Numerical analysis and machine learning for battery thermal performance cooled with different fluids

摘要

A system of parallelly placed Li-ion batteries placed at different spacings and cooled by different coolants is analyzed. The prime focus is on the effect of spac-ings between the batteries on battery thermal performance. The temperature and heat flux at the coolant and battery interface is taken as continuous. Coolants used belong to thermal oils, gases, conventional oils, liquid metals, and nanofluids. Finite volume method-based numerical analysis is performed upon validation with experimental work reported in the literature, by developing a code in the C language. The simultaneous effect of battery aspect ratio, thermal conductivity, Reynolds number (Ref), and heat generation at different battery spacings for each coolant on average Nusselt number (NuH) is explored. An analytical model is developed and artificial neuro-fuzzy interference system (ANFIS) prediction of NuH is carried out later. Battery aspect ratio, thermal conductivity, and heat generation for any given coolant do not affect the NuH. However, the coolant Prandtl number at different battery spacings has a prominent effect on NuH. Thermal oils and liquid metal coolants show a significant and non-linear variation of NuH with increased battery spacings. The effect of Ref on NuH was linear as expected for all the coolants at different battery spacings. However, NuH variations for liquid metal at different Ref were highly non-linear. The mathematical model developed and ANFIS regression analysis indicated a very comfortable prediction of NuH. The R-squared value of the mathematical model is up to 0.98 while from ANFIS it is up to 0.985 showing a very good fitness of these models.