fire resistance of post-tensioned structures journal/1973/march-… · other characteristics of...

FIRE RESISTANCE OFPOST-TENSIONED STRUCTURESArmand H. GustaferroConsulting EngineerThe Consulting Engineers Group, Inc.Glenview, Illinois

Reports on an analysis of 18 full scale fire tests ofconcrete slabs and beams prestressed by post-tensioning. In addition, results of tendon-anchorassembly tests performed at high temperatureswere analyzed. Recommendations on minimumcover thicknesses and member sizes are included forfire endurances of 1, 2, 3, and 4 hr.

More than 140 full scale fire tests ofprestressed concrete structural compo-nents have been conducted in theUnited States. In addition, researchershave studied the properties of steel andconcrete subjected to high tempera-tures. They have also measured andcalculated analytically the temperaturesthat occur in structural componentsduring fires. Currently, methods are be-ing developed to calculate the capacity,deflection, expansion, rotation, andother characteristics of structures sub-jected to fire.

ScopeThe purpose of this report is to pre-

sent an overview of pertinent informa-tion concerning the fire resistance ofstructures with post-tensioned rein-forcement. Information has been gath-ered from a number of sources. Resultsof fire tests of 18 slabs and beams withpost-tensioned reinforcement constitutea major source. These results were

compared with other information, i.e.,

tendon temperature data were com-pared with data on temperatures withinunreinforced slabs and beams, and thenthe resulting fire endurances were ana-lyzed. Because the temperatures of thetendons were in all cases cooler thantemperatures at comparable locationsin unreinforced slabs, it was possible tomake conservative recommendations onminimum dimensions for various fireendurances. In addition, results of re-cent high temperature tests of tendon-anchor assemblies makes it possible todetermine realistic cover thicknesses foranchors.

Standard fire tests of building con-struction and materials (ASTME119)1

The fire resistive properties of build-ing components are measured and spec-ified according to this common stan-dard. Performance is defined as the pe-riod of exposure to a standard firebefore the first critical "end point" isreached.

38

The standard fire exposure is definedin terms of a time versus temperaturerelation. At 5 min the furnace atmo-sphere temperature is 1000 F, at 30min 1550 F, at 1 hr 1700 F, at 2 hr1850 F, and at 4 hr 2000 F. The firerepresents combustion of about 10 lb ofwood (with a heat potential of 8000BTU per lb) per sq ft of exposure areaper hour of test. Actually, the fuel con-sumed during a fire test is dependenton the furnace design and on the heatcapacity of the test assembly. For ex-ample, the amount of fuel consumedduring a fire test of an exposed concretefloor specimen is likely to be 10 to 20percent greater than that used for a testof a floor with an insulated ceiling, andconsiderably greater than that for acombustible assembly.

The standard, ASTM E119, specifiesminimum sizes of specimens to be ex-posed in fire tests. For floors and roofs,at least 180 sq ft must be exposed tofire from beneath, and neither dimen-sion can be less than 12 ft. For tests ofwalls, either load-bearing or non-load-bearing, the minimum specified area is100 sq ft with neither dimension lessthan 9 ft. The minimum length for col-umns is specified to be 9 ft, while forbeams it is 12 ft.

During fire tests of floors, roofs,beams, load-bearing walls, and col-umns, the maximum permissible super-imposed load is applied. Floor and roofspecimens are exposed to fire from be-neath, beams from the bottom andsides, walls from one side, and columnsfrom all sides.

End point criteria for floors and roofsare:

(a) Specimens must sustain the ap-plied loading—collapse is an obvi-ous end point.

(b) Holes, cracks, or fissures throughwhich flames or gases hot enoughto ignite cotton waste must notform.

(c) The temperature of the unex-

posed surface must not rise anaverage of 250 F or a maximum325 F at any one point.

In 1970, new end point criteria weretentatively added to ASTM E119 forfloors, roofs, and beams fire tested in a"restrained" condition. "Restrained" inthis case means that thermal expansionof the specimen is restricted during thefire test. Two classifications can be de-rived from fire tests of restrained speci-mens, "unrestrained" and "restrained."Only "unrestrained" assembly classifi-cations can be obtained from tests ofunrestrained specimens. The tentativerevision of ASTM E119 includes aguide for classifying constructions asrestrained or unrestrained. In the guide,cast-in-place and most precast concreteconstructions are considered to be re-strained.

The new end point criteria are basedon critical steel temperatures. For struc-tural steel and reinforcing bars the criti-cal average temperature is 1100 Fwhile for cold-drawn prestressing steelit is 800 F.

For unrestrained members, the fireendurance is the time at which thecritical temperature is reached. For re-strained primary beams, which are de-fined as beams spaced more than 4 fton centers, the fire endurance is twicethe time at which the critical tempera-ture is reached. For restrained slabsand for restrained beams spaced 4 ft orless on centers, the steel temperaturesare disregarded.

Rational design proceduresIt was noted above that methods are

currently being developed for calculat-ing various parameters concerning thebehavior of structures during fires.Even though it is not the intent of thisreport to present a comprehensivetreatment of rational design proceduresfor fire endurance, a review of some ofthe principles involved may help inunderstanding the behavior of struc-

PCI Journal/March-April 1973 39

tures subjected to fire. For illustrationthe behavior during fire of three typesof flexural members will be discussedbriefly.

(a) Simply supported slabs or beams.Consider a simply supported reinforcedconcrete slab subjected to fire from be-low. Assume that the ends of the slabare free to rotate and expansion can oc-cur without restriction. Assume alsothat the reinforcement consists ofstraight bars located near the bottom ofthe slab. With the underside of theslab exposed to fire, the bottom will ex-pand more than the top, and the slabwill deflect. Also, the strength of theconcrete and steel near the bottom ofthe slab will decrease as the tempera-ture increases. When the strength ofthe steel is reduced to that of the stressin the steel, flexural collapse will occur.Such behavior has been clearly dem-onstrated in prestressed as well as rein-forced concrete members.2

It is apparent from the above de-scription that the steel temperature atwhich collapse occurs depends on (1)the stress in the steel, and (2) the typeof steel.

