I have tested room temperature austenitic stainless steels, and found temperature typically accumulates higher and does not cycle or drop.  your data is interesting, but the temperature cycles are extremely repeatable with the load.  the stainless steels diffuse heat slowly, and it looks like you are loading at a rate perhaps faster than thermal equilibrium would be achieved, so i would question exactly what you are measuring with the thermocouple.  are you sure the thermocouple was not bonded too tightly to the metal surface?  if so, you could be straining the thermocouple itself, and the change in length corresponds to a change in temperature-even if there is no actual change in temperature.  your thermal data looks so consistent with the load data, i would confirm that your thermocouple was not inadvertently measuring the mechanical strain.  cheers

Thanks Brian

The t/c is a sheathed and mineral insulated type N, which is wire locked to the specimen. If it was locked in two places then it is possible it might get streatched and the strain transferred from the sheath to the wire via the insulation. However I can't remember if it was wire locked in one or two places, I will check on Monday with the lab technician. 

I must admit my first thoughts were that because the thermal contact between sheath and specimen is limited that the t/c would not be able to respond to small changes of temperature on the specimen and would respond much more to radiant heat from the furnace wall.  I was not expecting the temperature and load to have the same period!  My first thought was the specimen moving slightly +-75 microns every cycle was changing its position in the furnace and experiencing a different temperature gradient.  

I observed this in two tests, and will look at others to see how reproducible it is.

Hi, your results make sense. In your case, the temperature evolution is essentially governed by three contributions:

- Mechanical dissipation, which is associated with the development of irreversible plastic strains. This contribution will always be positive, for both tension and compression. This contribution will thus be responsible for a temperature increase.

- Thermoelasticity, which is associated with the development of reversible thermoelastic strains. For an uniaxial loading, this contribution induces a temperature increase when the stress decreases and a temperature decrease when the stress increases.

- Heat exchanges due to the fact that you don't have perfectly adiabatic conditions. This is why the temperature does not continuously increase during the cyclic test.

If you consider these three contributions in a simplified heat diffusion equation, you should be able to explain your experimental observations. If you need a precise understanding of these aspects, you should have a look at the papers from A. Chrysochoos.

Thank you Charles

I will follow this up and look at more of my data.  

I would check your furnace for a gradient, especially since your frequency is so low.