Application of NFPA 130 for emergency evacuation in a mass transit station.

ABSTRACT

This paper presents an analysis of an emergency evacuation during a mid-platform train fire in a mass transit station. A time-based scenario is first developed, taking into account the inaccessibility of the stairs due to smoke obscuration. The exiting calculation approach in NFPS 130 is then applied to a parametric study including this time-based scenario, and the results are compared to a micro-simulation model. Both methods give consistent predictions of the exiting times. The results demonstrate that without the division of the station into two zones, the exiting time meets the code requirements; however, with the division, it could not meet the required exiting time. Such division is due to smoke blockage, which creates an uneven occupancy load between the zones. Therefore, appropriate consideration of the fire and smoke effects is important for emergency evacuation study. While the objective of station smoke management is to protect the evacuation routes at both ends of the station, future study would have to reconcile the available time and the required time of tenability.

INTRODUCTION

In an underground rail station, smoke generated from a fire travels up through the normal circulation elements toward the street level. During evacuation, passengers have been observed to follow the same route that they used to enter the station (Proulx 2002). If the evacuation routes were blocked by smoke, this would endanger the passengers during evacuation, as in, for example, the Daegu city subway fire in South Korea (Chien et al. 2004). Therefore, an effective smoke control mechanism is needed to protect the evacuation routes. Effective smoke control by emergency ventilation rather than other methods, such as compartmentation, has been typically designed for underground rail stations/tunnels and similar built environments. The use of emergency ventilation was demonstrated in the Channel tunnel fire accident (Comeau and Wolf 1997). Because of the unique configuration of underground mass transit stations, smoke management has been widely studied in field tests, using computational fluid dynamics (CFD) and other tools. Among others, these studies include the validation study on the London Kings Cross Station fire by Simcox et al. (1992), the CFD application for a light rail station by Zigh et al. (2000), and the study of the critical ventilation velocity in tunnel fires by Kennedy and Kang (2003).

In addition to effective smoke control, sufficient circulation capacity is needed to ensure a prompt evacuation of the station. Many concepts and computer models have been developed for emergency evacuation applications, including EXIT89 (Fahy 1999), Exodus (Galea and Perez Galparsoro 1994), Simulex (Thompson and Marchant 1995), and STEPS (Hoffmann and Henson 1997; Rhodes and Hoffmann 1999). Using the "hydraulic flow" concept, hand calculation of exiting time is also possible (Proulx 2002; Nelson and Mowrer 2002). This concept is done by assigning each circulation element a flow capacity. The total time required to clear a circulation element would be equal to the queuing time plus the travel time from the most remote point to that element. This method has been included as part of the guidelines on fixed guideway and passenger rail systems of the National Fire Protection Association (NFPA) in NFPS 130 (NFPA 2003).

While the micro-simulation evacuation models such as STEPS have included many features on human behavior and decision making, hand calculation using the hydraulic flow model is still a valuable tool in the life safety assessment. This is especially convenient for a large number of scenarios or alternative design options. However, few studies have examined the "coupling" of smoke spread in the station and evacuation accessibility. In this study, a time-based egress analysis is presented for an emergency evacuation in a mass transit station. The approach in NFPS 130 (NFPA 2003) is applied to a typical two-level rail station below grade, and five evacuation scenarios, driven by a train fire in the middle of the station platform, are developed and analyzed. A comparison of the calculated exiting times is also made to the STEPS simulation results.

DESIGN SCENARIO

Mass Transit Station

The station configuration, particularly with respect to emergency evacuation, is described here. Figure 1a a shows a typical transverse section. It has a center platform, and both the platform and the mezzanine are below grade. Each train consists of ten cars, which is about 50 ft (15.24 m) in length. The platform is 14 ft (4.3 m) wide and is about 550 ft (168 m) long. On the platform, there are eight stairways along the centerline, numbered sequentially from south to north as P1 to P8. The stairs are spaced from 30 ft (9.14 m) to 75 ft (22.86 m). The longest travel route on the platform is at the south end (Figure 2), which is about 160 ft (48.8 m) or about a three-car length. From the mezzanine to the street, there are four stairways (S1-S4) spaced evenly along the east side of the mezzanine. On the mezzanine, the longest travel distance is about 200 ft (61.0 m) from P1 to S1, as S1 is at the end of a rather long corridor. The numbering of these stairways is depicted in Figure 2 along with the longest travel routes.