The stress in the steel depends on theload intensity on the member. For ex-ample, if the steel stress is 50 percentof the initial yield strength, the criticaltemperature will be about 1120 F.However, if the steel stress is one-third of the yield strength, the crit-ical temperature will be about 1220 F.The temperatures would be differentfor cold-drawn steel or high strengthalloy steel bars. Thus, it can be seenthat if the load intensity is decreasedthe fire endurance will increase.Through rational design procedures, itis possible to estimate the increase infire endurance due to a decrease inload intensity.

(b) Continuous slabs and beams.Structures that are continuous or other-wise statically indeterminate, undergochanges in stresses when subjected to

fire.3 It should be noted that this is dif-ferent than simply supported memberswhere the applied moments at a sec-tion remain constant during fire expo-sure.

Consider a two-span continuous slabwith rocker-rollers at the outer sup-ports. During fire exposure from be-neath, the underside of the slab ex-pands more than the top. This differ-ential heating causes the ends of theslab to tend to lift from the outer sup-ports thus increasing the reaction at theinterior support. This action results in aredistribution of moments, i.e., the neg-ative moment at the interior supportincreases while the positive momentsdecrease.

During the course of a fire, the neg-ative moment reinforcement remainscooler than the positive moment rein-forcement because it is further from thefire. Thus, the increase in negative mo-ment can be accommodated. The re-sulting decrease in positive momentmeans that the positive moment steelcan withstand a higher temperaturebefore failure will occur. Thus the fireendurance of a continuous member isgenerally significantly longer than thatof a simply supported member havingthe same cover and load intensity.

(c) Members in which restraint tothermal expansion occurs. If a fire oc-curs beneath a small interior portion ofa large reinforced concrete slab, theheated portion will tend to expand andpush against the surrounding part ofthe slab. In turn, the unheated part ofthe slab exerts compressive forces onthe heated portion. The compressiveforce, or thrust, acts near the bottomof the slab when the fire first occurs,but as the fire progresses the line ofaction of the thrust rises as the heatedconcrete deteriorates. 4 If the surround-ing slab is thick and heavily reinforced,the thrust forces that occur can bequite large, but considerably less thanthat calculated by use of elastic prop-

40

100

80

U-0

60-cC,

in

0 40

20

070

(3) High Strength AlloySteel Bars(Tensile Strength)

\` (1) Hot-rolled Steel

.` (Yield Strength)

(2) Cold-drawnPrestressing Steel ♦♦♦(Tensile Strength) \ s

♦

200 400 600 800 1,000 1,200 1,400Temperature, ° F

Fig. 1. Temperature-strength relation for hot-rolled, cold-drawn, and high strengthalloy steels. (Curves 1, 2, and 3 from References 5, 6, and 7, respectively.)

erties of concrete and steel togetherwith appropriate coefficients of expan-sion. At high temperatures, creep andstress relaxation play an important role.Nevertheless, the thrust is generallygreat enough to increase the fire en-durance significantly. In most fire testsof restrained assemblies, the fire en-durance is determined by temperaturerise of the unexposed surface ratherthan by structural considerations, eventhough the steel temperatures often ex-ceed 1500 F.

PROPERTIES OF STEEL AND CON-CRETE AT HIGH TEMPERATURES

Physical properties of steel and con-crete are affected by the temperaturesencountered in fires. Strength, modulus

of elasticity, expansion, thermal con-ductivity, creep, and stress relaxationare all affected to some degree. Insofaras ultimate capacity during fires is con-cerned, strength is of primary impor-tance.

Steel strength at high temperaturesFig. 1 shows typical relations be-

tween temperature and strength forhot-rolled steel, i.e., reinforcing bars,cold-drawn prestressing steel, i.e., wireor strand, and high strength alloy steelbars. 5 , 6 , T Note that one-half of thestrengths are retained at about 800 Ffor cold-drawn steel, 1050 F for alloysteel bars, and 1120 F for hot-rolledsteel.

Concrete strength at high tempera-tures

Fig. 2 shows the temperature-


CoCC,

Oo-0rC,

CdN>

aE0U

Carbonate----nom'

Sanded Lightweight

5 Siliceous --+`

LStressed to 0.4 fL during heatingOriginal strength = fL

70ҟ400ҟ800ҟ1 wuҟ00

Temperature, °F

Fig. 2. Compressive strength of concrete at high temperatures (Reference 8)

0 1 1 I A

strength relationships for three kinds ofconcrete. 8 Carbonate aggregates in-clude limestone and dolomite whichundergo a chemical change at temper-atures above about 1300 F, i.e., carbondioxide is given off from the calciumand magnesium carbonates. Heat isused up during the reaction so thetemperatures within the concrete re-main somewhat lower than for noncar-bonate aggregates. Also, the resultingproducts are better insulators than theoriginal aggregates. Siliceous aggre-gates include quartzite, granite, andsandstone. The data for sanded light-weight concrete shown in Fig. 2 repre-sent concretes with a unit weight in therange of 105 to 115 lb per cu ft. Notethat at 800 F, most concretes retainmost of their original strength and at

1200 F carbonate and lightweight con-cretes have nearly all of their originalstrengths. Siliceous aggregate concreteretains more than one-half its initialstrength at 1200 F.

RESULTS OF 18 STANDARD FIRE TESTSCONDUCTED IN THE UNITED STATES

Data have been published from a to-tal of 18 fire tests conducted in theUnited States on post-tensioned pre-stressed concrete slabs and beams. Twofire tests of slabs were conducted bythe Fire Prevention Research Institutein Gardena, California.0.10 Underwrit-ers' Laboratories, Inc., Northbrook, Illi-nois, conducted three tests of post-ten-sioned structures, one was a slab" and

42

two were inverted tee beams. 12 ThePortland Cement Association fire testedseven post-tensioned beams, all ofwhich were modified tee beams span-ning 40 ft. Six tests conducted by theNational Bureau of Standards in 1953are of historical interest, 13 and werepart of a series sponsored by the BritishJoint Fire Research Organization andthe Building Research Station.

FPRI testsReferences 9 and 10 give pertinent

data about the Fire Prevention Re-search Institute tests. Both tests in-volved normal weight concrete slabs, 6in. thick, made with siliceous aggre-gates and post-tensioned unbonded ten-dons. One of the specimens was an in-tegral beam-and-slab assembly; the oth-er was a flat plate floor. The beamswere prestressed longitudinally and theslab was prestressed transversely withmoderate longitudinal prestress. Theminimum clear cover was 1 1/a in. forthe slab tendons and 2 in. for the beamtendons. In the other specimen, the b-in, slab was prestressed with post-ten-sioned tendons in two directions. Theminimum cover at midspan was 11/2 in.

