As an owner of one of approximately 140,000 Chevrolet Bolts (including EV and EUV models) sold in the U.S. that were recalled last November, it’s more than a little frustrating that 14 months later, I’m still only permitted to charge the battery to 80% of its capacity. Like all other Bolt owners, I purchased a vehicle rated for up to 240 miles on a full charge. On a sunny warm day in California, after fully charging the battery (to the permitted 80% charge capacity), my Bolt is reporting an estimated driving range of 180 miles.
While I have no issues with the overall build quality of the Bolt or its battery module, the question remains: why couldn’t GM recall only those vehicles that were manufactured using the faulty battery modules? Had they been able to trace which battery module was installed in each vehicle, then this $1.9 billion dollar recall could have been far smaller and less impactful to the majority of Bolt buyers, not to mention for GM and the battery maker LG Chem.
At Peaxy, we deal with battery customers in all stages of the commercial lifecycle – from ones performing R&D prior to scaling out their manufacture, to some operating pilot manufacture lines, to those manufacturing at scale and/or integrating gWh of batteries per year. While battery technologies may differ greatly between these companies, all share a common need to know what lot of a particular component or material was used to manufacture each single battery or cell, traced to the individual serial number.
What is battery lot traceability?
Battery lot traceability allows a company to trace exactly which lot each component belongs to, in the manufacture of a single serialized battery or cell. While it may sound simple, the ability to attain such a necessarily laser focused view into your data requires coordinated efforts and advanced software capabilities, as well as deep domain expertise.
Coming from a programming background, I like to think of tracing individual lots through the manufacturing cycle, be it small batch R&D, pilot or scaled out manufacture, as a large multidimensional array. This would consist of top-level serialized assets, individual components used in manufacture, together with the individual lots (or batches) for each component used to manufacture a single asset.
By capturing manufacturing, test and integration data at every step of the process by individual battery or cell, full traceability for any data point is available at the cell, module, string, or energy block level.
The benefits of battery lot traceability
For companies still focusing on R&D, implementing lot traceability provides several benefits. When a company is asked to demonstrate results to existing or future investors, having the ability to create reports that include the exact lot of each component used to manufacture the samples used in the DOE (Design Of Experiment) demonstrates that the company is using best practices. This in turn gives investors confidence that reported results are repeatable, rather than being a one off, random result.
In cases where results differ between samples of the same DOE, knowing that the samples were built using two or more anode lots, or even two or more lots of a particular chemical used in the electrolyte solution can provide clear direction when the R&D team investigates the test result discrepancies. Incorporating component lot information into machine learning algorithms (such as logistic regression) can provide additional insights for the R&D team when analyzing DOE results.
For companies that have started to scale out their production and are selling and/or installing their batteries, the ability to trace all components down to individual lots increases in importance tenfold, and takes us back to the opening paragraph of this article. Had GM been able to track individual battery modules to individual VINs (Vehicle Identification Number), they could have easily identified the affected units and reduced the scale of the recall to just the vehicles fitted with the suspect battery modules. Other areas where lot traceability is useful are managing warranties and triaging quality problems less impactful than a recall scenario but still important, including abnormal discharge rates.
An example use of battery lot traceability
To illustrate the benefits more, let’s create a hypothetical example of an energy company we’ll call Acme that has an install base consisting of numerous sites, amounting to multi-gWh in energy capacity.
Over the course of several weeks, Acme notices that a growing percentage of batteries are discharging at a rate higher than expected or predicted. At first glance, the number of failing batteries appears to be within the expected range, and therefore doesn’t initially generate any alerts. Using the ability to trace each lot used in the manufacture of every serialized battery (a feature in our own software), a thorough investigation can reveal relatively quickly if there is a common cause behind the battery failures, and to evaluate the potential risk across the entire fleet of deployed batteries.
The investigation could roughly follow these steps:
Create a collection or group in the battery analytics software, using the serial numbers of the failing batteries. This list can be easily obtained from the data fed from the BMS.
Look for commonalities between the failing batteries. A well-configured battery analytics solution that offers lot traceability should provide the capability to compare in-field data to the original formation test results. In this example, all batteries in the collection may have been manufactured using different lots for the casing, electrolyte, cathode and even the separator, but were all manufactured using the same anode lot. Armed with this information, Acme now knows it could be facing problems with batteries manufactured with a particular lot of its anode material.
Using the data obtained from the battery analytics software, a second query can find virtually all of the batteries manufactured using the suspect anode lot, regardless of where they were eventually deployed across multiple sites.
A portfolio view of deployed battery sites. Trace material lots used within each asset of the site, compare in-field data to the original formation test results, or find all batteries manufactured using a suspected anode lot, regardless of where deployed.
A key consideration in evaluating battery analytics software is the ability to integrate with existing software solutions and data sources. In the case of battery manufacturers, this includes MES, digital manufacturing travelers, and automated manufacturing processes. For energy companies it might include ERP, BMS and warranty management systems to name a few. Integrating lot traceability across all systems from manufacture through field services provides the ability to assess risks to all fleet deployed assets and gain a significant advantage over competitors who can’t.
Companies may assume if they already have all of this data that they can find the answers themselves. Battery analytics software with digital manufacturing traveler capabilities, however, can make it much easier (or even possible) to sift through enormous amounts of data from a complicated systems landscape. It provides not only a parameterized view of the data with easy point and click drilldown, but also insights into battery health and what specific components might be impacting it. Armed with these insights from their own data, companies in the battery value chain can realize the prospect of making negative outcomes such as massive recall scenarios a thing of the past.
For more insights into the Chevy Bolt recall, please see our previous article: How GM and LG could have made the Chevy Bolt recall less painful.