Peak shaving reduces a specified measured maximum electricity demand through load control, generation or energy storage. Its defining outcome is a lower relevant peak, not simply fewer annual kilowatt-hours or a battery operating during expensive hours.
Before evaluating battery storage for this purpose, ask which measured peak the proposal intends to lower and how it establishes that result. The answer needs a meter boundary, a measurement interval and a feasible operating sequence.
Our commercial demand-charge guide retains the account-specific tariff and commercial buying decision. This entry explains the peak-shaving concept and its measurement. All numerical examples are original hypotheticals, not utility rules, product specifications, Sunburst project results or savings promises.
Which peak is being shaved?
“Peak” requires a reference. It could describe a facility’s highest utility-meter demand, the highest demand during specified hours, a controller’s instantaneous reading or the loading on a particular electrical component. Those values may differ.
A proposal should name the boundary and the period. Reducing one appliance’s maximum does not prove that the whole facility’s billed peak changed. Another load may start at the same time, or another interval may become the determining maximum.
The Department of Energy identifies peak shaving as a storage application aimed at reducing peak demand. Its 2025 FEMP presentation supports that application distinction; it does not establish a particular site’s financial outcome. FEMP battery implementation presentation.
For a buyer, the most useful statement is specific: the model lowers a named demand quantity from one value to another under stated assumptions. “The battery shaves peaks” is a capability description. It still needs evidence that the relevant peaks occur when that capability is available.
How do kilowatts and kilowatt-hours fit together?
Kilowatts describe power. Kilowatt-hours describe energy accumulated over time. A peak-shaving action needs enough power to reduce demand and enough energy to sustain that reduction.
In a fictional constant-output example, supplying 4 kW for half an hour delivers 2 kWh. Supplying the same 4 kW for two hours delivers 8 kWh. The power reduction is equal, while the energy requirement is four times larger.
That distinction matters when two proposals offer similar battery energy but different output limits, or similar output limits but different energy. A large energy number cannot substitute for the required sustained power. A high short-duration output number cannot substitute for a longer energy requirement.
State the boundary of the quantities. Energy described as already delivered to the load should not receive another assumed discharge-loss deduction. Charging input and usable stored energy may use different boundaries and therefore need their own treatment.
A peak-shaving calculation is also not an appliance-starting calculation. The equipment may need to satisfy both, but their durations, measurements and constraints differ. Do not select a battery from the peak-shaving arithmetic alone.
Why does the demand interval matter?
A meter can integrate energy over an interval and report an average power for that interval. An instantaneous screen reading does not necessarily equal the quantity used to determine a demand charge.
Suppose a fictional account uses fixed 15-minute intervals and records 5 kWh in one interval. Fifteen minutes is 0.25 hour, so the corresponding average demand is 5 ÷ 0.25 = 20 kW. Treating the 5 kWh number as 5 kW would misread the demand.
Now suppose a supported battery supplies 4 kW throughout that full interval, delivering 1 kWh. With all other assumptions unchanged, meter imports fall from 5 kWh to 4 kWh, and interval-average demand falls from 20 kW to 16 kW.
This is an arithmetic illustration, not evidence that an actual account uses that interval or meter arrangement. Confirm fixed versus other interval methods, timestamps, applicable windows and how data fields are defined before reconstructing a billed result.
DOE’s rate-evaluation guidance recommends examining bills, interval data and the account’s rate structure. It distinguishes energy charges from demand charges and notes that different components respond differently to a project. DOE utility-rate evaluation.
Does a brief battery response lower the full interval by its rated power?
Not necessarily. If output lasts for only part of the measuring interval, its contribution to the interval average is smaller than its power during that brief action.
Continue the fictional 15-minute example. Supplying 4 kW for five minutes delivers 4 × 5/60 = 0.333 kWh, approximately. Spread across the full 0.25-hour interval, that reduces average imports by about 1.333 kW, not 4 kW.
If the rest of the example remains unchanged, measured demand falls from 20 kW to about 18.667 kW. The same output sustained for all 15 minutes would produce the larger 4 kW average reduction described earlier.
This example explains why duration and timing belong in the claim. It does not prescribe when a controller should respond or whether a particular battery can provide the output. Exact control behavior and operating constraints need documented support.
Our continuous battery power entry explains condition-qualified sustained output. A timed peak rating or surge-power rating answers a different question and cannot automatically demonstrate an interval-long contribution.
How is peak shaving different from load shifting?
Load shifting changes when electricity is used or supplied from storage. Peak shaving describes a reduction in a relevant maximum. One action can achieve both, but neither name proves the other result.
Moving a task from one hour to another may lower the original peak. If it joins other tasks in the destination hour, it may create an equally high or higher replacement peak. The schedule changed, but the relevant maximum may not have fallen.
