The PJM capacity construct produces the same strategic situation as ERCOT's four coincident peak, by a different route: a determinant fixed by your behavior during a small number of summer hours, applied for a year afterward, and unchangeable once set.
The details differ, the mechanism is reviewed and revised over time, and the exact measurement definition is a matter for PJM's own current documentation rather than for any summary. What does not change is the shape of the problem, and the shape is what a facility manager needs in order to plan.
The shape of it
Capacity in PJM is procured in advance, so that enough resources are committed to serve the region's peak with a reserve margin. The cost of that procurement is allocated to load, and the allocation to a given retail account is based on how much that account was drawing when its zone was at its highest.
The sequence:
- During a defined summer measurement period, the highest zonal peak hours are identified.
- Your load during those hours is measured.
- Scaling factors are applied — reconciling zonal to system values, accounting for losses, and applying the forecast requirement.
- The result is your capacity obligation for the following delivery year.
- That obligation is multiplied by a capacity price to produce the capacity charge you pay each month of that year.
Steps two and four are the ones a customer influences. Step two is a handful of hours next summer. Step four is arithmetic.
Why this is worth attention
Because the leverage is concentrated in a very small number of hours, and because the resulting charge is large enough to be visible in a supply contract.
What one summer's peak hours are worth
A site evaluating a capacity tag reduction program.
- Average load across the measurement hours, unmanaged3,200 kW
- Reduction achievable during a called curtailment1,100 kW
- (Reduced average across the measurement hours)2,100 kW
- Applicable scaling factor applied to the measured contribution1.09
- (Capacity obligation before: 3,200 × 1.09)3,488 kW
- (Capacity obligation after: 2,100 × 1.09)2,289 kW
- (Obligation avoided)1,199 kW
- Capacity price applied, per kW-year$61.00
Value of the reduction, for the delivery year$73,139
The capacity price varies by auction and by zone, and the scaling factors are set by PJM. Use the values applicable to your zone and delivery year — those above are illustrative and exist to show the structure.
The structure is what matters. A modest reduction, sustained across a small number of hours, produces a saving that persists for a full delivery year. Very few operational changes have that ratio of effort to return.
What makes it hard
You are predicting somebody else's peak. The hours that count are determined by the zone's load, not yours. Forecasts are available and alerts are a commercial product, but you are always acting on a prediction. The general problem is discussed in coincident and non-coincident demand.
The feedback loop is a year long. You act in July and find out whether it worked when next year's obligation is assigned. Nothing about the following month's bill tells you.
More than one measurement hour matters. Because the obligation derives from performance across several hours rather than one, a single successful curtailment does not carry the whole result. Consistency across the measurement period is what produces the reduction.
The rules are revised. Capacity market design in PJM is an active regulatory subject and both the market rules and the measurement conventions change over time. Anything you build should reference the current documentation rather than a fixed assumption.
Building a program
- Confirm from your supplier what your current obligation is, what rate is applied to it, and which delivery year it covers. If it is embedded in a bundled supply rate, ask for it to be broken out.
- Confirm the current measurement period and definition from PJM's own materials, for your zone.
- Arrange a forecast or alert source before the season starts, not during it.
- Quantify what you can shed and for how long, honestly. An hour of curtailment repeated across several afternoons is a different operational commitment from one event.
- Write the decision rule and name the person with authority to call it.
- Log every event: the alert, the action, the load achieved, and whether the hour turned out to matter. Without the log you cannot evaluate the program.
- Check the interaction with any demand response enrollment. Being paid to curtail and reducing your own obligation by curtailing are two distinct benefits, and program rules govern whether both can apply. See demand response programs.
The last point is genuinely worth checking rather than assuming. Curtailing during a capacity measurement hour may also coincide with a demand response event, and the terms of the program determine how that is settled.
Where storage changes the answer
The operational cost of curtailment is what limits participation for most sites. A battery removes it: you discharge into the peak hours instead of stopping production, so you can respond to every alert without a production decision.
That converts the sizing problem into something well-defined — a known number of hours at a known power level during a known season — which is a much easier specification to write than continuous monthly peak shaving. Both cases are handled in sizing a battery for peak shaving, and the capital appraisal in total cost of ownership for a behind-the-meter battery.
And the charge you still have
None of this touches your distribution demand charge, which is measured non-coincidentally against your own peak and is a separate problem entirely. A site can reduce its capacity obligation substantially while its monthly demand charge does not move at all. Price them separately, from your own tariff and your own interval data: how to read an industrial electricity bill and demand charges explained.
