Sizing a containerized energy storage system for peak shaving requires more than matching battery capacity to a facility’s highest recorded demand. Technical evaluators must examine interval load data, demand charge structures, peak duration, and operating constraints to determine the required power rating and usable energy capacity. This guide explains how to translate those inputs into a practical system specification, accounting for battery efficiency, degradation, and dispatch strategy while assessing the trade-offs between peak reduction and project economics.

Start with the bill, not the battery

A demand peak only matters financially if it affects a charge on the electricity bill. Before selecting a discharge rating, identify the meter and tariff components the project is intended to address. A facility may face a monthly maximum demand charge, a separate on-peak demand charge, or a charge influenced by a previous high-demand period. Some tariffs also bill demand in kilovolt-amperes rather than kilowatts, making reactive power relevant to the evaluation.

Obtain the tariff, recent bills and at least a year of interval data where available. Use the utility’s billing interval: a controller reacting to instantaneous power will not necessarily minimize a demand charge based on a 15-minute or 30-minute average. Check whether intervals are fixed or rolling, when billing windows begin, and whether weekends, holidays or seasonal periods are treated differently.

The historical maximum is a useful warning, but it is rarely a complete design basis. An isolated spike caused by an unusual restart may justify an operational fix rather than a larger battery. Conversely, a moderately high load that persists every afternoon can require substantial energy capacity, even if it never produces the year’s single highest reading.

Turn the load profile into power and energy requirements

Choose a candidate demand ceiling and calculate how much site demand exceeds it in each billing interval. The highest excess indicates the approximate discharge power needed at the meter. The accumulated excess during a peak event indicates the energy that must be delivered before charging or the end of the event.

For each interval, the basic calculation is: discharge power = site demand minus target ceiling, when that difference is positive. Multiply each interval’s discharge power by its duration in hours and add the results across the event. This is a starting model, not a dispatch guarantee: it assumes the battery can follow the profile, has sufficient charge and is not constrained by other services.

Consider a hypothetical facility with a 2,000 kW demand ceiling and a three-hour peak. Its average demand is 2,200 kW in the first hour, 2,400 kW in the second and 2,100 kW in the third. The ideal battery contribution is therefore 200 kW, 400 kW and 100 kW across those hours. That implies at least 400 kW of discharge power and 700 kWh delivered at the meter for this event. Actual selection would require more analysis and allowances.

Run this calculation across the full data set, not just a representative day. Closely spaced peaks may need to be treated as one event if there is insufficient time to recharge. Also test several ceilings. Lowering the target can produce diminishing returns: shaving another small portion of demand may require a disproportionate increase in inverter power, battery energy or charging time.

Convert the result into a defensible equipment rating

The 700 kWh in the example is energy delivered at the billing meter, not necessarily the capacity stated on a battery nameplate. Losses occur between the cells and the meter, and cooling, controls and other auxiliaries consume energy. The impact depends on where each load is connected and how the supplier defines its ratings.

Work backward from the meter using a consistent boundary. Account for discharge-path losses and auxiliary consumption during the event, then determine how much installed battery capacity is needed within the permitted state-of-charge range. Do not apply round-trip efficiency in full to discharge energy and then apply charging losses again: round-trip efficiency combines both directions. Charging efficiency belongs in the recharge and operating-cost calculation.

Degradation is another reason to distinguish installed capacity from usable capacity. If the project must meet the peak-shaving target near the end of its contract term, request an end-of-term performance calculation under the proposed duty cycle and operating conditions. A capacity guarantee alone may not establish available power at low state of charge, high temperature or after a sustained discharge.

For the power conversion system, specify continuous AC power at the relevant operating conditions and connection point. Check transformer losses, auxiliary loads, voltage limits and any thermal derating. If reactive power support is required at the same time as peak shaving, the inverter’s apparent-power rating may limit the active power it can deliver. A system rated for 400 kW at the inverter terminals is not automatically a system capable of reducing metered demand by 400 kW.

Make sure the battery can be ready for the next peak

Peak shaving is a scheduling problem as much as a capacity problem. A battery that empties during a morning production ramp may be unavailable for a more expensive afternoon peak. Likewise, aggressive recharging can create a new billing peak and erase the intended saving.

Model charging against the same interval demand ceiling used for discharge. Available charging power is limited by site headroom, the inverter, battery charge limits and any contractual import limit. Include charging losses and auxiliary consumption, and check whether the available low-load window is long enough to restore the required state of charge. Where demand remains high overnight, a larger battery does not necessarily solve the problem; a revised operating schedule or a higher target ceiling may be more realistic.

Control strategy matters when peaks are uncertain. A fixed schedule is simple but can discharge too early. A controller using load forecasts and a state-of-charge reserve may handle variable production better, provided the forecasts and meter signals are reliable. Test the strategy with imperfect forecasts and missing data, not only with perfect historical hindsight. If the battery is also expected to provide backup power or another grid service, reserve the required capacity and power in the model rather than assuming the same margin can serve both purposes simultaneously.

Check the container against the site, not the other way around

Containerized packaging simplifies some integration work, but it does not remove site engineering. Confirm the proposed location, access for installation and maintenance, foundation requirements, drainage, cable routes and proximity to the point of interconnection. A longer cable run or an undersized transformer can change losses, installation cost and the practical discharge available at the meter.

Ask suppliers for performance at the project’s actual ambient conditions and elevation. Cooling design affects both battery life and auxiliary load; its power consumption should not disappear from the financial model. Review noise, ventilation or thermal-management arrangements, emergency isolation, fire protection and emergency response with the relevant local authorities and insurers. Applicable codes, testing and certification requirements depend on the equipment configuration and jurisdiction and should be confirmed before the layout is finalized.

Electrical integration deserves an equally early review. Confirm protection coordination, fault-current implications, metering accuracy, communications interfaces and utility interconnection requirements. If the facility has on-site solar or generators, model how their output and operating restrictions affect the net load seen by the billing meter. Backup operation also requires a separate review of islanding, transfer equipment and critical-load arrangements; a grid-connected peak-shaving system should not be assumed to provide it.

Compare options on delivered value, not container size

A useful evaluation compares several power-and-energy combinations under the same tariff and dispatch assumptions. For each option, calculate the revised bill interval by interval, including all relevant demand windows and any ratchet provisions. Then include the cost of charging energy, conversion losses, auxiliary consumption, maintenance, capacity augmentation or replacement where applicable, and the treatment of degradation over the financial horizon.

Pay particular attention to the months in which the system misses its target. If a single uncontrolled interval establishes the monthly demand charge, excellent performance on every other day may have limited value. Sensitivity tests should cover load growth, production changes, equipment downtime, tariff changes and a battery that begins a peak with less charge than planned. For facilities with flexible processes, compare storage with alternatives such as staggered motor starts or rescheduled production; the strongest solution may combine operational changes with a smaller battery.

For cross-market comparisons, GPEGM’s focus on power electronics, distribution equipment and energy-market intelligence is a useful reminder to assess the complete electrical and commercial context. Inverter design, switchgear integration and local tariff rules can affect the decision as much as the advertised battery capacity.

Before requesting final quotations, issue a common specification: interval load data, tariff assumptions, target demand ceiling, required power at the meter, energy-delivery duty, operating environment and the date at which performance must still be met. Ask each bidder to state measurement boundaries, losses, auxiliary loads, degradation assumptions and exclusions. The most credible proposal is the one that shows how it will meet the target on the difficult days—not simply the one with the largest nameplate capacity.