In modular battery systems, cells are grouped into modules, making it easier to remove or replace individual parts. This can improve battery health and extend lifespan. However, manually diagnosing battery cells is both challenging and time-consuming. SEC doctoral student Fatemeh Hashemniya has developed methods for the automatic detection of faults in reconfigurable modular battery packs.
Faults in battery cells can originate from the production stage, where individual cells may already be defective, and others develop during operation, such as ageing or thermal issues. Traditional battery packs are built as fixed systems which means that cells are connected in a predefined structure that cannot easily be changed.
“Instead of having one fixed structure, you can divide the battery into modules and reconfigure them if needed,” says Fatemeh Hashemniya. “That gives much more flexibility, both in operation and maintenance.”
Structural diagnostic methods
To be able to detect and isolate the faults Fatemeh has worked with structural diagnostic methods developed at Linköping University. These methods are based on how parts are connected rather than detailed calculations and enables efficient diagnosable systems even in large and complex applications.
“I have extended the structural diagnostic methods to handle reconfigurable battery systems. This has led to the development of a multi-mode structural diagnostics approach that can handle multiple configurations of the system. It’s been tested in both laboratory settings and on real battery packs at Scania. In one case an issue was identified in a system that was initially assumed to be fault-free, which showed that it can detect problems that are not obvious,” says Fatemeh.
A typical modular battery pack can have hundreds of cells and sensors are crucial to identify the problems. In some cases, additional sensors are needed to make faults detectable. In other cases, there are already many sensors, and the challenge is how to use them effectively. It’s also important to distinguish between faults in the cells and faults in the sensors, which can show incorrect values due to bias or noise.
Analysing patterns in sensor data
“We look at patterns in the sensor data to evaluate faulty cells and the cause of the “illness”, while some patterns indicates a sensor problem. When everything is properly designed and configured, detection can be relatively straightforward. However, designing the system and selecting the right sensors and methods is complex. Some faults are easy to detect while others are more difficult, especially in real systems where measurement noise is present,” says Fatemeh.
The developed methods are ready to be integrated into battery management systems since the required data is already available from sensors. The main step is to implement the diagnostic logic that interprets the data.
Fatemeh will present the results at her PhD defense scheduled 25 September. “The main benefit from automated early detection is that it saves both time and cost when the vehicle is taken to a workshop. It also allows for more efficient maintenance, since you can target the exact faulty component.”
About Fatemneh Hashemniya
Fatemneh is originally from Iran and came to Sweden five years ago to start her doctoral studies at Linköping University within the SEC project “Diagnostics and supervision of dynamically reconfigurable battery systems”.
“I’m very grateful to have been part of the Swedish Electromobility Centre. The workshops, study visits, and collaborations have been very valuable for my development during these years.”
https://liu.se/en/employee/fatha93
Image: Shutterstock/allahfoto