Log/attempt for open source CFD / FEM framework for mast/sails simulation

Hello,

Since some months, I can't refrain a need to better understand mast / sails interactions for simple rigs like small landyachts, like very flexible free standing mast, or windsurfer rigs. Basically FSI - Fluid Structure Interractions <=> CFD + FEM. There are professional software packages for that. But as a hobbyist, this is not an option. I wondered if there would be open source options. Browsing some academic papers, I arrived to the below possibility:

• MBDyn for the FEM part. It can model beams (mast / boom / battens), membranes (sail), controls (mainsheet), has some aerodynamics features. MBDyn - MultiBody Dynamics - Homepage https://www.mbdyn.org/
• OpenFoam for CFD. This one seems the commonly used option for open source CFD and sails. https://openfoam.org/
• preCICE, coupling library for multi-physics simulations. https://precice.org/index.html

There has already been some coupling between the 3, to address some applications that have some similarities with my case, as in https://precice.org/community-projects.html#fluid-structure-interaction-of-inflatable-wing-sections or https://elib.uni-stuttgart.de/bitstream/11682/10645/1/FSI_Soft_Kites.pdf

I don't know up to were I will be able to go, as I'm a sailor, I like science, I know the aerodynamics basics applied to sails... but I have no experience with simulation software, CFD and FEM. However, I'm willing to learn. I expect progress to be slow ;-)

A researcher near MBDyn design team may provide some advive and support. This will be very helpful.

Any feedback and advice will be appreciated. Intention is to learn.

JMF

 

In the software section of the forum there is a recent thread about using Freecad for hull design.It also has workbenches for FEM and CFD,both of which are topics beyond my understanding.

 

I'm a FreeCAD user. I already used the FEM workbech for simple static simulations. Some people also work on a MBDyn workbench. This position FreeCAD in the landscape as an interesting piece to glue things together. However, from what I see the workbenches are not mature enough yet to be the central piece of the puzzle.

JMF

 

There's a 3 year trial of autodesk cfd for academics and students.

 

Thanks geo2 for the information.
Unfortunatly I'm neither a student nor an academic. 3 year trial still looks appealing (more than 3 or 6 month) to find a "workaround".

However, the OpenFOAM+MBDyn combo is still appealing.

I progress super slowly, but there is still no road blocker yet. I'm digging the MBDyn membrane modelling, and association with the flexible mast. The tool is powerful, but really oriented to "people that know". So reference documentation more than howto and tutorials... And a quite small community. Then the too is designed to be as generic as possible with (to my eyes at least) high level physics general laws considerations, that you then have to dertive to your basic case (exemple for a simple beam).

I think that once a use case will be set-up with the basics of a sail rig, then adapting it to different models should be much more easy.

Let's see if I can arrive to do someting. This will be the 2022 hobby project

Best regards,

JMF

 

Update about the slow process...

I have notr given-up but this is really tough for me. MBDyn certainly can do the job, but the lack of pre and post-processing makes the usage less intuitive and more difficult for the beginner.

The simulation of soft membranes seems a much more difficult problem than the solid simulation. It seems to easily not converge. Calculation time is quickly in minutes even on simple cases. The membrane needs to be tensionned all the time.

Then the assembly of curved panels (luff curves, panels to battens...) to straight structure: mast, boom, battens, is not straightforward.

I focus now on the structural simulation, and I'm investigating the Calculix solvers with Prepomax and Mecway. Someone helped me to build a very simple exemple with Mecway: flat triangular sail, simple mast, simple boom, uniform pressure, simple relaxing of the sail at the clew to create some draft (no curves, no batten, no roach, no coupling...) : Form-finding on tensile structures. https://mecway.com/forum/discussion/comment/5678/#Comment_5678

There is a lot of available papers on FSI, using CST or Shells... but not so easy to experiment for the hobbyist (at least me).

JMF

 

Hi JMF, the "SimWorks" cfd software is free and should cover the needs for your study. I have "scraped the surface" of it for comparison with others, and it seems to have a fairly straightforward workflow. Good luck!

