Written by Jan Figgener, Senior Battery Expert, ACCURE Battery Intelligence
This blog presents our publication in Nature Energy, one of the world’s most renowned energy-focused scientific journals. This study, conducted in collaboration with RWTH Aachen University, presents the findings of an eight-year analysis comprising about 14 billion data points, which you’ll find published alongside the manuscript.
You can access the full paper in Nature Energy.
You can download the full dataset here.
Members of ACCURE Battery Intelligence have been spearheading the development of battery diagnostic solutions for over a decade. Our journey began as scientists within a research project at RWTH Aachen University, involving 21 private households using first generation home storage systems. From there, we’ve continuously refined our battery diagnostic algorithms and tailored them to industry needs.
Home storage systems are a pivotal component in driving the decentralized energy transition, and more so – empowering individuals to take part in it. The global market has grown significantly in recent years, with Germany alone being home to 1.5 million installed systems. As the European Batteries Regulation now requires reliable estimates of battery health (typically measured by remaining capacity), this research couldn’t be timelier.
Home storage systems are ingeniously simple: they charge with surplus photovoltaic (PV) energy during the day and supply this energy to households at night.
While some systems charge as soon as there is surplus PV power (excess-charging strategy), others wait until noon to charge, targeting the PV generation peak (forecast-based strategy). The forecast-based strategy is advantageous for two reasons:
Figure 1 below shows examples of these two two operational strategy types. The excess-charging system on the left begins charging (positive power) early in the morning, causing it to reach a fully charged state around noon. As a result, it must curtail PV power during midday to comply with the 60% feed-in limit of the subsidy program in this case. In contrast, the forecast-based system on the right delays charging until noon, allowing it to fully utilize the peak PV generation.
Figure 1: Excess-charging (left) vs. forecast-based (right) operational strategy. Figure according to Figgener et al. (2024), https://www.nature.com/articles/s41560-024-01620-9
Home storage systems are an interesting application for battery diagnostics. Their regular full charge and discharge cycles make it possible to precisely determine capacity. Because of this operational behavior, our method tailors coulomb counting to the specifics of home storage operation:
As shown in Figure 2, the resulting algorithmic state of health (SOH) estimation (blue dots) closely matches the conducted field capacity tests (yellow dots). Using a simplified linear fit (illustrated by the red trendline in Figure 2), the system’s aging rate is determined to be 2.7 percentage points per year. Based on this rate, the system is estimated to reach its end of life of 80% SOH in about 7 years, which coincides with the warranty period given at the time.
Figure 2: State of health (SOH) estimation. Algorithmic capacity estimation (blue dots) vs. field capacity tests (yellow dots). Field measurements for the system began two years after commissioning (see start of x-axis). Figure according to Figgener et al. (2024), https://www.nature.com/articles/s41560-024-01620-9
Our method is validated for lithium nickel manganese cobalt oxide (NMC), lithium nickel manganese oxide (LMO), and lithium iron phosphate (LFP) batteries, delivering robust results across different systems. Overall, we found that home storage systems lose 2-3 percentage points of usable capacity per year. On a positive note, most first-generation products met their warranties by adding a capacity buffer.
Get battery analytics trends and industry knowledge delivered directly to your inbox. Because we batteries.
Scaling up from a scientific study to industrial applications is expectedly difficult. Regarding the study we’re presenting here, the following challenges can be highlighted among many others:
1. Data volume and runtime
In this study, we analyzed 21 home storage systems, having a cumulative energy capacity of 150 kilowatt hours. At ACCURE, we currently monitor more than 4 gigawatt hours, which is over 25,000 times more data.
Scaling from these early developments with a smaller--though not insignificant--data set to monitoring such a large storage fleet is difficult and one of the main obstacles we have effectively overcome at ACCURE. One of our core competencies is connecting compact databases to performant cloud-based algorithms using both physics-based models and artificial intelligence.
2. Variety of use cases
When analyzing batteries on a large scale, use cases need to be well understood to develop tailored diagnostic algorithms. For example, while home storage systems have regular full cycles, some utility scale storage systems and electric vehicles might not. This is why we use different analytical approaches to accurately estimate the state of health (SOH). These include various methods such as voltage relaxation analysis or open circuit voltage analysis.
3. Customer focus
Customer needs are our central priority. We translate complex scientific methods into easy-to-understand key performance indicators, and support our customers with user-friendly dashboards, error handling tools, and forecasts on the performance, lifetime, and safety of their assets.
The full paper detailing this 8-year study, can be found in Nature Energy.
Jan is a Senior Battery Expert at ACCURE covering battery diagnostics and science communication. He evaluates current market developments and closely collaborates with international companies, agencies, and ministries on the topic of battery storage. His research has been published in international journals. In his free time, Jan enjoys playing sports and music.
ACCURE helps companies reduce risk, improve performance, and maximize the business value of battery energy storage. Our predictive analytics solution simplifies the complexity of battery data to make batteries safer, more reliable, and more sustainable. By combining cutting-edge artificial intelligence with deep expert knowledge of batteries, we bring a new level of clarity to energy storage. Today, we support customers worldwide, helping optimize the performance and safety of their battery systems. Visit us at accure.net.