough 8.: Plot of the overage tempercluse 16 M eeodings of the thermocouples locoled of the end ol the top g4ao fionge for sections Sj, L,N d-200 ,;;gd s - e, - mn 4,,, l200 l l M1m O to 20 30 40 SO 60 TO 80 90 10 0 11 0 82 0

• 1000 y,,,, g ;,,,,,

.5 5k5 Nh FlG. 9-Temperature eveb of processedsteel bearnes. g 800 go,..,, pg s,. 6* 7p{; ,i

1. ,y Test Series Be through 8,g d

pset es the overage temperoture 200 -g I6 M teodengs of the thermoCouples located on the top of the bottom O 10 20 30 40 50 60 TO 80 90 00 0 110 12 0 1400 8Bo*9e 'o' sections S {.t. N g Time, Minutes FlG. e-Temperature roi sup paruge of protretrJ stre-I beams. yl200 ,s a noncomposite test assembly (Il-l) results primarily because the design } ooo I-t I tssumptions simply do not evaluate the bending resistance of the concretc i sLi ng stab. The noncomposite test assembly is not representative of actual con- {eoo g struction because all bond between concrete slab and stect beam was y climinated. This assembly was included in the experimental study solely 600 ,/ to evaluate the idealized noncomposite case. The pseudo-composite and composite assemblies were representative of actual construction, and 400 jf / there is every reason to believe that the measured stresses were indicative g/ 200 of the working stresses which would be achieved under designjoadings in actual construction. O 80 20 30 40 SO 60 70 80 90 10 0 HO 12 0 Load and Fire Exposure Time Versus Deflection 1,me, u,o. lee The midspan deflection versus superimposed load and fire exposure f' rio. to-7c,.eperatures of sop of bors=,e p re of pror<cs,J sic,1 bc.ms s.s time for all test assemblics is plotted in Fig. 7. shrouch B 12. i n. i

l 7s it ,1 MilHOO$ Ctrl 1CIlte ON PROttC ED sifft StAA6 ftOOs AseD eOOP Assi ft It is intercsiing to note the variation in the load-dellection relationship lionships were obtained on independent beam-slab assemblies under sini-of the several test assemblics as shown in Fig. 7. Considering the pscudo-l pic span bemling comlitions with varying degrees of restraint. composite assemblics, for wluch the vertical design load was 13.3 Lips, the deflection varied from 0.287 in. for 11-10 to 0.110 in, for in-4. The Endurance Time Versus End Restraint noncomposite assembly,11-1, deflected 0.555 in. at midspan untler the Evaluation of the effect of end restraint was the major chgective of srme 13.3 kip loads. These initial deflections did not appear to inihicnce the study. The criscrion used do judge fire endurance time was the sus-the ultimaic fire cmlutance end point. tained load carrying capacity of the test assembly. In a single span beans-stah clement shis uhimate failure condition occurs when three hinge i Test seo. e, eheough s,. mechanisms develop in the beam. For the simpic span unrestrained cases Plot os the overage temperature in this emperimental program, two hinges for ultimaic failure are built ' 4600 seedsage of the thermocouples locolod of the end set the t>ottosa 1400 Ilonge Ier sections S {. L. N k i Tm han g we g8 8600 Piet el Ihe overage semperes,er. 2 teodenga et the thermocouples l i 1200 ,s - er e coseg og sh, c..,,,,gn,g,,,,,,, 8e* Set 1400 80enge for sections s }, t, se{ Y,, 000 i t ; g U Ski 88) 1200 /d ( 800 g o,-o, g e,-o,,-- t' lOOO l., ] s g " 600 f sli al R 000 Me,- o, E s,- e.,

