When VSD Is the Wrong Answer for Your Duty Cycle

The short answer: Variable-speed drives save energy by matching motor speed to demand, but those savings assume a plant that runs at 30–60% average load. For a continuous-process operation running 85–100% load — asphalt batch plants, plastics extrusion, blow-molding, full-bay machine shops, busy paint booths — the VSD’s 3–5% drive conversion loss actively costs more than the savings. The breakeven sits around 65–70% average duty. Above that, fixed-speed plus load/unload control plus a properly sized receiver beats VSD on 10-year total cost of ownership.

What VSD actually does

A variable-speed drive takes incoming AC power, rectifies it to DC, then inverts it back to AC at a variable frequency, which sets the motor’s RPM. Slow the motor and the airend turns more slowly; a slower airend pulls in less ambient air per second and delivers less compressed air. The trick is that the drive can adjust speed continuously to match plant demand — so at 50% demand, the motor draws roughly 50% of full-load kW, instead of the 70–80% a modulating-control fixed-speed would draw or the 30–40% an unloaded load/unload fixed-speed would draw idling between cycles.

The marketing claim is straightforward: at part load, VSD beats both fixed-speed control modes. Independent CAGI testing confirms this. The claim that needs auditing is the assumption that your plant operates at part load enough of the time to harvest those savings.

The lie hidden in the brochure

VSD brochures and ROI calculators almost universally assume a load profile that looks like this:

• 30% of operating hours at full load
• 40% at half load
• 30% at quarter load or idle

This is a reasonable approximation for general manufacturing — auto parts, electronics assembly, light fabrication — where compressed air supports tooling that doesn’t run continuously. But it is a complete mismatch for heavy-industrial duty. A plant that produces extruded plastic on a 24/7 line uses the same air every minute the line is running. An asphalt batch plant cycles between full-bore and idle, but “idle” is a 10-minute window between batches, not a steady part-load state.

For continuous-process loads, the realistic profile looks more like:

• 85% of operating hours at full load
• 10% at three-quarter load (changeovers, cleaning)
• 5% truly idle (shift change, breakdown)

At full load, the VSD drive itself dissipates 3–5% of input kW as heat — the rectifier-inverter conversion is never perfectly efficient. A fixed-speed motor running at full load has no such intermediate stage; the contactor is closed, full line voltage is on the windings, and the only losses are the motor’s own efficiency rating (typically 94–96% for a modern IE3-class motor). At a 90%-duty plant, the drive losses on the VSD outweigh the part-load savings.

The breakeven

$50K$100K$150K$200K 20%40%60%80%100% Average duty cycle (% of full load) 10-year total cost of ownership ($K) crossover ~70% VSD rotary screw Fixed-speed + smart unload

Plant average load	Annual kWh: fixed-speed load/unload	Annual kWh: VSD	VSD advantage
30%	140,000	92,000	+34% (VSD wins)
50%	180,000	140,000	+22% (VSD wins)
65%	215,000	200,000	+7% (close to even)
75%	240,000	240,000	0% (even)
90%	275,000	290,000	-5% (fixed-speed wins)

(Numbers above are for a typical 50 HP screw running 4,000 hr/yr at 125 PSI, normalized to a CAGI-verified specific-power curve. Real plants vary ±15% based on airend efficiency, motor class, and load smoothness.)

The crossover for this class of compressor is around 70% average load. Below 70%, VSD wins on energy. Above 70%, fixed-speed with properly-tuned load/unload wins.

The other costs that aren’t on the brochure

• Drive maintenance. Power-electronics modules have a service life of roughly 80,000–100,000 hours under typical industrial conditions. Capacitor replacement is a scheduled item (every 5–7 years). IGBT failure is uncommon but expensive when it happens ($2,500–$8,000 module replacement). A fixed-speed motor and starter have effectively no scheduled maintenance beyond bearings and contactor inspection.
• Harmonic distortion. Six-pulse drives inject 5th and 7th harmonics back into the plant feeder. If the plant’s transformer is undersized or the upstream switchgear isn’t compliant with IEEE 519, harmonic filters or line reactors are required at additional cost.
• Drive cooling. Drives are derated above ~40 °C ambient and require active cooling — fans on the heat sink, often a separate cabinet HVAC system. The cooling load is itself a parasitic energy cost.
• Bearing fluting. High-frequency switching in the drive induces shaft currents that pit motor bearings (“electrical bearing fluting”). Shaft-grounding rings and insulated bearings are sometimes specified to mitigate, adding ~$500–$1,500 to motor cost.

None of these costs are dealbreakers — VSD has been a mature technology for two decades — but they shrink the headline ROI compared to brochure projections, especially at high duty.