[FIGURE 1 OMITTED]

The same station model was used previously for a platform thermal comfort study (Kang 2004) for which a certain degree of verification was obtained by comparing the CFD results with a one-dimensional network model. Both images in Figure 1 are extracted from the completed CFD model, which includes all the station's public area and two ten-car trains, as depicted in Figure 1b. A general purpose program (Fluent 2003) is used with the standard turbulence k-[epsilon] model. Smoke transport is solved using a separate species transport equation with the fluid modeled as a mixture composed of multiple components (i.e., air and smoke). Pressure-type boundaries are used for most of the model openings, except where there is forced ventilation when a velocity boundary condition is prescribed. Details of the CFD model can be found in Kang (2004).

[FIGURE 2 OMITTED]

Evacuation Configuration

The longest travel distance on the platform is approximately 160 ft, and on the mezzanine it is about 200 ft (Figure 2). In addition to stairways P1-P8, emergency exit stairways are located at both ends of the station platform. These emergency exit stairways are in enclosed shafts. Each shaft has two stairways leading directly from the platform to the street and bypassing the mezzanine. On the mezzanine are one service gate and 18 turnstiles between the paid and unpaid areas. The effective width of stairways P1-P8 is about 4 ft (1.2 m) and is 15 ft (7.5 m) for S1 and 7.5 ft (2.3 m) for S2 and S3. The street level is assumed to be the point of safety.

The station occupancy load is estimated from the entrainment load and the link load. The link load refers to the number of people on the trains in both directions during peak rush hour. The link load is estimated to be about 60% of the maximum train capacity; that is, 60% x 1,100 = 660 people per train per hour. A conservative assumption of 20% is made for the entrainment load. This gives, in the peak direction with one missed train headway, 2 x 20% x 1,100 = 440 people. In the off-peak direction, the entrainment load is 220 people. This gives a total occupancy load of 1,980 people. The travel speeds and/or the exiting capacities of typical circulation elements, such as stairs and gates, are taken from NFPS 130 (NFPA 2003) and are listed in Table 1.

Fire and Emergency Ventilation

The fire is assumed to be on the fifth car of the northbound train, which is in the middle along the platform (Figure 2a). The incident car is close to the platform stairways P2, P3, P4, and P5. Away from the fire, the smoke could simultaneously affect two of the four street exits, S2 and S3. Therefore, this fire location may be considered the "worst" scenario. Figure 3 shows the t-squared design fire of medium growth rate. The growth constant is [alpha] = 12 W/[s.sup.2], and the fire curve is Q = [alpha] [t.sup.2] + [Q.sub.0], in which the initial heat release rate, [Q.sub.0], is taken as the value at the end of the second minute (t = 120 s) on the medium growth curve. The fire is based on the burning properties of polystyrene, which has a heat of combustion of [DELTA][H.sub.c] = 39.2 MJ/kg and a smoke yield rate of [Y.sub.s] = 0.164 kg/kg-fuel (Mulholland 1995; Klote and Milke 2002). A volumetric approach (Zigh et al. 2000) is used to model the heat and the smoke release rates from the fire.

[FIGURE 3 OMITTED]

Emergency ventilation could use either the station ventilation system or the tunnel ventilation system. Station ventilation is done by smoke exhaust at the mezzanine ceiling, indicated by the line segment marked "RA" (return air) in Figure 1a. However, the duct size is limited by the available ceiling space to a maximum ventilation rate of 350 kcfm (165 [m.sup.3]/s). Since this might not be sufficient to contain the smoke, the tunnel ventilation system can also be used to exhaust the station from both ends of the platform, the so-called "all-exhaust" ventilation mode. The exhaust capacity is assumed to be 350 kcfm (165 [m.sup.3]/s) at each end of the station. The ventilation system's operational timeline is shown in Figure 3. It shows that the mezzanine exhaust starts at 30 seconds after the fire and is switched off at the fourth minute, while at the same time, the tunnel ventilation is switched on and reaches its peak within 30 seconds. Note that this timeline, such as the "switch of mode" at the fourth minute (Figure 3), is not pre-set but is based on the continuous monitoring of the smoke dispersion. This will be discussed further with the results.

