railcar fire size a

Upload: joe-halohali

Post on 08-Aug-2018

215 views

Category:

Documents


0 download

TRANSCRIPT

  • 8/22/2019 Railcar Fire Size A

    1/11

    Fourth International Symposium on Tunnel Safety and Security, Frankfurt am Main, Germany, March 17-19, 2010

    495

    Predictions of Railcar Heat Release Rates

    John Cutonilli & Craig Beyler

    Hughes Associates, Inc

    3610 Commerce Dr, Suite 817

    Baltimore, MD 21227 USAEmail:[email protected],[email protected]

    KEYWORDS: Railcar, Heat Release Rate, Modelling, Experimental

    OVERVIEW

    The design of tunnel ventilation and other fire safety systems for rail tunnels and stations depend upon

    knowledge of the heat and smoke production rate from the rail car. This process involves thedetermination of material fire properties of the railcar materials, predicting the fire growth based upon

    the size of the initiating fire, and determining the heat and smoke generation rate history of the car.

    This paper discusses a methodology that can be used to predict railcar heat release rates and discusses

    the key concepts that impact the results.

    The best method for determining the heat release rate history for a rail car is to physically test the

    railcar itself using various fire scenarios in multiple full scale fire tests. This method has a number of

    limitations. The primary limitation is in the cost. A new railcar is a multi million dollar piece of

    equipment. Even mock-ups would cost hundreds of thousands of dollars to construct and instrument.

    Most situations will also require multiple tests that reflect different situations, such as different

    ventilation conditions, different fire scenarios, or different materials in the cars. The size and

    configuration of the railcar require unique fire test facilities that can conduct such tests.

    Hughes has developed a methodology to overcome these limitations. The methodology involves a

    combination of computer fire modelling and small-scale fire testing to determine the smoke and heat

    release rate histories. The small scale testing is used to generate needed inputs to the computer fire

    models. Two validated computer fire models (HAIFGMRail, and HAICFMRail) are used to predict

    the heat and smoke generation during all stages of the fire, which may include the early stages of a fire

    (pre-flashover), occurrence of flashover, fully-developed (post-flashover), decay, and complete

    burnout. These computer models are be used to evaluate the potential for fire spread to adjacent

    railcars in the train. The models themselves have been published in the peer reviewed fire science

    literature [1,2], have been validated by comparisons with available data, and have been used for a

    number of rail systems in support of emergency ventilation design.

    The computer fire models used to determine the smoke and heat release rate histories require inputs

    that are best obtained from small scale testing such as the cone calorimeter test. The cone calorimeter

    data is used to develop model input parameters for all car materials, including thermal properties,

    ignition temperature, pyrolysis and burning properties, heat release rate, and smoke and species yields.

    Cone calorimeter tests are normally conducted in triplicate at three incident heat fluxes, in addition to

    the determination of the critical heat flux for ignition. This small scale testing requirement can be quite

    significant when as many as 10-12 materials often found in a rail car.

    The primary sources of heat and smoke in the railcar come from the interior finish materials. Early

    stages of the fire require that the spread of fire across these materials be determined. A pre-flashover

    flame spread model (HAIFGMRail) is used to determine this spread along combustible surfaces within

    a compartment, in this case a railcar. From this flame spread, the model predicts the amount of heat

    mailto:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]
  • 8/22/2019 Railcar Fire Size A

    2/11

    Fourth International Symposium on Tunnel Safety and Security, Frankfurt am Main, Germany, March 17-19, 2010

    496

    and smoke generated during the early stages of the fire (pre-flashover) up to the occurrence of

    flashover. The model incorporates heat feedback from the compartment to predict the transition to

    flashover. It is also capable of predicting burnout of the flames if the conditions in the compartment do

    not result in the transition to flashover within the compartment (decay of the fire).

    The pre-flashover model is used as a part of the risk assessment of the railcar. This analysis is typicallyperformed on a number of different types of fire scenarios [3] including accidental and intentional

    (arson) fire scenarios. Using the pre-flashover model, a determination of the fire size and associated

    quantities of combustibles needed to flashover the railcar for the various fire scenarios can be made.

    This information can then be used to determine the risk of the fully involved railcar. i.e. small fires

    occur more frequently but to not cause the railcar to flashover while larger fires are more rare, but may

    cause the railcar to flashover.

