Which software do you recommend for capillary channel flow?

[member 428835]

Hi PF!

I'm trying to model a fluid dynamics problem where there is flow in a wedge. Length scale is 160mm and two fluid phase: air sits on top of silicone oil. Boundary conditions can be to specify height conditions, flow rate, et cetera. Very low gravity, so surface tension is very relevant.

I've asked people in my lab and some say you could get OPENFOAM to do this, but to ask around. So that's what I'm doing here.

 

[bigfooted]

Sure, openfoam can handle such problems. It also takes a bit more time to learn than fluent because of the lack of a proper gui and the bad documentation. We had a student run a test case in fluent and in openfoam. He could validate the case in fluent in a day, with openfoam it took a couple of weeks to get the same result. So, if you just want to solve the problem and you have a license available, use fluent. If you are really keen on learning high-level c++ and the details of numerical methods (and maybe, after a couple of weeks, solve your flow problem), use openfoam. Also make sure you have some validation test cases to check the results.

You could also check out the gerris flow solver. It is much more specialized than openfoam, so you are not able to add your own physics models, but it is also much easier to use. It can also deal with two-fluid flows, surface tension dominant flow is actually it's specialty. It doesn't have turbulence models, so if you have high reynolds numbers you have to resolve all of the turbulence features, which is quite expensive. Check out if it meets your demands. It might save you some time compared to openfoam.
http://gfs.sourceforge.net/wiki/index.php/Main_Page

 

[member 428835]

Sure, openfoam can handle such problems. It also takes a bit more time to learn than fluent because of the lack of a proper gui and the bad documentation. We had a student run a test case in fluent and in openfoam. He could validate the case in fluent in a day, with openfoam it took a couple of weeks to get the same result. So, if you just want to solve the problem and you have a license available, use fluent. If you are really keen on learning high-level c++ and the details of numerical methods (and maybe, after a couple of weeks, solve your flow problem), use openfoam. Also make sure you have some validation test cases to check the results.

You could also check out the gerris flow solver. It is much more specialized than openfoam, so you are not able to add your own physics models, but it is also much easier to use. It can also deal with two-fluid flows, surface tension dominant flow is actually it's specialty. It doesn't have turbulence models, so if you have high reynolds numbers you have to resolve all of the turbulence features, which is quite expensive. Check out if it meets your demands. It might save you some time compared to openfoam.
http://gfs.sourceforge.net/wiki/index.php/Main_Page

Thanks for the detailed response! Let me elaborate: I'm a second yea PhD student in fluid mechanics and am responding to a grant solicitation that lasts for 2 years. My proposal last year was denied but they asked me to resubmit. The main weakness they identified was lack of a sophisticated numerical tool to simulate 3D transient flow problems with dynamic contact lines.

I'm saying all this detail because I would have a lot of time to develop a robust numerical tool that models the aforementioned physics if I receive the grant. Does this change your recommendation.

As of now, several of my lab members know OpenFOAM to some extent, though I mainly work in MATLAB and Mathematica. I'm open to new techniques, and have a good background in numerical methods. I don't need to worry about turbulence though.

 

[bigfooted]

Well in that case openfoam might be for you: it seems that you actually want/need to dive into the details of the solver, and that you *must* use a cfd code where you can access the source code because you need to implement your own model.
Still, there are several good and well maintained codes out there (kratos, gerris, code saturne or the caelinux project), and in some codes it might be easier to implement what you want than in others.
If there is already openfoam knowledge in your lab, and no experience with other open source cfd codes, it might be better to go for openfoam. The name is also well known and for a grant proposal this might be a benefit. But I think you have to do some more investigation to check how much is in openfoam and what you would need to implement to really start your research