TENABILITY/ACCESSIBILITY OF EVACUATION PATHS

During a fire in a rail station, tenability conditions in the station environment could be affected in a short time. This is especially important in considering the accessibility of the circulation stairs or certain areas in the station that might be affected by smoke. The required exiting time could be much longer if any of the evacuation routes were blocked by smoke. Because this accessibility is dependent on the fire development as well as the operation of the emergency ventilation (Figure 3), a time-based evacuation analysis approach is needed.

The objective of the emergency ventilation is to contain the smoke so that the egress routes farthest away from the fire, i.e., P1 to S1, P7 and P8 to S4, and the emergency exits, can be maintained tenable. A transient CFD simulation is conducted first to model the smoke transport inside the station. From the simulation results, the time is estimated when the station circulation stairs are affected by smoke (Table 2). The estimated times in Table 2 are based on the smoke visibility/obscuration levels.

It should be noted that smoke visibility is not the only parameter in determining tenability. In NFPS 130 (NFPA 2003), other parameters are also suggested, such as air velocity, temperature, radiant heat flux, toxicity, and other sub-lethal smoke effects (Gann 2004). However, visibility has been regarded as one of the most significant factors affecting safe escape in a fire. For example, when the visibility distance drops below 30 ft (10 m), statistical data have shown that 97% of British and 94% of US populations initiated the "turned back behavior" (Bryan 2002). This means that if a majority of the population would turn back from a stairway exit due to low visibility, that stairway is not accessible for evacuation.

In this study, a conservative 20% smoke obscuration is assumed, which would cause sufficient notice and concern such that passengers would turn back and stay away. While previous studies and some building codes have suggested a visibility of 30 ft (10 m), NFPS 130 (NFPA 2003) suggests a visibility of 100 ft (30 m) for illuminated signs and 30 ft (10 m) for doors and walls. Given that the stair openings on the mezzanine are about 10 ft (3 m) long, the "time of accessibility" for a platform stairway is estimated when the amount or concentration of the smoke in the stairway exceeds 20% light attenuation. This 20% smoke obscuration is about equivalent to a visibility distance of 50 ft (15 m) or less. On the platform, this visibility distance is approximately one car length. The smoke visibility is estimated using an approach based on "line of sight" (Kang 2005), following the definition of the light extinction (Klote and Milke 2002). This approach treats smoke visibility or obscuration as an exponential attenuation instead of using the correlation by Jin (2002).

Because the fire is closest to stairway P4, it can be seen (from Table 2) that smoke affects stairway P4 as early as t = 14 s. The two adjacent stairs, P3 and P5, are also affected in a rather short time, at about t = 30 s. Note that this is because the mechanical smoke exhaust does not start until t = 30 s. Stairway P6 on the north end becomes inaccessible at t = 160 s, and stairways P1, P7, and P8 are not affected for the first four minutes. Despite the low ceiling height, the ceiling smoke exhaust is able to withhold the smoke on the mezzanine for the first four minutes. After that time, the smoke exhaust by the tunnel ventilation fans clears the smoke in the mezzanine and is able to maintain the mezzanine smoke free thereafter.

In addition to smoke visibility, the condition in the station can also be examined using smoke concentration levels, such as smoke mass fraction, which is the mass ratio of smoke content over the mixture gas. Figure 4 shows the distribution of smoke mass fraction on the plane 6 ft above the platform and the mezzanine floor. The spatial distribution of smoke mass fraction, although not directly equivalent to visibility, is useful in confirming the tenability predictions from the visibility calculations.

