Commercial and industrial storage is usually sized against a typical day and a peak-valley price spread. Both simplifications discard the part of the bill that storage is best placed to reduce.EcoSizing sizes against the full 8760-hour sequence for this reason.
What the bill is made of
A C&I electricity bill has two components with different logic. Volumetric energy charges scale with consumption. Demand charges scale with the highest power drawn in the billing period, independent of how long that draw lasted.
Demand charges account for a substantial share of the total — in the range of 15 to 30 percent in the markets we work in.
The distinction matters because the two components respond to different things. Energy charges respond to how much is shifted. Demand charges respond to whether a single interval was clipped.
Why a typical day cannot see the demand charge
A typical-day profile is constructed by averaging. Averaging removes the extreme intervals by definition.
A demand charge is set by one interval in the billing period — often a single 15-minute window across an entire month. That window is precisely what averaging eliminates. A model built on typical days will report the energy arbitrage correctly and miss the demand charge entirely, in both directions: it will not identify the peak that needs clipping, and it will not verify that the storage configuration is capable of clipping it.
The consequence is a configuration sized for arbitrage that leaves the larger cost item untouched.
The second thing averaging removes
Seasonal mismatch between generation and load disappears under the same operation.
A spring weekend with high irradiance and a shut plant produces heavy curtailment. Averaged into a typical day, the same period appears as ordinary consumption absorbing ordinary generation. The model reports high self-consumption; the site experiences curtailment. PV capacity is sized against a self-consumption rate that does not occur.
Both failures share a cause. Sizing is determined by extremes, and typical days are constructed by removing them.
Sizing against the full sequence
Running the full 8760-hour load and generation sequence keeps the extreme intervals in view. Two things follow.
Peak identification becomes exact
The intervals that set each billing period's demand charge are identified rather than estimated, and the storage power and energy required to clip them can be checked directly.
The reservation trade-off becomes explicit
Energy held in reserve to clip an unpredictable demand peak is energy unavailable for arbitrage. These two uses compete for the same stored kWh, and the balance between them is a decision the model has to make at every interval rather than a policy set in advance.
A configuration optimised on this basis holds state of charge against demand events while shifting energy across the tariff structure, with the split resolved hour by hour under the actual load sequence.
What to ask for
- Interval-level simulation results across the full year, rather than a typical-day illustration.
- A reconstruction of the site's historical bills from the model, showing that tariff parameters and load data are correct before any projection is built on them.
- Confirmation that demand-charge savings and arbitrage revenue are not double-counted, and that both are separated from any market or incentive revenue.
- Whether the local tariff bills demand on maximum demand or on transformer capacity, since storage affects only the former.
The optimisation method and its modelling assumptions are set out in EcoSizing: Method and Scope. The scenario discussion is in White Paper on Intelligent Microgrid Planning and Sizing
