How to Size a C&I Battery Storage System Using 15-Minute Load Data

Sizing a commercial and industrial battery energy storage system should not begin with a preferred cabinet size. It should begin with the facility’s load profile, tariff structure and operating objective.

For many C&I projects, 15-minute load data is useful because demand charges are commonly calculated from average load over a defined interval. The data can reveal when a facility reaches its billing peak, how long the peak lasts and whether the battery would need high power, more energy, or a combination of both.

But 15-minute data does not automatically determine the battery size. It is the input to a sizing process.

The final design must still account for:

• The tariff’s demand-charge interval and peak windows
• The facility’s target demand level
• The battery’s maximum discharge power
• Usable energy within the operating SOC window
• Round-trip and charging/discharging efficiency
• PCS or inverter limits
• Solar generation, if the site is coupled with PV
• Backup or resilience requirements
• Forecast uncertainty and repeated peaks
• Warranty, degradation, auxiliary load and service requirements

First, define what the battery is supposed to do

A battery cannot be sized correctly until the project objective is explicit. “C&I storage" can describe several different operating goals:

Peak shaving

The battery discharges when the facility would otherwise exceed a target demand level. The primary design question is how much power must be delivered, and for how long, during the relevant billing interval.

Load shifting

The battery charges during one period and discharges during another. The design depends on the energy moved, the time gap between charging and discharging, tariff periods and the number of cycles permitted.

Solar self-consumption

The battery stores surplus solar production and discharges when the facility load is higher than the available PV output. The relevant load profile is combined with the PV production profile, not analyzed alone.

Backup power

The battery supports selected loads during an outage. This introduces a different sizing problem: critical-load power, required duration, transfer behavior, backup circuits and reserve SOC may become more important than monthly demand-charge reduction.

Multiple objectives

A project may combine peak shaving, solar self-consumption and backup. That can improve asset utilization, but it also creates control conflicts. A battery reserved for backup cannot always be fully dispatched for tariff savings.

Write the objective and priority order before selecting a product. Otherwise, a system may be optimized for one value stream while being expected to perform another.

What 15-minute load data actually tells you

A 15-minute interval series can be used to calculate the average facility load in each billing window. That makes it possible to reconstruct the demand value used by the tariff, locate recurring peaks and compare a baseline profile with a simulated battery-controlled profile.

The data is especially useful for four tasks:

1. Reconstructing billing demand
2. Locating the facility’s highest demand intervals
3. Testing a target demand threshold
4. Verifying results after the battery is installed

The time interval and the peak period are not the same thing. A 15-minute averaging interval describes how demand is measured. A peak period may cover a longer tariff window. A project can therefore have a 15-minute billing interval inside a several-hour peak-price period.

This distinction matters. A system that can respond quickly may reduce a short demand spike, but it still needs enough usable energy if the facility remains above the target for multiple intervals.

Step 1: Collect the right data set

Use the most granular reliable data available from the meter, utility, energy-management system or site monitoring platform.

Minimum data fields

The basic load file should identify:

• Timestamp and time zone
• Active power or energy consumption
• Measurement interval
• Missing or duplicated records
• Meter direction and sign convention
• Site operating status, where available
• Solar production, if PV is installed
• Generator or other on-site supply, if relevant

If the tariff is based on 15-minute demand, the load data should be at least as fine as the billing window when possible. NREL has used 5-minute load data to calculate costs where demand charges are determined over a 15-minute period. Finer data can help expose short peaks that might be hidden by coarser averaging.

For seasonal C&I facilities, analyze a sufficiently long period to capture operating changes. A full year is often useful for identifying seasonal peaks, shutdowns, production cycles and weather-related demand changes. Do not treat one representative week as a complete annual sizing basis unless the project scope explicitly limits the analysis.

Clean the data before sizing

Before calculating a battery size:

• Remove duplicate timestamps
• Identify gaps and abnormal zero values
• Check whether daylight-saving changes shifted the time series
• Confirm that the meter’s units are kW, kWh or another quantity
• Confirm whether the values are interval averages, interval totals or instantaneous readings
• Match the meter time zone to the tariff time zone
• Record any imputation or data-cleaning assumption

A clean-looking chart can still be unsuitable for sizing if the interval meaning is wrong.

Step 2: Read the tariff before reading the peak

The largest load point in the data is not automatically the point that matters financially.