#, 4 s 400 E

" 600 / # 2 / // 200 400 O 10 20 30 40 50 60 70 80 90 00 0 110 20 g 200 Time, Manuses l lG. t l-Tenuperatures of toe el botton, pange el p,noce ord un el ka,ns. O 10 20 30 40 50 60 TO 80 90 10 0 11 0 12 0 Fire Esposure Versus Ilearn Temperature 'Ihe temperature range versus fire exposure time for selected thermo-1 IG. 12-Tcenec,.s,cs cl e re,,c,.,p.,,,o,,, po,,rc of prosce scJ secci k,,es. couple locations on the proiccted sicci beams are ploited in Figs. 8 through 12. Table 2 shows the temperature recorded at load failure for into the assemblics from the beginning in Ictms of the hinged emi sup-each of the test assemblics. piuss. Thus, the fire endurance time is predicated on the clapsed sime 'Ihe appropriateness of temperature criteria for descrmining the fire under fire caposure acquired to devclop a plastic third i-inge mechanism cadurance end point of protected stect beams was considered in the due to Imsitive moment at midspan. study. 'Ihc temperature test data, summarized in Table 2 show that the the unrestrained lest assemblics all reached she point at which they highest steel temperature recorded at load failure was 1530 F. This oc-could no longer sustain the applied load at essentially the same time. *Ihc noncomposite assembly and the pseudo-composile assembly reached their cmacd in assembly 11-12 which was restrained under optimal conditions. ~ in the

• sis of the restrained pseudo-composite assemblics (114 through lue endurance end points at 90 min under vertical design loads of 13.3 g

11-11, constsuctions more nearly representative of ficiel conditions), the Lipt The composite ecst assembly reached its fire endurance end point at average of the indisidual high temperatiires at load failure was 1354 F. 3 93 min under vertical design loads of 21.5 Lips. In all sests, the high temperature at load failure occurred in she Imtlum P'sor to fire caposure, the worLing stresses measured in the bottom y flange of the protected steel beam. These temperature. load failure rela. Ilanges of she stect heams in the unrestrained mmcomposite (11-1),

~ Wue005 ctttzAcat) oss roosteteo safet etAas stoos Ape a00t Asseants SI

• gl ti 2 - S.mmary of serrf Arm.e armereaextr8 88 88*r "I i'*# l"'#"'r_

l N'c cadurance performance. Similarly, in'the case of slee pseudsman- - cluded that initial worting stress has sages 6caset inAssence on time tdtintale ~ $,2,...... n 4 s.a.. l,,,,,, ~ ~

• s*

m. i h. d - - - ~,,'"",. I u,,,l. posise assembly Il-2, a progressive dcoctioracion of partsal-comipassee ac-tion may sake place, ehus bringiat the inicial working seresses neore in i a,,,,,u, s.. 4 ve.e r m sa - ac= y *_s; ,ys ,1, line with shose of the coniparative compossee and ma=ca pe=ine test as-r,.g.. '.'.'.*/ semblics. Fueehermore, the thernial stresses induced in she sacel beans as a 1875 l HIM8 result of the monuniform temperasure disteilmsion between lap and bnt. 530 978 819 i O I noncompouse untc. N.lsment 1895 tom Ranges may be the overrideng factor in tellisante Ere 4 sumined L C$ak' Ij," ', 3 " ' ', " $ "M 'Ifg For the restrained cases, she application of negative end =a=ne=ne re. p l S. pimis 32 pse:udo composisc( N 1 twent il I 5 quired k, maintain the initial end slopes climinated alte initial assechanical h nge action provided by Ihe jackisig franses. Tleis negated the sienple ,,5 S ; point 1895 1875 Han 510 9m 1895 Span hending action which caisted at the start of flee test and, in cNect, p1 conipossle unre.1 point i 2 HH5 1 g,,,o,,,,, i ,onined 'm im H,. S, H., im ,,,, p _. _, _._,, j,,,,""],"i f - H b T, N point m5 im n5 m5 Hm m5 M++' 4 ++4 84 pscudo composene p g ) l S poing 1375 1315 1855 585 IHs 1175 '[ I 3, I," B.5 pseudo-co.osmsuc N point 1110 IM Il j w, o,,,,,, l S point l'ap tilo 1820 580 Has im* i 1285 1245 1075 Sjo isn4 1285 ,,gg[,fcd W,, W,, W,p j Cente 1315 12M lim 6m His? W5 [ S. ; point 1520 IM 1880 5M HM 33M I g.7 psca do composite N int s2co 12m IHn 565 un6 1260 g, "'"a' ( s.:, min, im 12w nm een Hn8 nm tyg5 12ml 1825 62n 1058 IM -O, ",',,,,;,','"d* *""*'*{ "df,""" e i I' nm irm n65 an na5 im ~ $ ce ',,,,, t,& v N i S point 1545 im lim m5 nI2 1345 p., 9 9 g9 pseudo.composene N point lied) 8135 1185 5'NI til8 ll*8 1185 Ha um IWd) 1201 1385 cose e come 8 case 3 ~ \\ $. t gwins l)M I245 l2N MS IIO I# FIG. I1-Tyrical hending co*educians smder verring applied ened meneernos. B '10 pseudo composite N.! poins 1165 1110 Ilm 620 til9 IW l ecstsained. Ocnict ilm l uun 100 M5 lin4 shortened the positive moment region of the beams span by creating } ( S point IM IN HM counternenure points somewhere within the original span length. Fegere i e il pseudo 4 omposiac 13 fieriher illustrates this point. Omceptually, this acceanes for the pro-5f cco!,c,oeng kn# be cerance d sim# span kam-M asseMs wie emi N p im im IIIe m5 lin im """"d"' so im Irw lim m in9 Im restsaint. Discounting the epect of shermal seresses and the cNeet of teni- """3 g s. ; imio sim ms im 6m 1074 83*0 e perature on she material properties, it can be seen in Feg.13 that if a j l N.1 point HIS 1185 1 14 1 8 54 ngative resiraining end moment equal 10 one half the positive bending i s.12 cmnposite restrained Ccne wo un Im k "**C"' C' eased by the verlical design loads is applied, the cNective posi. j Inc beneling moment at midspan is reduced to one half its original value. .} 11 a negative restraining end moment greater than one half the vahse of . rsiimakd- ',scudo-composiac (11-2), and coniposite'til-3) test assemblics were 17.5, ( Ihe pmitive bemling moment is applied, she effective bending suoment at l i4.7 and i9.s ksi, respectively; and the respective Gre emlurance emi nudsp,m is further reduced. The limit, in this case, is full limity at the end, p.;s we,e el. and 93.in. in view a a.uh e orde a ++: 1 0 <"nai'mhich develops a manimum end mosnent of 3PL/I6 and a the n,e cmiurance end points in these ihree tests, is should not be con: 'csulemg gmsalive moment of PL/16. l

