New facility to determine fire resistance of columns - National ...

+SerTH1 I CouncilN21dno. 914cop. 2Nat~onal Research Conseil nationalCanada de recherche CanadaNEW FACILITY TO DETERMINE FIRE RESISTANCEOF COLUMNSby T. T. LieANALYZEDReprinted fromCanadian Journal of Civil EngineeringVol. 7, No. 3, September 1980pp. 551-558NRC - ClSTlBLDG. RES.LIBR k2YDBR Paper No. 914Division of Building ResearchPrice $1.00OTTAWA NRCC 18423

This publication is being distributed by the Division of BuildingResearch of the National Research Council of Canada. Itshould not be reproduced in whole or in part without permissionof the original publisher. The Division would be glad to beof assistance in obtaining such permission.Publications of the Division may be obtained by mailing theappropriate remittance (a Bank, Express, or Post Office MoneyOrder, or a cheque, made payable to the Receiver General ofCanada, credit NRC) to the National Research Council ofCanada, Ottawa. KIA 0R6. Stamps are not acceptable.A list of all publications of the Division is available and may beobtained from the Publications Section, Division of BuildingResearch, National Research Council of Canada, Ottawa.KIA 0R6.

INew facility to determine fire resistance of columnsT. T. LIEFire Research Section, Division of Building Research, National Research Council of Canada, Ottawa, Ont., Canada KIA OR6Received February 11, 1980Accepted March 27, 1980The new Division of Building Research (DBR) column testing furnace is described and an doutline given of the proposed studies that this facility makes possible.Le four de mise a I'essai des poteaux de la Division des recherches en bkiment (DRB) est decritet les differentes etudes que cette installation peut permettre sont enumirees.Can. J. Civ. Eng., 7,551-558 (1980)0315- 1468/80/030551-08$01.00/0@I980 National Research Council of Canada/Conseil national de recherches du Canada6 22 2--"!5- gJ

552 CAN. J. CIV. E LNG. VOL. 7, 1980IntroductionOne of the safety requirements for buildings todayis the provision of appropriate fire resistance. Untilfairly recently, the fire resistance of building componentscould only be determined by costly andtime-consuming tests. Now it can be determined bycalculation, thanks to the development of mathematicalmodels. Because these models only partiallyrepresent reality, however, their validity must beverified.A versatile column testing furnace has recentlybeen installed at the Division of Building Researchof the National Research Council of Canada. Thenew facility will enable the Division to test thevalidity of mathematical models to predict fireresistance, and to obtain a better understanding ofthe many problems for which there are yet notheoretical solutions. Although designed for researchpurposes, the furnace is also capable of performingstandard tests on specimens in accordance with mostinternational standards.Calculation ProcedurePredicting the fire behaviour of structural membersinvolves the specification of fire temperatures and thecalculation of temperatures, deformations, andstrength of structural members. Because all thesequantities vary with time the calculation procedureis complex, but it has been simplified by the developmentof numerical calculation techniques and the useof high-speed computers. The various steps incalculating the fire resistance of a member and determiningthe appropriate protection are shown inFig. 1.Fire LoadFirst is the determination of the fire load, i.e., thecombustible contents, which should form the basisfor the fire resistance design of a building. Severalsurveys have been carried out recently to obtaininformation on fire loads in buildings (Bryl et al.1974; National Academy of Sciences 1977).Fire TemperatureAssuming that the design fire load is known, thenext step is to determine time-dependence of the firetemperature. Calculation of the fire temperaturecourse of compartment fires has been the subject ofseveral studies (Kawagoe and Sekine 1963; odeen1963 ; Magnusson and Thelandersson 1970; Tsuchiyaand Sumi 1971; Harmathy 1972). The fire temperaturecourse can be determined from a heat balancefor the compartment, taking into account the heatproduced and the heat losses to walls and openings.The most important factors are fire load and thedimensions of the openings through which air,IDETERMINATIONOFDESIGN FIRE LOADICALCULATION OFFlRE TEMPERATURE CURVEr+=7PROPOSED PROTECTION1CALCULATION OFTEMPERATURES OF STRUCTUREMODIFYPROTECTIONrl,CALCULATION OF STRENGTHOF STRUCTURE DURING FlREFro. 1. Calculation procedure for fire-resistant design.necessary for combustion of the fire load, can enter.How fire load affects fire temperature is shown inFig. 2, the influence of openings in Fig. 3. Fire loaddetermines the duration of the fire, whereas theopenings influence both intensity and duration. Thecalculated fire temperature course or, eventually, astandard fire temperature curve is used as exposuretemperatures for the calculation of the temperaturesin structural members.Temperature of MembersTemperatures of fire-exposed structural members 4can be calculated; the most common procedures arethe finite difference and finite element methods. Inthese, the dependence of material properties ontemperature is taken into account and, in some cases,a high accuracy can be obtained. For example, theaccuracy with which temperatures in protected steelcan be predicted is better than 1% if the materialproperties of the protection are known precisely(Lie 1977). In practice this is usually not the case, butan accuracy of 15-20% is possible, which in the fieldof fire is regarded as fairly accurate.Although, in principle, it should be possible toI

