Air Compressor Motor Power and Current Relationships

I = P / (√3 × U × cosφ × η). At 380V, cosφ 0.87, η 0.92, a 22 kW motor comes out to about 42A. A YE3-180M-4 nameplate says 41.8A.

Below about 40% load, both cosφ and η fall off a cliff and the formula becomes useless. At 30 Hz on a VFD compressor the power factor might be 0.62, might be 0.71, depends on the motor design and magnetizing current. Don't bother plugging partial-load numbers into this formula. Just clamp the cable and read what the meter says.

Quick Reference Table

Cable Sizing

This is the topic that causes the most grief on compressor installations, so it gets the most space here. Breakers and VFDs are comparatively simple to get right. Cables are where corners get cut because copper is expensive and conduit routing is a pain.

IEC 60364-5-52 Table B.52.4 gives the reference ampacities. For single-core PVC copper in free air at 30°C: 35 mm² is 126A (not 130A, the round number that gets passed around), 50 mm² is 153A, 70 mm² is 196A, 95 mm² is 238A. XLPE insulation bumps these numbers up by about 20%. GB 50055-2011 section 4.6 says motor feeder cable ampacity shall be no less than 1.25 times motor rated current. So that's the floor.

Now put that cable in a real installation. IEC 60364-5-52 Table B.52.17 gives grouping correction factors. Three circuits in one conduit: correction factor 0.70. Four circuits: 0.65. That 50 mm² at 153A in free air becomes 153 × 0.70 = 107A with two other circuits in the same conduit. A 55 kW motor at 103A technically passes at 107A. Four amps of margin. In a 30°C environment.

In reality, compressor rooms in Guangzhou or Chennai or Monterrey hit 42-45°C for four months of the year. The temperature correction factor at 45°C for PVC is 0.87 (IEC 60364-5-52 Table B.52.14). So 107A × 0.87 = 93A. The motor needs 103A. The cable is 10% undersized.

Going to 70 mm² gives 196A in free air, 196 × 0.70 × 0.87 = 119A after derating. 16A of margin over 103A. That holds up. The price difference between 100 meters of 50 mm² and 70 mm² four-core is maybe ¥3,000-4,500 depending on the supplier and copper prices that month. That's nothing compared to the cost of pulling the cable out and replacing it after a thermal failure.

XLPE insulation changes the math. Same 50 mm² cable but XLPE instead of PVC: free air rating is about 185A. After grouping and temperature derating: 185 × 0.70 × 0.94 (XLPE correction at 45°C is better than PVC) = 122A. Passes with margin. XLPE costs 15-25% more per meter than PVC but has higher ampacity and better thermal resistance. For compressor rooms that are consistently hot, XLPE cable on a 50 mm² cross-section can be cheaper than PVC cable on a 70 mm² cross-section because the copper weight is less.

Voltage drop. The standard says under 5% at the motor terminals. Per IEC 60364-5-52 Annex G, or more practically, by the resistivity method: copper resistivity at operating temperature (let's say 70°C for a loaded cable) is about 0.0213 Ω·mm²/m. A 50 mm² cable carrying 103A over 120 meters, one way, voltage drop per phase is I × 2L × ρ / A = 103 × 240 × 0.0213 / 50 = 10.5V. As a percentage of 220V (phase voltage): 4.8%. On the edge. And that's running current. During star-delta starting at 3x rated, the drop triples. Motor terminal voltage during starting drops to maybe 345V line-to-line. Starting torque drops with the square of voltage: (345/380)² = 82% of normal starting torque. On an Atlas Copco GA55 or a Sullair ShopTek ST55, the airend has low breakaway torque when warm, so 82% is usually enough. On a cold start with cold oil, breakaway torque goes up and 82% might be marginal.

A parallel cable run solves the problem without pulling new cable. Two 35 mm² cables give roughly 70 mm² effective cross-section and half the voltage drop of a single 35 mm² run. Each cable carries about half the current. The installation requires a junction box at each end to combine the two cables onto the terminal. This is common on retrofit jobs where the existing conduit is full and a second conduit is being run on a cable tray or in a new trench.

