Cutting Metered Demand Charges with AI-Driven BESS
Industrial plants run machinery — variable frequency drives, rollers, compressors, condensers — that draws sharp, momentary power spikes. Even when average consumption is modest, these spikes push up Metered Demand (MD), and DISCOMs levy fixed charges against the contracted demand that a high MD forces the facility to hold. The result is a heavy fixed-cost burden that has little to do with how much energy the plant actually uses.
In one modelled facility, momentary peaking drove an MD of around 3,300 kVA while the average load was only about 2,300 kVA — roughly 44% higher demand than the plant genuinely needed, with fixed charges billed on that peak.
How BESS Caps the Peak
A Battery Energy Storage System charges during low-load periods and discharges precisely when demand spikes, shaving the peaks so the meter never records them. For a plant averaging 5 MW with momentary peaks above 8 MW, an intelligently controlled BESS can cap MD at around 6 MW — cutting demand charges directly.
The savings compound beyond fixed charges. Capping the peak lifts the plant's Plant Load Factor (PLF) — in the modelled case from roughly 57% to above 95% — which qualifies the facility for a higher per-unit DISCOM rebate on its energy charges. Together, controlling peaks and optimising MD trimmed roughly 7–8% off the facility's total monthly DISCOM bill.
Grow Without Growing Your Contracted Demand
MD reduction also frees up headroom. By suppressing spurious peaks, a facility can add new machinery and expand production while staying within its existing contracted demand — avoiding costly upgrades to its grid connection. In the modelled plant, a 5 MW expansion that would otherwise require 5 MW of additional contracted demand needed only 2 MW once peaking was controlled.
Why the BESS Must Be Software-Defined
MD is unforgiving: a single unmanaged peaking event anywhere in the month sets the billed demand for the entire month. To deliver savings, a BESS must anticipate and offset every peaking event in real time — which is only possible with AI/ML-driven control tuned to the facility's load. Right-sizing matters too: in the report's illustration, a 1.25 MWh battery failed to catch four peaking events, while a battery with only 23% more capacity absorbed them all. Ingro Energy applies AI across pre-commissioning digital twinning, commissioning-stage sizing and procurement, and post-commissioning EMS operation — automating charge and discharge against real-time MD to protect both the savings and the battery's life. The full report details the load profiles, sizing analysis, and rebate mechanics.