SKODA designs and manufactures railway vehicles - electric locomotives, suburban trains and underground train units. SKODA places great effort in the design of railway vehicles that meet requirements for comfort and safety of passengers as well as adequate driving time. When trains are travelling with speeds of 250 Km/h, the effect of interaction with other trains and/or the external environment, require investigation to improve vehicle stability, aerodynamic noise and performance. All these factors are influenced by the aerodynamic forces on the vehicle. SKODA designers were interested on the influence of the front and rear locomotive nose shape and the influence of the gap between the locomotive and the first wagon on the aerodynamic loading. To assess the loading on their new designs, they used CFD for analysis and optimisation. Results from the complex CFD analyses of train aerodynamics included values which describe aerodynamic loads on the vehicle surface, flow wake - behind a train, behind the locomotive and pressure details on the locomotve surface.

SKODA used APUS-CFD to investigate the aerodynamic loading of their train designs when travelling in a tunnel.

A tetrahedral mesh was generated with 1.55 Million cells and the travelling speed of the train, for the results presented, was 200 Km/h. The dimensions of the model, together with the boundary conditions set are presented below.

Results obtained with APUS-CFD were compared and validated against previous computations with other commercial CFD software. The pressure distribution on the locomotive surface is illustrated below.

The figure below shows the velocity distribution on the symmetry plane in the form of contours. You can see the high velocities (red patch) at the rear-top of the locomotive.

Furthermore, the velocity distribution along the tunnels at y=4.6m (just above the train) in shown in the following plot, together with comparisons against another calculation.

Velocity contours at the front	Velocity contours in mid-region

The parallel version of the APUS-CFD solver was benchmarked and tested during the verification exercise. Execution times, for a mesh of 1.55 Million cells are shown in the chart below. The effect of number of processors on performance was assessed and the numerical efficiency of the solver was observed. For all the tests performed the solver converged with almost the same number of iterations, regardless of the number of processors used (from 1 CPU to 64 CPUs). The times (in mins) shown in the chart represent runs performed with the same number of iterations (2000).
The performance and efficiency of the parallel solver are presented in the bar-chart for up to 64 processors.
APUS-CFD was benchmarked on a Windows XP cluster (AMD Opteron 3.0GHz) consisting of two servers. Each server had 2-CPUs and were connected with an 100 Mbits/s ethernet switch (blue-coloured bars). The 16, 32 and 64 CPUs system was the SGI Altix 3700 Bx2 running Linux (orange-coloured bars).

Performance and Efficiency of APUS-CFD

Automotive
Marine
Motorsport
Power Generation