Modelling of H steel column using 3D Shell element

[BATUHAN]

I need to model H steel column using 3D Shell element but I am not sure how to define axial load and boundary conditions?

I found this procedure shown in below from an article.
For compression members, fully-fixed supports were considered in one end and free in the other, the displacements were also restrained in the transversal direction (Uy) and vertical direction (Uz) in the intersection of the web with the flange.

What do you recommend me? For example what is the difference between line and point constraints. Should I apply constraints to all edges or lower flange for simply supported beam?

hank you for your interest

 

[FEA way]

Which software do you use ? Boundary conditions should generally represent real life supports or assumptions of analytical model (used for reference) as closely as possible

 

[MegaStructures]

First of all I probably wouldn’t bother using shell elements unless you have a variable cross section, or some type of non-uniform loading that you need to check local effects for. For global design of a standard structural shape I would use BEAM elements and a design code that considered all linear and non-linear buckling effects. In my region AISC would be an excellent resource.

Is there something atypical we should know about your column? Also we’ll need to know your software package to advise you on how to add loads and constraints.

 

[BATUHAN]

I am supposed to use shell elements to study on the performance of column at elevated temperature. I use ABAQUS for my study.

 

[MegaStructures]

Are you very new to FEA? No offense, but I'm surprised by your original question if you are doing such an advanced study. The boundary conditions should match the behavior of the actual column connection. If your column is pinned you shouldn't pin all of the nodes on the bottom edge. Why? Because if you limit translation of each node you will limit rotation of the column. I've never used ABAQUS but there should be a rigid spider command. What you can do is create a node at the center of your bottom edge and send rigid spiders out to all of the nodes on that edge, then pin the center node. This should limit translation of the column, but allow for end rotation.

Make sure you understand the behavior of the column before plugging information into FEA and blindly trusting the results.

 

[FEA way]

In Abaqus such connection between nodes and reference point can be realized using either couplings (kinematic/distributed) or rigid body comstraints.