Motor Winding Failures and Nuisance Trips on a 250 kW PWM Drive with a Long Cable and a Nearby Capacitor Bank
A 415 V, 250 kW, 4-pole cage motor on a rolling mill is fed by a 6-pulse PWM voltage-source drive with a diode front end, default 4 kHz carrier and a 400 V output. The motor is 240 m away, connected by three single-core 95 mm2 cables in trefoil; the drive has no output reactor. After 14 months, two motor winding failures have occurred in the first two turns near the terminals. The drive trips on overcurrent at about 40 Hz during acceleration, and the DC-link overvoltage trip operates whenever the mill stops quickly. A 600 kVAR capacitor bank is connected to the same 415 V board, its fuses have blown twice, and the plant measures 9% fifth-harmonic current on the board. The 1600 kVA supply transformer has 5% impedance. Ambient in the drive room is 45 degC and the site altitude is 600 m.
1. Explain the winding failures using cable length and switching rise time, and state what you would check and specify.
Read the worked answer
Travel time. Wave propagation in a PVC/XLPE cable is roughly 5 ns/m, so one-way transit is 240 x 5 = 1.2 microseconds and the round trip is about 2.4 microseconds. An IGBT edge of 0.1-0.2 microseconds is much faster than the round trip, so the arriving wave is fully reflected at the high motor surge impedance and the terminal voltage approaches twice the DC-link voltage: 2 x 565 = about 1130 V, with ringing on top. Every switching edge applies this to the first turns, which carry the highest inter-turn stress - exactly where the two failures occurred.
What this is not. It is not a drive firmware defect, and it is not caused by the overcurrent trips; the failures are a voltage-stress phenomenon and would have occurred even with perfect current limiting.
Checks. Confirm the motor is inverter-duty to IEC 60034-25 and qualified to IEC 60034-18-41 for the impulse level at this cable length; measure the actual terminal voltage with a differential probe and a suitably rated divider, not with an ordinary meter.
Specification. Fit an output dv/dt filter (preferred at 240 m) or at minimum a 3% output reactor, and confirm the peak and rise time afterwards. If a long cable must be retained, a sine-wave filter removes both the stress and the cable charging current that contributes to the overcurrent trips.
2. Assess the capacitor bank and harmonic condition. Is the bank at risk, and what exactly should be installed?
Read the worked answer
Short-circuit level at 415 V. Ssc = 1.6/0.05 = 32 MVA.
Resonance. h = sqrt(Ssc/Qc) = sqrt(32/0.6) = sqrt(53.3) = 7.3. The parallel resonance sits just above the 7th harmonic, and with realistic damping the amplification band covers the 5th as well. A 6-pulse diode front end injects 5th, 7th, 11th and 13th currents (the 5th being the largest), so the bank is being asked to absorb amplified harmonic current and its fuses are acting as the weak link - and once one fuse clears, the remaining units are overloaded and fail in cascade.
Remedy. Install a detuned reactor with 7% reactance on the bank: tuning order = sqrt(1/0.07) = 3.78 (about 189 Hz). The capacitor voltage then rises to 415/0.93 = 446 V, so the units must be rated at 440 V or 480 V, not 415 V; a 415 V-rated bank will overheat and fail even with the reactor. Recheck the fuse rating against the increased fundamental current.
Source-side measures. Ask for a DC-link choke (typically reduces current distortion by roughly a third) or a 12-pulse/active front end on the drive. An active filter is justified only if other drives on the board also inject harmonics; a single 250 kW drive rarely justifies one on its own. Verify the result with a 7-day harmonic survey before and after.
3. Explain the overcurrent trip near 40 Hz and the DC-link overvoltage on fast stop, and decide whether the drive frame size is adequate for the site conditions.
Read the worked answer
Overcurrent at 40 Hz. Below base speed the drive is in constant V/f (constant torque) control. If the connected load torque rises with speed and the acceleration ramp is short, the drive reaches its current limit at the speed where load plus accelerating torque equals the motor capability. 40 Hz suggests a mechanical cause - a worn mill bearing, product jammed in the stand, or a load that needs more torque than the motor can give at that point - rather than an electrical fault. Check motor current waveform balance, verify the V/f boost setting, lengthen the ramp, and inspect the mechanical drive train. Note that the long cable's charging current also adds to the measured output current and can push the drive into current limit prematurely; the output reactor helps here.
DC-link overvoltage. A fast stop with a high-inertia mill makes the motor regenerate; with a diode front end there is no path for that energy, so the DC bus rises until the trip operates. Fit a brake chopper and resistor sized for the inertia, or extend the deceleration ramp within the drive's braking capability. Do not simply increase the trip threshold.
Frame adequacy. Typical drives are rated at 40 degC ambient; at 45 degC a derating of about 1% per degC applies, i.e. 5%, so the 250 kW rating falls to about 237 kW for a 250 kW motor - marginal at best. Altitude at 600 m is not a constraint (derating generally starts above 1000 m). Specify a 315 kW frame and verify with the manufacturer's derating curves for 45 degC and 4 kHz.