Fuses in parallel

This includes types of overcurrent, Short Circuit and RCD protection
Post Reply
User avatar
gregmcc
Site Admin
Posts: 189
Joined: Sun Apr 12, 2020 4:45 pm
Answers: 0
Has thanked: 18 times
Been thanked: 49 times

Fuses in parallel

Post by gregmcc »

So a bit of a technical question,
Keep in mind this is in a WORKS environment not an installation.

Network electrical engineer (I use the word engineer loosely here as it's a job title and the person may or may not be an electrical engineer - I don't know) has specified a service fuse to a new factory at 400Amps. As there are twin 185 AL cables per phase, the engineer has decided that in order to terminate these 2 cable in to the service pillar two of the fuse disconnect units will be paralleled together and each fitted with 200A fuses

So I'm assuming that he simply went 200A x 2 =400A.

I don't pretend to know everything about everything BUT I was lead to believe that fuses in parallel simply do not work like that, is there a specific formula or table to work out the effective fusing current?

The fuses are NH3 style din type, I don't recall the exact brand, but all 6 fuse are the same.
AlecK
Posts: 912
Joined: Thu Apr 16, 2020 11:24 am
Answers: 5
Has thanked: 2 times
Been thanked: 351 times

Re: Fuses in parallel

Post by AlecK »

Not my area of expertise either.
And I don't like the idea much.
However not significantly different from a dual-feed set-up with a bus-tie [Fig 7.1].

I think the suggested configuration will meet requirements for overload protection.
Each bit of cable is protected by a suitable device, so cannot carry sustained overload.

For short circuit and fault protection (which ESRs require the network to provide for mains);
both devices would need to operate to disconnect within required time.
For faults partway along the run, likely one fuse would blow before the other, due to different impedances in each leg.
But as long as both blow within required time all good.

Fault protection is 5 s [5.7.2]; so probably OK
The SC calc in 2.5.4.2 would need to be worked through using CSA for one of the 6 conductors; and compared to the operating characteristics for the fuse. Probably OK.
JBrtc
Posts: 2
Joined: Mon Apr 12, 2021 11:02 am
Answers: 0
Has thanked: 0
Been thanked: 0

Re: Fuses in parallel

Post by JBrtc »

Interesting consideration. From a theory perspective on the actual operation, rather than compliance:

My understanding is that fuses directly in parallel don't really work as there's not likely to be any way to enforce even current flow between them. Any tiny difference in the wiring to each fuse or the resistance of the fuse in the holder will lead to one carrying more current than the other. Not really an unsafe situation, just that the rating will be something unknown between 200A and 400A. If everything were perfectly balanced or manually adjusted to be correct, it 'could' be ok, it's just a kinda crappy idea. Fit a different fuse, have the holders age slightly differently and the balance between them could be significantly different.

On the other hand, if each individual cable has an individual fuse then I can't see how there would be an issue given that the two cables will likely have enough impedance to enforce a reasonable amount of current sharing. Each cable is ultimately protected correctly by its own fuse, as Alec points out, and the impedance of the cable itself is likely to be the biggest factor in determining how balanced the current will be between the two cables. Resistance of the fuse and holder is likely to be 1mR or below (else it'd be pretty damn warm at 200A). If the cable has 1% Vd on it then that's ~10mR at 200A.

In theory, I guess an argument could even be made that individually protecting parallel conductors is better than summing them, as any damage or degradation that changes the current balance can't cause an overload. No idea how much of a real-world consideration that would be, though.
User avatar
gregmcc
Site Admin
Posts: 189
Joined: Sun Apr 12, 2020 4:45 pm
Answers: 0
Has thanked: 18 times
Been thanked: 49 times

Re: Fuses in parallel

Post by gregmcc »

The cables are not individually protected, the load side of the fuse holders at the network end are bus tied together, at the main switch installation the cables are directly bolted to the live side of the main switch.
the set up effectively 2 parallel fuses(per phase) mechanically linked together so both fuses have to be closed together.

