IET logo
 
IET
Decrease font size
Increase font size
Topic Title: Inrush Currents on IT Equipment
Topic Summary: Problems With MCBs Tripping
Created On: 09 December 2013 10:08 PM
Status: Post and Reply
Linear : Threading : Single : Branch
Search Topic Search Topic
Topic Tools Topic Tools
View similar topics View similar topics
View topic in raw text format. Print this topic.
 09 December 2013 10:08 PM
User is offline View Users Profile Print this message



keithredpath

Posts: 426
Joined: 30 March 2002

We have numerous circuits of underfloor busbar systems supplying 16 Work Stations each. Each work station has a PC and a VDU.

The supply cables are 10 mm SWA.
All the MCBs are 20 amps.
Most are Type Cs (Some are Type As)

The problem is after a power failure, it takes numerous attempts to get the MCBs to come back on. They spark and arc extensively.

I will recommended upgrading the MCBs to 40 Amps (Type C) but I still have a feeling the inrush currents are the problem. Is there any way to measure these inrush currents, so I can then recommend type D MCBs if this is found to be the problem. The earth loop impedances are all below 0.33 Ohms.

-------------------------
keithredpath
 09 December 2013 10:16 PM
User is offline View Users Profile Print this message



MickeyB

Posts: 181
Joined: 18 January 2003

32A Type D is standard for commercial jobs....... you could use Type C if you kept the load down. ...
16 is way too many PC's...... limit to 6 or 8 or so..... earth leakage/ surge etc. ...

Unfused taps? with 4mm tails to desk distribution module?
 10 December 2013 07:49 AM
User is offline View Users Profile Print this message



Ricicle

Posts: 851
Joined: 23 October 2006

What type of VDUs are they ?

-------------------------
Empty barrels make the most noise.
 10 December 2013 08:38 AM
User is offline View Users Profile Print this message


Avatar for gkenyon.
gkenyon

Posts: 4489
Joined: 06 May 2002

Many manufacturers of PCs and associated equipment, that are intended for commercial / industrial use, supply this information (often on-line) as a matter of course these days, for precisely the reason you state.

If no information is forthcoming, then the following "rule of thumb" might be useful for maximum inrush currents:
- Flat-Panel Monitors up to 322 : 15 - 20 A inrush
- Small SMPSU (<80 VA): 10 A inrush
- Larger SMPSU (or PCs) 80 - 650 VA: 20-25 A inrush
- Heavy SMPSU >650 VA (or Server, each input if more than one PSU): 35 A inrush.

UNLESS the equipment is old (before around 2003), in which case even 240 VA SMPSUs sometimes had inrush currents in excess of 120 A !!!!

You need to assume that the maximum inrush current will flow for 20 ms, and (I probably don't need to add) that whilst for tripping times in fault, we use the "worst-case" of the operating curves for OPDs (i.e. for mcb's Type B 5 In, Type C 10 In, Type D 20 In), when looking at inrush, we need to consider the "best case" (i.e. for mcb's Type B 3 In, Type C 5 In, Type D 10 In).

Because a real OPD will operate somewhere between the "worst-case" and "best-case", if your calculations show something just above the "best-case", swapping the OPD is often a way of stopping a nuisance-tripping problem caused by inrush currents.

Measuring inrush currents is not easy, best left to the manufacturer or specialist test lab.

Inrush currents aren't just a problem for power-on, they can also be a problem with temporary power dips (brown-out) or surges, particularly with IT equipment, as they are often a result of charging large capacitors in the PSU.

And finally - there is another (some might say, a technically more "elegant") solution: use a fuse instead of an mcb! Simply because the fuse doesn't have the discontinuity in its operating characteristic as mcb's do: where in the mcb the magnetic element takes over from the thermal element.

-------------------------
Eur Ing Graham Kenyon CEng MIET TechIOSH
 10 December 2013 12:59 PM
User is online View Users Profile Print this message



AJJewsbury

Posts: 11768
Joined: 13 August 2003

Another approach is to time-delay on switch on, each load with a different delay, so you're only having to deal with the surge from one item (+steady load from the rest) rather than sum of all the surges at the same time. There are readily available 'sequential start' power distribution units (PDUs) for rack mounted equipment from various manufacturers and some UPSs have a programmable delay before switch-on when power returns - all to address this very problem. If you've a distributed system you can achieve the same effect with an on-delay timer at each outlet (floor box say), each set to a different delay.
- Andy.
 10 December 2013 05:07 PM
User is offline View Users Profile Print this message


Avatar for gkenyon.
gkenyon

Posts: 4489
Joined: 06 May 2002

Andy,

Just to add that the delay-starts are OK for solving the power-on tripping (as per original question in the thread), but may not help if tripping is also occurring during brown-out and surges.

And also add that the use of fuses instead of mcb's of course has drawbacks from an operational perspective, if indeed they can be retrofitted in the relevant DBs?

All due to progress ;-)

-------------------------
Eur Ing Graham Kenyon CEng MIET TechIOSH
 10 December 2013 06:02 PM
User is offline View Users Profile Print this message


Avatar for OMS.
OMS

Posts: 19900
Joined: 23 March 2004

Wouldn't you need to consider a crest factor as well Graham ?

ie the maximum inrush current (I@20ms) that an MCB will carry is :

In x minimum tripping threshold for MCB Type x crest factor

So for say a 20A Type C in the OP

20A x 5 x 1.414 = 142A

Say 6 No desktop PC's or 2 - 3 servers (max no of SMPS).

It's hardly suprising a 20A Type C is tripping with 16 PC's connected.


Regards

OMS

-------------------------
Failure is always an option
 10 December 2013 10:49 PM
User is offline View Users Profile Print this message


Avatar for gkenyon.
gkenyon

Posts: 4489
Joined: 06 May 2002

OMS,

Good point, but I think the answer needs to be ... I suppose it depends.

I suspect that inrush currents quoted by many manufacturers may be divided by the Crest Factor, which may be different from SQRT(2), to make them appear lower in the first place: not sure it's "regulated" by a standard.

-------------------------
Eur Ing Graham Kenyon CEng MIET TechIOSH

Edited: 10 December 2013 at 11:00 PM by gkenyon
Statistics

See Also:



FuseTalk Standard Edition v3.2 - © 1999-2014 FuseTalk Inc. All rights reserved.