Storage can reduce measured imports while the underlying appliance schedule stays unchanged. In that case, the facility still requires the same power for its operations, while another source supplies part of that demand at the meter boundary.
Load shedding describes disconnecting selected demand under a supported control sequence. It may help lower a peak, but the resulting service interruption and restoration need separate approval and design. Do not assume that every load is available to interrupt.
Keep the objective explicit. Reducing energy cost during expensive periods, lowering a demand maximum and preserving backup energy can require different control decisions. A proposal should show which objective governs when they conflict.
What baseline does the calculation need?
Use credible interval data for the relevant facility and billing periods. A monthly total does not show the shape, timing or duration of the peaks a control system must address.
Identify unusual events, missing data, changed operating hours, new equipment and expected future demand. If a business recently added charging or refrigeration, an older load profile may no longer describe the proposed operating case.
State whether the baseline is measured, reconstructed or forecast. Each can be useful, but the uncertainty differs. A forecast built from equipment schedules should not be presented as a measured record of actual demand.
The solar-versus-storage comparison retains the broader capital-allocation decision. For the peak-shaving portion, request the same baseline in every alternative so differences reflect the proposed actions rather than inconsistent load assumptions.
Also preserve required operations. A baseline and revised case should deliver comparable services unless a deliberate operational change is disclosed. Canceling a production shift or reducing required cooling is more than supplying the same load from a battery.
Why must the model examine more than the highest interval?
After one maximum is lowered, another interval can become the new maximum. The demand benefit is limited by that next interval unless the plan addresses it too.
Suppose a fictional month’s two highest interval demands are 20 kW and 19 kW. A battery lowers the first to 16 kW but does nothing during the second. The revised maximum is 19 kW, so the month’s measured peak falls by 1 kW, not 4 kW.
If the battery also lowers the 19 kW interval, the model must examine the next candidate. This process continues across the relevant period. It is not enough to show a single attractive event with a large reduction.
A useful proposal includes a ranked set of candidate peaks and the supported action during each. Ask how it handles events on consecutive days, depleted stored energy, different production conditions and any unavailable equipment.
This does not mean a missed event always determines every bill. The tariff defines the applicable periods and quantities. It means the model must identify the events that could actually control the relevant maximum instead of selecting only the easiest event to reduce.
How long must the battery sustain the reduction?
For a simplified event, energy equals the power contribution integrated over its duration. A flat 4 kW contribution for two hours needs 8 kWh delivered at the chosen boundary. A brief 4 kW contribution needs less energy but may not address a prolonged demand plateau.
Consider a fictional proposed 4 kW reduction for three hours. That needs 12 kWh delivered if the contribution is constant. A system with 8 kWh available at that same boundary can supply that contribution for only two hours under the simplified assumptions.
Extending the duration requires another supported source, a lower contribution or a different load plan. Do not solve the shortfall by treating nominal capacity as extra available energy or counting an untouched reserve as freely dispatchable.
Read usable battery capacity and battery reserve separately. Ask which energy quantity the model uses, what is held back and whether the starting state is achievable before each event.
A more realistic demand profile may require varying output. Then calculate each interval’s contribution rather than assigning one flat value to the whole event. The simplified examples teach the relationship; a qualified design needs the actual sequence and constraints.
Can charging create a new peak?
Yes, if charging adds demand during a period that matters to the chosen maximum. Calling an hour “off-peak” for energy prices does not establish that it is excluded from every demand calculation.
Suppose a fictional facility imports 14 kW during a charging interval. Adding 6 kW of battery charging raises imports to 20 kW if no other source or load change offsets it. That new interval can undermine a plan intended to keep the relevant maximum below 18 kW.
This result depends on the account’s measurement rules and the assumed simultaneous flows. It is not a universal prohibition on grid charging. A feasible plan might use a different window, limit charging power or use otherwise available solar, if the equipment and permissions support that sequence.
Require the charging schedule in the model, not only the discharge curve. Include the energy needed to refill storage, losses at consistent boundaries and the time available before the next demand event.
Check rebound demand after load control too. Deferred tasks may restart together, creating another maximum. The restoration sequence therefore belongs in the peak-shaving assessment rather than being treated as an unrelated operating detail.
Why can measured demand fall while billed demand stays the same?
The measured current-period peak and the number charged under a tariff can differ. Historical look-back rules, minimums, contract values or other provisions may control the bill even after current measured demand falls.
For a fictional example, assume billing demand is the greater of the current measured peak and 80% of a prior 30 kW peak. The historical floor is 24 kW. Lowering current demand from 22 kW to 18 kW leaves the billing quantity at 24 kW in this invented rule.
