| As to why I am so concerned about this, its not so much the drop in | volts (as long as we remain within the limits that appliances can work | within), but rather the fact that there is an increase in the adjacent | phase voltage. Measurements (also taken now) indicate that there is an | increase in the blue phase from 236v to 239v when the geyser is once | again energised, the geyser being on white phase. (Phase colouring in | South Africa is the same as Europe = Red, White (or yellow), Blue). | Is this quite normal? I ask this because we had a situation a couple | of years ago where the feeder neutral was faulty, and we ended up with | roughly 280v on red and blue phase, with roughly 150v on white. | I found out afterwards that we were only connecting to 4 other houses | via the neutral at that time. | Needless to say, all that was destroyed by the overvoltage was an old | sprinkler controller, but it could have been far worse.
In the US the problems of multiwire circuits are more common and well known because we have them even on single phase (2 hots 180 degrees to each side of a grounded neutral). The same problems also exist with three phase, but the calculations are a little more involved.
When you have an excessive load on the red phase, it "pulls" the neutral point towards the voltage point of the red phase (which pulling the red phase toward neutral). The degree of this pull is most extreme nearest the heavy load and decreases as you follow the power feed wires back to the source. The impedance of the wire and transformer(s) are part of why this happens. When current flows across an impedance, there is a voltage drop. This is simplest to figure when it is a pure resistance. The standard ohms law equation applies. Suppose for example the total resistance of the wiring is 0.1 ohms and you are drawing 13 amps through that wiring. The voltage drop will be 1.3 volts regardless of what the supply voltage is. Half of that will be on the hot phase and half will be on the neutral (assuming each has 0.05 ohms resistances). There will also be 16.9 watts of power heating up those wires, distributed across the length of them. That voltage on the neutral has a phase angle to it which is a vector in the direction of the phase having the extra load. That vector drawn on a polar plot shows a decreased distance to the end of the vector for the red phase, illustrating the voltage decrease seen in the red phase. That change in neutral vector position also shows an increase in voltage relative to the other 2 phases.
So, given the arbitrary parameters I picked, you'd have a vector change of 0.65 volts in the red phase back to the neutral, and a vector change of 0.65 volts in the neutral out to the red phase. That would raise the voltage between blue and neutral by 0.325687886 volts. Same for the white/yellow phase.
This gets more complicated when the loads are reactive and have current phase angles different than the voltage phase angles, such as you can see with motors. It gets even more complicated with non-linear loads that introduce harmonics (due to the 120 degree phase angle, not all the harmonics will cancel out even if all the phases are in balance). The vector diagrams in these cases would involve constantly changing vectors that spin and pulse.
Of course all the mathematical precision here is pointless with lots of voltage changes taking place with varying loads happening on the various phases. But the real point here is that a voltage RISE is very much to be expected on the opposing phase(s) as a result of the load on other phases. Keeping the phases in balance keeps the voltage drop vectors on the neutral close to zero (they add up as vectors). When neutral current is zero, voltage drop across the run of the neutral is zero.
There are things you can do about this. Running the water heater at
400 volts (if you can get a 400 volt element and the controls are rated for 400 volts) would help by spreading the load across 2 phases instead of 1, and avoiding the neutral entirely. In the US, we run our heavy appliances like a water heater mostly connected between the 2 hot phases (for a total of 240 volts). Although we have more issues with that due to our lower overall voltage and higher current (leading to more voltage drop that hits that lower voltage harder) the same idea could help in your situation.
The higher the voltage you can bring the power in with, the better the situation. You'd see noticeably less if you brought power in at 400/690 volts and transformed it down to 230/400 right there. But you'd also see noticeably less money in your bank account.