N-Phase Controller with Thermal Protect

After pondering controller designs sufficiently newfangled to capture my attention for the next few months, I came up with a quick MATLAB simulation of a 40-phase PWM controller. Each phase corresponds to an IGBT capable of switching ~500W at 18KHz and up to 192V. The output of each PWM driver is tied to a thermocouple glued to each IGBT. If the temperature on a given IGBT exceeds a temperture threshold, then the output of the PWM is forced low until the temperature falls below a lower threshold. Thermal hysteresis is, therefore, built in. The MATLAB model currently assumes that each device sources 40Amps at 144Volts near the thermal shutdown region for 30% of the IGBTs. Two plots are shown. The upper plot is Time vs. total controller Amperage, while the lower plot is Time vs. PWM channel. Black indicates OFF and white indicates ON. Notice how some PWMs drivers shut off prematurely, due to thermal overload of their respective IGBTs. The net effect of a controller pushed to it's thermal limit is the somewhat ragged total Amperage plot. When all IGBTs are working within limits the total output current of the controller is nearly smooth, so V = L x (dI/dt) is significantly lower than that of a 1-Phase PWM controller. I need the help of EV veterans to tell me what impact this has on the specifications for the Flyback Diode(s) and Capacitor bank.