VFD ground fault (related to lack of surge suppression or utility cap switching?)

[malachi constant]

Hi all,

I've been asked to look at a VFD ground fault issue that has plagued a client for over a year. Any thoughts you have are appreciated...

* Suburban school district. Possibly three buildings affected, all within two miles of each other.
* The mechanical systems at each of these schools were recently replaced (some in summer 2013, some in summer 2014). Still gathering info on the electrical services but at this point I believe they are 480V, were not upgraded, and I read a note that says they do not have surge suppression on them. Trying to confirm all that.
* There were no known issues prior to the mechanical upgrade, but now have experienced recurring issues on the VFDs and with motor burnouts.

The issues are:
* VFDs on multiple fans and pumps are tripping on ground faults. In some locations wire nuts have melted. At least one motor (toilet exhaust fan) has burned out.
* Electrician did a very thorough review of the situation in June and found two instances of wire nuts melting (at same motor, on different phases). Electrician stated this is typically caused by loose terminations but both nuts were solidly connected. Speculated that there was a short circuit or overvoltage. Did not find any evidence of faulty wiring.
* Electrician also found VFDs were programmed to 1.56% over FLA instead of manufacturer recommended 1.15%. Electrician reprogrammed a few of the VFDs and recommended someone come out and verify all were programmed correctly. When the electrician returned in August they noted the VFDs had reverted to their original settings.
* I have a fault log showing trips on VFDs at one school (the school with the most issues) over the summer. They appear to be very intermittent and not concurrent with each other. Each one notes that the voltage at the time of trip is between 600-700V on 480V units.
* The utility was engaged with little action to date. They did tell the Owner they were going to switch their cap bank during a given hour one morning, and the Owner confirmed three trips that hour, but only at one building (the one with the most issues). I am told the utility has stated that any problems resulting from the cap bank switching are the Owner's problem to resolve.
* The utility had applied data logging devices to each school, I do not believe we have seen copies of these but am trying to obtain.
* I do not believe to date that the VFDs have been properly reprogrammed/commissioned, even though this request has been repeatedly made since June.

So...
* Three buildings in close proximity.
* Frequent VFD ground vaults at all of them, no observed pattern.
* Possibly related to cap bank switching.
* Possibly related to improperly programmed VFDs. (Why would the handful that were reprogrammed later revert to old settings?)
* Possibly no surge suppression present. (If so seems like a pretty easy decision to add it.)

As you can see I'm still diving into it but any thoughts you all have on the matter is appreciated. Also curious how likely it is that there is widespread damage to the motors or drives. I'll keep you updated as everyone loves a good mystery... Thank you!!

 

[petersonra]

Hi all,

I've been asked to look at a VFD ground fault issue that has plagued a client for over a year. Any thoughts you have are appreciated...

* Suburban school district. Possibly three buildings affected, all within two miles of each other.
* The mechanical systems at each of these schools were recently replaced (some in summer 2013, some in summer 2014). Still gathering info on the electrical services but at this point I believe they are 480V, were not upgraded, and I read a note that says they do not have surge suppression on them. Trying to confirm all that.
* There were no known issues prior to the mechanical upgrade, but now have experienced recurring issues on the VFDs and with motor burnouts.

