VFD - Frequent Frequency changes

[redbarron55]

Two things to look at.
One the loop tuning is off.
The gain is probably too high and the time constant too short,.
The other is the bearing issue.
Motors are mostly inverter rated for the electrical windings, but not for the currents in the shaft and bearings.
If it were mine I would tune the loop and allow the pressure to drop a little more an smooth out the power and rpm fluctuations.
Secondly I would install an insulated rear bearing and and install an Aegis grounding ring on the output shaft.
If you still have the bearing take a zippidy doo blade and cyt eh outer race apart and look to see if the bearing had washboarding issues

If so (or even if not) insulating the rear bearing is a very good idea with VFDs.

 

[GTUnit]

I was thinking exactly the same thing. There is a study on Inverter Driven motor bearing failures floating around.
Basically VFDs can induce unwanted currents in the motor shaft which will then ARC from the shaft through the bearing to the bearing race, pitting it from a welding like discharge. This destroys the bearing over time.

It is also my theory that the power pikes generated when a VFD is commanded to do a rapid speed change will aggravate and increase the possibility of above.

Two things to look at.
One the loop tuning is off.
The gain is probably too high and the time constant too short,.
The other is the bearing issue.
Motors are mostly inverter rated for the electrical windings, but not for the currents in the shaft and bearings.
QUOTE]

 

[kamenges]

Does the speed only ever go to 30Hz or 60Hz? If so, I don't think it is as much a loop tuning issue as it is there is not continuous control loop at all. I'm thinking this is just a high/low control with hysteresis. If it were a continuous control loop you would think sooner or later the thing would settle somewhere other than the two speeds listed just by dumb luck.

Keith

 

[Tharon]

Those speeds were merely range examples. They fluctuate up and down anywhere from 30 to 60 hz. Most of the time the fluctuations are from about 35-48hz. And it does eventually settle down but can take anywhere from 10-15 minutes and then start back up 10-15 minutes later.

I have a 1:30 video of the drive output and pressure readout but I haven't got it off my phone yet to resize and post to something like YouTube.

 

[GTUnit]

The tanks should buffer the supply to where the drive does not need to follow line demand so closely.

Where is the pressure sensor that the PID references located? I hope its in the tank and not the output line.
What are the ramps in the VFD set to? If you have no control over the PID then maybe you can do a "patch" fix by extending the VFD acceleration and deceleration ramps

 

[Tharon]

The sensor is where ever it was placed by the compressor OEM. I am imagining somewhere inside the air compressor. It is not installed on our tank system.

 

[shooter]

Put the I on half the value it now has, and the D also on half the value.
It is oscillating on the pressure controller. and that is bad for the motor as it will get hot, due to the energy is put into it.
the control line should be after the tanks not before, so it will react slower on changes and it will give the real pressure that is going to the system.

 

[jraef]

Actually I can't see how the frequent speed changes are doing harm to the motor or drive. Forget the increase in heating, the drive is controlling current so accelerating is not going to involve a significant increase above the motor rated amps. Remember, if it is increasing from a lower speed, amps were lower as well so the increase to accelerate is not the same as if it were a load change at full speed.

I can't address the compressor wear issue however. A big concern with variable speed on compressors is the possibility of compromising the lubrication system. If that has been separated from the compressor speed, problem solved and I wouldn't worry about that.

 

[Tharon]

Tomorrow I am going to read through the manual and try to find where the adjustments for these values are.

I will also look at repositioning the pressure sensor, but am a bit more cautious about that because I am not certain what would happen if the compressor turned on, but for some reason the pressure was not getting to the tank (someone from maintenance maybe putting a shut off in line thinking they will help, or closing an existing shutoff, etc). With the sensor no longer on the compressor, would I be able to damage things even more?

Edit:
I talked with one of my pneumatic reps earlier and they basically said the biggest downside to the ramping up/down was the energy wasted. And I'm sure even if the physical wear was minimal or non existent, the energy loss from continually and quickly ramping up and down for no necessary reason would make it worth trying to fix if it's just a PID adjustment.

 

[OkiePC]

Don't break the cardinal rule of maintenance

 

[Tharon]

If it's working, poke around until it's working better?

If I break it we do have a spare waiting and ready to turn on. So I have a little wiggle room to play.

I find it more and more annoying now that I have been looking into it. Listening to that constant wind up and down echoing on my office wall. So now that I am thinking of it I cannot stop because every time I sit down to work on programming, all I hear is, "wrrrrrr wrrrrrr wrrrrrrr wrrrrrrr".

 

[Tharon]

Well, I have been too afraid to make any real changes to the system while it is working. I keep thinking I will break something and not be able to get it running again quickly enough. And production is afraid to run on the spare unless absolutely necessary.

So it might have to wait until the next Christmas / New Year shutdown.

For now I guess I'll just wear my ear plugs while sitting at my desk.

 

[jraef]

It's not noise or a debounce issue. I'm pretty sure it's just the PID is tuned tightly. The amount the output is changing with small changes in PV is huge. From what little I understand about PIDs, I would think it's the P or D portions that would cause it.

But my real concern is whether or not the frequent speed changes put any more wear on the motor or compressor.

Actually, it sounds as if might in fact be a "D" issue and people quite often ignore the D thinking it's unnecessary. The D is Derivative of the number of changes in time, so it is SPECIFICALLY there to mitigate high quantities of changes. Another way of looking at PID is;

P = Proportional to the size of the PV change, meaning the PRESENT error.
I = Integral to the rate of change in the PV, meaning PAST of that error.
D = Derivative of the quantity of PV changes in time, meaning predicting the FUTURE errors.

In your case there are apparently a lot of changes in the PV over a relatively short time and the PI programming is just responding to them unfiltered. The Derivative function is what is supposed to filter out those numbers of changes in time to make the overall response smoother.

Another more remote possibility is that when in that intermediate capacity range, the air demand is changing extremely rapidly by a lot, and the VFD is going into Current Limit because the external PID controller is giving it a very rapid change in speed. The way a VFD responds to a high current demand is to either trip off line, or override the speed command temporarily to allow the motor to catch up without exceeding the current setpoint. So the fact that you never see the motor overload might indicate that this is what's taking place. The system is demanding more of the motor than the VFD is capable of making it deliver, so the VFD is overriding the speed command to dampen the spike. But by the time the VFD is able to make the motor respond, the PID controller is satisfied and lowers the demand, which sends a signal to the VFD to slow down again.

 

[CalG]

Change the VFD acel and decel ramp time to 5 seconds

and let the proprietary control information fall where it may.

With that much air storage, sub-second response from the compressor speed is not needed.

 

[justblaze]

It could be skipping a certain frequency range where the compressor may resonate and make a lot of noise. We have several compressors. One is a 100hp IR Nirvana with a permanent magnet motor. It has a certain frequency range it will skip over and it will act exactly as you describe. Then demand changes and the cyclic effect stops.