TIME, hFIG. 2. Influence of fire load on fire temperature course.obtain for concrete the same accuracy as for steel, it* is in fact more difficult to predict concrete temperatures.The main reason is that the thermal propertiesof concrete depend strongly on the properties of theaggregate, which may vary over a wide range andare difficult to determine accurately. In addition,moisture in concrete will have considerable influenceon its internal temperatures.,Strength of MembersWhen the temperature rises in fire-exposedmembers, their strength lessens. If the fire load issufficient and the duration of the fire long enough,a stage will be reached at which the strength of themember will no longer be adequate to support thestructural load. The fire load that is just sufficient toreduce the strength to this critical point is definedas the critical fire load. The aim is to make the criticalfire load for a member sufficiently high to allow it towithstand the burn-out of the design fire load. Ingeneral, this can be achieved by selecting for themember appropriate protective materials of adequatethickness and appropriate dimensions.Test FurnaceThe test furnace was designed so that it couldproduce the conditions to which a member might beexposed during a fire, such as fire temperatures,structural loads, and heat transfer. It consists of asteel framework supported by four columns with thefurnace chamber inside the framework (Fig. 4;Table 1 gives details of furnace characteristics).Loading DeviceThree hydraulic jacks produce forces along thethree principal axes. The jack acting along the axisof the test column is designed for a maximum loadof 2200 kips (9790 kN). This capacity is sufficient tosubject concrete columns of approximately 30 in. x30 in. (760 mm x 760 mm) to their design load.This jack can also be used to restrain the thermalTIME, hFIG. 3. Influence of openings on fire temperature course.expansion of a column and thus simulate the conditionsin fires where column expansion is restrainedby upper floors. The other two jacks exert lateralforces at the top of the column, to simulate the forcesand displacements due to floor expansion during afire. Loads and movements of the jacks can be controlledmanually or by computer, and can be programmedto follow a predetermined course, makingit possible to simulate a wide range of practicalconditions.Furnace ChamberThere are 32 propane gas burners in the furnacechamber arranged in eight columns of burners containingfour burners each. The total capacity is16 000 000 Btu/h (4700 kW), which should besufficient to simulate any fire severity encountered inpractice.The Dressure in the furnace chamber is controlledautomatically. It is usually set somewhat lower thanatmospheric pressure to avoid a flow of gases fromthe chamber to outside. If necessary, for example tomeet ISO1 standards, the furnace can be run at anoverpressure.It is possible to test both columns and walls with afire-exposed length of between 10 and 14 ft (3.1 and4.3 m) in the furnace. Heating can be controlled insuch a way that one or more sides of the specimenare exposed to fire.The furnace chamber, whose floor area is 8 ft 8 in.(2.6 m) square, is made of an insulating material thatwill produce high heat transfer to the specimen. Theroof can move freely upwards and functions as ablow-out panel as a safety measure against explosion.Safety against explosion is provided by variousother measures. Before ignition of the pilot burnersthere is a purge period in which an amount of airabout three times the volume of the furnace chamberis blown through it. In each burner there is an ultra-'International Organization for Standardization.

