The effect of test conditions on the relative performance of cylindrical lithium-ion cells

The quickest and simplest way to benchmark cells against one another is through the comparison of datasheets and rapid standardised testing of cells. However, typical datasheets provide information based on cells tested under constant current load profiles in controlled ambient temperature. These test conditions are unrepresentative of most real applications. This work selects and benchmarks six 21700 cylindrical cells under four different test categories and two cooling conditions, to determine if relative performance of cells translates from standardised current and thermal conditions to more realistic currents and conditions. It is found that these cells perform very differently under different thermal boundary conditions, changing their capacity and thermal percentile rankings with respect to one another by up to 20 %. Furthermore, when tested under a drive cycle, many of these cells display percentile ranking changes of up to 30 % from a constant current equivalent cycle. This work demonstrates the need to test battery cells under accurate, application-focused conditions to correctly compare performance. Cell testing should use application-specific thermal and electrical setups, and application relevant load cycles. The assumption that the cell that performs best under standardised conditions will perform best under the real application is shown to be false. • Six cylindrical 21700 cells are tested under relevant electrical and thermal loads. • Capacity and temperature relative performance does not translate across conditions. • Linear relations can predict performance across varying power for a drive cycle. • Cell datasheets do not provide representative information for comparing cells. • Cells must be tested under conditions identical to the desired application.