State average incident power, peak envelope power, pulse width, repetition rate, duty cycle and modulation crest factor separately. For a rectangular pulse train, average incident power equals peak power multiplied by duty cycle, but that result only describes average heating. It does not prove peak voltage, current density, arcing or short-pulse survival.
Apply the published derating curve at its named temperature point: ambient, case, flange, baseplate or coolant. Include the thermal interface, mounting torque, heat sink, airflow or coolant flow and stabilization time. A 25 W rating at 25 °C is not automatically 25 W at a hot baseplate.
Acceptance should monitor incident and reflected power, waveform and temperature at declared locations, then repeat critical return-loss measurements after stress. Keep product-specific peak and pulse limits separate from the illustrative average-power calculation.