Both assemblies were mounted infixed restraining frames during the firetests. Structural end points were notreached during the tests which lastedmore than 4 and 3 hr, respectively. The

end point for the first test occurred at3 hr 51 min when the unexposed sur-face temperature rose an average of250 F. Although the second test wasstopped before an end point wasreached, the heat transmission endpoint would have been reached atabout 3 hr 15 min.

UL testsReference 11 gives pertinent details

of the fire test of a lightweight concretepost-tensioned flat plate floor conductedby the Underwriters' Laboratories. Dur-ation of the test was 3 hr 45 min withno end point occurring. The specimenhad been dried for 7 months at hightemperatures prior to the test and themoisture content of the concrete waslow. Based on the correction procedurefor nonstandard moisture content (Ap-pendix A5 of ASTM E119-71), the heattransmission end point would have oc-curred at about 4 hr 40 min. No spal-ling of the specimen occurred.

Reference 12 refers to fire tests ofinverted tee beams prestressed withpost-tensioned tendons. In one speci-men the tendon was bonded while inthe other the tendon was unbonded.The superimposed load on the un-bonded specimen was substantially low-er than the load on the bonded speci-men. Both tests were terminated at 4hr 15 min even though no end point

Table 1. Data from PCA Tests (Reference 7)

BeamNo.

Type ofReinforcement

Bonded orUnbon`ded

Type ofConcrete

SuperimposedLoad,

lb. per ft.

FireEndurancehr.:min.

80 Bars Unbonded Normal weight 1040 5:0282 Bars Bonded Normal weight 1535 4:2983 Bars Bonded Lightweight 1680 5:0176 Wires Unbonded Normal weight 1135 3:0478 Wires Bonded Normal weight 1750 3:2079 Wires Bonded Lightweight 1740 4:3389 Wires Bonded Normal weight 1760 3:18


LIGHTWEIGHT CONCRETE

140C

R

./ 5LL 1 10(0

-.

aEH

1

80

pJҟHI.^^•ҟ •mac•

JiOҟt

. mac•ҟ̂p

co,^ro

50C30ҟ60ҟ90ҟ120ҟ180ҟ240

Fire Test Time, Minutes

Fig. 3. Temperatures within concrete slabs during fire tests (expanded shale ag-gregates) showing strand temperatures in hollow-core slabs

was reached. At the ends of the teststhe midspan deflections were about 1in. for the 17 ft 5 in. spans. A compan-ion pretensioned specimen was also firetested in the same series. The behavior

of the pretensioned specimen was sim-

ilar to that of the post-tensioned com-

panions.

PCA tests

As part of a broad series of fire tests,the Portland Cement Association firetested seven 40-ft beams in which thereinforcement was post-tensioned.?Two types of reinforcement were used,high strength alloy steel bars and cold-drawn wires with button heads. Beams

44

were essentially rectangular, 14 in.wide, 25 in. deep, with 6 x 4-in. flanges.Tendon cover at midspan was 2 1/2 in.Table 1 gives some pertinent dataabout the specimens and tests.

Beams were simply supported onrocker-roller supports to minimize re-straint to thermal expansion. Includedin the series of tests were companionspecimens reinforced with Grade 40and Grade 60 bars, and three speci-mens with pretensioned seven-wirestrand. Among the conclusions reached

from the Portland Cement Associationseries of tests were:

1. Prestressed beams of lightweightconcrete had longer fire endurancesthan their normal weight companions,and

2. Beams with unbonded post-ten-sioned reinforcement had about thesame fire endurances as their counter-parts with bonded reinforcement.

NBS testsAs noted above, the six tests con-

1400

LL

1 i00

EF-

800

500'

____U-..."lu-I

^SFT-11-1/2 in.Cover

SFT-1>3 in.i Cover

30ҟ60ҟ90ҟ120ҟ180ҟ240Fire Test Time, Minutes

Fig. 4. Temperatures within concrete slabs during fire tests (siliceous aggregates)showing slab tendon temperatures in FPRI-SFT-1

PCI Journal/March-April 1973ҟ 45

SILICEOUS AGGREGATE CONCRETE

Iҟ 2 -S.•ҟCi

a

7.i`ҟGore

Q tiro• 5,ti^o eiGo

Jeioo^

30ҟ

60ҟ90ҟ120ҟ

180ҟ240


Fig. 5. Temperatures within concrete slabs during fire tests (siliceous aggregate)showing tendon temperatures in FPRI-SFT-2. (Adjusted for furnace temperature

lag.)

1400

U-

1100

EH

800

500

ducted at the National Bureau of Stan-dards in 1953 involved beams manu-factured in England and fire tested inaccordance with the 1932 edition ofBritish Standard 476. The test proce-dure of BS476-32 is similar to that ofASTM E119 except for one major dif-ference. The loading requirement ofBS476-32 called for a superimposed

load of one and one-half times the de-sign live load rather than one times thelive load. Recent editions of BS476have revised that requirement to onetimes the live load. This stipulation isthe same as ASTM E119. Thus, the fireendurances of the six National Bureauof Standards tests were probably sig-nificantly shorter than might be ex-

46

petted if the normal loading had beenapplied. The beams were rectangularwith or without a composite slab. Thesteel consisted of wires, 0.1 or 0.2-in.diameter, post-tensioned and grouted.The span was either 10 ft or 16 ft andbeams were simply supported. Twobeams were coated with 1 in. thick ver-miculite concrete. Fire endurances (notadjusted for loading) ranged betweenabout 1 112 and 6 hr.

ANALYSIS OF TEST DATA

In 1971, Undewriters' Laboratoriesdeveloped criteria for fire resistanceclassifications for precast prestressedconcrete units. The classifications werebased on results of standard fire tests ofhollow-core slabs, various stemmedunits, and inverted tee beams. The newcriteria are based on the tentative re-visions of ASTM E119-71, i.e., for sim-ply supported unrestrained memberswith cold-drawn prestressing steel thefire endurance is the time required forthe steel to reach 800 F during a stan-dard fire test. For restrained beamsspaced more than 4 ft on centers, thefire endurance is twice the time re-quired for the steel to reach that tem-perature. For other restrained units, thesteel temperature is disregarded andthe structural end point governs.

