Peaxy Lifecycle Intelligence for Batteries, a new predictive battery analytics platform, powered by machine learning.


Why did you decide to launch this new offering? Why now?

Peaxy has worked in the industrial analytics space for eight years, and we’ve accumulated significant expertise in creating solutions for power generation, aviation, oil and gas, and automotive industries — in many cases with machine learning and digital twins. Our use cases focus on the maintenance of large, serialized, industrial machines or systems. In these environments, increased electrification and battery storage are where our customers want to go. 

In the battery space specifically, we’re working on transportation applications, behind the meter battery installations or larger utility-scale storage installations. For all of those scenarios, our customers require solutions that can track tens of thousands of serialized batteries. Even a modest, 10 MWh battery storage installation will include some 20,000 serialized batteries, each one requiring 24/7 monitoring, each one having its own degradation curve, each one having its own manufacturing and assembly pedigree, and each one having a 5-10 year lifespan. In short, this is a large-scale, by-serial-number data problem that our team and our product — Peaxy Lifecycle Intelligence — are uniquely qualified to solve.


What is the background on the development of the product? How long did it take to create, and what were some of the key inputs that went into designing it?

I don’t believe that fleet-level models are good at predicting and optimizing industrial equipment. Large-scale industrial equipment is by its very nature unique. Batteries are no different. The inception of our battery solution came from modifying our Peaxy Lifecycle Intelligence (PLI) product for a few early energy storage customers. These lighthouse customers have allowed us to deep-dive into the battery space, letting us apply our decades of combined experience in power generation, modeling and industrial analytics. 

PLI for Batteries has been in development for about two years now, but we’ve had battery customers using the software during this time. We addressed early on some of the most obvious pain points, importantly with capabilities aimed at tracking degradation curves in real-time. We have cut our teeth on the industry-wide problem of managing cycle statistics, which is a huge challenge for battery packagers, system integrators, and OEMs — given that customers rarely operate batteries in the way they are tested for in the lab.


Who do you imagine taking the most advantage of PLI for Batteries? Give an example of how it can most help address a pain point.

Our main focus with the launch is around grid-scale operators, power producers and OEMs that have a vested interest in optimizing charge/discharge regimes and extending asset life. Enabling battery leasing programs and battery warranty management are also focus areas, and our software has features that help lessors make sure they protect the long-term residual value of battery assets. 

Peaxy Lifecycle Intelligence also eliminates paper on the OEM factory floor, at a system integrator’s facility and within R&D labs. For example, we automate and normalize battery cycler data across different cycler brands and model series: that alone has a transformative effect on making it easier to understand what electrolyte recipes or bill of material configurations have the most promise. Linking all of this together is our ability to capture the entire data value chain, in order to create a single source of (performance) truth. This is absolutely essential in driving machine learning insights and high fidelity digital twins.


What role will machine learning play in designing solutions for your customers?

There’s understandably a lot of hype and excitement around machine learning. We believe in using it as a powerful tool to solve practical problems and to advance the feature set of our products, versus it being a means to an end by itself. One area where we are actively deploying ML is for anomaly detection at the battery level, string level, block-level, and site level. Depending on the quality and completeness of the data available, we can train our proprietary algorithms to drive insights into, for example, knowing when a battery or string of batteries will fail, or what its remaining useful life is under certain operating regimes.


Where do you see the energy storage industry heading in the next five years, and how will Peaxy play a role?

For energy storage, it’s all about moving past 4 hours of storage capacity and moving towards 10 hours and beyond. This is essential if battery solutions are expected to compete with more traditional carbon-based power generation, like combined-cycle peaker power plants. 

Innovations driving battery solutions in the coming years will come from materials science, with new and more efficient anode and cathode materials coming online, particularly for lithium-ion batteries. Digitizing battery cycler data, tracking all manufacturing and sourcing details, understanding ambient conditions and operating conditions will all prove critical to being able to compete within the battery industry. 

We strongly believe that any business driven by our software will have a competitive advantage over businesses served by legacy solutions. Applying the latest technology capabilities in machine learning and predictive analytics will dramatically improve overall battery performance, reduce costs and drive the overall growth of the energy storage market. We are at the forefront of these advances and we intend to be the world-class partner of choice for battery companies.