This example is not a South Carolina tariff. It demonstrates why a current measured reduction cannot be converted directly into dollars without reproducing the applicable billing calculation.
DOE’s rate guidance describes demand look-back methods and several demand-charge structures. Use the account’s actual documents to determine which, if any, apply. The commercial demand-charge guide covers that detailed verification process.
Do not conclude that a measured reduction has no possible value merely because one bill remains unchanged. It may matter under other conditions or periods. The proposal should state which value is supported now and which depends on future events, without promising those events.
What would a simple demand-charge calculation look like?
Suppose an invented tariff charges $12 per billed kW for one monthly demand component. If a verified model lowers that billed quantity from 20 kW to 18 kW, the gross reduction in that component is 2 × $12 = $24.
The calculation assumes no other rule overrides the quantity and that the stated price applies to the difference. It excludes energy charges, other demand components, fixed charges, riders, taxes and project costs.
If the model proves only a measured reduction, the $24 conclusion is premature. First establish that the measured change reaches the billed quantity. If a second demand component uses a different window, calculate that component separately.
Do not multiply the most favorable month’s result by twelve without examining the other months. Seasonal operations, production, demand windows and prior peaks can change the opportunity. A year-long model needs a year-long sequence of supported assumptions.
Our commercial solar ROI guide owns the full investment analysis. This glossary provides the narrower demand-reduction test that a financial model should receive as an input.
Does solar always shave a demand peak?
No. Solar must contribute during the relevant demand intervals to reduce measured imports at that boundary. Annual production does not show whether that timing relationship exists.
An evening peak may receive little direct solar contribution. A daytime peak can occur during weak production. A different interval can become the maximum after a sunny interval is lowered. Those possibilities belong in the model.
Compare synchronized facility demand and solar production rather than subtracting nameplate solar capacity from the facility’s maximum. Nameplate capacity is a rating, not a statement of production during a particular interval.
Where storage is proposed alongside solar, show the incremental case. Identify what solar already changes, what storage adds and how storage charging interacts with imports or exports. Avoid crediting both technologies with the same reduction independently.
A conservative proposal can assign little or no demand-charge value when the evidence is weak while evaluating solar’s energy value separately. An unsupported demand assumption should not be used to make an otherwise uncertain project appear financially attractive.
What should the controller and measurement system support?
Ask which measured quantity the controller observes, where that measurement occurs and how its actions relate to the utility’s demand quantity. An equipment-local measurement may not capture all facility loads.
Describe supported actions, response behavior and dependencies. The system may control a battery, selected loads or charging equipment; that does not prove it can control every load required by the proposed strategy.
DOE’s fleet charging guidance emphasizes vehicle schedules, operational readiness, building data, rates and communication among system elements. Those are useful planning principles for controllable charging, not proof that a particular charger or protocol is installed at your site. DOE smart-charge management guidance.
Our EV load-management guide covers the related charging-capacity decision. A peak-shaving design still needs to show that the required vehicle energy can be delivered before departure while respecting the relevant demand target.
Ask what happens when measurement, communication or equipment is unavailable. Require documented fallback behavior and commissioning evidence. Do not assume a generic “smart” label guarantees that every demand event will be handled.
How do backup and other operating priorities interact?
Stored energy assigned to peak shaving may be unavailable for an outage unless the plan preserves a reserve or supports another documented priority. A financial model must not treat the same energy as unrestricted daily dispatch and untouched emergency supply.
Similarly, an external event or another operating mode may change dispatch. The design should state which objective takes precedence and what that choice does to the peak-shaving estimate.
Use separate scenarios when needed: ordinary operation, higher reserve, low production, depleted energy and unavailable control. The reader should be able to see which savings assumptions depend on ideal conditions.
The point is not to maximize one metric at the expense of the facility’s purpose. Essential loads, business operations and agreed resilience requirements belong in the constraint set before selecting a strategy.
What should South Carolina buyers verify locally?
Identify the utility account and current tariff before assigning a financial benefit. Residential and commercial accounts need their own checks; a demand charge should not be assumed merely because storage is being considered.
Record the measurement interval, billing windows, demand units, historical provisions, minimums and applicable solar or storage conditions. Confirm charging and export permissions separately from the manufacturer’s general capabilities.
Use the utility directory to start the relevant account discussion. A nearby business’s tariff or another state’s rate cannot establish your own billing treatment. This entry intentionally publishes no local demand price, eligibility promise or utility-specific controller setting.
For a commercial facility, confirm the project scope with the assessment team rather than assuming that every storage size or operating application is offered. A glossary definition describes the concept; it does not establish an equipment lineup or installation commitment.