In the Underwriters' Laboratoriesstudies that led to the classification cri-teria included in the January 1972"Fire Resistance Index," 14 measuredsteel temperatures were compared witheach other and with published data ontemperatures within concrete membersduring fire tests. For example, steeltemperatures during full scale fire testsof 14 hollow-core slabs were comparedwith temperatures measured withinplain concrete slabs. Data from four ofthe tests of lightweight concrete slabsare shown in Fig. 3. The basic slabtemperature data on which the chartsare based were developed in the test

program described in PCA ResearchDepartment Bulletin 223.15 Charts sim-ilar to Fig. 3 for carbonate and siliceousaggregate concretes were also preparedand analyzed. Note that the measuredsteel temperature curves are roughlyparallel to the slab temperature curves.In each case the steel temperatures aresomewhat lower than those estimatedfrom the slab data for the same dis-tance from the exposed surface. Thus,the cover requirements based on theslab concrete temperatures are slightlyconservative. Based on Fig. 3, the coverrequirements for unrestrained light-weight prestressed concrete slabs areapproximately 1 in. for 1 hr, 1 5/s in. for2 hr, and 2 in. for 3 hr.

A similar, though more complex, pro-cedure was used in analyzing the datafor stemmed units and inverted teebeams.

The procedures used by the Under-writers' Laboratories in developingtheir classification criteria are essential-ly those used below for analyzing slabsand beams.

Analysis of slab dataPertinent steel temperature data

from the three tests of slabs are shownin Figs. 4, 5, and 6. Fig. 4 shows thetemperatures of the tendons in thebeam-and-slab assembly (FPRI SFT-1).Note that the temperature of the slabtendons with 1 1/z-in. cover reached anaverage of 800 F after 3 hr of fire expo-sure. This temperature is far lower thanwould be expected for 1 1/z-in, coverbased on the concrete slab temperaturedata. In fact, the beam steel tempera-tures were lower than the 2-in, line onthe plot. The low recorded tempera-tures might be due to either or both ofthe following items. First, the recordedtemperatures reflect the average tem-perature of the tendon rather than themaximum temperature that would oc-cur at the bottom of the tendon. Sec-ond, the tendons were greased and


1400

U-0

j 1100f0

aEG)H

800

500

P'AA •A1A'L4

ivAr1ruVIIIII I1I

30ҟ

60ҟ90ҟ120ҟ

180ҟ240


Fig. 6. Temperatures within concrete slabs during fire tests (expanded shale ag-gregates) showing tendon temperatures in UL 5084-3

wrapped and the lubricant and wrap-ping materials might have kept thetendons cooler during fire exposure.

Fig. 5 shows the tendon tempera-tures recorded during the test of thenormal weight concrete flat plate floor(FPRI SFT-2). The curvilinear shapeof the temperature curves may be dueto the furnace atmosphere tempera-

tures which were somewhat low duringthe first 21/z hr and high thereafter.Again, the recorded steel temperatureswere lower than would be expectedfrom the data on which the curves aresuperimposed.

Fig. 6 shows the tendon tempera-tures recorded during the test of thelightweight concrete flat plate floor

48

Table 2. Summary of Data on Slab Tendons

"B"Corresponding

"A" Test Time Distance fromCover, to Reach Exposed "B"—"A"

Test in. 800F,hr:min Surface, in. in.FPRI-1 1-1/2 3:12 2-5/8 1-1/8FPRI-1 3 5:07* 3-5/8 5/8FPRI-2 1 1:44 1-3/4 3/4FPRI-2 2 3:30* 2-5/8 5/8UL-R5084-5 1 1:31 1-3/8 3/8UL-R5084-5 1-3/4 2:47 2-1/4 1/2* Extrapolated

(UL R5084-3) superimposed on a plotof temperatures within lightweight con-crete slabs during fire tests. Again thetemperatures are lower, but to a lesserextent than those in Figs. 4 and 5, pos-sibly because the specimen was kiln-dried prior to the test.

In an attempt to determine requiredcover thickness for unrestrained slabswith post-tensioned tendons, an analy-sis can be made of test times at whichthe tendons reached 800 F. These val-ues can be compared with correspond-ing test times at which concrete at vari-ous levels reaches 800 F during firetests. Table 2 provides a basis for com-parison.

The values of `B" in Table 2 are thedistances from the exposed surface inplain concrete slabs at which the tem-perature is 800 F at the test time indi-

cated in the third column. The valuesof `B"-"A" in the last column indicatethe magnitude of reduction of coverpossible. Note that those values rangefrom 3/s to 11/s in. Thus a reduction ofat least 3/s in. is warranted. Resultingcover thicknesses for simply supportedunrestrained slabs with post-tensionedreinforcement, based on temperatureswithin slabs with a 3/s-in, reduction,are given in Table 3.

Cover requirements shown in Table3 apply to tendons 1/z-in, or larger insize. The values for 1 hr for carbonateand lightweight aggregate concretes aregoverned by minimum cover require-ments for slabs (see the provisions inACI 318-71).

Analysis of beam dataFig. 7 shows tendon temperatures at

Table 3. Cover Requirements for Unrestrained Slabs with Post-Tensioned Reinforcement

Cover Thickness, in., forFire Endurance of

Aggregate Type 1 hr 1-1/2 hr 2 hr 3 hrCarbonate 3/4 1-1/16 1-3/8 1-7/8Siliceous 3/4 1-1/4 1-1/2 2-1/8Lightweight 3/4 1 1-1/4 1-5/8


1400

LL 1 100ci4-

1.a)a-EH

800

500

CARBONATE AGGREGATE CONCRETE

77-p77__

QO

/ I,. ___

midspan of the three inverted teebeams fire-tested at Underwriters' Lab-oratories. It is interesting to note thatthe temperatures of the pretensionedstrand (UL R4123-12) correspond tothe temperatures that might be antici-pated for strand centered about 2 1/4 in.above the bottom of a slab even though

the cover was only 13/4 in. and thestrands were in a beam rather than in aslab. Temperatures of the post-ten-sioned tendons (UL R4123-12A) werelower yet, possibly because the bondedand unbonded tendons were centered23/4 and 2% in., respectively, above thebottoms of the beams. Nevertheless, the

