Optimization of the Flow Process in Engine Bays - 3D Fan Modeling Strategies

Licentiate thesis, 2009

In todays automotive industry there is an apparent need and demand for low fuel consuming vehicles. This is a fact every automotive OEM is aware of and the heavy duty truck industry is no exception. With a cooling fan in today’s trucks that typically consumes of the order of magnitude 50bHp fully engaged - reducing this loss is very important. Adding to this is increasingly heavy emissions legislation where a lot of the new exhaust gas aftertreatment systems requires more cooling. To fulfill this demand for cooling, increasing the ventilation and airflow through the engine bay serves as a cost effective way of at least partially reach this goal. Most easily is this done by a more powerful fan, however new legislations requires quieter fans, so today there is an apparent need for fans with higher efficiency. To design fans with a higher efficiency in complex environtments, CFD offers today a great opportunity and potential. However, computer power is still limited for solving the complete problem without any simplifying models, and this leaves the end-user heavily depending on the performance of these models. This is what is addressed in this thesis - an investigation of different fan modeling strategies and how these interact with other common CFD modeling strategies, such as turbulence models, mesh size and end-user working procedure. Main focus is kept on modeling the fan with the stationary MRF (Multiple Reference Frames) model or the transient Rigid Body Rotation model. These models are then compared to eachother on the basis of performance over time consumption. In the work comprised by this thesis it was found that it is very difficult to get the fan MRF model to perform well in a complete vehicle installation. This result was independent of choice in turbulence modeling and mesh size. The lack of performance in this model was found to be due to a limited space for fitting a valid rotational domain for the fan. It was also found that the choice of rotational domain played a major role in the behavior of this model. The MRF model was however found to be a consistent fan model in the sense that the classical fan laws still applies, hence from this conclusion it was proposed to correct the deficiency of this model with a simple speed correction. It was shown that a fan rotational speed increase of 14% was suitable for this type of simulations with a consistent choice of rotational domain. These 14% were tested and shown to apply to fans of different types, speeds and diameters. The fully transient (URANS) Rigid Body Rotation model performed well, even for small case sizes. It was found in this work that a small case of 3 million cells sliding mesh performed superior to a case of 16 million polyhedral cells with MRF, noteworthy is that the latter consumed twice as any CPU-hours for converging.