BS 3871 Miniature Circuit Breakers Let Through Energy

Let through energy is a bit problematic for breakers - a fuse gets faster as the fault current rises,  and tends towards a constant let- through energy (I squared r times time, but the resistance r is a fixed parameter of the fuse, as is the weight of metal to be raised to melting point to start it breaking)

For an MCB, although at low currents increasing faut current increases breaking speed, there comes a mechanical limit of how fast the contacts can be separated,and it more or less stops getting faster.

For this reason, while fuses can be cascaded, and if the thin one blows, the fatter one behind it will not,  for all values of fault currrent, for MCBs no such statement can be made. Indeed it is common to find a random selection of breakers open under fault, and for really high current faults the fuse at the origin fails instead.

So I2t is actually a function of the prospective fault current, and is either depicted graphically, or listed as spot  values . For mcbs  made to the IEC standards since 1999 or so this is a spec parameter, and guaranteed by design. for earlier devices it isn't

IF you put in the maximum breaking time of 10ms you will come up with a very large minimum cable requirement,and may well (incorrectly) conclude the cable is not protected.

It may be cost effective to replace the breakers or back them with a fuse, but first post the make and rating of breaker here - some folk on here have breaker test data going back to mists of time and may be able to advise.

Mike,

Very well put, but to further complicate the issue the let through energy for a circuit breaker can potentially be greater with a smaller fault current. Normally the maximum let through energy can be calculated from I²t where I is the maximum fault current and t is the opening time of the contacts (in this case I assume the 10ms of the OP). However if the maximum fault current is not much greater than the instantaneous (i.e. short circuit) trip time of the CB then a slightly lower current which remains below the instantaneous trip time, keeping within the overload part of the characteristics curve with a much longer trip time, can end up with a higher I²t since the trip time may be in hundreds of ms, if not seconds. 

I suspect that this latter condition will not be a problem in this case as this is for a specific fault current below the maximum that is already within the scope of the existing installation, so it will be sufficient to do a calculation for the increased let through energy with the new short circuit current.

Alasdair

mapj1:


IF you put in the maximum breaking time of 10ms you will come up with a very large minimum cable requirement,and may well (incorrectly) conclude the cable is not protected.

 

I think that should be the minimum time ie the definite minimum time to operate (DMT) of the MCB is generally assumed to be 0.01 seconds provided the fault current is above the minimum value to give disconnection within 0.4 to 5 seconds - and also below the breaking capacity of the MCB (which may make a demand on further upstream protection providing a cut off characteristic)


Regards


OMS

curse the short duration edit window. or I would go back and reword to be less ambiguous. So its a second post.

To be clear, the 10ms is the limit of the   curves  no longer in the current BS 7671 show that as the bottom of the graph, these  curves are the maximum time it might take, at a particular fault current.

The problem of the way the graph is drawn  does not help us to work out how it may behave when when driven harder - when  not only are we in the fast magnetic region, but we are saturating the magnet core so the force stops increasing much and the contacts are moving as fast as they can, all we can say is that we would be below the 10ms line, but we cannot say by how much. Also a time to reach an open circuit is not the full story, as once an arc is established there is an effective series resistance (a complex and current dependent one, that is almost a constant voltage but not quite) that starts to affect the let through.

Thanks for your responses guys, I tried to be fast in the first post and did not provide enough information. The fault current is in excess of X times of multiples quoted for the type of MCB. So as per the regulations a trip time less than 0.1s (instantaneous) should be proved with energy let through. I have contacted Dorman Smith which appear to be the manufacturers of these MCB with no positive response - picture attached. They did not have any data. 


There are about 100 MCBs of the same make and manufacturer so it would be of benefit if there is a way around the lack of data. I have not looked into the fuse providing backup but I will, discrimmination is a must in this installation. Would not want to loose the whole board with a blown fuse.


Indeed as Mike mentioned 10ms gives a very large number (for 3kA and 10ms -->> energy let through is 90000).


