A new approach to determining passenger rail vehicle design fires.
As the United States looks to add high speed rail and enhance commuter rail and subway systems, protection of passenger rail infrastructure is more critical than ever. Designing for CBR threats, bombs, and active shooters have been a focus over recent years to meet security goals; however, fires remain a significant threat to rail infrastructure and the traveling public. Consider the following events:
* 11 November 2000: fire in the ascending car of the Kitzsteinhorn funicular in Kaprun, Austria, resulted in the deaths of 155 persons (Schupfer, 2001);
* 18 February 2003: a subway fire in Daegu, Republic of Korea, resulted in 192 deaths and 148 injuries (NEMA, 2011)
One of the most important input parameters to life safety and emergency ventilation system designs to mitigate catastrophic events in rail tunnels and stations is the design fire. While there are a wide range of fire scenarios that can occur within these facilities, one of the more significant hazards is an interior passenger rail vehicle fire. How large the fire may become is dependent upon the initiation fire, vehicle interior linings and contents, vehicle configuration and ventilation.
Full scale tests can be used to evaluate fire hazards represented by different interior lining materials, contents and ventilation scenarios, when exposed to various initiation fire scenarios. However, full scale tests are costly and time-consuming. Small-scale fire testing can offer a cost- and time-effective solution. Through the application of a screening tool, it is possible to quickly assess a material's propensity to spread flames over a range of initiation fires. Additionally, small-scale test data can be used to estimate material pyrolysis and combustion properties for use as input in models with flame spread capabilities. Through careful model calibration and an in depth understating of a given model's limitations, such tools, which may include computational fluid dynamics (CFD) programs, provide techniques to predict flame spread and characterize overall fire development based on small-scale data alone.
This paper presents a method to use small scale testing to estimate fire hazards in passenger rail vehicles. The method combines: (1) characterization of postulated initiation fires in terms of thermal insult, (2) a simplified approach for predicting potential for flame spread based on small scale testing, and (3) hazard-based fire and flame spread modeling to characterize design fire scenarios. The methodology lends itself to supporting threat, risk and vulnerability assessments of transit systems. A risk-based framework can also allow owners and operators to mitigate exposure through the vehicle design process by defining specific threats or initiating events to which the vehicle ought to be resistant. This paper intends to overview the process; details of the method, implementation of the method, or combining the method with risk techniques are given elsewhere (Meacham et al., 2010, 2011a, 2011b).
The Department of Homeland Security (DHS) outlines an approach which considers risk to be function of consequence, vulnerability, and threat (DHS, 2009). Meacham et al (2011a) describes this approach for passage rail vehicles:
Threats. As used here the threat is fire, or more correctly an initiating fire. For purposes of characterizing the threat, fire in a passenger rail vehicle can be accidental or deliberate. Passenger rail systems experience accidental fires due to ignition from equipment failure within the train (e.g., locomotive power systems failure, equipment failure in the internal electrical systems, lighting systems, heating systems, brake systems, etc), from failure in the system infrastructure (e.g., power systems, station related hazards), and from external exposure fires (e.g., trash fires, brush fires). Deliberate actions such arson are also a significant threat to passenger rail systems.
Vulnerability. There are several dimensions of vulnerability. With respect to deliberate events, the vulnerability exists due to the generally open nature of the systems, particularly subways and commuter rail systems, with limited capacity to screen all passengers for security threats. Rail vehicles are also vulnerable in the sense that they use combustible materials, which can be ignited and contribute to the magnitude of the fire event and the extent of resulting damage. Passengers in the vehicles and stations are vulnerable due to the limited means of escape.
