ASHRAE NY-08-023-2008 Assessment of Fire Heat Release Rate for Train Fires《火车释热率评估》.pdf

1、172 2008 ASHRAE ABSTRACT This paper provides a discussion on current methodolo-gies used to predict train fire heat release rates by discussinghow the various fundamental combustion thermophysicalproperties vary as a function of temperature and equivalenceratio. This paper illustrates that the therm

2、ophysical propertiesvary but are being assumed constant, and it identifies that thereis a need to reconsider how to properly use these properties andto develop an extensive database of fundamental properties inorder to improve the quality of the answers. In addition, thispaper points out that combus

3、tion must be taken into accountwhen determining the fire heat release rate and that there is nouniversal answer, so each case needs to be evaluated on its ownmerits. Knowing answers on a per-unit-area basis is notenough to scale off to a full three-dimensional domain.INTRODUCTIONFire safety has been

4、 at the forefront of infrastructure proj-ects. Public safety is top priority when a design is postulated.Means of evacuation is a major requirement. But, what to usefor fire design criteria? In mass transit systems, what toassume when a train fire is the governing design constraint ofthe project is

5、of most concern. Should we have a conservativeapproach? This has been the attitude of many designers, butthe question remains: how much conservatism is too much? Orbetter yet, on what should we be conservative?The International Building Code (ICC 2000) requires thatdesign fires not be less than 5.2

6、MW unless there is a rationalanalysis performed. NFPA Standard 130 (NFPA 2003)requires an engineering analysis that includes the fire heatrelease rate (FHRR) produced by the combustible load of atrain car, any combustible materials that could contribute tothe fire load at the incident site, the fire

7、 growth rate, station andtrainway geometries, and operation of mechanical ventilationto control the spread of smoke. Note that NFPA Standard 130does not specify a minimum FHRR, but it requires the FHRRto be determined by engineering analysis.It is well known that a fire has ignition, growth, flashov

8、er,fully developed, and decay stages. In each one of these stages,we need to predict the FHRR. In the early years of fireresearch, the FHRR was determined by burning materials ona load cell, where the FHRR was estimated from the mass lossand heat of combustion. Today, the method of oxygenconsumption

9、 is used very widely. The underlying assumptionof this method is that the FHRR is almost constant per unitmass of oxygen consumed. But even this methodology has itspitfalls, as will be explained below.This paper focuses on fire science as a branch of combus-tion engineering and highlights key issues

10、 that need to be takeninto account whether we are trying to estimate the FHRRexperimentally or numerically.THERMAL PROPERTIESThere is no doubt that we cannot predict the FHRR unlesswe know the thermal properties of the materials. In general,fire scientists have focused on the heat of combustion, Hc,

11、 theFHRR from a test sample, and the CO and CO2yields.Certainly these are key parameters, but it should be noted theyare not constant. The heat of combustion is a function of thefirst and second laws of thermodynamics and chemical equi-librium. As such, the heat of combustion is a function oftempera

12、ture and the equivalence ratio, , (ratio of stoichiomet-ric air-fuel-ratio over actual air-fuel-ratio). Walton andThomas (2002) show the effect of on the flashover compart-Assessment of Fire Heat Release Rate for Train FiresJ. Greg Sanchez, PEMember ASHRAEJ. Greg Sanchez is principal mechanical engi

13、neer with MTANew York City Transit Capital Program Management, New York, NY.NY-08-0232008, American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc. (www.ashrae.org). Published in ASHRAE Transactions, Volume 114, Part 1. For personal use only. Additional reproduction, distribut

14、ion, or transmission in either print or digital form is not permitted without ASHRAEs prior written permission.ASHRAE Transactions 173ment fire temperatures following a method based on Babraus-kas (2002) work.Figure 1 plots the calculated heat of combustion for gaso-line at various temperatures and

15、equivalence ratios, . Notethat the heat of combustion could become zero and could evenswitch from exothermic (releasing energy) to endothermic(absorbing energy). Therefore, when a material is tested, whatis very important to determine is its heat of formation. Thiswould become input to the thermodyn

16、amic model in order tobe able to predict the heat of combustion more realistically,and thus the FHRR. The surface ignition temperature is also important and isusually determined under a pilot condition. However, what ismost important is to determine the spontaneous surface igni-tion temperature (non

17、-pilot). This is the temperature at whichthe fuel surfaces would release fuel vapors due to convectionor radiation. We know that the pilot surface ignition tempera-tures could be as much as half of the spontaneous surface igni-tion temperatures (Drysdale 1998). Without attempting tomeasure the spont