Anyway, I can remember when I was an apprentice (a really long time ago) we were told that putting 2 x 10A pieces of fuse wire in a re-wire able fuse holder does not equal 20A, it equals much more. I was thinking that the same situation would apply here?
AlecK
Posts: 912
Joined: Thu Apr 16, 2020 11:24 am
Answers: 5
Has thanked: 2 times
Been thanked: 351 times

Re: Fuses in parallel

Post by AlecK »

Would need the fuse link characteristics to work that out
And not sure my brain would be up to it.
HRCs are still basically thermal devices, but far different from a simple bit of tinned copper
In fact my understanding is that each HRC link is, internally, a whole bunch of small, carefully-engineered, fusing elements in parallel.

But I still believe each "leg' is protected for overload.
Regardless of minor imbalances in load-sharing between the two legs of each phase-pair; no way can more than 200 A be drawn through any leg for long enough to cause damage.

For SC & Fault protection; if we consider a fault part-way along the run; both fuses will blow .
Maybe not at exactly the same time, due to the different impedance in each path to that point;
but the difference will be tiny.
Faults downstream of the combined phase-pair will similarly cause both fuses to blow.
Or for A-A Shorts, likely 3 or even all 4 fuses.
JBrtc
Posts: 2
Joined: Mon Apr 12, 2021 11:02 am
Answers: 0
Has thanked: 0
Been thanked: 0

Re: Fuses in parallel

Post by JBrtc »

gregmcc wrote:
Wed Sep 15, 2021 5:01 pm
The cables are not individually protected, the load side of the fuse holders at the network end are bus tied together, at the main switch installation the cables are directly bolted to the live side of the main switch.
the set up effectively 2 parallel fuses(per phase) mechanically linked together so both fuses have to be closed together.

Anyway, I can remember when I was an apprentice (a really long time ago) we were told that putting 2 x 10A pieces of fuse wire in a re-wire able fuse holder does not equal 20A, it equals much more. I was thinking that the same situation would apply here?
I can't figure that that would be the case. Assuming the two piece of fuse wire are the precise same resistance and carry the same current, they should clear at twice the current of the 10A wire alone. Without seeing the curves, I don't know that 2x 10A fuse wires would be identical to 1x 20A fuse wire for the entire range of fault currents, but I'd guess it wouldn't huge hugely far off. Not enough to be hazardous, I'd think.

On the other hand, if you imagine the scenario where the two lengths are extremely different, say ones ultra-short and then the other takes a route that makes it twice as long, then the purely theoretical answer is that the short one will carry twice the current of the longer one. Given that 10A rating, the equivalent would then be roughly what a 15A fuse would be. If it takes 100A for 1ms to clear the 10A wire, then that scenario would be reached at 150A of fault current (100A in the 'short' one, 50A in the 'long' one), with the short one clearing and then the long one taking all the current and clearing shortly thereafter. That raises an interesting consideration that the trip curve of such an arrangement might look a little bit odd, but I can't see how it would ever be 'worse' than the perfect combination case.

I found a Cooper Bussman application note that mentions some of these considerations:
https://bussfuses.net/pdf/APPLICATION.pdf
It says paralleling can be used to obtain higher current ratings than existing ranges or reduce stock variety but that care should be taken to keep the connections as symmetrical as possible for current sharing purposes. It notes that the temperature coefficient of the fuse assists which is an aspect I hadn't considered. Positive temperature coefficient devices get more resistive as they get hot, which is one of the ways fuses are engineered to clear faster. Negative temperature coefficient devices (like a lot of traditional semiconductor devices) don't share as well, as the hotter they get, the lower resistance, more of the shared current, more loss, gets hotter, etc. This is why failures in things like UPS units will often only have one blown-up MOSFET, that's the one that had the 'thermal runaway' condition.

The app-note says that it's prudent to allow for a 5% de-rating for each parallel path, which should indicate how accurately the sharing is expected to be. So given that, I'd say it seems like it would likely be fine, providing they're next to each other and connected largely the same.

Again, that's all from a pretty purely theory perspective, of course, any time I've needed to design fusing it has been within a range where I could just use a single fuse.
Post Reply