CFD Method Validation: Simulation of Defroster Performance Testing

Abstract

An accurate prediction of the ice melting pattern would be useful for the performance evaluation and energy optimization during the design phase of the vehicle defroster. It can also be used to ensure that the design passes the legal requirement as well. A transient CFD simulation method suitable for defroster performance simulation is therefore verified and validated in this study. The aim is to achieve as accurate and efficient simulation as possible for the use in industry. In this study, the geometry is first prepared in ANSA and the simulation and post-processing are executed in STAR-CCM+. Fluid film model is used to model the ice layer on the windscreen and its melting and solidification model plays an important part in obtaining the ice melting pattern. The model is first verified regarding the optimal mesh setting, time step size, flow solver setting and the choice of turbulence models. It was found that the mesh and turbulence model can affect the simulation stability greatly. Realizable k-epsilon, k-epsilon Lag EB and SST k-omega model are examined and Realizable k-epsilon is selected as the most suitable one for this simulation. The time step size and flow solver can then be adjusted to maximize the solution efficiency. With steady state nature of the flow field, the time step size can be increased up to 5 or even 15 s depend on the required level of accuracy. With the flow solver frozen after a steady state initialization, the solver time per time step can be further reduced by 30 – 35%. With 15 s time step size, the 42 minutes testing time transient simulation can be executed within 2.5 hours with 960 computational cores and still yields results with acceptable accuracy. Next, to improve the result accuracy, boundary conditions and the domain geometry are validated against the test data. It was found that the velocity and turbulence profile at HVAC-defroster connection is needed to achieve accurate flow impingement pattern and ice melting pattern on the windscreen. The ice layer thickness can also affect the melting rate considerably so it should be verified in future tests. In contrast, it is apparent that the simulation is not too sensitive to the defroster inlet temperature profile and so a variation within 5 – 10 ℃ is acceptable for future application. With these verifications, highly accurate ice melting pattern is achieved for both the windscreen and front side windows at the end of this study. The application of this final scheme on another vehicle model also confirms the method reliability.