Consequences. Fires can remain small, contained to the initial item burning, or grow to encompass all vehicles in the train. The ultimate size (energy, intensity) of a fire will be influenced by the size of the initiation fire, the type, amount and characteristics of fuel available to burn within the vehicle, the size and location of ventilation openings, and the compartment (vehicle) configuration. Factors such as the effects of adjacent materials on fire spread, interaction of the fire with the geometric elements of the car, and the potential for flame spread away from the fire origin must be considered. Fire can result in significant damage to a passenger rail vehicle, its occupants, the occupants of a station or surrounding area, and to physical infrastructure such as rails, bridges, tunnels, catenary, cable and other services. The extent of damage in terms of life safety, physical infrastructure and operations can be catastrophic.
Risk. Overall fire risk includes characterization and assessment of potential accidental and deliberate initiation events, fire-related vulnerabilities, and the consequences of the fire events. As noted in the NRC review (NRC, 2010), many risk assessment methods exist; selection of a suitable approach should be based on the availability and reliability of data.
Figure 1 summarizes the proposed approach for evaluating the risk - based on the threats to and vulnerability of material contents - associated with rolling stock. In such an approach, rolling stock is a part of the overall transit safety system where decisions pertaining to vehicle design can help mitigate fire risks that can impact the entire transit system. A discussion of this method follows; this discussion uses the risk framework described by Meacham et al (2011a) to provide context for the method.
Hazard Assessment * Failure/hazard analysis of vehicle systems * Review of fire incident history * Identification of "Important" initiating scenarios Initiating Event * Literature review or Characterization intermediate-scale testing to define initiating scenarios in terms of heat release rate * Calculate incident flux * Determine flux-time product Component Material Assessment * Small-scale (cone calorimeter testing) of component materials at multiple incident fluxes * Estimate critical heat flux & flux-time product * Calculate flame spread ("b") parameter Material Ignition * Compare material critical heat Vulnerability flux to maximum initiating event heat flux * Compare material and initiating event flux-time product Material Flame Spread * Identify sign of material "b" Propensity parameter at initiating event incident flux * Screen or select materials based on "b" (among other criteria Initial Flame Spread * Utilize simplified flame spread Consequence Evaluation algorithm to evaluate early fire growth * Estimate likelihood of accelerating flame spread Event Consequence Analysis * Estimate heat release rate history from simplified algorithm; or * Utilize alternate techniques [post-flashover models, CFD models! to estimate heat release rate Determine Risk * Calculate risk associated with a given initiating event based on probabilities and appropriate consequence metrics Figure 1: Proposed rail vehicle fire assessment methodology.
Threat: Initiation Fires Characterization
Different initiating fires will produce a range of different outcomes. For example, a small trash fire may only spread along a seat, up the adjacent wall, and then self-extinguish. Alternatively, a gasoline fire from an arson event could be large enough to cause full vehicle involvement. The ultimate fire size associated with these fuels is an important factor in determining if the initiating fire will lead to involvement of interior lining materials. Scenarios of interest in a passenger rail vehicle setting tend to include trash bags, suitcases, vehicle seats (Babrauskas, 2008), and flammable liquids.
After defining the size of the initiation fires, it is important to then be able to characterize the fires in order to assess the likelihood of fire spread. In general, issues of concern are intensity (temperature and radiant heat flux) and duration of fire in relation to the material which may next become ignited.
The cone calorimeter is a standardized testing apparatus (ASTM, 1997; ISO, 2001) that uses oxygen consumption calorimetry to measure the heat release rate of a small-scale material specimen. By measuring the concentration of oxygen, carbon monoxide, and carbon dioxide within the apparatus flue gases, it is possible to estimate the heat release rate of the burning specimen (Huggett, 1980; Janssens, 2008). The involve exposing a specimen measuring 0.1 m by 0.1 m (~0.01 m2) to a constant external radiant flux through the cone-shaped heating element. This surface heat flux can be can be varied up to 100 kW/[m.sup.2] allowing for fire performance and burning characteristics, such as heat release rate per unit area, smoke and species production, and time to ignition to be evaluated under different heat flux conditions (Babrauskas, 2008).
