Thanks again for your reply. I understood your nice explanation and I agree: by applying just a temperature distribution on one surface I am not considering the temperature distribution inside the body. To solve this first problem I performed a steady state thermal simulation, and by applying some boundary conditions I can obtain the temperature distribution on all nodes. Let's assume this distribution is "good enough" for my application. I then imported this temperature distribution in a coupled field simulation and defined it as initial condition. Again I apply some boundary conditions in both the thermal and mechanical field (fixed cube base + convection on the sides and a temperature constrain on the base). The problem that I still have and that I would also have with your last suggested strategy is that the geometry I am importing in my simulations is the one obtained with that temperature distribution. That geometry is already hot and deformed, and I don't know the standard geometry. This means that if I would follow your proposed strategy my part would undergo additional thermal expansion and the result I will obtain is not the one I am looking for.
With what I tried so far I managed to obtain the thermal distribution in the inside of the body as said earlier in the message. The problem is once again that even if I apply it as being the initial condition, it will be applied in the simulation time instant "0+" (let's call it that way, where I mean the first simulation instant). Once again the result is that the geometry is treated as being cold and undeformed, and therefore at the time instant 0+ the geometry undergoes a thermal expansion, and if the simulation is performed for a long enough period, and the part cools down completely, the shape gets back to a perfect cubic shape without residual strains, whereas I expect a contracted shape for t=infinity.
In the project tree, under "Initial Physics Options" I found the "Reference Temperature" under structural settings. By applying a high temperature value (I picked the maximum temperature of my superficial temperature distribution, which I assume to be the highest temperature in my body) I have a thermal contraction as a result. My questions is: how wrong will my result be in the end? Because as far as I understand my thermal distribution inside the body is exactly what I should define under "Reference Temperature".
Thank you very much for your help, and sorry for my many questions, this is my first Ansys experience and I am still exploring all the many available settings.