A hybrid model for simulation of lithium-ion batteries using artificial neural networks and computational fluid dynamics

Abstract

Chemical reactions inside lithium-ion batteries generate heat and cause temperature rise. Hence, it is necessary to monitor battery time dependent heat generation. In this work, a hybrid model for simulating heat generation inside a pack of lithium batteries has been developed. An artificial neural network (ANN) has been employed to simulate electrochemical and thermal behaviors of a Panasonic NCR 18650 lithium-ion battery. In order to develop the hybrid model, the designed ANN has been inserted into ANSYS Fluent software through a C source code. A 3-D computation fluid dynamics (CFD) has been developed to simulate temperature distribution in the battery pack. Experimental data has been obtained using a NEWARE battery test system at different C-rates. The outputs of the proposed ANN consist of heat generation inside the battery as well as the electrochemical parameters. The combination of the ANN and CFD modeling, which led to a hybrid model, can be mentioned as the major contribution of this work. The results show an excellent consistency between the proposed model and test data. The simulation estimates the range of the manufacturer’s working temperatures (-20 to 60 oC), regarding the considered batteries.