I had a look through "Concepts for Gage R&R Studies" by Larry Barrentine. The following is suggested for electronic temperature equipment. You'll have to assess it's usefulness based on your system
1) If there is no opportunity for the operator to impact the measurements then there is no reproducibility element - so the gage r&r study becomes a gage r study (repeatability only)
2) If it is not possible to run meaningful repetitions of measurements with the equipment during operation, then the repeatability has to be assessed using the calibration system.
3) Incorporate the following into the calibration: Use a calibrated energy source to simulate the hot product. Take at least 20 measurements using the unit. If the unit is used at a number of different settings this must be repeated at each setting.
4) Plot a moving range chart to look for statistical control, and calculate standard deviation using standard control chart factors
5) Calculate the Gage R&R % as (5.15 x standard deviation) / tolerance
6) Before doing any of this, maybe you could get in touch with the temp unit manufacturer and get his advice about any procedures he would use. Also, I guess if you are using an outside firm for calibration, you could speak to them and ask for their advice.
So, basically his suggestion is to incorporate a repeatability element into your calibration procedures. So, if you think it's necessary, take a look at your current calibration procedures and try and work out how to do it.
Another thought - I'm guessing that there's some problem with measuring the impact of this heating process on the product. Otherwise, you would be determining whether the heating process was working adequately or not, based on the results of this "final product" measurement, and you would run an r&r study on that final product measurement system.