Solar Peak Shaving: Cut Demand Charges With BESS

Last updated: November 2026

A large solar array can reduce annual electricity purchases while leaving an expensive evening demand peak largely unchanged. That is a timing problem, not proof that solar has no demand value. The useful question is how PV and a battery change every interval that contributes to the customer’s actual bill.

Solar peak shaving combines a load profile, a tariff and a dispatch strategy. Start by reproducing the existing bill, then calculate the residual load after PV and the battery power and energy needed to reduce it. A battery energy storage system, or BESS, should be selected from that analysis rather than from a standard cabinet size.

This guide retains a five-step sizing method and a manufacturing-facility example. The facility, costs and modeled results are illustrative design assumptions, not a real customer project or a completed annual simulation. They show how to check the calculations without promising a universal savings percentage or payback period.

SurgePV publishes this post and sells solar design software. Its verified capabilities include PV layout, shading, yield and financial modeling, electrical workflow support, BOM output and proposals. Native interval-data imports, tariff ratchet simulation and automatic battery-size sweeps require separate verification.

How Commercial Demand Charges Price Your Peak

Demand charges price a tariff-defined measure of power, while energy charges price accumulated consumption. Kilowatts describe power; kilowatt-hours describe energy. The billing determinant may use a maximum interval average, specified time windows, a contract level, a historical floor or another rule. Read the actual schedule before choosing a dispatch target.

For a simple illustrative tariff based on the highest fifteen-minute average, a facility at 500 kW for one complete billing interval can establish a 500 kW monthly determinant. A brief instantaneous reading of 500 kW is different. Its effect on the interval average depends on how long it lasts and on the other demand within that interval.

A sourced example of tariff complexity

PG&E Schedule B-19 defines maximum demand using fifteen-minute averages for the specified customer class, with provisions for shorter measurement in certain fluctuating-load situations. Its published March 1, 2026 rate sheet separates summer peak, part-peak and overall maximum-demand components. Customer eligibility, voltage class and rate option matter.

The table below reproduces selected secondary-voltage total bundled demand rates from sheet 4, effective March 1, 2026. They are separate components, not alternative all-in prices. Do not use them for a different account, rate option or billing period without checking eligibility and the relevant tariff version.

A battery can affect the components differently because their maxima may occur in different intervals. Recalculate each determinant and then apply the corresponding rate. Multiplying a single assumed shaved kW value by every listed rate can overstate savings if the same reduction is not achieved in each relevant window.

Create a tariff register

Record the account’s schedule, service voltage, applicable options and effective dates. Add billing-cycle boundaries, demand measurement, time-of-use periods, seasonal definitions, holidays, export compensation, minimum charges and any standby provisions. Keep the source tariff and a representative bill with the model.

If the account has separate supply and delivery charges, reconcile both. A published bundled rate may not match a customer buying supply through another arrangement. Do not transfer a tariff from a nearby facility simply because the sites share a utility territory or have similar annual energy consumption.

Ratchets can carry a peak into later bills

A demand ratchet uses a historical demand measure to set a billing floor according to the tariff. For an illustrative 80% ratchet, a qualifying historical peak of 800 kW would create a 640 kW floor during the specified lookback period. If the month’s measured demand is 300 kW, the floor can still govern that component.

The percentage, lookback, season and affected component must come from the actual tariff. An 80% ratchet is not the same as a bill that is 80% above current demand. In this example, 640 kW is more than twice 300 kW. Keep the floor calculation explicit instead of describing it with an ambiguous percentage increase.

The model also needs the historical peak at the beginning of the study. A new battery does not necessarily erase an existing ratchet immediately. Show how the floor changes as historical intervals leave the lookback window, and distinguish first-year savings from a later steady-state year.

When Solar Alone Reduces Demand Charges

PV reduces demand when generation overlaps a relevant load interval, but the bill depends on the highest remaining interval under each tariff rule. Lowering the original peak can reveal another peak later in the day. Recalculate the whole series instead of subtracting PV output only at the original maximum.

