Ultrasonic and vibrational methods to determine changes of state of lithium-ion cells



Journal Title

Journal ISSN

Volume Title



Lithium-ion batteries (LIBs) are the chosen power source for battery electric vehicles and battery energy storage systems. These high-power, high-capacity applications subject LIBs to challenging operating environments where mechanical, electrical, and thermal abuse is likely. In these applications, thousands to hundreds of thousands of cells are connected in series and parallel, which creates a challenging monitoring problem. The search for improvements to the battery management system (BMS), including new sensing modalities, is a very active and growing field. This work investigates the use of mechanical inspection of lithium-ion batteries using dynamic mechanical loading for state estimation. Ultrasonic inspection is used to monitor cells as they undergo normal charge-discharge cycling and different amounts of thermal loading, sometimes to thermal runaway. By specifically monitoring the ultrasonic signal characteristics of the signal amplitude (SA) and time of flight shift (TOFS), we can monitor changes to the cell's stiffness, density, and attenuation which result from changes in the cell's state of charge (SOC), temperature, or the presence of damage from thermal abuse. We find that ultrasonic signal characteristics warn of impending cell failure up to 25 minutes in advance of traditional monitoring sensors. A transfer matrix model employing a Bloch-Floquet formalism which accounts for the repeating layered scheme of the cell is introduced to explore ultrasonic wave dispersion due to the layered structure and internal losses due to the cell's polymeric components. Experimentally obtained ultrasonic signal characteristics were corroborated with this periodic transfer matrix model (PTMM) which can simulate SA and TOFS by using the appropriate SOC or temperature-dependent material properties of cell components. The PTMM validates experimental measurements, and helps demonstrate which cell components dominate the characteristics of ultrasonic wave propagation in the thickness direction of LIB pouch cells. The results from US inspection demonstrate its applicability to provide advanced warning of cell failure and also to detect the presence of damage from previous thermal abuse. The same chemo-mechanics that drive changes in the cell's ultrasonic response should also affect the cell's modal response. One can imagine implementing modal testing on cell packs as a part of routine maintenance, or making use of ambient vibrations as the excitation for modal testing in applications like BEVs . As such, this work also explores the viability of vibrational inspection for state estimation, focusing primarily on SOC and state of health (SOH) estimation. The surface velocity of lithium-ion pouch cells confined in a fixed-fixed configuration is measured with a scanning laser Doppler vibrometer (SLDV) while the cells are subjected to base excitation using an electrodynamic a shaker. SLDV scans are performed after the cell has been charged or discharged to specific SOC and repeated across numerous cycles. Results from these experiments show that the modal frequency of a cell shifts towards higher frequency with increasing SOC. These results were corroborated with an effective material model of the cell which was created with multiscale homogenization of the cell's components including their microscale heterogeneity. This material model is created using material properties of constituents at full charge and at full discharge, and is input to a finite element simulation of the resonance frequency of the cell. We find good agreement between the resonance frequency predicted by the multiscale model and transmissibility measurements at 0% and 100% SOC. The experimental results for continued cycling showed an increase in modal frequency at the fully charged and fully discharged states with cell aging. While identifying the chemo-mechanical cause of the changing cell modal response with aging remains a challenge, the correlation between SOH and modal response illustrates how the technique can be used for both SOC and SOH estimation.


LCSH Subject Headings