Performance Benchmarking of Lithium-ion Batteries: The Effect of Test Conditions on the Down-Selection of Cylindrical Cells

One of the most critical steps in designing a Li-ion battery pack is choosing the optimal cell for the application. There is no established scientific process for determining the best cell for a given application. It is acknowledged that thermal boundary conditions and electrical load profiles can greatly affect the performance of Li-ion cells, yet data specification sheets and cell benchmark databases only offer comparisons under standardised ambient conditions using standard electrical loads.

This study aims to identify whether cell performance rankings made under one set of conditions, defined by the cooling and load specifics, translates to cell rankings obtained under other conditions. This work selects a series of beginning of life (BoL) performance tests, designed to represent different types of use. These tests are as follows:

• 1C Discharge
• 5A Discharge
• 1C Charge & 2C Discharge Cycle
• Drive Cycle, formed by a repeated 3-step profile comprised of rest (6s), base discharge (8.25s, 27W) and pulse discharge (0.75s, 63W), with an average current of approximately 5A

Each test is performed while the cells are under two different cooling conditions, base cooled (cooling on) and natural convection (cooling off). The cells are wire bonded, attached at their base to water cooled blocks and enclosed in an insulated box. Six different 21700 cylindrical cells of varying energy/power densities undergo the same series of tests, allowing a comparison of their performance.

A ranking system is defined to score the cells based on the performance of the cells in each combination of test conditions, rather than across all of them. The cells are ranked for capacity extracted under each condition (100 – highest capacity, 0 – lowest capacity), and for thermal performance, which is an average of the maximum measured temperature and the axial thermal gradient (100 – lowest temperature and thermal gradient, 0 – highest temperature and thermal gradient).

It is found that comparative performance between cells can change by up to 20% between the different cooling conditions. Cells display comparative performance changes of up to 30% between a drive cycle and a continuous load equivalent. This results in different cell rank depending on the specific set of test conditions. Further to this, it is shown that using the datasheet to make initial judgments on which cells should perform the best leads to inaccurate predictions. Datasheets do not provide a meaningful reflection of real performance of cells. This is partially due to the lack of standardisation and definitions, but also as shown in this work, due to the dependency of cell performance on specific test conditions.