VFD failure

[Rsteenson]

Hey guys,

I would like to get some input from everyone. We have had two Square D Altivar VFD's blow up on the same motor in a month. Square D hasn't even gotten back to me with any advice or suggestions on where to go from here. My customer ordered a dust system for their train loading system at a grain elevator about two years ago. I ordered my MCC equipment at the same time and it showed up about 6 months later. The mechanical equipment finally showed up last March and we had it running in late April. We had a VFD failure after about 200 hours of run time and of course since it was sitting in packaging for a year and a half there was no arguing warranty replacement. My company has installed maybe a hundred VFD's over the years from 1 hp to 400 hp and I havent ever had to deal with an event like this. The wires for this motor leave the MCC room underground for about 40' through an existing PVC conduit that houses some other smaller 480 volt motors. It emerges and changes over to IMC conduit for about 100' before it hits the motor. This is the only VFD that comes out of this electrical room that has about a 50-60 motor MCC. We pulled THHN wires from the VFD to the motor un-spliced. Altivar VFDs have an internal line reactor in them so I would think it would filter any unexpected small spikes? Here is a summary of the events that happened, I would appreciate any suggestions.

VFD #1.
Went into commission on 4/20. All startup went well.
The VFD ran for 19 trains which take about 11 hours run time on average, about 209 hours.

The train crew showed up at roughly 6 am on 11/1 and when they went to start the filter fan from the plant PLC HMI it wouldn't start (it hadnt ran in a couple of weeks). The employee went to the adjacent MCC room and saw the breaker was tripped. When he went to reset it the VFD bucket blew up. When I opened this VFD casing up I saw had an obvious line side short. We were not able to analyze the last fault occurred so we do not know the direct cause. The breaker inside the bucket had parts rattling around so it was one hell of a short.
When we wired this system in April, the feeder conductors were pulled through 40' of existing conduit with existing wires. From there we ran new conduit another 100 feet to the new dust filtration system 60 hp motor. My employees that showed up the morning that the VFD failed were the same employees that wired the system in April. Upon arrival they tested the motor, tested the feeder conductors, and opened every enclosure that the conductors passed through to make sure there wasn't an obvious wiring short. All megger measurements were 2g ohms, which is perfect. They borrowed a 60 hp across the line starter from a different motor on site and all started and ran just fine.

VFD #2.
We swapped out the borrowed across the line starter for the new VFD and it went into commission on 11/14. I also changed the breaker inside the bucket that was on the line side of the drive to avoid any chance that this could be part of the problem. All startup went well.
This VFD ran for 7 trains, or about 77 hours.

The train crew ran for roughly 6 hours of a nighttime train on 12/10. This time the filter fan tripped mid-train (roughly 6 hours) at around midnight. When the employee went to reset the breaker on the MCC bucket the same thing happened as the first time, it blew up.
My same employees went over on Monday morning 12/12. They tested the motor, wires, and junctions points the same as before. Everything checked out perfectly like it did the last time. They borrowed the same motor starter as they did the first time and everything has been running fine since.

So, the other strange thing is that we had a 150 hp Altivar VFD in a different electrical building fail about 6 months ago. This VFD is in an electrical room about 300' away, but is fed from the same utility transformer. This one only lost its control board but it happened when it was sitting in stop mode just like VFD #1. I have always seen VFD's protect themselves in the case of a short but two of these VFD's weren't even in run mode when they failed?
I reached out to the utility to ask some questions and they offered to put a power monitor on the xfmer. A couple of weeks later they reached out and said that they weren't seeing any odd events going on. I am hesitant to put another VFD in until I figure something out!

Thoughts???

 

[__dan]

Sounds like both drives failed in the same way, at the line input / rectifier side. The drive input transistors / diodes can fail shorted commonly and that's what you're describing. Trying to reset the MCC bucket into a shorted faulted drive rectifier, it can be expected to go boom.

The drives normally have fast fuses at the input which would have been taken out. And, resetting the MCC breaker into a faulted drive, while not desirable, the MCC bucket is expected to not itself blow up in pieces. I have seen a pretty new drive in that size range fail early by shorting an input diode. And under orders they wanted me to close power into it with new fuses, which I did, but seeing how the fuses and the breaker both taken out, I advised it would go boom, and why.

Generally when the drive takes out the input fuses / breaker, shorted diodes at the front end are suspect and the drive is done. There is a non powered diode test that usually works if you have practice with it, instead of testing it by closing the industrial power system into a suspected short fault. Normally I would say get the fault codes from it, but the drive would have to power up for that. Try doing the non powered diode test at the front end, just for practice for the next time.

