Ultrathin Strain and Temperature Sensors for Li-ion Batteries

The goal of the GE led research team, in partnership with University of Michigan, and the Ford Motor Company, is to develop a smart, cost effective sensing system that significantly increases the utilization of stored energy in battery packs while maintaining or improving upon system lifetime for vehicle applications. The team addressed three technical challenges – 1) an ultrathin sensor array capable of measuring cell strain and surface temperature across multiple cells within a battery pack, 2) reduced order modeling for real-time computation and observability analysis for a minimum number of sensors, and 3) sensor-pack integration and system evaluation of sensors and adaptive battery control. The elements of the team’s innovative approach are illustrated in the Figure 1.
 

The team evaluated different polymer substrates and different deposited metals for the sensors, with the goal of reducing thickness 20x compared with state of the art battery system sensors, which are typically 2-3 mm thick. The resulting co-located strain and temperature sensor array is under 100 mm thick, enabling the sensor array to be placed between the battery cells within the pack. The sensors achieved accuracy to <0.1 C and <0.1 mm in displacement. A photograph of a battery pack is depicted in Figure 2.
 

The U of Michigan team developed reduced order physics-based models to utilize the information from the GE sensor array, extracting the thermal and stress features to be used in a new type of battery control algorithm. Predicting the swelling of the entire cell as it charges and discharges in an operating vehicle environment would normally be too computationally intensive for practical use. The team overcame that challenge through observability analysis and estimation techniques that span many physical scales from the electrode level phenomena (5 µm, 50 msec) to the cell level (10 cm, 1 sec), to the pack at the vehicle level (1.0 m, 5 sec).  To quantify cell swelling, the team developed innovative experimental methods and specialized laboratory fixtures that measure the battery free and constrained swelling along with its thermal behavior.

The sensor data, paired with simplified, reduced order physics-based model, are utilized in real-time to optimize pack performance under operation, and make predictions on the state of health of the pack. To evaluate performance, the team instrumented a full battery pack with 76 cells from a Ford Fusion Hybrid Vehicle with their new sensors and control system for testing at Ford Motor Company. The results immediately enabled multiple innovations in real time management. These include setting power limits, fast warm-up, and state of health estimation of capacity fading based on monitoring shifts in bulk stress. Analysis from the team’s demonstration on the Ford battery pack indicated that these innovations can enable downsizing of the battery with associated increase in energy utilization by 19% per cell and a projected decrease in capacity of only 0.5% after 100,000 miles. The initial testing on a hybrid electric vehicle (HEV) pack was a useful demonstration of this approach to improved battery systems, and the results are promising that the integration of advanced sensors with model-predictive controls can improve performance in EV battery systems, but it will require continued development and deployment on larger battery electric vehicle (BEV) packs to reap the maximum benefits and value from this technology.