l g3 (# si METHODS StillAct.It ON PSOTECitD sitti MAM ftOCa AND COOf AssrMM2 03 l,g and l'L/8, the moment at midspan will vill govern, and f ailure would be

l -

g l by the plastic hinge formation at midspan. The smaller end moment. I g would still create a prolonging cifcca on fire performance, but not so significant an increaw as the former instance. From this analogous situa. l. g'- (,, f tion, without hes.t and temperature cliccts, it is readily apparent that i I M j -'. *.,,,,'

• il

= theoretically the time dillercnce between the restrained and unscstrained 4 .g , N.*", assembhes is p imardy due to the magnitude of end restraint isnposed. i a' ' ' p m-- "15 ._ g i 7. k '.9[ hjr .s \\s t- ,i l N 4 s.'.5 o., 7 -e..M i g, t' '

• s
• - 'ly. rjy y... s i. J. : ' ' ' "' m.

gy e' ,n ;- s A '_m=L %, .~ k.f.. \\.Q'-{.;, i d.

i. -

e l l ~~

• \\

% 6bf ' . 't y." l l l E H & CEI H E Nh... m.f DY U d3 b{ 3 M p,,gp gj, ff f: I' ..,.,;p} ?' .~ h h k(/, i i ]D$ s p -wan b h v. g; o , %.. e..... gm. gf 4 1 1 g-udig;,g gg.... -~j

i. + - ' hf_.. ;

e mAA....- h$ '- via. ta-continuest. - I The major portion of restrained pseudo-composite test assemblics (11-4,11-5,11-6, and 118) reached the fire endurance limit in the range FIG. t4-Tygwas verws of aucent.1,cs afscr rest. ffom 111 to 113 min. Test Assembly 11-11 would have also reached the fire endurance limit in this time range had the applied load not been re-Using the example in Fig.13, when the negative restrainigg end mo-l ment is between l'L/8 and 3f'L/16, ultimate f.iilure cannot occur until duced at 108 min. The increase in fire endurance time demonstrated by these restrained assemblies compared with the unrestrained assen'iblics j hinge mechanisms develop in the steel 14am at the center of the span as w.is approximately 25 per cent. Test Assemblies Il-7 and B-10 reached ) well as in the negative bending region near the ends. 'Ihis would provide their fire endurance limit at 107 and 106 min, respectively. Figure 14 the range of maximuni increases in the fire endurance time. shows various typical views of assemblics after the test. 8 In the case where the negative restraining end moment is bewicen 0