The issues are:
* VFDs on multiple fans and pumps are tripping on ground faults. In some locations wire nuts have melted. At least one motor (toilet exhaust fan) has burned out.
* Electrician did a very thorough review of the situation in June and found two instances of wire nuts melting (at same motor, on different phases). Electrician stated this is typically caused by loose terminations but both nuts were solidly connected. Speculated that there was a short circuit or overvoltage. Did not find any evidence of faulty wiring.
A melted wire nut suggests more current was flowing through the connection than should be, by a lot. Most VFDs will trip pretty quick if that happens.
* Electrician also found VFDs were programmed to 1.56% over FLA instead of manufacturer recommended 1.15%. Electrician reprogrammed a few of the VFDs and recommended someone come out and verify all were programmed correctly.
Why is the electrician fooling with something he clearly does not understand?
When the electrician returned in August they noted the VFDs had reverted to their original settings.
not that unusual. Many VFDs will revert to a previous setting on powerup if the new parameters are not explicitly saved.
* I have a fault log showing trips on VFDs at one school (the school with the most issues) over the summer. They appear to be very intermittent and not concurrent with each other. Each one notes that the voltage at the time of trip is between 600-700V on 480V units.
If you are getting 700V on a 480V line, you have a utility problem. period. However, it may be that what the fault log is telling you has to do with the DC bus voltage, and not the incoming AC.
* The utility was engaged with little action to date. They did tell the Owner they were going to switch their cap bank during a given hour one morning, and the Owner confirmed three trips that hour, but only at one building (the one with the most issues). I am told the utility has stated that any problems resulting from the cap bank switching are the Owner's problem to resolve.
That is true. It is not up to the utility to fix problems the owner has with their own equipment as long as the voltage the POCO is supplying is within the parameters specified by the utility commission or whoever it is that regulates the POCO there..
* The utility had applied data logging devices to each school, I do not believe we have seen copies of these but am trying to obtain.
Would be helpful.
* I do not believe to date that the VFDs have been properly reprogrammed/commissioned, even though this request has been repeatedly made since June.
Who has made the request and who is the request being made to? It may be that the person who is being requested to reprogram the VFD just does not know how.
So...
* Three buildings in close proximity.
* Frequent VFD ground vaults at all of them, no observed pattern.
* Possibly related to cap bank switching.
* Possibly related to improperly programmed VFDs. (Why would the handful that were reprogrammed later revert to old settings?)
* Possibly no surge suppression present. (If so seems like a pretty easy decision to add it.)
I doubt surge suppression would make all that much difference but it is relatively cheap to add to the service point.

As you can see I'm still diving into it but any thoughts you all have on the matter is appreciated. Also curious how likely it is that there is widespread damage to the motors or drives. I'll keep you updated as everyone loves a good mystery... Thank you!!

Let me ask you a few questions.

What brand/type of VFDs are these?

Do you have extensive experience working with these or any other VFDs? If not, why are you involved? Based on some of your questions (and failure to ask other questions) it does not seem like you have much experience with VFD installations.

Has anyone talked to the tech support people at the VFD manufacturer?

Is the incoming line 480V delta or wye, or even corner grounded?

if these are on ventilation fans it is possible that wind might be blowing hard enough to generate energy that has nowhere else to go so shows up on the DC bus. If there is no way to get rid of the excess energy it can trip on excess DC bus voltage. If it is a short term situation braking resistors could be added as a means of bleeding off the excess energy.

 

[malachi constant]

Hi Bob,

ABB is the manufacturer for the VFDs in question. I understand there is a different VFD manufacturer on one of the RTUs but have not confirmed what that is or if is is experiencing issues.

I'm a licensed PE (electrical) w/ 15 years experience in the architectural consulting field. Every firm I've worked for has made the decision to have VFDs specified by Div 23 and supplied by the mechanical contractor, so this kind of falls under "not my jurisdiction". But yeah, I don't expect the mechanical guys to figure out anything related to a ground fault. On my end I have been involved in resolving a handful of VFD-related issues over the years, so I'm no stranger to this world but am certainly not holding myself out as an expert.

I was asked to get involved by our architect, who had subbed this project out to a different electrical engineer (who is still engaged in this process). Apparently no one has taken the time to step back and gather all the data - just people reacting to angry emails for six months. I just found out about it last week and have been reviewing documents (emails, logs, bid documents, maps, etc) to better understand the situation and make sure the right questions are being asked. (My week has gone like this: "Hey! It's three buildings, not just one! Hey! It's not just ground faults, we've had units burn out! Hey! There's no surge suppression! Wheee!")

The VFD supplier/installer and controls contractor have been asked to look into this (by the GC, EC, electrical engineer, etc) multiple times in a number of emails, but as near as I can tell none have responded. I have calls into various folks to see if this is correct and to try to engage these parties.

