The only way by which a thin hollow cylinder can experience thermal stress is- if the thermal expansion (i am assuming that expansion happened due to a heated fluid flowing inside the cylinder) of the cylinder is constrained by some external means (press-fitted inside the casing let us say). For this case, you can apply the standard press-fit analysis for calculating thermal stress. In all the other condition, the hoop stress experienced by the cylinder will be equal to (P*R/t), where P is the pressure of the fluid, R is the radius and t is the thickness of the hollow cylinder.
Now let us take the case of a thick cylinder (this is more tricky now).
For this case, the temperature gradient across the thickness of the cylinder will lead to differential expansion, and the cylinder can generate thermal stress (depends on the slope of the temperature gradient) without even any external constrains. This problem can be solved by assuming the thick cylinder as a series of thin cylinders stacked coaxially to each other, such that, the inner thin cylinder is press-fitted by the subsequent external thin cylinder. The same standard press-fit analysis can be applied between the each thin cylinder interface to calculate the stress. Now the question comes, how to select the thickness for a thin cylinder for constructing series of thin cylinder. For this, a constrain (R/t < 10) can be applied initially to get the number of divisions (let us say "N"). You can stop here itself, but if you want to get a more accurate solution. Once formulation is constructed using N, to increase accuracy, a limit on N tends to infinity should be applied for getting an improved solution.