What Is Peak Shaving With Battery Storage?

How do demand charges work?

The U.S. Energy Information Administration defines a demand charge as the portion of a customer bill for electric service based on the customer maximum electric capacity usage under the applicable rate schedule. A 2017 National Renewable Energy Laboratory survey explains the mechanics: demand charges are typically based on the highest average usage within a defined interval, usually 15 minutes, during a billing period.

That survey, covering more than 10,000 tariffs, found that demand charges typically represent 30 to 70 percent of a commercial bill and that nearly 5 million commercial customers can subscribe to tariffs with demand charges above $15 per kW. Lawrence Berkeley National Laboratory describes the design variants: non coincident charges on the monthly maximum regardless of timing, peak period charges in a utility defined window, seasonal charges, averaging intervals from 15 minutes to one hour, and ratchets, where billing demand cannot fall below a fixed percentage of the peak billing demand in the previous 12 months.

How does a battery shave the peak?

The controller reads the utility meter continuously and sets a target for net grid draw. When load rises above the target, the battery discharges the difference so the meter sees only the target. When load falls below it, the battery recharges, slowly enough that charging does not create a new peak.

Two ratings decide whether this works. Power in kW must be at least the gap between the facility peak and the target. Energy in kWh must cover that gap for the full duration of the excess, after losses. A battery with enough kW but too few kWh runs dry partway through the peak, and the meter records full load for the rest of the interval.

What does a worked example look like?

Take a facility with a 500 kW monthly peak, a 100 kW battery, and a demand charge of $20 per kW; the $20 figure is illustrative, though the NREL survey found tariffs at or above $20 per kW in several states. The controller sets the target at 400 kW. During the peak the battery discharges up to 100 kW and the meter records 400 kW. Savings are 100 kW times $20, or $2,000 for the month, and $24,000 per year if the battery holds the line in all twelve billing periods.

Now size the energy. Suppose load exceeds 400 kW for two hours, averaging 75 kW above target, so the battery must deliver 150 kWh. At the 85 percent round trip efficiency the NREL Annual Technology Baseline adopts for commercial storage, that takes about 176 kWh of charge, so a 100 kW, 200 kWh system fits with modest margin. If the same peak lasted four hours, 200 kWh would fall short and the month would be billed at a higher peak.

Why does forecasting matter?

Energy is the constraint. The battery can discharge its stored kWh only once before it must recharge, so the controller has to decide which intervals deserve that energy. If it discharges at 2 p.m. against a modest rise and the real peak arrives at 4 p.m., the battery is partly empty when it matters. One missed 15 minute interval sets the demand charge for the month, and under a ratchet it can set a floor for the next year.

A load forecast from interval history, weather, and occupancy tells the controller when the peak is likely and how long it will last, so it can hold energy in reserve and set a target it can defend. On a solar plus storage site the forecast must include solar too, because a passing cloud can lift net load by hundreds of kW in minutes.

What are the failure modes?

Peak shaving fails in predictable ways, and each one shows up on the bill as a higher kW than planned.

• Mispredicted peak: the target is set too low, load crosses it by more than the battery can supply, and the meter records the excess.
• Depleted battery: energy is spent early or the peak outlasts the kWh rating, so the last intervals go unshaved.
• Ratchet clauses: one high interval raises billing demand for up to 12 months, so a single failure costs far more than one month of savings.
• Charging into the peak: recharging too fast, or at the wrong time, creates a new peak the battery itself caused.
• Capacity fade: a battery sized with no margin loses usable kWh over its life and stops covering the duration it was sized for.

Peak shaving vs load shifting vs demand response: what is the difference?

The three strategies use the same battery but target different line items, and they can conflict in the same hour. Load shifting moves energy from high price to low price periods under a time of use rate. Demand response reduces load on request from a utility or aggregator for a payment. Peak shaving targets only the maximum kW, and a controller doing all three must rank them.

What should you look for in peak shaving software?

The software, not the battery, determines how much of the demand charge is removed. Look for a controller that runs locally and keeps shaving when the internet drops; forecasts load and solar and adjusts its target as the month unfolds; respects the limits published by the battery management system; and manages recharge so it never creates a new peak.

Intellect Operate is an energy management system built for this job. It runs its control algorithms on ruggedized edge hardware so demand charge management continues through network outages, forecasts load and solar, reads battery management system limits every second and constrains dispatch to them, and stacks peak shaving with time of use optimization and demand response in one control loop.