Peak shaving is a power problem first and an energy problem second. The power rating must cover the amount by which load exceeds the target; the usable energy must sustain that reduction for the full peak event, including conversion losses and operating reserve.
A monthly bill rarely shows enough detail. Sizing should start with interval demand data and the exact tariff rules used to calculate billed demand.
Direct answer: Use at least one full year of interval data. Identify recurring peaks, exceptional events and seasonal production changes before choosing the target demand limit.
Define the boundary
For Factories and C&I EPCs in tariff-driven markets, the first task is to turn the intended service into measurable inputs and acceptance limits. The following table keeps the decision tied to evidence instead of a broad product label.
| Input | Unit | Use in model |
|---|---|---|
| Interval facility load | kW by timestamp | Find peak magnitude and duration |
| Billing demand rule | kW and interval length | Reproduce utility bill |
| Target grid limit | kW | Set discharge command |
| Peak duration | hours | Calculate required usable kWh |
| System losses and reserve | percent or kWh | Convert usable requirement to installed capacity |
Acceptance and failure tests
The commercial offer should state its assumptions, exclusions and measurement boundary. Buyers can then compare systems on the same basis and keep later design changes under document control.
| Example step | Calculation | Result meaning |
|---|---|---|
| Power requirement | Measured peak minus target grid limit | Minimum PCS discharge power |
| Event energy | Sum of excess load across peak intervals | Usable discharge energy |
| Operational allowance | Event energy plus reserve and losses | Required available energy |
| Cycle count | Number of dispatch events per year | Warranty and degradation input |
| Savings check | Avoided demand charge minus energy and service costs | Project value before financing |
Action points for the project team
- Model the utility demand interval exactly; a 15-minute rule and a 30-minute rule can produce different dispatch needs.
- Keep backup reserve separate from the energy allocated to economic dispatch.
- Test the control strategy against days with multiple peaks.
- Use a sensitivity table for production growth, tariff changes and battery degradation.
Why peak-shaving savings disappoint
These are the practical objections and failure modes most likely to stop approval, delay commissioning or create an avoidable service call for Factories and C&I EPCs in tariff-driven markets.
- A monthly bill is used instead of interval demand data.
- Recharge creates a second billed peak later in the day.
- The same battery reserve is counted for both backup and economic dispatch.
Questions the buyer should ask before approval
- Can the model reproduce the current bill before storage is added?
- How many separate peaks occur on the worst operating days?
- Which tariff ratchet, billing interval and seasonal rule drives the result?
Turn interval data into a dispatchable design
Clean at least twelve months of interval data before sizing. Mark shutdowns, meter changes, missing intervals, production expansions and exceptional events. Recreate the billed demand from the utility rule and compare the model with invoices. If the baseline does not match the bill, adding a battery only makes the error harder to find.
For each interval, calculate the load above the proposed grid limit. The highest excess determines the required discharge power, while the sum of excess energy across the event determines usable kWh. Then add conversion losses, auxiliary demand, control margin and the state-of-charge reserve. Repeat the calculation for days with several peaks, because the battery may not have enough time or grid capacity to recharge.
Do not optimise one historic year and call the result final. Run production-growth, tariff, degradation, hotter-weather and lower-availability cases. Keep backup reserve separate and test what happens when an outage follows an economic dispatch. The investment case should show the demand saving, added recharge energy, service cost and any unserved target peaks rather than reporting only a simple payback.
Project workflow
- Validate interval timestamps, time zone, meter units and missing records.
- Reproduce the utility demand calculation before simulating storage.
- Test base, growth and stress years with reserve and degradation constraints.
- Convert the selected dispatch into PCS, usable-energy and EMS acceptance requirements.
Additional commercial check
The EMS acceptance test should replay representative load traces or use a controlled site test. Record the demand setpoint, response delay, metering location, battery limits and recharge rule. Include an alarm for insufficient reserve or unavailable power so the operator can distinguish a control limitation from equipment failure. After the first billing cycles, compare measured demand, dispatch energy, recharge cost and unavailable events with the model. Use that review to tune the target without violating warranty or backup reserve. A successful project needs an operating review, not only a commissioning certificate.
Keep the calculation reproducible
Save the raw meter or load file, cleaning log, tariff or operating rules, model version and every assumption that converts site data into power and energy. The owner should be able to rerun the base case when production, tariffs, loads or reserve policy change instead of treating the original result as permanent.
- Raw and cleaned time-series data with time zone and interval documented.
- Base, growth and stress scenarios with losses, reserve and degradation inputs.
- Selected control setpoints and the reason rejected alternatives were not chosen.
- A commissioning trace that connects the model to measured site behaviour.
At handover, give the owner the model inputs and an update trigger. A material tariff, load, production, weather or reserve change should lead to a controlled rerun rather than an unsupported promise that the original saving or runtime still applies.
Evidence base
Research review date: 30 September 2026. Quantitative statements below are tied to the named source and should be rechecked if the project is procured later.
Utility bills can combine energy, demand and time-of-use components, so savings models must use the actual tariff rather than a generic peak/off-peak spread. See DOE utility-rate guidance.
The IEA identifies batteries as a versatile source of short-term power-system flexibility, but the required duration depends on the service being delivered. See IEA Electricity 2026 flexibility analysis.
The U.S. Department of Energy checklist separates early project development, technical specifications and interconnection work for commercial lithium-ion systems. See U.S. DOE BESS procurement checklist.
IRENA reports a 2024 global average installed cost of USD 192/kWh for utility-scale battery storage, 93% below 2010. Buyers should not treat that global benchmark as a project quotation. See IRENA Renewable Power Generation Costs in 2024.
Related HMX Product and Project Pages
For product-level comparison, review the commercial and industrial energy storage range, 100kW/215kWh air-cooled system, 150kW/315kWh air-cooled system, Malaysia factory storage project. These pages show available HMX categories or references; final suitability still depends on the project specification and written confirmation.
Closing View
The useful output of the study is not one battery size. It is a repeatable model with stated inputs, sensitivities and an acceptance test that the owner can audit.