3U bU

Fire Test Time, Minutesy^ 120 180 240

1. Pretensioned 1-3/4 in. cover.2. Post-tensioned unbonded 1-7/8 in. cover.3. Post-tensioned bonded 1-15/16 in. cover.

Fig. 7. Temperatures within concrete slabs during fire tests (carbonate aggregate)showing tendon temperatures in inverted T beams (UL 4123-12-12A)

50

CARBONATE AGGREGATE CONCRETE

7

__^^

aSJ

e77

4ti

140C

weLL0

mI-C,

Ea)H

80C

,4

30ҟ60ҟ90Fire Test Time, Minutes

1. Pretensioned strand.2. Post-tensioned wires, bonded.3. Reinforcing bars.

120ҟ180ҟ240

4. Post-tensioned bars, bonded.5. Post-tensioned bars, unbonded.6. Post-tensioned wires, unbonded.

Fig. 8. Temperatures within concrete slabs during fire tests (carbonate aggregates)showing temperatures of corner bars, wires, or strand during PCA tests of 40-ft t

beams

temperatures correspond to those ofabout 3Y4 and 3 in. above the bottomof a slab. The low temperatures forthe bonded tendon might result fromthe insulation afforded by the high wa-ter content of the grout within the duct.

Figs. 8 and 9 show temperatures of

the corner bars or strands of the 40-ftbeams tested at the Portland CementAssociation. The temperatures shownrepresent the maximum bar or strandtemperatures because the corner barsor strands, i.e., those with 2% in. sideand bottom cover, were the hottest in


LIGHTWEIGHT CONCRETE

/7 I77H

1400

LL 11005,

CDaE9

800

500

1

2

3

4

30 60Fire Test Time, Minutes

90 120 180 2401. Pretensioned strand.ҟ3. Reinforcing bars.2. Post -tensioned barsҟ4. Post-tensioned wires.

Fig. 9. Temperatures within concrete slabs during fire tests (expanded shale ag-gregates) showing temperatures of corner bars, or strand during PCA tests of 40-

ft beams

each of the tests. Note that in Fig. 8(normal weight concrete) the time-tem-perature relations for the post-tensionedbars, bonded or unbonded, the reinforc-ing bars, and the unbonded post-ten-sioned wires are grouped closely to-gether. Temperatures of the corner pre-tensioned strands were higher andthose of the bonded post-tensioned

bonded wires were lower. The same ap-proximate relationships are also true forthe lightweight concrete specimens(see Fig. 9).

From Figs. 8 and 9 it can be seenthat the corner bar temperatures of thepost-tensioned units are essentially thesame as (or lower than) slab tempera-tures at a distance of about 2i/z in. from

52

FixedҟHeado is iii.ҟ-'

4 ^2"x 12 ҟConcreteql

D D i^'000.CylinderҟwithLongitudinal o;I' n'Center Hole

i

o

o a -_.r i

Bottom Anchor xo ҟoxAssembly

ElectricFurnace

Steel BearingPlate

FurnaceThermocouple

p Anchor Assembly

^ҟn i

Movable Head

IIIII

Strand___;___l

IIWire orBar

Bottom Anchor AssemblyShowing Locationsof Thermocouples

0

11 Base of Testing Machine

Fig. 10. Arrangement for high temperature tests of tendon-anchor assemblies

the exposed surface. Since that was thecover of the corner bars, the cover re-quirements for slabs, as determinedfrom the Portland Cement Associationconcrete slab data, should be adequatefor beams with dimensions roughlycomparable to those tested. On thisbasis, for beams with post-tensioned re-inforcement wider than about 12 in.

the cover requirements for unrestrainedclassifications would be those shown inTable 4.

The above tabulation assumes thatthe minimum cover would be 1% in.for all beams, and that tendons are %-in. or larger in size. For narrowerbeams the cover would have to besomewhat greater in some cases. For


Table 4. Cover Requirements for Unrestrained Beams at Least12-in. Wide and Prestressed with Post-Tensioned Reinforcement

Steel Concrete

For Beams at Least 12-in. Wide,Cover Thickness, in., for

Fire Endurance

Type Type* 1 hr 2 hr 3hr 4 hr

Cold-Drawn NW 1-1/2 2 2-1/2 3Cold-Drawn LW 1-1/2 1-3/4 2 2-1/2

H.S.A. Bars NW 1-1/2 1-1/2 1-1/2 2H.S.A. Bars LW 1-1/2 1-1/2 1-1/2 2

*NW = normal weight; LW = lightweight

beams 8 in. wide, comparable coverrequirements could be those shown inTable 5. For beams with widths be-tween 8 and 12 in., cover requirementscan be obtained by direct interpolation.For example, for a 10-in, wide beam

of lightweight concrete with cold-

drawn steel, the cover for 3 hr would

have to be 2% in.

The values for 8-in, wide beams were

derived from the relationships of cover,

beam width, and temperature based on

results of tests at the Portland Cement

Association and the Underwriters' Lab-oratories, some of which have not yet

been published. For cold-drawn steel, atemperature limit of 800 F was used.For high strength alloy steel bars, atemperature limit of 1000 F was used,making the results somewhat conserva-tive.

Protective coatings

A 1972 report' s gives analyses of firetests of slabs, beams, and joists andconcludes with recommended thick-nesses of sprayed insulation for pre-stressed units. The thicknesses ofsprayed mineral fiber, vermiculite TypeMK, or intumescent mastic are givenfor slabs and beams of various widths

Table 5. Cover Requirements for Unrestrained 8-in. WideBeams Prestressed with Post-Tensioned Reinforcement

Steel Concrete

For Beams 8-in. Wide,Cover Thickness, in., for

Fire Endurance

Type Type 1 hr 1-1/2 hr 2 hr 3hr

Cold-Drawn NW 1-3/4 2 2-1/2 4-1/2*Cold-Drawn LW 1-1/2 1-3/4 2 3-1/4

H.S.A. Bars NW 1-1/2 1-1/2 1-1/2 2-1/2H.S.A. Bars LW 1-1/2 1-1/2 1-1/2 2-1/4

*Not practical but shown for interpolation purposes.

54

o 8Cc

0

°6Ca0

vC)

4C

rnas

U2C

)

Relationshipstrand

for(P CAҟR

old-drawnsearch

Depart ient Bulletin 134)

•o

Series No.Series No.

