Most demand charges bill something you can see in your own data. Four coincident peak does not. It bills your behavior during four specific quarter-hours that are chosen by the entire Texas grid, announced only after the fact, and used to price your transmission charge for the following year.
It is the clearest example anywhere in United States electricity billing of a determinant you cannot measure at the time and cannot change afterward.
The mechanism
Transmission cost in the ERCOT region is allocated to load using a coincident peak method. In outline:
- ERCOT identifies the single highest fifteen-minute interval of system-wide demand in each of the four summer months: June, July, August and September.
- For each of those four intervals, a customer's own demand during that interval is recorded.
- Those four figures are averaged. The result is the customer's transmission billing determinant.
- That determinant is applied to the transmission rate for the following calendar year.
Four intervals. One hour of metered time in total, spread across a summer, setting a charge that appears on twelve subsequent bills.
Why it is designed this way
The reasoning is the same cost causation argument that underlies any coincident charge, applied to the part of the system where it holds best.
The transmission network exists to move power at the moments of greatest stress. It is planned and built against the system peak, not against any individual customer's peak. A facility drawing heavily at three in the morning imposes no burden on the transmission peak whatever; a facility drawing heavily during the hottest afternoon of August imposes a very direct one.
Allocating transmission cost by contribution to the system peak therefore matches cost to cause more closely than a volumetric charge or a non-coincident demand charge would. The general argument is in coincident and non-coincident demand.
The design has an intended side effect, and it works: it gives large customers a strong reason to reduce load at exactly the moments the grid is most stressed.
What managing it actually involves
This is a forecasting exercise, and it should be run as one.
Prediction. Summer system peaks in Texas correlate with temperature, and with the interaction between temperature and the working week. Forecasts are available — from published system load information, from providers, and from commercial services that issue 4CP alerts as a product. Most large participants buy the forecast rather than build it.
A decision rule agreed in advance. When an alert arrives, somebody has to decide whether to curtail, and there is no time for a meeting. The rule, the authority and the notification path should be written down before June.
A curtailment plan that can actually be executed. Which loads shed, in what order, how quickly, and what the operational consequence is. A plan that requires forty minutes of preparation is not usable against a fifteen-minute interval that has already been forecast for a specific hour.
A tolerance for false positives. You will curtail on days that turn out not to contain the monthly peak. That is the cost of the strategy, not a failure of it.
Why a high false-positive rate is still economic
A site evaluating whether to run a 4CP program.
- Average demand across the four intervals, unmanaged2,400 kW
- Achievable reduction during a curtailment900 kW
- Transmission rate applied to the determinant$4.85 / kW-month
- (Annual value if all four are caught: 900 × 4.85 × 12)$52,380
- (Value if three of four are caught: 675 × 4.85 × 12)$39,285
- Curtailment events actually called in the season14
- Estimated production cost per unnecessary curtailment$1,200
- (Cost of 10 unnecessary curtailments)$12,000
Net value at a 10-of-14 false positive rate$27,285
Even catching only three of four intervals while curtailing ten times unnecessarily leaves a clear positive. Rates, loads and costs are illustrative — the structure of the calculation is the point.
That structure is why 4CP programs are common among large Texas industrials despite the apparent absurdity of curtailing on a guess. The asymmetry is large: catching an interval is worth a great deal, missing one costs the whole year, and an unnecessary curtailment costs an hour of output.
What it interacts with
Your ordinary demand charge does not go away. The distribution-level demand charge on your tariff is a separate determinant, measured non-coincidentally, and 4CP management does nothing for it unless your own peak happens to coincide. Both need pricing separately: facility, on-peak and billing demand.
Batteries change the calculation considerably. A battery removes the operational cost of curtailment — you discharge instead of stopping production — which collapses the false-positive penalty to almost nothing and lets you respond to every alert. That changes the sizing logic: the requirement is four events of a known duration rather than continuous monthly shaving. See sizing a battery for peak shaving.
On-site generation does the same, subject to whatever standby charge the tariff applies to a site that can serve part of its own load — a cost that has to be netted off the benefit before the program is evaluated.
The determinant is sticky. Because it is set once and applied for a year, a bad summer cannot be recovered by good behavior in the autumn. In that respect it behaves like a ratchet with a twelve-month memory: ratchet clauses.
Before you act on any of this
Two cautions, both serious.
First, allocation methodology is set by regulatory rule and is periodically reviewed. The mechanism described here is the four coincident peak method as it has operated; proposals to change coincident peak allocation appear from time to time in Texas regulatory proceedings. Confirm the current rule with the Public Utility Commission of Texas before designing a program around it.
Second, confirm the method applies to your service at all. Not every account in ERCOT has transmission cost allocated this way, and a smaller commercial account may have it recovered volumetrically. Reading it off your own bill is the first step, which is what how to read an industrial electricity bill is for.
If you are outside ERCOT, the equivalent question is how your own region allocates capacity and transmission cost. In PJM the analogous mechanism runs through peak load contribution: PJM capacity tags.
