If one bad 15-minute interval has ever spiked your bill, this is for you
If you run a commercial building, an industrial facility, a manufacturing plant, a hospital, a university, or a data center — and the power bill keeps jumping around in ways you can't quite explain — this guide is for you. You're spending more time thinking about electricity than any normal person should, because the number moves and no one has told you why.
Here's the short version up front: peak shaving is not about how many kilowatt-hours you burn over the month. It's about how high your kilowatts climb in the single worst 15 or 30 minutes — because on many tariffs, that one interval sets your entire demand charge.
What's the most cost-effective mix of operational changes, on-site generation, and storage to pull down your worst kilowatt intervals — while keeping your core operation intact and paying back the capital inside your target window?
By the end you'll understand kW vs. kWh, why demand charges are so large, the three levers you actually have, how to size and cost a solution, and the four questions to put to any vendor before they size a battery for you.
Everything about peak shaving rests on one distinction. Energy is the volume you use — kilowatt-hours, the area under your load curve over the month. Demand is the height of that curve at a moment — kilowatts, averaged over a 15- or 30-minute interval. You can hold your monthly kWh flat and still get two very different bills, depending on whether you let a sharp peak register.
Flatten that one red spike and your kWh barely changes — but your demand charge drops with the height. That gap is the entire opportunity.
Utilities build for the highest moment of demand, not the average day — like building a 14-lane highway for the worst traffic a few days a year, because blackouts aren't acceptable. That capacity gets billed back to big users in proportion to how much they push the system at its worst. It's why demand charges run 30–70% of a large C&I bill, and why every kW matters: at $20/kW, one kilowatt shaved off your peak is $20/month — $240/year — per kW.
The whole job is clipping the top off your load curve during the intervals that set your bill. You have three mechanical levers to do it.
Reduce or reschedule non-essential loads during peak windows. HVAC is usually the biggest lever — pre-cool, then let the setpoint drift 2–3°F for a few hours. Add lighting control, process rescheduling (a steel mill times its arc furnace around the system peak), pump scheduling, and smart EV charging.
Run diesel or natural-gas generators — or ramp CHP — during the intervals you care about, so part of the facility comes off the grid. A data center might run generators 2–4 hours on hot afternoons. CHP fits when you have steady demand for both power and heat.
Charge off-peak, discharge during the peak to hold grid demand under a set ceiling. A 500 kW / 1 MWh battery covers a 500 kW peak for two hours. A controller watches the meter in real time and discharges whenever load approaches the threshold.
None of the three works well without controls. An energy management system watches your meter, forecasts peaks from history, weather, and schedules, and dispatches automatically. And you don't need perfect prediction — set a rule like "never exceed 1.5 MW" and the battery discharges whenever load climbs toward it. You're clipping the top of the curve every time it reaches for the ceiling.
Sizing flows straight from your load profile. Pull 12–24 months of bills and interval data, and sub-meter to find what drives the peaks. Then a handful of numbers tell you whether this pays.
| Metric | What it tells you |
|---|---|
| Demand charge rate ($/kW) | The most critical number. Under $5/kW to over $30/kW by region — it sets the value of every kW shaved. |
| Load factor | Average ÷ peak demand. 0.4 or below = tall peaks, high shaving potential. (500 kW avg / 2,000 kW peak = 0.25.) |
| Peak duration | Sets battery energy size. A 2-hour peak with a 500 kW target needs a 1,000 kWh battery. |
| Peak frequency | Daily summer peaks justify capital far better than one or two events a year. |
| Round-trip efficiency | Batteries lose 10–20% — to discharge 1 MWh you charge 1.1–1.2 MWh. |
| CapEx | Batteries ~$500–$1,000/kW; generators ~$300–$600/kW. A 1 MW/2 MWh battery: $500k–$1M+. |
| Payback | Most C&I energy projects target 3–7 years — on your real numbers, not a generic example. |
Operators usually get pitched one technology by one vendor. Here's the frame to judge the pitch — technology follows the duration and shape of your peaks.
| Technology | Best fit |
|---|---|
| Battery (BESS) | Frequent, short peaks (1–4 hrs). Fast response, no on-site emissions, and it can stack value — peak shaving + demand response + TOU arbitrage + solar self-use. |
| Generator | Longer peaks (4+ hrs) or when you also need critical backup. A 1 MW diesel for 4 hrs burns 140–160 gal ($500–$700) plus maintenance — worth it when avoided demand beats that comfortably. |
| CHP | Facilities with steady simultaneous demand for electricity and heat — baseload power that ramps at peak. |
| Hybrid solar + storage | Charge the battery by day, discharge into evening peaks — strong in time-of-use regions. |
The risks that quietly kill the business case
A MW-class system isn't a rounding error. Generators need fuel, testing, emissions compliance; batteries need thermal management and eventual replacement.
