Search Results

Computational fluid dynamics has been used to study the flow pattern in a Volvo S80 passenger compartment. The main purpose of this work is to secure a method for future use of CFD in developing climate control systems in cars. The effects of mesh resolution and mesh size were studied by varying the number of cells from 1 million to approximately 5 million. It was found that at least 2 million cells are needed to approach a mesh size independent solution. The other focus of this study was the outlet boundary conditions. Since a passenger compartment is not air tight, outlets were assumed to be around doors, through the floor, through the backseat, as well as the evacuation at the rear of the passenger compartment. It can be seen that the solution is only sensitive to drastic changes in the leakage.

Flow simulations of an air distribution system have been carried out using the CFD code FLUENT/UNS [1]. The purpose of this study is to validate this complex flow problem versus experimental data. Two modes of the climate system are investigated; the Ventilation mode and the Floor/Defroster mode. The complete geometrical model contains all ducts, central unit, heat exchangers, defroster and nozzles of the air distribution system. A high level of geometrical detailing in the mesh, consisting of 2.1 - 3.3 million cells, is used. The study shows that CFD has a potential to give reliable results, even for complex systems, like air distribution systems, if used in a controlled manner.

For the computation of the flow around a car there is typically an overprediction of stagnation and base pressure coefficient by 0.02 and 0.07 respectively causing an underprediction of total drag by about 0.01 - 0.02. To determine the cause of the error in stagnation pressure several different effects have been investigated: length of inlet section, boundary conditions on the floor, mesh resolution, turbulence models and different flow solvers. One major effect was found to be a short inlet section in the computational model, which caused an overprediction of the stagnation pressure. The second major effect was insufficient resolution of the mesh along the stagnation line. More than 100 nodes along the stagnation line were necessary to avoid a significant pressure drop. Finally the k-ε turbulence model caused overprediction of total pressure very close to the stagnation point. The use of an explicit algebraic Reynolds stress model removed this error.