What Exactly Does Battery Analytics Do?

Business management, retirement planning, investing, weather forecasting, road network design, airline ticket pricing, government budget balancing, and more are sciences, where we observe phenomena and create mathematical models that match the observations as closely as possible to predict future outcomes.

They all work the same way: Data goes in, the model does its thing, and insight comes out. Mathematical models run our society. Why aren’t we applying the same rigor to battery packs?

Battery analytics software collects, processes, and analyzes data from lithium-ion batteries to help us understand battery behaviors, predict performance, identify potential issues, enable proactive maintenance, and support continuous improvements. 

Product builders can use the insights to ensure there’s enough power when needed to get the job done, but not so excessive that it kills the business case. Unfortunately, we aren’t leveraging battery analytics’ full potential to make electrification more efficient.

Many battery packs are mis-dimensioned. Here’s a classic example: The average car covers 12,000 miles annually, or 33 miles daily. Yet, battery packs designed for 300+ miles are the norm. If airlines were to fly with 90% of their seats empty, ticket prices would be ten times higher. But that doesn’t happen because airlines have collected data and done their math.

Why aren’t people aware of the gross underutilization of battery packs?

First, nobody is collecting, analyzing, and presenting battery data for business decision-making. But why isn’t anyone doing that? Traditional battery solutions don’t provide the flexibility to reassign battery capacity — you can sit on a ton of insights and still can’t do much about equipment already in the field.

By now, you’ve probably figured out that SDB can provide the capability to adjust battery capacity and behaviors on the fly, making it possible to respond to analytics insights in timely and meaningful ways. That leaves us the first piece of the puzzle — collecting and turning data into actionable insights.

A large part of operations planning is iterative. You try something, look at the results, tweak a few things, then rinse and repeat. Models emerge from insights, which come from seeing patterns in data. Hence, it all starts with data.

For batteries, the process of collecting data, mining it for insights, and presenting the findings to support (business) decision-making starts with telemetry. Then, we draw insights automatically and manually.

The automated aspect involves acquiring data to build predictive models like we do with weather forecasts. Weather forecasts were notoriously unreliable, but today, they can predict when precipitation starts down to the minute. What changed?

More data and better models enable more accurate predictions. Numerous sensors collect telemetry like wind speed, temperature, and precipitation. We have buoys in the oceans and sensors hanging on telephone poles to beam data to institutions like the National Oceanic and Atmospheric Administration (NOAA).

Computers ingest data, run it through an algorithm, compare forecast results with actual phenomena, and identify discrepancies. Then, the self-learning model improves the algorithm to predict the most likely outcomes when it next sees similar data.

Battery longevity insights work similarly, requiring massive amounts of data — now possible thanks to a sizable installed base.

Tanktwo’s software-defined battery (SDB) solutions incorporate AI and machine learning technology to continuously fine-tune battery behaviors based on operating conditions and requirements of individual battery packs to turn real-time data into insights and actions to optimize battery life, performance, safety, and other parameters.

The manual aspect of battery analytics involves getting data insights to stakeholders and decision-makers to act on the information. Instead of granular, minute-to-minute adjustments, these insights often lead to substantial shifts in product development or service delivery. 

For example, an airline wants to keep planes producing revenue as close to 24 hours a day as possible. Decision-makers must know where the planes are, how long it takes to turn around a flight, and when to schedule maintenance. Accurate data helps inform their decisions to maximize revenue.

Same for battery pack design. If an operator sees that v1.0 of its garbage trucks return to the depot every time with 75% SoC (state of charge), they’d be confident to say that v2.0 can have a battery pack of half the size.

Long story short, the machine-targeted analytics make things work well under the hood, and the manual aspect enables decision-makers to improve operations, such as reallocating resources to improve profitability.

Many interacting factors impact battery longevity and behaviors, requiring numerous data points to draw meaningful conclusions. That’s not a task that can be handled manually — which takes us back to the importance of battery analytics and self-learning mechanisms for optimizing cost efficiency.

But why didn’t people use analytics for equipment powered by internal combustion engines (ICEs)? As we’ll explore further in the next section, the different cost distribution means that the extra cost of building a bigger fuel tank and driving a full tank around is insignificant compared with building and lugging around a grossly over-dimensioned battery pack.

As such, there hasn’t been an incentive to analyze fuel tank capacity and usage patterns to optimize the size and weight of fuel-carrying mechanisms. Unless a mindset change happens across industries to prioritize battery analytics, we will continue to lack access to data to understand how to dimension battery packs for optimal efficiency.