Help me troubleshoot a three-phase heater issue

The troubleshooting is after the fact since I already have the problem fixed but I’d still like to figure how it happened.

Basically, I have a machine with a bank of heating elements fed by 3-Φ 575V. Each group of three heaters is connected in wye. Phase A to heater A, phase B to heater B, etc. The opposite ends of the heaters are shorted together. The machine has been running for many years so it’s not a new design or anything and each heating element measures 135Ω when cold.

What happened is the element in heater A shorted out. For some reason heater B then got hot enough to melt the aluminum block it’s mounted in. I can’t for the life of me figure out how that could happen.

This might be beside the point but there is a thermocouple monitoring the temperature which is mounted in the block for A (the one that shorted). Heater A wouldn’t have been producing heat when it shorted since there was no current through the element (the short occurred at the connection where the leads enter the heater). Heaters A and B are physically side by side so I would have thought the heat from B would still be enough to heat up the thermocouple enough to shut things down but the temperature readout never went above 190˚C. The only other possibly pertinent factor I can think of is that heater B is connected to a current monitoring relay. I don’t think that could cause the problem I saw but replaced it anyway.

No matter how I think about this I think a failed element should have resulted in less current, not more (kind of like a Scott-T transformer). I can’t think of a scenario which results in heater B getting hot enough to melt aluminum, especially since heater C was fine. FWIW the aluminum block is about the size and weight of a typical hardcover novel.

I’m not so concerned with the thermocouple reading as with how heater B could get that hot and also with why C didn’t get that hot even though it should be seeing the same input. Heaters A and B were replaced, heater C and the thermocouple were left in and are working normally right now. I meggered and ohmed out the element in heater B after removing it and it looks fine aside from the block it was in having melted.

Any thoughts, WAGs?

There may be a confusion of terminology here; hopefully you can clarify. A “short” is when you get an inadvertent connection that bypasses the main part of your circuit - in essence, the current takes a short-cut between supply and return. If I understand you correctly, you have a burned out element A, which results in an open-circuit, not a short-circuit.

Is this correct?

The heating elements may have temperature-sensitive resistance. It might be helpful to measure voltage and current across each element once everything’s up to temp; possibly element B his higher resistance (at temp) than element C, and so that is causing more power to be dumped there than in element C (when element A is open-circuit).

The heater element B didn’t heat up and melt the block. A short going through the aluminum block melted the block. That’s the best WGA I have based on logic and I would look for the two points current would have had to flow from and too to melt the block. You may have had an arc welder situation even.

Heater A was shorted, ~zero ohms between the two leads.

You’d get zero ohms on a open circuit too. Actually, it’d be infinite ohms, but most meters would just sit there and read nothing if the circuit is open. If you really had a short your meter would beep or indicate zero ohms. Is that the case here?
You sure it’s a short? I can’t picture how an element would go from X ohms to zero ohms. The breaker or fuse should have opened.

I’m trying to restate your position in a simpler form.

You are stating that you had 1 phase faulted to a floating neutral and 2 other phases with equal loads from phase to floating neutral?

Then your assumption would be that the two loaded phases should have equal currents and therefore have equal damage?

But you have one loaded phase showing overload damage and one that didn’t.


Bank A failed causing the floating neutral to no longer sum to 0 volts and it somehow shorted to ground.

The ground path went through the B bank mounting block.

Second SWAG -

Bank B had some kind of thermal induced change in resistance. If so, it may appear normal in cool tests but would carry more current than Back C when voltage increased.

Did the current monitoring relay trip? You said you replaced it but you didn’t say if it showed any signs of over-current wear.

Sorry, I could only make a drive-by response earlier.

Yes, that’s exactly the case. Thanks for making the terminology clearer.

One wire connected to the other just inside the mica or whatever it is they fill those heaters with. I don’t know why the insulation would break down, usually heaters fail open not closed. The fuse eventually opened but it’s not a dead short to ground, the return path still goes through the other two resistances. it’s easier to see with a schematic but I didn’t have time to find one online that I could link to.

That relay is to detect an open circuit. I’m going to suggest to my boss that we install current relays and/or thermocouples on each heater. Unfortunately that adds up to hundreds of extra devices throughout the building. I hope they’ll do it anyway since the risk is quite huge if we can develop that kind of temperature in a building full of flammable material.

That, and a combination of BubbaDog’s WAG could be the case. I wish I had seen signs of arcing between the block and something around it but I didn’t. The melted aluminum burned a nylon conveyor in the area but I don’t see arcing damage. It could have arced within the block itself though, even though it meggers OK now doesn’t mean it was always OK since a short circuit could extinguish itself by melting away whatever connects it (another WAG I know). It’s a better hypothesis than anything I’ve come up with so far.

You didn’t mention what kind of contactors you have feeding the heaters, solid state or mercury relays, or some other. If they are solid state, it is a typical failure mode for the controlling triac to fail as a short, supplying power continuously even it off. The same can happen with standard dry contacts if they weld together.

Also, if for whatever reason, the thermocouple was not getting a proper reading (is this on an extruder?.. Depending on the design some extruders have problems with the barrel thermocouples), it could have set the controller to just keep firing full on. Even a short in the thermocouple wire will cause it to read ambient only, and never see the heater heating up.

In any event, yes, even one single element of a three phase heater in an aluminum casting can (and will) melt the aluminum unless there is some external method of removing excess heat. We’ve done it here quite a few times. Very messy too.

Thanks for the response (to everyone else too).They are indeed solid state relays. In this case they are shutting off properly.

The thermocouple was not reading correctly when it happened because the block it was mounted in was short-circuited and the current was bypassing the heating element. So I do think the heater was staying on continuously as the controller tried to bring the temperature up to set-point. I didn’t think one heater would be enough to get that kind of heat in the block. That corrects one error in my reasoning at least. That also points out another major flaw in our set-up. There is nothing to detect an overtemperature in the blocks farthest away from the center.

Even though I still don’t get why that one block melted and the other didn’t, there is a lot of good thinking coming out of this. I am going to push hard for a thermocouple on each block following this.

And yes, very messy when that block melts. I hope not to see that again. :slight_smile:

575 Volts? I have never seen 575 is this a production voltage?

If your heaters are large enough to reach the upper limit easily this could be the problem. With the loss of heater A, and the upper limit sensor neat A, B and C can stay on longer before reaching the upper limit. And the area around B and C will exceed the upper limit, overheating B and C. this does not mean that B or C are over amping.

I have seen this on fan powered boxes with reheat.

575 is a common voltage here. It’s getting harder to find equipment rated for it. I don’t know if it’s true but an Allen-Bradley rep told me they only make 575V rated drives and motors for just a few markets such as here, Ontario, and at least one state, Wisconsin maybe? Newer equipment is all 480 and we step it down at the machine. The entrance is 575 though.


When I worked at AB-Rockwell, the various voltages that we would need to provide for the test cells where 690, 575, 480, 400, 240 and 208.
I do electrical work in Wisconsin and have not seen 575 in use as of yet. Your rep wes probably from Wisconsin which may have led to the confusion.:wink:

:stuck_out_tongue: That was probably it, thanks for the info.