Neutral Point of Heater Bank

I'm not sure what's standard practice, but I can see a couple of advantages if having N connected even if it doesn't end up doing much most of the time. Obviously if the elements could be switched individually you'll need N connected - but there even if they're normally all switched together having a N would ensure correct operation of the other elements if one (or two) failed for any reason (perhaps due to a local thermal cut-out tripping, the element failing to open circuit, or loss of one phase of the supply).


And if all else fails, see what the manufacturer's instructions say...


  - Andy.

Also if they're electronically (e.g. thyristor) controlled, you might find that a N connection is necessary even for normal operation - as the altered waveforms might not quite sum to zero in the way that a nice sine wave does.

  - Andy.

If the heater is made up of groups of parallel banks of elements in each phase then it is essential that you connect the star point of the element banks to the supply neutral.  If you do not then if a single element in the parallel bank fails the load will become unbalanced and the 'star point' of the load will shift with respect to the star point of the supply.  This will cause a voltage rise on some of the elements and this in turn will lead to more failures.


This is the same situation as a lost neutral on a three phase distribution system that feeds single phase loads (unbalanced). 


You can calculate the voltage distribution using Millman's theorem.


I had to deal with a situation like this when I was an electrical officer in the merchant navy.  We took delivery of a couple of new lubricating oil heaters.  These were three phase banked elements connected in star - this is not a suitable design for most ships as the engine room services are generally connected to an IT system that does not have a distributed neutral.  A delta connection does not cause a problem.


The elements in these heaters started to fail, first  in one phase and then in a second.  The end result was complete failure in two phases and none in the third..  The manufacturer's rep was called in and after I demonstrated the fault using banks of 15 watt lamps and presented him with the calculations two new delta connected heaters were supplied on free issue - a very costly mistake on their part.


Regards


Geoff Blackwell

I would say that the star point of the heater bank should be connected to the supply neutral, UNLESS the instructions specifically state otherwise.

The neutral should also in my view be full sized, since a control fault or the loss of two phases of the supply cold result in the neutral being fully loaded.

So we had better start connecting the star point of motors to the supply N too then in case one of the windings develops turn to turn shorts, or perhaps otherwise begins to draw excess current and that would do the same thing.

In that situation the motor has failed and will probably require a rewind - further failures in the winding do not change that. 


If heaters are connected in banks and one element in the bank fails then a replacement element would return the heater to full capacity.  The omission of a  neutral on this type of load can cause the failure of the whole bank as the voltage rise will stress the remaining elements.  If a neutral is connected the voltage will remain stable and further failures due to voltage stress would not occur.


This is not a safety issue and is not AFAIK the subject of any regulation - however, the heaters I mentioned were replaced because the manufacturer accepted that the method of installation contributed to premature failure.


Regards


Geoff Blackwell

Not always failed completely though Geoff, I've seen motors running quite happily with 20-30% imbalance in the current draw across the phases.


It is certainly not required in any of the machinery electrical standards and this would be out of the scope of BS7671 anyway.


To me it's not a neutral point, a neutral point originates at the supply.

This is a common point of connection in a load.

If the supply is being skewed that much by a single heater failure then I would think you have got bigger problems.

I've just seen that your example was ship board, sorry I don't think that's comparable.

In that scenario I would agree totally that a different approach is required, but fed by the national grid I don't agree.

Well why don't you try the demonstration I did using banks of lamps connected in star on a three phase supply. Use 5+ in each bank and then simulate the fault by removing one lamp at a time from the same bank - note the change in brightness.  However, have a care, you may cause real failures in the remaining lamps in that bank. ?


Regards


Geoff Blackwell

Sorry Geoff, I don't think that a lamp and a resistive heater element are comparable either.

A lamp has a very different current/temperature/resistance characteristic to a big resistor, which is basically all most of these heating elements are.


Apart from that, it's a bit of a mute point anyway because likely the Eurotherm unit will have in its instructions what is required, so you'll have to follow that otherwise it will throw a freaky fit.

They are often SSR based.

Assuming the heating elements are resistive loads, there are three basic single-fault failure modes:


(a) Open-circuit element. The voltage drops on the other two elements to 200 V each.


(b) short-circuit to Earth. THis one is more tricky, as it may well not be a "fault of negligible impedance" as assumed by BS 7671 ... and overcurrent protective device may not operate. It depends how the break happens, but either the protective conductor connection restores Neutral through the earth... and the overcurrent protection might not operate. The shortened element could then eventually blows open-circuit, and rapiring the lement will then remove the "star-point to Earth" fault.


(c) shortened element. (resistive element resistance decreases). This will put an overvoltage on the other two elements, and they are potentially at risk of over-voltage and damage.



So I guess, in most of the failure modes, resistive elements wouldn't require a Neutral, but just a caution that element faults don't always follow the assumptions made in BS 7671.