Berkeley Lab’s July 2017 publication, Exploring Demand Charge Savings from Commercial Solar, describes how savings vary with tariff design, customer loads and PV profiles. It does not justify assigning one guaranteed reduction percentage to all schools, factories or warehouses.

Load shapes to investigate

Daytime offices, schools and some cooling-driven loads may have useful overlap with PV. Evening processes, overnight charging and continuous operations may have less overlap with the intervals that determine their bill. These are hypotheses for the analysis, not fixed savings bands by building category.

Review the actual causes of demand peaks. A school can have evening events, a warehouse can operate overnight refrigeration cycles, and a factory can change shifts. Building labels are helpful for forming questions, but the interval data and operating schedule should determine the model inputs.

Why combined solar and storage can help

The 2017 report Solar + Storage Synergies for Managing Commercial-Customer Demand Charges, NREL/TP-6A20-70360 reports complementary demand-management effects and emphasizes that reductions are customer-specific and tariff-dependent. The linked publication record is hosted by the National Laboratory of the Rockies; the report retains its original publication number.

PV can lower residual daytime load or provide charging energy, allowing the battery to serve other constrained periods. Storage can address demand when PV is low, within its available power and energy. The combined value must be simulated; it is not automatically the sum of independently optimized solar and battery savings.

Compare alternatives against the same baseline

Model the existing site, PV alone, storage alone and the combined system using the same load and tariff assumptions. If the customer’s operations change, add that as a separate scenario. Otherwise, a shift in production or occupancy may be incorrectly attributed to the installed energy system.

Also consider operational measures such as staggering equipment starts or shifting a flexible process. These may reduce the battery requirement, though they can affect productivity or customer service. Record the operational constraints and obtain the site’s agreement before treating a load change as available dispatch flexibility.

The Five-Step Peak-Shaving Sizing Method

Use a load-first process that separates power, energy and financial value. The following steps produce an auditable candidate design; equipment selection and electrical engineering follow from the requirements they identify.

Step 1: Obtain interval data and billing history

Request data covering the relevant seasons and any tariff lookback period. Twelve months is a useful starting scope for seasonal review; additional history may reveal unusual peaks, operational changes or the historical ratchet state. Do not assume every utility portal provides the same resolution or retention period.

Match the data interval to the demand measurement. Hourly averages cannot reconstruct a fifteen-minute maximum reliably. If sufficiently detailed data are unavailable, obtain metering or label the analysis as preliminary and use justified sensitivity cases. Do not present an interpolated curve as measured short-interval demand.

Check units before importing data. Interval kWh must be divided by interval hours to obtain average kW. For example, 100 kWh consumed in fifteen minutes corresponds to 400 kW average demand. A file that already contains kW must not be converted again.

Normalize time zones, daylight-saving transitions, interval-start versus interval-end timestamps and meter identifiers. Flag missing or duplicated records. Reconcile monthly energy and demand against bills, documenting differences caused by meter conventions, adjustments or unavailable data. This validation should precede PV or storage optimization.

Step 2: Build the demand-event distribution

Use a load-duration curve to see how frequently different demand levels occur, but retain the chronological series. Sorting demand values removes the sequence needed to calculate battery depletion and recharge. A duration curve is a screening tool, not a complete dispatch model.

Identify the intervals that drive each monthly billing component and group adjacent high-load intervals into events. Record their timing, duration, shape and likely cause. Review both isolated spikes and broad plateaus, then check whether the operating team expects those patterns to continue.

Percentiles can summarize the data, but a 95th- or 99th-percentile threshold is not automatically the correct design target. The economically preferred target depends on the tariff, battery cost and tolerated misses. Model the residual cost of events the selected design cannot fully cover.

Step 3: Calculate the solar-only residual load

Develop a PV production series for the proposed location, orientation, shading and equipment. Align it with the load timestamps and electrical boundary. If the production model is hourly but the tariff is finer, explain the resolution limitation and examine sensitivity to intra-hour variation.

Subtract modeled PV from the load at each interval, applying export or curtailment rules as necessary. Then recalculate all demand determinants. Keep the PV assumptions and loss model available for review, and distinguish weather-driven uncertainty from changes in the customer’s load.