I12

• 200 400 600 800 1000 1200

Temperature , F

Fig. 11. Relation between temperature and tensile strength of cold-drawn strand(from Abrams and Cruz) together with test results of tendon-anchor assemblies

and concrete cover thicknesses.Even though none of the tests ana-

lyzed in that report were of memberswith post-tensioned reinforcement, thedata should be directly applicable toany member with cold-drawn prestress-ing steel. It should be noted that twoof the National Bureau of Standardstests were of specimens coated withvermiculite concrete. In each case thefire endurance of the coated specimenwas more than double that of its un-coated counterpart. The data are notdirectly applicable for beams or slabswith high strength alloy steel bars, butwould be conservative if applied direct-ly.

ANALYSES OF RESULTS OF TESTS OFTENDON-ANCHOR ASSEMBLIES AT

HIGH TEMPERATURES

Several tests have been performed todetermine if anchors commonly used inNorth America for post-tensioning con-tinue to function at temperatures thatoccur during fires. Reports of these testsare not readily available, so much of thepertinent data is included here.

Tensile tests of tendon-anchor as-semblies

Three series of tests were conductedat the Portland Cement Association


e

a80

.rn

0

060o

800 F Valuefor Strand

Hot -rolled215K Stee

vo)C

° 404)

XL

C,Ca> 20

,7M

2 3 4 5 6 7 9 10 II 12

Series Number

Fig. 12. Results of tests made at 800 F compared with value obtained for cold-drawn strand at same temperature

Laboratory. In two of the series, testswere performed at various tempera-tures between 600 F and 1000 F, andat 70 F. In the other series, 12 types oftendon-anchor assemblies were testedat 800 F and at 70 F. Results of thesetests were compared with results of ten-sile tests of tendons in which the an-chors were not heated.

Fig. 10 shows the test setup. Notethat the bottom anchor assembly wascentered within the electric furnace.The concrete cylinder was used to pro-vide uniform bearing for the anchor.In fact, several of the types of bearingplates must be cast into the concrete.The cylinder also served to locate theanchor within the furnace. Some of thecylinders were jacketed with steel

pipes. In such cases, the bearing plateswere machined to a maximum diameterof 3'7/s in. to ensure that the plates didnot bear on the steel jacket. A load ofabout 1000 lb was applied to the speci-men at the start of the test and main-tained during the heating period. Aperiod of 2 to 31/2 hr was required toheat the specimen to the desired testtemperature. When thermocouples 1, 2,and 3 (Fig. 10), located on the anchorhousing and on the tendon, reachedthe test temperature with a variation of15 F or less, the tensile load was in-creased at a rate of about 8000 lb permin until failure occurred.

Results of the two series of tests con-ducted at 70 F, 700 F, 800 F, 900 F,and 1000 F, are shown in Fig. 11.

56

36°

Anchor

1fBearing Plate

8 uҟ8uCD

r -^ O^ҟ6

0)

X=24ga. Chromel-alumel thermocouple

PLA N

r--- r

V..

ҟ

L--'UL_J m 0

Q

ELEVATIONDimensions . in inches

Dimension A B C D E F Strand SizeSpecimen A 3 1 I/2 2 1/4 17/8 '8 1 1/8 0.6 in.

Specimen B 1 1/2 2 /2 3 3/4 I I I/2 in.

Fig. 13. Details of strand-anchor assemblies embedded in slab


Series No. 1 consisted of 0.6-in, diam-eter strand and a rather massive anchor-bearing plate assembly. Both the strandsize and the anchor are among the larg-est in use in North America. They wereselected for this series of tests becausethe investigators felt that large tendonanchor assemblies might be more vul-nerable to heat than smaller ones. Series12 consisted of '/z-in. strand and smallanchor-bearing plate assemblies. Dupli-cate tests were conducted at 70 F,700 F, and 900 F and triplicate tests at800 F. Relatively small differences inthe breaking loads occurred for dupli-cate and triplicate tests at a specifictemperature. Fig. 11 shows the resultsof these tests compared with the ten-sile strength-temperature relation ofcold-drawn steel strand determined byAbrams and Cruzs from tests in whichthe anchors were not heated. It can benoted that the test results compare fav-orably with those for strand, differingby three percentage points or less inall cases. Thus it appears reasonable toassume that the temperature-strengthrelationships of tendon-anchor assem-blies are about the same as those for thetendon alone.

In the third series of tests, 12 typesof tendon-anchor assemblies weretested at temperatures of 800 F and70 F. Eight assemblies made use of '/z-in. diameter seven-wire strand, twoused 0.6 in. strand, one used 1/4-in, di-ameter button-headed wire, and one as%s-in. diameter deformed bar. Partici-pants who supplied tendons and an-chors were:

Atlas Prestressing Corp.Dyckerhoff and Widmann, Inc.Freyssinet Company, Inc.Inland-Ryerson Construction Prod-

ucts Co.Prescon Corp.Stressteel Corp.Stresstek Corp.VSL Corp.Western Concrete Structures, Inc.

Fig. 12 shows the results of the 12tests (at 800 F) compared with thecomparable value reported for Abramsand Cruzb for strand at 800 F. Thevalue shows for Series No. 9 is not di-rectly comparable to the others be-cause the tendon was a 5/s-in. diame-ter hot-rolled bar having a tensilestrength of 215 ksi. The other tendonswere 270 ksi cold-drawn strand or 240ksi wire. Note that the average varia-tion from that for strand was less thantwo percentage points and the maxi-mum variation was about four percent-age points.

As noted above, the results of SeriesNo. 9 are not directly comparable tothe others since the temperature-strength relation for 215 ksi hot-rolledsteel is probably different than that ofcold-drawn strand. From Fig. 1it can be seen that high strengthalloy steel bars (145 ksi) haveabout 80 percent of their 70 F strengthat 800 F. Even though the value of 68percent for Series No. 9 is lower thanthat for 145 ksi bars, it is considerablyhigher than the value for 270 ksi cold-drawn steel, and thus seems to be rea-sonable.

From these tests it appears that theanchor does not influence the tempera-ture-strength relation significantly. Itshould be noted that the tendon-anchorassemblies represented a wide spec-trum of those in use in North America.It does not appear that the mass of theanchor has a significant influence on thebehavior at high temperatures.