By measuring time to ignition at varying heat fluxes, it is possible to estimate the minimum heat flux at which ignition of the material may occur. The thermal exposure generated within the cone calorimeter can then be defined as the product of the time to ignition and the incident flux less the material's critical heat flux: ([q.sub.e]" - CHF) x [t.sub.ig] With knowledge of the critical heat flux and thermal exposure experienced during testing, one can then compare against the thermal exposure generated by a given initiation fire to assess whether the initiation fire is an adequate ignition source to involve. If the thermal exposure from an initiation fire exceeds that developed during a cone calorimeter test, ignition and subsequent involvement of the material would be expected; if less, then material ignition would not be expected.
Vulnerability: Flame Spread
While susceptibility to ignition of a material by a given initiation fire is a key indicator of risk, it does not fully define the vulnerability of a material or an entire vehicle to a particular hazard. Depending on the victim material's combustion properties and the strength of the initiation fire, the ensuing fire could either decelerate or accelerate. For a decelerating fire, the burn-out rate exceeds the flame spread rate and the amount of material actively involved in combustion decreases with time, often leading (self) extinguishment. In such a fire, the amount of material that becomes involved and the extent of damage can be limited. For accelerating fires, the spread rate is faster than the burnout rate resulting in more and more material becoming actively involved in combustion. In this scenario, the resulting fire can become very large and potentially lead to a critical event such as flashover, in which most/all contents of the vehicle are simultaneously actively burning.
Determining which materials under certain exposure conditions would result in accelerating flame spread is important in evaluating material use in rolling stock. A first analysis screening approach based on Quintiere's (Cleary and Quintiere, 1991) flame spread parameter is proposed to identify which materials are likely to support flame spread at various heat fluxes. This initial screening process quickly identifies a material's relative fire performance without extensive fire testing. The equation, b = 0.01 E" - 1 - [t.sub.ig] / [t.sub.end], utilizes fire parameters measured or derived from cone calorimeter results: average heat release rate, time to ignition, and total burn duration. The b-parameter predicts a material's tendency to accelerate or decelerate flame spread at various heat fluxes. Flame spread is considered to accelerate if the b-parameter is greater than zero. For values less than zero, flame spread is considered to decelerate such that it could eventually self-extinguish. Figure 2 demonstrates the dependence of the b-parameter on incident flux for a number of typical ceiling (a) or (b) wall lining materials. The b-parameter is typically directly proportional to incident heat flux indicating that materials will tend to support accelerating flame spread at higher incident fluxes.
[FIGURE 2 OMITTED]
The b-parameter can be used in at least two ways: (1) as a screening tool to inform decisions about material selections, and (2) as a first-order hazard assessment. For the former, consider two wall lining materials in Figure 3(b) - the fiber reinforced plastic (FRP) and the painted phenolic FRP. At incident fluxes less than 35 kW/[m.sup.2], decelerating flame spread would be predicted for both materials. At incident fluxes greater than 50 kW/[m.sup.2], the b-parameter for the FRP gelcoat turns positive, suggesting the material will support accelerating flame spread. At similar incident heat fluxes, the painted phenolic consistently has lower b-parameter values; at 50 kW/[m.sup.2] this is still negative. This means that under similar thermal exposure conditions the painted phenolic is less likely to support accelerating flame spread than the FRP. As such, material selections could be made based on relative values of the b-parameter. As a first-order hazard assessment tool, the b-parameter can crudely predict material response to certain initiating fire events by comparing the incident heat flux from an initiating fire event (a function of heat release rate, flame height, view factor, and flame and material optical properties) to that utilized in the cone calorimeter experiments. In this way, the b-parameter can be used for rapidly evaluating consequences of potential initiation fire scenarios: specifically decelerating or accelerating flame spread in response to a certain initiation fire.
