This paper focuses on the applicability of numerical prediction of sound radiation caused by an axial vehicle cooling fan. To investigate the applicability of numerical methods, a hybrid approach is chosen where first a CFD simulation is performed and the sound radiation is calculated in a second step. For the acoustic simulation an integral method described by Ffowcs-Williams-Hawkings is used to predict the sound propagation in the far-field. The simulation results are validated with experiments. The corresponding setup in experiments and simulation represents an overall system which includes the cooler, the cooling fan and a combustion engine dummy. To optimize the economical applicability in terms of simulation setup and run time, different approaches are investigated. This includes the simulation of only one blade using a periodic boundary condition as compared to the whole fan geometry. In the CFD simulation an SAS-turbulence-model is applied. The results show that this is a very useful approach considering the challenges in prediction of numerical sound. On the one hand, the turbulence model has to solve small scales which cause sound in high frequency ranges which leads to a small time increment and a high grid resolution. On the other hand, a long simulation time for predicting sound in low frequency ranges is needed. Additional to the sound propagation in the far-field, it is a benefit to be able to localize the acoustic sources with regards to geometrical optimization. For this, the divergence of the Lighthill Tensor is calculated on the CFD grid.