Most building owners and plant operators want to use less electricity. But how much you use is only part of the cost. When major equipment runs—and what else is running at the same time—can also affect the bill.
AI and predictive controls can help teams anticipate demand and coordinate equipment more effectively. The value comes from reducing costly peaks while keeping buildings comfortable and production running.
Why timing matters
A facility can improve its energy efficiency and still face a costly demand peak.
Consider a plant where compressors, process cooling and other large loads all ramp up together. Each system may be doing its job, but their combined demand deserves a closer look. Can some equipment start earlier? Can a flexible load wait? Which operations cannot move?
Electricity consumption, measured in kilowatt-hours, tells us how much energy was used. Demand, measured in kilowatts, tells us the rate of use over a particular interval. Managing one does not automatically manage the other.
For participating Ontario Industrial Conservation Initiative customers, Global Adjustment costs depend on their contribution to the province’s five highest-demand hours during the base period. The IESO’s Peak Tracker explains why timing matters for these customers.
What does the research show?
A 2026 study from Lawrence Berkeley National Laboratory examined predictive controls that combined a building model with machine learning to coordinate heat-pump operation under dynamic electricity pricing.
In experiments at an instrumented test facility, the approach reduced daily HVAC energy costs by 8.7% and peak demand by 29.2% compared with the baseline controls.
These figures describe the tested HVAC system and pricing conditions, rather than savings on an entire building’s utility bill. They demonstrate why demand reduction and cost reduction should be measured separately.
For an owner, the useful question is what a similar approach could achieve under their own tariff, equipment configuration and operating requirements.
Where AI can help
Building and plant teams already work with schedules, meters and controls. AI can help them make better use of that information.
Weather, occupancy and operating data can support forecasts of upcoming loads. Predictive controls can then help determine when equipment should run, within limits set by the operating team.
In a commercial building, that might mean adjusting cooling ahead of a peak period while maintaining temperature and ventilation requirements. In a plant, the review might focus on how flexible support loads overlap with production.
Reliable data and functioning equipment come first. Some improvements can be achieved with conventional controls; the choice should follow the problem being solved.
Compressed air: a measurable savings opportunity
In industrial plants, compressed air is a good place to look.
The U.S. Department of Energy’s guidance on compressed-air leaks notes that leaks can waste 20–30% of a compressor’s output. That is lost air capacity, not an equivalent percentage reduction available on the facility’s electricity bill.
Capturing savings requires more than finding leaks. Repairs need to be followed by controls that match compressor operation to the reduced air requirement.
Better sequencing, appropriate pressure settings and storage can help avoid unnecessary compressor operation. For peak-demand management, teams should also examine how compressors overlap with cooling, pumps and other large loads.
Any adjustment must protect required air pressure, air quality and production continuity. Our article on compressed-air optimization explores the broader efficiency opportunity.
An interesting development: storing cooling as ice
One interesting development evaluated in the U.S. is modular ice-based thermal storage: making ice ahead of peak periods, then using that stored cooling when electricity demand is high.
Berkeley Lab evaluated Nostromo’s IceBrick system at California’s Beverly Hilton. Its September 2025 report provides preliminary findings from one site.
Ice storage already exists commercially. The opportunity is to coordinate stored cooling with forecasts, electricity prices and operating needs. Suitability depends on space, required cooling temperatures, installation costs and the tariff.
Start with the operating data
Review electricity data, equipment schedules and the bill. Identify what creates the peak and which loads offer real flexibility.
Then test a defined change. Track demand, consumption and cost separately, alongside comfort or production performance.
That gives owners a practical basis for deciding what to improve next—and whether a larger investment is justified.
Contact Kontrol Technologies to discuss your building or industrial energy-management priorities.