    A second computer fire model (HAICFMRail ) is used to determine the heat and smoke release rate

    histories after flashover. It is capable of predicting the window failure during the fire exposure and will

    modify the ventilation into the railcar based on the failure of windows. Window failure is based on

    some experimental results [4]. The change in ventilation into the railcar has a significant impact on the

    heat release rate history [5]. If sufficient fuel is available, increasing the ventilation into the railcarwould result in an increase the heat release rate of the fire. Conversely, increasing ventilation into the

    railcar could also cause the fire to transition into the decay stage if insufficient fuel is available to

    support a fully-developed fire under the higher ventilation conditions.

    Fire spread to adjacent cars can occur through hot gas/flame projections out of side or end windows.

    Calculations based on these two models are conducted to evaluate the potential for fire spread to

    adjacent railcars based on window failure times. If fire spread to an adjacent car is predicted, then the

    heat and smoke release rate histories of the newly ignited railcar will be predicted and added to the

    railcar already burning.

    FLAME SPREAD MODEL

    The flame spread model, HAIFGMRail, is a computer fire growth model developed by Hughes

    Associates, Inc. (HAI) [1, 6-10] to determine pre-flashover heat release rates from combustible

    finishes. It has been in development for over twelve years and is the primary tool for calculating the

    upward and lateral spread of fire on combustible wall and ceiling surfaces in a corner configuration in

    the presence of a hot gas. It uses various sub-models based on published data and methodologies to

    address the following aspects of the calculation:

    The flame and thermal plume heat fluxes to the wall and ceiling surfaces;

    The ceiling and wall boundary temperatures;

    The compartment temperature; and

    The ignition and pyrolysis of combustible lining materials.HAIFGMRail computes the fire spread in a combustible corner on an elemental basis. A corner region

    is defined by two walls and a ceiling; each wall and the ceiling are subdivided into a number of

    elements or material cells over each of which the temperature, flux, and pyrolysis conditions are

    assumed constant. The cell size is user selectable and is typically on the order of 0.1 m or less. Figure 1

    depicts a typical corner region.

    Flame spread is governed by the thermal properties and ignition temperature of the exposed material.

    When the surface temperature of a material cell reaches or exceeds the ignition temperature of the

    surface material, the cell is ignited. The surface temperature is calculated by a heat balance between the

    incident heat flux, the heat flux conducted into the material, and the heat flux convected and radiated

    back into the compartment is performed at the element surface. This heat balance is performed using a

    transient finite difference calculation through the total cell thickness and results in an array oftemperatures that approximate the temperature distribution through the boundary at that cell location.

  • 8/22/2019 Railcar Fire Size A

    3/11

    Fourth International Symposium on Tunnel Safety and Security, Frankfurt am Main, Germany, March 17-19, 2010

    497

    The incident heat flux is comprised of three components: the heat flux direct from the source fire, the

    heat flux from burning wall or ceiling cells, and the heat flux from the hot gas layer, if present. The hot

    gas layer is computed using a two-zone model that is based on the work of McCafferey, Harkleroad,

    and Quintiere [1113]. The key parameters that are calculated using this sub-model are the upper layer

    temperature, the lower layer temperature, layer interface elevation above the compartment floor, and

    the neutral plane elevation.The heat release rate from a burning cell is determined using cone calorimeter transient heat release

    rate data measured at a reference incident heat flux. The data heat release rate and the time at which

    this heat release rate occurs are scaled from the reference heat flux to the current incident heat flux and

    the fire duration at any one location is a function of the total energy evolved at the scaled heat flux. The

    approach effectively provides for a variable heat of gasification and allows a reasonably accurate

    simulation of the burning of materials that char or undergo a physical change.

    Figure 1. Typical Corner Model Grid.

    POST FLASHOVER COMPARTMENT MODEL

    A sophisticated single room, one-layer model is used to predict the burning rates within the fully

    involved rail car. As fires grow, railcar conditions become the dominate factor in determining fire

    growth and the subsequent burning rates. Typical models are not sufficiently sophisticated to predict

    the burning rate [14, 15]. Instead they leave this important parameter for the user to determine. The

    present model, HAICFMRail, allows the interrelationship between the compartment temperature, the

    airflow rates, and the burning rate to be determined to properly model the burning rate based on

    compartment conditions.

