MECHANICAL SMOKE EXTRACTION FOR LARGE FIRE LOADS IN THE GOTTHARD BASE TUNNEL

Transcription

1 MECHANICAL SMOKE EXTRACTION FOR LARGE FIRE LOADS IN THE GOTTHARD BASE TUNNEL Brander, Ch. Pöyry Infra Ltd, CH ABSTRACT The new Gotthard Base Tunnel will be used by passenger and freight trains, whereby the latter ones may carry heavy good vehicles with large fire loads of up to 250 MW. As a case study, the impact of these fires was studied with respect to the mechanical smoke extraction, the resulting concrete and air temperatures, and their influence on airflow and subsequent fan operation points. One-dimensional flow simulations combined with a radial shell and a linear layer model for concrete temperature simulation were performed. The following topics will be addressed in the article: Modelling of 40 and 250 MW fires, the smoke extraction system and the entire concrete plenum system. Air and concrete temperatures in the smoke exhaust plenums and requirements for the concrete structure and technical equipment, such as doors and exhaust fans. Heat fluxes and balances throughout the duration of the fire. Impact on the overall pressure loss, volume and mass flows, and operating points of the smoke exhaust fans. Simulation of further effects, such as buoyancy and heat penetration into walls of concrete. Impact on the design of the smoke exhaust fans and the concrete structure. Measures to ensure the function of the smoke exhaust fans. Keywords: mechanical smoke extraction, large fire loads, modelling of concrete plenum system and smoke extraction system, one dimensional flow simulation 1. INTRODUCTION The new Gotthard Base Tunnel (GBT) consists of two 57 km long single-track tunnels, which are connected by cross passages. Two multifunction station are placed at 20 and 40 km from the northern portal of the tunnel. Changing lanes and emergency stations as well as technical rooms accommodated for rail operations and ventilation installations are situated in the multifunctional stations. The GBT will be in operation by the end of Each multifunction station consists of two emergency stations, which can be fully ventilated by air inlet through the escape doors and by smoke extraction. The smoke extraction is designed for a volume flow up to 250 m 3 /s. The fire load of a passenger train is assumed not to exceed 40 MW. When a fire is located on a passenger train, the train in intended to stop in the next emergency station and the exhaust fans are activated. Beside the passenger trains, the GBT will be primarily used by freight trains and occasionally by trains carrying heavy good vehicles with large fire loads of up to 250 MW. In case of a fire, these trains are not meant to stop in the emergency stations but to get out of the tunnel. In

2 case of the stopping of a freight train at the emergency station, the exhaust fans will not be activated, nor will any damper be opened. Still, as a case study, the impact of a fire with a fire load of 250 MW, combined with the activation of the exhaust ventilation system, was studied. The impact of these fires was studied with respect to the mechanical smoke extraction, the resulting concrete and air temperatures, and their influence on airflow and subsequent fan operation points. Various simulations were performed for all emergency stations of which are two at Sedrun and two at Faido. The concept of the smoke exhaust plenums is demonstrated in Figure 1. Figure 1: Geometry of the smoke exhaust plenums 2. SIMULATION AND MODELING 2.1. Numerical flow simulation The impact of the fires was studied with numerical flow simulations, which involved a large number of boundary conditions, dependencies and influences. Therefore, a complex aerodynamic and thermodynamic simulation system was needed. The aerodynamic conditions of the smoke flow had to be linked to the thermodynamic behavior of the smoke and the surrounding concrete structure. The aerodynamic model used was one dimensional in the direction of the air flow and transient; no branches such as leakages or parallel flow were simulated. Using a conservative approach, it was decided not to simulate any leakages, as they would result in lower gas temperatures in the smoke exhaust plenums. The air volume was discretized in length and time. A typical element length was 20 m, time steps varied between 0.1 and 0.5 s. Due to the high temperature and large altitude difference, the density was variable and calculated by pressure and temperature for each time step and element. An explicit finite volume approach was used to solve the differential equations. Each element was determined by geometrical and physical parameters like altitude difference, pressure drop coefficient, surface area et cetera. Further influences like standing trains in the tunnel, exhaust fans and fire loads were also placed in the simulation model. The thermal condition was calculated for every timestep and every element of the smoke exhaust plenum. Two thermodynamic models for concrete and rock were used to simulate the impact on the surrounding concrete: A radial shell model was used for all tunnels, caverns, shafts surrounded by concrete and rock. The rear boundary condition was thereby a constant temperature at a depth of 1 m.

