Starting induction motor from non-zero speed

Hi all,
i need to start an induction motor from non-zero speed, without any jerk torque. The main question is how to find initial frequency of the voltage that has to be injected.
Any comment and suggestion is appreciated.

Thank you.

Hello, I'm really interested in this kind of situation. I have built a speed adjustable control for a single phase induction motor, using H-bridge drive, dc/dc converter, an pic controller. Meanwhile, the bootstrap Diode just fail and the mosfet driver go bad. I hope to replace it soon. However, which method are you using in your design?

I am working on V/F control of 3-phase induction motor. i have developed a 10KW drive. now, adding features. Afterward I will work on DTC.

The only way to do what are you asking for (a smooth start) is to implement a vector control.
The problems are the difference of the rotor's resistance, which you should estimate and measurement of the actual speed in such low conditions.

According to induction motor characteristic, there's a specific slip frequency to achieve maximum torque. Ideally you'll start the VFD with this frequency.



Industrial VFDs have often means to estimate the actual rotor speed and control the field frequency respectively.

- - - Updated - - -

The only way to do what are you asking for (a smooth start) is to implement a vector control.

Click to expand...

Not necessarily. True vector control is surely preferred and tt's absolutely required for synchronous machines. For asynchronous machines and standard load cases, V/f control (possibly with some auxiliary functions) is usually sufficient.

Hi all,
thank you for your reply.
I thought about vector control but in such condition I doubt its performance!estimation flux position is very troublesome in this condition. the voltage induced to stator by residual magnetic is high enough to make good estimation? I think auxiliary function would work as well.

The question is how can I estimate rotor speed?

The drawn curve is the torque-speed of an induction motor in specific voltage and frequency. This curve changes by changing frequency and voltage. This is why we can have maximum torque in any speed by VFD.

The problem is I wanna start a motor that is rotating already. and speed estimation is my major problem.
I would be grateful if you give me hint about this issue.

kappa_am said:

The problem is I wanna start a motor that is rotating already. and speed estimation is my major problem.

Click to expand...

I assume you have at least DC voltage measurement and perhaps phase current measurement too. Also, motor has been coasting (=no inverter modulation) so long that motor back emf has reduced to a small value. Let's say motor has been coasting at least three rotor time constants before a new starting attempt.

Idea is to supply voltage to the motor with decreasing frequency starting from the high frequency. Voltage is your normal three phase voltage that I guess is something like sinusoidal pwm on voltage source inverter. For example on 50 Hz motor you can start on 50 Hz. Most likely free wheeling frequency will be lower than that. I would use a constant voltage amplitude instead of voltage according to v/f ratio. 10-30% of nominal voltage is something to start with. Motor current is small as long as frequency of the voltage you supply is higher than motor rotating frequency. When supply frequency reach rotating frequency, current increases and you should detect it as DC voltage dip or in current measurement.

Change to v/f mode after detection. Sometimes it is necessary to increase frequency a bit to avoid regenerating mode especially if there is no braking resistor or the supply is a diode rectifier. Perhaps 1/5 -1/3 of nominal slip frequency added to detected frequency is enough to ensure that the motor starts in motoring mode. Motor has no flux yet. One possibility is to modulate on the "detection + additional freq" a second or so to allow flux to build up. On the several hundred kW range more that one second might be needed. After that ramp to frequency reference.

Usually nominal slip decreases when motor size increases. It means that the detected frequency has to be closer to real motor rotation frequency on a big machine than on a small machine. On a small motor search/scanning frequency can be decreased with faster rate than on a big motor.

If the motor rotates slowly detection does not work. I have seen this solved by simple starting from zero Hz if there is no detection and minimum search frequency has been reached. Minimum search frequency is typically a few hertz. If it is anticipated that the motor can rotate to negative direction then search is done also from high negative frequency and only after that the motor is started from zero frequency if there is no detection.

the voltage induced to stator by residual magnetic is high enough to make good estimation?

Click to expand...

Yes, unless you managed to demagnetize it before. In this case, a short current pulse would be sufficient to refresh the remanent field, that's what we do with an asynchronous generator.

- - - Updated - - -

As far as I know, a "searching mode" as described by tsan500 has been used by previous Siemens VFD, may be still used today.

Thank you for your comprehensive reply. I'll think about your suggestion, and I'll be back with any probable question.

Tsan and FvM, Thank you for your great help.