‘Application of Prognostics to the EV industry and Li-Ion batteries’

by Manobhava Baruah 30 Apr 2022


Autocar Professional's two-wheeler EV Forum webinar kicked off with a presentation by Lohit Dhamija, Project Lead -R&D, Varroc who spoke about the opportunities for applications of Prognostics for vehicle health management in EVs.

Dhamija has an MTech degree in mechanical engineering with a specialisation in CADA from IIT, Bombay. At Varroc, his key focus is on model-based control and data analytics. With over 12 years of experience, Dhamija has worked on various projects including traction motor controller, BMS, DC-DC converter and advanced analytics in vehicles.

While some failures in a vehicle’s physical health can be seen with the naked eye, Prognostics helps to provide a deeper inspection of a vehicle’s system. It is the process of predicting when a system or component will no longer perform its intended function. It is also tasked with monitoring a system's health and predicting its remaining useful life (RUL) by evaluating the degree of degradation from its expected state of health in its expected usage conditions.

According to ISSO 13381: Prognostics of future fault progressions requires prior knowledge of the probable failure modes and future duties to which the machine will or might be subjected. It also requires a thorough understanding of the relationships between failure modes and operating conditions.

Dhamija said, "There are multiple opportunities in which prognostics and Preventing Health Maintenance (PHM) can come into an electric vehicle. Some of the potential applications where Prognostics can be used is Traction motor, Li-ion batteries and inverters."

"In li-ion batteries, there are some failures which can cause battery degradation. The failures are namely battery ageing which can cause capacity degradation. The other ones are internal and external short circuit faults and thermal faults which basically may lead to fires in the vehicles. So, we need to understand the mechanism and the way to predict these faults and then attempt to collect data specific for those faults. These can then be modelled using any technique either model-based, data driven or hybrid,” said Dhamija.