Battery energy storage is being installed behind-the-meter to reduce electrical bills while improving power system efficiency and resiliency. This paper demonstrates the development and application of an advanced optimal control method for battery energy storage systems to maximize these benefits. We combine methods for accurately modeling the state-of-charge, temperature, and state-of-health of lithium-ion battery cells into a model predictive controller to optimally schedule charge/discharge, air-conditioning, and forced air convection power to shift a electric customer's consumption and hence reduce their electric bill. While linear state-of-health models produce linear relationships between battery usage and degradation, a non-linear, stress-factor model accounts for the compounding improvements in lifetime that can be achieved by reducing several stress factors at once. Applying this controller to a simulated system shows significant benefits from cooling-in-the-loop control and that relatively small sacrifices in bill reduction performance can yield large increases in battery life. This trade-off function is highly dependent on the battery's degradation mechanisms and what model is used to represent them.
Lithium ion batteries have a well-documented tendency to fail energetically under various abuse conditions. These conditions frequently result in decomposition of the electrochemical components within the battery resulting in gas generation and increased internal pressure which can lead to an explosive case rupture. The 18650 format cell incorporates a vent mechanism located within a crimped cap to relieve pressure and mitigate the risk of case rupture. Cell venting, however, introduces additional safety concerns associated with the flow of flammable gases and liquid electrolyte into the environment. Experiments to quantify key parameters are performed to elucidate the external dynamics of battery venting. A first experiment measures the vent burst pressure. Burst vent caps are then tested with a second experimental fixture to measure vent opening area and discharge coefficient during choked-flow venting, which occurs during battery failure. Vent opening area and discharge coefficient are calculated from stagnation temperature, stagnation pressure, and static pressure measurements along with compressible-isentropic flow equations and conservation of mass. Commercially sourced vent caps are used with repeated tests run to quantify repeatability and variability. Validation experiments confirmed accuracy of opening area and discharge coefficient measurement. Further, trials conducted on vent caps from two sources demonstrate the potential for variation between manufacturers.