ProductionVerified Real deployment at a named or precisely described company, with a measured result from a primary source.

AI cuts plant energy cost: demand-based ventilation control, price-aware production plans and cross-plant energy benchmarking (Stellantis, Casuso Propellers, Jotun)

Added 2026-11-09 · Last updated 2026-11-09

Manufacturers in this case: Casuso Propellers · Jotun · Stellantis · all manufacturers using AI

Role

Energy management, plant engineering and production planning · Typical users: plant manager, energy manager, facilities or utilities engineer, production planner, group operations lead

The problem

Energy is a large and volatile cost, and the people who plan production rarely see it. In foundry processes, according to the vendor Evenbytes, energy can account for up to 40% of total production costs, and with hour-by-hour electricity prices it is hard to know when to melt and pour. Paint shops are among the most energy-intensive parts of automotive manufacturing: at Stellantis' Rüsselsheim plant (Opel and DS Automobiles) ventilation had to be cut without hurting working conditions. At Jotun, a paint and coatings maker with more than 40 plants, each plant had its own energy monitoring but there was no common baseline, so nobody could say which factory was most efficient for a given product, and production planning ignored energy.

How AI helps

Three different ways to use AI on the same cost line. (1) Predictive control of utilities: at Stellantis, etalytics' etaONE platform adjusts ventilation rates and cooling in real time from production needs and ambient conditions, schedules fresh-air intake when conditions are favourable, uses the building's thermal inertia to limit daytime cooling and allows ventilation to be shut down safely during low-load periods. (2) Price-aware production planning: at Casuso Propellers, a marine propeller manufacturer, Evenbytes' Forewatt plans production around hour-by-hour energy prices, available resources and plant capacity, and produces an optimal daily or weekly consumption plan with no new hardware. (3) Cross-plant benchmarking: at Jotun, Greenovative integrated energy and production data from plants in India, Oman, the UAE, Saudi Arabia, Qatar and Egypt into one model, set energy baselines by equipment, process, plant and product category normalised for batch size and throughput, and its AI layer shows how product mix and batch size influence energy so that energy becomes an input to regional production allocation. In each case people keep the decisions: the AI recommends or controls within limits that engineers set.

Step by step

  1. Find out where the energy goes: sub-meter or at least list your five largest consumers (furnaces, compressors, ventilation and cooling, ovens, paint booths) with their hourly consumption.
  2. Get your electricity tariff and, if you buy at spot or time-of-use prices, the hourly price history, and compare it with when your biggest loads actually run.
  3. Write down what cannot move: quality and temperature limits, safe start-up and shutdown rules, delivery dates, labour shifts. These become the constraints for any AI recommendation.
  4. Pick one lever for a pilot: ventilation or HVAC control, shifting a flexible load such as melting or heat treatment to cheaper hours, or comparing similar lines or plants on energy per unit produced.
  5. Run the AI in advisory mode first: compare its recommendations with what your engineers would have done, and switch to automatic control only inside limits they set.
  6. Measure against a baseline normalised for production volume, weather and product mix (energy per unit, not total kWh), and ask for the method to be auditable.

Tools

  • etalytics etaONE (AI-based predictive control of ventilation and cooling): used at Stellantis Rüsselsheim
  • Evenbytes Forewatt (prescriptive AI that plans production around energy prices): used at Casuso Propellers
  • Greenovative (AI energy intelligence and cross-factory benchmarking): used at Jotun
  • Other tools to evaluate: the energy management module of your SCADA or MES, your electricity supplier's price and demand data, a sub-metering system on your biggest consumers
  • To test the idea without new software: hourly meter data and a production schedule in a spreadsheet, plus any LLM (ChatGPT, Claude, Copilot, Gemini) with the prompt below

In the AI tools index: etalytics etaONE · Evenbytes Forewatt · Greenovative · all AI tools used by manufacturers

Result

Stellantis Rüsselsheim paint shop, etalytics etaONE: over 60% reduction in ventilation energy consumption and a significant drop in compression cooling, with stable indoor climate; the project was part of the DiNaPro research initiative with TU Darmstadt, and the page gives no kWh or euro figure (etalytics success story, a vendor publication). Casuso Propellers (marine propellers), Evenbytes Forewatt: electricity bill cut by 8% to 17%, described as verifiable and auditable (Evenbytes case page; no baseline or period published). Jotun (paints and coatings, 40+ plants), Greenovative: 18 to 20% improvement in group-level energy performance, driven by better production allocation and fewer process inefficiencies, plus identification of best-in-class factories, processes and equipment per product category (Greenovative case study, 24 February 2026, a vendor publication; the metric is not defined in kWh).

As reported by the source; most figures are vendor or self-reported and not independently verified.

Human review: cautions

  • All three results come from vendor publications. etalytics and Evenbytes give a percentage with no kWh, euro value, baseline or period; Greenovative's 18 to 20% is an 'energy performance' improvement without a stated unit.
  • The three cases are different levers (utility control, production planning, benchmarking), so their percentages cannot be compared or added.
  • Production planning around energy prices can conflict with delivery dates and labour shifts: keep the constraints explicit and keep a planner in charge.
  • Cutting ventilation or temperature set-points touches worker comfort and process quality: set limits with EHS and quality before allowing automatic control.

Try it yourself

You are an energy manager at a manufacturing plant looking for the first AI pilot to cut energy cost.

I will paste: (1) hourly or daily energy consumption by main consumer for the last 3 to 12 months [furnaces, compressors, ventilation and cooling, ovens, other]; (2) our electricity tariff or hourly prices; (3) our production schedule or output per day; (4) the constraints that cannot change [temperature and quality limits, shift patterns, delivery dates].

Do this:
1. Rank the consumers by energy and cost, and say which formulas to use in a spreadsheet to compute energy per unit produced; do not compute them yourself.
2. Point out the three biggest gaps between when we use energy and when it is cheap, or between similar days with different consumption, quoting the rows.
3. For each gap, say whether the lever is control (ventilation, cooling), planning (moving a flexible load) or benchmarking (comparing lines or plants), and what could stop us from using it.
4. Propose one pilot with a baseline, a measurement method that corrects for volume and weather, and a safe limit for automatic control.
5. Write 10 questions to ask a vendor (how the savings are measured and audited, hardware needed, how the system behaves if data is missing, who keeps control, reference plants like ours, price).

Do not invent savings figures for us. Use only the data I paste in; if something is missing, write 'not found' or list it as a question instead of guessing. Do not send, order, post or change anything in any system: produce drafts for my review and approval.

Want more like this? Join the newsletter

Related cases

All 21 Production use cases →

Sources

Weekly: 1–2 practical AI use cases for your plant

Newsletter in

Free, one email a week. By subscribing you agree to receive the weekly newsletter (and its sponsor message) from AI Agents for Manufacturing. Every email includes an unsubscribe link, and you can unsubscribe anytime. See our Privacy Policy.