In short
- Energy is priced hour by hour, linked to wholesale market prices rather than to a fixed schedule.
- It rewards a site that can move load away from expensive hours and penalizes one that cannot.
- The annual average price is usually attractive; the risk lives in a small number of extreme hours.
- Demand charges, transmission and capacity charges continue to apply separately.
- Evaluate on the worst weeks in the historical data, not on the average year.
Time-of-use pricing divides the day into a few periods with prices fixed a year in advance. Real-time pricing removes the approximation: each hour is priced at or near what the wholesale market actually charged for it.
The consequence is a much stronger incentive to move load, and much more variance in the bill. Whether that is a good trade is a question about your operation, not about the tariff.
What you are taking on
Under a fixed rate, the supplier absorbs the risk that wholesale prices spike, and charges a premium for doing so. Under real-time pricing, you absorb it and keep the premium.
That is the entire economic proposition. The average is usually favorable — you are no longer paying an insurance premium — and the tail is entirely yours. In practice, most of the annual variance in a real-time bill comes from a very small number of hours: cold snaps, heat waves, and the occasional supply event.
The evaluation that matters
The mistake is to compare annual averages. The right test is to price your historical consumption against the historical hourly prices for the same period, and then look at the worst weeks rather than the mean.
Testing exposure rather than the average
Same consumption, priced against a year of historical hourly prices.
- Annual consumption3,100,000 kWh
- Cost under the current fixed schedule$204,600
- Cost under real-time pricing, same consumption, whole year$186,300
- (Apparent annual advantage)$18,300
- Consumption during the 40 most expensive hours of that year61,000 kWh
- Average price during those 40 hours71 ¢/kWh
- (Cost of those 40 hours alone)$43,310
- (Cost of the same 40 hours under the fixed schedule)$4,026
Exposure concentrated in 40 hours$39,284
The annual advantage is $18,300 and the exposure in forty hours is more than twice that. A site that can curtail during those hours does very well; a site that cannot has bought a lottery ticket. Figures illustrative.
That is the shape of the decision. Forty hours out of nearly nine thousand determine whether the schedule was a good idea, and your ability to respond during those forty hours is the only thing that changes the answer.
Who it suits
Suits: sites with substantial, genuinely deferrable load; operations with storage, thermal or electrical, that can be discharged on price signal; sites already running a demand response capability, since the operational machinery is the same; organizations able to tolerate month-to-month variance in a cost line.
Does not suit: continuous processes with no latitude; operations where a curtailment decision has to go up two levels of management; sites without automation, since responding to hourly prices manually is not sustainable; businesses whose budgeting process cannot absorb an unusual month.
That last one is not a technicality. A schedule that saves money on the year and produces one alarming month can be abandoned for reasons that have nothing to do with its economics. If the finance function is going to be asked to absorb variance, it should agree to that before the election rather than discover it in February — and the way to get that agreement is to show the historical worst month rather than the historical average.
The machinery you need
A price feed and an automated response. Prices published day-ahead have to reach a control system that acts on them without a human in the loop for routine decisions.
A pre-agreed curtailment plan. Which loads shed, at what price threshold, for how long, with what operational consequence — decided before the first expensive day rather than during it.
A threshold that reflects your own economics. The point at which shedding is worthwhile is where the price exceeds the value of the production you would forgo. That number is yours and it is not the same as anyone else's.
Someone accountable for the outcome. Real-time pricing rewards attention. Left unattended it drifts toward the outcome in the calculation above.
What it does not change
Real-time pricing applies to the energy component only. Demand charges continue under their own rules, as do transmission and capacity obligations, and those are frequently the larger part of an industrial bill. A site that moves to real-time pricing and ignores its demand determinant has optimized the smaller number: demand charges explained.
There is also an interaction worth noting. High wholesale prices tend to occur during system stress, which is also when coincident peak determinants are set. A curtailment triggered by price may therefore reduce a capacity obligation at the same time — genuine stacked value, but only if the two are evaluated together rather than by different people: coincident and non-coincident demand.
The sensible progression
Most sites should not go to real-time pricing first. Prove the capability under a time-of-use schedule, where the expensive periods are known a year ahead and a missed response costs a known amount: time-of-use rates for commercial accounts.
A site that consistently shifts load under time-of-use has demonstrated the machinery works, and can then evaluate real-time pricing on evidence rather than on intention. The full comparison method applies either way: how to choose a rate schedule.
