What is a demand charge?

The mechanics

An energy charge bills how much electricity a facility used. A demand charge bills how much capacity the utility had to stand ready to deliver, and it is set by one interval.

The meter records average power over each fixed interval — commonly 15 minutes, sometimes 30, occasionally 5 — for the whole billing period. At the end of the month, the utility takes the single largest of those values and multiplies it by the demand rate. Everything the facility did in the other 2,879 intervals is irrelevant to that line.

This is why demand charges surprise people. A plant can cut its consumption by 8% and see the demand line unchanged, because the reduction happened in intervals that were never going to set the peak.

Four tariff mechanics decide what the number actually is:

• The interval length. A load that runs for 5 minutes at 3,000 kW registers as 1,000 kW on a 15-minute meter and as 3,000 kW on a 5-minute meter. The interval is a tariff term, and a demand-reduction strategy built against the wrong one produces nothing.
• Time-of-use windows. Many tariffs bill on-peak demand at a high rate and off-peak or all-hours demand at a lower one. A peak set at 3 a.m. may cost a fraction of the same peak set at 4 p.m.
• The ratchet. A ratchet clause sets a floor on billed demand for the following months, typically 50% to 80% of the highest peak set in the previous 11 or 12 months. One bad August interval can raise the demand line for a year.
• The power-factor clause. Many tariffs scale billed demand up when power factor falls below a threshold, which means an interval can set a higher billed demand than the actual kW drawn. See what is power factor.

Worked example, including the ratchet

A facility on a tariff with a $14.50 per kW-month demand rate and an 80% ratchet over 11 months.

• August peak interval: 2,400 kW → demand charge = 2,400 × $14.50 = $34,800
• Ratchet floor for the following 11 months = 2,400 × 0.80 = 1,920 kW
• November actual peak: 1,400 kW. Billed demand: 1,920 kW, because the ratchet floor is higher.
• November demand charge = 1,920 × $14.50 = $27,840, against $20,300 if the ratchet did not exist. The August interval cost $7,540 extra in November alone.

Now shave 150 kW off the August peak:

• New billed peak: 2,250 kW → $32,625, saving $2,175 in August
• New ratchet floor: 1,800 kW → saving a further $1,740 in each ratcheted month
• First-year total: roughly $21,300, of which only $2,175 appears in the month the change was made.

That lag is why demand work is hard to fund and easy to abandon. The saving from a peak reduction is mostly invisible until the ratchet releases.

Load factor: the number that tells you whether this is worth doing

Load factor is average power divided by peak power over the same period:

Load factor = kWh ÷ (peak kW × hours in the period)

For the facility above: 850,000 kWh in a 730-hour month against a 2,400 kW peak gives 850,000 ÷ (2,400 × 730) = 48.5%.

A load factor under about 50% means the facility is paying for capacity it uses less than half the time, and peak management usually has real money in it. A three-shift plant at 85% load factor is running flat and has very little peak to shave — for that site the demand conversation is about the 15% of the time it spikes, not about the base.

The 2026 Mid-Atlantic version of this problem

For facilities in the PJM footprint there is a second demand-driven charge that is not on the demand line at all. PJM capacity cleared at $28.92 per MW-day for the 2024/25 delivery year and at $325.00 for 2028/29 — roughly $119,000 per megawatt per year, contracted through May 2029. (Figure anchored on the 2028/29 clearing price and therefore conservative; the intervening delivery years cleared at $329.17 and $333.44.)

The price is settled and nothing a facility does now changes it. The quantity is a different matter: it is set by the facility's peak load contribution, measured on roughly five summer afternoons. See PJM demand response.

For context on why this is urgent locally: Pennsylvania industrial electricity prices moved +23.0% year to date through June 2026 against 2025, against a US industrial average of +6.1%. (EIA Electric Power Monthly, Table 5.6.B.)

What a demand charge is not

It is not a charge for using more electricity. A facility can raise its consumption and lower its demand charge by flattening its profile, and can cut its consumption while leaving the demand charge untouched. The two lines respond to different behaviour.

It is not the same as your peak load contribution. The demand charge is set by the facility's own highest interval, whenever it occurs. Peak load contribution — the quantity that drives capacity charges in PJM and comparable markets — is set by the facility's load during the system's peak hours, which may be intervals when the facility was not near its own maximum at all. Reducing one does not automatically reduce the other.

It is not a fixed percentage of the bill. The industry commonly quotes "30–70% of a commercial bill." That is a range across tariffs and load shapes, not a statistic, and quoting the top of it to a three-shift manufacturer at 85% load factor destroys credibility immediately. Twelve months of statements answer the question exactly for one facility in about ten minutes.

It is not reduced by an energy-efficiency project unless the project touches the peak interval. Lighting retrofits in a plant whose peak is set by compressor staging change the demand line by very little.