[FIGURE 3 OMITTED]
Consequence: Initial Flame Spread Assessment
A simplified upward flame spread model based on the work of Mowrer and Williamson (1991) has been developed to represent the initial flame spread that might occur in passenger rail vehicles (Meacham et al., 2010; 2011a, 2011b). Upward flame spread is a complex phenomenon which involves thermal exposure from an initiation fire to a surface (e.g., a wall), sufficient heating of the wall material to release combustible gases (pyrolysis) and ignite them, and the ability of the material to continue to burn in the absence of an initiation fire by creating a sufficiently large and hot flame to propagate the pyrolysis process. Figure 3 illustrates the characteristics of upward flame spread as adapted from Mowrer and Williamson (1991). The model uses parameters measured in small-scale tests and includes expressions to demonstrate how flame height, pyrolysis height, and burnout height relate to one another and change over time. The model assumes one-dimensional flame spread along the wall. The overall flame height, [x.sub.f], consists of the flame created on the wall by the external source fire and the flame extension up the wall. The pyrolysis height is represented by [x.sub.p]. When fuel is consumed, it no longer supports a flame and a burnout front develops, indicated by [x.sub.b]. The initiation fire imposes a heat flux on the vertical wall. After the first exposed material is ignited, the flame extends up the wall, emitting a flame heat flux to the unburned fuel above.
The flame spread model uses the b-parameter and heat flux levels to assess the flame zone (xf) and pyrolysis zone (xp) to determine if sufficient burning will occur, in the absence of the initiation fire, to have the pyrolysis zone accelerate faster than the burnout zone (xb), leading to self-propagating flame spread (Meacham et al., 2010, 2011a, 2011b). Details regarding the flame spread model, and how the outcomes of the model can be incorporated into a CFD model to assess how large a passenger rail vehicle fire might grow, and the potential impacts of that fire on life safety and critical infrastructure, are presented elsewhere (Meacham et al., 2010; 2011a, 2011b).
Consequence: Design Fire and Outcomes Modeling
Thus far the methodologies described address individual components and their vulnerability to defined initiation fires. These tools address material ignition and "early" fire development within the vehicle. However, for emergency ventilation system design it is necessary to understand the ultimate outcome of a fire scenario in order to define a design fire criterion. This requires the ability to predict the involvement and interaction of multiple factors in the course of fire development including vehicle material contents, ventilation conditions, among others. Numerous methodologies have been used throughout the years for estimating rail vehicle design fires including (1) the summation of fuel loads divided by some selected, arbitrary burn time; (2) the area-scaling of cone calorimeter test results to the entire vehicle interior area; (3) relatively simple post-flashover, single-zone or heat-balance models.
With exponential increases in computing power and storage and the release of Fire Dynamics Simulator (FDS), computational fluid dynamics (CFD) based fire modeling has been used extensively worldwide for both building design and post-fire investigation. Although using CFD to model heat and smoke transport in buildings is now standard practice in the fire community, fire modeling to predict surface flame spread has not yet been widely applied. Modeling flame spread between different materials and objects, and the resulting growth in fire hazard is challenging due to its strong dependence on material thermal properties, model geometry, and computational mesh size. Yet, there is substantial and growing interest in using first-principles computer modeling to predict real-scale fire behavior from small-scale fire test data.
While FDS contains the basic physics necessary for fire growth modeling, the determination of the numerous material properties for use in FDS remains a challenge. There are no simple or direct means for determining all of the parameters required for use in the pyrolysis models used in FDS or other CFD codes. The current trend is to utilize optimization techniques to estimate values of each of the parameters. The approach is heuristic and work is ongoing to understand and quantify uncertainties, which compound uncertainties already associated with FDS modeling, and CFD modeling in general.
The current work provides means for bringing more confidence to fire growth modeling results. The b-parameter and simplified flame spread model can be used as a pre-cursor to more detailed fire growth modeling in a number of ways. The simplified model can potentially be used to determine if and/when a critical event such as flashover may occur, marking the point at which an alternate technique such as a post-flashover energy balance model could be applied to determine the maximum or peak heat release rate potential of a vehicle. Alternatively, the results of the simplified model can be translated as input to a CFD model such that the initial heat release rate of the CFD model is sufficiently large to be insensitive to many model grid resolution issues. The proposed approaches are outlined here.
