Single Phase UPS Output Short Circuit

[richanton]

We have been mulling around an apparent problem with UPS designs. We generally always use industrial UPS's with single phase output at 120 volts going to a circuit breaker panel. The question that came up is that there is not enough short circuit to trip a branch breaker on the panelboard. This issue is even more noticeable on a European system that has 230 volts, single phase as the output voltage. I read a prior post on the subject that implies that the response of the UPS when it sees a fault on an output branch breaker is to switch to the bypass mode, under which normal system short circuit would be available. This is all somewhat confusing. What I am trying to figure out is once the UPS does transfer to the bypass system, what will trip first, the branch breaker or the main breaker at the output of the UPS. Do I just treat both breakers as if it was a normally coordinated system, or is there some other factor involved.

 

[davidbeach]

You have to look at the coordination of the breakers. The UPS goes into bypass in a fraction of a cycle as the output inverter can't supply the fault and at that point you have a situation where you can essentially ignore the UPS and just look at the bypass circuit. You might also want to ponder the condition of a fault on an output circuit while there is no alternate source for the UPS to switch to bypass, such as during the loss of utility power.

 

[itsmoked]

Would seem the UPS needs to 'command' a breaker to trip when it detects an overload.

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[davidbeach]

Keith, that could be done, but which breaker? With the UPS in bypass there is a possibility of the breaker nearest the fault clearing first. Without bypass it would be extremely difficult to determine which breaker to trip. If the bypass circuit is not available, the UPS will protect itself and shut down, no worries about tripping breakers to protect the UPS, but no loads being served.

 

[itsmoked]

I see. Here's a long question with probably a short answer..

Lets say you have a UPS that is 10kVA output. 10000 / 120 = 83.3A

It runs a panel with (8) 20A breakers.

Loading: Each breaker is running 10A loads. 80A

Now you get a fault on one load. The UPS can't supply more than 83.3A. It is already supplying 80A there isn't enough supply to get the faulted circuit to clear the breaker. UPS shuts down and instantly becomes a expensive waste of money as it's not "U"ing anything anymore.

Is there a current monitoring solution to this? Are there 20A breakers that are thermal magnetic that would let you adjust each circuit to its maximum so when it starts drawing 13A it can trip?

And thanks for that info David.

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[rbulsara]

richanton:

Your observations are pretty much accurate. Welcome to the imperfect world. As you mention best you can do is try coordinating the load breaker with the UPS bypass breaker.

If you are lucky the, UPS will go to bypass without dropping the entire load and hopefully the brach breaker will open. If it does not, the bypass breaker would open still droppnig the entire load.

This is no worse than if there were no UPS and trying to coordinate say a 60A breaker with a 20A branch breaker. Such MCCBs are difficult to coordinate at current in excess of 1000A or 2000A and there will always be a race.

The point is you are not going to have any better coordination without an UPS. Now UPS is installed primarily for having a reliable power. UPSs are of little help if the fault is on the load side. For the size of UPS system you are describing, the reality is it serves a small load to begin with. So losing a part of it or entire UPS will not make much of the difference. If the UPS is large compared to the brach circuit rating, you would have a better chance of opening a 20A brach CB on a fault and still stay on the UPS.

Having said that I have had UPS manufactures demonstrate that they can clear a short circuit (even if it goes to the bypass) on a branch circuit under a controlled test. Two of them passed and one did not want to do it. If you want to be really sure, the only way is to perform a test on a like unit at the factory.

One of the leading manufacturer of large UPS system will not do the SCC test citing safety/liability reasons. Incidently they are the ones we specify most because their reliablity and service are otherwise proven.

 

[richanton]

Thanks for the observations. By the way, one of the alternate suggestions is to use fuses for the branch circuits instead of breakers since fuses are faster. I think it's time for me to do a white paper.

 

[rbulsara]

rich:

There again (with fuses) you will have competing goals. Now you will be fighting for space. For one or two circuit it may seem OK. Think of larger data centers.

Have you seen a panelboard with (42) 20A fuses? I have not. If you find one imagine putting several of them on data center floor, a premium space.

Beyond all this technical analysis on paper, also take a stock of history of reliability or failure rates of so many systems in place using circuit breakers. You will find that circuit breaker system would fare just as good. There are always reasons why certain type of systems have become so widely used and accepted. Granted none them are perfect. But all things considered, only best value systems become popular.

Much have been said about fuses vs. breakers in other threads, I would imagine, so I will not go there.

 

[davidbeach]

Keith, your description is what happens if there is no bypass source available. If the bypass source is available, the UPS will switch to that to allow that source to supply the necessary fault current to clear the breaker(s) necessary to interrupt the fault.

I'm not aware of any circuit breaker that would allow an adjustment down from 20A; lots of shunt trips and overcurrent relays could get you there but at tremendous cost. A much more practical solution would be to use dual corded equipment with completely independent sources to the two sides, with full load capacity available on both sides. That way if there is a fault on one side, the other side can carry the entire load until the fault is removed.