My only hope is now if someone has any brochures stored somewhere with let through tables/graphs. Otherwise I will be replacing them


Thanks,

Mike

OK - I think we are getting at cross purposes here, chaps


The so called instantaneous tripping time of the MCB is 0.1 seconds (100ms)  - it may operate faster, but we don't know how fast


The definite minimum time to operate is 0.01 seconds (10ms) - it cannot operate any faster


For the region in between, the only accurate assessment is the energy let through determined by testing and published by the manufacturer (ie the I2t characteristic)


For an adiabatic assessment, provided you have enough current to drive the operation of the MCB (and that must be true in this case as the concern is of in increased fault level being applied to the MCB) then if you assess the system based on definite minimum time to operate then you have the probable worst case energy let through in the absence of manufacturers data.


the derived 90,000A2s from above is quite modest - any conductor above 2.0mm2 will handle that OK


Regards


OMS

Let us work to a minimum permissible copper CSA.

cross sectional area is S in the text book, in sqmm.

S= root(I2t)/k

K~ 145 for rubbers  and 115 for PVC insulation , with certain assumptions about starting temperature.


Take our I2t value  of 90,000  (3kA and 10msec )and square root it = 300


cross section minimum is  300/115 = 2.61mmsq.

, so that implies with that breaker 2.5mm is not OK, but you are comfortably OK with all cables of 4mmsq up and larger.



I don't for one moment  think we need to use 3000 A for 0.1 seconds in this case but if you did then the 300 becomes more like 1000, and the minimum cable size is probably 10mmsq....


This does not make the installation dangerous to use, but means that if you have  short circuit fault within a short distance of the origin,  you need to inspect to see if the insulation near the CPC is damaged. Equally if the cable is damaged within a few metres of the box with the breaker in, it is not such a hard job   to replace it and as you get further out the PSSC falls due to cable resistance ( as a ready reckoner figure for cable resistance for PSSC assume 16 milliohms/metre for a single core of 1mmsq and scale for other lengths and cross sections- this under estimates, use 19 or 20 milliohms per metre for volt drops, this tends to over estimate slightly.)


(16 is a nice number for rules of thumb ,  1.6mm is nearly 1/16 of an inch and almost exactly the diameter of 16 standard wire gauge and 16 Birmingham sheet metal gauge, though not american wire, but AWG 14 is ~ 1.6mm dia. )


The saving grace is that for very thin wires the assumption that no heat leaves the copper while it is heating is not true - as the diameter shrinks, the surface area to volume ratio favours more rapid heat transfer - We know this is also true in reverse, it is why thin bacon cooks faster than thick sausages (I often make a cooked brekkie on weekends you see, so I know this sort of stuff ..)

So the idea that all in an instant the whole length of copper rises by 100 degrees and has enough stored energy to damage the surrounding insulation while doing so is something of a simplification. (and the assumption that the Adiabatic formula works from 0.1 sec to 5 sec hides some similar "let us  assume all wires are the same diameter" slight of hand)

OK - I was assuming 90C/250C (ie XLPE insulation and K of circa 150) so you should be OK for 2.0mm2 and upwards - which should exclude a whole load of circuits from your assessment.


Which leaves you with the lighting circuits - how far down the cable do you need to go, if the 3kA value is present at the DB, using 1.5mm2 conductors before you are safe again - not far I suspect.


If you are looking for reason to keep the existing despite the mains upgrade, then you will need more data, but don't be surprised if it doesn't help much - do you know for example if it was compliant before the upgrade took place (ie at the previous fault level) - have you tried a random MCB is say Amtech (or data from a different supplier)


If you are looking for a reason to scrap the existing, then absence of that data is as good a reason as any


How accurate is the 3kA value, and is that subject to cut off characteristics from upstream devices


For information, if these are the old black cased loadmasters to BS 3871, they have an M3 (ie 3kA rating) and for all sizes to 70A, they have A2s values of 10,000, 40,000 and 90,000 for 1kA, 2kA and 3kA fault levels, respectively


Regards


OMS

I should also add that if they are slightly more modern Loadmodule or Loadlimiter devices (particularly the latter at 9kA rating) they will have significantly better cut off characteristics at 3kA - something like 10,000A2S on a 10A Type C on a 3kA fault level


Regards


OMS