I've been watching a discussion on a LinkedIn group that is addressing some writers' concerns about variable speed drives (VSD).
The discussion is titled "Are variable speed drives harmful to motor? [sic]" and part of the discussion can be found here. Note that you need to be a member of the group to read it. If you're not, the cogent part of the discussion from the original poster is summarized thusly: How will the PWM-based variable voltage/variable frequency excitation from a typical inverter-based VSD affect a typical induction motor? Several people provided thoughtful answers that gave details on what could go wrong and how it could go wrong.
VSDs are used to vary the speed of induction motors. Induction motors have traditionally been constant speed devices. They run at a speed that is synchronous (mostly) with the frequency of the applied current, although the phase relationship lags. To vary the speed, you need a power source (probably high power) with frequency (and probably voltage, too) that can be varied. Do this and you have a VSD. With a VSD controller, the applied voltage level is usually adjusted downward as the frequency is lowered: Since back EMF decreases at lower speeds, a lower applied voltage is needed (or else motor current would rise excessively).
So far, this is pretty straightforward. But how do we get a variable frequency? We only have 50Hz or 60Hz available. Easy as pie. Or possibly easy as π radians. Take the AC line voltage, full-wave rectify it, filter it, and apply that bus voltage to an H-bridge (or a 3-phase H-bridge) made from power FETs or IGBT devices. As described, this is the heart of a power inverter. So far, so good.
Control the drive signals to the H-bridge with an appropriately timed and phased arrangement of PWM signals. By varying the pulse width/duty cycle, the H-bridge outputs will produce current in the motor windings that, on average, appears sinusoidal, mostly. The motor's winding inductance and its inertia will tend to smooth things out. But this is where things can get a little troublesome.
Those pesky PWM waveforms are, on average, OK -- but the fact that they are switching on-and-off rapidly means that there is a lot of high-frequency content in them. This may cause excessive heating in the motor windings. The high di/dt also means there will high voltage transients across the motor windings. If the motor is not designed for inverter duty, it may be damaged.
Damage can occur where you probably would expect it -- the insulation on the motor windings may deteriorate or break down completely. Smoke and arcing are the telltale signs.
Damage can also occur where you probably don't expect it -- in the bearings. Again, the high-frequency content of the PWM current is the culprit. It can capacitively couple current through the bearings to the motor housing. This can cause damage that will appear to be pitting or corrosion -- bearings aren't designed to carry current, just mechanical loads.
One more concern -- EMI radiating from the connections between inverter and motor; or conducted back through the incoming AC power lines. Be aware of that so you can deal with it. A good resource is a paper written by several Rockwell Automation engineers, EMI Emissions of Modern PWM AC Drives.
Adding L-C filters between the inverter and the motor can help with the heating and insulation issue cited above. Sometimes this just moves the heating due to circulating currents from the motor to the filter elements, so be aware of that, too. Enlisting applications engineering help from the inverter and motor vendors is essential when dealing with this issue.
Have you had to deal with VSD/inverter problems like this? How did you resolve the problems?