New Technologies applied to the design and optimization of tunnel ventilation systems Ing. Justo Suárez Area Manager - ZITRON Dot. Ing. Massimiliano Bringiotti Managing Director - GEOTUNNEL Ing. Ana Belén Amado Responsible of CFD and Virtual simulation - ZITRON ABSTRACT: Construction of new transport infrastructures is being a social demand during the last decades; therefore many countries have decided to carry out important investments on this field. Among these transport infrastructures, road and railway tunnels and Metro lines are the most significant ones. The amazing improvements and fast growing technology in mechanized tunnel excavation with TBM (Tunnel Boring Machine) has made possible the excavation of tunnels with several tens of kilometers. Subsequently, the technology applied in the electro-mechanic installations to be installed in these tunnels, such as the ventilation, should also be improved. This is a serious challenge in front of us, in order to provide safe and efficient solutions to new and complex problems related to the passenger s safety in these very long transport tunnels. This paper is a general approach to the new technologies that are being currently applied for the design and optimization of tunnel ventilation systems. KEYWORDS: Tunnel, Ventilation system, Computational Fluid Dynamics, Fire, Safety. 199
1. INTRODUCTION During last years the length of the transport tunnels (Road and railway) has increased significantly. Not so long time ago, any tunnel over 2 km length was considered as a long tunnel. Nowadays, tunnels over 10, 20 even 40 kms are being constructed and some of them are already in operation. Technology for the construction of these long tunnels has also developed tremendously. The use of TBM is now considered as the standard method of excavation, so there is almost no limitation on the length of the tunnels to be excavated. In parallel, the electro-mechanic installations for these new very long tunnels should be improved and adequate rapidly to the new tunnel conditions (long tunnel distances, very high volume of traffic and passengers simultaneously in the tunnel, ) otherwise we could arrive to the following paradoxical situation: the length on the new tunnels to be constructed will be restricted, not due to limitations on the tunnel construction technology, but because the electro-mechanic installations will not able to satisfy minimum conditions of comfort and safety for the passengers. Among all the electro-mechanic installations in a tunnel (lighting, communication system, video and surveillance system, fire detection and extinguish system, ), the ventilation is one of the critical ones. Ventilation system in a tunnel (road and railway) is directly related to the comfort and safety of the passengers. Additionally, from an electrical point of view, ventilation is the most energy demanding system. Therefore, the design of a safe and high efficient ventilation system becomes a very important task to be studied carefully. To achieve this goal, new technologies, such as CFD (Computational Fluid Dynamics), 3D scanners, should be applied to fan design. Also, to verify this optimised design, fan manufacturers should offer the possibility of aerodynamic fan testing at real scale and full power. 2. BASIC PRINCIPLES FOR FAN DESIGN The main input data for the aerodynamic fan design are: duty point/s (Air Flow Vs Pressure), efficiency and reversibility 200
Based on the above data, fan designers should carry out the fan selection, which means to select the most aerodynamically optimised and less power consuming fan that satisfies these requirements. This is a challenging work that involves the mechanical design of all different fan elements. Among all of them, impeller and guide vanes are the most affecting ones to the fan performance (Air flow and Pressure) and to the motor consumption (efficiency). The aerodynamic profile of the impeller blades and guide vanes should be designed to optimize the fan aerodynamic performance at minimum power consumption. The objective is to achieve the highest lift force (F L ) with the lowest drag force (F D ). Figure 1. Layout showing Drag and lift forces When the blade angle of attack is adjusted to move air, a force (F) is affecting the blade. This force F can be divided into two main components, one in the same direction as the air, called drag force (F D ), and another with perpendicular direction, called lift force (F L ). The shape of the upper part of the blade causes an increase in the air velocity locally and therefore a reduction in static pressure, generating then lift force. If the angle of attack is increased in exceeds about 20º (depending the blade profile) a severe flow separation occurs, and the drag force increases rapidly, this phenomenon is known as stalling. Another general rules to take in account in fan design are: 201
