ECN Forum Forums Code Related Discussion NEC & other Code issues Bonding of water piping.

Bonding across a dielectric union _must_ to some extent defeat the purpose of the union. You have two _different_ metals in contact with a conductive fluid; net result is an electrochemical cell. Short the two terminals of this cell together, and you will have current flow.

That said, think about the distances involved. Without the dielectric union, you have the two metals in direct contact with a _very_ short path through the electrolyte. Add the union, and now you greatly extend the distance that ions have to travel through the water, increasing the resistance of the electrochemical cell substantially.

Lots of times the hot water and cold water pipes are bonded at the fixtures anyway, although the intentional bond is an especially good idea as more and more fixtures are connected using flexible hose.

I wonder if it would be possible to design an approvable water pipe bond that was something like a self triggering SCR, something that would be an insulator up to about 1V, and then break over and start conducting? This would appear to be a short at electrical system voltages, but an insulator at the potential of the electrochemical processes that the dielectric union is supposed to stop.

-Jon

Hi,
I thought a union is installed for future replacement purposes only, and played no other role.

The HW tank insulates the two connections according to my MASTER PLUMBER firend.

-regards

Greg

Winnie, a silicon diode has a voltage drop of .6 or .7 volts. They don't conduct below that voltage.

Mustangelectric, bonding the cold water within 5 feet of entrance is for use as a grounding electrode. Bonding at the water heater is usually for bonding piping systems in the building, in which case the 5' doesn't apply.

(edited 01-15-2005).]

[This message has been edited by Physis (edited 01-15-2005).]

Admittedly I had not considered a pair of ordinary silicon diodes, 'back to back' for bidirectional current transfer, but I don't think that this would be suitable.

Ordinary silicon diodes have _some_ conductivity right down to 0; the current versus voltage curve is an exponential, so it is reasonable to approximate a diode as a device with no conductivity up to a threshold voltage, and a fixed forward voltage drop...but the reality is that at low currents you have a very low Vf, and you have some reverse leakage current.

Then, when the time comes to conduct a fault current, you will have that 0.5-0.7V...at a whole bunch of amps.

Ideally the device would have zero conductivity up to a threshold of perhaps 1V, and then would 'trigger' and have zero voltage drop for any current above a few 10's of mA, and then once the current falls below the limit, would go into the high resistance state again.

I don't know if a semiconductor could be tailored to meet these goals. Also, I don't know if a simple silicon diode would be good enough for this task without a) introducing safety problems or b) costing too darn much.

-Jon

Physis,

What do you mean? Do you want more information? A friend of mine is a plumber and his customer was having a lot of "pin-hole" leaks in his house. The plumber would fix a leak or two almost every year. He finally asked me to look into the situation. I checked all through the house and all that I found was a plastic water filter in the back of the house. I asked the plumber if the leaks were in the front or the back of the house. He told me that all of the leaks were in the back of the house. I put a jumper across the plastic water filter and there hasn't been a leak since.

Harold,

Yeah, that's all. More information. I don't think it's impossible, just very unlikely. Like, for instance, were any other theories considered? Was anything unusual measured? Is it possible it was a chemical reaction?

Winnie,

What does that? FET's or MOSFET's. It's interesting to think about, but makes a pretty complicated jumper. You could power it with the fault.

Physis,

When I was looking for problems, I tried to get any kind of a reading across the plastic water filter. I put a VOM acrossed the copper pipes, (Both digital and analog) I tried to put am ampmeter around the copper water pipe, but I didn't see any kind of readings at all. My thought was that the current (What ever the type) was intermittant and it just wasn't present when I was looking for it.

I would expect that the current was present when something electrical, connected to the plumbing, was "on". A defective water heater element could put quite a bit of power into the piping without tripping the breaker, the same could be true of a disposal, washing machine or dishwasher. The current would only be there when the faulty equipment was on.
This is the kind of thing that causes those nuisense trips of GFCIs.

Harold,

Your situation sounds to me like what happens in boats in salt water.

Basicaly, any metal you put in salt water turns into 2/3 of a battery cell. The only other thing it needs is another piece of metal that's a different kind.

In boats you put a piece of zinc on the bottom of it to "if I recall correctly" to be more negative than the other metals and therefore do all the oxidizing.

If you put two leads from a battery into water, the water molecules get seperated into oxygen and hydrogen. I think it's the negetive lead that gets the oxygen.

Anyway, I'm thinking your situation is electrochemical. And I don't think you were looking for DC.

Shunting the "electro" half would likely stop it.

Maybe I was miss reading you Harold, because I think we're in agreement.

Ok, the problem with dissimilar metals is when the water is a conductor or a dielectric. So I disagree with a plumber who says the jumper would cause his bimetal unions to not work as intended.

You'll need to set your meters to DC to see this stuff.

An electrochemical cell produces DC, so if you are getting corrosion caused by dissimilar metals in water, I would expect DC current flow.

A 'dielectric' is an insulator, not a conductor. Two dissimilar metals in an electrolyte is a cell; if current can flow than one metal will dissolve into the electrolyte. Two dissimilar metals isolated by an insulator is an open circuited cell, and the only current flow will be the 'leakage' through the electrolyte. Adding the bonding jumper will clearly 'close the circuit' around the cell, and let current flow.

To expand what I said in a previous post, the characteristics of an electrochemical cell depend upon things like the metals involved, the conductivity of the electrolyte, the length of the path through the electrolyte, etc. Even in an electrochemical cell, current still flows in _complete_ closed loops. However the current flow through the electrolyte is in the form of ions physically moving, _not_ electrons alone. If you increase the space between the anode and cathode in the cell, you will greatly increase the internal resistance of the cell.

If you have different metals in direct contact in an electrolyte, you have a _very_ short path through the electrolyte, and thus a very low 'internal resistance' for the galvanic corrosion current flow.

If you have two different metals in an electrolyte, separated by an insulator, but electrically bonded together, then you have a much higher internal resistance in the electrochemical cell, and thus much lower current flow.

But I would still expect _some_ current flow.

Harold,

After you bonded around the water filter, did you measure any current flow on the bond? Here is a possible theory: there was a minute DC current causing the pinhole corrosion. But after you fixed the bonding, you got a bit of AC current flow. AC current electroplates for half the cycle, and probably smooths out any pinhole corrosion into much slower overall corrosion of the entire pipe.

-Jon