Circuit Breakers

1.2-1.3 times motor rated current, round up. D-type for motor feeders. Size to protect the cable and motor, not to survive the starting surge. An MCCB like an ABB Tmax XT3 160A on a 55 kW motor at 103A provides 1.55x margin on the thermal trip, which is within the acceptable range per GB 50055-2011 section 4.3. A Schneider NSX160F at 160A frame with a 125A thermal magnetic trip unit (TM125D) would be tighter, at 1.21x, and that's fine too. Either one needs D-curve or motor-rated trip characteristic.

Oversizing breakers is common in old plants and it means the motor has no overcurrent protection. A 250A breaker on an 85A motor, which is a configuration that exists in more facilities than anyone would like to admit, won't trip until the motor is drawing 200A+ sustained. The motor stalls at 120A. The winding burns. The breaker doesn't know.

VFD Sizing

Match to output current, not kW label. The kW label is a marketing number.

Specific example: Danfoss FC302 series. The 55 kW model (P55KT4E20H2) has a rated output current of 106A in normal duty and 90A in high overload duty. Siemens G120 PM240-2 at 55 kW does 110A. ABB ACS580-01-106A at 55 kW does 106A. Yaskawa GA700 at 55 kW does 112A. So "55 kW" gives you anything from 106A to 112A depending on the manufacturer.

Going up one frame: the Danfoss FC302 75 kW (P75KT4E20H2) does 147A. That's 43% margin over the motor's 103A. It'll never derate into trouble. The price difference between the 55 kW and 75 kW FC302 is somewhere around $800-1,200 depending on the distributor and the region. A single production shutdown from a VFD overcurrent trip costs more than that.

For compressor applications specifically, the VFD needs heavy-duty overload capability. The Danfoss FC302 in high overload (HO) mode gives 160% for 60 seconds and 180% for 0.5 seconds. In normal duty (ND) mode it's 110% for 60 seconds. Compressor loading needs HO mode because the pressure spike when the solenoid opens sends a torque transient through the airend. If the VFD is configured for ND mode (which is the default out of the box on most drives), the first loading event after startup may trigger an overcurrent trip. This catches a lot of people. The compressor starts fine in unloaded mode, runs for a minute, the controller commands a load, the solenoid opens, pressure slams the rotors, current jumps to 135%, and the VFD throws fault code A-81 (overcurrent, Danfoss) or F0001 (Siemens) or OC-1 (Yaskawa). Change the application mode from ND to HO in the VFD parameters and it goes away.

There's a whole separate issue with VFD output cables that most compressor installers don't know about because it's a power electronics problem, not a compressed air problem. The VFD output is a PWM waveform. Voltage pulses with rise times in the hundreds of nanoseconds on modern IGBT drives, sometimes under 100 ns on SiC-based drives. When a fast pulse travels down a cable, the cable has characteristic impedance determined by its geometry and insulation. The motor has a different impedance. At the impedance boundary, part of the pulse reflects. On a 380V system the DC bus is about 540V. The reflected pulse adds to the incident pulse at the motor terminals. Peak voltage can hit 2x bus voltage: 1080V. The Siemens motor catalog (catalog D 81.1, section on converter-fed operation) rates standard 1LE1 motors for 850V peak at the terminals, or 1300V peak with reinforced insulation (VFD-duty option, add suffix "-8VY" to the order code). So a standard Siemens motor on a long cable from a VFD is seeing voltage spikes 27% above its rating. That erodes insulation.

Below 50 meters of cable, the pulse propagation time is short enough that the reflected wave doesn't build up to full amplitude. 50-100 meters, a dV/dt reactor at the VFD output (Danfoss part number 130B1066 for 55 kW, Schaffner FN 5040 series, or TCI KDR series) slows the pulse edges to 2-5 μs and limits the peak to about 800V. Above 100 meters, a sine wave filter converts the PWM output to a near-sinusoidal waveform. Motor terminal voltage stays below 450V peak. A 55 kW sine wave filter from Schaffner (FN 5020 series) or Danfoss (130B1268) costs in the neighborhood of $1,500-2,200.

VFD-duty motors (Siemens 1LE1 with -8VY suffix, ABB M3BP with Option 413, WEG W22 Inverter Duty) have reinforced insulation rated for 1300V peak and phase insulation paper between the first turns. They can handle long cable runs without a reactor up to maybe 100-150 meters depending on the drive's switching frequency and IGBT characteristics. They cost 10-20% more than the standard frame.

