Underhood thermal management is a challenging problem in automotive industry. In order to make sure that vehicle works efficiently, there should be enough airflow through the cooling system so that the consequent heat rejection would be adequate. In idle condition the required air flow is provided by the cooling fan so a better understanding and an accurate predictive CAE tool for fan is very beneficial. Computational Fluid Dynamics (CFD) has been extensively used in predicting aerodynamic performance of automotive components. In the current work, the airflow performance of a fan, shroud and radiator assembly was simulated using Moving Reference Method (MRF) method. Although it is less expensive than Sliding Mesh (SM) method, the CAE results compare well with the test data. The simulation was carried out over 10+ different shrouds and the effect of geometrical parameters on airflow was investigated. The CFD data show that the smoothly converging shroud will lead to higher flow rates while cavities and steps will act as a restriction and degrade the efficiency. Besides, it is seen that decreasing the fan tip clearance up to 17 mm will improve the air flow as it prevents the leakage of the pumped flow, but if we go further in reducing the clearance, the airflow does not increase and may even decrease, which may be explained based on the interference of blade and shroud boundary layer. Extending the shroud will not help the air flow and it is recommended to immerse almost 70% of the fan into the shroud. This is a rule of thumb but may vary based on the details of fan assembly.