Why this happened, it's just a guess or working premise but the transistors are fragile for transient overvoltage. You could be in a area with something big routinely switching and putting transient noise on the bus. Transient suppressors, SPD's, can be a gimmick in systems that large. The noise could be bigger than the suppressor and it also goes poof. The drive would need series inductance, a line reactor, at the front end to mitigate transient overvoltage.

Troubleshooting is a process of elimination. Listing all the possible causes then working your way through the list. Other than transient overvoltage at the input and line side input line reactors as a common recommendation, I don't see another possible cause and mitigation indicated.

Sq D should be involved as much as possible in looking at the list of possible causes and then working their way through it. You definitely want to exhaust the process with them. They know the drive better than anyone.

 

[petersonra]

Even if you have a dead short the overcurrent protection device is supposed to be able to clear the fault without flames coming out of the bucket. I wonder if there is more available fault current in the over current protection device can handle.

 

[GeorgeB]

Going based on what OP and @__dan say, I'd like to be sure the system is grounded Y. @Jraef has explained that the corner grounded delta can cause problems on input circuitry.

 

[Rsteenson]

Even if you have a dead short the overcurrent protection device is supposed to be able to clear the fault without flames coming out of the bucket. I wonder if there is more available fault current in the over current protection device can handle.

It’s a 64k rated breaker. AFC is 24k. Flames weren’t out of the bucket, but there was soot all over.

 

[Rsteenson]

Going based on what OP and @__dan say, I'd like to be sure the system is grounded Y. @Jraef has explained that the corner grounded delta can cause problems on input circuitry.

Yes, it is a grounded wye tub.

 

[Rsteenson]

It’s a 64k rated breaker. AFC is 24k. Flames weren’t out of the bucket, but there was soot all over inside the bucket.

 

[Rsteenson]

Sounds like both drives failed in the same way, at the line input / rectifier side. The drive input transistors / diodes can fail shorted commonly and that's what you're describing. Trying to reset the MCC bucket into a shorted faulted drive rectifier, it can be expected to go boom.

The drives normally have fast fuses at the input which would have been taken out. And, resetting the MCC breaker into a faulted drive, while not desirable, the MCC bucket is expected to not itself blow up in pieces. I have seen a pretty new drive in that size range fail early by shorting an input diode. And under orders they wanted me to close power into it with new fuses, which I did, but seeing how the fuses and the breaker both taken out, I advised it would go boom, and why.

Generally when the drive takes out the input fuses / breaker, shorted diodes at the front end are suspect and the drive is done. There is a non powered diode test that usually works if you have practice with it, instead of testing it by closing the industrial power system into a suspected short fault. Normally I would say get the fault codes from it, but the drive would have to power up for that. Try doing the non powered diode test at the front end, just for practice for the next time.

Why this happened, it's just a guess or working premise but the transistors are fragile for transient overvoltage. You could be in a area with something big routinely switching and putting transient noise on the bus. Transient suppressors, SPD's, can be a gimmick in systems that large. The noise could be bigger than the suppressor and it also goes poof. The drive would need series inductance, a line reactor, at the front end to mitigate transient overvoltage.

Troubleshooting is a process of elimination. Listing all the possible causes then working your way through the list. Other than transient overvoltage at the input and line side input line reactors as a common recommendation, I don't see another possible cause and mitigation indicated.

Sq D should be involved as much as possible in looking at the list of possible causes and then working their way through it. You definitely want to exhaust the process with them. They know the drive better than anyone.

I hooked up 24vdc to the control terminal board and got the drive hmi to power up. I could see parameters on the hmi itself but I couldn’t access anything in the VFD.

 

[Jraef]

One possible scenario is that the SYSTEM is ungrounded (it has nothing to do with the bucket) and as a result, the line side MOVs on the drive failed catastrophically. When they explode (and yes, they can), they spew conductive material all over everything in the vicinity. The breaker usually trips but as soon as it is reset, it is closing into a dead fault and there is an arc blast, hence the flames coming out of the bucket. Because this failure is on the line side of the VFD, the drive electronics have no visibility of the fault and could not have done anything about it, nor record anything other than the loss of line power.

This by the way is one reason why UL has recently changed and is now REQUIRING high speed fuses ahead of drives, even if there is a circuit breaker. The idea is that by making people open the enclosure to change the fuses, they will see the damage and (hopefully) know not to re-energize it until the damage is cleaned up.

