Neutral Point of Heater Bank

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

May I add a fourth - or a subdivision of (a) if you prefer - where several individual elements are connected in parallel on each phase. (probably not unlikely for a 52kW unit),  If one element fails open circuit the current on that phase drops and the star point moves towards the other two phase (towards, the point where the elements on the other two phases would see 200V, but not quite), The remaining elements on the phase with the faulty element then see an increased voltage and start to overheat (with probably undesirable consequences). At least I think that was Geoff's example.


  - Andy.

As Andy has suggested, I am considering banks of elements in each phase.


This is a common method of construction for this type of heater.  In the example I gave the heater was used to raise the temperature of lubricating for the main engines on a ship (actually a cross channel ferry).  The heaters are inserted into tubes which are contained in a large cylinder which the oil is pumped through.  The actual elements were long spring like coils supported in multiple ceramic insulators.  The mechanical construction was not ideal for a ship as the elements were orientated horizontally and this may have caused resonant vibrations induced by vibrations in the vessel.


The method of control was a simple on / off contactor - this is significant because the control system can do nothing to mitigate any voltage rise caused by a shifting star point.  A power electronics control system could be design to compensate for voltage rise so the suggestion by Broadgage that the manufacturer's instruction might override the need for a neutral is perfectly true and if such instructions exist they should be followed.


The generating system on even a small cross channel ferry is perfectly capable of maintaining proper output if a heater of this size and type fails in any mode.


I demonstrated the mode of failure to the manufacturer's rep using lamps as it gives a very graphic illustration and his eyes appeared to glaze over a bit when I started drawing phasor diagrams to explain what had happened.


The problem we had was fairly long standing and a large number of failures had occurred.  Always with one phase completely at fault and some times with two but the third remaining unaffected.


Now to the fun part ?.


Some years ago developed (mainly for my own amusement) some routines to handle complex numbers using the Pascal programming language.  At that time the Free Pascal Compiler (fpc) did not have a complex number data type so I had to create my own using a array data structure.  The fpc now does have a full suite of complex number data types and functions base on record structures.  One of the things I did with my system was to apply it to Millman's Theorem to solve three phase star connected load problems.  Be of a certain age I used Red, Yellow, Blue for references to the line conductors, n for neutral conductors and the supply network start point, and s for the star point of a star connected load.


This software can be used for this problem and here are some results.


Assume we have three banks of heaters each consisting of five elements per bank at, say, 10 kW per bank and 2kW per element.


Conductance per phase for an intact bank  = 0.19 seimens


input power 10 kW per phase


Voltages are described with reference to the red phase and this is at 0°.  With a neutral connected phasors for phase voltages are Vrn, Vyn and Vbn.  With the neutral disconnected these phase voltages are now wrt the load star point which I call s.  So we have Vrs, Vys and Vbs.  These are the voltages that appear across the element banks (or your house if the neutral breaks in the supply network)


Neutral connected 

Vrn, Vyn and Vbn all equal to appoximately 230 volts each displaced at Vrn @ 0°, Vyn @ 240° and Vbn @ 120° - rotating anti clockwise, angles referenced anti clockwise


Phase Currents are approximately 43A in phase with each voltage.  In (neutral  current) = 0.


If I now open the neutral the phase voltages and currents do not change because the load is balance  (In) can no longer exist as there is no connection between the load star point and the supply neutral.  There could, however, be a voltage.  This would appear if the load becomes unbalanced.  I call this Vsn ie: the voltage between the load star point and the supply neutral.


If I now reduce the power on the red phase by 2kW to simulate the loss of one element in the five element bank we have.

Red phase - 8 kW load - Vrs = 247.5 @ 0°, Vys = 223.21 @ 243.67° and Vbs =  231.21 @ 116.33°


Note the rise in Vrs and the shifting angles in Vys and Vbs


If we continue to reduce the load in the red phase the voltage progressively increases

6 kW load - Vrs = 266.54 @ 0°, Vys = 215.44 @ 248.21° and Vbs =  215.21 @ 111.79°,  Vsn = 35.54 @ 180°

*

2 kW load - Vrs = 315 @ 0°, Vys = 202.51 @ 261.05° and Vbs =  202.51 @ 98.95°, Vsn = 84 @ 180°

0 kW load - Vrs = 346.5 @ 0°, Vys = 200 @ 270° and Vbs =  200 @ 90°. Vsn = 115.5 @ 180°,


Note that as the load decreases Vrn increases and Vsn increases @ 180° to Vrn 


If you continue by reducing the load of one of the two remaining phases Vsn tracks along either Vys or Vbs depending on which load you are reducing.


My software produces quite reasonable phasor diagrams but I have not included any facility to print them!.


In summary,  I agree with the comments that you should follow any manufacturer's instructions and I am not claiming that this is some earth shattering safety issue.  However, simply connecting a neutral can provide a measure of protection that might not otherwise be in place.


Regards


Geoff Blackwell

Thanks Geoff ... for those interested to follow up with the calculations, Excel can be used to grind complex numbers these days also.