CFD and Neural Network-Based Predicting for Oil Delivery Performance of Oil Pump in Engines

Abstract

Oil delivery pefromance of oil pump in engines are investigated by a new prediction method based on computational fluid dynamic (CFD) and artificial neural network (ANN). CFD analysis was done by using Fluent commercial code and distribution of velocity of pump’s internal flow field was achieved by the solving of pump’s CFD model. Infromation data about oil pump’s rotate speed, supplying pressure, oil temperature and oil flow rate were obtained by CFD simulation analyzing. ANN model that used to describe the delivery performance of oil pump was employed, and the model was trained by learning samples from those CFD simulation results. Predicting for the delivery performance of oil pump under various operating conditions were carried out by this model. Experimental results were also used to validate the obtained simulation results. The studies show that the ANN(trained by CFD learning samples) predictions are in very close agreement with the oil flow obtained experimentally or predicted by CFD. The method introduced here can give useful supports for optimization designing of oil pump’s dimension.