554 CAN. J. CIV. ENG. VOL. 7, 1980FIG. 4. Test furnace and specimens.violet sensor that can detect failure of a burner. Ifthis happens, the gas supply to the column of burnersin which the failure occurred will shut off automatically.It is possible, however, to restart theburners manually, without purging, provided thefurnace temperature has reached a level at whichthere is no danger of explosion. A temperature of1000°F (588°C) was selected as being safe. Thistemperature is usually reached after 5 min in astandard test. Being able to restart the burnerswithout purging prevents the loss of a test in anadvanced stage.InstrumentationFurnace temperatures are measured with the aidof Chromel-Alumel thermocouples installed 1 ft(0.3 m) from the test specimen at various heights.There are two thermocouples placed opposite each

NOTESTABLE 1. Furnace characteristicsLoading deviceCapacity axial load [kips (kN)] 2200 (9790)Capacity lateral load (x-axis) [kips (kN)] 70 (310)Capacity lateral load (y-axis) [kips (kN)] 25 (110)Maximum movement of axial jack [in. (mm)] 12 (305)Maximum movement of lateral jacks [in. (mm)] 6 (152)Maximum speed of jacks [in./min (mm/rnin)] 3 (76)Accuracy of controlling and measuring loads(% at low loads) Approx. + 5Accuracy of measuring displacements [in. (mm)] kO.004 (0.1)Furnace chamberInner height of chamber [ft (m)] Up to 14 (4.3)IFloor area [in. (m)] 104 (2.6)Capacity of 32 burners [Btu/h (kW)] 16000 000 (4700)Temperature uniformity (% deviation fromaverage temperature after 5 min) Less than 2Pressure in chamber at 9 ft (2.7 m) heightAdjustable between - 0.2 and+0.2in. (-5and +5mm)H20Properties of material of floor, ceiling, and doors (ceramic blanket)At 932°F (773 K)Density [Ib/ft3 (kg/m3)] 6 (96)Specific heat [Btu/lb. "F (J/kg.K)] 0.26 (1 090)Thermal conductivity [Btu/ft .h. "F (W/m .K)] 0.007 (0.12)At 2000°F (1352 K)Density [Ib/ft3 (kg/m3)] 6 (96)Specific heat [Btu/lb. "F (J/kg.K)] 0.27 (1130)Thermal conductivity [Btu/ft . h . "F (W/m .K)] 0.24 (0.42)Properties of material of walls (insulating fire brick)At 932'F (773 K)Density [Ib/ft3 (kg/m3)] 45 (720)Specific heat [Btu/lb. OF (J/kg. K)] 0.24 (1000)Thermal conductivity [Btu/ft . h "F (W/m .K)] 0.15 (0.26)At 2000°F (1352 K)Density [lb/ft3 (kg/m3)] 45 (720)Specific heat [Btu/Ib.OF (J/kg.K)] 0.31 (1300)Thermal conductivity [Btu/ft h. "F (W/m.K)] 0.28 (0.48)other every 2 ft (0.6 m) along the height of thefurnace chamber, starting from the floor. The temperaturesmeasured by the thermocouples areaveraged automatically and the average temperatureused as the criterion for controlling the furnacetemperature.As every burner can be adjusted individually a hightemperature uniformity can be obtained. At presentthe uniformity is such that after 5 min the deviationof temperature at the various measuring points fromthe average furnace temperature is less than 2%.Triaxial loads are measured by pressure transducersthat supply an electrical signal to recorders.Axial deformations of test specimens and lateraldisplacements at the top of the specimen are measuredwith linear potentiometers. In testing restrainedspecimens, the deformations of the steel frameworkof the furnace are measured with strain gauges.All electrical signals, i.e., those from temperature,load, and deformation sensors, can be used tocontrol these quantities. A computerized control anddata acquisition system is usedfor this purpose andfor measuring the important variables that determinethe performance of test specimens duringexposure to fire.A problem of long standing is how to measureaccurately the deflection of a specimen in a furnace:the temperature is too high to use conventionalmethods. A method that has been developed recentlyby the Division of Physics, NRCC, is used to measuredeflections in this furnace. In this method (Chapmanand Green 1977), whose general purpose is tomeasure movements in a hostile environment (Fig.5), a laser beam is split into two beams with the aidof a beam doubler. The beams are focused on atarget, in this case the centre of the column. Thus twospots are formed on the target. If the target moves,the distance between the two spots changes. Withthe aid of lenses and mirrors an image is formed ofthe two spots. The image of each point is reflected,