The electrician was "fooling" with the VFD because their wiring was shot and they were the only ones who would respond to the situation. I won't throw them under the bus because after six months and ~100 emails they are the only ones who communicated clearly and concisely about what they saw, documented everything, and made strong recommendations ("get someone out here to program all the VFDs"). As it is ultimately "not their jurisdiction" they have stepped out of the picture but up until then have been a great resource. I wouldn't go so far as to say they don't know what they are doing as they were the ones who installed the drive supplied by Div 23 and seem to be very conversant throughout all this. That said the programming is supposed to be by the controls vendor.

Incoming service is almost certainly 480V wye, but I will add that to my list to double check.

Not just happening on fans - having issues with pumps too.

I'll check on whether the voltage on VFD log is AC or DC. The spreadsheet sent to me does not say. I have a screenshot of the VFD, it also does not say - will have to check with the manufacturer.

Are we yelling at each other? I hope not. I'm trying to answer your questions and hoping I'm not coming across as snarky or defensive. Trust me, your insight is appreciated. I'm just trying to bring sanity to a situation that needs it.

Thanks!

 

[petersonra]

Are we yelling at each other? I hope not. I'm trying to answer your questions and hoping I'm not coming across as snarky or defensive. Trust me, your insight is appreciated. I'm just trying to bring sanity to a situation that needs it.

Thanks!

Sometimes I am a little more direct than is considered tactful.

It sounds to me like no one has bothered to issue a P.O. to anyone who is able to actually look at the problem in depth. That is not an uncommon problem once a project is done and signed off on.

Gathering data is a good place to start, but one of the things I have learned from dealing with field people over the years is that often in cases like this what they think they saw is not what they really saw. It is best to go into this looking at it as if no one has ever looked at it before rather than getting preconceived and potentially misleading ideas from other people.

You mentioned that there are ground fault trips but the fault log you cited had nothing to do with ground faults. How did you get the idea there are ground faults if it is not what is in the log?

If the problem turns out to be switching transients from the power company, about all you can do is install load reactors on the VFDs that are tripping.

It may also be that you have a series of problems, some of which are related to improper VFD settings and some which are external problems.

The melted wire nuts are a very curious situation. I guess I would want to see what they mean by melted before worrying a lot about that though. Generally, a VFD will trip long before you get enough heat in a wire to melt a wire nut.

personally, I would get someone who has a lot of experience with these particular drives to go with you on site. All brands and even models within the same brand seem to have their own set of quirks.My suggestion is bite the bullet and issue a P.O. for a couple of days for an experienced guy to go with you onsite, and not necessarily a guy from ABB. There is no good way to know what you are getting if you do that. You might get a kid who just graduated from school last week, or a guy with 30 years of ABB VFD experience.

 

[ptonsparky]

We have solved problems with utility capacitor switching by installing line reactors for the VFD.

Load side reactors might help those motors that were not originally designed for use with VFDs. A good motor shop should be able tell you why the winding failed. I believe there are some indicators of over voltage spikes to the motors.

 

[Jraef]

1) I personally would NEVER use a wire nut on a VFD circuit. I never use wire nuts on motor connections period, but that's just me. Wire nuts and vibration don't play well together.

2) 156% of motor FLA? Hello! Someone has NO CLUE what they are doing. It's a very unfortunately common problem when setting OL relays for electricians to read the NEC and think that is a design guide, it is not. It is a list of minimums and maximums. The NEC says you CAN set up an OL for up to 125% of a motor FLA when the motor has a 1.15 Service Factor. BUT, that is the "pick-up point" of the OL relay, the point at which is BEGINS to count down to a trip based on time and current (squared). What the NEC doesn't tell you is that the OL relay mfrs will almost ALWAYS have ALREADY factored in the pick-up point in their settings / selection tables. So YOU don't need to add it again, and if you do, the pick-up point becomes 125% of 125% and the motor burns out! In the case of VFDs, it's the same for the built-in electronic overload protection, The VFD ALREADY has the pick-up point factored in.

Here is what it actually says in an ABB VFD manual with regard to setting the motor OL protection, they are all the same.

With the default value 100%, motor overload protection is functioning
when the constant current exceeds 127% of the parameter 9906
MOTOR NOM CURR value.

That 127% is ALREADY the pick-up point of the OL protection, as allowed for in the NEC (actually it's technically too high... but see below).

