Critical Velocities for Smoke Control in Tunnel Cross-Passages

Transcription

1 Paper presented at the First International Conerence on Major unnel and Inrastructure Projects, May 2000, aipei, aiwan. Critical Velocities or Smoke Control in unnel Cross-Passages Fathi arada, HBI Haerter Ltd., Consulting Engineers, Switzerland ABSRAC his paper addresses the issue o critical low velocities required to stop the ingress o smoke into tunnel cross-passages used or passenger evacuation. Current engineering practices or estimating critical velocities in tunnels (empirical correlations, phenomenological methods, Computational Fluid Dynamics) are briely reviewed. he paper proposes a correlation or the critical velocity in cross-passages, and presents supporting evidence or the correlation using CFD. In addition to the usual geometrical and ire heat release parameters, the longitudinal velocity through the tunnel is ound to be an important parameter or the estimation o the critical velocity through a cross-passage. Some o the assumptions and limitations o the proposed model are discussed, and practical recommendations or preliminary design work are given. 1 INRODUCION During a ire scenario in a vehicle tunnel, it is o paramount importance to maintain escape routes ree rom dangerous smoke. In many road, railway and metro tunnels, cross-passages between two parallel tunnels are employed as escape routes during a ire emergency. hese cross-passages must be guaranteed ree o smoke in order to provide a visual indication o sae evacuation paths to escaping passengers, to protect passengers while they are traversing the cross-passages, and to ensure that the nonincident tube is kept clear o smoke. One important question that tunnel ventilation specialists and other parties concerned with tunnel saety have to answer is: what is the minimum resh air velocity required to maintain smoke-ree conditions in a cross-passage? he accurate estimation o this critical velocity allows a better balance to be struck between the desired saety level and the cost o ventilation installations (ans, doors, ducting etc). In order to discuss the underlying issues and challenges behind this question, a review o the current practice in estimating critical velocities or smoke control is presented, ollowed by a discussion regarding how the particular issue o critical velocities through crosspassages can be handled. o o neighbouring tube tube -1-

2 Paper presented at the First International Conerence on Major unnel and Inrastructure Projects, May 2000, aipei, aiwan. Fig. 1: ypical Rail-unnel Cross-Passage (courtesy o Alpransit Gotthard AG) -2-

3 2 NOMENCLAURE Symbol A C p grade G H Q & c V Greek Meaning unnel cross-sectional area Heat capacity o air Gradient o tunnel Acceleration due to gravity Height o tunnel cross-section Heat release rate due to ire emperature Flow velocity Units m 2 J/(kg K) % ms -2 m W K ms -1 ρ Density kg/m 3 Subscripts (none) c d Ambient conditions Critical value Cross-passage door Fire conditions Main tunnel 3 SAFEY FEAURES Beore discussing the speciic issue o critical velocities in cross-passages, it may be o beneit to outline the general saety eatures applicable to tunnel cross-passages used or evacuation purposes, as outlined in approximate chronological order below: Immediately ater ire incident in a vehicle Cross-passages are usually protected at one or both sides by doors which act as passive ire and smoke barriers. hese doors are designed to withstand high temperatures (204 C according to NFPA 105, 1993) during the entire evacuation period. Upon activation o the emergency ventilation system A positive pressure dierence is developed across the cross-passage. Fresh air lows across any cross-passage openings and leakages rom the non-incident to the incident tube (Fig. 2). However, the pressure dierence must be limited in order to acilitate manual opening o the doors (133 N maximum orce according to NFPA 92A, 1996). Ater opening the cross-passage doors he cross-passage doors in the region o the ire are opened either manually by escaping passengers or sta, or via remote control rom the traic operations centre. Due to the reduced low resistance, the air-low in the direction o the incident tube is signiicantly increased, and this serves to block the low o smoke into the cross-passage. In addition, a bubble-eect is generated in the incident tube, where the resh air jet clears the smoke in the vicinity o the cross-passage door and thus serves to direct -3-

