[cyrusIII]

Hi everyone,
My major is in solid mechanics but I am very interested to start learning about modeling physical phenomena involving solid fluid interactions or even more than that, i.e multiphysics simulations like chemo-electrical-solid-fluid interactions.
Anyway, I was wondering what would be your suggestions for a total beginner in this field.
I know about numerical methods but I never have done it for example for navier stokes equation. Is there any textbook and open source code that has done it step by step for Navier stokes so that I can learn it easily? Any code available for Matlab or Mathematica or C++ ?
I appreciate any software suggestions. I remember some years ago people were using Fluent Gambit. Is it still the same? I see some new software out there like Comsol. Which software do you suggest? I know it depends on the project and software but I like to hear your personal experience and opinions.
Sorry for my long post
thank you

[thegauravonline]

Well i am new to cfd fsi too...

Using OpenFoam.

[cyrusIII]

Originally Posted by thegauravonline

Well i am new to cfd fsi too...

Using OpenFoam.

Would you please explain about Openfoam? I checked the website and it looks like the tutorial section is for those who register for the workshops or something like that. would you please explain why you chose Openfoam? Is it because it free and open source? Do you use it in windows or Linux?

[thegauravonline]

Yes!! because its open source.
We can really make our own solvers.

Using in Linux Ubuntu 14.04

[cyrusIII]

Originally Posted by thegauravonline

Yes!! because its open source.
We can really make our own solvers.

Using in Linux Ubuntu 14.04

Have you tried installing it on windows?

[davidwilcox]

how about NEK5000? it's free and because it uses the spectral element method, it gives you less numerical diffusion compared to finite volume (in Fluent).

[FMDenaro]

Just a comment .... Nowadays a software is supposed to be the best teacher, no one think that learning CFD would be a task of a student in a classical course...
that makes me crazy ...

[Reeb]

Hi,

If you want a open source software I would recommend Openfoam but it is not an easy task to start using it. I would rather go for a commercial software and use the saved time to studie the literature for knowledge.

I think that the biggest commercial software's are Star-CMM+, Ansys Fluent, and Ansys CFX. In my opinion these software's are great for pure CFD applications but not for FSI simulations since they must use a third party solver to connect the solid and the fluid. Often it is quite hard to setup a FSI simulation and a lot of parameters that needs to be set in order to get the simulation running, speaking about CFX and Fluent since I have not used Star-CMM+ for FSI simulation.

For FSI simulation I would recommend LS-DYNA, they are mostly know for the structural explicit capabilities but they have developed a new CFD solver that shows really good FSI capabilities. They do not use a third party solver to connect the solid and the fluid since the CFD solver was made for FSI simulations from the beginning. It is really easy to activate the FSI simulation when the fluid and solid simulations are defined, you only specifies two additionally commands and then you are set for a FSI simulation. I know that the CFD solver is quite knew and may need some validation against experiments or other softwares but if you are interested in FSI simulation I would give it a chance.

[cyrusIII]

Originally Posted by davidwilcox

how about NEK5000? it's free and because it uses the spectral element method, it gives you less numerical diffusion compared to finite volume (in Fluent).

I am a total beginner, Would you please explain more about the three things you mentioned:
1.Spectral Element Method
2.Numerical diffusion
3.Finite Volume

[cyrusIII]

Originally Posted by FMDenaro

Just a comment .... Nowadays a software is supposed to be the best teacher, no one think that learning CFD would be a task of a student in a classical course...
that makes me crazy ...

It is interesting how you look at it. From your point of view, which software (free or commercial) is best to start learning? What is your favourite and why?

[cyrusIII]

Originally Posted by Reeb

Hi,

If you want a open source software I would recommend Openfoam but it is not an easy task to start using it. I would rather go for a commercial software and use the saved time to studie the literature for knowledge.

I think that the biggest commercial software's are Star-CMM+, Ansys Fluent, and Ansys CFX. In my opinion these software's are great for pure CFD applications but not for FSI simulations since they must use a third party solver to connect the solid and the fluid. Often it is quite hard to setup a FSI simulation and a lot of parameters that needs to be set in order to get the simulation running, speaking about CFX and Fluent since I have not used Star-CMM+ for FSI simulation.