    The one layer model is a classical method for calculating compartment conditions during a fire [16]. A

    one layer model is typically used to calculate post-flashover fires because the interface has already

    moved to a height near the floor and conditions are generally uniform during the fully developed

    burning period. Several researchers have also shown success in using a one-layer model to approximatecompartment conditions [17-19]. The one-layer model was chosen because it is a simple proven

  • 8/22/2019 Railcar Fire Size A

    4/11

    Fourth International Symposium on Tunnel Safety and Security, Frankfurt am Main, Germany, March 17-19, 2010

    498

    method that can adequately represent a range of fully developed fire conditions.

    The most important feature of the model is the ability of the model to predict the burning rate of

    multiple materials. An unlimited number of materials can be modeled. The model allows a user

    specified burning rate, but the model will always limit the maximum burning rate based on the surface

    area, mass and compartment conditions. During the early part and late stages of the fire (growth anddecay stages), the burning rate is based on radiative feedback from the compartment and from the fire

    itself. The burning rate will solely be based on radiative feedback from the compartment if the

    compartment reaches fully developed burning. The model also predicts the failure of windows and

    reflects the effect of the change in ventilation on burning rate in the car.

    The model also has sophisticated flow rate and heat transfer routines. The flow rate routines calculate

    flows into and out of the compartment based on compartment temperatures. An unlimited number of

    vents can be specified and the openings of each of these vents can change with time. The heat transfer

    routines allow a compartment to be divided into an unlimited number of different heat transfer regions

    or boundaries. For example, the walls, ceiling, floor, and windows can all be specified as different

    regions. Each of these regions can be specified with multiple materials and each material specified with

    temperature dependent properties. This allows a window to be specified as a single material, whilewalls, ceilings and the floor specified with multiple materials (i.e., lining and insulation).

    EXAMPLE

    To demonstrate this methodology, a sample analysis was conducted on a representative intercity type

    railcar. Figure 2 depicts the layout of the railcar. The primary fuel source for the railcar fires was the

    interior finish materials. Fire performance data was measured using the cone calorimeter using

    representative samples of the interior finish materials.

    Figure 2. Railcar layout.

    Railcar Modeling Results

    The fire growth modeling was performed to predict whether the railcar would reach flashover. This

    model has three types of input, information on the materials of the railcar, initiating fire sizes, and

    ventilation conditions. For most situations the materials of the railcar are given and the minimum

    initiating fire size to cause flashover needs to be determined for different ventilation conditions. The

    model could also be used the other way around. A maximum design basis fire size could be chosen and

    different materials could be chosen to prevent the railcar from flashing over.

    The typical used of the model involves modeling the conditions that develop inside of the railcar with

    different types and sizes of initiating fires in the railcar. Different initiating fires have different growth

    Window Pairs

    0.60 m x 1.37 m

    Exit Door

    0.76 m x 1.91 m

    2.84 m

    24.5 m

    20.5 m

    2.22 m

    0.72 m

    2.49m

    1.17m

    1.44 m

    Seat Pair

    0.75 m x 1.0 m

    Window Pairs

    0.60 m x 1.37 m

    Exit Door

    0.76 m x 1.91 m

    2.84 m

    24.5 m

    20.5 m

    2.22 m

    0.72 m

    2.49m

    1.17m

    1.44 m

    Seat Pair

    0.75 m x 1.0 m

  • 8/22/2019 Railcar Fire Size A

    5/11

    Fourth International Symposium on Tunnel Safety and Security, Frankfurt am Main, Germany, March 17-19, 2010

    499

    curves, which influences the size of the initiating fire needed to flashover the railcar. An arson fire

    such as a flammable liquid spill fire might have a rapid fire growth, but burn for only a short duration.

    An accidental fire such as a trash bag fire or carryon luggage would have a slower growth rate, but will

    burn for a much longer duration.

    For each fire type there is a minimum fire size that leads to flashover. Figure 3 shows a flammableliquid arson spill that is not able to flashover the car. Figure 4 shows a slightly larger fire that flashes

    over the railcar in about a minute. From this information, a determination of the creditability of the

    scenario and determine the speed at which flashover will occur.