 

I just had a look, ans it seems to me a GUI for Openfoam, and handle the CFD part. From the website, it does not looks to address the Structure part. I'm interested in the coupling of both, and I now struggle with the structure part (even if not the more difficult at the end ?).

JMF

 

About slow progress: please keep in mind that you have been working less than a year on this and all of the relevant prior art is by post docs or the result of PHD studies. It would be amazing if you achieved the easy general result in less than 8 years and a shame it doesn't earn you an advanced degree.
The software and methods you are pursuing look capable, but that doesn't mean that you are close to anything more than sending your computer into a coma.
I want to applaud your efforts but also feel the need to warn you how this calculation blows up even when you do it right.
I am not qualified, but if I was doing this I would try minimize each of the analyses and be sure they run efficiently separately before I linked them. Do the CFD on a rigid rig, then apply that loading on the mast to get it's deflection with a zero stretch sail, return that result to CFD....
Only add sail stretch when the spar and CFD stabilize. I honestly would look at the forces and see if I could find a sail material that made stretch irrelevant.

 

I fully agree with you Skyak. I don't focus yet on sailcloth stretch. However this is an embedded feature of the elements used in the FEM tool.

And as you propose, I start with only one side of the simulation: the structural part first and the interaction betwee the cloth part and the spar (skeleton).

Always learning something in the process

JMF

 

Like I said, I am not qualified but....I can only tell you what will be obvious pretty quickly.
If you can't avoid stretch embedded, at least you can set the cloth to super stiff to minimize the interactions/nonlinearity.

About getting comments from more qualified posters
-show some promising work and try to ask small concise questions
-you might even open it to opinions but that can get depressing or even insulting
There are quite a few here that have made these rigs with some success but it's an art and I would say even the great are not confident they can do a completely new design right in one try.
The only person I have seen post a complete rig with CFD linked to sail stretch is Mikko B. That was on (expensive) commercial software in the context of an Olympic sailing class.

(43) Sailing aerodynamics: From 6 to 20 knots in 10 seconds. A dynamic Finn simulation - YouTube

 

Hi,
I'm still very part time on that topic. I still believe that the target is achievable. However, because of my lack of academic knowledge on mechanics, I hit many walls. Main one up to know is FEA simulation of the sail cloth whith shell/membrane elements and the lack of bending stiffness that creates convergence issues.

To better understand the process and convergence conditions, I switched to the free version of Mecway (Mecway - Finite Element Analysis https://mecway.com/) that provides pre and post processors, and a clear workflow. When I will have someting working there, I will either switch back to "pure" Calculix solver or MBDyn.

One of the non obvious things is that in my use case with flexible mast, the t_zero condition is an already tensionned rig where the luff curve of the sail (shell) has to tie to a curved mast beams. I'm starting with straight mast and no luff curve... but this looses some of the application interest.

For the aero part, and to reduce the computational time, I identified a new interesting code, DUST (DUST – An aerodynamics solution for complex configurations https://www.dust-project.org/). "DUST is a novel, flexible solution to solve aerodynamics problems. Started as a collaboration between A3 by Airbus and the Department of Aerospace Science and Technology of Politecnico di Milano, it is meant to deliver affordable, reliable solution of aerodynamics problems on complex configurations. Focused on scalability and flexibility in the input, geometry definition, movement, and solution phase DUST allows the users to analyze diverse configurations with increasing and scalable levels of fidelity."

There is a coupling with Precice and MBDyn: "The main set of features introduced in this version are related to the coupling of DUST with MBDyn, using preCICE. This coupling allows to create a rare (if not unique) opportunity to use only free open-source software to build a complete aeromechanic tool chain, capable of simulating complex aircraft configurations with their complete physics, from flexible structures to aerodynamics, from dynamics to control. All these developments have already been extensively used and validated internally (see the publications), and now they are available to everybody."

Maybe this will end in something usefull... At least, in the process, I learn a lot ;-)

JMF