The emails usually read the same way. Someone writes that their electricity bill has climbed again, asks whether a variable speed drive unit would fix it, then asks how many years until the machine pays for itself. Two questions, and the second one is the harder one.
Disclosure before anything else: we manufacture and sell both fixed-speed and variable speed rotary screw compressors, so we are not a neutral referee. What follows is aimed at oil-injected screw units in roughly the 11 to 37 kW band with load/unload control on the fixed-speed side. Piston, scroll and centrifugal machines follow different arithmetic. If your plant runs centrifugal units or you are sizing a brand new compressor room, much of this still holds conceptually, but the numbers will not transfer.
The short answer is that VSD pays off sometimes, and the deciding variable is not the price of the machine but how your air demand moves through the day.
What a VSD actually changes
A fixed-speed screw compressor has one working speed. When the discharge pressure climbs to the upper set point, it goes into unload: the inlet closes, the airend keeps turning, and the motor keeps drawing power. When pressure drops to the lower set point it loads again, and that cycle repeats all day.
A variable speed machine drops the motor speed instead. Less air being consumed means lower rpm and therefore less power going in, and there is no unload cycle in the traditional sense because the machine simply produces less air.
That is the mechanism. It is well documented, and there is a whole body of work on it under ISO 1217 for performance testing. If you want the technology itself explained at greater length, we covered the drive internals separately. The interesting question is not whether it works, but whether it works often enough at your plant to matter.
The power you pay for while nothing is being made
Here is the part that gets overlooked: a screw compressor in unload mode is not off, because the motor still spins and still pulls current. That draw typically lands somewhere between 20 and 40 percent of the full-load kilowatts, depending on the airend design, the discharge pressure, and how the unload relief is set. Twenty to forty percent is a wide band on purpose. Check it against your own machine rather than borrowing a number from a brochure, including ours.
Run the arithmetic on one machine. A 22 kW fixed-speed unit drawing roughly 6 kW while unloaded, working 4,000 hours a year with about half of that time in unload, burns close to 12,000 kWh a year producing no usable air. At USD 0.10 per kWh that is about USD 1,200 annually, and it repeats every year you keep the machine.
Working out the payback yourself
You do not need a consultant for this, only three numbers and a formula.
Annual saving, kilowatt-hours per year:
Saving = P × (1 − LF) × UL × H − P × LF × d × H
Where:
- P is full-load input power in kW, taken from the spec sheet or an ISO 1217 test report rather than the motor nameplate alone
- LF is load factor, your average air demand divided by the machine’s full output
- UL is unloaded draw as a fraction of full load. Use 0.30 if you have nothing better, and measure it properly later
- H is annual running hours
- d is drive loss, roughly 0.03 for VFD electronics
The first term is what you win: the unloaded hours that become partial-load hours. The second term is what you give back, because a drive sits between the supply and the motor and loses a little energy even at full output. Most brochure-level comparisons quietly leave the second term out, which is how you end up with savings estimates that never appear on a bill.
Then divide:
Payback years = price difference ÷ annual saving
You will notice we have not quoted a price difference. We are not going to invent one, because machine configuration, freight, duty and local certification all move it, and a number pulled from nowhere is worse than no number. The table below sidesteps that entirely. Work out your saving first, then compare it against whatever gap your suppliers quote.
Three duty profiles, three answers
All three rows use the same machine: 22 kW, UL of 0.30, drive loss 0.03, electricity at USD 0.10 per kWh. Only the load factor and the running hours change. These are worked examples, not measured plants. Swap in your own figures.
| Plant A | Plant B | Plant C | |
|---|---|---|---|
| Average load factor | 0.90 | 0.60 | 0.35 |
| Annual running hours | 6,000 | 6,000 | 4,000 |
| Unloaded hours recovered (kWh) | 3,960 | 15,840 | 17,160 |
| Drive loss given back (kWh) | 3,564 | 2,376 | 924 |
| Net annual saving (kWh) | 396 | 13,464 | 16,236 |
| Net saving at USD 0.10 | USD 40 | USD 1,346 | USD 1,624 |
Plant A is running too steadily for VSD to earn much. There are almost no unloaded hours left to recover, so the drive losses eat most of the gain. Plant B is the case where VSD earns its keep. Plant C saves more in absolute terms despite fewer hours, because it spends most of its life well below full output.