3 A linear layer model was used for all partition walls/ceilings that were vented on the rear via the makeup air. The rear boundary condition was a constant air temperature. The heat transfer coefficient between the air and the wall is calculated on the basis of forced or free convection. The two discretization models were then combined to give a good approach of the real geometry. A typical discretized model of an access tunnel can be found in Figure 2. Figure 2: Typical discretized model of the smoke exhaust plenum in an access tunnel showing makeup air below the suspended ceiling. The heat transfer coefficient α, which describes the convective heat flow from the air/smoke into the concrete surface, is critical parameter for the calculations. The heat transfer coefficient α is calculated, using the following formula (Incropera 2002, [1]): α = with μ 3 LL.027*Re *Pr * μ LW λ d hydr μ LL : Dynamic viscosity of air at air temperature μ LW : Dynamic viscosity of air at wall temperature Re: Reynolds number Pr: Prandtl number λ: thermal conductivity of air [W/mK] d hydr = hydraulic diameter Due to the use of correction factors and temperature dependent values for the calculation of the Reynolds number, this formula is valid for a large temperature range. The heat transfer coefficient for the present simulations is W/(m 2 /K) Modelling of the fire The simulations were based on two fire load curves for a passenger train (P peak = 40 MW) and a freight train (P peak = 250 MW). The fire load curve of the freight train corresponded to the simultaneous fire of 12 freight trailers. The fire load of the passenger train rose in its first 70 min to a size of 20 MW from where it further increased to a maximum size of 40 MW after 360 min before it finally stopped after 480 min. (1)

4 For the freight trains, the fire load at the beginning of the simulation was 180 MW from which it grew after 30 minutes to its peak of 250 MW, where it remained for 150 minutes. The total heat released by the fire was calculated by the fire load multiplied by a factor of 0.9 for the degree of conversion of the burning material. The heat that the fire emitted by radiation heat transfer was calculated as a fixed percentage amount of the total fire load. The radiation heat transfer from the hot gases into the smoke exhaust plenums was neglected, resulting in higher exhaust gas temperatures, as the hot gases are significantly warmer than the surrounding walls. The duration of the simulations covered the entire duration of the fire. Due to the immense amount of heat stored in the concrete exhaust plenums, even after hours, the exhausted air was still relatively hot Modelling the exhaust fans The smoke exhaust plenums were separated from the tunnel through exhaust dampers. The conservative assumption was that the train would already be burning while entering the emergency station and that the exhaust fans were already operating. The exhaust fans in Faido are placed at the outer end of the smoke exhaust plenum. At Sedrun, the fans are located at 400 m from the outlet in an underground cavern. The exhaust fans are equipped with a blade angle control system and frequency converters. In case of an event the exhaust fans run at a constant speed on a characteristic curve with a maximum flow rate. This characteristic was incorporated into the simulation program and served to determine the air flow in the system. 3. RESULTS Simulations of fires in passenger trains (40 MW) and freight trains, carrying heavy good vehicles (250 MW), in all four emergency stations were examined. The results are illustrated using the example of Faido East emergency station with fire on a freight train. In this case, several effects can be well studied. Where it makes sense, results from other emergency stations are added Exhaust gas temperature One important variable is the exhaust gas temperature, which strongly depends on the distance from the fire. In Figure 3 these temperatures are shown in relation to the distance to the fire. Due to the large length of the exhaust system (the access tunnel length is 2700 m), the air significantly cools down on the way to the portal. The maximum air temperatures shortly behind the fire area depend on the fire load and the mass flow. The moved mass flow decreases notably after the outbreak of the fire, because the hot air creates a greater resistance to the flow (seen also Section 3.3). Due to the powerful exhaust system, which can extract large amounts of exhaust gas, the maximum exhaust gas temperatures do not exceed 730 C. The temperature variation over time is shown in Figure 4. After 180 minutes the fire diminishes. At this time the maximum air temperatures are reached. An exception is the area near the portal, which has a distance of more than 3000 m to the fire. The gas temperatures in this area continue to rise and reach their maximum after 210 min, due to the enormous amounts of heat, which remain stored in the concrete surrounding. At Faido the fans are located at the very end of the exhaust duct where the exhaust gas temperature could reach a maximum of 100 C. At the Sedrun exhaust fan central, much higher smoke temperatures of 240 C are achieved. The heat resistance requirements for all fixtures and exhaust fans could be derived from the smoke exhaust temperature. Since the temperatures represent mean values, a safety margin had to be added.