4 escaping passengers. his eect was observed during the Channel unnel ire (Channel unnel Saety Authority, 1997). We shall now ocus on the issue o estimating the critical velocity or smoke control ater opening the cross-passage doors. Burning train Cross-passages Non-incident tube Air-low ++ Overpressure in non-incident tube Fig. 2: Emergency Ventilation across Rail unnel Cross-Passages 4 CURREN PRACICE IN ESIMAING HE CRIICAL VELOCIY Current engineering practices or calculating the critical velocity or smoke control in tunnels include: Empirical correlations he most widely-used correlations or estimating the critical velocity are based upon a non-dimensional Froude number analogy (homas, 1970). he Froude number is deined as the ratio between the buoyancy orces generated by the ire and the inertial orces due to the imposed ventilation air low (Eqn. 1). gh ( ρ ρ ) Fr = Eqn. 1 2 ρv According to the experimental measurements o Lee et al (1979), Froude numbers o less than 4.5 are required to preclude the movement o smoke against the imposed ventilation low direction. By relating the density dierence between the hot gases rom the ire to the ambient air ( ρ ρ ) to the convective heat release rate rom the ire Q & ), Kennedy (1996) proposed a ormula or the critical velocity, such: ( c V c ghq& c = ρcp A Frc 1/ 3 Eqn. 2 where the critical Froude number (Fr c ) is given by -4-

5 0.8 3 Fr c = 4.5( min( grade,0) ) Eqn. 3 is estimated rom the enthalpy conservation equation: Q& c = ρ C AV + p c Eqn. 4 Equations 2 to 4 orm a coupled set that are solved within many well-used tunnel ventilation programmes including the Subway Environmental Simulation (SES) Computer Programme (Version 4, 1997). Nonetheless, Grant et al (1998) have pointed out several methodological weaknesses in this model, including its ailure to account or the complex near-ire low ield and its interaction with the ire source and the particular tunnel under consideration. For the particular case o tunnel cross-passages, the convective heat release rate Q c rom a vehicle ire occurs outside the cross-passage so it is not immediately obvious how this model should be used. In addition, it is not clear which low velocity should be used in Eqn. 4 (through the tunnel or through the cross-passage?). Section 5 proposes a development o this model to account or cross-passage lows. Phenomenological methods hese are typically two-dimensional methods that employ multiple zones (resh air layer, smoky layer(s)) or predicting the smoke spread rom ires, as described by Charters et al (1994). Although they provide signiicantly more inormation than simple empirical correlations or the critical velocity, their extension to deal with cross-passage low is inherently problematic due to the strong threedimensional nature o such lows. Computational Fluid Dynamics (CFD) CFD is an engineering tool or solving the governing Navier-Stokes equations or the low, temperature and low species in virtually any system o interest. Its power and lexibility has led it to its increasing use or tunnel ventilation applications, including the resolution o the near-ire low ield (e.g. uovinen and Holmstedt, 1994). However, CFD oers no panacea the underlying mathematical models rela t- ing to turbulence, combustion and radiation are still being actively developed and hence have to be careully validated or the proposed engineering application. Most tunnel ventilation specialists still employ one-dimensional low networks using computer programmes such as hermoun or SES to develop their emergency ventilation concepts, and only use CFD to conirm their one-dimensional design or to answer critical questions (e.g. relating to threedimensional low and smoke patterns) that cannot be dealt with using simpler tools. 5 EMPIRICAL MODEL FOR CROSS-PASSAGE CRIICAL VELOCIY As made apparent in the previous discussion, empirical models or the critical velocity are still o engineering interest, despite the availability o phenomenological methods and CFD. hey oer a ast and robust means o estimating the required airlow, which can be used or preliminary ventilation designs. I required, more sophisticated engineering calculations or physical model tests can be carried out to conirm the inal design parameters. In order to develop the empirical model, we shall consider the enthalpy balance in a control volume straddling both the cross-passage and the main tunnel (Fig. 3). -5-