18、aneous surface ignition temperature, wewould only be able to assume pilot ignition data, which couldlead to predicting faster growth rates.Along the same lines is the mass loss rate. The mass lossrate is a function of the surface temperature. The hotter thesurface, the more mass given off by the mat

19、erial. A commonapproach is to use an Arrhenius approximation (Equation 1).A, E, R, and T are the pre-exponential coefficient, activationenergy, universal gas law constant, and local temperature,respectively. (1)We need to evaluate the pre-exponent constant and theactivation energy for each potential

20、 fuel in order to properlyaccount for the mass loss fluctuation as a function of thesurface temperature. Other properties of importance are themass density, thermal conductivity, specific heat, and heat ofgasification. Gross (1985) provides some data, but more isneeded, especially on traincar materi

21、als.The US DOT (Lyon and Janssens 2005) has recentlymade available a publication that lists flammability informa-tion for polymers. In this report, it is stated that the pilot igni-tion temperature is almost constant regardless of the externalincident heat flux. However, depending on the material, t

22、heignition temperature would vary between 225C and 645Cwith an average of 423C and a standard deviation of 91C. Itshould be noted that these temperatures are for pilot ignitionand not for spontaneous ignition.BURNING RATE VS. MASS LOSS RATEOne of the most misunderstood variables is the burningrate.

23、The FHRR is calculated by multiplying fuel mass burningrate times the heat of combustion of the fuel (Drysdale 1998). (2)Many interpret fuel mass burning rate to be the fuel massloss rate (Ingason 2005; Walmerdahl and Werling 2003);others interpret it to be the fuel mass flow rate embedded intheir d

24、efinition of . These rates are similar to each other butare very distinct at the same time. It should be clear that fuelmass burning rate is the fuel mass that reacts in the flame zoneand enters into combustion, provided the fuel and air arewithin the flammability limits and the temperature above th

25、efuel gaseous autoignition temperature. The assumption ofusing the fuel mass loss rate, or even the fuel mass flow rate,may be acceptable under a well-stirred reactor and premixedconditions. However, most fires we deal with in the fire protec-tion industry are diffusion flames (non-premixed) and spr

26、eadover a large space, unlike internal combustion in engines. Thismeans that the premixed approximation would not be theproper methodology to assess the FHRR in trains or other largespaces. To assess this properly, the fire scientist needs to lookinto flammability limits. Based on combustion princip

27、les andthe equivalence ratio (Glassman 1987), the flammabilitylimits have been found to range between 0.5 (lower limit) and3.3 (upper limit), although some could be higher and somelower. A lean diffusion flame can be sustained between 0.5 1.0, while a rich diffusion flame can be sustained between1.0

28、 3.3. Therefore, the flammability limits of materialsneed to be evaluated in order to properly account for the flam-mable range. Then, based on the first and second laws of ther-modynamics and chemical equilibrium, the chemical equationwould need to be balanced as a function of lean, stoichiomet-ric

29、, or rich conditions and temperature.COMBUSTION PRODUCTSThe combustion process of a fire is commonly simplifiedas a pseudo first-order reaction, as shown in Equation 3. Thisreaction depends on the type of fuel and the amount of airrequired to react, called the air-fuel-ratio (AFR). Manyapproximate t

30、he chemical reaction assuming a stoichiometricformulation as shown in Equation 4 (Perhson 1998). Stoichio-metric conditions are for an equivalence ratio of 1.0 ( = 1.0).Figure 1 Heat of combustion for gasoline as a function of .mfuelAeE RT()Qmfuel burning rateHc=174 ASHRAE TransactionsThe heat of co

31、mbustion tabulated in most books is at = 1.0and an ambient temperature of 298 K. Many assume theseproperties in their analyses, which could yield errors if one isnot careful.1 kgfuel+ (AFR) kgairJ (1 + AFR) kgproducts + heat (3)Fuel + a(O2+ 3.79N2) J bCO2+ dH2O + gN2+ heat(4)Some approximate the che

32、mical reaction following amore detailed chemical reaction (Equation 5) and using exper-imental measurements of CO and CO2. They performed theanalyses holding the measured yields constant for all ranges of and temperatures. Others even program the CO and CO2curves measured in the experiments.Fuel + a

33、(O2+ 3.79N2) J bCO2+ cCO + dH2O + eO2 + f H2 + gN2+ heat (5)The fact is that the products vary as a function of temper-ature and . Figures 2, 3, and 4 illustrate how they vary forT = 500, 1000, and 1500 K, which are typical flame/reactiontemperature ranges. It should be noted that at 500 K, COformat