 

[Rsteenson]

One possible scenario is that the SYSTEM is ungrounded (it has nothing to do with the bucket) and as a result, the line side MOVs on the drive failed catastrophically. When they explode (and yes, they can), they spew conductive material all over everything in the vicinity. The breaker usually trips but as soon as it is reset, it is closing into a dead fault and there is an arc blast, hence the flames coming out of the bucket. Because this failure is on the line side of the VFD, the drive electronics have no visibility of the fault and could not have done anything about it, nor record anything other than the loss of line power.

This by the way is one reason why UL has recently changed and is now REQUIRING high speed fuses ahead of drives, even if there is a circuit breaker. The idea is that by making people open the enclosure to change the fuses, they will see the damage and (hopefully) know not to re-energize it until the damage is cleaned up.

Thanks for jumping in Jraef. The utility xfmr is a grounded wye. Based on the chain of events I explained it sure sounds like everyone is pretty comfortable that this is due to a line side surge. The utility is adamant they aren't seeing any spikes so I don't know where else they could be coming from? What would be the proper way to to add some protection for the next VFD that goes in?
One other thing I should mention- there are two 2000 amp MCB motor control centers in this room. There is a surge protector on one of the motor control centers, but not the other (this VFD is on the one without). I didn't design this room or equipment but I cant see how that would suffice? The Square D surge protection bucket that was in there needed all three trip units replaced a couple of years back. They were probably 10 years old so I'm not sure if that is the extent of their lifetime or not.

 

[Rsteenson]

There are about (5) wye start-delta run starters in this MCC for 75 through 150 hp. Would the startup of these create dangerous spikes?

 

[kwired]

Thanks for jumping in Jraef. The utility xfmr is a grounded wye. Based on the chain of events I explained it sure sounds like everyone is pretty comfortable that this is due to a line side surge. The utility is adamant they aren't seeing any spikes so I don't know where else they could be coming from? What would be the proper way to to add some protection for the next VFD that goes in?
One other thing I should mention- there are two 2000 amp MCB motor control centers in this room. There is a surge protector on one of the motor control centers, but not the other (this VFD is on the one without). I didn't design this room or equipment but I cant see how that would suffice? The Square D surge protection bucket that was in there needed all three trip units replaced a couple of years back. They were probably 10 years old so I'm not sure if that is the extent of their lifetime or not.

Lifetime is probably more defined by what transients they have been subjected to than how long they have been in service.

 

[__dan]

The Square D surge protection bucket that was in there needed all three trip units replaced a couple of years back. They were probably 10 years old so I'm not sure if that is the extent of their lifetime or not.

That's pretty solid evidence that the noise on the bus, transient overvoltage, is bigger than the industrial grade SPD first in line for the hit. That SPD would be quite larger or better rated than the built in SPD elements inside the drive front end.

There are about (5) wye start-delta run starters in this MCC for 75 through 150 hp. Would the startup of these create dangerous spikes?

My first thought was noise on the bus generated by the other loads switching, rather than transient spike coming in through the utility transformer which also does happen.

A recording instrument directly on the MCC bus should show something nasty going on for switching transients. The transient are very fast, like 20kHz.

I have seen that with smaller more fragile devices with their own internal SPD MOV's attached to an MCC bus with large switching loads. The 75 hp drive I would expect has a beefier front end. But if the transients take out the intended dedicated MCC SPD, the drive would be smaller and more fragile than that.

The Y delta starters are going to be noisy. With the motor spinning it is generating Voltage when it closes to the delta run starter. Guaranteed they are not in synch when that happens. Worst would be when there is no load on the bus to absorb the noise and then switching. You would see that if for example the MCC swapped from utility to generator, switching the entire thing from no load.

It could be difficult to intercept the bus to put line reactors between the bus and the drive. If you decide to put a large SPD on the MCC which probably would be advised, especially if you can see it happening on a recording instrument, the SPD should be large enough to have its own external can. The guts, the MOV's, can fail catastrophically inside the can.

 

[__dan]

Also, the breaker that tripped is likely toast and should be changed. It sounds like the let through was too much.

The arc blast leaves behind a carbonization which is conductive. I would be happy the thing was able to turn off and would be afraid to turn it back on if it looked like the blast escaped the breaker and went everywhere. I would be afraid of the conductive carbonization on the insulators. I would be afraid of flashover over the damaged insulators.

