Intelligent EV Charging Network Optimisation

A fast-growing EV charging network operator needed to improve charging throughput, reduce grid strain during peak demand periods, and offer dynamic pricing to drivers — all without requiring hardware changes to existing charging infrastructure. Sequoia delivered an AI-powered intelligence layer that improved efficiency by 45% and reduced peak grid load by 30%.

The Challenge

The operator's network of 200+ stations was experiencing two converging problems. First, peak-period demand was causing grid stress events at high-usage sites, creating reliability risks and incurring demand charge penalties from grid operators. Second, charging sessions were unmanaged — vehicles were charged at maximum rate regardless of departure times or grid conditions, leaving performance optimisation opportunities untouched.

The operator lacked the data infrastructure to understand demand patterns, forecast peak periods, or price sessions dynamically. Decisions about station investment and capacity planning were made on lagging operational data rather than predictive intelligence.

“We had 200 stations generating enormous amounts of session data that we weren't using. We needed to turn that data into decisions — about pricing, about capacity, about when to push charging sessions to avoid peak demand.”

— Chief Technology Officer

Our Approach

Sequoia built a software intelligence platform deployed as a service layer on top of the operator's existing station hardware and OCPP infrastructure:

• Demand forecasting engine — Time-series ML models trained on 24 months of session data, grid pricing signals, and local demand patterns — producing 72-hour ahead forecasts per station cluster with 94% accuracy.
• Smart charging scheduler — An optimisation layer that shifts non-urgent charging sessions into off-peak windows, using driver-consented departure time inputs to guarantee vehicles are charged by need time.
• Dynamic pricing API — A real-time pricing engine that adjusts session rates based on current grid costs and station demand, surfaced through the driver-facing mobile app and displayed on charger screens.
• Operations analytics dashboard — A network-level visibility layer showing real-time utilisation, forecast demand, demand charge risk, and revenue optimisation opportunities across the full station estate.

The Outcome

Grid stress events at peak-demand sites were eliminated in the first month of smart scheduling deployment. Demand charge penalties reduced substantially, with the savings financing the platform cost within the first two quarters.

The forecasting engine now informs capital deployment decisions — the operator uses network-level demand models to identify sites requiring capacity upgrades 6–12 months before utilisation thresholds are reached.

Published by the Sequoia Team