Drive/Servo Multi-Motor Application

[JRoss 84]

This is a question for anybody out there who has an in depth understanding of AC motors and how they work with VFDs, particularly in a multi-motor application. I'm putting this in the General Topics forum even though we're using Mitsubishi equipment because the problem is related more to the motors than anything. I'm working on a job upgrading the drives on a massive pavement testing rig. The rig weighs in the realm of 1.4 million pounds, and rides on two parallel railroad tracks that straddle the testing area. The rig moves up and down the tracks on 16 sets of "bogeys" (8 per track) that are each driven by a 50HP constant torque motor. The motors are connected in groups of four to 350HP VFDs (total of four VFDs). Originally, There was an encoder on one of the motors that was split out to all four drives (for speed feedback) and the PLC (for position feedback). Moving the rig involved turning on all four drives with the same speed and direction; stopping involved pneumatic disk brakes on each bogey. The system is 12-15 years old. The upgrade replaced the original Mitsubishi drives with Mitsubishi A700 VFDs (the most advanced VFD line they make) of the same size, as well as adding braking resistors to use "dynamic braking" and save the disk brakes for emergencies. In addition, we sought to improve the system by adding three encoders, for a total of four. Now each group of motors has its own encoder going back to its own VFD for speed regulation. For a variety of reasons, we took this a step further by adding motion option cards to the drives so that we could also close the position loop and control the system similar to servo. We had to send three motors back to the motor OEM to add extended shafts for the encoders. The customer also chose to have their spare motor modified in the same way (total of four motors modified). After some initial difficulty tuning the system due to mechanical backlash, we had the system running fairly well. We took several weeks off while some other work was being done in the area, and then the fun began. After sitting idle for a couple of weeks, the customer powered up the rig to move it to a different section of the track. It got about 10 feet at a fairly slow speed, when one of the encoder motors failed catastrophically. Sparks, smoke, the whole nine yards. We came in and helped them mount the spare, then moved the rig about 4 feet to test motion. All was well, so we started to move again, and got a Ground Fault error on one of the drives. Turns out that a second encoder motor (NOT the spare) had failed, though very quietly. At this point, we megged all the motors, and found that most of them had more than adequate insulation (including the spare). However, a third encoder motor was nearly ready to fail, measuring at about 0.05 megaohms. I'm told readings should be in the 4-5 megaohm range. The three failed (or almost) motors were the ones that had been modified by the OEM. So now we are facing the unenviable task of trying to figure out why the motors failed. One theory is that something happened during the modification to damage the insulation, and the OEM has all three motors and is trying to determine that. A second theory is that something about the design of the upgrade is damaging the motors, and this is the one I'd like input on. Other than that, I'm not entirely sure what questions to ask. Any help would be appreciated! I will also be posting this on the plctalk forums.

[GT6Steve 4]

Did the other work in the area involve welders??

[JRoss 84]

Not to my knowledge. The customer's engineer was doing some controls upgrades on the loading cells that are the business end of the test rig. All the physical work was done parallel to our install, and as far as I know, he was just working out the programming kinks.

[GT6Steve 4]

So these are standard shaft encoders, optical disc and photoeye? It is the encoder motors specifically that failed. Not the drive motors. Right The drive motors are the originals, merely having the shafts modified for the encoders. Same style of encoder as the original one? Are the encoders electrically isolated from the motor or bonded. Transient currents Is where that's going. I'm purposely ignoring the obvious connection faults and such as you've certainly got the expertise on hand to find the obvious. All that's essentially changed is the drive package.....

[JRoss 84]

All motors are original, one already had an extended shaft for an encoder. Three of the others were modified to add an extended shaft to accept an encoder. It's these three motors that failed. We used standard optical hollow-shaft encoders, and the wiring is electrically isolated. Changes: We replaced the drives, added braking resistors, added line reactors, added overload protection for each motor individually, added the additional encoders, and are now using a motion controller over SSC-Net III to command the drives.

[Veganic 8]

No alarm history on the VFD? Motors failed due to temperature rise in overload conditions? Can you run tests on the undamaged unit or can it only be run with the others? Ask them to modify one of the "good" motors with an extended shaft - see what reaction you get? Can you use multiple motors with a single encoder? Edited 14 Aug 2013 by Veganic

[JRoss 84]

No alarms on the first one that I know about. Second one just a ground fault. We were running at low speed, but not at full load, just jogging along. All three units have been sent back to the motor OEM for diagnosis and repair, still waiting to hear back. We can't even consider modifying another motor because of the time involved. Now.... I want to look past the obvious culprit of the motors. Is there anything about the design that could have stressed the motors to this point? Misaligned flux in the four motors on a VFD? The servo-like control pushing the motors harder? Etc. etc....

[Veganic 8]

Have you spoken to Mitsubishi? I had a quick google and the only reference I found was to a Siemens VFD: "Operation with more than one motor is possible with all inverters, providing the following points are observed: .... encoder operation is not possible under any circumstances." Which doesn't explain why it would be the encoder motors in each case - or why one survived. If the problem is with the motor modification you'll be a long time looking for alternative causes. I think you need to be convinced that the motor modifications are not the root cause. That's why I suggested getting another motor modified. The reaction may tell you all you need to know about their confidence in the soundness of the modification.

[gleblanc 1]

I wonder if something in the drive implementation might be causing issues. The newer VFDs almost certainly run on a different PWM frequency. They may also be switching on and off faster (steeper slope on the switching, not different frequency), causing higher voltage transients as part of the switching process. Is there any communication between the 4 VFDs, so that they know how fast each is running? The 4 different VFDs could now be running different speeds. I'll assume that the 1.4 million pound mass is rigid. If the speed differential is enough, could one set of motors now be pushing the entire machine, while the others aren't doing anything, or trying to slow it down? Could differences in wear on the bogey wheels account for some speed difference? Mind you, none of that explains why the encoder motors would be the ones that failed. Given that the motors didn't fail in drive groups, and that the encoders seem to be the common link between the problem motors, it seems unlikely that the problem lies elsewhere.

[GT6Steve 4]

Forgive me for pushing this to the top but enquiring minds wanna know..... How was this resolved? Steve

[JRoss 84]

This was "resolved" but never fully explained. The engineer at the motor manufacturer (the ones who originally modified, then later examined and repaired the failed motors) did not believe that our system destroyed the motors, but could not otherwise explain the destruction. We elevated the issue to Mitsubishi corporate. They discussed internally and stated that while they don't believe that multi-motor vector mode would destroy a motor, they do not support it and recommended replacing the SSC-Net III network with CC-Link. We discussed the application with other engineers who all came to roughly the same conclusion: multi-motor vector mode is not recommended, but shouldn't destroy motors. Our conclusion is that the combination of the modification on 15-year old motors and the stress the motors were under during commissioning (tuning, testing, etc.) served to finish them off. Nevertheless, we chose to play things safe and followed Mitsubishi's recommendation to "down-grade" to CC-Link. We lost some of the advantages of the positioning control, but the system is operational and the customer signed off. If anybody needs a Q172D motion CPU and some FR-A7S option cards that are only slightly used, let me know...