    The post-flashover fire model was used to predict the gas temperatures, ventilation flow rates, window

    failure, and heat release rate of fires inside of the intercity railcar. Heat release rate curves from the

    flame spread model were used to determine the fire growth rate prior to flashover. Once the

    compartment flashed over, the post flashover model predicts the heat release rate curve. Inputs for this

    model revolve around the materials and the ventilation. Most of the time the materials are given, but

    the model can be used to help select materials if a certain heat release rate is needed.

    The ventilation, including the initial ventilation and window failure times affect the peak heat release

    rate and the time at which it occurs. This effect of ventilation has been demonstrated in scale model firetests [5] which confirmed that the spread and size of the fire inside the railcar was mainly controlled by

    ventilation.

    The windows typically fail after flashover creating significant increases in the ventilation.

    Unfortunately there can be a large variation in window failure times given small changes in the type of

    the window. Test data was used to determine window fallout times. The test configuration consisted of

    0.050 m thick polycarbonate windows 0.60 m high and 1.37 m wide exposed to a line fire that

    produced a heat flux of 25-30 kW/m2 [4]. Window fallout times took approximately 6 minutes if the

    entire window was constructed of a single sheet of polycarbonate. If the windows were made of two

    smaller (0.60 m high and 0.68 m wide) sheets of polycarbonate reinforced in the center of the window,

    the window failure time doubled to around 12 minutes. These results were used to deduce window

    failure criteria in terms of the back face temperature at window failure. The window failure criteria for

    glass windows would differ from the polycarbonate window results.

    To evaluate the effects of the ventilation on a railcar, the model was used to evaluate the impact of the

    number of doors initially open (one door or two) and the time that the polycarbonate windows fallout.

    Figures 5 and 6 show the heat release rate and gas temperatures for a fire in the railcar with one door

    open. The results show that the delay in the ventilation has a significant effect on the heat release rates

    and temperatures in the compartment. The reasons for this can be seen in Figures 7 and 8, which show

    the remaining masses of the various interior materials. The majority of these materials burn away

    around the time the smaller windows fail, limiting the heat release rate of the railcar.

    Initial openings also have an effect on the heat release rate and temperatures in the compartment. For afire in the railcar with two doors open the heat release rate is in the 15-20 MW range with peaks to 35

    MW. If the doors remain closed, heat release rates in the 5 MW range have been calculated. The

    differences in the heat release rate can clearly be seen when contrasted with the one door open case

    (Figure 5).

    Comparison of Data

    There is a limited amount of heat release rate data on fully developed fires inside of actual railcars. Most of

    this testing has focused on railcar fires located inside tunnels and has been used to support tunnel design

    projects [20-23]. A detailed description of the tests is provided in Ref. [23]. Several different variations of

    railcars were evaluated including an aluminum subway railcar (18.0 m long, 2.8 m high, 3.0 m wide) and

    two steel intercity railcars (26.1 m long, 2.4 m high, 2.9 m wide). The subway railcar and one intercity

    railcar (IC-train) contained older interior finish materials, while newer interior finish materials werecontained in the other intercity railcar (ICE-train). All of the tests were run with the doors closed and only

  • 8/22/2019 Railcar Fire Size A

    6/11

    Fourth International Symposium on Tunnel Safety and Security, Frankfurt am Main, Germany, March 17-19, 2010

    500

    0

    500

    1000

    1500

    2000

    2500

    0 200 400 600 800 1000

    Time (sec)

    HeatReleaseR

    ate

    0

    100

    200

    300

    400

    500

    600

    Temperature(C)

    Total

    Source

    Wall

    Ceiling

    Temperature

    Figure 3. Heat Release Rate and Temperature for Fuel Spill with no flashover of the railcar.

    0

    500

    1000

    1500

    2000

    2500

    3000

    3500

    4000

    0 10 20 30 40 50 60 70

    Time (sec)

    HeatReleaseRate(kw)

    0

    100

    200

    300

    400

    500

    600

    700

    800

    Temperature(C)

    Total

    Source

    Wall

    Ceiling

    Temperature

    Figure 4. Heat Release Rate and Temperature for a Fuel Spill causing flashover of the railcar.

    Time After Flashover (min)

    0 5 10 15 20 25 30

    HeatReleaseRa

    te(MW)

    0

    5

    10

    15

    20

    25

    30

    35

    40

    45

    12 min. Window Fallout

    6 min. Window Fallout

    Windows Fallout

    Windows Fallout

    Figure 5. Heat Release Rate for One Door Open with Large Windows (6 min fallout) and Reinforced

    Windows (12 min fallout).

