How to reduce peak demand charges: a practical playbook

For most commercial and industrial accounts, the single largest lever on the power bill is not how much energy you use. It is how hard you pull at your worst moment. Demand charges bill you on that peak, and they commonly run 30 to 70 percent of a commercial electricity bill. Cutting the peak, even a little, moves real money every month.

This playbook walks through what a demand charge actually measures, how to find your own peaks, the load-management and peak-shaving moves that reduce them, and how to keep the savings from drifting back.

What a demand charge actually measures

Energy charges bill the total kilowatt-hours you consume over the month. A demand charge is different: it bills the highest average power draw, in kilowatts, across a short window. Most utilities use the peak 15-minute interval in the billing period. One bad quarter-hour sets the charge for the whole month.

Rates vary widely by utility and region, but $10 to $25 per kW per month is common, and some high-cost markets exceed $30 per kW. At $15 per kW, a 400 kW peak is $6,000 a month in demand charges alone, before a single kilowatt-hour of energy is counted.

Ratchet clauses make one peak last a year

Many tariffs add a ratchet: your minimum billable demand is set at a percentage, typically 70 to 80 percent of your highest peak over the prior 11 to 12 months. Set a new peak in one hot afternoon and you can pay against that floor for the next year, even in months when your real demand is far lower. That is why controlling the single worst interval matters so much.

Step one: find your peaks before you try to cut them

You cannot manage a peak you cannot see. The demand line on the bill tells you the number, not the cause. Interval data tells you the story: which quarter-hour set the peak, what else was running at that moment, and whether it was one event or a repeated pattern. Start here.

1. Pull at least 12 months of interval (kW) data alongside the demand line on each bill.
2. Identify the peak interval in each billing period and what it coincided with (startup, shift change, a hot afternoon).
3. Separate structural peaks (they repeat) from one-off spikes (a single event you can prevent).
4. Confirm the tariff: interval length, ratchet percentage, and any coincident-peak or time-based demand windows.

Step two: cut the peak

Stagger equipment startup

Simultaneous startup of chillers, air handlers, lighting and elevators is one of the most common causes of a needless peak. Sequencing those startups over several minutes, rather than all at once, is among the fastest ways to shave demand in commercial properties, and it costs nothing but a controls schedule.

Shift flexible loads off the peak window

Battery charging, pumping, non-urgent process loads and EV charging can often move to hours when the building is already below its peak. The goal is to flatten the curve so no single interval spikes above the rest.

Pre-cool and use thermal mass

Cooling the building or a thermal store ahead of the daily peak, then coasting through it, shifts HVAC demand out of the expensive window. Reviews of peak load management in commercial buildings report that these methods can deliver meaningful reductions in total electricity cost, with peak load management commonly cutting 20 to 30 percent off electricity expenses for sites with sharp, controllable peaks.

Peak shaving with battery storage

Where peaks are short and predictable, a battery can discharge to cap demand. Analysis from the U.S. National Renewable Energy Laboratory found that more than five million U.S. commercial customers face demand charges above the roughly $15 per kW threshold where storage starts to pay for demand management. Commercial peak-shaving projects commonly report demand-charge reductions of 30 to 50 percent at facilities with sharp peaks.

Fix power factor

If your billed demand is driven up by low power factor, correction can lower it without changing a single operation. That is a separate mechanism worth checking on its own; see our note on power-factor penalties.

Step three: match the effort to the payoff

Not every tactic fits every site. Match the move to your load shape and to how much the tariff rewards it.

Step four: hold the gains

Peaks creep back. New equipment, a warm summer, a changed shift pattern, and last year's win quietly erodes. The fix is to make peak review a monthly habit rather than a one-time project: check each bill's peak interval against target, watch the ratchet floor, and catch a new peak the month it happens instead of a year later.

That habit only works if the data is clean and current. MartinAI structures your bills and interval data, ties each peak to the tariff in force, and surfaces the intervals that set your demand charge, so the monthly review takes minutes instead of a spreadsheet marathon.