5

We currently have two cabinets of equipment at our datacenter, and each cabinet is served by two 30A circuits. The facility is relatively new, and each circuit is (supposedly) redundant at the source.

Most of our equipment uses redundant power supplies, but some devices not deemed super important have been left without redundant power, so that our mission crit equipment can continue humming if we lost an entire circuit.

However, based on best guess (which includes accounting for a 20% redundant-to-non-redundant efficiency improvement based on some prior reading), two of the four circuits would probably be over the 80% mark in this scenario. Specifically, I would have estimated readings of 21A, 24A, 25.5A, and 28A.

Are loads such as these acceptable for a disaster scenario like this? I don't know what it would take to lose a redundant circuit, but I'm trying to plan for it.

Matt Beckman
  • 1,502
  • 18
  • 33
  • 30a @ 240 or 120v? – Mark Henderson Mar 20 '14 at 23:30
  • Single phase or three phase? – toppledwagon Mar 21 '14 at 00:15
  • @MarkHenderson 30A 120V – Matt Beckman Mar 21 '14 at 00:27
  • @MattBeckman if I have time later I will expand this to a proper answer; but 30a@120v@80% isn't very much power, but is probably OK, but > 80% I'd be concerned – Mark Henderson Mar 21 '14 at 01:06
  • Why are you allowing this to be your problem ? Presumably you have an agreement that says you can use up to X amps power with your datacenter provider. If they have represented they will provide you with up to X amps, it is reasonable to be able to use the full amount - and if you have any concerns, they are the only people who can provide you with an answer. If you are building a DC yourself, you really need to ask your sparky and not rely on random advice from the Internet. – davidgo Mar 21 '14 at 01:18
  • @davidgo At this facility, overage fees are charged if your 95th percentile of power usage each month goes over the 80% mark (per circuit). So, the only "guarantee" is likely the 80% mark. However, you might have a point about simply asking them directly what their breaker situation is at the source. – Matt Beckman Mar 27 '14 at 00:18

1 Answers1

2

I'm not an electrician and certainly shouldn't be considered as an authority on such matters, but I'll jump in anyhow.

There are different kinds of circuit breakers (thermal, thermal magnetic, others). They behave differently under different types of loads. (I've seen sketchy welders that draw 22A that don't blow a residential 15A circuit because it only ran for short bursts, for instance).

Also there are "80%" and "100%" circuit breakers. A 80% CB may blow if you are over 80% load for more than 3 hours. I've also been told that circuit breakers aren't typically tested as though they're precision devices - they don't go through a test suite where it passes QA if it can run at 99% of the rated load and blows at 100% of the rated load. Instead, they're tested such that they pass if they don't blow at 80% load and do blow at 100% load, and what happens in between is ... undefined.

Another issue is that you may be at 82% of max load now, but what if the DC is really hot? Often servers have fans that change speed depending on temperature -- and when they spin faster they'll draw more power, so now instead of 82% of max load you're at 88% of max load. When things go wrong they tend to go wrong in surprising and unexpected ways.

So, depending on the equipment, probably you want to size circuits to either have non-critical stuff on them or make sure that they're under 80% under all scenarios.

http://www.wincogen.com/CB

chris
  • 11,944
  • 6
  • 42
  • 51