The two alternative approaches - utilizing the output of the simplified model to initialize the model or conducting intermediate-scale calibration tests - are proposed to address uncertainties in the process that are not yet fully understood or quantified. Once the CFD model has been appropriately initialized and/or calibrated, the model can be used to explore the outcome of numerous initiating fire events in order to either establish the risk profile for the vehicle or establish design criterion for system-wide components, such as tunnel ventilation systems. Figure 4 shows an exemplar rendering from an FDS model that utilized the testing calibration approach. The rendering demonstrates the geometric complexity that can be incorporated into a CFD model. Figure 5 shows exemplar heat release results from FDS models utilizing the specified burner approach; the results show how material selections can affect the net consequence of a fire scenario event.
[FIGURE 4 OMITTED]
[FIGURE 5 OMITTED]
Risk: Calculate Risk of Outcomes
Through historic data, the probability of a given initiation fire can be assessed. Risk techniques can then be used to determine the likelihood of a given design fire size. NRC (2010) provides many risk assessment methods exist, and selection of a suitable approach should be based on the availability and reliability of data and method. Meacham et al. (2011a) provide detail on risk techniques for assessing passenger rail vehicles.
This paper proposes a methodology to screen materials, to determine flame spread, to assess fire growth, and to determine order-of-magnitude fire sizes given an initiation fire. The method has a range of potential applications, including the following (Meacham et al, 2010):
* Fire hazard assessment of existing stock. This approach provides insight into what size initiation fire might lead to full vehicle involvement. This would be foundational to a threat, vulnerability and risk assessment (TVRA) of vehicles to understand risks and impacts of fires to passengers and other vehicles.
* Fire hazard assessment of critical infrastructure. The methodology can be applied to help understand the resultant fire size from a fully-involved vehicle. This could be used as part of a threat, vulnerability and risk assessment (TVRA) in stations, tunnels and other rail infrastructure and to support assessment of mitigation options.
* Options analysis for new vehicle design. The methodology can be applied to help assess the performance of different interior lining materials at the vehicle design stage. This information can inform judgments about material options and use in rail vehicles.
* Regulatory support. Current regulations generally require testing of material flame spread at a single heat flux level (NASFM, 2008). As demonstrated by this research; however, a material may perform differently at different heat flux levels: resisting ignition and self-propagating flame spread at lower levels, but igniting and self- propagating flame at higher heat flux levels. Based on the level of risk or hazard deemed tolerable from a regulatory perspective, this methodology could be used as support for modifying material test requirements, such as requiring a range of incident heat fluxes and reporting outcomes. This could lead to new or more robust testing which aims to develop an understanding of how materials perform as part of regulatory benchmarking.
Portions of the material presented in this paper are based upon work supported by the Science & Technology Directorate, U.S. Department of Homeland Security, under Award Number 2009-ST-108-000013.
The views and conclusions contained in this document are those of the authors and should not be interpreted as necessarily representing the official policies, either expressed or implied, of the U.S. Department of Homeland Security.
American Society for Testing and Materials (ASTM). 1997. Standard Test Method for Heat and Visible Smoke Release Rates for Materials and Products Using and Oxygen Consumption Calorimeter. ASTM E-1354. Annual Book of ASTM Standards. Volume 07.07.
Babrauskas, V. 2008. "Heat Release Rates," in SFPE Handbook of Fire Protection Engineering, Fourth Edition. Section 3, Chapter 1. Quincy, MA: National Fire Protection Association.
Cleary, T., Quintiere, J. 1991. "A Framework for Utilizing Fire Property Tests." Third International Symposium on Fire Safety Science, 647-656.
