Hi all!

I started thinking about the different systems for grounding and stumbled across a potential problem.

In the US and much of Europe the standard way of grounding a building is to connect the ground wire to the neutral at the panel, meter or main fuse. (I won't use the terms grounded and grounding wire, since I only end up confusing myself. I'll also ignore the ground rods in this case.)

At the panel you have single pole breakers, which trip if there is too much current in the hot wire. There can never be a dangerous short between neutral and ground, since they are tied together.

Now: If I add a subpanel, using a cable with a separate ground wire I will no longer have neutral and ground tied together next to the breakers. When there is load on the circuits at the subpanel, there will be a voltage drop in the neutral in the cable from the main panel. Hence, there is a potential between the ground and neutral in all circuits on the subpanel. This includes those without load. If I short circuit neutral and ground at e.g. a socket, there will be current flowing due to this. This current never goes trough the breaker. Isn't this a danger? Have I missed something somewhere?

If we assume the voltage drop is 10V in the cable from the main panel, there vill be a 5V difference between ground and neutral. Suppose there is a 2.5 mm2 cable (13 AWG) cable with a neutral-ground short circuit 3 meters (10 feet) from the subpanel. The current flowing in this will then be:
I=V/R=5/(2*3*0.017/2.5)= 123 Amps (!)

A GFI will of course eliminate the problem, but most circuits don't have GFI:s.

[This message has been edited by C-H (edited 09-30-2002).]

If you short the neutral to the grounding conductor, all you have done is create a parallel path for the neutral current. To find the current that will flow in the grounding conductor, you have to do a parallel circuit calculation based on the current flow in the neutral prior to the short and the impedance's of the neutral and grounding conductors.
Don

C-H,
As Don has said, shorting the neutral and ground at the sub-panel creates a parallel path for the existing neutral current. It doesn't increase that current.

The flaw in your logic is that the 5V potential difference between neutral and ground that you have assumed is actually no longer present once the neutral & ground are shorted.

The reason for isolating the ground and neutral in a sub panel is so the ground wire doesn't carry any neutral current. Imagine if the neutral that fed the subpanel were to break, the full load of the sub panel would be going through the ground wire. If someone were to disconnect the ground while it was under load they might not live through it. This happens to plumbers when they replace the water meter, now we jump it to prevent that from happening.

Perhaps I should have taken an example:

A 25 mm2 (3 AWG) 3-wire cable going to a panel, carrying 100A. The voltage drop is 10V. Then I connect a 2.5 mm2 (13 AWG) wire between neutral and ground at the load side. If we assume that this wire is very short, the resistance can be neglected.

If the ground and neutral conductors has the same area, the current will then be shared evenly between them, right? The current going through the 2.5 mm2 wire is now 50A, right?

But, indeed there is no longer a potential difference.

Yes, if the wires are the same length, the same uniform cross-sectional area and made of the same material, then the current will divide equally (assuming also that the connections at both ends also offer the same resistance to each path).

Tsolanto:

Yes, separating ground and neutral is usually assumed to be a good thing. Among the reasons for doing so are the ones you mention.

People now talk of using hot, neutral and ground (TN-S) rather than just hot and neutral (TN-C) to supply houses, since a separate ground wire eliminates the dangers of a loose or broken neutral, just like you have pointed out. This means that the potential difference between ground and neutral at a receptable will be half of the total voltage drop from transformer to receptable.

Pauluk:

From your answer above I take it that you agree that the small conductor beween ground and neutral in the previously described case will carry a current which is much higher than it can handle.

If the load is in fact a panel and the small wire is a short cable to a receptable next to the panel, does this change anything? The resistance in the cable is no longer neglible, but nevertheless very low.

sorry to drift off topic here but...

aside from codes...can we assume that in a real situation with proper wire size, etc. that it would be safe to supply a sub panel, that is in close proximity, lets say 5-10 ft for example, where voltage drop is not an issue, with 3 wires?

Definitely! If the 100A neutral current were divided equally between neutral and ground paths then the 2.5 sq. mm link wire would get warm very quickly.

Not sure I'm quite with you here.....

Do you mean what would happen if the neutral-ground short were at the receptacle end of a short cable run from a sub-panel?

