Can you confirm that your series of measurements exclude heat inertia in the alumina bricks as origin of the difference?
The switch on/switch off times were identical. The fuelled reactor heats faster than the un-fuelled one, and tends to overshoot the PID-set temperature is all. Any difference in the stored thermal energy is marginal, and does not explain the difference in the curves. As is usual with high-grade lightweight insulating materials, its actual heat storage capacity is quite small.