SimScale CAE Forum

Is it possible to simulate a flow around body changing its shape during simulation?

My interest is a water and air and the interface between them as well.
The vehicles I am exploring are floating on the water and are driven by the wind.
I can provide many answers about how it can work, however, sometimes it looks very difficult or impossible.
One of such problem is that the CFD simulations I have been in contact with are not supporting objects which like to change the shape during simulation.
I do understand well that we are still a long way from simulating everything but a progress will always open unexplored areas of the world of fluids around solids…

Back to the problem I like to get a help with:

I got a design of a variable shape wing. This wing can change its AoA, thickness, camber and twist within seconds.
This also means that the flow around such wing and especially the laminar part can be influenced by manipulating the wing shape.
The goal will be to improve the aerodynamics of the wing and better understand this kind of wings in general.
Right now I do not have any answers what might work but the is all about - exploring…
So the question is: do you know if this kind of simulation is possible within Simscale?

If you like to get some feeling about the wing in action it can be observed on youtube:

Best Regards
Waldemar

1 Like

Hi @WaBe, Just wanted to say this is a really cool application. As for whether you can perform this analysis the way you describe, I believe the answer is no. You could do a parametric study and graph how each parameter change alters the wings aerodynamics, this would be a good starting point. However, I believe what you need in the long run is the ability to transiently alter these parameters to see how fluid reacts to change of the wing. I have had no exposure to Fluid Structure Interface simulations, however, I believe that an FSI simulation still wouldn’t be what you want, maybe someone in the know could further comment on this? FSI is not currently implemented on SimScale.

I think there was a webinar using CAESES that you may be interested in, essentially using CAESES and SimScale to do a parametric study of a design (such as yours).

Looking forward to seeing the results :slight_smile:
Darren

1 Like

Hi 1318980

Thank you very much for your kind answer - it helps to understand better where we are - you understand well what it is about.
Generally, I do believe that the problem is in meshing - when the object becomes dynamic in shape then the mesh must not only follow
but also change on the “fly” and that will increase the necessary computing power enormously.
However, I do believe that simulation power will increase in the near future so I can simulate the Coandă effect and explore the air mass inertia effects with variable shape bodies :wink:
On small dinghies (sailing) we have a phenomenon called “pumping” - this is when the AoA of the sail is alternated by sailors muscles.
The movement of the sail surface is changing the air flow so higher lift force can be produced. This is also this kind of effect I want to study.
I do have the CAESES on my computer but it looks that one needs to become dedicated dynamicist.
If the community dealing with such odd areas is small then we will have not enough confidence in the results…and this is holding me.

I will communicate about any progress if any as my projects are open to anybody.
Best Regards
Waldemar

1 Like

Hi Waldemar (@WaBe),

as Darren mentioned, this feature is not (yet ;)) implemented in SimScale. The idea you have in mind reminds me a bit of topology or shape optimization but that the shape “optimizes” or “worsen” during the mesh deformation process in your case. In Open$\nabla$Foam, which SimScale also uses as a solver, there is a feature called “Adjoint-Based Topology and Shape Optimization”, which is not part of the SimScale capabilities but which might be added if there is a demand from the user.

So the basic routine of your simulation would be that you either deform the mesh during the simulation which implies high computational costs or that you create a model for each deformed shape of the wing and be able to upload it here on SimScale which will be solved step-by-step from the solver which is not very time-efficient.

You could also use this Shape/Topology optimization to see what your wing has to look like with specific boundary conditions and change those boundaries and make the wing function in a way that only those “optimal” states can be achieved by deforming the wing. Or you could couple Topology optimization and shape optimization by getting a rough form of your optimal design with the TopoOpt and then using the ShapeOpt to fine-tune your component.

Does this somehow make sense to you? Very interesting project in my opinion!

Best,

Jousef

1 Like

Hi Jousef,

Nice to have contact with you Jousef and YES, it make sense what you wrote, as you try to see the opportunity
instead to say NO.

The difficult part of this method we are talking now is that what is expected from the simulation is the dynamics of the fluid during the simulation when the shape gets changed.
Therefore any discrete solution to simulate it partly in steps will not give the answers we are looking for as we will miss the effect of delayed behavior of the fluid due to its kinetic energy, inertia and more.
Imagine that when an airfoil got a laminar layer and the thickness of the airfoil is suddenly changed together with AoA the laminar layer may extend instead to be shorter and begins the turbulent phase sooner…and when such control gets repeated then it may lead to higher lift force.
This is not what I claim but it is only an example what difficult problem it is even difficult to test it in wind tunnel as we do not know what to expect.

When it comes to a request of adding such functionality to Sim Scale is not for me to do it - I am really a little player.
But my opinion is that the customers not always know what they want before they got an iPAD. (who need that was the first question).

We can see that the mechanical designs are less straight lines anymore due to 3D printing ability.
We can see that the objects like to move during simulations - Aeroplane just turning in the air?

I do believe that we can easily agree on this NEW feature and that this will be moving the CFD simulations forward and the customers will find many of application for that.
OK, they don’t know that yet…

Best Regards
Waldemar

1 Like

If you can accept solutions of a rigid body at varying AOA, use a solver with dynamic mesh, e.g., pimpleDymFoam https://www.youtube.com/watch?v=-UBLo158AzU. Apply a very small time step to predict the ‘transient’ effect.

But such solution may not be as representative as real world applications, where the body of interest also undergoes elastic deformation. In this case, you would need an FSI solution, more expensive but do-able in foam-extend.

2 Likes

Hi Dylan

It is always fun to see something well done - thank you for showing me the youtube clip.
Yes, you touch the sensitive button = more expensive - it is where my attention always goes down…but I will keep one eye on the FSI.
And you increased my way of thinking with the youtube clip - Thanks!

It is always possible to do it in the real world but that expensive too and that is telling us also something about the CFD technology as well - yet lot to do in this field!

Best Regards
Waldemar

1 Like