A CFD SIMULATION AND OPTIMIZATION OF SUBWAY STATION VENTILATION L Wang 1,4*, G Tu 2,4, T Zou 3,4 and J Yang 4 1 The School of Environmental Science and Engineering, Tianjin University, Tianjin, China 2 The School of Environmental Science and Engineering, Tianjin University, Tianjin, China 3 Dept. of Refrigerating and Air Conditioning Engineering, Tianjin University of Commerce, Tianjin, China 4 The School of Environmental Science and Engineering, Tianjin University, Tianjin, China ABSTRACT In rebuilding Tianjin Metro CFD simulations are performed to evaluate the original design of the ventilation system of Southwest Station and a new system is given to optimize the velocity and temperature fields at the station. Field measurements are conducted to validate the turbulence model and acquire boundary conditions. It is found that two-equation turbulence model is acceptable in simulating complex flow at the station. A method of simplifying the simulation of the transient flow at the station into steady flow is used and the validation criterion for this simulation is also given. INDEX TERMS CFD, Field measurements, Turbulence model, Validation criterion INTRODUCTION Subway has been a powerful tool for urban transportation since 1863 when the London Underground, the first subway system, began service in London, England. With the increasing population and the greatly developing economy of China subway is acting a more and more important role. Tianjin Metro, the secondly built subway in China, is going to be rebuilt to meet the local urban development and expected to serve in Beijing 2008 Olympic Games. Definition of problem Although different existent stations of Tianjin Metro have different geometries Southwest Station is the most typical one. Its geometry is shown in Figure 1. The platform mechanical ventilation is realized by two jet openings locating at each end of station and the supply air jets towards train and track. There is no mechanical exhaust system at the station and air is removed mechanically by tunnel fans and naturally by the exits of the station. The original design for rebuilding Southwest Station is shown in Figure 2. The design volume flow of Southwest Station is 400,000 m 3 /h. For most existent station the platform height is only 2.9m, which is too low to set ceiling ducts. So in this original design there are two grille vents at the each end of the platform to supply fresh air along the platform length direction and two grille vents to jet air breadthways towards trains. The design velocity of each lengthways grille vent is 5.54m/s and for each breadthways vent it is 5.28m/s. Under the platform 80 grille vents of the same design velocity (4.62m/s, 40 for each platform of the station) are responsible for exhaust. * Contact author email: lzwang@tju.edu.cn 836
Figure 1. The existent Southwest Station Figure 2. The original design for the new Southwest Station Although it requires relatively little structure rebuilding of the station experientially this design seems to have some problems. a) The speed of lengthwise vents is up to 5.54m/s, which is so great that it may cause unbearable speed at some points of the platform. b) The lengthwise supplied air may mostly directly flow outside platform through the exits so that the area between two exits may become Dead Zone where unbearable temperature may be caused. In this paper CFD techniques are used to evaluate the velocity and temperature fields of the platform and a new ventilation scenario with better thermal conditions is then suggested. Field tests at the existent Station are performed to acquire the boundary conditions for the later simulation. Preceding simplifications and presumptions Because of mechanical ventilation and the existence of train-driven wind the turbulence on the platform is transient and complex. While ensuring the reliability of the computation results some preceding simplifications and presumptions have to be taken. 837
a) Since the maximum air velocity is reached at the period when train stop at or start away from the station the period the simulation concerns about is from the point when at the section of x=0.0m (Figure 1.) the air velocity begin to change under piston-effect to the point when train totally stops at the station (defined as a Pulling-in Cycle). This transient process is simplified to steady process in which train has stopped and the air velocity of field measurements at these two ends of the station (x=0 m and x =74.4m in Figure 1.) and passenger exits (Figure 1.) are processed to be time-averaged when these velocity boundary conditions for later simulation are determined. b) The volume flow (V piston ) driven into the station by pulling-in train is determined by such factors as BR (Blocking Ratio, the ratio of train cross-section area to tunnel cross-section area), the length of the train (L train ) and the resistance of station R station etc. For existent and new stations BR are almost the same. Although the L train of the latter doubles the one of the former which may increase V piston, the R station of the latter is greater than the one of the former which may counteract this increase. So it is presumed that V piston is the same for both new and existent station and the volume flow through the passenger exits (V exit ) is also the same. Based on this presumption the results of the field measurements at the existent station can be used as velocity boundary conditions to predict velocity field of new station. METHODS Theory The flow at the station is taken as incompressible steady turbulence. Although generally the k ε turbulence model is not very sound to predict turbulent flow, the validation of the simulation at the existent station (as shown in Figure 3) shows that this model can be used. The basic governing equations are the Mass Conservation Equation, Momentum Equation, the Transportation Equation of k, the Transportation Equation of ε and Energy Equation. The governing equations are integrated on the spatial control volumes. Except that Pressure uses body force weighted scheme other dependent variables (momentum, k and εetc) use second-order upwind schemes when they are computed at the cell faces. Pressure and velocity are coupled by SIMPLE Algorithm. Field measurements At the existent Southwest Station the Multichannel Anemomaster Hotwire Anemoscope is used to measure air velocities. The layout of measuring points and the locations of measuring sections are shown in Figure 1. Once each two second the anemoscope scans all points and automatically print the results. All data are recorded during a complete Pulling-in Cycle. Then