Coincident peak and capacity charges explained

Most energy managers can read a demand charge: it tracks the highest average kilowatt draw their building hit during a billing period. Coincident peak and capacity charges are different, and they catch experienced teams off guard because the cost driver is not your own peak at all. It is your consumption during a handful of hours when the whole grid was at its busiest, hours you had no way to see marked on a calendar.

That single distinction changes how a bill behaves. A capacity charge measured during grid peaks can be locked in for a full year and billed in equal monthly slices, so a warm afternoon last summer can quietly sit on every invoice you pay this year. Because the charge often lands as a flat line item rather than a metered quantity, it rarely gets the scrutiny it deserves.

What a coincident peak actually is

A coincident peak is your facility's electrical demand at the exact moment the wider system reaches its highest load. Grid operators and utilities recover the cost of generation and delivery capacity from the customers who were drawing power when that capacity was most stressed. In practice, they measure your average demand during a small set of system peak hours and turn it into a capacity obligation, sometimes called a capacity tag or peak load contribution.

This is the opposite of a non-coincident demand charge, which looks only at your own maximum draw regardless of when it happened. With a coincident charge, the timing is everything. You could run a hard peak at 6 a.m. and pay nothing extra for it, while a moderate load during a hot late afternoon, when the grid itself peaked, sets a charge that follows you for months.

How capacity and transmission charges get set

The mechanics vary by market, but the pattern is consistent. A grid operator identifies the highest system demand hours over a defined window, measures each large customer's demand during those hours, and converts it into an annual obligation. In the largest organized wholesale market in the United States, that obligation is built from the five highest summer peak hours, commonly written as 5CP, and then billed across the following delivery year.

The sequence usually works like this:

• The operator watches for the highest system load hours over a base period, often the summer months when air conditioning drives demand.
• Your average metered demand during those specific hours becomes your capacity or peak load contribution value.
• That value is multiplied by a capacity price, which is set through a forward auction or a regulated rate.
• The result is fixed once, then charged in equal monthly installments across the delivery year.

Because the price is set in advance and the quantity is set by a few past hours, neither number moves with your day to day operations. You can cut usage in the fall and winter and still see the same capacity line every month, because it was locked to last summer's peak-hour behavior.

The price side has been moving sharply. In the largest US wholesale market, capacity prices settled roughly nine times higher in one recent auction than the year before, driven by rising demand and tightening supply. A customer whose peak-hour contribution stayed flat could still see the capacity line jump simply because the price per unit of capacity climbed. That decoupling of price and quantity is part of what makes the charge hard to budget: two independent levers move it, and neither is visible on a monthly meter read.

The numbers behind the charge

The dollars are not small. Demand-related charges can make up 30 to 70 percent of a commercial electricity bill according to national laboratory research, and capacity is a large slice of that. In the largest US wholesale market, capacity prices recently cleared at a record of about 329 dollars per megawatt-day, a level that flows straight through to the capacity line on commercial invoices. Analysts have described the pool of these charges as a multibillion dollar annual cost that many customers do not track hour by hour.

Why these charges are hard to see coming

Three things make coincident peak charges slippery. First, the trigger hours are only known after the fact. The grid peak might land on any hot afternoon, and you learn which hours counted weeks or months later. Second, the charge is decoupled from the current month. A steady, low-usage month can still carry a large capacity line because the value was set earlier. Third, the charge often appears as a flat fee or a per-account allocation rather than a metered quantity, so it does not move when your kilowatt-hours move, and a quick glance at the bill does not reveal what set it.

For a portfolio, the problem compounds. Different sites can sit in different markets, each with its own definition of the peak window and its own price. A finance team reconciling dozens of accounts has no easy way to confirm that a capacity value matches the metered demand it was supposedly built from, unless the underlying peak-hour data is collected and lined up account by account.

What steady, well-timed operations are worth

The upside of a coincident peak structure is that trimming load during the specific peak hours can lower the charge for a whole year, not just a month. In one large Canadian province, eligible large customers pay a substantial share of provincial supply cost based on their contribution to the top five demand hours of the year, measured over a twelve month base period. Sites that predict those hours and reduce demand during them can move their allocation meaningfully, which is why some operators publish running trackers of the current peak hours.

None of that works without accurate interval data and clean historical bills. You cannot manage what you cannot see, and peak-hour strategy depends on knowing your demand at fifteen minute or hourly resolution, matched against the hours the market actually counted. When that data is scattered across PDF bills and portal exports, the analysis stalls before it starts.

There is also a prediction problem. Since the counted hours are only confirmed after the base period closes, reducing load on the right days depends on forecasting which hot afternoons are likely to set the system peak, then acting on those forecasts before the fact. A miss cuts costs on a day that did not count, and a hit avoids charges for a full year. That asymmetry rewards teams that keep a tight, current view of both their own demand and the grid's likely peak conditions, and it penalizes teams working from bills that arrive weeks after the hours that mattered have already passed.

Turning peak-hour data into a plan

This is the gap MartinAI is built to close. It collects utility bills and interval data across any commodity and any utility, then normalizes the line items so capacity, transmission, and coincident peak charges are broken out as structured fields rather than mystery totals. You get the demand values, the charge basis, and the period each charge covers, aligned across every account in a portfolio.

With that foundation, your team can tie each capacity charge back to the metered demand behind it, watch how peak-hour behavior tracks over time, and feed clean, hour-level data into whatever demand management or forecasting tools you already run. The point is not another dashboard for its own sake. It is having the peak-hour numbers ready and trustworthy so the strategy work can actually happen.