A thermodynamic framework to rapidly determine remaining discharge time in Li-ion batteries

Abstract

The magnitude of accumulated entropy generation until complete discharge (AEGD) is applied to rapidly estimate the remaining discharge time (RDT) of lithium-ion (Li-ion) batteries. This approach operates on real-time prediction of RDT during a single discharge cycle and is applicable across diverse operating conditions and battery types. Experimental validation tests were conducted in 18650 and 27000 Li-ion batteries with different capacities and discharge rates. Additional verification test results are presented using independent data from 14500 polymer Li-ion batteries. The effectiveness of the proposed method is established with equivalent circuit model (ECM) and a machine learning (Random Forest) model using the same benchmark dataset. It is demonstrated that the method accurately identifies RDT for (i) variable operating conditions, (ii) from an arbitrary discharge voltage point, (iii) fluctuating voltage profiles, and (iv) for different temperature conditions ranging from 10 to 50 °C.