I think I disagree here and would like you to explain further. Each 120 degree separated line is positive for 50% of the period (period 20ms in your case, positive (or negative) for 10ms.
True. For continuous DC current, the conducting path from say, L1 to L2 will change when L2 voltage exceeds L1 voltage. That's where the 120deg comes from.
Current will flow when the line voltage (minus forward drop) is greater than load voltage, almost certainly less than the 50% per diode element in operation.I completely agree with the different shape ... if the capacitor bank is connected to a diode element powered by a 100VAC RMS supply, ignoring forward drop I would be supplying current when the waveform is above the capacitor bank voltage at that instant. Current will flow until the peak of the waveform is reached, approximately sqrt(2) * 100V or approximately 140V. The primary ripple frequency will be (assuming a bridge) 100 Hz with single phase, 300 Hz if 3 phase ... but percent of conduction for each diode will not vary ... WILL IT? My thick skull only sees that as significant while the line-load difference is in the approximately 0-2V range ... being very generous on real device characteristics.
Good points.
I'm about to have dinner - stuffed Shitake mushrooms with ground beef, Bulgar wheat herbs....etc
Mrs B is an amazing cook.
So not much time to answer all of your points.
Just one. At 300Hz the diode is likely to be forward biased for more of the conduction period. It starts at 0.866 of peak voltage.
With single phase it starts at zero.