Electrochemical modelling for prediction of long-term battery power

The battery management system (BMS) of an electrified vehicle contains model-based algorithms to estimate and predict battery status, such as state-of-charge (SOC) and available power. Since the on-board computing power is limited and the number of cells is large, it is common practice to use simplified equivalent circuit models to describe the current-voltage characteristics of the battery cells. These models have the advantage of being simple enough for on-board implementation, and the parameters are also observable from current and voltage measurements. This is important since it enables updating of the model to handle changed characteristics due to ageing or varying operating conditions.

A drawback of the equivalent circuit models is that they do not consider the underlying electrochemistry and transport phenomena in the cell, which means that there are behaviours which they cannot predict. One example is the voltage drop observed when discharging a cell with high power for a longer time. The voltage drop depends on both time-independent and time-dependant processes behind the transient behaviour of mass transport in both electrolyte and electrode active material. Capturing this behaviour is important for long-term (~30s) power prediction needed for safe and efficient operation of electrified vehicles.

The aim of the project is to solve the problem of long-term battery power prediction for use in the BMS, which is still an open-ended research question. Starting from physics-based electrochemical models and applying methods from Automatic Control, the goal is to find a simplified model suitable for on-board implementation where the age dependent parameters are observable from measurements available in the vehicle.

In a longer perspective, the goal is also to establish a collaboration between the cell level research performed at KTH and system level research performed at Chalmers and Volvo Cars.