This posting illustrates a common mistake for calculating thermal effects.By default, the temperature of every point (node) in the model is set equal to the value specified by "Reference temperature at zero strain" in the Flow/Thermal Effects tab of the properties of the study dialog.
A common mistake is that users specify temperatures to faces and/or edges/vertices and run the study. The results will not look right due to the sudden change in temperature between the boundary and all other inner nodes since these nodes will have the reference temperature.
For example, consider a static study of a bar fixed at two opposite ends as shown:
The user applied temperatures of 600 and 300 F to the right and left ends, respectively, and ran the study. A von Mises stress plot for this case is shown below:
The stress is highly concentrated at the faces where the temperature has been specified. The user may have thought that the static study will vary the temperature linearly between the two end. That is not true.
The proper way to handle this is to create and run a thermal study first to calculate the temperature profile and then use the thermal study to define the thermal loading for the static study.
The von Mises stress plot in this case looks as expected.
The concentrations at the corner is due to the "fixed" restraint.
The note holds for buckling, frequency, and nonlinear studies. For nonlinear studies, it is proper to use results of a transient study that spans the same time period. Note also that the meshes of the referenced thermal study and the referencing study DO NOT have to be similar.