Also, the manual is NOT saying that you SHOULD set it for 115%, you are misinterpreting that. Here is the only mention of "115%", it says:

Example: If the constant protection level needs to be 115% of the motor
nominal current, set parameter 3007 value to 91%
(= 115/127*100%).

So what does that mean? It means, as per the NEC, IF the motor is rated for 1.0SF, then the OL pick-up point must be NO MORE THAN 115% of FLA. But because the ABB drive defaults to 127%, YOU must adjust it accordingly. So to ATTAIN 115% as a pick-up point, you must DE-RATE the value in that parameter to 91%.

How does this relate to reality? Look at the motor. 99.9% of motor manufacturers will tell you that if you use a motor with an VFD, the Service Factor becomes 1.0, even though it is marked 1.15 on the nameplate. So essentially, ALL motors run by inverters are 1.0 Service Factor, ergo the OL pick-up point per the NEC must be no MORE than 115%, so that parameter in the ABB drive must be set to 91%.

If the people making this bad decision are setting the motor FLA in the VFD to 156% of the actual nameplate (which by the way is 125% of 125%), then they are cooking the motor and all components in the circuit! Assuming by this that they are NOT actually reading the manual and understanding it, we can assume the OL setting is at the default of 127%, so the VFD is not even going to BEGIN thinking about protecting from an overload until the current reaches 156% x 127%, so 198% of FLA! I'm not surprised that the wire nuts are melting, it's likely the weakest link and the first thing to go. So as soon as they do, the motor leads are touching the frame and you get a GF trip.

Bottom line, I doubt the GF tripping is the problem here, it's just the last line of defense against foolish decisions!

 

[ptonsparky]

Given Jraefs lesson...who the heck designed all this that the motors are overloaded? Pumps and fans would lead me to believe they are being run at above normal speed, be it pulley ratios or HZ. Mechanical design problem from the start or VFD programming error at install of VFDs.

 

[Jraef]

Given Jraefs lesson...who the heck designed all this that the motors are overloaded? Pumps and fans would lead me to believe they are being run at above normal speed, be it pulley ratios or HZ. Mechanical design problem from the start or VFD programming error at install of VFDs.

That would be my guess too.

"Hey, we need more flow, turn up the speed on those VFDs!"

I had a MECHANICAL ENGINEER tell me to do that once to make up for the fact that he had selected too small of a pump. He insisted that I increase the pump speed to 90Hz. I pointed out that if I did, his 100HP pump was going to demand 272HP, but the motor was still a 100HP motor. After a half day of me showing him how the Affinity Laws work (which I would have thought an ME would know), he finally figured it out.

 

[Iron_Ben]

* The utility was engaged with little action to date. They did tell the Owner they were going to switch their cap bank during a given hour one morning, and the Owner confirmed three trips that hour, but only at one building (the one with the most issues). I am told the utility has stated that any problems resulting from the cap bank switching are the Owner's problem to resolve.

You have the best username ever. Well done. That's about my third or fourth all time favorite book. I've read it half a dozen times I'm certain. But as to the topic at hand.....

The cap bank switching could be a contributor to the issues. Is the bank "fixed", in that the poco switches it manually, or is it "switched"? We had both types on our system. The controllers that operated the switched cap banks could be configured to open or close on any one of voltage, current, temperature, vars, amps, time of day and more. Some of our banks might have operated once every couple of months, while others came in and out several times a day. We had antique controllers from the 60's and 70's on our system but also some of the latest and greatest. Those would have a date and time stamp for each cap bank switching operation, which could help in troubleshooting if you compare to your VFD fault logs.

Lastly, the power company's alleged response on this subject is inappropriate. I almost can't believe that one of their people would say such a thing. You need to find the right person at the power company, the person who will work with you to determine if in fact the capacitor switching is contributory to the VFD problems or not.

 

[malachi constant]

If the people making this bad decision are setting the motor FLA in the VFD to 156% of the actual nameplate (which by the way is 125% of 125%), then they are cooking the motor and all components in the circuit! Assuming by this that they are NOT actually reading the manual and understanding it, we can assume the OL setting is at the default of 127%, so the VFD is not even going to BEGIN thinking about protecting from an overload until the current reaches 156% x 127%, so 198% of FLA! I'm not surprised that the wire nuts are melting, it's likely the weakest link and the first thing to go. So as soon as they do, the motor leads are touching the frame and you get a GF trip.