L%. es roe rt 'inoos CffliCitte ON PSOfECTED sitEt gram ptoog ann 8007 Asstanet75 l The performance of test Assemblics 15-4 through 11-10 and cve:s test f m, k,eps. Aher the 408-in.-kip restraining end moment was attained ~ ' Assembly 11-12 cicasly imiscated that cml restraint alone wouhl set pro. at was hchi constant and the end rosatiim was cased. The Ere enduranE a duce entended Ere endurance perfosmance such as had been experienced cml point aclueved usas i15 min-the kmgest time attained to that point in some AS~lM stamlard fisc tests. 'Ilieschwe, since design lo.uls calcu-en &c caperiments, but mW a significant increase. 1:ted by accepted engineering proceduces were appised to the asscenblics On Assembly 1112, a similar gest procedure was used, but &c end is these fire tests, additional factors inherent in the construction must nunnent buiMup was limited to 300 in.-kips. The Ere endurance end i li d loads, . have come into play to clicctively reduce or redist:ibute lhe app e Imint achieved was 141 min. Assembly 11-12 was a compassee camstruc-moments, or wo Ling stress as the 6te test psogresses. To verify this, the '"* "*I' OC',CI"'c, the vertical loads sclected for the purpose of tinis test applied vertical knads on Assembly 11-1I were arbitrarily reduced from were detenmned by measurement to imposed stresses in the bottom l 13.3 kips to approximately 5 kips after 108 min of nec caposure. Ihis Mange o{ &c protected secci heam representative of the pseud &c-: 7 =* repscsented a reduction in the applied load of approximaicly 65 per cent. 85SC*hlses. The vertical loads were 14.8 kips. The purpose of limik the i As a result, the test assembly sustained she reduced load witimut failure restrammg cnd monient buildup to 300 in.-kips was to case the assembly i for a ental of 176 min of Erc exposure. 'Ihc results, coupled with those into a simultaneous hiemation of the two negative moment henges near of the vasions restrained test assemblics, cleasly indicate that the extended fire endurance espesienced on some full-scale assemblics must h: ve re-g 5 i sulted from load or nuwncnt redistribution. 5 e40 a 5 Failure Mechanism f it became apparent, as the c.perimental program progressed, that within a fairly bsoad range the magnitude of restraint imposed had only a 1 minor clicct on the 6te endurance limit achieved. Again, using the anal-AN - ogy illustraicd in Fig.13, when the end moment is between the limit of j w 800 PL/8 and full 6xity, the plastic moment capacity of the stect is a constant. g-

• Ihus, the only factors which could be capcceed to inermluce a time varia-

[ g l ti<m in the Ese endurance performance wesc the axial thrust, which would o 4oo son u... . o... n,,,,,,, cause added compressive stresses to timsc induced by the negative cml l(1%I

• tu t) = samu )

o moments in the bottom Hange of the beam near the emis, and the tem-j I'lG.15-fsegrcr of rcserains i,crms pre c,,J ,,,,,,g ,~ pcrature clicct in flic sanic region. The nature of slic initial structural failure which attended cach of the the ends with the posisive nmment hinge at the center. The prolonged j test assemblics was a buckling phenomenon in the web and Imasom Ilange emlurance time cahibited was altributed primarily to this procedure. This j of the protcceed steel beam at approximately 2 ft from cach end. 't his in. is etimvalent to lhe anak>gous comlition simwn in Fig.13 isuposing an dicated that the formation of the initial pair of hinge mechanisms was j caused by the cosubined amial thrust and negative cml moment which j cml nmment of PL/8 with the thermal effecis included. prmluced excessive compression in the lowcr flange of the sicci beam. E#cct of End Restraint I l Altimugh such failure maics have been emperienced during actual fire One appniach to defiining the characect of end restraint imposed in j j tests, they are the execplion rather than ihc rule. This led to a pre-this experimental program consists of summing the absolute vahacs of liminary evaluation of the combined bemling and thermal clicces, which imsauve and negative end moment applied by the rams and ploiting this l indicated that some relatively low magnismic of restraint wouhl prmfuce suon versus ' ire cmlurance time. Such a plot does reflect, in a general a significant further increase in fire endurance time. fonn, an increasing sense of cml restrains and includes the effect of imah i Expcsiments on two lest assemblies were pcsformed in an attempt to dunst and nunnent but does not distinguish helwcen the clicci of shaust dernonstrate a further increase in the fire endurance time. On Adembly and nunncnt. A plot of the maximum value of positive and negative re-Il-9, tlic test procedure involved maintaining the end rotation attained straining momcots achieved in cach lest assembly in this program versus after the application of the superimposed loads by the gsadual huihlup Grc cndurance tune is shown in Fig.15. of negative end moment by the use of the 50 ton ram pair to a limit of *, I'he time at which the maximum value of positive and negative re-

