Ground Fault on Large Switch Gear

[electrical chick]

We service industrial companies with large 2000 to 4000 amp switch gear. Of these companies, several are plastic extrusion plants. All large switch gear is required to have Ground Fault protection. The problem that some of our customers are having is that any small short downstream trips the main in the switch gear and shuts down the extrusion line or the plant. This causes thousands of dollars of down time and clean up or wasted product. Why do the individual breakers serving these down the line pieces of equipment not trip instead of the main? Is it a series versus fulling rated issue? It is because the AIC rating of subpanels is lower than that of the main switch gear, or is it totally due to the sensitivity of the ground fault protection in the main versus the sensitivity of branch panels and breakers? Can this problem be stopped without violating code or endangering employees? I am trying to fully understand where the problem is before offering a possible solution.

 

[haskindm]

You will need to determine why the breaker is tripping. I suspect it is tripping due to a Ground fault, so the main is tripping not the branch circuit breakers as the main is the only breaker that is capable of sensing the ground fault. You may need to install GFCI on the branch circuits in order for them to trip first in the event of a ground fault.
There is an exception (230.95) that allows the Ground Fault to be omitted for continuous industrial processes "where a non-orderly shutdown will introduce additional or increased hazards". This will require a long talk with the AHJ to determine if this exception should be applied and if ground fault detectors and alarms would be required. Keep in mind that expense does not equal hazard.

 

[raider1]

You may also want to look into whether or not the setting of the ground fault device is correct.

I have been on jobs where the GFP was set to the lowest setting from the factory and this caused the main to trip due to a ground fault in a 20 amp 277 volt lighting circuit.

The Ground Fault Protective device should be set at a point to protect the large service equipment from a phase to ground fault but not lower enough to trip before the down stream OCPD's trip.

Chris

 

[jim dungar]

You need to familiarize yourself with what breaker ratings mean and how their tripping functions operate.

The fault current handlijng capability of a breaker, its AIC, has almost nothing to do with its tripping functions. Yes it is true that some high AIC breakers trip faster, than lower AIC ones, in the instantaneous region (i.e. >3x long time pickup).

Whenever you question why one device operates before a different one, you are dealing with the coordination of the devices. It is surprizing how few companies are willing to pay to have a coordination study performed.

The probable solution is not to disable the GF protection on the main, but rather to add GF protection to the feeder and or branch breakers instead. And of course, have a coordination study performed.

 

[kingpb]

As others pointed out it sounds like GF settings are not coordinated. The IEEE Buff Book would be a good place to start for in depth coverage on plant protection.

Has an arch flash study been performed? If so, it could be GF settings were changed to try and accommodate the results of the study. If not, I'd say the last person checking/setting the replays did not understand what they were doing.

To do it right, a full coordination study needs to be done which will require modeling the entire system and performing short circuit analysis. Why they are at it they might as well do arc flash as well since it won't take much more effort.

This is a relatively expensive undertaking depending on size of plant or equipment involved, but cost wise, I would imagine much cheaper then plant down time and lost product.

 

[zog]

Excellent comments by everyone, I agree with most of them. Simply put you need to know what type of trip the main is experiencing, what type trip unit is on the main breaker? Is the GFP part of the trip unit or an external GF panel?

I also suspect the main is the only breaker with GF protection (A common "cost savings" in the design stage that I never understood) in which case you should add GF protection to the feeders and have everything coordinated.

 

[brian john]

This is not going to be corrected installing GFCI. GFCI is typically a 120 VAC device, the GFP is most likely on a 480/277 ground wye.

1. I would check the setting on the GFP.
2. Verify the GFP is set per the coordination study, assuming there is one.
3. Test the GFP (have a testing company do this.
4. Very difficult to implement now, but multi-level GFP with Zone Selective control, will minimize this issue.

What size CB is causing the trips 20 or 30 amp or larger. Assuming it is 100 amp or larger minimizing trips can be almost impossible with a single main GFP.

 

[defears]

If all your equipment is one manufacturer, you may be in luck. Call them to have a study performed. They have been through this before. Cheaper and faster than an engineer.

 

[brian john]

A manufacture cheaper that a local engineering firm? NOT IN MY EXPERIENCE, support your local EE's and contractors.

