Peak Shaving Battery Storage Sizing Guide

Convert the load curve into a duty cycle

Choose a candidate grid setpoint and calculate the load above that setpoint for every interval in the measured load profile. The highest difference indicates the required discharge power. The accumulated energy above the line indicates the discharge energy needed during each event. Repeat the calculation across the year rather than sizing from one monthly bill peak.

For a first estimate, required battery energy equals the sum of shaved kW multiplied by interval duration, then adjusted for conversion losses, usable SOC and margin. Final modeling should use the actual dispatch time step and equipment constraints.

Peak shaving calculation: a worked screening example

Use P_shave(t) = max(P_load(t) - P_grid,target, 0) for each interval. Event energy at the AC point of connection is E_AC = sum(P_shave x interval duration). A screening estimate of nominal battery energy is E_nominal = E_AC / (eta_AC x DeltaSOC_shave), where the SOC window excludes any protected reserve.

For a hypothetical 15-minute event with site loads of 420 kW, 460 kW and 440 kW against a 350 kW grid target, the required shave is 70 kW, 110 kW and 90 kW. The screening PCS requirement is therefore at least 110 kW before project derating, overload duration and reactive-power obligations are checked.

The event requires 67.5 kWh at the AC point: (70 + 110 + 90) x 0.25 h. If the screening assumptions are 92% AC-path efficiency and a 60% SOC window available to peak shaving, nominal energy is approximately 67.5 / (0.92 x 0.60) = 122.3 kWh, before temperature, degradation, availability and design margin are applied.

This is a worked screening example, not a product recommendation or savings forecast. Replace every assumed value with the site’s complete interval data, supplier efficiency curves, approved SOC policy and operating constraints, then test the worst repeated event and recharge window.

Size power and energy separately

A short motor or process peak can require high PCS power but little energy. A long afternoon plateau can require modest power and much more energy. Treating both projects as the same kWh system produces weak results.

Check PCS continuous rating, overload duration and reactive-power obligations. If the PCS must also support voltage, power factor or backup transfer, those functions may reduce the power available for peak shaving.

Cabinet or container for peak shaving?

The screening result also points to the equipment format. When the required shave is in the range of roughly one hundred to a few hundred kW, documented outdoor cabinets such as the 105 kW / 233 kWh, 125 kW / 261.2 kWh and 250 kW / 522.5 kWh platforms are the usual starting point. Combining several cabinets is only appropriate when the supplier confirms the parallel architecture, controls and protection. When the peak shaving duty moves into the megawatt range, battery containers such as the documented 1.25 MW / 3.354 MWh, 1.725 MW / 5.016 MWh and 2.5 MW / 5.016 MWh systems become the practical format. Compare the trade-offs in the cabinet vs container BESS guide and the 20ft and 40ft BESS container checklist.

Test charging and reserve constraints

A battery that shaves the morning peak may not have time to recharge before the afternoon peak. Define the peak shaving SOC policy separately from any protected backup reserve, then model charger or PCS limits, import headroom, PV surplus and the site operating schedule. Reserve energy must remain protected unless the approved operating policy explicitly allows its use.

For a grid-setpoint control mode, the EMS should measure demand at the agreed point of connection, discharge above the target plus deadband, charge only when headroom exists, and enforce SOC, power and ramp limits. Forecast error, meter latency and command response should be included in performance testing so the system does not chatter between charge and discharge.

Measure the result after commissioning

Acceptance should confirm meter location, setpoint logic, sampling interval, response time, SOC limits and data export. After commissioning, compare baseline peaks, controlled grid demand, battery power and SOC. This evidence is required to tune the strategy and verify economic performance.