For FSI simulation I would recommend LS-DYNA, they are mostly know for the structural explicit capabilities but they have developed a new CFD solver that shows really good FSI capabilities. They do not use a third party solver to connect the solid and the fluid since the CFD solver was made for FSI simulations from the beginning. It is really easy to activate the FSI simulation when the fluid and solid simulations are defined, you only specifies two additionally commands and then you are set for a FSI simulation. I know that the CFD solver is quite knew and may need some validation against experiments or other softwares but if you are interested in FSI simulation I would give it a chance.

Thank you for putting time. I think, so far, you have been the only one that named couple of different softwares and their strengths on different applications. Here are my questions:
1.When you say pure CFD would you bring some examples? Because when I think about it, any simulation is involved fluid and chemical reactions and solid structure even simulating water running in the pipe at least it affects boundary conditions. I appreciate some examples.
2. Would you name some of those "third party solvers to connect the solid and the fluid" if I choose to use Fluent, Ansys or Star CMM ? I have heard the name Gambit always with Fluent, is that what you mean or it is just for meshing?
3.What are the examples of "parameters that needs to be set" for the FSI. Do you mean physical parameters? BCs? or parameters like geometrical and meshing stuff within the software?

[FMDenaro]

Originally Posted by cyrusIII

It is interesting how you look at it. From your point of view, which software (free or commercial) is best to start learning? What is your favourite and why?

a software is just a tool and has its handobook...actually, CFD is a discipline and has its time and its requirements to fulfill with that

[cyrusIII]

Originally Posted by FMDenaro

a software is just a tool and has its handobook...actually, CFD is a discipline and has its time and its requirements to fulfill with that

I understand your point, but I think I have the basics but I have not used them or not applied them to real problems like what I do in solid mechanics. In fact I never learnt nonlinear modeling of solids in school but I learnt is through the software and wherever I needed more explanations I just referred to some texts or handbooks. I would never learn it if it was not because of the software and the real industrial projects.

[Reeb]

Originally Posted by cyrusIII

Thank you for putting time. I think, so far, you have been the only one that named couple of different softwares and their strengths on different applications. Here are my questions:
1.When you say pure CFD would you bring some examples? Because when I think about it, any simulation is involved fluid and chemical reactions and solid structure even simulating water running in the pipe at least it affects boundary conditions. I appreciate some examples.
2. Would you name some of those "third party solvers to connect the solid and the fluid" if I choose to use Fluent, Ansys or Star CMM ? I have heard the name Gambit always with Fluent, is that what you mean or it is just for meshing?
3.What are the examples of "parameters that needs to be set" for the FSI. Do you mean physical parameters? BCs? or parameters like geometrical and meshing stuff within the software?

Hi,

Sorry for late reply.

1. When I say pure CFD I mean when you see the fluid as a continuum. The pipe walls only affect the flow with a non-slip boundary condition and depending on implementation and application a wall roughness coefficient can be used.

2. Gambit is for mesh and not for connecting the CFD and mechanical solver. Regarding Ansys you can look at this page: https://support.ansys.com/staticasse...teractions.pdf
It is for version 14.0 and I do not know how much that has changed to the current 17.0 version.

What I mean is that both Star CMM and Ansys Fluent/CFX uses the finite volume method and the structural solver uses the finite element method. This means that the loads and displacement needs to be "translated" between the different element formulations. If the CFD solver also uses FEM this step is not needed, can not say which method that is best.

3. I have not used Star CMM so can not say anything about how that works but I think that it uses a similar approach. What I mean is the number of coupling loops and field loops see page 26 of previous pdf, in some cases you also need to define a damping factor that regulate how much of the displacement/load is transferred between the solvers. This is done to get a better convergence in the coupling and to avoid to big changes. A lot can of course have changed since I performed the FSI simulation in the beginning of 2014 with the 14. something version.

I hope that gives you an idea about the coupling procedure, in my opinion I think that it is much easier to perform a FSI simulation in LS-DYNA, one more advantages is also that when you pay for the software all solvers are included in the price compared to many others where you pay for each solver. This is only my personal opinion and as I said a lot can have changed since I used Ansys, I have not used any of their newer versions.

Best regards

[thegauravonline]

Originally Posted by cyrusIII
Have you tried installing it on windows?

uh Yess!!
I installed in Windows as well.
Success.

