Battery chemistry prediction with short measurements and a decision tree algorithm: Sorting for a proper recycling process

Abstract

Given the push for electrification in every front, the demand for batteries is going to keep increasing. The current rates of growth for production of raw materials are not expected to keep up with this increasing demand. Therefore, recycling has to be a significant part of the value chain in providing the raw materials. Presorting the batteries collected prior to a recycling operation would greatly improve the efficiency of the process. Given users of consumer batteries cannot classify their spent batteries by their chemistries, a chemistry identification will need to be implemented at the industrial scale. We are reporting on a decision tree that is based on a density measurement along with electrical measurements in order to classify batteries by their chemistry in addition to a provision for reusability. 109 batteries are measured in total from a battery recycling bin, including batteries from commercial retailers that were used for various applications.