A VFD retrofit on a pump that trips in the morning and reads a noisy signal
A 7.5 kW, 415 V, four-pole, 1440 rpm three-phase centrifugal pump motor, previously direct-on-line started, has been retrofitted with a 7.5 kW variable frequency drive to save energy on a partial-load duty. Commissioning data: motor full-load current 14.5 A; the drive's motor overload parameter was set to 14.5 A; acceleration ramp 3 s; minimum frequency 25 Hz; maximum 50 Hz. The motor cable is 65 m of unshielded three-core armoured cable laid in the same perforated tray as the 4–20 mA cable from the discharge pressure transmitter. The drive panel and the motor frame are both bonded to the panel earth bar. After two weeks: the drive trips on overcurrent every morning when the operator opens the discharge valve fully before starting; the drive trips at random times when the pump is held near 30 Hz; the pressure signal wanders erratically, occasionally reading 3.2 mA; and the motor frame measures 78 °C after an hour at 30 Hz in a 34 °C ambient.
1. Explain the repeatable morning overcurrent trip and state the parameter and operating changes you would make.
Read the worked answer
Why it is repeatable at that moment. A centrifugal pump draws its least power at shut-off and its most power at high flow. Starting with the discharge valve already wide open puts the pump at the far right of its curve, so the required torque and current are high precisely when the drive is trying to accelerate from standstill. A 3 s ramp on a loaded pump forces the drive into its current limit almost immediately, and it trips to protect itself.
Parameter changes. (1) Enter the motor nameplate data properly - rated voltage, 14.5 A, 1440 rpm, power factor and rated frequency - so the drive's motor model is correct; a drive left on default data controls the wrong machine. (2) Extend the acceleration ramp to 10–15 s, which costs nothing at start-up and removes the current-limit excursion. (3) Set the overload protection to the motor's rated current with the correct service factor rather than a token figure, and confirm the current limit is set high enough to accelerate but not so high that it masks a mechanical fault. (4) Set a minimum frequency that keeps the pump above its thermally safe region.
Operating change. Start the pump with the discharge valve closed and open it slowly over ten to twenty seconds. That is standard centrifugal-pump practice, and the drive's ramp then has only the shut-off power to deal with. Write it on a laminated card beside the starter; the operator is not being careless, he was never told.
Do not 'cure' it by fitting a bigger drive or removing the trip. The trip is correct behaviour for a bad start sequence.
2. Account for the erratic 4–20 mA pressure reading and the random drive trips, and say what you would change in the installation.
Read the worked answer
The coupling path. A VFD output is not a sine wave; it is a pulse-width-modulated square wave switching at several kilohertz with fast edges. An unshielded motor cable radiates that switching energy, and 65 m of it running alongside a low-level analogue signal in the same tray will couple noise capacitively and inductively into the 4–20 mA loop. The symptom is exactly what is reported: a drifting value, an occasional impossible 3.2 mA (below the 4 mA live zero, so the transmitter or the input is being disturbed), with no fault in the transmitter itself.
The random trips are the same disturbance arriving as common-mode current on the earth system, upsetting the drive's own sensing.
Remedial measures, in order of cost-effectiveness. (1) Physically separate the signal cable from the motor cable - a minimum of 300 mm, and cross at right angles where they must meet. This alone often fixes it. (2) Replace the signal cable with shielded twisted pair and earth the shield at one end only, at the drive or PLC, to avoid a shield earth loop. (3) Bond the motor cable armour and any screen at both ends - at the drive gland and at the motor gland - because the screen's job here is to return high-frequency current, and a one-end bond leaves it to travel through the building earth instead. (4) Check the drive manual: 65 m of unshielded motor cable is around the length where makers recommend an output reactor or du/dt filter to limit reflected-wave overvoltage at the motor terminals. (5) Keep the 4–20 mA signal rather than converting to 0–10 V, which is far more noise-sensitive.
3. Is the 78 °C motor frame temperature at 30 Hz a fault? Analyse it and justify whether the pump may run continuously at that speed.
Read the worked answer
Where the heat comes from. A standard totally enclosed fan-cooled motor has its cooling fan on the shaft. At 30 Hz the shaft turns at 60 % speed, so the fan delivers roughly 60 % of its airflow and the cooling falls with it, while the motor still carries a substantial magnetising and load current. On top of that, the drive's PWM waveform adds harmonic losses in the winding and iron compared with a sinusoidal supply.
Is 78 °C a fault? Not necessarily. With class F insulation the winding is rated to a 155 °C total temperature, and a frame at 78 °C in a 34 °C ambient implies a rise of about 44 K, which is within the motor's design rise for many machines. The frame temperature is only an indicator; the winding hot-spot is higher and cannot be read from the outside. The real question is whether the drive's thermal model, loaded with correct nameplate data, is showing the motor near its limit - that is the number to trust.
Continuous duty at 30 Hz: the honest answer. Not on a standard motor without checking. Because a centrifugal pump's power falls with the cube of speed, the load at 30 Hz is only around 13 % of rated power, so the electrical stress is low; but for a duty that runs for hours at low speed the safe answer is to fit an inverter-duty motor with an independently driven cooling fan, or to raise the minimum frequency to a speed the motor can cool itself at. 30 Hz as an occasional operating point is fine; 30 Hz as the permanent duty needs the motor and the drive's thermal protection to be checked, and the decision recorded.