5 Figure 3: Air temperature in the exhaust duct 3.2. Concrete and rock temperature Figure 4: Temperature at points (1), (2) and (3) in relation to time The concrete temperature is the critical variable for assessing the impact of fire on the concrete, particularly in relation to the explosive spalling of concrete. The most important factors are the concrete temperature (critical from 180 C to 250 C an higher), the temperature gradient in the concrete depth (critical from 5 K/mm) and the heating rate (critical from 2.5 K /min). All values can be extracted from the simulation results. A typical temperature profile for the vault and the concrete partition wall at the beginning of the partition wall (Pos. 1 in Figure 1) is shown in Figure 5 and Figure 6. While the exhaust gas runs in the first few hundred meters in special excavated galleries where the supply air and exhaust air plenums are separated by a partition wall. The escape routes follow the supply air side. This would result in high security requirements on that wall. The comparison of Figure 5 and Figure 6 shows that due to the short fire duration of 180 min, only a very small fraction of the heat arrives at a depth of 30 cm. It is found that the use of different models (radial shell and linear layer model) leads to almost identical results. This distinction is important in geometry for simulations with much larger observation periods, such as climate calculations, or case of geometries implying small dimensions. Figure 5: Temperature in the vault at the beginning of the partition wall (Pos. 3 in Figure 1) Figure 6: Temperature in the wall at the beginning of the partition wall. (Pos. 3 in Figure 1)

6 Mass and volume flow and exhaust fan operating point As the air temperature increases rapidly, the air density in the exhaust system decreases. After 30 min, the density along the first half of the exhaust plenum is less than 0.8 kg/m 3, close to the fire it is significantly lower. At the fan station, the air density is > 1.0 kg/m 3. According to the mass conservation, a very large flow volume is achieved near the fire. This particular density leads to a higher system pressure loss, so that the fan runs on the characteristic curve in an area with higher pressures. The fans would need aerodynamic (distance to stall point) and engine power capacity to absorb these fluctuations of the operating point. This effect is partially compensated by buoyancy, which takes values in Faido of up to 1670 Pa. At Sedrun the situation is even better, as buoyancy pressures can reach up to 5000 Pa due to the 800 m high shaft. By neglecting the buoyancy, the operating point of the exhaust fan Faido would rise within 30 min to a total pressure of 7870 Pa compared to 6800 Pa without fire load. At that time, the first part of the channel is dominated by high air temperatures. Far from the fire, especially at the position of the fan, still low temperatures are found. At the GBT the influence of buoyancy due to the differences in height between the fan and the central tunnel would be sufficient to ensure the proper function of the exhaust system even in such a scenario. In case of the 40 MW fire load, the operation point of the exhaust fan Faido would rise to 7000 Pa, which is only 3 % more than without fire load. Figure 7: Volume flow in the exhaust duct Figure 8: Pressure drop, fan pressure, buoyancy 3.4. Heat flow In Figure 9 the released heat and the large heat flux into the concrete wall are shown. Accordingly, it takes a long time before the vault temperature has decreased. This effect was investigated in long-term simulations whereby even 5 hours after the fire had stopped, a heat flux of 12 MW was transmitted from the walls to the air.

7 MEASURES Figure 9: Heat flow into the wall and released heat The exhaust system was designed for the 40 MW fire load according to a passenger train. Based on the current results for the 40 MW fire load and further studies, various measures have been designed and implemented to protect passengers and infrastructure. Polypropylene Fiber Concrete was used in some parts of the exhaust duct. Where necessary, the concrete structure was protected by an insulation acting as a fire protection layer. However, when set up in large areas, this method prevents less heat from flowing into the concrete, resulting in higher smoke temperatures. All components of the ventilation system (dampers, fans) were designed so that they could fulfill 100 % of their capacity at the given temperatures. The fans have the necessary reserves of aerodynamic and engine power capacity to absorb the fluctuations of the operating point described above. 5. REFERENCES [1] Incropera, F.P., DeWitt D.P. (2002) Fundamentals of heat and mass transfer. John Wiley & Sons. Inc. 5th edition.