At t = 80 s, Figure 4 shows that the smoke on the mezzanine is largely between stairway P2 and P5, and dense smoke mostly comes through stairway P4, as it is closest to the incident car. Note that the smoke has already caused an isolation of the platform and the mezzanine into two zones, north and south of the fire. With the growing fire, the smoke on the mezzanine starts to close in on stairways P1 and P7 at t = 240 s, despite the fact that on the platform the area near stairways P1 and P7 is not yet affected by smoke. Based on the observed results at t = 240 s, a decision is made to switch off the mezzanine exhaust and to activate the tunnel ventilation in exhaust at this time (Figure 3). This is to protect the two station-end evacuation routes, P1 to S1 and P8 to S4.

EVACUATION ANALYSIS BY NFPS 130 (NFPA 2003)

Five scenarios are selected to calculate the station exiting time (Table 3). Some of these scenarios take into account the accessibility of the stairways and/or utilization of the emergency exit stairways. The vertical circulation elements (VCEs) include the stairways from the platform to the mezzanine and the corridors/stairways from the mezzanine to the street level. The emergency exit stairways are not classified as VCEs and are in enclosed shafts providing access directly from the platform to the street.

The first four scenarios in Table 3 use a fixed number of VCEs, or a constant exiting capacity. The second and fourth scenarios assume that five of the eight platform stairways are not accessible due to smoke and that two of the four mezzanine exits are blocked by smoke. The number of inaccessible stairways is based on the timing from Table 2, which shows that the five platform stairways, P2-P6, and the two mezzanine exits, S2 and S3, are affected by smoke within the initial four minutes. The emergency exit stairs are utilized in all three "emergency" scenarios in Table 3, which provides additional exiting capacity. For all scenarios, a uniform occupancy distribution is initialized on the platform.

[FIGURE 4 OMITTED]

Figure 5 shows the exiting times for these scenarios. In general, except for the second case, the calculated exiting time is around or below four minutes for the platform and below six minutes for the station. For the second case, with the reduction in the number of accessible VCEs, the predicted exiting time is nearly nine minutes for the platform and more than ten minutes for the station. As suggested by case 2, the exiting time would be much longer if the emergency exit stairs were not used. These emergency exit stairs provide additional exiting capacity with a shorter travel distance and are, therefore, important. Comparing case 4 to case 2, the exiting time can be reduced by nearly five minutes. This is significant considering the exiting time is only about three to four minutes for the platform and is only about four to five minutes for the station.

The last scenario, case 5, uses a time-based egress approach. There are two main differences between the last case and the first four cases. First, in the last scenario, the exiting capacity is adjusted by the accessibility of VCEs as fire progresses (Table 2). The other difference is the consideration of the station as two separate zones. This is because of the station center fire, which isolates the north and the south areas of the station. This can be seen from Figure 4. With respect to evacuation, these time-based events are described using the following three stages and are used in case 5. Note that the time referred to here is consistent with that of fire development (Figure 3), and the VCEs accessibility is based on that in Table 2.

* Stage 1 -- Initial 30 seconds: From the very beginning, platform stairway P4 is not accessible due to its proximity to the fire. Evacuation starts from time zero, and the people initially near the fire move toward either the nearest stairways (P3 and P5) or the stairways farther toward the ends of the platform (P2 or P6).

* Stage 2 -- Second 30 seconds: At t = 30 s, the public area on the platform and the mezzanine are separated by fire into two isolated zones, north and south of the fire; at the same time, platform stairways P3 and P5 become inaccessible due to smoke. Evacuation continues, with those people in the middle of the platform moving toward the north and the south ends of the platform. At t = 60 s, platform stairway P2 and the mezzanine exits S2 and S3 become inaccessible for evacuation.

* Stage 3 -- After 60 seconds: Platform stairway P6 fails at t = 160 s. In addition to the emergency exit stairways, only two egress routes are available beyond t = 160 s, one at each end of the station P1 to S1, and P7 and P8 to S4).