Using Current for Diagnosis

Discharge pressure set too high is the overwhelming majority of high-current complaints on compressor service calls. A machine rated 0.8 MPa / 8 bar running at 10 bar overloads the motor. The operator set it there because the production floor is getting 5.5 bar instead of the 6 bar they need, and instead of fixing the distribution piping (undersized 2" header running 180 meters with twelve tee takeoffs, a strainer that's been in service since 2015 without cleaning, and four 90° elbows where 45° long-radius bends should have been used), they cranked the compressor up 2 bar. The compressor draws 112% rated current. The overcurrent relay trips. The maintenance electrician raises the relay from 105% to 115% to stop the trips. Now the motor runs continuously at 112% with no protection. The winding runs 20-25°C above its thermal class rating. Insulation class F is rated for 155°C. At 175-180°C the insulation degradation rate is roughly 4x the rate at 155°C.

Oil separator differential pressure. New element: 0.02-0.05 MPa. Service limit: 0.1 MPa. Every compressor manufacturer publishes this in their O&M manual. Atlas Copco GA series calls for element replacement at 1.0 bar (0.1 MPa) differential. Ingersoll Rand R-series says 15 psi (about 0.103 MPa). The differential gauge or indicator is on the control panel of any machine built after about 2005. Older machines might not have one, and the operator has no way to know the element is plugged until the motor trips on overcurrent.

Running an element past 0.1 MPa differential to stretch its service life is a false economy. At 0.15 MPa the motor pulls maybe 5-8% more current. At 0.2 MPa it's 10-12% more. On a 55 kW machine running 6000 hours a year, 8% more power consumption is 4.4 kW × 6000 hours = 26,400 kWh. At the industrial electricity rate in eastern China (about ¥0.65/kWh) that's ¥17,160 per year. An Atlas Copco DD/PD filter element for a GA55 costs about ¥1,200-1,800. The element saves its own cost in electricity in about five weeks. The extra three months of "savings" from running a plugged element costs ¥4,000-5,000 in electricity alone, not counting the accelerated bearing wear from higher airend back pressure.

Bearing wear in screw compressor airends is gradual. Rotor-to-housing clearance is 0.04-0.08 mm depending on the airend manufacturer and size. The clearance is set at manufacture by the bearing preload and the machining tolerances on the rotor and housing bores. As bearings wear, the rotors shift position. The leakage path between the high-pressure and low-pressure sides of the compression chamber gets larger. Internal recirculation increases. The airend needs more input power to produce the same output pressure and flow. Current creeps up, maybe 0.5-1% per month once the wear becomes measurable. The vibration level rises at the same time because the rotor dynamics change. The discharge temperature goes up because more internal leakage means more re-compression of already-hot gas.

Low current with adequate pressure is just light load. Not a fault.

Low current with poor pressure: intake valve problem. The intake valve on an Atlas Copco GA is a butterfly valve actuated by a pneumatic piston. Control air comes from the compressor's own internal pressure. If the control air solenoid (Atlas Copco part 1089 0702 12 on GA30-90 series) doesn't energize, or the piston seal is leaking, or the control air tube has cracked, the butterfly stays closed. The motor spins, draws no-load current, and produces very little air. The controller sees pressure not rising and may try to load again and again, with the intake valve ignoring it each time. This gets logged as a "motor running but no pressure" complaint.

Clogged air filter. Same symptom, milder. Current drops a few percent, capacity drops noticeably, loading time increases. The filter differential indicator (a pop-up pin on most Atlas Copco and Ingersoll Rand machines, or a gauge on Kaeser) shows the restriction. Element replacement on a GA55 is a ¥150-300 part and fifteen minutes of labor. There's no reason to ever run a visibly restricted air filter.

Supply voltage. Below 360V on a 380V system, current goes up noticeably. The motor draws more amps to maintain shaft power at reduced voltage. At 350V, current is up roughly 8-9% compared to the same load at 380V. Below 342V (10% undervoltage, which is the limit in most motor specifications per IEC 60034-1 clause 7.3), the motor should be shut down. In practice, industrial parks in developing areas see sustained voltage of 355-365V during peak hours, especially in summer when everyone's air conditioning and compressor loads coincide. Voltage stabilizers or tap-changing transformers on the incoming supply address this. Or reduce discharge pressure to bring the motor load down enough that the current stays within rating at reduced voltage.