Push HVAC or lighting too hard and you hit comfort, safety, or output. A rushed process shift creates bottlenecks that cost more than the savings.
Generators, CHP, and big batteries trigger environmental review, fire code, and utility interconnection — time and money.
Buy a 15-year asset against today's demand charge, and a tariff shift toward TOU can stretch your payback well past plan.
- It's not about cutting total kWh — it's about the peak kW.
- It's not only for giant plants — a 200 kW cut pays at a mid-size building.
- Generators are not the only tool — batteries often win on speed and stacked value.
- You don't need perfect peak prediction — a threshold rule rides the peak in real time.
- It's not set-and-forget — tariffs, operations, and equipment all drift.
Geography materially changes the math. The same asset that barely pays in one territory clears easily in another once you stack program revenue — often $50–$100 per kW per year — on top of avoided demand charges.
| Region | The mechanism to work |
|---|---|
| Ontario | Industrial Conservation Initiative — Class A customers pay Global Adjustment on their share of the year's top 5 peak hours. Cut those hours, cut the fee. (Opt-in May 1–Jun 15; >5 MW auto-Class A.) |
| New York | NYISO ICAP is set by the single highest grid peak hour of the year. Shaving it can cut electricity costs up to 30%. |
| Massachusetts | ISO-NE ICAP works similarly; the Clean Peak standard issues certificates to resources that cut load or supply power during peak windows. |
| California | Demand response as Proxy Demand Response or Reliability DRR; PDR-LSR lets storage bid load down and up. |
Peak shaving cuts the top off your highest spike — with a battery, generator, or brief curtailment — and reduces total grid energy in that moment. Load shifting (its own guide, Energy Decision #11) moves consumption to cheaper hours without reducing total kWh. They're complementary — a plant might shave with a battery and shift by running pumps at night — but when a vendor uses them interchangeably, slow down.
When peak shaving is a winner — and when it's a poor fit
- Your demand charge is a real $/kW number and your load factor is low (tall peaks).
- Peaks are frequent and predictable enough to justify the asset.
- You've tried low-cost load moves first and measured them.
- The project pays back in your window — and stacks program revenue where available.
- Peaks are rare — once or twice a year won't carry the capital.
- A vendor sizes hardware before anyone read your tariff or load profile.
- Shedding would hit comfort, safety, or output.
- A tariff shift could strand a long-lived asset.
The best peak-shaving strategy uses your real $/kW rate and load profile to clip your worst intervals with the simplest mix that keeps operations intact and pays back in your window.
Read the bill and the load profile first. Try the free load moves next. Earn the right to spend capital last.
This is Energy Decision #12 in the complete C&I energy management series — 100 decisions, every one that matters. Read the rest of the library at Energy Answers.
| Peak shaving | Cutting the highest kW spike during the intervals that set your demand charge. |
| kW vs. kWh | kW is the height of your load (what demand charges bill); kWh is the total volume (energy charges). |
| Demand charge | A charge on your highest 15- or 30-min demand — 30–70% of a large C&I bill. |
| Load factor | Average ÷ peak demand. Below 0.4 signals tall peaks and strong shaving potential. |
| BESS / CHP | Battery Energy Storage / Combined Heat & Power — two of the shaving assets. |
| Round-trip efficiency | Energy you get back out of a battery, ~80–90%; the rest is loss. |
| Demand response | Getting paid ($50–$100/kW/yr) to cut load on request — stacks on demand savings. |
| Stranded-asset risk | A long-lived asset losing its payback when the tariff structure changes. |
Energy Answers · by Daniel Burke · Energy Decision 12 · C&I Peak Shaving