EPO Nuisnace Trips

Status
Not open for further replies.
bhsrnd

bhsrnd

Senior Member
Location
Fort Worth, TX
I figured since the threads covering cords under the floor keep going astray that I'd start a new one concerning EPO systems.

I keep seeing the phrase "nuisance trips" in various posts. Is there some ballpark number somewhere indicating the amount of those trips encountered every year by data centers? I ask because we have been in this building for 10 years now and have had no such occurrence. *knocks on wood*

Should one type of component versus another be used when designing an EPO system to lessen the chance of false trips? (e.g. - PLC versus a simple contact/relay setup)
 
There is no national database of trips.
Every one that I've heard of was a result of human error. Never have I heard of one that was done to save life.

I design them with relay based logic. Always power to activate, so that if someone trips a control source breaker, the shunt tripping doesn't happen accidentally.
 
Last edited:
EPO Nuisance Trips

EPO Nuisance Trips

Nuisnace trips are a direct result of 1 of 2 things. 1st is ambient temperature. All overcurrent Protection Devices are rated at 25 degrees C. If you raise the ambient (summetime) the OCPD can not carry rated current, thus tripping out the overloads.
2nd is having FLA (full load current) too close to the OCPD rated current. All OCPD are rated to carry 100% continous.
In the case of a Circuitbreaker, it is only rated for 1 trip in a Fault condition. Once that breaker trips, all the published specs go out the window. You now have carbon & pitting that occurs, and degration of that breaker, causing additional / multiple nuisnace tripping. Best to replace it , when that happends.
 
EPO operations have nothing to do with OCPD. They are a required feature to comply with 645.10 (Disconnecting Means) if the room is going to take advantage of 645. The purpose is to provide a grouped means to disconnect all dedicated power and HVAC to the room. A typical set up is using a single means of control (push botton) to operate a contactor or other device which shuts down the UPS and HVAC systems.

With no specific data to back this claim up, the most likely cause of nuisance trips in human error.
 
Background: An IT professional, with most of my roles over the last couple of decades involving keeping IT and networking systems running, usually for large or very large organisations.

EPOs are a necessary evil, I too have never seen or heard of one used appropriately, though I've known them to be accidentally triggered several times, once very nearly by me leaning on a wall and missing an EPO by inches.

Having said that they are a necessary evil, I don't actually understand what their purpose is in a data centre.

There are two distinct types of EPO systems, the aformentioned shunt variety, and I've never had any problems with this arrangement, though my sample size is one. The series type I've had fail a couple of times, and thus the facility power collapse. On one occassion it was contactor failure, on the other accidental wiring damage. The series arrangement is great for a workshop or whatever where there are real safety-of-life issues, and the need for the safer system overrides the need for continuous electricity. For a data centre, the probabilities fall the other way.

There is a third occasion I've seen a contactor fail, but on that occasion it wasn't the contactor's fault. The neutral had come adrift, so the contactor (as well as everything on the PDU) got subjected to overvoltages, and eventually failed. Mind you, by the time the contactor failed, there were many dozens of switch mode power packs that were dead, so loss of contactor was no real loss... That was a very expensive day.
 
Fire dept.'s like 'em... Talking to a few they told me of a few UPS battery, and server room fires where they came in handy. As it was the Fire dept that required the EPO's, not our electrical inspectors...

For some reason these guys like to kill the power before they start bashing holes in walls and filling them with water...:rolleyes:

I haven't done any in a while but the few I have done I could use the provided dry contacts off the UPS for the purpose and used a shrouded NO push-button.
 
I read an article a while back (I think it was IAEI) that talked about the purpose of an EPO.

It assumed the fire was electrical in origin. Shutting off the source of ignition was the goal. All that cable going up in flames under the floor.

hopefully someone else remembers that same article or can confirm or counter what I remember.

edit add;

I see this was discussed in another thread about safety of firefighters, which relates to the purpose I remember of the EPO.
 
Last edited:
For some reason these guys like to kill the power before they start bashing holes in walls and filling them with water...
Click to expand...
If that is really the concern and I am the fire fighter, I am not going to assume that the EPO is functional...I will fight the fire the same way with or without the EPO. If I think that there is a serious electrical hazard in fighting the fire, then I will protect the exposures and let the data center burn, however it is likely that my preplan will call for fighting this fire using wide pattern fog stream in place of the more normal straight or solid streams. Tests have shown that water fog can be used on energized electrical equipment at voltages in the 100k range. The plan would also call for the firefighters to stay out of the area until the power is known to be off. While the fog stream is not an electrical conductor, the run off water is.
Don
 
1. George, thank you for welcoming me.
2. UL489 dictates that circuit breakers be calibrated at 40 Deg. C ambient. My mistake, for fuses it is 25 degrees C.
3. Wanderer2001, my bad, appologies....
 
ron said:
There is no national database of trips.
Every one that I've heard of was a result of human error. Never have I heard of one that was done to save life.

I design them with relay based logic. Always power to activate, so that if someone trips a control source breaker, the shunt tripping doesn't happen accidentally.
Click to expand...


That goes against the "fail safe" logic we usually use when designing emergency shutoffs for other equipment. But, Article 645 doesn't require the EPO to be fail safe. And given the number of false trips vs. the number of times these are actually used in an emergency, I think I like your suggestion of making them "power to activate".

Steve
 
EPO?
Is that "Electrical Power Off", just trying to learn.
I am thinking off something like a pushbutton near an entrance door which when pressed will de-energize all equipment in the room. Is this correct?
 
EPO - Emergency Power Off. (beaten to it!)

You'll usually find an EPO outside each door (similar to a fire alarm button), and then several scattered around inside the data centre or whatever, plus one on the PDU itself.

If you ever get to fit EPO buttons, please, please, please make sure they are shrouded, and not just mushroom emergency stops like you would fit in a factory.

The cost of an accidental EPO shutoff can be enormous. (Most) computers systems dont like just being switched off, and when they get power back it may take many hours (possibly involving restoring data from tape) before they are usable for business purposes again. Some data may never be recovered, so for example someone might "lose" their house or car insurance, as the records relating to that transaction can never be recovered. They think they are insured, but aren't, and it's not till a few weeks later when they bug the insurance company about no paperwork that the fact they are uninsured is discivered.

Generally, you need to be talking about a reasonable size of data centre before the size warrants "proper" power design, and that means the data centre is supporting five, six or even more zeros worth of business a day, with hundreds, thousands, or tens of thousands of workers (locally, around the country, or maybe around the world) reliant on the computer services provided. The cost of these workers doing nothing is also non-trivial, not to mention customers who cant be served or processed, bad PR, overtime to fix the mess, automatic bank payments not made, cashflow,... the list goes on.

Thus as an IT pro to the electrical pros - please make sure the EPO wont embaress us all.
 
Last edited:
Status
Not open for further replies.
Top