Cooling degradation. Blocked cooler, failed fan, ambient above 40°C, direct sun on the unit. Motor winding temperature rises. Copper resistance rises at about 0.393% per °C. More resistance means more I²R losses in the winding, which raises temperature further. On a 55 kW motor the phase resistance at 20°C might be 0.082Ω. At 155°C (class F limit) it's about 0.125Ω. The extra resistance at elevated temperature increases copper losses by roughly 50% compared to cold. This shows up as higher current draw for the same mechanical output. Seasonal variation: the same compressor, same load, same pressure, might draw 101A in January and 108A in August. If 108A exceeds the overcurrent relay setting, the machine trips on hot afternoons. The relay didn't drift. The cooling deteriorated or the ambient changed.

Starting Methods

The star-delta switchover deserves more detail. The motor disconnects for 50-80 ms during contactor changeover. When delta reconnects, the phase angle mismatch between supply voltage and the motor's back-EMF determines how bad the transient is. Bad luck on the phase angle and the transient spike is as bad as a direct start. Good luck and it's barely visible on the ammeter.

Timer setting for the switchover: empirical. Start around 6-8 seconds for a 30 kW unloaded compressor. Adjust based on the ammeter reading during switchover. If the spike is harsh, add time. If the motor sits at near-synchronous speed in star for several seconds burning energy and not accelerating, shorten it. The optimal setting is specific to the motor, the load inertia, and the supply impedance, which is why no manual gives a universal number.

Closed-transition star-delta (Siemens calls this the "resistor transition" method, catalog section on motor starting) adds a set of resistors and a fourth contactor that bridges the open interval. The motor is never disconnected from the supply. Transient is mild regardless of phase angle. More expensive, more contactors to maintain, more wiring. Used on applications where the open-transition transient is unacceptable and VFD or soft starter is not in the budget.

Soft starter details: an ABB PSTX or Siemens 3RW55 on a 55 kW motor dissipates about 500-600W through the thyristors during running. With the bypass contactor engaged, zero. The bypass closes 5-10 seconds after the motor reaches rated speed. If the bypass contactor fails (coil burns out, auxiliary contact feedback doesn't confirm closure), the soft starter stays on thyristors and dumps that 500W as heat into the MCC panel indefinitely. Panel temperature climbs. Other equipment in the same panel drifts above rating. This is one of those failure modes that doesn't cause an immediate trip but degrades everything in the panel over months.

Power Supply Capacity

Three compressors: 55 + 37 + 22 = 114 kW. At cosφ 0.87, η 0.92, simultaneity 0.8: 178 kVA. A 200 kVA transformer covers it on paper. A 250 kVA covers it in reality. A 315 kVA covers it plus the 37 kW machine that gets added in three years when the production line expands.

Capacitor bank for power factor correction. Bringing cosφ from 0.87 to 0.95 frees up 9% of transformer capacity and eliminates the utility surcharge for low power factor. In China the surcharge is 0.5% of the electricity bill for every 0.01 below cosφ 0.90, per the State Grid tariff structure. At cosφ 0.87 that's a 1.5% surcharge. On a factory electricity bill of ¥200,000/month that's ¥3,000/month or ¥36,000/year. A 50 kvar capacitor panel costs ¥8,000-15,000 installed. Payback period: three to six months.

Harmonics from VFDs. A 6-pulse VFD rectifier generates 5th and 7th harmonic current primarily. With three VFD compressors on one transformer, the total harmonic distortion on the bus can reach 12-15% THDv, depending on the transformer impedance and the ratio of VFD load to total load. IEEE 519-2022 recommends THDv below 5% at the PCC for general systems and below 8% for dedicated systems. At 12-15% the transformer heats significantly above its nameplate loading because eddy current losses increase with the square of frequency, and 5th harmonic at 250 Hz generates 25 times the eddy current loss of fundamental at 50 Hz. The transformer can be thermally overloaded at 70% of its nameplate kVA rating.

Line reactors on VFD inputs, 3% impedance, reduce the 5th harmonic from about 25-30% of fundamental to about 10-12%. A 55 kW 3% line reactor from Schaffner or TCI costs $400-700. A K-rated transformer designed for harmonic loads costs 15-25% more than a standard transformer of the same kVA.