 

[GeorgeB]

Maybe I need to add blind to my deaf and dumb, but I don't see the model of your Altivar drive in the thread. A quick perusal of the se website shows that SOME versions require an external line reactor.

In any event, adding an external reactor should help knock the peaks down. Perhaps you will know more when se gets back to you.

 

[Rsteenson]

Maybe I need to add blind to my deaf and dumb, but I don't see the model of your Altivar drive in the thread. A quick perusal of the se website shows that SOME versions require an external line reactor.

In any event, adding an external reactor should help knock the peaks down. Perhaps you will know more when se gets back to you.

It is an altivar 630 process drive. It’s spec’d with a 3% impedance filter.

 

[__dan]

Not sure but I seem to recall, the 75 hp drive I saw fail early after about two years was also a Schneider Altivar.

I would not say that one was a transient overvoltage event. It ran the day before and was found failed shorted for the start of the next shift. It failed by itself just sitting there doing nothing. That was 2021.

At the time I saw it as a failure that should not have happened. It was too new. It was located at the furthest point in the building from the utility service, over 1000 ft away, so transients should have been dissipated by the extreme long distance. We changed it first thing in the morning after trying to run the plant. That site those guys did not have or want any clue. They only cared if it ran or not but not why or how. Was likely not reported to Schneider. The front end diode or transistor failed shorted.

 

[Jraef]

It is an altivar 630 process drive. It’s spec’d with a 3% impedance filter.

Do you actually SEE a line reactor in the bucket or are you going by some piece of sales literature? Unless the drive was ordered as a specific engineered packaged drive built into the MCC bucket with an external accessory line reactor, it's not exactly the same. There is an internal DC bus choke, which is the equivalent of a 3% line reactor in terms of reducing harmonics, but because it is AFTER the rectifier, it does not provide the same line side protective functions.

There are about (5) wye start-delta run starters in this MCC for 75 through 150 hp. Would the startup of these create dangerous spikes?

Absolutely Wye-Delta starters can produce MASSIVE transients called "transition spikes" on the AC line, which is why they should never be used if there are VFDFs or other power electronics on the same bus.

 

[Rsteenson]

Do you actually SEE a line reactor in the bucket or are you going by some piece of sales literature? Unless the drive was ordered as a specific engineered packaged drive built into the MCC bucket with an external accessory line reactor, it's not exactly the same. There is an internal DC bus choke, which is the equivalent of a 3% line reactor in terms of reducing harmonics, but because it is AFTER the rectifier, it does not provide the same line side protective functions.

Absolutely Wye-Delta starters can produce MASSIVE transients called "transition spikes" on the AC line, which is why they should never be used if there are VFDFs or other power electronics on the same bus.

So I typically order AB VFD’s with stand alone line reactors in the bucket. The last few times I ordered square d mcc with VFD’s (to go in to existing systems) I was told by the distributor that I don’t need them because the altivar drives are built with 3% internal reactors. The MCC that I added this VFD into had the wyeStart delta runs in them when I came on site. I should’ve done more homework on this I guess . Do you think a separate stand alone Line reactor would prevent the transients from the wye/delta starters? Thank you all for your input and advice. You are all wise men! I always learn something on this forum and I appreciate that.

 

[Jraef]

I was told by the distributor that I don’t need them because the altivar drives are built with 3% internal reactors.

Typical salesman misinformation. Again, they (like most non-Asian drives) have an internal CHOKE, but it is on the DC bus, so provides no protection for the rectifier. Not the same. AB, ABB, Siemens, Danfoss etc. drives that are built for EU and North American markets all** use DC bus chokes mostly because they are cheaper than adding more capacitors, but also provide SOME functions of a line reactor. Just not that one and that’s the one you likely needed here.

What LINE reactors do is to slow down the rate of rise of any line transients so that they do not cause damage to the solid state components. In the case of the rectifier, that would be the diodes or SCRs and the MOVs that protect them. The MOVs alone will stop the WORST transients, for a while, but the damage they receive is incremental so eventually they fail and often do so catastrophically.

So yes, with Wye-Delta starters in the same MCC, where you don’t even have conductor impedance to slow down the switching transient, you want line reactors ahead of your VFDs. In general, Line Reactors are “cheap insurance” in general when you have expensive drives and /or expensive down time.

And stop believing everything a salesperson tells you…

** all of these brands also offer small cheap “throw-away” drives that are Asian designs that use more capacitors rather than DC bus chokes, but those generally stop at around 30A.