Bottom line, I doubt the GF tripping is the problem here, it's just the last line of defense against foolish decisions!

jraef: Ok, let's say the overload is set at 156% (not really a hypothetical - I think that is the case here). I understand how a high overload setting would fail to protect the motor. But what is calling for the motor to run that hard in the first place? Does the OL relay double as a protection point AND an upper-bounds control point? (Not sure how to word that, hope you can follow it...)

Bob: I will definitely push for a third party to take a look. I have a great resource in the local VFD world that works for a different manufacturer but is someone I trust to be very honest, and frankly the best at solving VFD issues that I have worked with. I should have thought of that on my own. (But hadn't - so thanks!)
Also, regarding why we suspect "ground fault" - the ABB VFD fault logger says "earth fault", "DI STATUS AT FLT", Status at FLT: 0000 hex, Fault: 16, Freq @ fault: 34.9 Hz, Speed @ fault: 1032 rpm, torque at fault: 25.5%, current at fault: 7.5A, voltage @ fault: 659.4V. This is typical of a half dozen faults in a seven hour period. Appears to be a fairly normal fault full them.

Tom: I'll add line reactors to the list of most plausible solutions. Since these motors were all designed for use with VFDs I assume we can dismiss load side reactors. As to who designed it all, the mechanical engineer specified new motors with new VFDs sized to match (suchas 5HP motor w/ a 5HP VFD). Obviously our intention is that the VFDs would be correctly programmed in the field. I didn't even know you can get a correctly sized VFD but program it to overdrive a motor to the point of damage. For what it's worth I did find an email where the VFD installer swears the units were programmed correctly and wonders if someone has changed the settings. So appears something is resetting them, perhaps the original settings were never explicitly saved as Bob originally suggested.

Iron Ben: Thanks. I've secured it as a username on a couple sites now. I need to go back and re-read that - any Vonnegut, really. It's been a decade or so since I picked up anything of his, but they always get your mind going. Regarding utility switched vs fixed cap banks, I'll have to ask. They say "we switched the banks this morning" but do not indicate if that is auto or manual. I agree that the utility has a duty to prove their power is reasonably clean and stable - as there is a similar problem at three buildings I haven't ruled them out yet.

Thanks all!

 

[Jraef]

I should have added:

The GF detection for a VFD is on the OUTPUT part of the circuit, they use a Residual Current Detection method looking at the three output currents, usually using Hall Effect transducers on the common collector side of each pair of transistors, although some use CTs or Rogowski Coils. So the VFD GF detection would never see anything on the input side, and anomalies on the input side, while they may affect the rectifier and capacitor bank in the VFD, will NOT cause the VFD to trip on GF.

 

[mike_kilroy]

So to be clear:

1) your client has had 3? 5? 12? 27? false 'ground fault' errors on various vfds?
2) your client has had ONE motor actually fail? Do you know the exact fault displayed in THAT case?
3) 3? 4? 10? wirenuts between vfd & motors melted

So. 1 motor failure in 2-3 schools in 6 months? 1yr? 2yrs? 1 motor out of 100? 200? 300? Perhaps this fact should not unduly influence all the other theories?

Wirenuts where they really should never be, melting? If loose connection (Did YOU PERSONALLY vouch for their tightness?), it will melt... Keep in mind what that energy looks like: pretty nice sinewave CURRENT waveform at motor/vfd rating, and 650v nominal voltage spikes going to 1000-1500v peak to peak at pwm rate of the vfd (2khz? 4khz? 6khz?) = pretty nasty stuff to rely on a wirenut! Take them Jraef's recommendation to NEVER do this in future.

So sifting thru the problems, it seems MAYBE your issue is random false vfd output ground fault errors without real damage?

I would recommend you ask ABB engineers (not salesmen) exactly what causes "ground fault" errors. Might give clues.

