What would be a smart way to do convergence test for this model?

[drennon236]

I have three cylinder; inner steel part, middle concrete part, and outer steel part. I want to do a convergence study on the concrete cylinder in the middle (in red). What loads should I apply to the model and what output should I check the convergence for? I was thinking of applying normal pressure on the cylinders to activate coulomb friction, turning gravity on and checking for convergence of max and min principal stress at the bottom of the concrete cylinder. What else sort of convergence study would be smart to do, I am open to any suggestions.

 

[FEA way]

Convergence depends on load and boundary conditions so it has to be checked using the same settings as the analysis that you want to verify.

 

[drennon236]

The bottom steel part (Monopile) will be fixed, there will be hydrodynamic pressure on the cylinders, and I will apply moment and loads in xyz direction to the transition piece (upper steel part) on top of it via kinematic constraint. Then I have to check for converge of max/min principal stress in the grout. Perhaps the best place to check for convergence is at the bottom of the grout (middle concrete part) because there the hydrodynamic pressure will be largest?

 

[IceBreakerSours]

Some background: Space and time are discretized in this class of models (i.e., FEM). In space, the solver spits out a variety of fields (displacement, strain, stress, etc.) that evolve over time (or load steps in a static analysis). Based on those fields you may want to focus on a quantity of interest to you (e.g., reaction force, surface strains, max. principal stress, etc.) in a particular location (in geometric space, not a node).

The goal of a "convergence analysis" in this context is to refine the spatial and time discretization to make sure the quantity of interest reaches a sort of a steady state at a chosen location. In other words, the quantity of interest at a specific location must be (almost) independent of the number of elements and time step size. It is tricky to pick a quantity of interest at a predefined geometric location as you refine the mesh but if you pick the quantity of interest from a node, then you have to be aware of the impact of the extrapolation and averaging that happens during post-processing.

The more you think about it, the trickier it gets. In practice, one ends up making some approximations and leveraging engineering judgment to say the quantity of interest is converged.

*********************************************************
Are you new to this forum? If so, please read these FAQs:

http://www.eng-tips.com/faqs.cfm?fid=376

http://www.eng-tips.com/faqs.cfm?fid=1083