Low Ze, high PFC in industrial application

[alexcarnes]

Hi - just after thoughts on this please. I've been doing some testing on a 20 year old MCC. On the original EIC, Ze was reported as 0.04 R, PFC 5.75 kA.

Obviously this is a three phase supply, so shouldn't they have doubled the PFC and reported it as 11.5 kA, or is this business of doubling the phase fault current to estimate the short circuit current between all phases a recent thing?

The second problem - is there a reliable means of measuring such low Zes, or is it a fool's errand and you're best just calculating it? From a number of different tests, I'm gaining the impression that the real impedance might be nearer 0.03 R - a small difference, but that would make IPF 50% larger (230/0.03 = 7,666.7 x 2 = 15.3 kA). Even with all best practice my Fluke 1664 reads anything from 0 R to 0.03 R. Note that it's never given a higher reading, which is partly why I think the Ze on the EIC is suspect.

The third problem - I bet you know what I'm about to say! - there're a couple of 400v distribution boards full of 3 pole MCBs with 10kA breaking capacities. Zdb won't be meaningfully greater than Ze at the origin, one is fed off the bus bars, the other from a short run 16mm2 cable (Zdb is still so low I'm struggling to measure it - 0.02 R is commonly returned). Unless I've got something wrong or misunderstood something, the breaking capacity is inadequate - or was is OK to rate them on the single phase fault current 20 years ago? Even if the Ze on the EIC is correct, 2 x 5.75 kA = 11.5 kA.

Thanks.

 

[timhoward]

It is possible for the supplier fuses to solve the problem in some circumstances, as the PFC can cause a supplier fuse to blow before the fault current reaches the 10kA breaking capacity of the breakers.
So to be sure, more info about the supplier fuses would be needed. If it's an 100 amp supply or less, at first glance it seems that the fuses would interrupt proceedings at 5 kA if you had 15kA PFC.

 

[alexcarnes]

It is possible for the supplier fuses to solve the problem in some circumstances, as the PFC can cause a supplier fuse to blow before the fault current reaches the 10kA breaking capacity of the breakers.
So to be sure, more info about the supplier fuses would be needed. If it's an 100 amp supply or less, at first glance it seems that the fuses would interrupt proceedings at 5 kA if you had 15kA PFC.

Thanks, I wondered about that. From a purely electrical point of view then, if any fuses upstream of the 400V distribution boards had sufficient breaking capacity AND would blow faster than the MCBs' disconnection time at PFC, we'd be home and dry?

It's just that the regs don't make any mention of what's upstream of the DB. As far as I can make out, the letter of the law is that the MCBs should be rated for the PFC at that point?

 

[timhoward]

It's just that the regs don't make any mention of what's upstream of the DB. As far as I can make out, the letter of the law is that the MCBs should be rated for the PFC at that point?

I think 434.5.1 , 2nd paragraph, is the bit you are searching for.
I'm not saying it isn't a problem, just that it might not be a problem depending on what rating fuses are where.

 

[pc1966]

Having fuses or MCCB up-stream of the MCB can result in a safe situation, but you really ought to check with the manufacturer for the cascaded breaking capacity. Some are good at supplying this (e.g. the Hager commercial catalogue has tables near the back, Schneider has an on-line tool to compute selectivity and break limits, etc) but others are rubbish at real technical support.

In some cases you can take fuse graphs and plug in your estimated values to get some idea. This was one I posted here a bit back about domestic CU and how the typical 6kA MCB are safe (by design/certification) to 16kA PFC as a result:

Of course the details can be more complicated than simply looking at peak current limit of the fuse versus symmetric PFC equivalent, etc, but it shows the basic idea.

 

[pc1966]

Obviously this is a three phase supply, so shouldn't they have doubled the PFC and reported it as 11.5 kA, or is this business of doubling the phase fault current to estimate the short circuit current between all phases a recent thing?

It depends on how you measure it, and to some extend the nature of the supply (i.e. how impedances limiting the current arise).

If you have a typical measured single-phase L-N or L-E PFC value then you are seeing the result of U/(R1+Rn) or U/(R1+R2)

If you then have a "bolted three phase fault" you are seeing no impedance on the "virtual neutral" line as all three L currents sum to zero at the fault, so the limit is now U/R1

Assuming R1=Rn then that is double the L-N PSCC value, hence that approach. But if Rn is not comparable for whatever reason, it is not such an accurate measurement.