Number of impeller blades: More blades More pressure Impeller diameter: Bigger impeller diameter More air flow Hub diameter: Bigger hub diameter Lower air flow and higher pressure 3. COMPUTATIONAL FAN AERODYNAMIC DESIG (NUMECA ) The modern fan design is assisted by CFD (Computational Fluid Dynamics) technology. One of the most powerful software on this filed is NUMECA. As a developing science, CFD (Computational Fluid Dynamics) has received extensive attention throughout the international community since the advent of the digital computer. CFD interests are mainly driven by the desire to model complex physical fluid phenomena, which couldn't be easily or cost effectively, simulated with a physical experiment. Computational techniques differ from analytical or theoretical solutions in the sense that they only solve equations at a finite number of points rather than for the entire flow field. Choosing these points, of the entire flow field, may become quite difficult, especially for a complex geometry, and it may require hundreds of thousands or even million of points. In general, a dense grid with many points will give a solution of great detail, but require more computer resources and time to reach a solution. Since a compromise between computer resources and solution quality is required, the current trend is often to use a dense grid in areas where the solution may change rapidly such as in the boundary layer or near a shock wave, but the use of a coarser grid with fewer computational points in areas where the solution is expected to change more gradually. The resolution of Computational Fluid Dynamics (CFD) problems involves three main steps: spatial discretization of the flow domain, flow computation and results visualisation. 3.1. Example of application CFD was used to obtain the performance characteristic curve of a new fan model previously selected, with the following main features: fan diameter (Ø 2000 mm); hub diameter (Ø 1200 mm); power (160 kw) and rotation speed (990 r.p.m - 50 Hz - 6 poles motor) To prepare the geometrical model used in the CFD simulations the fan is divided into 2 different parts: Rotor (Impeller) and Stator 202
Rotor This part includes the rotating part of the fan, that means, the impeller. To generate the rotor block has been used the following data: Fan diameter (Ø 2000 mm), hub diameter (Ø1200 mm), quantity of blades (12) and tip clearance (7 mm). The first step is to generate the blade profile in 3D format. Figure 2. 3D impeller blade profile, showing pressure side and suction side Stator This block simulates the static part of the fan, downstream of the impeller. In this block the main element are the guide vanes. To generate the stator block, it was necessary the following data: fan diameter (Ø 2000 mm), hub diameter (Ø1200 mm) and quantity of guide vanes (11). Also the guide vane profile is generated in 3D format. Figure 3. 3D guide vane profile. Left is pressure side and right is suction side. 203
3.1.1. Step 1: Spatial discretization of the Air Flow Domain The quality of the grid is definitely important, since it strongly influences the solution including whether or not a solution can be found at all. What determine the grid quality are mainly two parameters: level of skewness (or orthogonality) and expansion ratio. It is important to minimize the number of cells (finite elements) containing a high level of skewness (cells with very low internal angles) because the calculation of fluxes can become significantly erroneous under such conditions. By other hand, it is particularly important, in regions of high gradients, such as boundary layers, free shear-layers and shocks, to keep the expansion ratio (the ratio of adjacent cell side) within an absolute range of about 0 to 1.6. To achieve a good grid quality in all Air Flow Domains, it was necessary divide each part of the total domain (rotor and stator) into smaller blocks. The mesh and the shape of some of these blocks are shown in the following figures. Figure 4. Spatial discretization of several blocks of Air Flow Domain on the rotor part 3.1.2. Step 2: Flow Model To define the project flow model, it is necessary to choose: fluid model and turbulence model. The fluid model used in these simulations is air with the specific physical properties (heat conduction, dynamic viscosity, temperature, ). Also, it is chosen a turbulence model that simulates the behaviour of complex flows. 3.1.3. Step 3: Flow equation resolution and boundary conditions The process to obtain the complete fan performance curve involves also several flow equations that have to be solved for different values of flow rate, so it is necessary to run as many calculations as fan performance points (Flow Vs Pressure). 204
