Here is my "electrishun's" take on the matter of harmonics. I hope this will be the spark for some discussion in which I may learn a thing or two.
First, assume we have a 208Y/120 3phase 4wire system. If A, B, and C have single phase loads connected to neutral, and have the same current flowing in each load, we have the following graph for current (Black = A, Red = B, Blue = C). In this instance, the neutral current is zero because any current flowing in equipment connected to a given phase is "returning" through the equipment connected to the other two phases. Everything is balanced, and essentially we have two loads in series between any given two phases. All of the loads are conducting current continuously.
Now, let us look only at A:
But if we have equipment connected from A to neutral that contains "switching mode" DC power supplies, (which in today's world would include most pieces of electronic equipment), then the situation is different. Because the DC power supply in the equipment only conducts near the peak of the sine wave, the original sine wave current is no longer a sinewave. The actual current waveform looks like this:
I'll go ahead and post the graphs of B and C also, first the normal sine wave of linear load, and then the graph of current from equipment from phase to neutral with DC supplies, otherwise known as non-linear load.
B phase:
B phase current with DC power supply as connected load:
C phase with non-harmonic connected load, normal sine wave: Oops, the forum only let's me put in 8 photos. I will omit the C phase normal sine wave.
C phase with "harmonic" DC power supply as connected load:
And here is the conglomerate graph of all three phases with switching mode DC power supplies:
Now a key point is to recognize that the current in A phase cannot balance on B phas or C phase because those loads are not conducting at the time when A phase loads are conducting. All of the A phase current must return to the source via the neutral. This same thing holds true for B phase and C phase. The phase currents cannot balance.
And finally a graph of the neutral current:
Notice:
1. The neutral current frequency is 3 times that of a given phase. This is called the third harmonic, or "triplen".
2. The current flow in the neutral is at or near - or + peak continuously. The RMS value of current would be larger than the phase currents RMS values. This could definitely cause heating of the neutral. Note that the neutral current is hardly ever near zero like that of each phase conductor.
3. The neutral current looks more like a square wave than a sine wave to me.
Now what about the higher order harmonics? Seems to me they originate with the voltage drop that will be experienced near the peak. Take A phase and a DC supply. When current flows only at the peak, the only voltage drop experienced will be at the peak. This causes a dip in the voltage at the peak and in my mind is the source of the higher level harmonics.