In English

Numerical simulations of generic train in crosswind using AVL Fire

Per Ringqvist
Göteborg : Chalmers tekniska högskola, 2011. Diploma work - Department of Applied Mechanics, Chalmers University of Technology, Göteborg, Sweden, ISSN 1652-8557; 2011:67, 2011.
[Examensarbete på avancerad nivå]

Many modern trains today have distributed engines in every wagon throughout the train making the leading wagon relatively light. This can cause a risk of derailment, when traveling at high speeds and exposed to strong crosswind or sudden gusts of wind. To study the forces acting on a train as a result of side wind, low Reynolds number experiments on a simplified train geometry has been conducted at the RTRI institute in japan. The experiments includes both stationary yaw angles as well as a train moving through a region of crosswind. The experimental data consists of pressure measurments at 38 locations distributed along the train body. The simulated scenarios consists of a stationary train exposed to side wind of 5 m/s at yaw angle 90 degrees (Re = 16000), as well as a train moving at 5 m/s perpendiculary through the side wind (Re = 23000). To simulate the moving train a moving velocity profile is used to as a model for the side wind. PANS and LES are run on identical meshes for both cases. The standard Smagorinsky model is employed in the LES. For the RANS part of PANS the ζ-f-model is used. The parameter fk defined as the unresolved to total ratio of turbulence kinetic energy, is recomputed before each timestep. The mesh for the stationary case is slightly to coarse for LES, which makes it possible to see the advantage of PANS. All simulations are performed using the commersial software package AVL Fire. Good agreement with the experimental data is found with both LES and PANS for the stationary case. However, PANS makes a substantially better prediction of the pressure at the lateral ends of the train, where the spanwise mesh-resolution is coarser. In the moving case LES and PANS displays very similar results, having some discrepancies but showing fair agreement to experimental data. The computing of fk assumes a stable time averaged flow, which is the probable reason why PANS does not show the same advantage in the moving scenario.

Nyckelord: PANS, LES, train, crosswind



Publikationen registrerades 2012-01-16. Den ändrades senast 2013-04-04

CPL ID: 153126

Detta är en tjänst från Chalmers bibliotek