The exiting time for the last scenario is calculated separately for the south and the north zones. Figure 5 shows that it could take close to 6 minutes to evacuate from the south zone of the station. This is much longer than the exiting time for the north zone, which is approximately 3.5 minutes. For the platform, the exiting time in the south zone has exceeded four minutes, while that of the north zone is only slightly over two minutes. In addition, the result indicates significant queuing on the platform of the south zone, i.e., at stairway P1 and the emergency exit stairways, where the wait time is about 3.5 minutes on the platform and is nominal on the mezzanine. In the north zone, the queuing occurs at the fare barrier (turnstiles) near exit S4 on the mezzanine, and the wait time is about 1.8 minutes on the platform and about 1.5 minutes on the mezzanine.

Note that in all the calculations, the entire occupancy load is initiated on the platform. The effect of the people density on the walking speed and the exiting capacity is not taken into account. However, the entrainment load gives the initial density as about 0.09 people/[ft.sup.2] (1.0 people/[m.sup.2]) on the platform, which is "moderate" for a corridor, and the assumption of 200 fpm (1.0 m/s) walk speed is quite reasonable (Proulx 2002). It should also be noted that the calculated exiting times are based on simultaneous evacuation. Scenarios 3-5 in Table 3 would be more of a "controlled" evacuation because of the use of emergency exit stairways. This is likely, as with two trains stopped in the station, there would be at least four trained transit personnel on the platform, two per train, to execute the evacuation.

NFPS 130 (NFPA 2003) specifies the exiting time requirements for a station as follows: (1) the exiting time for the platform is four minutes or less and (2) the exiting time from the most remote point in the station to a point of safety is six minutes or less. The results suggest that this requirement can be met with the additional capacity from the emergency exit stairways; however, with the inaccessibility and the division of the station into two zones due to smoke blockage, a longer exiting time is shown to be needed, especially for the platform in the south zone, which has exceeded the NFPS 130 (NFPA 2003) requirements.

Comparing the time-based egress results of case 5 to case 4, the predicted exiting time is much longer for the south zone and shorter for the north zone. This is probably because of the uneven occupancy load between the two zones, the south zone being of larger area but less circulation capacity than the north zone. Given the "hydraulic" flow approach in NFPS 130 (NFPA 2003), this result suggests that consideration not only of the accessibility of the stairways but also of the "area accessibility" is important in predicting exiting time. Indeed, NFPS 130 (NFPA 2003) itself has recognized the existence of "zone of tenability." In addition, the calculated exiting time is often referred to as the "minimum" exiting time, as it does not include the time needed to initiate the evacuation, that is, the pre-movement time (Nelson and Mowrer 2002). This pre-movement time can be estimated separately. The study by Proulx and Sime (1991) has shown a 15-second pre-movement time with the announcement of evacuation. NFPS 130 (NFPA 2003) also suggests consideration of the pre-movement time and other time intervals, but no specific guidelines are given.

COMPARISON WITH STEPS

In this section, the predicted exiting times are compared to the micro-simulation model STEPS (Simulation of Transient Evacuation and Pedestrian movementS). The STEPS program allows for people movement simulation under both normal and emergency situations. While Fruin's level of service (LOS) concept (in Proulx [2002]) can be used for both normal and emergency conditions, many other features have been implemented in STEPS to account for the complex human behavior during emergency situations such as a fire. These features include correlations of the people-moving speed with the load density, customizable occupancy group, awareness of certain exits/routes, and accessibility for evacuation. A detailed description of the program can be found in application studies such as Hoffman and Henson (1997) and Rhodes and Hoffman (1999).

The basic elements of a STEPS model include planes with exits and paths that link the planes together. Each plane is divided into rectangular grids for determining the shortest exit distance, and each exit is designated an exiting (flow) capacity. A STEPS model is essentially a mixed network of two- and one-dimensional flows. Figure 6 shows the station model, which is similar to that used for CFD (Figure 1b), with the exception of the emergency exit stairs at both ends of the platform. The same circulation capacities as in NFPS 130 (Table 1) are applied. The time-of-accessibility in Table 2 is used to control the evacuation availability of the VCEs; for example, stairway P4 is 14 seconds and P2 and P5 are 30 seconds. In a fashion similar to that used in case 5, the station is divided into two isolated zones at t = 30 s. Initially, the people load is distributed uniformly between the platform and the two trains. The evacuation condition at t = 10 s is depicted in Figure 6a. Note that simultaneous evacuation without any delay is assumed, which is the same as was used in the NFPS 130 (NFPA 2003) approach.