SurgePV supports the PV geometry, shading and yield portion of this study. Confirm any required time-series export and downstream analysis workflow before relying on it. Native fifteen-minute load imports, dispatch optimization and tariff ratchets are not assumed capabilities in this guide.

Step 4: Size battery output power

Choose a candidate import target and calculate the residual power above it at each interval. If residual demand is 850 kW and the target is 500 kW, the battery must provide 350 kW at that operating point. This is a same-boundary arithmetic example, not an equipment specification.

Check the selected system’s continuous AC output across the required temperature and state-of-charge range. Use manufacturer limits for the battery and power conversion system, including shared-inverter constraints where relevant. Do not apply an unsupported universal temperature derating percentage or inverter-oversizing rule.

Define any margin in relation to uncertainty or service requirements. A larger inverter is useful only if the battery and other system components can supply the added output. Extra converter capacity alone cannot guarantee deeper peak reduction or a specific percentage improvement in savings.

Step 5: Size usable energy and recharge

Integrate the required battery output over every constrained period. For a rectangular illustrative event requiring 350 kW for ninety minutes, delivered energy is 525 kWh. Real events are usually better represented by the sum of interval power multiplied by interval duration, rather than the maximum power multiplied by an arbitrary duration.

Convert delivered energy to the appropriate nameplate requirement using supported usable-state-of-charge, efficiency and aging assumptions. For illustration, 525 kWh delivered with a 90% usable window and 95% discharge efficiency needs 614.0 kWh before any end-of-life allowance: 525 / (0.90 × 0.95).

If the vendor already specifies usable AC energy at the required conditions, do not apply those factors twice. Check whether reserve, auxiliary consumption and degradation are included. Verify both energy and output power at the warranty milestone relevant to the design.

Finally, simulate recharge. A battery must recover sufficient energy before the next constrained period without creating a new billable peak. Include PV variability, other loads and permitted grid-charging windows. A model that resets the battery to full every morning without an energy source is not a feasible operating model.

Sizing the Solar Array for Demand-Charge Reduction

Optimize PV for the full project objective, including energy value, demand coincidence, available area and export constraints. Annual energy matching alone cannot show its demand-charge contribution. Conversely, maximizing demand savings alone can overlook valuable energy savings or other owner priorities.

Use the residual-load profile

For each candidate array, examine the load remaining during the relevant demand windows. Check whether a larger array reduces the governing peak or mainly adds midday surplus. The battery’s required power and energy can change differently as PV grows, so update both rather than scaling the storage size in proportion to array capacity.

A hypothetical 1 MW facility peak, 600 kW battery and 400 kW PV contribution could theoretically reach zero grid import at that moment if all outputs are available on the same boundary. If the target is a nonzero import, less PV or battery output may suffice. The actual target and availability must be stated before inferring a required PV size.

Use site-specific output at the relevant hours rather than claiming that every array produces a fixed fraction of nameplate in the afternoon. Include weather and seasonal variation. A design that needs PV at a particular level to respect a hard service limit also needs an accepted fallback when that level is unavailable.

Evaluate exports explicitly

Use the customer’s actual export arrangement and compensation schedule. Export value can vary by time, tariff and project eligibility. Avoid a universal cents-per-kWh assumption or a claim that every commercial project is subject to the same net-billing treatment.

For each PV size, report direct consumption, battery charging, export and curtailment. These flows have different value and loss pathways. If export is prohibited, show the curtailment control and its effect on production, rather than valuing all generated energy at the retail purchase rate.

Inverter loading ratio is a variable, not a preset

The inverter loading ratio is DC array capacity divided by AC inverter capacity. Increasing DC capacity relative to the inverter can change clipping and output across the day, but it does not automatically create the best afternoon profile or maximize demand savings. Orientation, shading and weather also matter.

Compare candidate ratios with the same site and tariff model. Record which hours gain production, which hours clip and whether those changes affect demand determinants. In a shared-inverter arrangement, check how simultaneous PV production and battery discharge compete for the available output capacity.

Battery Power vs Energy: What Each Controls

Power determines how much the battery can reduce demand at a moment. Energy determines how long it can sustain that output before its usable state of charge is depleted. Both ratings need a defined boundary and operating condition.

