Novel Method for Using Force in Incremental Capacity Analysis for Capacity Fading Estimation

Abstract

Health monitoring techniques in batteries rely on voltage measurements. Several methods haven been introduced in literature for the evaluation of the aging in battery. In Cyclic Voltammetry (CV), the electrode potential is increased linearly with time. The resulting cyclic voltammogram is a plot of current vs. voltage. The shift in the plot peaks is correlated with aging. Statistical methods can also be used to track aging by fitting a probability density function (PDF) to a histogram of voltage data during charge/discharge. As the cell degrades, the PDF curve shifts which allows for aging detection. Another widely known method is the differential voltage (DV) method. The method plots the differential of voltage over capacity (dV/dQ) with respect to capacity and monitors peak shifting to relate with degradation. The inverse of the DV method is the incremental capacity analysis (ICA). In many cell chemistries, the cells are characterized by a voltage plateau for a wide range of SOCs. The ICA method plots the incremental capacity over voltage (dQ/dV) with respect to voltage, which allows for clearly identifiable peaks that correlate with capacity fading. This method has been recently shown to predict capacity fading with less than 1% error (1). Although the ICA method has been shown to be accurate in estimating capacity fade, it still has some major drawbacks. The method is sensitive to voltage measurement noise and accuracy since dV is small. This is especially the case in Lithium iron phosphate (LFP) cells, which are characterized by flat voltage curves. A second limitation of ICA arises from the peak location. In the case of an NMC cell, the ICA peaks in discharge are centered around 40% SOC. This means that the battery has to operate in the low SOC range in order to estimate and monitor the peak location and hence the capacity fading. Here we report a novel method of monitoring electrode expansion in the ICA method. The method uses force measurements to derive the IC (incremental capacity) curves. In lithium-ion batteries, charging causes volume change or swelling of the electrodes as the lithium ions intercalate in the negative electrode. In applications where the batteries are constrained or compressed to prevent expansion, the swelling causes a measurable stress. This stress can be measured using a force sensor (or strain gauge). Experiments are performed on 4 different fixtures each containing 3 battery cells, clamped together between two end plates (Figure 1). The fixtures are operating under an aggressive power cycle (Imax = 36 C-rate, Iavg = 12 C-rate) centered at various SOC ranges and preloaded under two different forces. The degradation of the cells was used to develop a model of the peak location as a function of capacity. Figure 2 shows the model error in estimating capacity fade based on the peak location using the measured force, coulomb counting, and voltage. As shown in Figure 2, the incremental capacity analysis based on force measurements (ICF) is promising since it can be used to predict capacity fade with 0.42% accuracy. Also, in the case of an NMC cell, the ICF peaks occur around 70%SOC while those of the ICV occur around 40%SOC (Figure 1). Hence the proposed SOH monitoring could update estimates more frequently within the regular use of an electric vehicle in urban and most driving conditions, given that drivers typically spend more time in the higher battery SOCs than in the lower ones. It is also shown that bulk force measured on the fixture holding 3 different cells balanced within 3% capacity can be used to estimate individual cell capacities of each fixture when individual cell voltages are measured. Results show that the maximum error is 3.1% with an absolute average and standard deviation on the error of 0.42% and 1.14% respectively as shown in Figure 2. Future work would include implementing this method in on-board state of health monitoring prognostic algorithms. Acknowledgment The information, data, or work presented herein was funded in part by the Advanced Research Projects Agency-Energy (ARPA-E), U.S. Department of Energy, under Award Number DE-AR00002691.