A proposal worksheet for the peak-shaving claim
Request a short evidence record that connects the claimed benefit to the operating plan. It should be possible to trace a dollar result back to a billed quantity and then to a supported physical action.
| Field | Required explanation |
|---|---|
| Objective | Which maximum is intended to fall |
| Boundary | Meter or other measurement point |
| Interval | Duration, timestamps and applicable windows |
| Baseline | Measured, reconstructed or forecast data |
| Contribution | Supported power and duration in each event |
| Energy sequence | Starting energy, discharge, recharge and reserve |
| Competing peaks | Other intervals that could control the maximum |
| Billing conversion | Exact rule connecting measured and billed demand |
| Exceptions | Overrides, outages, missing data and fallback behavior |
| Verification | Commissioning and ongoing comparison records |
Keep unknowns visible. Missing interval data or an unconfirmed tariff should remain an unresolved input rather than becoming a guessed number hidden inside an annual savings figure.
How should costs and verification be assessed?
Separate gross demand-charge reduction from the cost of achieving it. Hardware, installation, required electrical work, controls, commissioning, support and ongoing costs need the actual project scope. Do not assume every option has the same expense.
Compare operational changes and other feasible measures where appropriate. A supported schedule adjustment can differ materially from installing storage. Preserve required service in every alternative so the comparison remains useful.
For verification, retain synchronized meter, generation and dispatch records where available. Note changes in occupancy, production, weather and equipment. A lower peak in a quiet month is not automatically evidence that the controller caused the entire difference.
Ask for a documented explanation when the model and actual bill diverge. The cause might be a different determining interval, a billing provision, an unavailable control action or a changed load. Identifying the cause is more useful than treating every discrepancy as normal variation.
What should you bring to an assessment?
Bring current bills, available interval data, tariff documents, operating hours, major load schedules and any existing solar or storage details. Include planned equipment additions and the operations that cannot be interrupted.
A free Sunburst assessment can begin the scope discussion around your site and utility context. Ask for the measured-demand case and the billing-demand case to be shown separately, with unresolved inputs clearly marked.
The useful outcome is a feasible operating plan and a financial comparison supported by the account’s rules. A claimed peak reduction should be reviewable from the data, not accepted on the strength of a feature name.
Peak shaving FAQs
Is peak shaving only for commercial buildings?
The physical concept can apply wherever a specified demand maximum is reduced. Its bill value depends on the individual account. Do not assume a homeowner has a demand charge or that a commercial account uses a particular rule. Verify the relevant measurement and billing structure first.
Is a battery required for peak shaving?
No. Supported load control, scheduling or generation can lower a relevant maximum. Storage is one option. Compare feasible actions against the same operating requirements, measurement boundary and baseline before deciding which option merits a purchase.
Does a large battery capacity guarantee a lower peak?
No. Energy capacity does not establish sustained output, event timing, starting energy or control behavior. Another interval may determine the maximum. The proposal must connect the available power and energy to every relevant candidate event.
Does a high peak-power rating prove demand-charge savings?
No. A timed equipment rating does not establish an interval-long contribution or a billed-demand reduction. Check the supported duration and operating conditions, then apply the account’s measurement and billing rules to the resulting meter demand.
Can charging outside expensive energy hours create a demand problem?
It can if those hours still matter to a relevant demand measurement. Adding charging to existing demand may create a replacement maximum. Examine the full charge/discharge sequence and each applicable demand component separately from time-of-use energy prices.
Why can one missed interval reduce the expected benefit?
If that interval becomes the relevant maximum, lowering other intervals may not lower the determining quantity as much as expected. The effect depends on the tariff’s rules. A model should evaluate competing events rather than showing only its strongest example.
Are peak shaving and outage backup the same function?
No. One targets a demand maximum; the other supplies supported loads during an outage. They can compete for stored energy and controls. State the priority, reserve and equipment requirements for each instead of assigning the same energy independently to both.
What evidence should I request first?
Request the actual tariff and a credible interval-load record. Then ask which measured and billed quantities the proposal intends to change. Without those inputs, a precise demand-charge savings claim is not adequately supported.
Sources and methodology
Official sources were reviewed on October 6, 2026. DOE’s rate-evaluation guidance supports the distinction between bill components, interval-data review and look-back concepts. Its fleet guidance supports the need to align controls with schedules, readiness, data and communication.
The August 5, 2025 FEMP storage presentation supports peak shaving as a demand-reduction application. Vendor project examples, foreign installation details, maintenance intervals and performance claims in that presentation are not adopted here.
Every calculation is an original fictional example with stated boundaries. No local tariff price, rule, product availability, first-party outcome or savings guarantee is asserted. Account-specific financial conclusions require current official documents and a feasible supported design.