[kaya]

I am not the most experienced one ever, but here's my contribution.

i think there is a fine line in between using CFD and learning CFD.

for the usage alone, just pick an engineering software from the shelf and it will do the job, don't try to understand the details of different software packages, you most likely won't get them anyway. you fllow some intro tutorials and shouldn't take you more than some days that you get introduced to the world of colorful fluid dyanmics.

If you can't spare the time / afraid of math or getting your hands dirty / lazy but still have the 'I want to learn' mind set, I suggest pushing the openFoam path. A steeper learning curve, and it will take longer to set up simulations where you have more freedom, more user defined picks, more frustration but the community is huge and the best part is you can dig it as long as your knowledge allows you. In my area industry also seems to be leaning on openFoam-derived codes.

Having said all these the actual answer to your question is, get a cfd book - red Ferziger is an easy to read one. FMDenaro can perhaps tell you which book is good on what but I bet this asked millions of times in the forum.

[cyrusIII]

Originally Posted by kaya

I am not the most experienced one ever, but here's my contribution.

i think there is a fine line in between using CFD and learning CFD.

for the usage alone, just pick an engineering software from the shelf and it will do the job, don't try to understand the details of different software packages, you most likely won't get them anyway. you fllow some intro tutorials and shouldn't take you more than some days that you get introduced to the world of colorful fluid dyanmics.

If you can't spare the time / afraid of math or getting your hands dirty / lazy but still have the 'I want to learn' mind set, I suggest pushing the openFoam path. A steeper learning curve, and it will take longer to set up simulations where you have more freedom, more user defined picks, more frustration but the community is huge and the best part is you can dig it as long as your knowledge allows you. In my area industry also seems to be leaning on openFoam-derived codes.

Having said all these the actual answer to your question is, get a CFD book - red Ferziger is easy to read one. FMDenaro can perhaps tell you which book is good on what but I bet this asked millions of times in the forum.

Thank you for the software and Book suggestion. I definitely give them a try. It has been a long time from my graduation in Mechanical Eng area and I do not afraid of math, the math is Algebraic and Continuum mechanics both of which I am/was good at! I want to learn CFD because I am interested in it otherwise my job has nothing to do with CFD.

[cyrusIII]

Originally Posted by thegauravonline

uh Yess!!
I installed in Windows as well.
Success.

Thank you, I will try it on windows do you know if other packages or plugins that I should install it? I mean for example in Fluent I should install Gambit and Exceed too is that the same for OpenFOAM?

[cyrusIII]

Originally Posted by Reeb

Hi,

Sorry for late reply.

1. When I say pure CFD I mean when you see the fluid as a continuum. The pipe walls only affect the flow with a non-slip boundary condition and depending on implementation and application a wall roughness coefficient can be used.

2. Gambit is for mesh and not for connecting the CFD and mechanical solver. Regarding Ansys you can look at this page: https://support.ansys.com/staticasse...teractions.pdf
It is for version 14.0 and I do not know how much that has changed to the current 17.0 version.

What I mean is that both Star CMM and Ansys Fluent/CFX uses the finite volume method and the structural solver uses the finite element method. This means that the loads and displacement needs to be "translated" between the different element formulations. If the CFD solver also uses FEM this step is not needed, can not say which method that is best.

3. I have not used Star CMM so can not say anything about how that works but I think that it uses a similar approach. What I mean is the number of coupling loops and field loops see page 26 of previous pdf, in some cases you also need to define a damping factor that regulate how much of the displacement/load is transferred between the solvers. This is done to get a better convergence in the coupling and to avoid to big changes. A lot can of course have changed since I performed the FSI simulation in the beginning of 2014 with the 14. something version.

I hope that gives you an idea about the coupling procedure, in my opinion I think that it is much easier to perform a FSI simulation in LS-DYNA, one more advantages is also that when you pay for the software all solvers are included in the price compared to many others where you pay for each solver. This is only my personal opinion and as I said a lot can have changed since I used Ansys, I have not used any of their newer versions.

Best regards

Thank you, Reeb, for the PDF file. Which one you prefer personally, LS-Dyna or Ansys? Which one is more user-friendly and have better community for help or build in help?

[thegauravonline]

Originally Posted by cyrusIII

Thank you, I will try it on windows do you know if other packages or plugins that I should install it? I mean for example in Fluent I should install Gambit and Exceed too is that the same for OpenFOAM?

I am using Salome for geometry

and Paraview for post processing