Use of Ultrasonic Pulses to Detect State of Charge and State of Health in Lithium-Ion Batteries

The rapid growth in the adoption of lithium-ion batteries since their introduction in the 1990's is expected to continue in the coming years. Electric vehicles in particular will account for a large market share, being considered clean transportation technology compared to the internal combustion engine. Lithium-ion batteries change their internal state during cycles of charge and discharge. Estimation of the state of charge is commonly performed by battery management systems that rely on charge counting and cell voltage measurements. Determining the physical state of the battery components is challenging.

Recently, the response of an ultrasonic pulse through a battery has been successfully correlated with both change in state of charge and state of health, the approach is now well established. This study assesses the qualities contained within an ultrasound signal response by investigating the behaviour of ultrasonic waves as they pass through the components in a layered battery structure, as those components change with battery charge. This is possible based on acoustic sensitivity to material property changes such as electrode density during charge cycling. Captured reflections from pulsed signals will accumulate in the signal response, with observed differences providing the potential for real-time, non-invasive, non-destructive measurements of internal changes in the battery cell.

A model has been developed to understand the nature of the ultrasound response and the features that provide a particular characteristic. Modelling was based on the one-dimensional wave equation, scripted in MATLAB. Layered properties simulating a battery cell are required to run the model, the outputs of which include simulated wave responses, wave development plots and animations.

Experiments were conducted to establish methodologies in detecting a relationship between battery state of charge and ultrasound response. Data analysis and visualisation methods were developed during this study to determine the optimal methods of gaining battery charge measurements. The accuracy of charge monitoring is affected by temperature, either from the cell or ambient conditions. Evaluation of temperature effect is possible using analysis contained in this work, with a method of calibrating signal for temperature suggested.

It was concluded from experimental testing combined with modelling that small changes in battery parameters could cause significant changes to an expected signal response. The variety in cell construction and manufacturing discrepancies presents a serious problem for the application of ultrasound monitoring in practice. A smart peak selection method was developed to ensure that regardless of the nature of the ultrasound response, state of charge measurements are optimised by ensuring the regions of signal with best battery charge correlation are identified. This can greatly help with the automation of the process in a sensor-based battery management system.

Finally, ultrasonic monitoring was explored as a method of early warning of the onset of thermal runaway during thermal abuse testing. The signal responses show deviations from expected signals at temperatures consistent with expected cell reactions to temperature increases, such as the onset of self-heating and electrode delamination. Instrumentation calibration and limitation tests were conducted to provide confidence that changes in ultrasound responses are battery and not sensor related.