Fire tests to study the effects of cov-er on tendons and anchors

In most post-tensioned structures, ex-posure to fire is likely to be less severeat the anchor than at other locations inthe beam or slab. However, in somecases, the anchors are situated in vul-nerable locations. Because the anchorsrepresent concentrations of metal, it islikely that the strand temperature at the

58

i

60ҟ90ҟ120ҟ180 240Fire Test Time, minutes

(a) Specimen A

IIOC

90C

70C

50030

___.1f

___

60ҟ90ҟ120ҟ180 240Fire Test Time, minutes

(b) Specimen B

u_0

a)4-va,aEa,I-

1100

90C

70C

50030

U-0

a)4-va)nE

I-

Fig. 14. Temperatures within concrete during fire tests compared with strandtemperatures at and away from anchors


Table 6. Suggested Concrete Slab Thickness Requirements forVarious Fire Endurances

Aggregate

Slab Thickness, in., forFire Endurance Indicated

Type 1 hr 1-1/2hr 2hr 3hr 4hr

Carbonate 3-1/4 4-1/8 4-5/8 5-3/4 6-5/8Siliceous 3-1/2 4-1/4 5 6-1/4 7Lightweight 2-5/8 3-1/4 3-3/4 4-5/8 5-1/4

anchor can be different from that awayfrom the anchor.

To study the magnitude of the tem-perature difference, fire tests were per-formed on two slabs in which three ten-don-anchor assemblies were embedded.Slab specimens were 3 x 3 ft in planand 6 in. thick. Strands in the tendon-anchor assemblies were horizontalthroughout as shown in Fig. 13.

Fig. 14 shows the results of the tests.Temperatures of the tendons at the an-chors were higher than away from theanchor for three of the tendons. Thedifference was insignificant for two ten-dons, and for one tendon, the tempera-ture of the tendon at the anchor wascooler than away from the anchor. Dis-regarding the tendon that was cooler atthe anchor, the tendons were up toabout 70 F warmer at the anchor thanaway from the anchor. To compensatefor the higher temperature at the an-chor, the cover to the tendon can beincreased by about 1/4 in.

Fire tests to study the effects of dif-ferent sheathing materials for un-bonding

In North American practice, un-bonded post-tensioned tendons are gen-erally greased and sheathed with eitherkraft paper or plastic. Paper sheathingis generally spirally wrapped whileplastic sheathing is usually in the form

of a continuous tube. A fire test wasconducted at the Portland Cement As-sociation Laboratory to determine ifthe sheathing material affects the ten-don temperature during exposure tofire.

The fire test specimen consisted of aconcrete slab in which some strandswere sheathed with paper and somewith plastic. The 4-in. thick slab speci-men, which was 3 x 3 ft in plan, con-tained two layers of sheathed strand.Four strands in the east-west directionhad 1-in, cover and four in the north-south direction had 2-in. cover. At eachlevel, the first and third strands weresheathed with paper and the other twowith plastic. The 4-in. thick slab speci-positioned on each strand, one at mid-span and the other 12 in. away. Thethermocouples were located on thestrand within the sheaths.

The slab specimen was exposed to astandard (ASTM E119) fire exposurefor 21/ hr. During the test, thermocou-ple readings were monitored and com-pared. With 1-in, cover, strands withpaper sheathing were about 15 F to30 F cooler than those with plasticsheaths. With 2-in, cover, strands withpaper sheaths were 10 F cooler to 15 Fwarmer than those with plastic sheaths.These differences are not consideredto be significant because of the usualvariations in temperature readings of

60

embedded metal in concrete. Thus itappears that the type of sheathing ma-terial (paper or plastic) has only a mi-nor influence on the strand temperatureand does not affect the concrete coverrequirements significantly.

RECOMMENDATIONS FOR MINIMUMDIMENSIONS FOR VARIOUS FIRE RE-

SISTIVE CLASSIFICATIONS

SlabsFor heat transmission, i.e., tempera-

ture rise of 250 F of the unexposed sur-face, the thickness requirements forconcrete slabs should be the samewhether the concrete is plain, rein-forced, or prestressed. Table 6 givesslab thicknesses suggested in PCA Re-search Department Bulletin 223.15

Cover thicknesses for post-tensionedtendons in unrestrained slabs are de-termined by the elapsed time during afire test until the tendons reach a criti-cal temperature. For cold-drawn pre-stressing steel that temperature is 800F. For restrained slabs there are notemperature limitations. Fire tests ofrestrained slabs indicate that slabs withpost-tensioned reinforcement behave

about the same as reinforced concreteslabs of the same dimensions. Accord-ingly, the cover for post-tensioned ten-dons in slabs should be the same as thecover for reinforcing steel in slabs. Ap-plying these criteria to slabs with post-tensioned tendons made of cold-drawnsteel, cover thicknesses are suggestedin Table 7.

BeamsMinimum dimensions for beams with

post-tensioned reinforcement for vari-ous fire endurances are functions of thetypes of steel and concrete, beamwidth, and cover. For very wide beams,the cover requirements should be aboutthe same as those for slabs.

For restrained beams spaced morethan 4 ft on centers, the fire enduranceis twice the elapsed time during a firetest at which the steel reaches the criti-cal temperature. The suggested coverthicknesses in Table 8 are based onthese criteria.

For beams or joists less than 8 in.wide, the Underwriters' Laboratoriesrequirements for pretensioned stemmedmembers can be used for members withpost-tensioned cold-drawn steel. Beamsor joists with post-tensioned highstrength alloy steel bars and narrower

Table 7. Suggested Concrete Cover Thicknesses for Slabs Prestressedwith Post-Tensioned Reinforcement

Restrained or Aggregate

Cover Thickness, in., forFire Endurance of

Unrestrained Type 1 hr 1-1/2 hr 2 hr 3 hr 4 hr

Unrestrained Carbonate 3/4 1-1/16 1-3/8 1-7/8 — — -Unrestrained Siliceous 3/4 1-1/4 1-1/2 2-1/8 - - -Unrestrained Lightweight 3/4 1 1-1/4 1-5/8 — — -Restrained Carbonate 3/4 3/4 3/4 1 1-1/4Restrained Siliceous 3/4 3/4 3/4 1 1-1/4Restrained Lightweight 3/4 3/4 3/4 3/4 1


Table 8. Suggested Cover Thickness for Beams Prestressed withPost-Tensioned Reinforcement