CAN. J. CIV. ENG. VOL. 7, 1980PlaneMirror Lens LensBwmt]&ler A M' L' LzSychmnw5 1 7and Micrometerme.---1UIIMotor md . \, % Snde mand lnta, ... "t Mirror4 .i. ----afrmer-a d IR MZW-cell P b W iand Firnd SlitFIG. 5. Device for measuring deflections (Division of Physics, National Research Council of Canada).TABLE 2. Information on test columnsNumberof Cross section, Moisturecolumns [in. x in. (mm x mm)] Load condition Aggregate Purpose of test3 12 x 12 0 Atmospheric Siliceous Repeatability(305 x 305)1 8x8 0 Dry Siliceous Heat transfer(203 x 203)1 12 x 12 0 ~ i y Siliceous Heat transfer(305 x 305)1 16 x 16 0 Dry Siliceous Heat transfer(406 x 406)1 12 x 12 Design Dry Siliceous Loaded model2(305 x 305)12 x 12loadTo be Atmospheric Siliceous Influence load(305 x 305) selected1 12 x 12 0 Atmospheric Carbonate Influence aggregate(305 x 305)2 12 x 12 To be Atmospheric Carbonate Influence aggregatex (305 305) selectedNOTE: Column lengths are all 12.5 ft (3.8 m).one after the other, on a photocell with the aid of arotating mirror. In this way, pulses are generated bythe cell. Because the time interval between two pulsesis proportional to the displacement of the column, thedeflection of the column can be determined bymeasuring this time interval.The deflections of a column are measured alongthe two orthogonal axes perpendicular to the columnaxis. Tests in an existing furnace show that the deviceis capable of measuring displacements with anaccuracy of k0.004 in. (0.1 mm) at a distance of61 in. (1550 mm).Test ProgramConcrete ColumnsThe first series of columns intended for testing, asa part of a joint study, are reinforced concretecolumns constructed by the Portland CementAssociation. Mathematical models for calculatingthe temperature distribution in square columns andtheir strength during exposure to fire were developeda few years ago (Lie and Allen 1972; Allen and Lie1974). By subjecting the columns to the fire con- 'ditions and loads assumed in the models it will bepossible to verify the validity of these models.Figure 4 shows one column installed in the furnace ,and two, awaiting tests, on the right. In total 12columns were constructed for the first series of tests;information on these columns is given in Table 2.Three columns of 12 in. x 12 in. (305 mm x 305mm) cross section were constructed to examine therepeatability of the tests. Three columns 8 in. x 8 in.(203 mm x 203 mm), 12 in. x 12 in. (305 mm x305 mm), and 16 in, x 16 in. (406 mm x 406 mm),will be tested unloaded to verify the heat transfermodel. One column will be tested under the designload to compare the calculated with measured fireresistance. Two columns will be tested under lower

