Effect of ambient temperature changes on air-flow calorimetry

Figure 1 below shows how difficult it is to use air flow calorimetry when the ambient temperature is not stable. This is a 30 W calibration. Average input power (left axis) is 29.7 W. From 3:50 to 4:20, the temperature (right axis) rose from 25.9°C to 27.0°C. The apparent output power fell from 35.0 W down to 22.5 W.

From 10:00 to 3:49 the next day, the ambient temperature was fairly stable. Average output was 29.0 W. After adjusting the output power for estimated losses from the calorimeter chamber walls, the average was 30.5 W, an overestimate of 0.8 W.

I have estimated this instrument can measure to the nearest ~2 W. That is an unsophisticated approximation made by me. Based on:

1. When ambient changes are reasonably small, such as in Fig. 7, fluctuations are around 2 W (apparent 2 W, but not real).
2. The calibration constant is 0.05522 K/W. The temperature difference is measured with one Pt100 RTD at the inlet, and two Pt100 RTDs at the outlet. The average value of the two outlet RTDs is used. I think we can rely on these for precision to the nearest 0.1°C, so that's 0.1°C / 0.05522 K/W = ~1.8 W. I wouldn't trust a temperature difference smaller than that.
   

You can see why I had some doubts about the 12 W of excess heat reported at ICCF21, shown in the lower graph.

Figure 1. Effects of ambient temperature changes on a 30 W calibration. (Oops! I had the wrong figure.)

Figure 2. 12 W excess, ICC21 result.

Here is a look at the effects of ambient temperature fluctuations on a 200 W calibration. This is not good, but I am confident the instrument can detect 40 to 250 W of excess heat with confidence, despite these problems.