9 titirAcate OH reOitCID sifft efAu f tOOa AND a007 AssfMMCs 87 slet IIst MitHODs to prmluce simultaneously the negative moment hinges near the beam raining moment was achieved varied for each test sssemHy, buil ng g for those assemblics muse highly restrained. Th.- va ucs I n oment versus time for Assembly 114 are shown in Fig.16 , Egers of ApplicJ Verticallmds strais i m are typical on the moment buddups for all of the restiamed test as. The vertical loads theoretica'dy calculated by accepted enginecting emblics. In Assembly 15-8, the maximum restraming moment value was procedures to impose design allowable stresses actually imposed meas-chieved et 80 min while the fire endurance time was l12 mm. lhe inured initial working stresses in the range of 18 to 36 per cent below the crence is that the combined axial stress plus bending stress in the sfcc design stresses for the assemblics involved in this study. These re'ults .I : s . cam incacased to the point that, for the temperature in the cn,tical fibers' \\ were consistent with the measured working stresses observed in other nci 3ient yicid of the secci ensued. The initial pair of plastic hinge mecha-studies of sisuctural performance; as well as measured working strenes s nctr the ends of the beam developed at or about 80 min of fire ex observed in assemblics subjected to the ASTM stand' rd fire test. Struc-msure From that time, restraining me, ment, generated decicased unti aural mechanisms inherent in construction, as opposed to design assump-p a iis [ N he final midspan hinge mechanism developed and ultimate structura tions, play a significant role in reducing the actual working $ tresses to a in i level below the allowable stresses. It is also faidy well substantiatett that ([ . ilure occurred. t Om further point should be noted with respect to end restramt. Test a , similar relationships between design allowable stresses and workicg Assemblics ib4 through 11-7 were all constructed with the concrete slah strewes exist in actual buildings. The restats of the study indicate that P l' the magnitude of the applied vertical load is a raore significant parameter _l._ _! of fire endurance performance than measured initial working stress. f aoo ?

• T:

~Q R.T Temperature Criteria ag 24 u E From the results of the study, it is clear that for load bearing assem-l . onn' ~ ' soo.r a 1 {' b]jes, temperatt,Te per Je does not provide a meaningful indc1 of fire 16 n, j. g performance. Temperature a,d,the temperature distribution are factors, gd - I

• • "f--

as are t?.c vut? cal design loads, which combine with the magnitude of I g p 4o. so - ao end restraint to determine the forces and moments to be resisted by the - im, umm structural assembly. In those instances where an asserd.ly is tested with-m 8 6-Af an.. a' h"'"t"""'"""'""""" out load, or the structural mechanisms of a tested assembly is to'bc al-8d tered by substitution c.f dher structural clements, the use of temperatue cut back at the ends to climinate restraint generation by capansion of the may provide a vah.shic means of evaluating fire performance. For ex- ,i shb. *Icst Assemblics *.U8 through 11-12 were constiucted with the con ampic, temperatures may be used as an end. point criterion for evaluating certe slab fulllength astd grouted between the end pcking frames In cac

• protected steel beams not tested under Imul hearing conditions or when instance, on Assemblics Ib8 through 11-12, as the test progressed, a crac h

is i dd k m a & md he h I formed and widened between the concrete slab and jadmg frame as a ihme with which it was tested. If the individual high temperatures were result of the positive end rotation of foc test assembly. 'Ibc capansion o g g g g;,;g gg g g g gg the concrete slab, the cmdest clement in the assembly, was not sulhesent Nm h ali d M W F M p a mi a h n e k to ovcicome the rotational crack. Consequently, essentially no end re-g ,ggg

g g straint was generated as a result of the concrete stab.

