There is a particular way an energy manager loses the confidence of a finance function, and it is not by proposing a bad project. It is by proposing a good project with the wrong number attached to it.
The measures below are all worth doing. Each reduces consumption, often substantially, and most have defensible payback on the energy line. None of them reliably reduces the highest fifteen-minute interval of the month, which is the only thing a demand charge prices.
The test is always the same, and it takes one minute with interval data: was the equipment running at full output during the interval that set your demand charge? If it was, reducing its draw reduces demand. If it was off, idling, or already modulated down, it does not.
1. Lighting retrofits
The standard opening move of any efficiency program, and rightly so — the savings are large, continuous and easy to verify.
Whether they touch the demand charge depends entirely on the site. In a windowless warehouse on a fixed lighting schedule, the lights are on during the peak and the retrofit does reduce demand. In a daylit facility with automatic dimming, lighting load at two in the afternoon may already be a fraction of its rating, and at a plant whose peak is driven by process equipment, lighting may be a few percent of the interval average regardless.
The energy case is usually strong. The demand case has to be established rather than assumed.
2. Standby and off-hours load reduction
Switching off equipment that was left running overnight and at weekends is genuinely valuable work and often finds surprising amounts of consumption.
But it acts precisely when the site is at its lowest, which is by definition not when the demand determinant is set. Kilowatt-hours fall; the peak does not move.
There is a variant that makes it worse. A program of aggressive overnight shutdowns, without a startup sequence, creates a larger simultaneous restart the following morning — and that restart can set a higher peak than the site had before the measure was introduced. The energy saving is real and the demand charge rises. The fix is staggered startup, and it should be designed at the same time as the shutdown program rather than after the first bill.
3. High-efficiency motor replacement
Replacing a motor with a higher-efficiency unit reduces losses whenever it runs, which is a real and permanent saving.
The reduction in drawn power is proportionally modest, because the difference between a standard and a premium efficiency motor is a few percentage points of input. Applied to the peak interval, a few percent of one motor's load is rarely visible in the demand determinant.
Motor replacement earns its place on reliability, on maintenance and on energy. It is not a peak measure, and a business case that treats it as one will be embarrassing at the review.
4. Building envelope improvements
Insulation, glazing, air sealing and roof work reduce the heating and cooling load continuously, and they are among the most durable investments a facility can make.
Their effect on peak demand is real but indirect and hard to predict: a lower cooling load means the chillers work less hard at the peak, but by how much depends on the plant's staging, its control strategy and the weather on the day. Envelope work is normally justified on energy and on comfort, with any demand effect treated as an unquantified upside rather than as a line in the model.
5. Compressed air leak repair
Frequently the highest-return maintenance activity available in a plant, and it is included here with a caveat because it is the one entry on this list that often does reduce demand.
Leaks make compressors run more, which means more energy. Whether that translates into a lower peak depends on whether the compressors were running at the peak interval and whether they were loaded at that moment. At many plants they were, and leak repair reduces both lines.
The reason it belongs on this list at all is that the demand benefit is routinely asserted without being checked. It is checkable: look at the compressor load during the peak interval before assuming.
6. Power factor correction
The most commonly misunderstood entry, because it looks like a demand measure and frequently is not.
Correcting power factor reduces apparent power, in kVA. Whether that reduces your bill depends on how your tariff treats it. If demand is billed in kW, correction changes the demand determinant not at all — the real power doing the work is unchanged, which is the whole point of the correction. If demand is billed in kVA, or if a power factor penalty or multiplier applies, correction does reduce the charge, sometimes substantially.
Same equipment, same physics, opposite economics, decided by a clause in the tariff. This is worth reading before any capacitor is ordered: power factor penalties explained.
The underlying point
None of this is an argument against efficiency. It is an argument for putting each measure on the correct line of the bill, because the bill has two lines and they are driven by different things.
Efficiency reduces the area under the load curve. Demand charges price its height. Measures that reduce the area without reducing the height save real money on the energy charge and nothing at all on the demand charge, and the whole distinction is set out in load shifting is not energy efficiency.
The practical discipline is to apply any proposed measure to the interval series and recompute the determinants from the modified series, rather than applying a percentage to a monthly total. It takes an hour, it produces a projection that survives contact with the next bill, and it is the same exercise that any credible savings claim will later be verified against: measurement and verification.
A projection that turns out to be right is worth more than a projection that is large. It is what buys the second project.
