Hey. A uniform ΔT is a first check. Steel coefficient is ~12e-6/K /K, so a ΔT of ~60 K gives free strain ≈7.2e-4; if fully restrained, that implies ≈151 MPa (E≈210 GPa), near yield for many steels. Do detailed steady-state or transient FEA when: 1- internal frames or attachments significantly restrain in-plane expansion; 2- you expect through-thickness or longitudinal gradients (e.g., stratified fluid, top surface cooling); 3 – welds and local attachments could concentrate loads; or 4- uniform-ΔT results are close to yield, buckling, or fatigue limits.

I suggest modeling a uniform ΔT global case first. If stresses approach limits, produce a realistic temperature map (conjugate heat transfer or measured data), include realistic restraints for supports and welds, check membrane vs bending stresses, run buckling checks on panels, and assess weld details and fatigue.

Allow expansion with sliding supports, decouple stiffeners, add expansion joints, or locally increase the section to avoid restraint-induced high stresses. If fully restrained thermal stress exceeds roughly 0.6·σ_y (about half yield), provide expansion allowances or revisit stiffener/layout and weld design.