Department of Homeland Security 2010. "DHS Risk Lexicon - Risk Steering Committee" Department of Homeland Security.
Huggett, C. 1980. "Estimation of the Rate of Heat Release by Means of Oxygen Consumption," Fire and Materials 12: 61-65, 1980.
ISO 5660-2. 2002. Reaction to Fire Tests - Heat Release, Smoke Production, and Mass Loss Rate Standard. International Organization for Standardization, Geneva, Switzerland.
Janssens, M. 2008. "Calorimetry," Section 3, Chapter 2, in DiNenno, P., Ed., SFPE Handbook of Fire Protection Engineering, Fourth Edition, NFPA, Quincy, MA, pp3-60 to 3-89.
Meacham, B.J., Dembsey, N.A., Schebel, K., Johann, M., and Tubbs, J. 2010. Rail Vehicle Fire Hazard Guidance, Final Report, Developed under DHS Grant 2009-ST-108-000013, Department of Homeland Security, Science and Technology Directorate, International Cooperative Programs Office, Washington, DC.
Meacham, B.J., Dembsey, N.A., Johann, M., Schebel, K. and Tubbs, J. 2011a. "Use of Small-Scale Test Data to Enhance Fire-Related Threat, Vulnerability, Consequence and Risk Assessment for Passenger Rail Vehicles," Journal of Homeland Security and Emergency Management, (Manuscript 1850, under review).
Meacham, B.J., Dembsey, N.A., Tubbs, J., Johann, M. and Schebel, K. 2011b. "A Simplified Approach for Assessing Initial Fire Development and Spread in Passenger Rail Vehicles," Transportation Research Record, (Manuscript 11-1513, accepted February 18, 2011, in press, expected publication December 2011).
Mowrer, F. W., & Williamson, R. B. 1991. "Flame Spread Evaluation for Thin Interior Finish Materials." Fire Safety Science - Proceedings of the Third International Symposium. England: Elsevier Science Publishers LTD, pp689-698
National Association of State Fire Marshals (NASFM). 2008. "Recommended Fire Safety Practices for Rail Transit Materials Selection." Prepared for the U.S. Department of Transportation, Federal Transit Administration, Office of Safety and Security.
National Emergency Management Agency (NEMA) 2011. "Daegu Subway," National Emergency Management Agency, South Korea, July 2011. <http://eng.nema.go.kr/sub/cms3/3_4.asp>.
Schupfer, H., 2001. "Fire Disaster in the Tunnel of the Kitzsteinhorn Funicular in Kaprun on 11 Nov. 2000," in Proceedings of the Fourth International Conference, Safety in Road and Rail Tunnels, Madrid, Spain.
Jeffrey Tubbs, PE Member ASHRAE Jarrod Alston, PE Associate ASHRAE Matthew Johann, PE Brian Meacham, PhD, PE Nicholas Dembsey, PhD, P EKurt Schebel
Jeffrey Tubbs, PE, FSFPE is an Associate Principal with Arup in the Boston Office (Cambridge, MA). Jarrod Alston, PE is an Associate with Arup in the Boston Office. Matthew Johann, PE LEED AP is a Senior Engineer with Arup in the Boston Office. Brian Meacham, PhD, PE, FSFPE is an Associate Professor at Worcester Polytechnic Institute (Worcester, MA). Nicholas Dembsey, PhD, PD, FSFPE is a Professor at Worcester Polytechnic Institute. Kurt Schebel is a Fire Specialist in Arup's New York Office (New York, NY).
|Printer friendly Cite/link Email Feedback|
|Author:||Tubbs, Jeffrey; Alston, Jarrod; Johann, Matthew; Meacham, Brian; Dembsey, Nicholas; Schebel, Kurt|
|Date:||Jan 1, 2012|
|Previous Article:||NYCT R-142 train design fire scenario characteristics.|
|Next Article:||Modeling of household refrigerator performance with low global warming potential alternative refrigerants.|