Another approach is you measure the L-L PSCC values instead, in this case to get the worst-case "three phase bolted" fault current you multiply L-L values by 2/sqrt(3) = 1.15 This approach has the advantage of not including Rn or R2, so it is more accurate, assuming your tester is safe & able to do the PSCC measurement on a nominal 400V supply.

 

[alexcarnes]

Thanks very much, both. I'll look into this further over the next day or two, it's getting close to my bedtime now.

Thanks again, big help.

 

[pc1966]

For doing big stuff you can get special high current loop impedance testers.

I have never used one but the DNO style folks have access. Basically it has both a bigger on/off test current to get more delta-V to measure, and uses 4-lead "Kelvin" clamps to measure the impedance on the terminals without crude corrections for the test cables and their contact impedance.

 

[alexcarnes]

It depends on how you measure it, and to some extend the nature of the supply (i.e. how impedances limiting the current arise).

If you have a typical measured single-phase L-N or L-E PFC value then you are seeing the result of U/(R1+Rn) or U/(R1+R2)

If you then have a "bolted three phase fault" you are seeing no impedance on the "virtual neutral" line as all three L currents sum to zero at the fault, so the limit is now U/R1

Assuming R1=Rn then that is double the L-N PSCC value, hence that approach. But if Rn is not comparable for whatever reason, it is not such an accurate measurement.

Another approach is you measure the L-L PSCC values instead, in this case to get the worst-case "three phase bolted" fault current you multiply L-L values by 2/sqrt(3) = 1.15 This approach has the advantage of not including Rn or R2, so it is more accurate, assuming your tester is safe & able to do the PSCC measurement on a nominal 400V supply.

Many thanks for that. Most modern testers with the high current two wire test allow one to measure phase to phase Ipf so I'd argue that's probably the best way.

In this particular application it's a TN-C-S and there's no measurable difference between the phase-neutral and phase-earth impedances (although as I mentioned earlier, the instrument gives different readings every time I do the test at this very low Ze!).

On the specific point of what was written on the original EIC though, I presume was technically incorrect to record the single phase PFC? It's certainly not what I was taught or in line with what it says in GN3 etc.

 

[pc1966]

Many thanks for that. Most modern testers with the high current two wire test allow one to measure phase to phase Ipf so I'd argue that's probably the best way.

In this particular application it's a TN-C-S and there's no measurable difference between the phase-neutral and phase-earth impedances (although as I mentioned earlier, the instrument gives different readings every time I do the test at this very low Ze!).

Really anything below about 0.1 ohm is not that accurate on your typical MFT. The better ones give an accuracy specification but it is not that good at low impedances, largely as the test leads and contact resistance can vary by 0.01 ohm quite easily.

On the specific point of what was written on the original EIC though, I presume was technically incorrect to record the single phase PFC? It's certainly not what I was taught or in line with what it says in GN3 etc.

There are two things:

• The PFC is L-E and that is "as measured" for the Zs/Ze value with the view to deciding if supply fuse/MCCB can meet ADS on an earth-fault, and to allow computation of down-stream Zs by adding R1+R2
• The PSCC is worst-case fault to decide if breaking capacity is sufficient. That would be typically be double the max measured L-N value for the 3P case. L-E can be higher in some cases (TN-S with many parallel earth paths, for example) but rarely that much higher to be considered.

 

[davesparks]

The second problem - is there a reliable means of measuring such low Zes, or is it a fool's errand and you're best just calculating it?

Yes, a high resolution loop impedance tester would be more suitable, Megger make a reliable one.

I have known of people introducing a known resistance into the test circuit to move the reading into a testers band of greater accuracy.

Note that it's never given a higher reading, which is partly why I think the Ze on the EIC is suspect.

Why do you think it's suspect? A change of 0.01 ohms in 20 years doesn't seem like the most unlikely thing in the world.

 

[alexcarnes]

Why do you think it's suspect? A change of 0.01 ohms in 20 years doesn't seem like the most unlikely thing in the world.