[FIGURE 6 OMITTED]

In general, the STEPS simulation results are consistent with the NFPS 130 (NFPA 2003) results. Both indicate similar locations of queuing. This can be seen from Figure 6b, which is an instantaneous snapshot of the evacuation at t = 90 s. Consistent with the NFPS 130 results, the queuing on the platform is at stairway P1, P7, and P8 and on the mezzanine at S3 and S4. The queuing of people can also be seen at the entrances to the emergency exit stairways at the south end of the platform.

Table 4 compares the predicted exiting times with those of case 5 by NFPS 130 (Table 3). In general, the STEPS predicted exiting times are 5% higher for the entire station and are 15% higher for the platform than those by the NFPS 130 (NFPA 2003) approach. As noted for the NFPS 130 results, the south zone has a higher people load than the north but less exiting capacity. In the STEPS model, the introduction of the time-based accessibility could create a more "uneven" people load between the zones that is initially "random" or uniformly distributed. This could further increase the evacuation time in the south zone. Another reason could be that in the STEPS model it is assumed that only 85% of the people are aware of the emergency exit stairs. Other factors would include the effects of the people density and the possible discrepancies in the actual travel distance. In a detailed comparison by Wall and Waterson (2002), the difference between NFPS 130 and STEPS has also been evaluated. Their study showed about a 10% difference, with the STEPS model being slightly conservative.

CONCLUSION

This study presents an analysis of an emergency evacuation for a platform train fire in the middle of a mass transit station. The station has two levels below grade. The train fire is approximated as a t-squared fire of medium growth rate. Taking into account the stair inaccessibility due to smoke obscuration, a time-based evacuation scenario is developed. This is used in a parametric study applying the NFPS 130 (NFPA 2003) exiting calculation approach. In the time-based evacuation scenario, the smoke generated from the fire not only renders a number of the stairways inaccessible over time but also isolates the station into two zones north and south of the fire.

The time-based egress analysis has demonstrated that, if the division of the station into two zones is considered, the exiting time could not meet the code requirements. Without the division, it does meet the required exiting time. Such division is caused by smoke blockage, which creates an uneven occupancy load between the zones. Therefore, consideration of the fire and smoke effects is important in estimating the emergency evacuation time and in evaluating the required time of safe egress. A comparison has also been made between the NFPS 130 (NFPA 2003) results and those predicted by the micro-simulation model STEPS. While both methods give consistent predictions, the exiting time predicted by STEPS is in general higher by about 5% to 15%. The reasons for such difference would include the people load balancing and their awareness of the emergency exit stairways.

The tenability requirements for a station fire are to allow for safe evacuation and safe access for fire fighting. In the case of the station center fire, while the objective of smoke management is to protect the evacuation routes at both ends of the station, the results from this study suggest that the emergency ventilation capacity might be only marginal, and further study would have to reconcile the available time and the required time of tenability.

REFERENCES

Bryan, J.L. 2002. SFPE Handbook of Fire Protection Engineering, 3d ed., P.J. DiNenno, ed., chapter 12, Behavioral response to fire and smoke. Quincy, MA: Society of Fire Protection Engineers and National Fire Protection Association.

Chien, S.-W., W.-L. Chen, T.-S. Shen, C.-C. Cheng, D.-C. Lee, Y.L. Hsuse, and T-K.Chen. 2004. A study on the people evacuation safety for the underground MRT station. Proceedings of the 6th Asia-Oceania Symposium on Fire Science and Technology, Daegu, Korea, March 17-20.

Comeau, E., and A. Wolf. 1997. Fire in the Chunnel! NFPA Journal, pp. 58-64.

Fahy, R.F. 1999. User's Manual, EXIT89, v 1.01, An Evacuation Model for High-Rise Buildings. Quincy, MA: National Fire Protection Association.

Fluent. 2003. Fluent, v6.1 User's Guide. Lebanon, NH: Fluent, Inc.

