Learning Forum

[Kyle Benson]

We regularly run 3D thermal - mechanical simulations with hundreds of bolts. We have always used simplified cylindrical 3D bolt and nut bodies, but I would like to know if it is possible to achieve the same results with line bodies with shell/beam elements. I believe the problem will be in the transient thermal side of the analysis. On our 3D models, we typically use convective heat transfer on all the exposed surfaces and a macro to apply the combined convection/radiation effective heat transfer between the bolt holes and bolt shanks. I setup a simple model to start exploring whether its possible to do something similar on a line body, but even with only a convection applied to my block and all heat transfer coming through the contacts, I see a big difference in response between the 1D and 3D bolt elements.

The line body is on left and coarse and fine 3D bodies are on the right. This is relatively slow heat transfer (.001 BTU/s-sqin-F at 900F on all sides of the block for 30 min), but the line body is responding much faster than the 3D bodies (which are a decent match given the difference in mesh density).

[John Doyle]

I am suspicious of the convection loading between the bolt shanks and holes for the two methods. Also, it is not clear what you are doing with the shell elements, as it relates to the beams in this model. I believe WB-Mechanical is using SURF152 elements automatically on the 3D bolt shanks and holes surfaces, for the convection in the 3D solid model. I would recommend a careful review of the MAPDL elements reference manual for SHELL181, BEAM188/189 and SOLID185, SURF152 and commands SFBEAM and SFE in Commands Manual.
Also, do these models match for conventional structural loads like bolt pretension (load in LS1 and Lock in LS2) without any thermal loads applied? That might be a good test to help confirm where the source of the differences are.