45 kW pump drive tripping at 40 Hz after an impeller change
A 45 kW, 415 V, four-pole IE3 motor drives a cooling-water pump through a voltage-source PWM drive rated 90 A at 40 °C ambient. The motor nameplate full-load current is 82 A. The motor cable is 65 m of 4-core 35 mm2 unshielded cable in a tray shared with control and instrumentation cables, with no output reactor or dv/dt filter. Carrier frequency is 8 kHz, the acceleration ramp is 10 s and the V/f curve is linear. The drive trips on overcurrent consistently at about 40 Hz, but only when the pump is started with the discharge valve partly open; it runs at 35 Hz and at full speed without tripping. The trips began after a new impeller was fitted. Motor insulation resistance is 500 Mohm and winding resistances are balanced. The motor frame is earthed at the motor end only, and the gland plate at the drive end is not bonded. The panel filter is clogged and the panel interior measures 50 °C.
1. Separate the electrical from the mechanical evidence. What would you ask the operator, and what test would you run first to decide which side the fault is on?
Read the worked answer
First question: what does the drive display say? The trip code is the single most useful piece of information.
- OC (instantaneous overcurrent, typically 200%+ within microseconds) points to a short circuit, earth fault, cable flashover or a very fast current spike.
- OL / thermal overload (110–150% sustained for seconds to a minute) points to genuine load torque or a derated drive.
Mechanical evidence is strong: the trip is reproducible at one speed band, depends on valve position (part-open = far from best efficiency point, with recirculation and possible cavitation), and began immediately after an impeller change. That is a load-torque signature, not a cable signature. Cable and dv/dt problems give intermittent, load-position-independent trips.
Decisive test: run the pump in local/hand mode with the discharge valve closed and sweep slowly through 40 Hz while logging drive output current, torque and DC-bus current. If the current stays near the running value and no trip occurs, the fault is hydraulic. Then repeat with the valve partly open and record the current at each frequency; a sharp hump around 40 Hz confirms a torque peak or resonance. Cross-check motor current with a true-RMS meter rated for PWM output, or trust the drive's own calibrated readout.
Do not start by changing drive parameters - establish the mechanism first.
2. Assess the cable, earthing and carrier-frequency arrangements. Are they the cause of this trip, and what would you change regardless?
Read the worked answer
Are they the cause? Probably not - but they are real defects.
Long unshielded cable. 65 m of unshielded cable on a 400 V PWM drive is beyond normal practice (about 50 m is the usual ceiling without mitigation). Effects: high cable charging current that the drive must supply, steep dv/dt at the motor terminals with reflected-wave overvoltage approaching twice the DC-bus voltage, increased common-mode current through the motor bearings, and radiated noise into the adjacent control and instrumentation cables in the shared tray. With an 8 kHz carrier the charging current and drive losses are higher than at 4 kHz.
Earthing. Motor frame earthed only at the motor end, gland plate unbonded at the drive end, is poor high-frequency practice: the common-mode return path is long and inductive, which raises noise, causes erratic earth-fault and overcurrent trips, and accelerates bearing erosion.
Why they do not explain this symptom: they produce random trips unaffected by valve position and not repeatable at one frequency. This drive trips at 40 Hz with the valve part-open, every time.
Changes to make regardless: reduce carrier frequency from 8 kHz to 4 kHz; fit an output load reactor or dv/dt filter (a dv/dt filter if the motor is not inverter-duty rated); bond the motor frame to the drive protective earth with a separate conductor run alongside the power cores; bond the gland plate at 360°; ideally replace with shielded cable and route it away from the instrumentation tray.
3. The panel is at 50 °C with a clogged filter. Quantify how this changes the drive's capability, and set out the corrective actions before the pump is returned to service.
Read the worked answer
Derating arithmetic. The drive is rated 90 A at 40 °C, and a typical figure is 1.5% per kelvin above 40 °C:
- At 50 °C internal: 90 × (1 − 0.015 × 10) = 90 × 0.85 = 76.5 A.
- Motor full-load current is 82 A.
So the drive's usable current is now below the motor's rated current. Wherever the load torque peaks - which, on the evidence, is around 40 Hz with the valve part-open - the drive reaches its limit before the motor reaches nameplate current. That is enough on its own to produce an overload trip in that speed band, and it explains why the symptom appeared now, when the impeller change altered the torque curve, rather than earlier.
Corrective actions, in order:
- Clean or replace the panel filter and restore airflow; re-measure panel internal temperature. Consider a larger enclosure or forced ventilation if it cannot be held near 40 °C.
- If the site cannot hold 40 °C, apply the manufacturer's derating table formally and either reduce the continuous load or fit the next drive size up.
- Resolve the hydraulics: start with the discharge valve closed and open it above about 45 Hz, or set a skip band of 37–43 Hz so the drive accelerates through the problem region, or trim the impeller.
- Check for cavitation (noise, fluctuating current, NPSH margin) after the impeller change.
- Then implement the cable and earthing improvements from the previous step.
Verify by a full current-versus-speed sweep from standstill to full speed with the valve in its normal starting position, confirming the current stays within the derated drive rating throughout.