csiterion of failure for protected stect beam assemblics tested under l proper vertical design loads. Conclusions l 'l. Dnign Ellect of Restraint ) Review of the data on the unrestrained test assemblics, which involved The cifcca of restraint on the fire cadurance performanc.: of the beam, tesk of consernctions designed as composite, pwudo-composite and n,n. slah anemblies was an incicase of approximately 25 per cent m ihc cn-composite by accepted engineering procedurcs,ineheate that the three dif-( durance time, as compared to unrestrained assembhes. The,mcrease in Ierent designs exhibited essentially identical fire endurance performance cmhiras.cc time for an opeimal kvci of restraint was 50 per cent. 'Ihe op-even though the measured stress in the bottom llanges of the steel beam timum occurs when the magnitude of restraint generated is just sullicient

o,m ,,_,m,_,,,,,_ se .rr irsi armoos lt. DISCUSSION ~ lemn the applied design loads viricd from 14.7 to 19.8 ksi. The magni-tude of design verticalload was 13.3 Lips applied at the Iwo outer quarter points of the beam span in the case of the pseudo-compwitc and non- ~, comgxwife assembly, and was 21.5 Lips in the ene of the compositeN. S. Penice' (writtero discussion)-The author af this paper is to be assembly. The componite assembly performed as wcll as the pseudo.congratulated on having presented a detailed and well-documented paper comgwnite ami noncoinposite asseinbhes, yet carried 62 per cent addi- } concerning a fxet of fire testing which has hitherto not been catensively emplored. tional vertical design loads. Perhaps the most significant of the conclusions derived from the series I of fire tests described by this paper is the fact that the unrestrained as-AcAnowledgments ' Ibis investigation was part of a scscarch program (9l spamsored by sem > lies cahibited an appreciably lower degree of hee endurance than American Iron and Steel Institute. 'lhe research contract was adminis-did the assemblics which were subjected to both waial restraint and posi-tered on behalf of the sgumsor under the Commitice on fluilding Research tive or negative restraining moments. This is contrary to a previously hsid and Technology by W. G. Ki:Llan<l. vice. president, buihling research, belici which is reflected in current testing practices in accordance with E. l'ahy, chariman of the l' ire protection Subcommittee. L A. the provisions contained in ASTM Method E 119. and II. llenjamin was chairman of the task group representing the stect industryIt is generally agreed that an ideal objective is that the restraint condi-tions imposed during a fire test of a roaf or floor and ceiling constructiost in this research program. Consultation and comtruction assistance on the fire protective insula-should be representative of the restraint generated in actual buildings tion was provided by L. A. liarron, managing director, Vermiculite In-under fire conditions. The test method described by this paper illustrates stitute.The sicci stud shear connectors were provided and installed by thethe practical possibility of applying a particular or programmed degree o Ncisem Stud Welding Division, Gregory Indmities, R. C. Singicton, vice-resu aint to a relatively simple cicment of building construction. The ques-tion that should now be asked is whether we are to proceed with estab-president for engineering. Research cemsultanis were C. II. Smith aml L F. Lindley Department lishing " typical" or "most detrimental' conditions of restraint, or whether of Civil Engineering, and 11. E. Gatewixx1, Depaitment of Acronautical we should recognize the increasing complemity of such a task anaf icsolve and Astronautical Enginecting. Research associate was W. W. Lanc. 'lhe to modify the present test method to provide what is essentially a mean-test assemblics were constructed and the tests performed by the fluilding ingful comparisou between the various designs. Research Laboratory stall under the supervision of L G. Ilirle. R. W. filerwAer (arateor)-The comments by Mr. Pearce are appre. ciated aml he points out the dilemma which now must be considered in terms of the "propcr* degree of restraint to be imposed in fire tests' R'I*"S pl Ar nual of seul c.mormenm. 6th ed, Amentan institute of Sacel Condrue-ASTM Methtml E 119. This brings into focus the philosophy which un-til IEr's / r i tc.I/, emperaen,c Sernre. 4th poniing. timied Siaies Steel Corp, deslies the Icst requirements. Should the test method attempt to provide 3 r m b m sh,ra,i961. results in terrus of fire endurance of construction systems in actiaal IJi t~scudenitial. A. M. The lar'ainer uran.ior of t.ncturrrene Arairriuis und sir =- buiWings undey Grc conditions, or should the results attempt to pro Frames-t ficit on Iise Resistante of floor and a comparative amicx of fire endurance of construction systems under some (<l tite $4Ie R W., Is I n re.hng Anembhes t ur i n huolocy. vol 2. No. I, Feb.1966. pp i5-23. standardized condition? It would be inappropriate for the author, at this ist Gaicwo.=1. n. E, and Gehiing R. w., -Arrowable Au d I oads and tien hns M dd M l@ m k a W m M % mu h M k Mimienn for incl.naic $sostenses under Non umfosm temperatme Descobu-li P e fournal of shc Arrmemr Jurm es. Vol 29, No 5, May,1962. pp 58 3-and Mr. Pearce and their colicagues on ASTM Committee E-5 are deeply engaged in precisely this subicct. It must be recognized, however, that if