Reading the price gap
Once you have your annual saving, this tells you what a VSD premium is worth:
| Annual saving | 1-year payback needs a gap of | 3-year payback | 5-year payback |
|---|---|---|---|
| USD 1,600 | USD 1,600 | USD 4,800 | USD 8,000 |
| USD 1,350 | USD 1,350 | USD 4,050 | USD 6,750 |
| USD 800 | USD 800 | USD 2,400 | USD 4,000 |
| USD 300 | USD 300 | USD 900 | USD 1,500 |
Most industrial buyers we work with treat anything inside two years as a clear yes, and three years as a reasonable maybe. Past four years is usually a no, unless there is a second reason to change the machine anyway, such as noise limits or air quality requirements.
When VSD is the wrong answer
This is the section we wish more buyers would read before requesting a quote, because we have talked customers out of variable speed upgrades more than once.
Your load factor sits above roughly 0.85. Straight production lines, continuous processes, anything that never really backs off. There is very little unloaded time left to convert, and you would be paying for capability you cannot use.
Your demand swings between nothing and everything. A few hydraulic presses firing eight times an hour will not reward a VSD. The drive has to ramp the machine up every time, and a larger receiver tank plus sensible sequencing usually solves the same problem for less money. We have recommended exactly that over a VSD sale. It is the correct answer, so we say it.
You already run several compressors badly sequenced. Fixing how multiple machines hand off load to one another often beats replacing one of them with a variable speed unit, so sort the sequence control out first and reconsider afterwards.
Electricity is cheap where you are. Run the formula with your own tariff before deciding. In markets where industrial power sits well below USD 0.10 per kWh, the same saving stretches over many more years, and the money may do more good elsewhere.
The compressor room runs hot and dusty. Variable frequency drives are electronics. A room that sits above 40 °C or carries conductive dust will shorten their life considerably. The relevant standards here are the IEC 61800 series for adjustable speed drives and IEC 61000-3-2 on harmonic limits. Sort the ventilation and filtration before worrying about the drive, and check the drive’s own derating curve for high ambient temperatures.
What VSD will not fix
Changing the machine does nothing about the things that usually waste more energy than the control method ever will.
Leaks. A compressor running into a leaky distribution network wastes air continuously, and no control strategy recovers it.
Unnecessarily high discharge pressure. The common rule of thumb is that every 1 bar of set-point pressure you can give up saves around 6 to 8 percent of energy input. It varies with the application, but it is almost always cheaper to lower the set point than to change the machine.
Drying capacity that runs continuously. Refrigerated and desiccant dryers cycling with no flow are their own line item. We keep a full range of units in the air dryer category for this reason.
Poor pipe sizing and excess pressure drop. Undersized mains force you to raise discharge pressure to compensate, which pays a penalty twice over.
The broader efficiency picture is worth reading separately if you have not already. Our piece on permanent magnet versus conventional variable frequency drives covers the motor technology side, and there is a companion article on where variable frequency technology actually gets applied.
What to bring to your supplier
Anybody quoting you a VSD machine should be asking for these. If they are quoting without asking, treat the number with suspicion.
- Full-load input power in kW from a documented test, not a moter plate rating
- Measured unloaded draw, taken with a clamp meter over a few load/unload cycles
- Run hours and load hours from the controller. Most machines log both
- A demand picture. Even one week of logging beats a guess
- Your lowest usable pressure. What the most demanding tool or process actually needs at the point of use
- Your electricity tariff, including any peak and off-peak structure
Item 4 is the one most often skipped, and it is the one that decides the answer. A machine quoted against a load factor you guessed will perform exactly as badly as that guess.
A reasonable way to decide
Measure first. If your load factor is below about 0.75 and your plant runs more than roughly 3,000 hours a year, VSD probably pays. If those two conditions do not both hold, spend the money on leaks, set-point pressure, pipe sizing and sequencing before touching the compressor.
Then ask for two quotes instead of one. A fixed-speed proposal and a VSD proposal, same output, same pressure class, both with input power figures you can verify. The difference between them, set against the number the formula above gives you, is your payback. It is not a complicated decision once the inputs are real.
If you want a second opinion on your own figures, the screw compressor range is linked here for reference, and you can send your demand profile and tariff through our contact page. We will tell you when VSD is not worth it. That happens fairly often.
Reviewed by the SAYI application engineering team. Figures in these examples are illustrative inputs for the formula, not measured results from a specific installation. Verify against your own machine data and current local electricity tariffs. Last reviewed September 2026.