Solar proposals for commercial buildings often include a line for demand charge savings. The reasoning sounds right: the building peaks in the afternoon, the panels produce in the afternoon, so the peak should fall.
It usually does not fall by much, and the reason is not the panels. It is the way a demand charge is measured.
The measure and the resource do not match
A demand charge is set by the single highest fifteen-minute interval of the month. Not the average afternoon, not the typical day — the worst interval, wherever it lands.
Solar output is the opposite kind of quantity. It is judged by its totals: kilowatt-hours per year, capacity factor, the good days that carry the average. A panel that produces well on twenty-eight days and badly on two has had a good month for energy.
For demand, those two bad days are the month. If the building's highest interval arrives during a passing cloud, or on the one overcast Tuesday, or at 5:30 pm when output has already fallen away, the solar system contributes almost nothing to the number the utility bills. The meter only has to see the building at full load once.
What the modeling found
Lawrence Berkeley National Laboratory, working with researchers from the laboratory then known as NREL, tested this systematically in 2017. The study simulated fifteen commercial building types in fifteen cities across seventeen years of weather, with a range of system sizes and panel orientations, and calculated the effect on several demand charge designs.
Under a basic demand charge — the building's maximum in the month, whenever it occurs — the result was modest. For systems sized to produce half of the building's annual consumption, rooftop solar reduced demand charges by 7 percent in the median case, and by less than 15 percent in about nine cases out of ten.
Half of annual consumption is a large system. It will cut the energy portion of the bill substantially. It will barely move the demand portion.
The building type mattered. Schools, whose load lines up well with the sun, reached around 18 percent. Most other building types landed between 5 and 10 percent.
Bigger is not proportionally better
The obvious response is a larger system. The study found demand savings do rise with system size, but with diminishing returns, and for reasons that follow directly from how the charge works.
A larger array pushes the building's net peak later into the day, toward the hours when solar output is falling anyway. It also pushes the peak onto cloudy days, because on sunny days the array has already flattened the afternoon. In both cases the new peak is an interval the solar cannot reach, and every additional panel is bought for energy, not for demand.
Orientation does little to help. Turning the panels southwest or west, to catch more of the late afternoon, raised demand savings by no more than 3 percent in any case modeled.
Where the answer changes: the design of the charge
The same study shows that the demand charge design matters more than the solar system does.
When demand was measured only inside a fixed noon-to-4 pm window, the median reduction rose to 19 percent, and to 40 percent or more in some cases. A window that ends before solar output fades removes the early-evening intervals that defeat solar under a facility-wide charge. Longer averaging intervals helped for a similar reason: they smooth over the passing cloud.
So the first question for any solar proposal claiming demand savings is not about panels at all. It is which kind of demand your tariff bills — a facility maximum at any hour, or a maximum inside a defined on-peak window — and when that window ends. Facility demand versus on-peak demand sets out how to tell. And if your on-peak window runs into the evening, it works against solar rather than for it: the window then contains exactly the hours the panels cannot cover.
What this means for a proposal
Treat the demand line in a solar proposal with suspicion unless it was calculated from your own interval data against your own tariff, month by month, including the cloudy days. An annual average hides exactly the intervals that set the bill.
The honest way to present solar alone is the way measures that do not cut demand recommends for efficiency projects: claim the energy savings, which are real and substantial, and claim little or nothing on demand unless the tariff design supports it.
Where demand is the real target, solar is usually half of the answer. The 2020 follow-up work by the same research group examined solar paired with storage, and the logic is straightforward: the battery covers the intervals the panels miss. A battery sized to hold the peak through a cloudy afternoon and into the early evening turns an unreliable demand reduction into a reliable one. Sizing a battery for peak shaving works through that arithmetic.
A note on on-site generation and standby
A site large enough to install substantial solar should also check whether its tariff imposes standby or supplemental service charges on customers with on-site generation. Those charges are priced in dollars per kilowatt too, and on some schedules they can offset part of whatever demand saving the solar achieves. Standby charges covers what to look for.
None of this is an argument against solar. It is an argument for putting its savings on the right line of the bill. A project sold on the energy line and delivered on the energy line builds credibility. A project sold on the demand line and delivered on the energy line spends it.