tion,

[6l t trect of 1.ong Fnposnee of Concreic to elich Tempera me." C,mercir infor-we auempt to alevciop the he endurance of a construction system in 52n (7i $.I[Ie c",Ni, nf.'rI amn. ST 99. Portland actual buildings under fire conditions we would not obtain a single-valued n.r re answer, but rather we would have to measure a range of performance remeni Associai.on. chicago til. j l8) llaonathy T. Z. and Hesnds, f E, "flydrated Portland rement and Iight. levels depending uptm methods of structural framing existing in a single " ' '"' *' " " " ' " '^ ' " ', vol 61

• "8)' $*",'f,'6d" '91 buiWing as wcll as the methods of $ttuctural framing of any and all a

[9] Ilictrader, R. W., "f#ar of Ssrurtural Restraint,m she fire Remruner of g Protn ord Steel Urum I'laar and Kool A ormbhrs. Ilmastmg Rew nh I atura- . Assistant chief engineer. Underwriters

• I ahoratories of Canada, Scarborough, 246/266. Sept.19 >6, I:ngincesing limpesimens Ssation, lhe tmy Report ii S Ontario, Canada.

Ohio State tiniversity, Columbus, Ohio.

.Am NN ' 8 23" '" j._ N. I I 1 ,o em arrucos f, i h { $ I buildings into which the construction system under consideration could , e j.E l.i be incorporated. II, on the other hand, wn.necupt to develop a compara- ,,, h live index of fire egdurance of construction systems under some stand t { ardized condition, there will invariably be some construction systems for-I. t 9 which that standardized condition would be inappropriate and some con-9 ,k struction systems for which that standardized condition would be either .I i -l gr unduly beneficial or unduly detrimental. It is obvious there are no simple t I answers, and it is inevitable that both the test method and the evaluations ['..!l. resulting therefrom will become more complex as the state of our knowl-i L edge advances. t The author does not agree with Mr. Pearce that the most significant r. e conclusion to be derived from this research is the fact that the unr l l'. strained assemblics demonstrated appreciably lower fire endurance than the restrained assemblics, since this was fully expected and is a com 3 I. ( pletely predictable result under the conditions existing in this research-1 program. Possibly Mr. Pearce's statement that this was contrary to a 1 previously held belief refers to the reported expect.3 ion by the members f L l '.((. of Committec E-5 when the stipulation regarding restraint was first in-bdOO i it l serted in the test method (presently contained in Section 23(6) of ASTM Metluxi E 119-65) that this would be a "more detrimental" con i of test. This has not proved to be the case, however. Fire tests on steel

3 beam and steel bar joist assemblics. almost without exception, have been I

I ?l catended as the result of imposing conditions to generate restraint during j [ j i the test. In the case of concrete structural assemblics, and particularly g. j,, j* prestressed concrete systems, moderate restraint has increased the fire en-I, j; '. durance more than four. fold according to the work of the Portland Cc- ,e j i ment Association Laboratories reported by Carlson and Selvaggio. On the other hand. instances when " full" or " ton.d" restraint was imposed 1 f g ~A 4 have resulted in premature failure due to czplosive spalling caused by k % ".. t 3 the large axial restraint and compressive stresses in the concrete me I / mber. a a A ? l j l,. dl ,1 I gl l J ' ., y q ) -,tl l i l u i i I I -(h ,,}}