these data are processed to yield time-averaged velocity of per section. Infrared Thermometer is used to measure the temperatures of the walls of the station which are taken as the constant temperature thermal conditions in the simulation. Computation and simulation Full-scale models of the existent and new station are created. First by taking the results of the field measurements as boundary conditions the air velocities of the existent platform are computed and the velocities of field measuring points are used to validate the two-equation model and the simulation. Some boundary velocities are unmeasurable such as the Section U in Figure 1. So the flow at this section is estimated by trial computations. Secondly these boundary conditions are applied to the simulation of the original design of the new station. Then the suggested ventilation scenario is computed under the same conditions. The temperature fields are only computed in both the original and the suggested design at the new 838
station because the heat load at the existent station is not typical and difficult to be determined. Since it is impossible to get exact values at every point and all computed velocities are timeaveraged, the simulation is acceptable if the average velocities at corresponding cross-sections approach the time-averaged measured velocities. This is the validation criterion of the simulation. RESULTS Figure 3. Validation of the simulation at the existent station (see Figure 1. for section locations) Figure 4. Predicted point velocities (y=3.25m, the original design, the new station) Figure 5. Predicted point velocities (y=3.25m, the suggested design, the new station) Figure 6. The suggested ventilation system (the new station) 839
DISCUSSION Figure 3 shows that the two-equation model and the preceding simplifications are acceptable in predicting flow at the station. Although the computed velocity at Section C and E don t equal to the values of the field measurement, the difference are acceptable and the overall variance trend of the velocity along the platform agrees with the measurement. In the simulation of the original design for the new station as shown in Figure 4 it is apparent that at the height of 1.7m above the platform the velocity of some spots reaches to 3.57m/s while beyond the exit the air speed reduces to 0.35~0.65m/s. This can be explained by the fact that much of the fresh air directly flows out of the station through the exits. Although the velocity of the zone between the two exits is not too low the ventilation efficiency of the original design is thus unsatisfactory. The simulation also shows that the maximum velocity exists at the exits, which agrees well with the references and the experience. In the suggested design at the height of 1.7m above the platform the maximum velocity is only 1.68m/s and the overall velocity field is relatively balanced (refer to Figure 5). The supplied air from the breadthways vents also very well prevents the hot contaminated train - driven air to spread onto the platform. Although there are still some zones where the velocity are only 0.2~0.4m/s they can be compensated by the adjacent zones where the velocity is higher. On the other hand the fresh air from the breadthways vents passes through the passengers, takes away heat and humidity, collects and is exhausted by the under-platform vents. The ventilation efficiency is quite satisfactory. From the references the heat loads from the train condenser and the brake resistance respectively are about 320kw and 200kw which are the greatest heat sources. From people and lighting they are 44kw and 30.6kw. The temperature of fresh air, tunnel air and the station walls are 27, 25 and 24 respectively (the outdoor ventilation design temperature in summer is 27 according to the references). Under the above thermal boundary conditions the simulations show that for the original design the temperature on the platform is about 32~34 and for the suggested design it is about 28 when the whole station flow remains the same. The design temperature of the original system is 30. So it is apparent that the suggested system is also much better in satisfying the design requirements than the original one. Since under the temperature of 28 passengers may feel uncomfortable when they flow into the station from the outside environment of higher temperature it is very promising to save energy in the suggested design by reducing the whole station flow and increasing the platform temperature. Technically this new design is also applicable. In Figure 6 under the platform the duct with the cross-section of 2.0m 1.6m is used for the breadthways ventilation. The maximum air velocity in the duct is about 14m/s, which is acceptable for brick ducts in subway. 840
CONCLUSION AND IMPLICATIONS It is applicable to use two-equation turbulence model to predict time-averaged flow at subway station. The simulation results are influenced greatly by the boundary conditions such as boundary velocity and pressure. As for the transience of the airflow at the platform it is also feasible to simplify it into steady flow in order to get time-averaged velocity field. In the above simulation the field measurement results at the existent station are applied into the simulation of the new station. It is suggested that careful analysis and comparison should be conducted to ensure the applicability of this method. Some corrections may be used if necessary. The breadthways ventilation have more advantages than the lengthways ventilation. The former is suggested when the height of platform is too low to have ceiling ducts. Before a design is accepted it is very necessary to evaluate and optimize it by such techniques as CFD etc. They will help a lot to build a reliable, effective and economical system. ACKNOWLEDGEMENTS Thanks for the great help from Senior Engineer Hong Liu at Tianjin Metro Company, Tianjin, China and Professor Shijun You at Tianjin University, Tianjin, China. Also thanks very much to those classmates at Tianjin University who helped to conduct the field measurements at Southwest Station. REFERENCES No. 3 Design Institute of China s Ministry of Railway. 2001.The Feasibility Research Report of the 1 st line of Tianjin Metro. Tianjin, China: No. 3 Design Institute of China s Ministry of Railway. Suhas V. Patankar. 1980. Numerical Heat Transfer and Fluid Flow. New York: Hemisphere Publishing Corporation. Transit Development Corporation. 1975. Subway Environmental Design Handbook. Vol. I, pp.150-164. Wenquan Tao. 1986. Numerical Heat Transfer. Xi an: The Publishing House of Xian Jiaotong University. 841