Power sets a lower bound on residual demand

Against a 1,000 kW residual load, a battery delivering 500 kW leaves 500 kW of grid import, while a 300 kW output leaves 700 kW, ignoring other changes. If the target is 400 kW, neither output reaches it. The shortfall does not automatically eliminate all savings; the tariff determines the financial effect of the remaining demand.

For example, a battery that achieves a 550 kW reduction against a planned 600 kW reduction can still deliver substantial demand savings under a simple linear demand rate. It misses the planned target by 50 kW. A minimum charge, ratchet or other rule can change the outcome, which is why the bill model must evaluate it rather than treating every miss as total failure.

Energy sustains the reduction

A system with 600 kWh of usable energy delivered at a constant 300 kW supports two hours on that boundary. If 600 kWh is instead a nameplate DC figure, usable AC duration may be lower. Reserve and operating limits can reduce the energy available for peak shaving further.

Extra energy may help with consecutive events or recharge constraints even when the single longest event is already covered. It can also support another service, subject to conflicting obligations. Assess chronological dispatch before concluding that additional kWh has no value.

C-rate does not replace the product specification

A same-basis ratio of 300 kW to 600 kWh is 0.5 per hour, commonly described as 0.5C in a simplified system comparison. Actual permitted output depends on the battery and conversion equipment. Do not assume a chemistry permits a particular continuous or burst rate without supplier evidence.

Compare power and energy at consistent temperature and aging conditions. A product may limit power at low state of charge even while some energy remains. That restriction matters near the end of a long event and should be represented in the dispatch model rather than hidden by a single nominal ratio.

Illustrative Design Example: A Manufacturing Facility

This is a modeled planning example, not a real Massachusetts project or a National Grid rate quotation. The assumed facility operates two shifts and consumes 5.2 GWh annually. The figures below preserve a useful sizing exercise while making the inputs and unresolved conditions explicit.

Load and PV assumptions

The candidate PV substantially lowers the assumed daytime average but only reduces the scenario maximum by 30 kW. This illustrates a peak with little coincident solar. It does not imply a specific annual demand-bill reduction, because other months, time windows and historical floors have not been supplied.

At an assumed DC-to-AC ratio of 1.30, 800 kW DC corresponds to about 615.4 kW AC. This is a calculation of nameplate ratio, not a production prediction. The assumed 985 MWh annual yield would need a site-specific model before it could support an investment decision.

Battery power requirement

Suppose the design target is 600 kW import during a residual-load event of 1,150 kW. Required battery output is 550 kW. A candidate 600 kW output rating provides nominal power margin, but the designer must verify that the specified output is available at the relevant temperature, state of charge and age.

Assume the event requires a constant 550 kW for 1.8 hours. Delivered energy is 990 kWh. In a real series, calculate the area above the target rather than assuming a rectangle unless the load actually supports that approximation.

Energy requirement and the candidate cabinet

With an illustrative 92.5% usable state-of-charge window and 95% discharge efficiency, beginning-of-life nameplate energy is 990 / (0.925 × 0.95) = 1,126.6 kWh. A 1,200 kWh candidate would provide about 1,054.5 kWh at the stated output boundary under those simplified assumptions.

If the design must cover the same event after capacity falls to an assumed 80% of initial capacity, the initial nameplate requirement rises to 1,408.3 kWh. The original 1,200 kWh candidate would provide only 843.6 kWh on the same assumptions. It would not sustain the full event without another source or a higher import target.

The retained-capacity fraction is an illustrative design assumption, not a warranty claim for a particular chemistry. The appropriate procurement response could be a larger battery, an augmentation plan or a revised target. Record the decision and the supplier evidence rather than declaring the initial candidate adequate for its whole life.

Illustrative annual bill inputs

The following values are assumed annual bill totals for arithmetic demonstration. They are not outputs of a supplied dispatch simulation. A project study must derive them from a complete chronological model and the actual tariff before using the savings in a proposal.

The $211,100 difference is gross bill savings on the stated assumptions. It excludes maintenance, financing and other project costs unless those have been explicitly included elsewhere. Storage recharge and conversion losses must be reflected in the post-system energy bill rather than subtracted inconsistently in multiple schedules.

