Open-source platform seeks to speed up production of safer EV batteries
Ann Arbor — Since Greg Less, technical director of the University of Michigan Battery Lab, was hired in 2014, research and automotive partners have called for the creation of a library for the results of groundbreaking work in battery development.
“People were coming here to say, ‘We need this. We need this. We need this. You have to develop it,’” Less said. “And I kept saying, ‘I don’t know how to develop this. I can’t get enough data. I can’t access the data and give it to you. “”
Voltaiq Inc., a California-based electric vehicle battery analysis software company and UM lab partner, may have the answer.
The company on Thursday announced its “Community Edition,” an open-source platform intended to house data sets collected from battery cells and computer code for the algorithms and models used to evaluate their performance. Executives hope the new tool will get academic and industrial research more quickly into the hands of manufacturers so they can leverage it and increase the rate at which they ramp up EV battery production to meet aggressive automaker launch timelines. automobiles.
“It takes four to five years to get these plants up and running and producing at high efficiency,” Voltaiq CEO Tal Sholklapper said during a tour of UM’s Battery Lab. “And it’s just not fast enough.”
Voltaiq’s business platform has customers like Google from Alphabet Inc., Amazon.com Inc., Mercedes-Benz from Daimler AG, Meta Platforms Inc. from Facebook and Microsoft Corp. She has also worked with two of Detroit’s three automakers. Voltaiq’s software gathers data collected from battery cells and formats it in a usable way in real time to qualify cells, check for anomalies that could indicate a fire hazard, and predict battery longevity that matters for warranties. vehicles, used car buyers and second life applications. .
However, validating the accuracy of machine learning and the algorithms used to perform these functions can require huge amounts of data. And at the rate that new techniques and materials in battery technology are being developed, there aren’t years to collect that data from eligible batteries when vehicle launches are to take place in the immediate coming years. to meet regulatory requirements for emissions and fuel economy.
“In order to get to the launch dates of the products, in order to really meet the deadlines that are coming up, and in order to have certainty in the guarantees that you have provided so that you minimize that in order not to have all the responsibility,” said Nicole Schauser, Senior Battery Data Scientist at Voltaiq, “we need better ways to qualify batteries faster.”
Much of this data, however, already exists and is even publicly available. It’s just scattered and can take days or even months to find. The Voltaiq Community Edition integrates with GitHub Inc., an internet hosting service for software development, and seeks to create a searchable location for available battery cell data as well as algorithms and models of machine learning that battery engineers can use to test or develop.
It’s valuable to the work collected and performed at UM’s Battery Lab, which tests hundreds of battery cells for automakers, startups, and other organizations.
“What Voltaiq came up with is this solution that people have been asking for,” said Less, who helped build the platform and uses Voltaiq’s business platform in the UM lab. “That’s huge for the data industry and for solving these issues that they’re talking about.”
The platform offers data and information on formation reduction, the process of stabilizing a battery cell which is the most time-consuming part of manufacturing, taking days; algorithms evaluating battery longevity; and models to detect anomalies. The hope is that the platform will help reduce the multi-year schedule that battery makers like LG Energy Solution and Panasonic Corp. have taken to set up battery gigafactories. Voltaiq is also discussing with universities to integrate the platform into courses.
Andrew Weng, a third-year research student at UM whose training work is available on the platform, says Voltaiq’s platform has allowed him to do research remotely during the pandemic and now community editing offers a place to share them.
“We can take the time to answer deeper questions,” Weng said, “and everyone benefits.”