6 Fire Main tunnel Longitudinal air low Smoke Cross-passage Fig. 3: Control Volume or the Enthalpy Balance in a unnel Cross-Passage In the absence o appreciable conductive heat loss through the walls, the enthalpy equation or the above control volume can be written as ( m & C ) + ( mc & ) + Q& = ( mc & ) Eqn. 5 p p d c p Cross-low where the subscript reers to the main tunnel, the subscript d reers to the cross-passage door and all other terms are deined in the Nomenclature. hrough analogy with Eqn. 1, the Froude number at the cross-passage door may be written as gh Fr = grade d ( ρ ρ ) = 4.5( min(,0) ) 2 ρvd Eqn. 6 where, is the mixed-out temperature downstream o the tunnel ire. Eqns. 5 and 6 can be solved in a coupled manner to estimate the critical velocity. In the absence o a longitudinal low in the tunnel ( m& 0), equations 5 and 6 are equivalent to the Kennedy correlations (Eqns 4 to 6). he critical velocity thus calculated would thereore correspond to a ire located just inside the cross-passage door-rame, i.e. a worst case scenario. With increasing longitudinal low in the tunnel, the hot gases in the tunnel would be cooled down, hence their density (ρ ) would rise. his eect reduces the critical velocity through the cross-passage V d in Eqn. 6. Numerical Example A brie numerical example will demonstrate the hypothesised relationship. C p = 1040 J/(kg K) = 300 K Q c = 20x10 6 W ρ = 1.1 kg/m 3 grade = 0% A d = 4.4 m 2 A = 33 m 2 (tunnel annulus area) H d = 2.2 m -6-

7 2.5 Value given by Kennedy Correlations, Eqns 2-4 Critical Velocity through Cross-Passage (m/s) Curve given by Eqns Longitudinal Flow Velocity through unnel (m/s) Fig. 4: Variation o Critical Velocity through a Cross-Passage with Longitudinal unnel Flow (Example) Fig. 4 shows the critical velocity through a cross-passage would be signiicantly reduced with increasing longitudinal low through the tunnel. his would imply that tunnel ventilation designers have a choice o dierent strategies or controlling smoke propagation across parallel tunnel tubes or within stations: - either increase the longitudinal tunnel low to cool the hot gases while maintaining a small crosspassage low velocity, or - ensure a strong cross-passage velocity, when the longitudinal tunnel velocity is small or cannot be controlled. he irst strategy has the added beneit o controlling smoke within the vehicle tunnel as well as within the cross-passage, as long as the airlow within the vehicle tunnel exceeds the required critical velocity. On the other hand, the generally large cross-sectional areas o the vehicle tunnel means that high longitudinal ventilation lowrates may be required. he second strategy is by deinition a more robust strategy or controlling the smoke spread through a cross-passage, since no assumption is made regarding the longitudinal tunnel velocity. In practice however, the choice o emergency ventilation strategy is intimately related to the escape paths chosen (e.g. along the platorm, through cross-passage doors, up escape staircases) and hence has to be careully evaluated or each project. Assumptions and limitations o model An empirical model or the critical velocity as presented by Eqns. 5-6 has many assumptions and limitations that the reader should be aware o, in addition to those mentioned or the original Kennedy correlations in section 4. - he hot smoke is assumed to ully mix with the longitudinal tunnel low beore reaching the crosspassage door. his in turn implies that the ire is located upstream o the cross-passage. Fires that are downstream o the cross-passage are unlikely to be the most critical in terms o smoke ingress into the cross-passage. -7-

8 - he door height, H d, is used as the relevant length scale in the deinition o the Froude number (Eqn. 6). For the limit o zero longitudinal low through the tunnel, this choice o length scale produces a critical velocity that is consistent with Kennedy's correlations. - he empirical model neglects all instationary eects. In a real ire scenario, vehicle traic low changes rapidly and the emergency ventilation ans would be put into action, leading to highly instationary conditions. Use or preliminary design For practical engineering purposes, a worst-case estimate o the required critical velocity through a cross-passage can be obtained or preliminary design purposes by - Assuming no longitudinal low through the tunnel, i.e. m& = 0 ; - Setting the convective heat release rate o the ire Q & c equal to the total expected heat release rate. he net eect o these two assumptions is to provide a critical low velocity and emergency an capacities that are somewhat conservative. Ater development o the overall emergency ventilation concept, urther optimisation may be carried out using CFD to check and conirm the preliminary estimates. 6 COMPUAIONAL FLUID DYNAMICS A limited CFD study was undertaken to investigate whether the results rom the correlation or the critical velocity (Eqns. 5-6) are tenable. A section o a rescue station in the Gotthard Base unnel was modelled in the CFD analysis, as indicated in Fig. 5. Area under investigation Incident train 80m Incident tunnel Incident Coach 30m Escape door Cross-passage 46.6m Non-incident tunnel Rescue station with incident train Cross-passage with escape door (2m x 2.2m) -8-