Capital-cost arithmetic

These are budget assumptions, not current supplier prices. The PV line is about $1.115/W DC for the assumed 800 kW DC array; it should not be labeled $1.45/W AC without specifying the exact AC capacity and rounding basis. Check that additional soft costs are not already included in the equipment scopes.

Without any incentive, $1,535,000 divided by $211,100 gives 7.27 years of gross simple payback. This is before operating costs, financing, degradation, tax and discounting. It is not a four-year result. Any incentive-adjusted figure requires a separately supported eligibility and timing assessment.

If all costs stayed the same but annual gross savings were 20% lower, they would be $168,880 and gross simple payback would be about 9.09 years. If savings were 20% higher, $253,320 would imply about 6.06 years. These sensitivities illustrate arithmetic only; they are not probabilities or performance forecasts.

Control Strategies That Determine Actual Peak Shaving

The energy management system, or EMS, must operate within battery, inverter, grid and customer constraints. A feasible offline calculation is not enough: the controller needs the right measurements, permissions and failure behavior to reproduce the intended strategy.

Setpoint-based control

A setpoint controller discharges when measured or predicted net import exceeds a chosen threshold. It may recharge when sufficient headroom is available, subject to its other limits. This can be straightforward to explain and commission, but a fixed target can deplete the battery too early during a long event.

Evaluate the control using realistic state of charge and imperfect measurements. Define deadbands, response behavior and priorities with the equipment supplier. Do not claim near-perfect reliability or a universal capture rate without evidence for the complete installed system.

A missed target can change the remaining value of dispatch within the billing cycle. If a monthly maximum is already set, the controller may choose a different economic target for the rest of that component’s period. Other time-window charges or future ratchets may still justify discharge. The decision needs the tariff state, not only the current meter reading.

Predictive control

A predictive controller uses forecasts of load, PV and prices to plan charging and discharging. It can reserve energy for a later, more valuable event instead of spending it on every early threshold crossing. Its benefit depends on forecast quality, equipment availability and the alternative strategy used for comparison.

Separate perfect-foresight optimization from an implementable forecast-driven simulation. Perfect foresight can provide an upper-bound reference, but it should not be presented as a guaranteed customer result. Include forecast errors, missing data and unusual operations in the evaluation.

Request evidence for forecast inputs, update frequency, telemetry requirements and how the controller responds when a forecast is wrong. A machine-learning label alone does not prove lower bills. Compare measured or reproducible results on a common dataset and document any tuning that uses future information.

Multiple tariff components and reserve priorities

Different demand windows can justify different targets. The controller may need to manage an overall monthly maximum, a time-window maximum and a ratchet simultaneously. Preserve the correct calendar and season definitions, including any holidays or billing-period boundaries.

Assign explicit priorities among bill reduction, backup reserve, export limits and external grid-service commitments. A reserve percentage should come from a defined service requirement. Energy kept for an outage is unavailable for routine dispatch unless the owner permits its use.

Solar charging and grid charging

Evaluate charging sources through their cost, availability, efficiency and contractual permissions. Solar charging can use otherwise exported or curtailed energy, while grid charging can restore readiness when PV is insufficient. Neither is automatically the preferred strategy at every interval.

Round-trip losses apply to energy passing through storage regardless of whether it originated from PV or the grid. The exact conversion path depends on the architecture. Do not claim that solar charging avoids all losses or automatically creates eligibility for a particular tax-credit percentage.

Charging must also respect demand targets. A low energy price is not enough reason to charge at full power if that action sets a costly new demand maximum. Include charger auxiliary loads and the facility’s other consumption when calculating available import headroom.

Common Design Mistakes That Erode Savings

These checks address the logic of a peak-shaving study without claiming an undocumented project audit. Use them to review both the model and the supplier proposal.

1. Using hourly averages for a finer demand interval

A fifteen-minute interval at 1,200 kW followed by forty-five minutes at 600 kW averages to 750 kW over the hour. The hourly value hides the higher short-interval demand. It does not prove a universal battery undersizing percentage, but it shows why the data resolution must match the tariff.