Have you verified that all installed motors have a GROUND WIRE coming directly back to each VFD along with the 3 phase leads? Maybe it is missing. Assuming your 3ph wye input, have you confirmed all power leads to vfds include the X0 neutral to vfd PE terminal?

I vote too for a cheap and fast potential fix as throwing a 3% line reactor in each VFDs AC input as a mask to the possible capacitor switching crud on the line. sometimes you will never get to the real answer when it comes to noise and just solving the symptom is the best solution.

 

[petersonra]

Have you verified that all installed motors have a GROUND WIRE coming directly back to each VFD along with the 3 phase leads? Maybe it is missing. Assuming your 3ph wye input, have you confirmed all power leads to vfds include the X0 neutral to vfd PE terminal?

Make sure it is an EGC going to the PE and not a N!!!!!!! It would be a serious code violation to run a N to the PE terminal on the VFD.

In any case, GF errors are not caused by missing ground wires.

 

[mike_kilroy]

Make sure it is an EGC going to the PE and not a N!!!!!!! It would be a serious code violation to run a N to the PE terminal on the VFD. In any case, GF errors are not caused by missing ground wires.

Sorry, guess I should have explained better. OP said he is fairly certain power source is wye. That means there is an X0 terminal on the wye. That is called "nuetral" and 99% of the time is tied immediately to ground, PE, EGC, whatever you want to call it. This Xo connection is seldom openly available to a plant technician to tie to, but the grounds are. So my shorthand reply meant to say, be sure the X0, called nuetral, center of the wye power source, is indeed tied to ground, and therefore to your PE terminal on your VFD, as it provides the low impedance tie to each phase to reduce noise problems.

Recall GF detection is just a method of watching for non 0 current (imbalanced) in a 3 phase circuit, so it CAN be applied to either the input or output of the vfd. I think most vfds implement it on the output, but ABB is a high end company that may use the input side in order for their drives to provide GF protection even on HRG systems; I do not know.

Actually missing or incorrectly grounded VFD, motor, or motor cabling CAN cause false GF faults.

Since my previous reply was confusing, I will add more here too. There is a chance those schools have high resistance grounding (HGR), used to reduce arc flashes if a phase gets grounded, and thus the Xo nuetral is NOT tied directly to ground, but rather, through a high resistance. But I did not mention this possibility - because it then becomes VERY UNLIKELY for any VFD to EVER give a GF alarm; most vfds have the GF detection circuit set to around 50% of the drive rating. With HGR grounding, this is usually higher than that value, so the drives GF circuitry is null and void and no longer works. A good description of this is here: http://www.plantservices.com/articles/2010/09groundfaults/

For this reason, there are GF detection circuits that are applied to the INPUT power to the vfd rather than the output side. Attached is a picture of such a method. Recall I suggested calling ABB and asking an application or design engineer how their GF circuit operates: this is the reason. If it is on the OUTPUT side of the drive, then it is unlikely as Jraef said for false GF alarms to be caused by anything on the input side. But without knowing the exact method YOUR vfd drives use, you cannot simply assume the protection is on the output side, thus input power may still be a concern.

As to grounding causing the false GF protection alarms, even ABB in their technical writings list it as a very valid potential reason. I have copied pages 16 & 17, THEIR list of potential sources of false GF alarms, here as they list possible causes of false GF alarms. This from: http://c.ymcdn.com/sites/www.ncsafe...BDFD-B3A2ACC137E7/OM_Tues_AM_0800_Stanley.pdf

 

[kwired]

A melted wire nut seems to be a good place for a ground fault to occur, especially for as little of a fault that can trip many VFD's.

Though it sounds like overloading is a problem here, also consider the distance from the drive to the motor - even with a VFD rated motor and especially on 480 volt supply - as the peak of the spikes that can be produced is well over 480 volts, where if a dual voltage motor were connected to 208 or 240 volts this spike is significantly lower when it comes to it's effects on the same insulation.

 

[milldrone]

here as they list possible causes of false GF alarms. This from: http://c.ymcdn.com/sites/www.ncsafe...BDFD-B3A2ACC137E7/OM_Tues_AM_0800_Stanley.pdf

Page 12

Where’s the squirrel?

•