Cover Thickness, in., for

Restrained or Steel Concrete Beam Fire Endurance of

Unrestrained Type Type* Width,'* in.1hr 1-1/2 2hr 3hr 4hr

Unrestrained Cold-drawn NW 8 1-3/4 2 2-1/2 4-1/2*** — — —Unrestrained Cold-drawn LW 8 1-1/2 1-3/4 2 3-1/4 - - -Unrestrained H.S.A. Bars NW 8 1-1/2 1-1/2 1-1/2 2-1/2 - - -Unrestrained H.S.A. Bars LW 8 1-1/2 1-1/2 1-1/2 2-1/4 - - -

Restrained Cold-drawn NW 8 1-1/2, 1-1/2 1-3/4 2 2-1/2Restrained Cold-drawn LW 8 1-1/2 1-1/2 1-1/2 1-3/4 2Restrained H.S.A. Bars NW 8 1-1/2 1-1/2 1-1/2 1-1/2 1-1/2Restrained H.S.A. Bars LW 8 1-1/2 1-1/2 1-1/2 1-1/2 1-1/2

Unrestrained Cold-drawn NW >12 1-1/2 1-/34 2 2-1/2 3Unrestrained Cold-drawn LW >12 1-1/2 1-1/2 1-3/4 2 2-1/2Unrestrained H.S.A. Bars NW >12 1-1/2 1-1/2 1-1/2 1-1/2 2Unrestrained H.S.A. Bars LW >12 1-1/2 1-1/2 1-1/2 1-1/2 2Restrained Cold-drawn NW >12 1-1/2 1-1/2 1-1/2 1-3/4 2Restrained Cold-drawn LW >12 1-1/2 1-1/2 1-1/2 1-1/2 1-3/4Restrained H.S.A. Bars NW >12 1-1/2 1-1/2 1-1/2 1-1/2 1-1/2Restrained H.S.A. Bars LW >12 1-1/2 1-1/2 1-1/2 1-1/2 1-1/2

NW = normal weight; LW = lightweight" For beams with widths between 8 and 12 in-, cover thickness can be determined by interpolation"`Not practical for 8-in- wide beam but shown for purposes of interpolation

than 8 in. should have the same coveras joists of the same size and fire en-durance.

Anchor protection

The cover to the prestressing steel atthe anchor should be at least 1/4 in.greater than that required away fromthe anchor. Minimum cover to the steelbearing plate should be at least 1 in. inbeams and 3/4 in, in slabs.

REFERENCES

1. ASTM Designation: E119-71,"Standard Methods of Fire Tests ofBuilding Construction and Materi-als," Part 14, ASTM Book of Stan-dards, American Society for Test-ing and Materials.

2. Gustaferro, A. H., and Selvaggio,S. L., "Fire Endurance of SimplySupported Prestressed ConcreteSlabs," PCI JOURNAL, Vol, 12,No. 1, February, 1967, pp. 37-52.PCA Research Department Bulle-tin 212.

3. Gustaferro, A. H., "TemperatureCriteria at Failure," Fire Test Per-formance, ASTM STP 464, Ameri-can Society for Testing and Ma-terials, 1970, pp. 68-84.

4. Selvaggio, S. L., and Carlson,C. C., "Effect of Restraint on FireResistance of Prestressed Concrete,"Fire Test Methods, ASTM STP No.344, American Society for Testingand Materials, 1962. PCA ResearchDepartment Bulletin 164.

5. Brockenbrough, R. L., and Johns-ton, B. G., Steel Design Manual,U.S. Steel Corp., Pittsburgh, Penn-sylvania, 1968, 246 pp.

6. Abrams, M. S., and Cruz, C. R.,"The Behavior at High Tempera-ture of Steel Strand for PrestressedConcrete," Journal of the PCA Re-search and Development Labora-tories, Vol. 3, No. 3, September,1961, pp. 8-19; PCA Research De-partment Bulletin 134.

7. Gustaferro, A. H., et al., "Fire Re-sistance of Prestressed ConcreteBeams. Study C: Structural Behav-

62

for During Fire Tests," PCA Re-search and Development Bulletin(RD 009B), Portland Cement As-sociation, 1971.Abrams, M. S., "CompressiveStrength of Concrete at Tempera-tures to 1600 F," Symposium onEffect of Temperature on Con-crete, American Concrete InstitutePublication SP-25, Detroit, Michi-gan, 1971.

9. Troxell, G. E., "Fire Test of Pre-stressed Concrete Floor Panel No.1, "Fire Prevention Research In-stitute, SFT-1, Gardena, California.

1.0 Troxell, G. E., "Fire Test of 6-in.Hardrock Concrete, Post-Tension-ed, Prestressed Flat Slab", Fire Pre-vention Research Institute, SFT-2,Gardena, California.

11. "Report on Unbonded Post-Ten-sioned Prestressed, ReinforcedConcrete Flat Plate Floor with Ex-panded Shale Aggregate," Under-writers' Laboratories, Inc., R5084-3. Reprinted in PCI JOURNAL,Vol. 13, No. 2, April 1968, pp. 45-56.

12. "Report on Prestressed Preten-

sioned Concrete Inverted TeeBeams and Report on PrestressedConcrete Inverted Tee Beams Post-Tensioned," Underwriters' Labora-tories, Inc., R4123-12-12A. PCIPublication R-119-66.

13. Ashton, L. A., and Bate, S. C. C.,"The Fire Resistance of PrestressedConcrete Beams," ACI Journal,Vol. 32, May 1961, pp. 1417-1440.

14. "Fire Resistance Index," Un-derwriters' Laboratories, Inc.,Northbrook, Illinois, January, 1972.

15. Abrams, M. S., and Gustaferro,A. H., "Fire Endurance of Con-crete Slabs as Influenced by Thick-ness, Aggregate Type, and Mois-ture," Journal of the PCA Researchand Development Laboratories, V.10, No. 2, May 1968, pp. 9-24,PCA Research Department Bulle-tin 223,

16. Abrams, M. S., and Gustaferro,A. H., "Fire Endurance of Pre-stressed Concrete Units Coatedwith Spray-Applied Insulation,"PCI JOURNAL, Vol. 17, No. 1,January-February, 1972, pp.82-103.

Discussion of this paper is invited. Please forward your discussion toPCI Headquarters by August 1, 1973, to permit publication in theSeptember-October 1973 issue of the PCI JOURNAL.


fire resistance of post-tensioned structures journal/1973/march-… · other characteristics of...

Documents