Where higher-resolution data cannot be obtained, disclose the uncertainty and avoid a precise savings guarantee. Metering additional representative periods can be more valuable than refining a battery-size sweep around an unsuitable dataset.

2. Omitting historical tariff state

Carry ratchets and other historical determinants into the simulation correctly. Run the first billing year with the actual initial state, then assess later years separately if appropriate. A model that starts with no historical peak can overstate early savings.

Check that the controller and the billing model use the same state. If the billing engine knows a demand floor but the controller does not, the simulated dispatch may differ from what can actually be deployed. Record how tariff changes and meter resets are handled.

3. Assuming nominal inverter output is always available

Use the manufacturer’s operating envelope rather than an invented universal thermal correction. Check sustained output, overload limits, ambient conditions, state of charge and any shared PV/storage constraints. A generic percentage margin is not a substitute for this evidence.

When comparing equipment, specify the required AC output and duration at the design conditions. This prevents one supplier from quoting favorable room-temperature peak power while another quotes continuous output at the site condition. The procurement comparison should describe the same service.

4. Selecting a standard cabinet before defining the load

A catalog size is a candidate, not the design objective. Compare viable combinations of power and energy and evaluate lifecycle cost, achievable savings and unmet targets. The economically preferred result may differ from the system with the shortest simple payback or the lowest first cost.

Avoid claiming a typical capital saving from optimization without a dataset. Report the actual difference between the quoted candidates and explain its cause. A small equipment-cost difference may be outweighed by civil works, control integration or maintenance scope.

5. Assuming one PV ratio is optimal

Evaluate array size, orientation and inverter capacity against coincident load and export rules. A higher DC-to-AC ratio does not guarantee better late-afternoon value, and a lower ratio does not automatically maximize lifetime energy. Keep the output profile and financial assumptions connected.

If the proposed battery shares conversion capacity with PV, evaluate combined output constraints. A system can have adequate separate nameplate ratings yet be unable to deliver both at once. The single-line diagram and supplier operating limits should agree with the model.

6. Ignoring growth and adverse operating conditions

Add scenarios for planned equipment, changed shifts, unusual weather, outages and reduced battery capacity. Do not apply a blanket percentage above one observed annual maximum and assume all uncertainty is covered. Different changes affect event duration, timing and recharge opportunity differently.

Show the cost of misses and the chosen service level. Perfect suppression of every conceivable peak may be uneconomic, while an overly aggressive target can repeatedly deplete storage. The owner should understand the accepted tradeoff before equipment is ordered.

ROI Math: Evaluate the Whole Cash Flow

Peak-shaving economics depend on the actual tariff, load, equipment cost, dispatch and financing. There is no defensible universal payback band based only on state, building type or a headline dollars-per-kW rate. Use the site’s complete cash flow and compare alternatives on the same basis.

Separate gross savings from net benefit

Start with the difference between baseline and project electricity bills. Then include incremental operating costs, service subscriptions, insurance where applicable, augmentation or replacement and other project obligations. Keep financing and tax treatment explicit rather than hiding them in a net installed-cost figure.

Simple payback divides an upfront amount by a recurring annual benefit under simplifying assumptions. It does not capture discounting, changing savings or cash-flow timing. Use net present value and other appropriate measures where the investment decision requires them, and report the assumptions behind each result.

US incentives require project-specific review

IRS Notice 2025-42 on Sections 45Y and 48E addresses changed timing rules for applicable wind and solar facilities, including the July 4, 2026 construction-start boundary and December 31, 2027 placed-in-service cutoff for affected projects. Storage has separate treatment under the cited provisions. Do not use an older blanket solar-plus-storage incentive assumption.

Have the tax adviser assess project eligibility, qualified basis, construction and commissioning dates, labor requirements, sourcing restrictions and any claimed bonus. This guide does not assume that domestic content or an energy-community location automatically produces a particular combined credit for every project cost.

If an incentive is supported, model when it can be used or monetized and any fees, conditions or recapture exposure. A tax credit is not necessarily cash received when the equipment invoice is paid. Show a case without the benefit so the owner understands the project’s dependency on it.