  • 8/22/2019 Railcar Fire Size A

    7/11

    Fourth International Symposium on Tunnel Safety and Security, Frankfurt am Main, Germany, March 17-19, 2010

    501

    Time After Flashover (min)

    0 5 10 15 20 25 30

    GasTemperature(oC)

    400

    500

    600

    700

    800

    900

    1000

    1100

    1200

    12 min. Window Fallout

    6 min. Window Fallout

    Figure 6. Temperature for One Door Open. with Large Windows (6 min fallout) and Reinforced

    Windows (12 min fallout).

    Time After Flashover (min)

    0 5 10 15 20 25 30

    Mass(kg)

    0

    50

    100

    150

    200

    250Seat

    Seat Shroud

    Floor Carpet

    Wall Carpet

    Window Mask

    Window

    Wall Lining

    Ceiling Lining

    Window Drape

    Figure 7. Mass of Interior Material for One Door Open with Large Windows (6 min fallout).

    Time After Flashover (min)

    0 5 10 15 20 25 30

    Mass(kg)

    0

    50

    100

    150

    200

    250

    Seat

    Seat Shroud

    Floor Carpet

    Wall Carpet

    Window Mask

    Window

    Wall Lining

    Ceiling Lining

    Window Drape

    Figure 8. Mass of Interior Material for One Door Open with Smaller Windows (12 min fallout).

  • 8/22/2019 Railcar Fire Size A

    8/11

    Fourth International Symposium on Tunnel Safety and Security, Frankfurt am Main, Germany, March 17-19, 2010

    502

    one window open. The windows were made of glass. During the subway railcar test, the aluminum skin

    melted, which increase the ventilation into the railcar. The subway railcar fire had a peak heat release

    rate of 35 MW, while the longer intercity railcars had peak heat release rates of 13-20 MW. A plot of

    the heat release rates measured in these tests can be seen in Figure 9.

    A direct quantitative comparison could not be made between the results of the sampleHAIFGMRail/HAICFMRail analysis and the test data due to limited information on the interior finish

    materials (including windows) denoted in the test reports and limited information on the ventilation

    conditions during the test. A qualitative comparison can be made however. The peak heat release

    rates, between the model and the tests, are in the same range as each other. The largest difference is in

    the time it takes to reach the peak and the duration of the fire. HAIFGMRail/HAICFMRail predicts

    increases in heat release rate (2-4 minutes), while the testing time to peak varied between 5 and 20

    minutes. The long duration, low severity fires seen in the intercity cars are indicative of the limited

    ventilation available as evidenced from the 35 MW peak of the subway car when the roof vented.

    Duration is dependant on ventilation which is discussed above.

    Some of the difference in the heat release rate histories can be attributed to test conditions. It was

    decided to analyze a subway style railcar in addition to the intercity analysis presented above tofacilitate a more even comparison with what was tested. The results of the subway railcar analysis are

    shown in Figure 10. The subway railcar in the tests is similar to the railcar used in the sample analysis.

    This test shows rapid increases in the heat release rate, with a peak heat release rate of 35 MW reached

    5 minutes after ignition, which is similar to the sample subway railcar analysis.

    The intercity railcars that were tested show some inconsistencies with other tests. In another series of

    tests [24], a different intercity railcar showed results that were more inline with the subway rail cars.

    Flashover conditions were measured inside within 140 seconds and full involvement of all materials in

    the car was observed at 175 seconds. While the heat release rate in this particular test (Ref. [24]) was

    not recorded, the times to flashover were indicative of the rapid heat release rate predicted in the

    sample analysis.

    In the intercity fire tests [23] ventilation conditions were constantly changing. Windows were heard

    breaking as early as 2 minutes into the test and as late as 42 minutes although exact window breakage

    times were difficult to determine from the test reports. The limited ventilation along with burnout

    limited the peak heat release rates and extended the duration of the fire.

    Some of the differences in heat release rates may be attributed to differences in materials used.

    Subsequent modeling conducted by Hughes on different versions of the same railcar suggests that

    newer railcars may produce heat release rates substantially higher than older railcars. There is a general

    trend to replace metals with plastic composites and glass with polycarbonate. These represent real

    concerns with new car designs with respect to the design fire size. The intercity railcar tests were

    conducted on older railcars, which may explain differences between the results.