 

[zog]

A manufacture cheaper that a local engineering firm? NOT IN MY EXPERIENCE, support your local EE's and contractors.

LOL, my thoughts exactly. Cheaper and faster than a local EE firem? I think not.

 

[charlie b]

The problem that some of our customers are having is that any small short downstream trips the main in the switch gear. . . .

Any small ?short?? Short to where? If you are talking about a wire within some motor breaking free and touching the case or touching a wire associated with another phase, then we are not dealing with a ?ground fault.? Why are placing the GFP device under suspicion? Do you have evidence that that is what tripped the main?

This could be a simpler problem, but alas one that does not have a quick or cheap solution. Others have already mentioned it: a coordination study. The basic truth is that when a short circuit takes place, a high current will flow from the source to the fault point, passing through (where applicable) the main breaker, the DP?s main breaker, the DP?s feeder breaker that serves the branch panel, the BP?s main breaker, and the BP?s branch breaker serving the load. Every single one of these breakers will see the fault current. Every single one is going to get the command to open (i.e., from its individual trip mechanism). Once the first one trips, whichever one it is, the others will no longer see fault current, and their trip mechanisms will abort their Open commands. Why would you expect any one of them to always win the race?

In a properly coordinated distribution system, the breakers are selected, and their trip settings adjusted, in such a way as to assure that the one that wins the race will be the one closest to the fault point. This type of coordination is mandated in certain emergency and legally required standby systems (refer to 700.27 and 701.18). But it is not required elsewhere, so your facility might not (and probably does not) have it. That could easily explain why a short in a 20 amp circuit can take out the main. I would say it is worth the cost to the owner to have the coordination study done, and to have breakers replaced or reset per the results of that study.

Welcome to the forum.

 

[erickench]

Are you sure it's a short and not an overload? I would suggest looking at the time-current curves for each circuit breaker and then comparing them. Selective coordination which includes the study of time-current curves indicates performance during overload and low level fault conditions. By comparing the time-current curves you will be able to determine which circuit breaker is faster. But for high level fault conditions other engineering methods must be used.

 

[G._S._Ohm]

causes thousands of dollars of down time and clean up or wasted product.

Depending on how often this happens on average, the company bean counters can decide how many thousands to spend in one lump sum on a study to get this fixed right the first time.

Another less theoretical way would be to monitor all currents everywhere with data loggers, all to be triggered on the fault, with a time readout to the nearest millisecond and all data loggers synchronized.
This should be done anyway to confirm the study results.

 

[rafael rivera]

We service industrial companies with large 2000 to 4000 amp switch gear. Of these companies, several are plastic extrusion plants. All large switch gear is required to have Ground Fault protection. The problem that some of our customers are having is that any small short downstream trips the main in the switch gear and shuts down the extrusion line or the plant. This causes thousands of dollars of down time and clean up or wasted product. Why do the individual breakers serving these down the line pieces of equipment not trip instead of the main? Is it a series versus fulling rated issue? It is because the AIC rating of subpanels is lower than that of the main switch gear, or is it totally due to the sensitivity of the ground fault protection in the main versus the sensitivity of branch panels and breakers? Can this problem be stopped without violating code or endangering employees? I am trying to fully understand where the problem is before offering a possible solution.

This is the first time I post or reply in the forum. I send all my respect and appreciation to the knowledge and experience that all you demonstrate. I have been involve in the construction and start up of a lot of refineries and petrochemical plants around the world and the only comment that I may add is to susjest to change the funtion of the EGF Tripping to a funtion of an alarm and install a high impedance neutral system at the substation to make this alarm operate. The coordination study is a must to set and adjust all other downstrem breakers to operated to confined faults at their locations.
This will prevent the main breaker from tripping and cause a nonorderlly shut down.

 

[G._S._Ohm]

I couldn't readily find the average success rate of a coordination study, so. . .

if the fault happens once/month or once/week you'd have to have 3 consecutive months or weeks of no failures after the fix to be 95% certain the fault is gone and 4 consecutive months or weeks with no failures for 98% certainty.

If the fault does come back within this interval you should probably have a data logger hooked up at the time so you can see exactly why it came back.