When VSD is the wrong answer

VSD is the wrong tool when one or more of these is true:

• Plant load is above 70% average across operating hours.
• Demand is binary, not variable. If your air load is either “full bore” or “off entirely” (asphalt batch, blow-molding) the smooth modulation a VSD enables has no value — load/unload is the right architecture, with the motor stopping fully between batches.
• The plant already has a master controller sequencing multiple fixed-speed compressors. A four-unit cascade can deliver part-load efficiency without VSD by staging the right number of units online for the current demand. See the master-controller guide.
• The plant feeder is borderline on transformer or switchgear capacity. Adding harmonic-generating equipment to an already-stressed feeder is asking for trouble; the harmonic mitigation cost can exceed the VSD savings.
• The compressor is below ~15 HP. Below that size, the drive cost as a percentage of the compressor cost is high enough that simple payback rarely beats five years even at low duty.

What to do instead

For high-duty heavy-industrial plants, the energy-optimization stack looks different than the VSD-default playbook:

1. Right-size the compressor for full-load duty. A 50 HP fixed-speed running at 85% is more efficient than a 75 HP VSD running at 55%.
2. Oversize the receiver. 8–10 gal/CFM, not the rule-of-thumb 4 gal/CFM, for any plant where load/unload control is the primary mode. The larger receiver lets the compressor reach its auto-dual timeout and shut the motor off entirely between batches — eliminating unload running.
3. Tune cut-in/cut-out spread. Wider band (15–20 PSI) for plants that tolerate pressure variability; narrow band (5–10 PSI) only where the process demands it.
4. Audit plant pressure. Every 2 PSI above the actual tool requirement is ~1% of full-load kW wasted. Most plants are 10–30 PSI above what they actually need.
5. Consider supplemental hardware for unload running. Where the receiver and tuning playbook is exhausted, an aftermarket unloader controller intervenes on the unload cycle directly — a different lever than VSD, addressing a different loss.

Real-world examples from the catalog

Among the rotary screws we list, the Kaeser CSD 75 (461 SCFM at 110 PSI) ships as a fixed-speed unit by default — the CSDX SFC is the variable-speed variant, sold separately for plants where the load profile justifies it. Atlas Copco’s GA-series follows the same model: the GA5-125 and GA7-125 are fixed-speed, with the GA+ VSD+ packages available as an upgrade. Ingersoll Rand’s R-Series — the R5.5i-125 through R7.5i-125 — is fixed-speed; the Nirvana and Sierra lines are the VSD-capable big siblings. The Quincy QGS-7.5 and Quincy QGS-10are fixed-speed; the QGV-series adds VSD. In each case the manufacturer makes both available because the right answer depends on the plant’s load profile, not on the badge.

Common questions

How do I know my plant’s actual average load?

If your compressor has a Modbus / Ethernet / Profibus link to the controller, the controller logs cumulative loaded-hours and unloaded-hours. The ratio of loaded to total is your average load. If you don’t have that logging, the field method is to install a current-clamp logger on the motor for one to two weeks and post-process: minutes-loaded / total-minutes = load. Anything above 70% should give you pause before specifying VSD.

Isn’t VSD always more efficient at part load?

Always more efficient than modulating, yes. Compared to load/unload with a properly sized receiver and an auto-dual timer, the gap is much smaller — and disappears entirely above 70% duty.

What if I have variable load but expect to grow into full load?

Then VSD is hedging both ways: efficient now at part load, less efficient later at full load. The honest tradeoff is to project years 1–5 separately, weight by likely operating hours, and run an NPV. Don’t pick VSD on the assumption you’ll stay at part load forever if your business plan says otherwise.

Is a soft-starter the same as a VSD?

No. A soft-starter ramps motor voltage up over a few seconds at startup, then steps out of the circuit entirely — once the motor is up to speed, line voltage is direct, no conversion loss, no part-load modulation. Soft-starters cost ~10% of a VSD and address only the start-current issue, not the part-load energy issue.

If you want to dig further on the controls side

If the duty math points away from VSD but the unload-running cost is still galling, the question becomes build or buy. The aftermarket path is a purpose-built unloader controller — a small device between the motor contactor and the unloading valve that drops the motor sooner than the OEM timer would. We describe one implementation in a sister project. The engineer-built path is a PID or threshold loop on a PLC or microcontroller, watching discharge pressure or motor current and intervening on the unload contactor directly. Both lever the same loss; the hardware path is faster to commission and pre-tuned, the DIY path is cheaper if in-house controls staff can absorb the tuning cycle.

Compressor Controller