NOTES 5571200-Y1auuu'- CALCULATED30 60TIME, mlnFIG. 6. Temperatures at various depths along centre line in12 in. x 12 in. (305 mm x 305 mm) column as a function oftime.load to obtain information on the influence of overdesign.The other columns were made with carbonateaggregate to determine the influence of the aggregateon the fire resistance of the columns.Tests on a second series are planned. The columnswill be larger and the influence of the steel reinforcementcover thickness on the fire resistance of thecolumn will be examined. It is also intended to testmore columns made with carbonate aggregate, tosee if they have a higher fire resistance than thosemade with siliceous aggregate.Figure 6 shows the results of the first test of theseries. This was carried out on a column of 12 in. x12 in. (305 mm x 305 mm) to examine the temperaturedistribution in the column. In Fig. 6calculated temperatures are compared with temperaturesmeasured at the following depths: 1, 23,and 6 in. (25, 63, and 152 mm). Calculated temperaturesare higher than most measured temperatures,but the differences are relatively small. The resultsindicate that the heat transfer model is reasonably ,accurate, although it is somewhat conservative. Moretests under various conditions must be carried out,however, to establish a mathematical model that canbe used with confidence for such calculations.SteelOne of the mathematical models that has beenvalidated is for heat transfer in protected steel (Lie1977). Often a critical temperature of the steel can bespecified for use as a criterion of failure. In this casethe calculation of failure can be reduced to thecalculation of the temperature of the steel. NorthAmerican standards, for example, assume an averagesteel temperature of 1000°F (538°C) as the failuretemperature of steel columns. Although accurateprediction of steel temperatures is possible thisassumption creates an uncertainty in the predictionof failure. It is probable that factors such as load andslenderness of the column play an important role indetermining the failure temperature. With the aid ofthe new column furnace it will be possible toinvestigate this.Another project now under study is the fire resistanceof concrete-filled steel columns. These columfishave several advantages, one of which is thatconcrete filling can be provided at a lower cost thanconventional protection. In addition, by designingthe column so that the steel carries the load at roomtemperature and the concrete carries it at fire temperatureswhen the steel has lost most of its strength,a fire resistance of more than 2 h can be obtained.This resistance is obtained without added protection;unfilled the fire resistance of the steel would be notmore than about 10 min. Another advantage of fireprotection by concrete filling is that some floor spaceis gained because there is no outside protection.The concrete filling also substantially increases thestrength of the column but there must be enoughconcrete to obtain sufficient fire resistance, whichmay be more than is needed for structural loads.The construction of mathematical models for thecalculation of the fire resistance of concrete-filledcolumns is at an advanced stage. It is anticipatedthat the new column test furnace can soon be usedto study the performance of these columns.ConclusionDevelopment of numerical techniques and theintroduction of powerful computers have acceleratedthe transition from testing to calculation of fireresistance. Few facilities exist, however, that arecapable of producing the conditions necessary todetermine the adequacy of the mathematical modelsused for the prediction of fire resistance. NRCC'sDivision of Building Research has now installed aversatile testing furnace that can test full-scalecolumns and walls under various load and fireconditions.AcknowledgementPlant Engineering Services of the NationalResearch Council of Canada was responsible for thedesign and construction of the column furnace.Thanks are due to E. A. Cunningham for his majorcontribution in accomplishing this work. This paperis a contribution from the Division of BuildingResearch, National Research Council of Canada andis published with the approval of the Director of theDivision.

558 CAN. J. CIV. ENG. VOL. 7, 1980ALLEN, D. E., and LIE, T. T. 1974. Further studies of the fireresistance of reinforced concrete columns. Division of BuildingResearch, National Research Council of Canada, Ottawa,Ont., NRCC 14047.BRYL, S., DEMOL, L., CZECH, R., and buss^. 1974. Fire loads inoffice buildings. Fire Safety in Constructional Steelwork,European Convention for Constructional Steelwork, Paris.France.CHAPMAN, G. D., and GREEN, E. 1977. A non-interferometricoptical probe. Review of Scientific Instruments, 45(6), pp.617-624.HARMATHY, T. Z. 1972. A new look at compartment fires. Part Iand Part 11. Fire Technology, 8(3), pp. 1%-217; 8(4), pp.326-351.KAWAGOE, K., and SEKINE. T. I%3. Estimation of fire temperature- time curve in rooms. Building Research Institute,Ministry of Construction, Tokyo, Japan, BRI Occasional ReportNo. 11.LIE, T. T. 1977. Temperature distributions in fire-exposedbuilding columns. Journal of Heat Transfer, 99(1), pp.113-1 19.LIE, T. T., and ALLEN, D. E. 1972. Calculation of the lireresistance of reinforced concrete columns. Division of BuildingResearch, National Research Council of Canada, Ottawa,Gt., NRCC 12797.MAGNUSSON. S. E.. and THELANDERSSON, S. 1970. Temperature-timecurve; of complete process of fire development.Theoretical study of wood fuel fires in enclosed spaces. ActaPolytechnica Scandinavica, Stockholm, Sweden, Civil Engineeringand Building Construction Series No. 65.NATIONAL ACADEMY OF SCIENCES. 1977. Fire and live loads inbuildings. Committee on Fire and Live Loads in Buildings ofthe Building Research Advisory Board, National Academy ofSciences, Washington, DC.ODEEN, K. 1%3. Theoretical study of fire characteristics inenclosed spaces. Division of Building Construction, RoyalInstitute of Technology, Stockholm, Sweden, Bulletin 10.TSUCHIYA, Y., and SUMI, K. 1971. Computation of the behaviourof fire in an enclosure. Combustion and Flame, 16, pp.131-139.I