Thermal load with thermal gradient in structural analysis

yes, in aerospace structural analysis.
thermal loads from FEA are often high because FE models tend to be over stiff (missing lots of local flexibilities).
what type of model - beam, plate, shell, solid?
how are you defining CTEs?
how are you defining boundary conditions?

Thank you for anwsering.
My structural building is about 70*80 meters composed of buildings connected with expansion joints : shells and beams. My only structural material is concrete with 10^-5 K-1 for CTE and for boundary conditions , My foundation piles are blocked in translation and free to rotate.
I remarked some stress singularities in many corners in the lower parts of walls : vertical stress explodes by 4-5 times in very small area which explains the results I'm having .
I still don't have explanation for these singularities ( if it's mesh related or something else , even my mesh is quite fine with 0.5 meters elements )
Any Ideas ?

hmm, surely there is steel rebar in the building? you need to account for that in the analysis.

otherwise post some pictures of the areas with singularities, with stresses and boundary conditions. I suspect the boundary conditions are too rigid.

Impossible to answer without more information. But getting large loads or stresses due to thermal loads is not surprising in itself. The structure wants to deform in some manner due to the geometry and the thermal load. And the boundary conditions prevent that deformation.

One option is to remove the constraints and let it move freely, then you can see what your supports prevents.

the engineer from France said:

My foundation piles are blocked in translation and free to rotate.

Click to expand...

The resistance to translation should reflect the conditions of the soil the piles are embedded in. With typical granular soils, the soil won't provide much resistance with the amount of movement you'll get from thermal expansion/contraction. Soil is usually modeled as a series of non-linear springs. If your software can model it that way, that's the most accurate, but if not just standard linear springs will get you close.

Unless you've done some special detailing to provide a hinge at the pile connection, they are typically modeled fixed or partially fixed for rotation to reflect the restraint that's present and moment in the piles.