ATS

Turning an Energy Management System into an AI-driven flexibility platform

We helped ATS evolve its Energy Management System into an intelligent platform that automatically decides when energy should be consumed, stored or sold based on the conditions and priorities of each customer site.

ATS is a major Belgian engineering and integration group that designs and installs technical infrastructure for industrial customers, including electricity, industrial automation, HVAC and energy solutions.

The business challenge: making energy management proactive

Optimizing energy at an industrial site means balancing multiple variables at once. Production and consumption fluctuate, electricity prices change and assets such as batteries and EV chargers need to respond at the right moment.

Priorities also differ by site. Depending on its financial and contractual situation, the focus might be self-sufficiency, avoiding expensive consumption peaks or taking advantage of changing market prices.

The ambition: building intelligence into the existing EMS

ATS needed a forecasting and optimization layer that could turn these variables into automatic decisions for local energy assets and integrate with its existing EMS at scale.

The approach

1. Forecasting energy demand, production and prices

We developed machine learning models that forecast real-time imbalance and day-ahead market prices, as well as local energy production and consumption over the same horizons. Together, they give the EMS a forward-looking view of each site and the market conditions around it.

2. Automatically optimizing local energy flows

A multi-objective optimization engine uses those forecasts to determine how energy should flow between local assets, balancing objectives such as self-sufficiency, peak shaving, dynamic price arbitrage and real-time responses to volatile imbalance prices.

Batteries, EV charging stations, solar inverters and other controllable assets can then be steered automatically, without an operator deciding when to charge, discharge or curtail.

3. Creating a platform that keeps improving

The platform combines market and grid data, the optimization engine, and an MLOps layer with an API connecting it to ATS's existing EMS. Models are automatically retrained as market conditions, consumption patterns and asset configurations change.

Built on open-source technologies and running on a sovereign private cloud, the platform also lets ATS test optimization strategies against historical data and use those insights to size assets such as batteries for existing and new customer sites.

The outcome: creating more value from every energy asset

End customers can reduce energy costs by shifting consumption to cheaper periods, shaving peaks before they affect capacity tariffs and using or selling stored energy when its value is highest.

For ATS, an EMS that previously focused on monitoring can now actively optimize energy use. Because the intelligence sits behind the existing product APIs, these capabilities can be rolled out across its installed base, creating a stronger and more differentiated product proposition. They can also back-test optimization strategies and use real performance data to size new energy assets before investment decisions are made.