    The comparison of the sample HAIFGMRail/HAICFMRail modeling to test data and other modeling

    efforts show similarities and differences. All of the comparisons show peak heat release rates in the

    same range or higher compared to the sample modeling results. Differences are seen in the time to peak

    and duration, but these differences can be attributed to differences in the car, initiating source and

    ventilation conditions during the test.

  • 8/22/2019 Railcar Fire Size A

    9/11

    Fourth International Symposium on Tunnel Safety and Security, Frankfurt am Main, Germany, March 17-19, 2010

    503

    Time [min]

    0 20 40 60 80 100 120 140

    HeatReleaseRate

    [MW]

    0

    5

    10

    15

    20

    25

    30

    35

    40

    Intercity (IC-train)

    Intercity (ICE-train)

    Subway (Aluminum)

    Figure 9. Large-scale railcar test data [4]

    Time After Flashover (min)

    0 5 10 15 20 25 30

    H

    eatReleaseRate(MW)

    0

    5

    10

    15

    20

    25

    30

    Windows Fallout

    Figure 10. HAICFMRail Compartment Modeling of Subway Railcar

    CONCLUSION

    In conclusion, the methodology presented here provides a reasonable alternative to full scale railcar

    testing. The value of this modeling comes from the wide range of fire scenarios and ventilation

    conditions that can be evaluated so that a suitably conservative design basis fire can be selected. This

    modeling also has the ability to assess the contribution of new materials on car performance and to use

    the modeling in the material selection process.

    The modeling indicates that fully-developed fires inside of railcars are dependent on the fire properties

    of interior finish materials, the surface area and combustible mass of fuel inside of the railcar, and the

    ventilation conditions into the railcar. Changes in ventilation, such as window failure, can result in

    large increases in heat release rate.

    Comparisons of the modeling results to full scale testing show both similarities and differences. The

    differences are attributed to insufficient information on tested railcar construction, ventilationconditions during the tests, and no material fire property data. Future large-scale test programs on

  • 8/22/2019 Railcar Fire Size A

    10/11

    Fourth International Symposium on Tunnel Safety and Security, Frankfurt am Main, Germany, March 17-19, 2010

    504

    railcars need to report surface area of each interior finish material, initial ventilation opening area, and

    occurrence of window fallout or other ventilation path development. Cone calorimeter test data on

    interior finish materials should also be reported. Prior to testing, simulations should also be conducted

    to determine the ventilation conditions that will result in the worst-case heat release rate for the railcar.

    REFERENCES[1] Lattimer, B., Hunt, S., Wright, M., and Beyler, C.," Corner Fire Growth in a Room withCombustible Lining." Fire Safety Science - Proceedings of the Seventh International

    Symposium Worcester, Massachusetts, 2002, pp. 12

    [2] Lattimer, B, Beyler, C, Heat Release Rates of Fully-developed Fires in Railcars Fire SafetyScience- Proceedings of the Eighth International Symposium, 18-23 September, 2005,

    Beijing, China, International Association for Fire Safety Science, 2005, pp. 1169-1180.

    [3] ASTM E 2061-03, Standard Guide for Fire Hazard Assessment of Rail TransportationVehicles, ASTM International, West Conshohocken, PA, 2003.

    [4] Strege, S., Lattimer, B., and Beyler C., Fire Induced Failure of Polycarbonate Windows inRailcars, Fire and Materials 2003, 2003, pp. 269278.

    [5] Ingason, H, Model scale railcar fire tests Fire Safety Journal, Vol 42, Elsevier, 2007, pp.271282.

    [6] Williams, F.W., Hunt, S.P., Beyler, C.L., and Iqbal, N. (1996), Upward Flame Spread onVertical Surfaces, NRL Ltr Rpt 3902 Ser 6180/0065.1, Naval Research Laboratory,

    Washington, DC, March 8, 1996.

    [7] Beyler, C.L., Hunt, S.P., Iqbal, N., and Williams, F.W. (1996), Upward Flame Spread onVertical Surfaces, Thirteenth Meeting of the U.S./Japan Government Cooperative Program on

    Natural Resources (UJNR) Panel on Fire Research and Safety, NISTIR 6030, 1, National

    Institute of Standards and Technology, Gaithersburg, MD, March 1320, 1996.