Galea, E.R., and J.M. Perez Galparsoro. 1994. A computer-based simulation model for the prediction of evacuation from mass-transport vehicles. Fire Safety J. 22:341-66.

Gann, R.G. 2004. Sublethal effects of fire smoke. Fire Technology 40:95-99.

Hoffmann, N.A., and D.A. Henson. 1997. Simulating transient evacuation and pedestrian movements in stations. Proceedings of the International Conference on Mass Transit Management, Kuala Lumpur, Malaysia, May.

Jin, T. 2002. The SFPE Handbook of Fire Protection Engineering, 3d ed., chapter 4, Visibility and human behavior in fire smoke. Quincy, MA: Society of Fire Protection Engineers and National Fire Protection Association.

Kang, K. 2004. Analysis of platform spot cooling in an underground subway station. International Journal of Ventilation 3(3):235-44.

Kang, K. 2005. Modeling smoke visibility in CFD. Proceedings of the 8th International Symposium Of Fire Safety Science, Beijing, China, Sept. 18-23.

Kennedy, W.D., and K. Kang. 2003. Tunnel fire modeling comparing CFD and the Froude number method. Proceedings of the 11th BHR Group International Symposium on Aerodynamics and Ventilation of Vehicle Tunnels, Luzern, Switzerland.

Klote, J.H., and J.A. Milke. 2002. Principles of Smoke Management. Atlanta: American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc., and Society of Fire Protection Engineers.

Mulholland, G.W. 1995. Smoke production and properties. The SFPE Handbook of Fire Protection Engineering, 2d ed., P.J. DiNenno, ed., pp. 217-27. Quincy, MA: Society of Fire Protection Engineers.

Nelson H.E., and F.W. Mowrer. 2002. The SFPE Handbook of Fire Protection Engineering, 3d ed., P.J. DiNenno, ed., chapter 14, Emergency movement. Quincy, MA: Society of Fire Protection Engineers and National Fire Protection Association.

NFPA. 2003. NFPS 130, Fixed Guideway Transit and Passenger Rail System. Quincy, MA: National Fire Protection Association.

Proulx, G., and J.D. Sime. 1991. To prevent "panic" in an underground emergency: Why not tell people the truth? Proceedings of the 3d International Symposium on Fire Safety Science. London: Elsevier.

Proulx, G. 2002. The SFPE Handbook of Fire Protection Engineering, 3d ed., P.J. DiNenno, ed., chapter 13, Movement of people: The evacuation timing. Quincy, MA: Society of Fire Protection Engineers and National Fire Protection Association.

Rhodes, N., and N. Hoffman. 1999. Fire safety engineering for the International Centre for Life, Newcastle-upon-Tyne. Proceedings of Interflam '99, Edinburgh, UK, June.

Simcox, S., N.S. Wilkes, and I.P. Jones. 1992. Computer simulation of the flows of hot gases from the fire at King's Cross underground station. Fire Safety J. 18:49-73.

Thompson, P.A., and E.W. Marchant. 1995. A computer model for the evacuation of large building populations. Fire Safety J. 24:131-48.

Wall, J.M., and N.P. Waterson. 2002. Predicting evacuation times--A comparison of the STEPS simulation approach with NFPA 130. Fire Command Studies 1(1).

Zigh, G., I. Ong, and K. Kang. 2000. CFD application for station fires. Proceedings of the 10th BHR Group International Symposium on Aerodynamics and Ventilation of Vehicle Tunnels, Boston, MA.

DISCUSSION

Robert Till, Associate Professor of Fire Science, John Jay College of Criminal Justice, New York, NY: You state in your paper that announcements of evacuation require a 15-second pre-movement time. Was this for pre-recorded announcements or announcements directly from staff? Pre-recorded announcements are often suggested as the best option since there is no chance of microphone feedback and they should be easier for patrons to hear. However, these pre-recorded announcements may lack credibility with patrons, as they know the announcements are generated automatically and could be a false alarm. This could increase the pre-movement time, especially when the passengers may not be in close proximity to the hazard. Did you find any discussion of this in your literature search?