Mike,
If you run a 3-wire connection to a sub-panel, however short, you still have the problem that if the neutral connection goes bad at any point then the panel and all grounding conductors connected to it could rise to full line potential.

Thanks. Then I'm not entirely off track. I did start to doubt my sanity, since everyone including me have considered shorts between ground and neutral to be harmless.


Yes, exactly that.

I see where you're going with this now.

If we have 100A already flowing on the neutral at a sub-panel, then a neutral-ground short occurs at the receptacle on the short cable, then you have a parallel path again.

Some of the existing 100A neutral current will be diverted down the neutral of the cable and back along the grounding conductor, then via the ground connection from sub-panel to main panel to get back to the main neutral.

It's not going to be quite 50A in this case, because the resistance in the parallel path will be a little higher than before due to the length of cable to the receptacle.

It might be interesting to work out just how much of the current would take that path for different lengths, but it could certainly cause problems.

It will be much less then 50% in this case. The main path for the grounded conductor current is the feeder neutral to the panel. The secondary path would be via the feeder equipment grounding conductor, branch circuit grounded conductor and the branch circuit equipment grounding conductor. While the resistance for this path may be small it will be much larger then the main path. It is the relative relationship between the resistance's of the paths that determine the current flow. The secondary path could easily have 10 to 20 times more resistance then the main path. A #12 has 10 times the resistance of a #4 and 20 times the resistance of a 1/0. If the path has 10 times the resistance, then only 1/10 of the total current will flow in that path.
Don

Here goes:

50 meters of 3 x 25 sq. mm. copper cable.
100 Amps.

The resistivity of copper is 0.017 ohm mm/m
Hence, the resistance in each conductor is: 0.017 x 50 / 25 = 0.034 ohm

Just to check that the voltage drop is below the recommended 4% (= 9,2V at 230V)

2 x resistance x current = 2 x 0.034 x 100 = 6.8V ==> OK! Within spec.

At load side of the above cable:

3 meters of 3 x 2.5 sq. mm. copper to receptable.

Path length if neutral-ground short occurs at receptable: 2 x 3 = 6 meters
Resistance of this path: 0.017 x 6 / 2.5 = 0.0408 Ohm

The resulting resistances are:
Neutral conductor directly: 0.034
Ground condutor via receptable: 0.034 + 0.0408 = 0.0748 Ohm

Relative resistance: 0.0748/0.034 = 2.2

This means that the neutral conductor will carry 2.2 times as much current as the ground conductor. If we do the math it works out to a current of 100/(2.2 + 1) = 31A in the ground conductor and the cable to the receptable.

This is on top of the current the cable would be normally be carrying. If the cable is already carrying 16A, the currents in the different conductors would be:

Hot 16A
Ground 39A
Neutral 39A

{Corrections to above values made.}

Are my calculation above correct? Please check!


[This message has been edited by C-H (edited 10-06-2002).]

C-H,
What about the feeder equipment grounding conductor back to the power source? If this is at a panel where there is a main bonding jumper, there is no current flow on the branch circuit grounding conductor with a grounded to grounding short at the receptacle, other then the division of the grounded conductor current for the receptacle load itself. If there is a feeder equipment grounding conductor, then you have to add in that resistance. If we assume that the feeder equipment grounding conductor is the same size as the circuit conducotrs, which is not normal, then the relative realation ship is 5.4. This number will be even higher when the equipment grounding conductor is smaller than the circuit conductors.
Don

Don

I agree with this, but haven't you forgotten to take the length of the conductors into account? If the branch circuit has the same length as the feeder, there is probably no problem.


I take "main bonding jumper" to mean a connection between ground and neutral? If so, I fully agree with the above.


You've got a point in that the grounding wire is normally smaller than the other conductors for large cables. However, in the case of 3-phase systems the neutral conductor is often downsized too. Therefore I'm not sure the downsizing of the grounding wire will have a significant impact.

What puzzles me is where you got the figure 5.4 from?

C-H,
The first path is from the service main bonding jumper to the panel via the feeder grounded conductor to the neutral bus in the sub-panel. The second path is from the neutral bus in the sub-panel to the receptale via the branch circuit grounded conductor, to the branch circuit equipment grounding conductor, to the panel grounding bus and back to the main bonding jumper via the feeder equipment grounding bus. This feeder equipment grouning conductor is the additional 0.034 that added into th fault path.
Don