Financing sensitivity

Use lender quotations or clearly labeled assumptions for interest, fees, tenor and repayment. Calculate debt service from the actual structure. A single claimed percentage increase in lifetime cost cannot describe every loan because amortization, fees, timing and refinancing differ.

Keep project economics separate from owner-equity cash flow. Borrowing reduces initial equity paid but creates repayment obligations; it does not reduce the equipment’s physical cost. Reconcile debt draws, principal, interest and ending balance, and run sensitivity to delayed savings or higher operating cost.

A practical sensitivity table

When Peak Shaving Is Not the Best Strategy

Peak shaving may be less attractive where the tariff has little avoidable demand cost, the load stays high for long periods or storage cannot recharge without creating another peak. That does not mean batteries have no value; another service may be more relevant to the owner’s objective.

A flat continuous load is demanding for a finite battery because lowering demand for a long period requires sustained energy and a recharge opportunity. Time-of-use shifting, operational changes or efficiency measures may deserve comparison. Evaluate the actual profile rather than assuming a building category rules the strategy in or out.

Where backup is the primary goal, define critical loads and reserve first. Combining backup and bill reduction is possible, but the available capacity must be allocated and the tradeoff shown. Do not count the same energy as fully available to both services at the same time.

For tariffs without an avoidable demand component, focus the financial analysis on the values that actually exist, such as energy timing or self-consumption. There is no universal minimum share of the bill that must be demand charges before a study is worthwhile. A screening estimate can identify whether detailed modeling is justified.

Turn the Analysis into a Reviewable Design

Deliver the validated load data, tariff register, PV assumptions, battery operating envelope and dispatch results together. Include first-year historical tariff effects, later-life battery limits and the fallback strategy. A reviewer should be able to reproduce the bill and trace each major savings claim.

During operation, compare actual demand determinants, state of charge and dispatch with the model. Investigate changes in load, tariff, PV availability or controller behavior before attributing every variance to battery sizing. Preserve settings and revision history so that corrective actions can be evaluated against the approved design.

For the surrounding engineering work, use the commercial battery sizing guide, commercial solar design guide and solar EV charging station design workflow. Explore SurgePV’s PV design tools for the array and yield inputs, then validate the tariff and dispatch analysis with the appropriate specialist workflow.

Frequently Asked Questions

How does solar reduce demand charges?

Solar reduces demand charges when its generation lowers the intervals used to calculate billed demand. Recalculate the complete residual-load series because another interval can become the new maximum. The result depends on the tariff, load timing and PV output, not annual solar production alone.

How do you size a battery for peak shaving?

Calculate required battery power from residual demand above a chosen import target, then integrate that output over the constrained periods to find delivered energy. Apply supported usable-capacity, efficiency and aging limits. Validate recharge, consecutive events and the actual tariff before selecting equipment.

What is a demand charge in commercial electricity?

A demand charge prices a tariff-defined measure of power rather than total energy consumed. It may use an interval maximum, a time window, a historical floor or other rules. Check the account’s schedule and measurement method instead of assuming every tariff uses the same interval.

Can solar reduce demand charges without batteries?

Yes, solar can reduce demand charges when output overlaps the intervals that govern the bill. The benefit may be limited when another peak occurs with little solar generation. Calculate the solar-only case from aligned load and PV data before deciding whether storage adds value.

What is the difference between peak shaving and load shifting?

Peak shaving targets demand levels that affect billed power, while load shifting moves energy use between times with different value or constraints. A battery can pursue both, but available power and energy must be allocated. The preferred priority depends on the tariff and customer requirements.

How long should a peak-shaving battery discharge?

Required duration comes from the chronological load above the selected target and the opportunities to recharge. There is no universal two-hour requirement. Check consecutive peaks, usable energy, reserve and end-of-life capability at the required output power.

How do demand ratchets affect battery savings?

A ratchet can set billed demand using a qualifying historical peak even after present demand falls. Include the tariff’s exact lookback, percentage and affected component, plus the initial historical state. A new battery may take time to reduce the ratchet floor as earlier peaks leave the calculation.