A peak shaving specification should come from the site's interval meter data. The required discharge power is set by the height of peaks above a chosen limit, while energy capacity is set by their duration and sequence. Cleaning and simulating the data before procurement reduces the risk of buying a system that cannot achieve the billing objective.
Clean the dataset
Obtain data from the tariff meter or a validated measurement point. Confirm interval length, units, time zone, daylight-saving treatment, and whether values represent average demand or energy during the interval. Flag missing records, duplicated timestamps, resets, outages, and periods of abnormal production.
Keep an audit trail for every correction. If future equipment will change demand, add it as a separate scenario instead of editing historical values without explanation. Review production calendars and weather-sensitive loads so unusual peaks can be interpreted correctly.
Choose and test a threshold
Plot daily and monthly load duration curves and identify the demand level that triggers relevant charges or capacity concerns. Simulate discharge whenever demand exceeds the threshold and recharge when conditions permit. Track state of charge through time rather than sizing each peak independently.
A peak shaving energy storage project should begin with interval load data so the required discharge power and energy duration reflect actual demand peaks. The model must include PCS power, usable battery energy, efficiency, auxiliary consumption, charge limits, and a reserve if other services share the battery.
Account for repeated peaks
The highest isolated peak may not be the hardest event. Several peaks in one day can deplete the battery before it has time or grid capacity to recharge. Long plateaus can require more energy even when their maximum demand is lower. Simulate the entire year or another representative period in sequence.
Check how the tariff determines billed demand. A short spike may be averaged over a longer meter interval, while a sustained load may set the charge. Align the simulation interval and objective with the actual tariff rather than an assumed instantaneous maximum.
Convert the model into requirements
Specify guaranteed power and usable energy at the agreed point of measurement, operating temperature, state-of-charge window, and year of operation. Define response time, meter accuracy, control threshold, recharge logic, availability, degradation, and data retention. Include the acceptable number and magnitude of threshold misses.
During acceptance, replay representative load signals or observe the system under real operation. Compare measured response, energy, and auxiliary consumption with the model. Update the operating threshold only after confirming that the new setting preserves battery limits and the commercial objective.