Sail-Shell- Shear locking?

Before trying more difficult things with MPCs and curved edges, I started by more basic things under Prepomax. A simple shell modelled triangle (sail) tied to a vertical rod (mast), loaded by pressure on the triangle. The free end of the triangle is “fixed” with a little displacement to relax the sail cloth a bit.

TriangleMastShearLockingBC

However this is not working as expected. Instead of creating a nice bend in the thin shell/sail, the corner get pushed and “turned” in the initial plane, with very high stress in that area (and nowhere else). Even pushing the mast forward. Is it what is called shear locking ?

I tried with a quad dominated mesh. It improves a bit, but still does not creates the expected bend in the fabric. Prepomax implements the shell elements with S6 and S8.

Are there ways to improve my simulation ? Some obvious flaws in my approach. Better elements to use in Calculix for that use case ?

By the way, I wanted to try with membrane instead of shell, but Prepomax imports FreeCAD surfaces directly as shells. Is there a process to import membrane in Prepomax ?

Best regards,

JMF

What exactly do you mean ? Did you use a prescribed (non-zero) displacement value ? Can you share the file ?

Shear locking happens with first-order solid elements utilizing full integration so I wouldn’t worry about that here.

PrePoMax names parts “solids” and “shells” but this is just to differentiate between volume-based and surface-based geometry. You can still apply a membrane section to such part. Just keep in mind that membrane elements don’t have bending stiffness.

Non-linear analysis used?

The files are here: Dropbox - Triangle-Mast.zip - Simplify your life

I have put in the zip the prepomax and the .inp files.

In Prepomax, nlgeom is “False”

Trying to set it to “True” does not seems to converge…

JMF

For now, I would leave out that prescribed displacement (-10 mm) and solve it in just one step. The results look good then.

Hello [JMF],

That seems a complex problem to start.
There is a form finding example from Manuel in the MECWAY forum where I normally post.
I have also posted a file with an example that I hope can be useful to you too. It has less than 1000 nodes, so it is free to run.
I do not submit the inp file in this forum because for some reason the .inp files from MECWAY do not fully work when it’s runs directly in CalculiX.
The example uses CalculiX Nonlinear and the Tension only material as you suggest.

*MATERIAL,NAME=TENSION_ONLY

I have found that it is important to always keep some tension on the sail. If not, it “““buckles””” and easily fails to converge.
That’s the reason why there are some straps at the fabric perimeter.
You can play with mast stifness and the other parameters the see the effect on the fabric.

You have remind me my 80’s when I started to windsurf with my first Tiga Sail. Those shapes have now evolve so much. .:slight_smile:
Find also a small video of the result. Computation time should be no more than 5 minutes.

@Disla : thank you, I have found the example in MECWAY forum. I’m downloading Mecway and will try to execute it.

I have also sailed Tiga (and Sainval for those who would know about those old things). The objective would be to simulate today’s full batten sails from landyacht (free standing quite soft mast).

I thought that as this is explored in papers since at least 10-15 years, and present in sail design commercial software, this would be achievable with opensource softwares, at least for the simpler full batten sail. They even succeed to simulate spinakers which should be orders of magnitude more difficult than my case.

So… not so simple for somebody like me :frowning: I also discover the difficulty to have converging models.

@Calc_em : this prescribed displacement is an important feature for the sail. Very flat sails ave not efficient. Controlling the shape/curve of the sail is a big part of the game.

I will insist a bit more, but definitively not a problem for a beginner :wink:

JMF

@Disla: I just played with your example, ans this is definitively encouraging. Shapes are more realistic.

I have now to understand better how is built your model.

I understand that it runs with CAlculix solver, so no exotic Macway feauture used to achieve that result.

Some had very good results with MITC shell elements (Wrinkle development analysis in thin sail-like structures using MITC shell finite elements - Inria) but I have to explore first membrane tension only material way.

I’m happy it’s being useful and encouraging.
Don’t forget you are facing a complex problem.
It is not by accident that one of the most emblematic webpage images of Calculix is a paraglide (in many ways similar to what you are trying to solve).

Yes I know… and unfortunatly, I have not found the paraglide files as an example. At least it would have been interesting to have somer basics how it was done (which elements and simulation steps and characteristics…).

Let’s see up to where I can progress.

JMF