Extract the tariff rules that define the target:

• Demand-charge rate in the applicable unit
• Billing demand averaging interval
• On-peak and off-peak windows
• Seasonal periods
• Coincident or non-coincident demand rules
• Monthly or annual ratchets
• Minimum billing demand
• Demand tiers
• Export or reverse-flow rules
• Whether the battery can charge from the grid

EIA defines demand charges as a bill component based on maximum electric capacity usage under the applicable rate schedule. EIA does not publish utility-specific electricity rates or demand-charge schedules, so the project team should obtain the actual tariff or a validated rate-database record.

The correct workflow is:

1. Identify the tariff’s demand measurement window.
2. Aggregate the load into the same interval if necessary.
3. Apply the tariff’s time-of-use and seasonal rules.
4. Find the relevant monthly or annual demand peaks.
5. Simulate the battery against that defined target.

Step 3: Convert the tariff target into a power requirement

Let:

• L(t) be the facility load at interval t
• D_target be the selected demand target
• P_PCS be the battery’s maximum grid-side discharge power

A simple screening expression for the battery discharge power is:

P_BESS(t) = min(P_PCS, max(0, L(t) − D_target))

This is not a universal design formula. It is a way to visualize the power envelope that the battery would need to cover at each interval.

The expression shows two important constraints:

• If the load is below the target, the battery does not need to discharge for peak shaving.
• If the load exceeds the target by more than the PCS can deliver, the remaining demand still reaches the grid.

The PCS or inverter rating therefore places a physical ceiling on instantaneous peak-shaving power. A battery with ample kWh but insufficient kW may not reduce the billing peak enough. Conversely, a high-power battery with too little usable energy may begin shaving the peak but stop before the billing window ends.

Step 4: Convert the power envelope into an energy requirement

For each interval, estimate the energy required to sustain the planned discharge:

E_interval = P_BESS(t) * Δt

For a 15-minute interval, Δt is one quarter of an hour. Across a peak event, the candidate battery must cover the cumulative discharge requirement—not just the highest single interval.

A practical screening process is:

1. Select a candidate D_target.
2. Calculate the battery power needed in every interval above the target.
3. Apply the PCS power limit.
4. Sum the interval energy across each peak event.
5. Record the largest cumulative energy requirement.
6. Repeat the test across the relevant days, months and seasons.

The candidate nominal energy must then be larger than the required delivered energy when SOC limits, conversion losses, auxiliary consumption, degradation assumptions and reserve requirements are included.

Do not use a generic kWh-to-kW ratio as a substitute for this simulation. One LBNL/NREL study included a 3* battery-capacity-to-power relationship as a modeling assumption, but that assumption is not a universal design rule.

Step 5: Separate rated energy from usable energy

The nameplate capacity is not necessarily the energy available for every operating event.

EPRI’s DER-VET documentation expresses usable energy as:

usable energy = rated maximum energy * (upper SOC limit − lower SOC limit)

This means a system’s usable energy depends on the operating SOC window. If the control strategy preserves a reserve or avoids the top and bottom of the battery window, the delivered energy is lower than the nameplate value.

A sizing sheet should therefore show at least:

• Rated energy
• Upper SOC limit
• Lower SOC limit
• Usable energy
• Backup reserve, if any
• Auxiliary load
• Charging energy required
• Discharging energy delivered

Do not label the full nameplate capacity as “usable" unless the operating specification supports that interpretation.

Step 6: Include efficiency and the charging path

Round-trip efficiency describes the relationship between energy discharged and energy charged over a cycle. For sizing, the key question is not only how much energy the load needs, but how much energy the battery must receive to restore its state of charge.

A simple planning relationship is:

energy charged ≠ energy delivered

The difference can come from battery conversion losses, PCS losses, auxiliary consumption and other system components. DOE FEMP guidance also warns that charging and discharging should be measured separately, because both can occur within a time step.

For a peak-shaving project, ask:

• Is the battery charged from the grid, PV or both?
• Can it reach the required SOC before the next peak window?
• Does the tariff penalize charging during a demand window?
• How much energy is needed to restore the target SOC?
• Are auxiliary loads included in the meter boundary?
• Does the control system prioritize peak shaving over solar capture or backup reserve?

A system that can discharge enough energy once may still fail the operating objective if it cannot recharge in time for the next event.

