How Predictive Analytics For Battery Management Drives Operational Efficiency and Profitability

Recently, I was driving from Virginia to Texas when a big winter storm hit the country. We looked at the radar map, saw where the storm was heading, made a detour, and avoided getting stuck in a “rain + plummeting temperature + wind = icy road + no go” situation.

This is predictive analytics in action.

Predictive analytics is everywhere. Advanced car computers can tell you when something will likely go wrong so you can fix it before getting stranded on the side of a highway. Netflix can show you what you may like based on what you have watched. Amazon makes product recommendations based on your past purchases and browsing behaviors.

But that’s just the tip of the iceberg. Predictive analytics is used in many industrial applications. For example, IoT devices on factory floors gather data, which is analyzed in almost real-time to identify machinery that will require service or replacement to prevent costly downtime. 

Predictive analytics is the linchpin of modern resource/asset management and optimization. Any application that uses finite resources, including electrification solutions (which uses lithium batteries, an expensive asset,) can benefit from this powerful capability.

What is predictive analytics?

Predictive analytics is a disciple that uses artificial intelligence (AI) and machine learning technology to derive insights from historical data to understand patterns, identify trends, foresee potential problems, and predict future behaviors. 

The software can further interpret the insights to provide recommendations for improving processes, inform immediate actions, optimize resource usage, and maximize performance.

Predictive analytics in battery management

Predictive analytics can help operators monitor each battery’s real-time behaviors (or “health”) to know which cell needs to be replaced and when.

They can eliminate the need to preventively replace cells at fixed intervals (often long before the battery can no longer perform,) a costly and wasteful practice simply because operators don’t have data to know which cell has deteriorated. 

I once tested a batch of used batteries switched out from retail scanners during routine maintenance. 95% of the cells still held more than 98% of their designed capacity. 

The Tanktwo Battery Operating System (TBOS) uses software to analyze telemetry and create a heat map (like a weather radar map) to show the health of each cell in a battery pack. It flags those with declining performance, so operators can replace them to prevent costly downtime without the cost of preemptively replacing every cell frequently.

The importance of predictive analytics in battery solutions and management

Predictive analytics identifies patterns to predict performance in real-time. It allows operators to optimize the performance and usage of expensive and finite resources, such as lithium, used in most batteries. The optimization helps them lower the total cost of ownership, minimize costly downtime, and increase efficiency while reducing wastage. 

The data can also help operators identify usage trends, pinpoint issues, and take the guesswork out of resource management to increase profitability. Additionally, companies can avoid the labor cost and operational disruption of changing a battery not due for replacement.

The improved cost-efficiency and ease of maintenance make it commercially feasible to electrify more applications. The increased visibility reduces the complexity of power management, further streamlining many processes.

Predictive analytics for battery solutions: Turning insights into action

A software-driven battery solution is a foundational element for applying predictive analytics to support commercially viable electrification applications. Yet, battery management software in today’s battery packs doesn’t have the capabilities to achieve the level of insights and optimization we need to realize the promise and potential of electrification.

A battery solution must be able to perform the following to support predictive analytics in power management:

Collect granular data in real-time

First, you need the ability to access data to understand the behavior of each cell at any moment in time to make the right decision. Each battery pack must be able to collect telemetry and communicate the data to an operating system where the software can collate and interpret the information.

Analyze data and forecast behaviors

You need software that can analyze the telemetry in real-time, interpret each cell’s behavior, and understand how it affects the battery pack’s performance. The software should also include a predictive model to forecast what will likely happen to each cell based on usage patterns and specific applications.

Turn insights into action

The last piece of the puzzle is a user interface that can show field personnel which cell will likely experience issues and provides recommendations on the next step. With sufficient lead time, operators can schedule maintenance activities to achieve the highest operational efficiency while minimizing waste. 

It’s like when a car tells you when it needs an oil change. The light goes on early enough so you can schedule maintenance when it's most convenient. And you don’t need to be a specialist in lubricant deterioration to know that you need an oil change.

Using TBOS, operators can receive a warning (yellow flag) with sufficient lead time before a cell may experience issues to optimize operational scheduling based on insights and performance standards. Plus, our software takes care of the analytics behind the scene — you don’t need data experts in the field to perform deep analysis to determine the best course of action.

Who can benefit from predictive analytics for battery solutions?

Any operator of electrified equipment at a commercial or industrial scale can benefit from efficient resource usage, the lower total cost of ownership, just-in-time maintenance, and ongoing optimization. Here are some use cases and examples:

Battery solutions for the aviation industry

Due to the sector’s high safety standards, most aviation systems are triple redundant. But having three identical batteries may not provide the desired redundancy because they’re likely to deteriorate at the same rate.

Predictive analytics supports these systems by pinpointing the likelihood of failure — not only when but also on a “sliding scale” (i.e., between 100% capability and complete failure) so operators can make accurate redundancy determinations.

Battery solutions for medical equipment

Did you know that 50% of the ventilators in the national stockpile could not be used when COVID hit because the batteries deteriorated to the point where they failed to hold a charge? A system that can proactively flag cells needing maintenance could help ensure every piece of equipment stays functional. 

Battery solutions for commercial EV fleets

Fleet operations must account for maintenance and replacement schedules based on vehicles’ age and mileage. Today’s EV battery solutions can’t tell operators when failure may occur. As such, companies often replace and discard batteries long before they start to deteriorate. 

However, replacing expensive battery packs at regular intervals based on statistical information instead of actual deterioration leads to (1) massive wastage and (2) unexpected downtime from “outlier” battery packs that deteriorate faster than normal.

With predictive analytics and Tanktwo’s modular battery system, operators can replace cells and battery packs as required without risking unwanted downtime — reducing operating costs and lengthening a fleet’s lifespan.

Battery solutions for the retail industry

Battery packs in retail barcode scanners deteriorate at different rates based on usage. Retailers preventively replace all batteries every 12 months to avoid the high cost of failure (e.g., loss of business because customers can’t check out.) But as mentioned above, most batteries are thrown out while having most of their capacities almost intact.

With predictive analytics, operations only need to replace deteriorated cells — reducing wastage and expenses significantly without any loss in uptime.

Bottom line

A data-driven, software-defined battery solution is key to unlocking electrification on an unprecedented scale. Predictive analytics will become a massive enabler by allowing efficient resource management and allocation to ensure operational efficiency while lowering the total cost of ownership.