[cyrusIII]

Hi
I have seen some chemical engineers use COMSOL so I was wondering if any of you have experiences with this software. I guess my main question is, Is it capable of doing the CFD or FSI stuff? If yes what are the advantages and disadvantages in comparison to the other commercial/FREE CFD softwares?
Thank you

[elabbasi]

COMSOL is very good for CFD and FSI, especially if the flow is laminar. FSI is really good with COMSOL because you have the same interface for solids, fluids (and other physics), segregated and fully coupled (monolithic) solvers, and very good control over the deforming fluid mesh.

Here is an informative page about their CFD capabilities from their website (https://www.comsol.com/multiphysics/...okes-equations), and another one about their FSI (https://www.comsol.com/multiphysics/...re-interaction). Here is a link also to a recent webinar where I show several FSI examples (https://www.comsol.com/video/simulat...l-multiphysics) using COMSOL.

[cyrusIII]

Originally Posted by elabbasi

COMSOL is very good for CFD and FSI, especially if the flow is laminar. FSI is really good with COMSOL because you have the same interface for solids, fluids (and other physics), segregated and fully coupled (monolithic) solvers, and very good control over the deforming fluid mesh.

Here is an informative page about their CFD capabilities from their website (https://www.comsol.com/multiphysics/...okes-equations), and another one about their FSI (https://www.comsol.com/multiphysics/...re-interaction). Here is a link also to a recent webinar where I show several FSI examples (https://www.comsol.com/video/simulat...l-multiphysics) using COMSOL.

Thank you very much for the information and the links
What are the limitations of COMSOL in CFD and FSI? I basically mean if someone wants to choose between CMM or openfoam or ANSYS, or Fluent, etc why should he pick up COMSOL? what are the advantages and disadvantages?
I know you mentioned laminar flow, but do you have some more examples?

[elabbasi]

It really depends on your problem(s), and for relatively simple problems most of the codes will give you comparable answers. I find COMSOL ideal for most FSI and other Multiphysics problems, and very good with single physics models like CFD. Here is a link from the COMSOL website showing the list of capabilities they offer for fluid flow https://www.comsol.com/products/spec...ns/fluid-flow/.

If you look up any of their examples you’ll see how easy it is to set up a Multiphysics problem, including FSI. In an older webinar I showed how to set up and solve a MEMS switch problem that involves FSI with electrostatics and contact (http://www.veryst.com/project/Multip...of-MEMS-Switch). The set up was really easy. Convergence was a harder because of the complexity of the problem. I also have a peristaltic pump FSI example (https://www.comsol.com/paper/fluid-s...tic-pump-11574) that, depending on the characteristics of tube and fluid, can have very strong coupling between fluid and solid requiring a monolithic solver. At least when we did that example several years ago we could not find any similar pump models with any other code.

The main difference between COMSOL CFD and other CFD codes is that COMSOL is finite element not finite volume based. For CFD-only problems I found, like I mentioned, that turbulence problems are more efficiently solved with finite volume methods. I find some types of multiphase problems better solved with finite volume methods as well. Note that at some point in the past finite volume methods for CFD where way more developed than finite element methods, but that gap has been shrinking in my opinion.

I hope that helps!

[cyrusIII]

Originally Posted by elabbasi

It really depends on your problem(s), and for relatively simple problems most of the codes will give you comparable answers. I find COMSOL ideal for most FSI and other Multiphysics problems, and very good with single physics models like CFD. Here is a link from the COMSOL website showing the list of capabilities they offer for fluid flow https://www.comsol.com/products/spec...ns/fluid-flow/.

If you look up any of their examples you’ll see how easy it is to set up a Multiphysics problem, including FSI. In an older webinar I showed how to set up and solve a MEMS switch problem that involves FSI with electrostatics and contact (http://www.veryst.com/project/Multip...of-MEMS-Switch). The set up was really easy. Convergence was a harder because of the complexity of the problem. I also have a peristaltic pump FSI example (https://www.comsol.com/paper/fluid-s...tic-pump-11574) that, depending on the characteristics of tube and fluid, can have very strong coupling between fluid and solid requiring a monolithic solver. At least when we did that example several years ago we could not find any similar pump models with any other code.

The main difference between COMSOL CFD and other CFD codes is that COMSOL is finite element not finite volume based. For CFD-only problems I found, like I mentioned, that turbulence problems are more efficiently solved with finite volume methods. I find some types of multiphase problems better solved with finite volume methods as well. Note that at some point in the past finite volume methods for CFD where way more developed than finite element methods, but that gap has been shrinking in my opinion.

I hope that helps!

Great information and explanation. Do you know why they did not develop finite element in CFD from the beginning? I remember in equations there were something that could cause some weak points in solving problems.