Use of multiple regression analysis to develop equations for predicting Li-Al/iron sulphide cell performance

Abstract

A multiple regression analysis was conducted to develop predictive equations for the specific energy and specific power of Li-Al/iron sulphide cells over a wide range of cell designs and operating variables. The intent was to make these equations as general as possible such that one set of equations would predict the performance of Li-Al/FeS or Li-Al/FeS2 cells with bicell (one positive electrode and two facing negative electrodes) or multiplate cell configurations. Data from 33 cells were used in the analysis of specific energy, and 26 cells were used to develop the specific power equation. The calculated specific energy and specific power showed good agreement with the measured values for these cells. In general, the deviation between the calculated and measured values was within ±10%. A check of the predictive capability of these equations also showed good agreement. The specific energy and specific power calculated for 14 cells not used in the regression analysis deviated by ±10% from the measured values. These equations were used to identify the most likely cell designs to meet selected electric-vehicle battery performance goals. These designs were included in an experimental programme for further performance evaluation.