Step 7: Test repeated peaks and forecast error

Sizing against one isolated peak can produce an undersized system.

Industrial facilities may experience several high-load intervals in one day, repeated production shifts, start-up surges or weather-driven peaks. A control strategy that spends the entire SOC reserve on the first event may have nothing left for the second.

The simulation should therefore test:

• Multiple peaks in the same day
• Consecutive high-load days
• Seasonal operating profiles
• Production start-up and shutdown events
• Solar variability, if PV is included
• Forecast error and conservative dispatch
• Monthly demand ratchets
• Battery reserve requirements

The LBNL/NREL demand-charge research used an optimized control approach and highlighted the importance of forecast assumptions. Perfect foresight produces an upper-bound result; an actual project must account for imperfect forecasts and control response.

Step 8: Check the PCS, interconnection and site limits

A battery’s energy capacity is only one part of the system architecture.

The PCS must be matched to:

• Required AC discharge power
• Charging power
• Grid voltage and frequency
• Interconnection limit
• Power factor and reactive-power requirements
• Protection and control scheme
• Operating mode, including grid-following or grid-forming where applicable
• EMS commands and meter feedback

The site also introduces constraints:

• Available electrical capacity
• Transformer rating
• Cable and switchgear limits
• Space and access
• Ambient temperature
• Cooling and ventilation
• Fire-protection design
• Noise limits
• Maintenance access

For an industrial buyer, the right question is not “Which cabinet has the largest kWh?" It is “Which battery, PCS, EMS and site architecture can deliver the required power and usable energy within the interconnection and operating limits?"

Step 9: Compare candidate systems instead of selecting one number

A useful sizing study should test a matrix of candidate configurations rather than produce one apparently precise answer.

For each candidate, record:

The preferred candidate is not automatically the largest system. It is the system that meets the defined operating objective under realistic constraints and can be documented, commissioned and serviced.

EXLIPORC perspective

For buyers moving from a load profile to a quotation, the 261kWh liquid-cooled C&I battery storage system provides a relevant reference point for comparing product-level power range, capacity range, thermal management and application scope. The product page lists a 50kW-to-1MW power-output range and C&I uses including peak shaving, load shifting, backup power and demand-charge management; confirm the exact model and commercial specification before using those values in a project design.

Thermal management should be evaluated after the load and power requirements are defined. EXLIPORC’s liquid cooling vs air cooling comparison can support a separate discussion of temperature control, footprint and high-ambient considerations, but it should not replace a site-specific thermal, electrical or financial assessment.

A real project reference is also useful for checking whether the sizing logic is connected to a deployable operating scenario. The 200kW/412kWh industrial-park peak-shaving case documents a factory/industrial-park application with solar and off-peak charging, peak-period discharge and scalable system planning. Its reported result belongs to that specific case and should not be transferred to another facility without matching the load profile and tariff.

A procurement-ready sizing checklist

Before requesting final pricing, ask the supplier or EPC to provide:

• The cleaned load-data file and interval definition
• The tariff version and demand-charge assumptions
• Baseline monthly and seasonal demand peaks
• Selected target demand levels
• Candidate PCS power and AC operating limits
• Rated energy and usable energy
• SOC upper and lower limits
• Efficiency boundary and auxiliary-load treatment
• Charging source and recharge schedule
• Multiple-peak simulation results
• Solar-production assumptions, if applicable
• Interconnection and transformer checks
• Warranty, degradation and service assumptions
• Commissioning and performance-test method
• The exact model, configuration and delivery scope

If a quote provides only a kWh number without the load profile, target demand, PCS rating, usable-energy definition and control assumptions, it is not yet a complete sizing proposal.

Final takeaway

15-minute load data is a practical starting point for C&I BESS sizing because it can reproduce demand-charge windows and reveal the facility’s real peak behavior. But it does not eliminate engineering judgment.

A reliable sizing process separates:

• kW: how much peak power the battery can remove
• kWh: how long it can sustain the response
• SOC: how much of the nameplate energy is usable
• Efficiency: how much charging energy is required
• PCS: how the battery can exchange power with the site and grid
• Control: when the system charges, discharges and preserves reserve
• Tariff: which load intervals actually affect the bill

The goal is not to choose the biggest battery. The goal is to document a configuration that can meet the target demand under the facility’s actual load, tariff, electrical and operating constraints.