Most CFOs of mid-size Philippine businesses have never been told this. Most facility managers have figured it out — but rarely explained it back upstairs. The demand charge line on a Meralco commercial bill, billed in kilowatt-amps (kVA) or kilowatts (kW), is not a function of how much electricity you used in the month. It is a function of when you used the most of it, in any single 15-minute window.
1. Meralco meters every commercial account on a 15-minute averaging interval.
2. The single highest 15-minute kVA reading in the billing month sets your demand charge for that month.
3. On a typical GP-LV bill, the demand charge is 8–12% of the total. On GP-HV it can reach 15–20%.
4. Behind-the-meter battery storage timed to your own peak window cuts the demand charge by 30–50% directly — before you change a single piece of thermal plant.
How the meter actually works
Every commercial Meralco meter records consumption in continuous 15-minute averaging intervals — 96 readings a day, ~2,880 readings a month. Each reading is the average power demand across that 15-minute window. At the end of the billing cycle, the meter does not add up the readings. It scans them for the single highest value and sets that as your billed demand kVA.
The maths is brutal. If your factory runs at a steady 200 kVA average all month, your demand charge is calculated as 200 kVA × the published demand rate (currently around ₱700/kW combined for GP-LV in 2026). But if your factory averages 180 kVA for 29 days and 29 hours, then one 15-minute window hits 320 kVA because three motor starters fire simultaneously at the start of a shift — your billed demand is 320 kVA. For the entire month. Across all 2,880 intervals. From one window.
The difference between 200 and 320 kVA at ₱700/kW is ₱84,000 in additional demand charge, for a single 15-minute event that the operator probably didn't notice on the BMS dashboard.
One uncontrolled motor start. One walk-in cooler defrosting at the wrong moment. The whole month re-rated.
What the spike actually looks like
In our two-week site surveys (run free as part of any scoping engagement), the same handful of spike shapes show up across every commercial customer we meter:
- 6 a.m. cold-storage defrost cycle. Walk-in cooler evaporator fans + electric defrost heater + new compressor pull-down on the same feeder, simultaneously. Easily 80–120 kW above baseline for 15–25 minutes. Most operators have this scheduled at the worst possible moment.
- 5 p.m. office HVAC restart after lunch shutdown. Chiller cycling back up while every aircon in the office wakes from setback. Combined ramp is sharper than the morning startup because the load is hotter.
- 10 p.m. laundry night-shift kick-off. Multiple washer-extractor motors starting within 2 minutes of each other, plus boiler firing, plus tunnel finisher. We routinely see 200 kW spikes from hotel laundries that nameplate-load at 80 kW continuous.
- Bakery proofing room electric resistance heater. A bank of 30 kW electric elements firing on contactors as the proofer warms from cold start. No soft start. No staging.
- The "everyone arrives at 8" spike. Office HVAC, IT room cooling, kitchen pre-heat, water heaters — all pulled by the same human schedule.
None of these spikes show up on a monthly utility-bill summary. They are recorded second-by-second by the meter, then filtered out of the bill by being averaged into 15-minute bins. The customer sees only the result: one number, billed demand kVA, that they accept as "what we used."
What the spike costs you, in pesos
Take a generic mid-size commercial site with 800 kVA contracted demand, billed under the GP-HV schedule. Demand-charge component for May 2026, with combined fixed and variable demand charges of approximately ₱750/kW:
| Billed demand | Monthly demand charge | Annual demand charge |
|---|---|---|
| 500 kVA (typical operating average) | ₱375,000 | ₱4.5M |
| 650 kVA (modest 30% spike above average) | ₱487,500 | ₱5.85M |
| 800 kVA (single bad-start event) | ₱600,000 | ₱7.2M |
| 500 kVA + 40% spike shave with storage | ₱290,000 | ₱3.48M |
The "single bad-start event" line is the most useful one. The difference between a steady operating profile and a single bad 15-minute window in any one month is ~₱225,000 extra on the bill — for a one-off event that the operator probably has no procedure to prevent.
Why the grid actually cares about your spike (and how this scales)
This is not just a billing quirk. The reason Meralco charges this way is because of what those spikes do at the grid level. Across thousands of commercial accounts in Metro Manila, those individual 15-minute windows hit the grid simultaneously — 6 a.m. when cold storage defrosts, 5 p.m. when offices switch over, 10 p.m. when laundries fire up the night shift. The aggregate is what runs the grid closest to brownout.
When one customer shaves their own spike with a battery, it removes that customer from the simultaneous peak. When ten thousand customers do it — through controllers that know each customer's specific peak window — the grid stops needing more centralised coal capacity to handle the simultaneous worst-case. Distributed demand-shaving relieves the grid faster, and cheaper, than any new power station.
How you actually shave a 15-minute spike
The architecture has three parts and they have to work together:
- A meter, or a survey, that finds your specific spike. Most operators can't shave what they can't see. Two weeks of 1-second-resolution power-quality logging on your incoming supply is enough. Karnot includes this free in any scoping engagement; the report is yours regardless of whether you proceed.
- Behind-the-meter storage sized to the spike, not the daily load. A 100 kWh LiFePO4 battery sized to cover a 200 kW × 30-minute window is a different specification than a 100 kWh battery sized for daily solar shifting. For demand shaving, the C-rate (how fast it can discharge) matters more than total capacity.
- A controller that knows when your spike is about to happen. The battery has to start discharging before the spike arrives — if it waits for the kVA reading to climb, the 15-minute averaging window has already started and the damage is done. The Karnot iSAVE platform (rolling out 2026–27) is built around this — pattern-matching the customer's own historic profile against real-time loads to pre-empt the discharge.
Layer in a heat pump and thermal storage on top, and the same controller can shave the thermal portion of the spike before calling on the battery. Thermal load is the largest single swing variable in most commercial sites; cutting the thermal peak alone can take 60–80% off the kVA spike at source, leaving the battery to handle only the residual.
Solar cuts the cost of every kilowatt-hour. Storage + controller cuts the cost of the worst kilowatt-hour.
How a project starts
One free site visit. Our engineers spend a day on site, plug a power-quality logger into your incoming supply and key sub-panels, and let it record for two weeks at 1-second resolution. We pull 12 months of Meralco bills alongside. Within 14 days you receive a one-page indicative report — your actual spike profile, the kVA you could shave with each level of intervention, the storage capacity required, the controller logic, and the projected demand-charge saving. No procurement triggered. No obligation. The data is yours.
Want to see your own 15-minute peak?
Book a free two-week power-quality survey. We come back with your actual spike profile, the kVA you could shave, the storage capacity required, and the projected demand-charge saving — no commitment.
Book a free survey Savings calculator