I just meant I believe it to be lower. If I really make every effort to zero the test leads till I get a consistent resistance measurement for them, give the instrument 20 minutes to stabilise, give it a good couple of hours to equilibrate to room temperature, and average out a few separate measurements, it actually comes to about 0.02 ohms.

I don't believe it's that low. The 50mm2 cable from the transformer to the origin calculates to a smidgen over 0.02 ohms (assuming the cable is laid 'as the crow flies') and then there's whatever impedance the transformer secondaries have...

It's still extremely low though!

 

[pc1966]

I have known of people introducing a known resistance into the test circuit to move the reading into a testers band of greater accuracy.

I don't think that works. Typically most meters are specified as something like +/-X% + N digits, your X percentage is the same on the total, once you remove the added R and its known range you get an even wider percentage of the result. Similarly once your R is subtracted from the reading, you still have the N least digits to consider.

Sadly none of the MFT I have ever seen even have the option for Kelvin cables, even as extra-cost option, which would get rid of a bit of the low R uncertainty at relatively low cost.

 

[pc1966]

I just meant I believe it to be lower. If I really make every effort to zero the test leads till I get a consistent resistance measurement for them, give the instrument 20 minutes to stabilise, give it a good couple of hours to equilibrate to room temperature, and average out a few separate measurements, it actually comes to about 0.02 ohms.

I don't believe it's that low. The 50mm2 cable from the transformer to the origin calculates to a smidgen over 0.02 ohms (assuming the cable is laid 'as the crow flies') and then there's whatever impedance the transformer secondaries have...

It's still extremely low though!

Have you got the transformer specification?

0.02 seems very low, my lowest was 0.04 at the busbar chamber off a 500kVA transformer, and 10 (+/-2 or so) kA fault current. Close to the 11kA the DNO gave me.

 

[alexcarnes]

Have you got the transformer specification?

0.02 seems very low, my lowest was 0.04 at the busbar chamber off a 500kVA transformer, and 10 (+/-2 or so) kA fault current. Close to the 11kA the DNO gave me.

Not to hand but I should be able to get it. It's on my to-do list!

I found the original cable calcs for the installation in the commissioning manual and they actually used a Ze of 0.026 Ohms for the Zs calcs! I can't remember the exact figures, but I couldn't help noticing that that's pretty much the value one gets for the ~65m loop of 50mm2 cable from the transformer to the MCC...

 

[davesparks]

I don't think that works. Typically most meters are specified as something like +/-X% + N digits, your X percentage is the same on the total, once you remove the added R and its known range you get an even wider percentage of the result. Similarly once your R is subtracted from the reading, you still have the N least digits to consider.

Sadly none of the MFT I have ever seen even have the option for Kelvin cables, even as extra-cost option, which would get rid of a bit of the low R uncertainty at relatively low cost.

Yeah I have never been too certain of the method either but it was championed by the late Engineer54, you could probably find his posts on the subject with a bit of searching.

 

[Marvo]

I've also heard of adding a known resistance to bring the measurement more into the accurate band of the instrument but I've never seen the advantages of this where as PC1966 point out you're basically robbing Peter to pay Paul. Personally once you get much below 0.05 ohms I'd be derriving the PFC by calculation referencing the transformer published specs and known supply cable impedances unless I had access to a high res loop tester.

 

[loz2754]

I noticed that the new Megger MFT X1 has a higher Ze resolution:

The MFT-X1 has extended the low end of the loop impedance range from 0.01 ohms to 0.001 ohms resolution and 50kA current calculation.

Even so, I don't think I'll be getting one at that price.

 

[alexcarnes]

I noticed that the new Megger MFT X1 has a higher Ze resolution:

The MFT-X1 has extended the low end of the loop impedance range from 0.01 ohms to 0.001 ohms resolution and 50kA current calculation.

Even so, I don't think I'll be getting one at that price.

Yes I noticed that too, although the Fluke 1664 that I use claims the same precision (it has a milliohms mode in the high current loop impedance test settings). I think it's a bit of a gimmick to be honest - the accuracy is poor at very low impedances so you just get the wrong answer to another decimal place, basically. At higher impedances it's not really significant.

I wondered whether the MFT-X1 would work any better than a tool like the LTW425, and accordingly emailed Megger to ask, but they didn't even dignify it with a response.