Coincident peak demand vs non-coincident peak demand, explained

Non-coincident peak demand is the highest demand your own facility reached in a billing period, whenever it happened. Coincident peak demand is what your facility was drawing at the moment the wider system reached its peak. The first sets the demand charge on your monthly bill. The second sets your share of capacity and transmission costs, often for a full year. They can be the same number, and for most sites they are not.

The US Energy Information Administration's glossary defines coincidental demand as the sum of two or more demands that occur in the same time interval, and a coincidental peak load as the sum of two or more peak loads in the same interval. Read from the customer's side, your coincident peak is your contribution to that shared moment. The mechanics of the capacity charges that result are covered in coincident peak and capacity charges explained; this article is about telling the two peaks apart, finding each on a bill, and managing them differently.

The two peaks, with an example

Consider two sites in the same market during a week when the system peaks at 5 p.m. on Wednesday. Site A is a plant that starts every line at 7 a.m. on Monday and hits 900 kW for one 15-minute interval, then settles near 500 kW; at 5 p.m. on Wednesday it is drawing 550 kW. Site B is a cold storage warehouse running at a steady 700 kW all week.

Site A pays the larger demand charge, because its own peak is higher. Site B carries the larger capacity allocation, because it was drawing more when the system was most stressed.

How each one is measured and billed

Non-coincident peak: the demand charge on the monthly bill

The utility's meter records demand in intervals, and for most utilities billing demand is the single highest 15-minute average draw during the billing period. That figure, sometimes lifted by a ratchet, is multiplied by the tariff's dollars per kW and appears on the bill as a demand charge. It resets every month, it is entirely under the customer's control, and it is the subject of demand charges explained. A national laboratory survey of more than 10,000 US tariffs found that demand charges typically make up 30 to 70 percent of a commercial electric bill.

Coincident peak: capacity and transmission allocation set by system peaks

The grid operator or the province identifies a small number of hours when system demand was highest, measures each large customer's draw in those hours, and uses the result to allocate the cost of generation capacity or transmission for a following period. The customer cannot know the hours in advance, and the charge that results is fixed for months or a year regardless of what the site does in between. Three programs show the pattern.

Canada: Ontario's Industrial Conservation Initiative

Ontario allocates a large part of its Global Adjustment cost to its biggest customers this way. Under the Industrial Conservation Initiative, consumers with average monthly peak demand over 1 MW, as well as manufacturers and greenhouses over 500 kW, can become Class A customers; those above 5 MW are enrolled automatically, and those between 1 MW and 5 MW, or between 500 kW and 1 MW in the eligible sectors, opt in through their distributor. Class A customers pay Global Adjustment based on the percentage that their demand contributes to the top five system coincident peaks measured during a defined base period.

The province measures those five hours over a base period that runs May 1 to April 30, and applies the resulting share to an adjustment period that runs July 1 to June 30 of the following year. The share is called the peak demand factor: the customer's consumption during the five hours divided by the province's total consumption in the same hours. The provincial system operator publishes the five hours after the base period closes and, since the base period that began in May 2022, determines them from Ontario demand rather than from adjusted settlement quantities (its ICI backgrounder is listed in the sources). For the 2019 to 2020 base period the province reported that the five highest system peaks ranged from 20,956 MW to 22,368 MW, and credited the program with an estimated 1,550 MW of peak reduction in 2018. The Class A and Class B split, and what it is worth, is worked through in Ontario Global Adjustment: Class A vs B.

United States: PJM's 5CP and ERCOT's 4CP

PJM, the largest US wholesale market, defines its summer window as June 1 through September 30 and identifies the five highest non-holiday weekday unrestricted daily peaks, the 5CP, with the data typically released in mid-October. Distribution companies use those hours to assign each customer a peak load contribution, which then sets the customer's share of capacity obligations for the delivery year. PJM's own guidance notes that most distribution companies use a 5CP approach and that peak shaving on those hours is usually an effective way to obtain a lower PLC.

ERCOT, the Texas grid operator, allocates transmission cost through four coincident peaks. A Texas transmission and distribution tariff on file with the state regulator defines a customer's 4CP kW as the average of the customer's integrated 15-minute demands at the time of the monthly ERCOT system 15-minute peak for June, July, August and September of the previous calendar year, updated on January 1 and fixed for the calendar year; a customer with no history is billed on its non-coincident peak instead. ERCOT publishes the 4CP calculations for each distribution service provider for those four months.

Finding each one on your bill

The common thread is that the coincident figure rarely appears as a kW value you would recognize. It shows up as a dollar line, or as a factor with several decimal places, derived from hours that were published months earlier. Bill validation that recomputes those lines from metered data and the published hours is the only way to know the figure is right.

Forecasting and responding

Because the counted hours are confirmed only after the fact, managing coincident peak is a forecasting exercise. Operators publish demand forecasts and peak trackers; weather, day of week and season narrow the candidate days to hot weekday afternoons in summer-peaking systems. A site decides, the day before or the morning of, whether to curtail for a window of a few hours, accepting that some curtailments will turn out to have been unnecessary. Responding on the wrong day costs some production; missing the right day costs a year of allocation, which is why the arithmetic favors acting on the top candidate days.

• Build a candidate-day rule: operator forecast above a threshold, temperature above a threshold, a weekday, and the hours in which the past five years' peaks have fallen.
• Pre-define the curtailment plan: which loads, in what order, for how long, and who authorizes it.
• Use storage or on-site generation to cover the window where the process cannot pause, and check standby charges before assuming on-site generation can run without a penalty.
• Review after each summer: which days were called, which counted, and what each hour's draw was. That record is the input to next year's rule.
• Treat non-coincident peak separately. Staggered start-ups and demand limiting run every day; coincident response runs on a handful of days. The demand response programs that pay for curtailment often reward the same hours.

Knowing your exposure: interval data plus validated bills

Two data sets answer the question. Interval data at 15-minute or hourly resolution, from Green Button or a meter feed, shows what the site drew in every hour, so once the operator publishes the peak hours the site's coincident draw is a lookup rather than a guess, and next summer's candidate-day rule can be tested against last summer's actual profile. Validated bills show what was actually charged: the billed demand, the capacity or transmission line, and the factor behind it, extracted as fields and recomputed.

MartinAI does both. It pulls interval data through Green Button and utility feeds, reads bills of any layout, and its utility bill validation recomputes demand, capacity and Global Adjustment lines against the metered data and the published peak hours, so a wrong factor becomes an exception with the evidence attached rather than a number nobody can check. The interval side is covered in putting Green Button interval data to work and interval data vs monthly bills.