34、ion does not start to appear until 1.5 (Figure 2).Then, at 1000 K and 1500 K, CO starts forming at = 1.0(Figures 3 and 4). For all cases, CO formation increases as increases. Another point to observe is the formation of CO2. WhenT = 500 K, CO2formation increases up to = 1.5, after whichit starts to

35、decrease. This coincides with the starting of COformation. As the temperature increases, the CO2formationpeak shifts to the left, after which it drops all the way to zeroat = 3.1. The temperature increase shapes the CO2formationinto a decaying exponential curve.A last point to observe is how water v

36、apor (H2O) varies.This formation has a strong impact on the heat of combustion.All this confirms that the products of combustion are notconstant and must be taken into account in order to properlymodel and measure FHRR.OXYGEN CONSUMPTIONThe fire research concept of oxygen consumption hasbeen a commo

37、n practice among modelers and experimental-ists since the late 1970s. The basis is that for most organicmaterials, the amount of heat released is more or less constanton a per-unit mass of oxygen consumed. The value is about13.1 MJ/kg (Janssens 2002; Ingason 2005; Carlsson 2005;Lonnermark 2005). How

38、ever, this value is for completecombustion and at standard room temperature of about 25C.This method is now recognized as the most accurate and prac-tical technique for determining FHRR from experimentalfires. Correlations have been developed with a 10% accuracy,provided that combustion is complete

39、(Equation 4). However,the error is larger if CO and soot production is considerable(Equation 5) or if a significant amount of combustion productsconsist of species other than CO2or H2O (Janssens 2002).Most of the fires we deal with are very sooty and have incom-plete combustion, which leads us to co

40、nclude that perhapsoxygen consumption is not as accurate as we anticipate. If thecalculations were made, it would be found that the heat ofcombustion per unit of oxygen consumed follows a similarFigure 2 Combustion product for gasoline as a function of at T = 500 K.Figure 3 Combustion product for ga

41、soline as a function of at T = 1000 K.Figure 4 Combustion product for gasoline as a function of at T = 1500 K.ASHRAE Transactions 175pattern as that shown in Figure 1it would decrease as a func-tion of temperature and .Therefore, oxygen consumption is an accurate methodol-ogy as long as it remains w

42、ithin the combustion constraints inwhich it is defined.STATION AND TRAINWAY GEOMETRYIt would be convenient to determine a universal FHRR fora train. But the fact is that based on combustion principles, theventilation paths available for the air and smoke to be trans-ported in the enclosed space are

43、key. For instance, if a trainwere placed in a very tight tunnel (Figure 5a), the FHRRwould be much less than at a station. This is caused by theinability to displace the smoke easily and allow fresh air toreach the fire region. Smoke would still be the biggest prob-lem, but the FHRR would be lower.

44、If the same train were placed in a station with a tight ceil-ing but had the tunnel wall on one side and open platform onthe other (Figure 5b), the FHRR would be larger compared tothat of the train in the tight tunnel. If the station geometry weresuch that one side of the train was exposed to a trac

45、k while theother was exposed to the platform (Figure 5c), the FHRRwould be different as well (higher or lower, but different). Ifthe station had a smoke management system that could removesmoke, the FHRR would be different still. The differencewould be a particular answer for the specific configurat

46、ion.This indicates that when a train FHRR is estimated experi-mentally, there is no universal set of answers, but each locationhas to be evaluated individually.TRAIN GEOMETRYWhen trying to predict train FHRR, the train geometry aswell as the distribution of the flammable materials inside thetrain ar

47、e very important. The tunnel and station geometry areimportant in order to provide means of dispersion of the smokeand makeup air to feed the fire; the train geometry is importantto predict the growth and sustainability of the fire inside atraincar. The location, size, and number of windows and door

48、sof the train are fundamentally important. They are the inter-face between the train cabin and the tunnel/station environ-ment. If only windows are open, and if windows are locatedvery high above floor level, the smoke layer would not allowenough air to enter and feed the fire. If doors were open, f

49、reshair to feed the fire would come in the lower level, while hotsmoke would leave by the upper level. However, in order toobserve these effects, full combustion must be accounted for,and the FHRR needs to be calculated rather than prescribed.When a FHRR is prescribed, the laws of physics can conflictif overpredicting input parameters.Some scale experiments have been conducted (Ingason2005). However, when it comes to fire scaling and scalingcombustion phenomenon, some of the rules used to scalefrom the model to full scale are questiona