Also boundary conditions have to be set, such as: inlet mass flow corresponding to the flow rate that will be calculated, outlet static pressure corresponding to atmospheric static pressure and rotational speed corresponding to motor speed. 3.1.4. Step 4- Results: Predictive Fan Characteristic Curve, post-processing and graphic results. Once the CFD simulation is carried out, the predictive fan performance curve is obtained. Figure 5. Final results for fan simulation and predictive fan performance curve Although the aim of this computer simulation is to obtain the predictive fan characteristic curve, it is also very interesting to analyse other graphic results that the post processing of the flow solutions by means of CFD allows to obtain, such as 2D and 3D images of pressure and velocity distributions, vector fields, flow lines or turbulence inside of the flow domain. Figure 6. Static pressure distribution Figure 7. Air flow lines along the fan 205
4. BLADE VERIFICATION WITH A 3D SCANNER To achieve a good performance and high efficiency values on a fan, the impeller blades should be well design and optimised. To make effective this virtual design, the real blades should be verified, checking that their shape is inside tolerance with the designed blade shape. A 3D scanner should be applied to achieve this goal. This verification has two steps: 3D scanning of the blades and shape comparison software to verify that the real (scanned) blades and the theoretical blades are identical. Figure 8. Step 1-3D blade scanning Figure 9. Step 2 - Blade comparison (real scanned Vs theoretical) 5. FACTORY TESTS 5.1. Aerodynamic fan test The objective of this tests is to verify that the manufactured fan achieves the predictive performance curve (Flow Vs Pressure) obtained previously by means of NUMECA software. During this test it is also verified fan efficiency, power consumption, noise level and vibration values. 206
These tests are carried out according to international recognised standards, such as AMCA (Air Movement and Control Association). Figure 10. Layout of Aerodynamic test bench accredited by AMCA. Any deviation or possible problem would be detected during this test and corrected before fan delivery. 5.2. High temperature fan test Fans used in emergency tunnel ventilation systems should be able to operate in high temperature conditions, in case of fire. To certify that the fans are able to operate properly under high temperature conditions is required to carry out a test. These high temperature tests have to be carried out in accredited laboratories, according to European standard EN 12101-3. The following minimum condition should be observed during test performance: The fan should be able to operate at the high temperature (up to 400 ºC) during a period of 120 minutes. Heat up ramp from ambient temperature to test temperature must be achieved in short time, that is within 5 and 10 minutes. After 15 minutes running the high temperature test, the fan has to afford being switched off (during 2 minutes) and restarted again. The average value of the fan air flow during the high temperature period must be in the range of the nominal values (not beyond of 10% difference). 207
Figure 11. Photo and scheme of high temperature test bench for tunnel axial fans 6. OPTIMIZATION OF COMPLETE VENTILATION SYSTEM USING CFD (FLUENT ) Simulation of tunnel ventilation systems is challenging both because of the complex physics involved and the large domain sizes required for tunnels that may run for several kilometres. Even so, comparison with physical testing results shows that computational fluid dynamics (CFD) can accurately simulate flow patterns and pressure drop in tunnels of any length. Continual improvements in CFD software and high performance computing hardware have made it possible to optimise the design of increasingly challenging tunnel ventilation systems. One of the most powerful and commonly used CFD for tunnel ventilation is FLUENT. Given the size and complexity of tunnel geometry, it is difficult and expensive to build and test scale models to evaluate potential ventilation system designs. Simulation makes it possible to evaluate the performance of alternative configurations and to determine how the system should be operated under normal and emergency conditions all in less time and at a lower cost. Figure 12. Graphical result of FLUENT simulation in a tunnel ventilation system 208
