I'll try out eliminating the rectifier, thanks for filling me in on the SCR waveform!

Belay that. The title of the device might say "SCR", but the wiring diagram looks more like a TRIAC device; in which case you still need the bridge rectifier.

And the circuit breaker is indeed thermal-magnetic.

This explains the 20 second delay between power-on and tripping. The current peaks are not enough to cause the breaker to trip due solely to the magnetic function because while they are greater than 20 A, the duration of each current spike is shorter than the magnetic time constant. However, the energy in the spikes integrates in the thermal part of the breaker, and exceeds that trip point in 20 s.

when I reduce the capacitor count/values, they simply overheat trying to keep up with the TEC load,

Nope. The load current is not what is heating the capacitors. They heat up because there are fewer of them sharing the very high ripple current, a current that does not go through the load. It is in parallel with the load current. As you decrease the number of capacitors, the ripple current (and heating) per remaining capacitor increases.

ak

 

Alright, well thank you everyone! Very enlightening discussion (for me at least). I had qualms about my design handling that huge amount of power with such a simple design and now I know why. On to a SMPS I guess.