    [8]

    Williams, F.W., Beyler, C.L., Hunt, S.P., and Iqbal, N. (1997), Upward Flame Spread onVertical Surfaces, NRL/MR/6180-9-7908, Naval Research Laboratory, Washington, DC,

    January 13, 1997.

    [9] Beyler, C.L., Hunt, S.P., Iqbal, N., and Williams, F. (1997), A Computer Model of UpwardFlame Spread on Vertical Surfaces,Proceedings of the Fifth International Symposium on

    Fire Safety Science, Melbourne, Australia, pp. 297308, 1997.

    [10] Lattimer, B.Y., Hunt, S.P., Wright, M., and Sorathia, U. (2003), Modeling Fire Growth in aCombustible Corner,Fire Safety Journal, 38 (8), December 2003.

    [11] Walton, W.D. and Thomas, P.H. (2002), Estimating Temperatures in Compartment Fires,Sections 36, The SFPE Handbook of Fire Protection Engineering, 3

    rdEdition, P.J. DiNenno,

    Editor-in-Chief, National Fire Protection Association, Quincy, MA, 2002.

    [12] Karlsson, B. and Magnusson, S.E. (1991), Combustible Wall Lining Materials: NumericalSimulation of Room Fire Growth and the Outline of a Reliability based Classification

    Procedure,Proceedings of the 3rd

    International Symposium of Fire Safety Safety Science,

    Elsevier Applied Science, London, United Kingdom, 1991.

    [13] Steckler, K., Quintiere, J., and Baum, H. (1982), Flow Induced by a Fire in a Compartment,NBSIR 82-2520, Washington, DC, p 93, 1982.

    [14] Jones, W.W., Forney, G.P., Peacock, R.D., and Reneke, P.A, A Technical Reference forCFAST: An Engineering Tool for Estimating Fire and Smoke Transport, NIST TN 1431,

    National Institute of Standards and Technology, Gaithersburg, MD, April 2003.

  • 8/22/2019 Railcar Fire Size A

    11/11

    Fourth International Symposium on Tunnel Safety and Security, Frankfurt am Main, Germany, March 17-19, 2010

    505

    [15] McGrattan, K. and Forney, G., Fire Dynamics Simulator (Version 4) Users Guide, NISTTechnical Special Publication 1019, National Institute of Standards and Technology,

    Gaithersburg, MD, February 2005.

    [16] Drysdale, D.,An Introduction to Fire Dynamics, John Wiley and Sons, West Sussex, England,1998.

    [17] Floyd, J.E, Hunt, S.P., Williams, F.W., and Tatem, P.A., A Network Fire Model for theSimulation of Fire Growth and Smoke Spread in Multiple Compartments with Complex

    Ventilation,Journal of Fire Protection Engineering, 15 (3), August 2005, pp. 199229.

    [18] Hurley, M.J., Evaluation of Models of Fully Developed Post-flashover Compartment Fires,Journal of Fire Protection Engineering, 15 (3), August 2005, pp. 175197.

    [19] Yii, E.H., Buchanan, A.H., and Fleischmann, C.M. Simulating The Effects of Fuel Type andGeometry on Post-flashover Fire Temperatures,Fire Safety Journal, 41, 2006,

    pp. 6275.

    [20] Ingason, H., Design Fires in Tunnels,ASIAFLAM 95, Interscience Communications, 1995,pp.77-86.

    [21] Steinert, C., Smoke and Heat Production in Tunnel Fires,International Conference on Fires inTunnels, Swedish National Testing and Research Institute, 1994, pp. 104-122.

    [22] Blume, G., Smoke and Heat Production in Tunnel Fires Smoke and Hot Gas Hazards,International Conference on Fires in Tunnels, Swedish National Testing and Research Institute,

    1994, pp. 138-146.

    [23] EUREKA 1995, Fires in Transport Tunnels Report on Full-Scale Tests, EUREKA ProjectEU 499:FIRETUN, Studiengesellschaft Stahlanwendung e.V., D-40213 Dusseldorf, 1995.

    [24] White, N, Dowling, V, Barnett, J, Full-Scale Fire Experiment on a Typical Passenger TrainFire Safety Science- Proceedings of the Eighth International Symposium, 18-23 September,

    2005, Beijing, China, International Association for Fire Safety Science, 2005, pp1157-1168.