Kai Kang: In a more accurate definition, the 15-second time should refer to the delay time for passengers to evacuate after the voice announcement, that is, after detection of the fire incident. I confirm that in the associated study (by others), live messages are used communicating the accident type and location and providing evacuation instructions. I agree with the comments on the lack of credibility of pre-taped messages during emergency situations. I believe this is also one of the objectives of the required Public Address (PA) system in NFPS 130 (2003). The characteristics of delayed initiation to evacuate are important in an underground environment, which should be a topic by itself and is not the subject of my paper and its literature review.

Kai Kang, PhD

Member ASHRAE 

Kai Kang is a project manager in the New York City office of Hatch Mott MacDonald.

Table 1. List of Typical Travel Speeds and/or Exiting Capacities
According to NFPS 130 (NFPA 2003)

Type of
Circulation Element    Travel Speed         Exiting Capacity

Horizontal surface     200 fpm (1.0 m/s)     2.27 pim
  (platform)
Stairs (up-direction)  50 fpm (0.254 m/s)*   1.59 pim[dagger]
                                               (0.0626 p/mm x min)
Gates                  --                    2.27 pim (0.0894 p/mm
                                               x min)
Turnstiles             --                   25 ppm[double dagger]

* Travel speed is based on vertical distance.
[dagger] "pim" is person per inch width and per minute.
[double dagger] "ppm" is person per minute.

Table 2. Predicted Time of Accessibility When the Circulation Stairways
Are Affected by Smoke

                    Time of Accessibility* for the Stairways
Location            (seconds)

Platform stairs     P1            P2           P3
  to mezzanine
  w/ smoke exhaust  >240          80           31
    at mezzanine
Mezzanine stairs    S1                  S2
  to street
  w/ smoke exhaust  - ([dagger])        n/a ([double dagger])
    at mezzanine

                     Time of Accessibility* for the Stairways
Location            (seconds)

Platform stairs     P4            P5           P6
  to mezzanine
  w/ smoke exhaust  14            33           160
    at mezzanine
Mezzanine stairs                  S3
  to street
  w/ smoke exhaust                n/a ([double dagger])
    at mezzanine

                    Time of Accessibility* for the Stairways
Location            (seconds)

Platform stairs     P7            P8
  to mezzanine
  w/ smoke exhaust  >240          - ([dagger])
    at mezzanine
Mezzanine stairs        S4
  to street
  w/ smoke exhaust      - ([dagger])
    at mezzanine

* "Time of accessibility" estimated based on 20% smoke obscuration on
the mezzanine.
([dagger]) Not affected by smoke.
([double dagger]) Not applicable, as the makeup air keeps only the exit
stairs clear of smoke.

                         Evacuation Time (min)
                         Exiting Time for  Exiting Time for
Scenarios                Platform (min)    Station (min)

Normal                   3.24               5.28
Normal w/VCEs down       8.65              10.21
Emergency & All VCEs     2.16               4.04
Emergency w/VCEs down    3.71               5.27
Emergency & Tenability-  4.27               5.83
  Station South
Emergency & Tenability-  2.02               3.38
  Station North

Figure 5 Comparison of evacuation times for all the scenarios.

Note:  Table made from bar graph.

Table 3. Configuration of Station Evacuation Scenarios

                               No. of Vertical Circulation Elements
                               (VCEs)
                               Platform    Mezzanine
                               Stairs to   Stairs to   Emergency Exit
No.  Scenario Description      Mezzanine   Street      Stairways

1    Normal                    8           4           0
2    Normal w/ VCEs down       3           2           0
3    Emergency and all VCEs    8           4           4
4    Emergency w/ VCEs down    3           2           4
5    Emergency and tenability  Varies 7-3  Varies 4-2  4

Table 4. Comparison Between Exiting Time Predictions from STEPS and
NFPS 130

          Station Exiting Time  Platform Exiting Time
          (min)                 (min)
          North   South         North   South

STEPS      3.33   6.20           2.28    4.96
NFPS 130   3.38   5.83           2.02    4.27
%         -1.38%  6.35%         13.04%  16.16%