As electric transportation evolves from the initial phase to mass-scale implementation, battery performance has turned out to be a major issue. For electric fleets, specially for those operating in environments where electric mobility is constantly used, like in delivery and transport, battery malfunction is hardly considered to be just an issue of technology. Battery failure can put a car off the road and lead to disruptions in operations, failure to deliver services to customers and higher maintenance costs.
The issue is though that battery failures do not occur just by themselves. The performance of electric batteries, changes in charging behavior, temperatures, discharge cycles and conditions of use can signal when the battery is about to fail.
Predictive analytics can help solve this problem.
Transitioning from reactive maintenance to predictive maintenance
Conventional battery maintenance practices can generally be categorized into one of two categories regular maintenance performed periodically, and reactive maintenance done after an issue has occurred. Each practice presents its own challenges.
On one end, maintenance carried out according to a fixed schedule may mean that batteries may get attended to when they are functioning normally, while problematic batteries may keep working till they fail completely.
The introduction of predictive analytics changes the existing process of battery maintenance by using data from continuous monitoring of battery usage to identify probabilities of failures.
This will allow fleet owners and operators to ask different questions such as “What batteries are at risk of failing, and what do we do about them?”
The importance of recognizing a problem before it occurs
In a high-utilisation fleet, battery failure involves much more than just the battery. A flat battery can cause any instance of downtime, leading to failures in delivery, inconvenience for the rider, need for other vehicles and losses in revenue.
Predictive systems can evaluate historic and real-time data and find battery failures that behave differently from the expected ones. When the battery shows any sign of elevated risk, maintenance units can inspect the vehicle, the battery or arrange for the replacement of the defective unit.
Importantly, not every prediction entails the necessity of battery replacement right away. The whole point of predictive modeling is to stabilize the process of battery failure detection and allow companies to take the necessary action right on time. Consequently, it can lead to decreased number of unwanted incidents and improved availability of the fleet.
The battery issue may not be confined to the battery
One of the key advantages of predictive analytics is that it goes beyond the battery.
Battery performance is influenced by the operating conditions of a vehicle. Thus, any battery trouble might be due to the battery operation, charging tactic, excessive temperature, deep discharge, aggressive driving style, problems with a vehicle, problems with the controller, connection issues or electricity failure.
To illustrate, if a certain group of vehicles has performed abnormally in terms of battery, simply changing the batteries may not help solve the problem. The main reason for battery failure might be due to the way the cars get charged.
The use of predictive analysis combines data about batteries, vehicles and the way they are operated. Thus, such analysis provides greater clarity concerning the reasons of the problem and allows the maintenance teams to shift from repeated problem-solving to more targeted diagnosis.
Increase battery life by understanding how it’s used
Batteries may degrade due to various factors other than time. Batteries do not only degrade but also perform poorly due to the way they are being used and charged.
Patterns that may be associated with the battery fail to only recharge it, and thus contribute to a shorter life cycle must, therefore be analysed.
Once such patterns are identified, changes to the operating practices and processes of charging may be applied.
This is quite important because battery lifecycle management ultimately means extracting more value from each battery being used.
From maintaining every vehicle to prioritising actual risk
Managing thousands of vehicles adds another complication scale. The problem is that conducting a manual inspection of every battery cannot be efficient or effective when you have so many vehicles. Using predictive analytics enables you to create a risk-oriented maintenance approach.
Instead of treating all vehicles as equals, the fleet team will be able to determine which smaller percentage of vehicles with elevated risk needs to undergo an inspection.
This results in the change of maintenance staff’s role from an efficient response to a long list of scheduled checks to concentration of all resources on where they are believed to be more useful.
Such an approach allows higher efficiency of maintenance services and at the same time decreases the amount of unnecessary maintenance procedures and costs.
Data collection is not the issue but data comprehension is the hurdle.
Today, electric cars create huge amounts of data using battery management systems, telematics, and connected vehicles technologies. Fleet managers’ challenge is not data access but data comprehension.
Information on battery condition, performance performance, usage and environmental conditions can be construed in useful manner. However, these stand alone pieces of data are more useful when they are analyzed and integrated.
To put it simply, any good predictive analytical framework must do more than show reports. It has to help glean data in order to know and understand what is happening.
For example, the objective is to go from:
Battery data → Performance history → Risk analysis → Recommended action
This can assist predictive analytics not only with maintenance but with other operations such as service, warranty and production.
Creating a more proactive EV ecosystem
Predictive analytics won’t completely eradicate battery failure. Batteries work in complex scenarios, hence there remains a chance of unexpected occurrences. However, using predictive analytics allows for the recognition of early warning signs that would transform the way of dealing with battery reliability in the future.
The bigger opportunity here is achieving a more proactive approach in electric mobility, where the state of the batteries is permanently monitored over time, maintenance work is performed based on real risks, and operational choice is determined by data, not calculations.
As the scale of electric vehicle fleets constantly increases, this approach gets of greater importance. The future of battery management won’t mean waiting for battery failures to happen, but learning how to identify the warning signals in advance and acting quickly.
Predictive analytics can facilitate that process.