9 Fig. 5: Geometrical Model o Rescue Station with an Incident rain As a irst step, the CFD model (mesh, boundary conditions and physical models o heat transer and turbulence) was tested or the case o a 10 MW train ire with no low through the cross-passage. Fig. 6 indicates that a good agreement between the CFD computations and the Kennedy ormula or the critical velocity was obtained. 2.5 Flow Velocity through Station (m/s) A Kennedy Formula or Critical Velocity CFD Computation, with 19 m Smoke Backlayering CFD Computation, with No Smoke Backlayering Fire Heat Release Rate (MW) Fig. 6: Critical Velocities in Station unnel he behaviour o the smoke within the tunnel was then analysed using a range o longitudinal velocities through the rail tunnel and through the cross-passage, and using two ire heat release rates (Fig. 7). V=0 2 m/s Fire source: Q=10 20 MW m= kg/s =988 K P=1.01 bar adiabatic walls V=0.5 2 m/s Fig. 7: External Boundary Conditions Applied to the CFD Model O particular interest in the calculations was to investigate the conditions under which smoke ingress into the cross-passages may occur (Fig. 8). Summary diagrams were produced or each ire heat release rate comparing the results o the CFD predictions o smoke ingress with the critical velocities -9-

10 predicted by the correlations (Fig. 9). he results are in agreement which each other, which gives some conidence in the use o the proposed correlation o critical velocity in a cross-passage. Fig. 8: Smoke Contaminates 14 m o a Cross-Passage or the Case o No Imposed Longitudinal Flow through the Station unnel, 2 m/s through Escape Door, 10 MW Fire. 4 Flow Velocity through Escape Door (m/s) A Formula value or critical velocity through escape door (m/s) CFD Result, No Backlayering in Cross-Passage CFD Result, 14 m Backlayering in Cross-Passage Flow Velocity through rain/unnel Annulus (m/s) Fig. 9: Variation O Critical Velocity through a Cross-Passage with Longitudinal unnel Flow using CFD and Formula (Eqns. 5-6) or a 10 MW Fire -10-

11 -11-

12 7 CONCLUSIONS he maintenance o smoke-ree conditions in evacuation cross-passages is an important saety goal, and engineering correlations such as the ones discussed here, despite their many limitations, serve to - highlight the importance o parameters such as cross-passage door geometry and longitudinal low through the tunnel during the early design stage, and - provide a irst estimate o the low velocity needed to protect the evacuation cross-passages, and hence deduce the required emergency an capacity. he use o properly validated Computational Fluid Dynamics (CFD) models can provide deeper insights into the probable behaviour o smoke in complex three-dimensional tunnel geometry (tunnels, stations, cross-passages, escape doors etc.). In this study, CFD was used to conirm the trends suggested by the simple engineering correlations. In general, CFD can also be used to check certain design points that are identiied by the tunnel ventilation designer as being critical. 8 REFERENCES Channel unnel Saety Authority (1997) Inquiry into the ire on heavy goods vehicle shuttle 7539 on 18 November 1996, he Stationery Oice, UK. Charters, D.A., Gray, W.A. and McIntosh, A.C. (1994) A computer model to assess ire hazards in tunnels (FASI), Fire echnol. 30, pp Grant, G.B., Jagger, S.F. and Lea, C.J. (1998) Fires in tunnels, Phil. rans. R. Soc. London A, 356, pp Kennedy, W.D. (1996) "Critical Velocity : Past, Present and Future", One Day Seminar on Smoke and Critical Velocity in unnels, IC, 2 April. Lee, C.K., Hwang, C.C., Singer, J.M. und Chauken, R.F. (1979) Inluence o Passageway Fires on Ventilation Flows", Second International Mine Ventilation Congress, Reno, Nevada, 4-8 November. National Fire Protection Association (1996) NFPA 92A, Recommended practice or smoke-control systems. National Fire Protection Association (1993) NFPA 105, Installation o smoke-control door assemblies. homas, P.H. (1970) Movement o smoke in horizontal corridors against an air low. Inst. Fire Engrs Q. 30, pp uovinen, H. and Holmstedt, G. (1994) CFD Modelling o unnel Fires, International Conerence on Fires in unnels, Boras. -12-