30% Reduced False Positives in EV Fleet

Nearly one in three batteries a fleet reports as "failed" hasn't failed at all.

That was the first hard number to come out of a preventive-maintenance program Fawkes Energy built for one of India's largest battery-as-a-service operators and it reframed how the company thinks about every service call, every replacement, and every rupee of warranty exposure across its fleet.

The operator

Our partner is a leading Indian battery-as-a-service provider for electric two and three wheelers. They lease and manage battery packs for commercial use, with more than 20,000 assets deployed in the field. Their promise to customers is simple: affordable, reliable, accessible batteries-as-a-service, with end-to-end lifecycle management.

That promise gets harder to keep as the fleet grows. Every additional battery in the field is another asset that can throw a fault, trigger a complaint, or pull a vehicle off the road. At fleet scale, the operational drag compounds fast.

The problem: Service breaks as the fleet scales

Before the program, battery operations were fundamentally reactive. A driver reported a problem, a service team investigated, a fault was diagnosed manually, and often the battery was replaced. That model works at a few hundred assets. However at twenty thousand, it breaks.

Five pressures were building at once:

The common thread: without asset-level intelligence, the operator was managing symptoms instead of causes, and paying for it in cost, downtime, and unnecessary replacements.

What we built

Fawkes Energy built a custom, fleet-scale preventive-maintenance layer on top of FawkesCore, our battery intelligence platform. It ingests raw telemetry from deployed assets and turns it into decisions service teams can act on.

Under the hood, the platform moves data through four layers of battery intelligence:

• Descriptive: what happened
• Diagnostic: why it happened
• Predictive: what is likely to happen next
• Prescriptive: what to do about it

On top of that stack sit the analytics the operator actually uses day to day: fleet-level asset health monitoring, failure prediction and risk scoring, a service-prioritization engine that ranks assets by real risk rather than by complaint volume, and plain-language maintenance recommendations: monitor, inspect, intervene, investigate, or replace.

The program began as a 100-vehicle pilot and is scaling through the fleet toward full 20,000-asset coverage.

The shift: from reaction to prediction

The single biggest change isn't a feature - it's the operating model.

Most fleet operators run on a reactive loop:

‍

Every battery becomes a continuously monitored asset whose health, risk, and economics are actively managed rather than passively observed. Three findings from the field show what that shift is worth.

Proof 1: 30% of "battery failures" weren't the battery

Drivers reported the familiar symptoms: reduced range, poor acceleration, unexpected shutdowns. The default response was to inspect or replace the battery, because the battery is the most visible component.

By combining voltage response, current behaviour, temperature signatures, cohort comparison, and service history, FawkesCore found a large cluster of these "failed" batteries were electrochemically healthy: normal capacity retention, stable internal resistance, normal charge acceptance, no abnormal degradation. The real culprits sat elsewhere: elevated motor current demand, drivetrain inefficiencies, controller calibration, connector and contact resistance.

Roughly 30% of reported battery failures were not battery failures at all. The operator could redirect those cases to vehicle-level diagnostics instead of battery replacement, therby cutting unnecessary swaps, spare-inventory pressure, and investigation time, and pointing engineering effort at the actual root cause.

Proof 2: catching a fast-degrading cohort before it failed

Not all batteries age at the same rate. Across different vehicle platforms, duty cycles, environments, and battery generations, degradation is uneven. Historically it only became visible after failures had already piled up.

Using cohort-level analysis, FawkesCore compared batteries of similar age, mileage, charging behaviour, and environmental exposure, and flagged a specific cohort pulling away from the fleet baseline: faster resistance growth, higher operating temperatures, quicker capacity fade, growing imbalance. Conventional BMS still classified these batteries as healthy. Physics-informed degradation models showed deeper insights.

That gave the engineering team months of advance warning: time to increase monitoring, review supplier-level manufacturing variation, and prioritize inspections before the cohort turned into a wave of field failures. At fleet scale, a single cohort issue can touch thousands of assets at once; catching it early is the difference between a planned review and an emergency downtime.

Proof 3: finding the environmental stress that quietly ages batteries

Degradation is often blamed on chemistry or manufacturing. In the field, one of the largest drivers is environmental stress. It does its damage silently, long before capacity loss becomes obvious.

Analysing temperature distributions, charge behaviour, daily utilization, geography, and rest periods, FawkesCore isolated a subset of batteries under chronic thermal stress. They weren't overheating. They were simply spending long stretches at elevated temperatures during peak ambient conditions, back-to-back fast-charging sessions, and high-payload duty cycles. Without contextual analysis they looked completely normal.

That reframed the goal from replacing degraded batteries to preventing degradation in the first place through charging-policy changes, operational scheduling, and driver-behavior guidance. For a battery-subscription business, extending useful life across thousands of assets flows straight into asset ROI and residual value.

Why it matters

Taken together, the three findings mark a shift from service management to fleet reliability engineering. The operator stops reacting to whichever component shouts loudest and starts managing each battery as a measurable asset - its performance, its risk, and its lifecycle value.

The same intelligence does double duty. The cohort and stress analysis feeds procurement and R&D - quantifying how different cell chemistries, suppliers, and pack designs actually perform in the field. And the same health data underpins end-of-life decisions: which batteries are fit for resale, which for second-life stationary storage, and which for recycling. Batteries stop being depreciating assets and become managed lifecycle assets with measurable residual value.

Looking ahead

As the program scales from its pilot toward the full 20,000-asset fleet, these insights compound across operations, warranty, procurement, servicing, and second-life value. This value only grows as their operations scale to 1 lakh assets eventually.

We're not building a battery monitoring dashboard. We're building the predictive operations layer that lets electric fleets scale efficiently, reliably, and profitably. As India's electric-mobility ecosystem enters its next phase of growth, preventive maintenance and lifecycle intelligence stop being nice-to-haves and become the foundation the whole business runs on.

‍