Fire simulations using CFD software are useful for optimization and validation of tunnel ventilation systems. A fire growth model is defined within the tunnel in a location strategically chosen to account for worst-case conditions. The designed tunnel ventilation system is activated with a chosen time delay from the moment that fire starts. The simulation predicts the performance of the ventilation system in order to control the smoke towards the fumes exhaust areas and prevent the back-layering effect. The passenger evacuation areas in the tunnel should be kept smoke-free during the assumed evacuation time. Other key simulation outcomes are velocity distribution and temperature maps along the tunnel. In CFD simulation, meshing is a very important issue, particularly in our case of tunnels that can extend several kilometres. There are critical areas, such as the fire location and the areas upstream/downstream of fans, where the mesh must be fine to accurately capture the physics. In other areas, cells can be larger to reduce the total cell number and reduce the computational and processing time. Accurate simulation of a tunnel ventilation system today typically requires about 1 million cells per kilometre of tunnel. High-performance computing and recent advances in parallel algorithms have enabled quasi-linear scalability of fluid dynamics calculations. Continual development in both hardware and ANSYS FLUENT software have contributed, and will continue to contribute, in solving more complex and higher fidelity tunnel ventilation systems. High-performing hardware allows the clustering of processing units in a cost-effective way, while more efficient computational methods permit prediction of more complex physical phenomena in longer tunnels. The final result is an accurate simulation of complex tunnel ventilation systems in less computational time. 7. CASE STUDY: VENTILATION SYSTEM FOR GUADARRAMA TUNNELS (2X28 KM) Guadarrama tunnels are the most important infrastructure on the high speed railway line between Madrid and the north-west region of Spain. They are the fifth longest tunnels in the world and entered into service on 23 rd December 2007. They consist in two parallel tubes, one for each direction in which the high speed train runs, with a length of 28.4 kms and a section of 52 m 2. The most innovative characteristic of Guadarrama tunnels is the fact that the excavation was made with 4 TBMs starting simultaneously from the tunnel portals and breaking through 209
at approx. 15 kms inside the mountain, that means there is not any intermediate shaft or access galleries. This is something unique for this kind of very long tunnels. From one side, the excavation costs and time of construction is reduced significantly, but on the other hand, the design and installation of an effective ventilation system becomes a very difficult and challenging task. The tunnels are interconnected by means of emergency passageways every 250 m and a 500 m long emergency area, located at an equidistant point from both tunnel portals. In case of tunnel evacuation, this emergency area can hold up to 1200 people Originally, the emergency ventilation system was designed considering two main ventilation plants located at both tunnel ends, with closing doors at the tunnel entrances, in order to pressurise one of the tunnels, in case of emergency (i.e. a fire event) in the other parallel tunnel. However, after a detailed risk analysis of this original ventilation design, it was decided to reject it, due to the problems associated with a possible failure in the opening and closing of the doors and the catastrophic effects in case of crashing the high speed train at over 250 km/h. Subsequently, it was decided to carry out a complete re-design of the emergency ventilation system. The idea was to replace the tunnel doors by a pneumatic closure system using jets of air. The key point for a successfully re-design of the emergency ventilation system was the use of CFD (Computational Fluid Dynamics). Using this innovative technology, it was possible to perform several computer simulations in order to achieve the most effective design for the tunnel ventilation system. The civil construction of all structural elements involved in the tunnel ventilation system, have been carried out following strictly the results of the CFD simulations. Figure 13. CFD meshing model Vs real photo of the subsequent ventilation constructions 210
It should be especially remarked the optimization in the air flow injection angle of the Saccardo nozzle located at the top of the false tunnels constructed on the four tunnel portals. Figure 14. CFD meshing model Vs real photo of Saccardo nozzles located above tunnel portals. LITERATURE [1] European standard EN 12101-3 Smoke an heat control system Part 3: Specification for powered smoke and heat smoke ventilators. [2] Fire and Smoke control in road tunnels. PIARC committee in road tunnels (C5) 2111