[JasonG]

I have recently installed OpenFOAM-1.6, and have started going through the documentation and tutorials to learn how to use this resource. Before I invest too much time trying to learn the ropes I wanted to ask more experienced users if my plans were feasible within the realm of OpenFOAM.

I need to simulate the spray atomization of a fluid (water in these initial tests) as it is expelled from an injector at various pressures. My goal is to simulate the effect changing certain geometries of the injector has on the spray geometry.

Does this sound like something I should be able to model in OpenFOAM?

I do have access to Ansys, and would prefer to do any preprocessing with that if it is possible to interface my model and any mesh information from that.

Thanks

-Jason

[alberto]

Hi Jason,

what kind of model do you want/need to use? VOF? Euler-Lagrange? Euler-Euler?

It is usually possible to export meshes from Ansys-FLUENT to openFOAM.

Best,
Alberto

[JasonG]

Originally Posted by alberto

Hi Jason,

what kind of model do you want/need to use? VOF? Euler-Lagrange? Euler-Euler?

It is usually possible to export meshes from Ansys-FLUENT to openFOAM.

Best,
Alberto

Alberto,

I am not yet sure which model will best suit my needs, I know I will have high pressures (1,000-2,000 bar) with multiphase flow. I'm just beginning to really get started on this project and still have a lot of research and reading on the different model types.

I just discovered that my university only has an Ansys license for the FLOTRAN solver within Classical Ansys and the CFX mesher within Workbench. I have been familiar with Ansys for structural analysis, and would like to do what I can with in it; but I was under the impression it would not be ideal for multiphase flow.

Thanks,

Jason

[alberto]

You should be able to export from Ansys to openFOAM. Check the formats OpenFOAM can convert from and those to which Ansys can export.

[JasonG]

Originally Posted by alberto

You should be able to export from Ansys to openFOAM. Check the formats OpenFOAM can convert from and those to which Ansys can export.

Thank you. Do you think that what I am attempting to accomplish is something that I should be able to represent within OpenFOAM? I am still learning the basics, but from what I have gathered I believe I will have to determine which solver will best suite my needs. From there I will have to start a new case and write the necessary files that the solver will use as an input.

[alberto]

Hi Jason,

it really depends on what model you decide to use. OpenFOAM at the moment has a good VOF and Lagrangian implementation, while the Euler-Euler approach (two-fluid/multi-fluid) is not as advanced.
For high-pressure sprays in engines, I saw some work done with Lagrangian methods (or similar, like the KIVA code), but also VOF was employed for the primary atomization.

Writing the solver configuration files is not that hard. You can take example from the tutorials and eventually ask on the discussion board, if you can share the details of the work.

What might be more complicated is adding functionalities to the solvers, if you need to do so. In such a case, the learning curve might be long, and more considerations are necessary to decide if OpenFOAM is the right tool or not.
Speaking frankly, if you plan to use OpenFOAM in the future, and it has the functionality you need (or it is not too far from that), it is probably OK to invest time at the beginning to learn how to use it. The time required, of course, depends on what background you have in CFD and numerics.

If it is a "one-time" task, or you choose to use an approach for which you can find a specialized code which can work more easily for you, it is more convenient to choose accordingly and save time.

Best,
Alberto

[JasonG]

Originally Posted by alberto

Hi Jason,

it really depends on what model you decide to use. OpenFOAM at the moment has a good VOF and Lagrangian implementation, while the Euler-Euler approach (two-fluid/multi-fluid) is not as advanced.
For high-pressure sprays in engines, I saw some work done with Lagrangian methods (or similar, like the KIVA code), but also VOF was employed for the primary atomization.

Writing the solver configuration files is not that hard. You can take example from the tutorials and eventually ask on the discussion board, if you can share the details of the work.

What might be more complicated is adding functionalities to the solvers, if you need to do so. In such a case, the learning curve might be long, and more considerations are necessary to decide if OpenFOAM is the right tool or not.
Speaking frankly, if you plan to use OpenFOAM in the future, and it has the functionality you need (or it is not too far from that), it is probably OK to invest time at the beginning to learn how to use it. The time required, of course, depends on what background you have in CFD and numerics.

If it is a "one-time" task, or you choose to use an approach for which you can find a specialized code which can work more easily for you, it is more convenient to choose accordingly and save time.

Best,
Alberto

Thank you for the input. I am currently a nearing completion of my undergraduate degree in mechanical engineering and I have a stronger background in structural analysis, but I have had an interest in CFD. I may continue my education to either the masters or doctoral level, in either case I would be continuing this research that requires this simulation.

My advisor was aiming to have some simulations by November, so I may need to re discus the time table.

Thanks again,

Jason

[alberto]

Originally Posted by JasonG

Thank you for the input. I am currently a nearing completion of my undergraduate degree in mechanical engineering and I have a stronger background in structural analysis, but I have had an interest in CFD. I may continue my education to either the masters or doctoral level, in either case I would be continuing this research that requires this simulation.

Well, at this point, learning OpenFOAM would be a good idea, since you can afford to take spend some time to learn it and deepen your knowledge of C++ (if you don't know it already).

My advisor was aiming to have some simulations by November, so I may need to re discus the time table.

Well, it is a very optimistic deadline for someone who never worked on multiphase flows, but if everything is in the code already, it might be feasible.

Probably answering to the following questions will help to decide what model to use and to understand if it is available in OpenFOAM:

- Do you need to study the primary atomization or can you assume an atomizer model?
- Do you need to account for evaporation or other forms of mass transfer?

Best,
Alberto

[JasonG]

Originally Posted by alberto

Well, at this point, learning OpenFOAM would be a good idea, since you can afford to take spend some time to learn it and deepen your knowledge of C++ (if you don't know it already).



Well, it is a very optimistic deadline for someone who never worked on multiphase flows, but if everything is in the code already, it might be feasible.

Probably answering to the following questions will help to decide what model to use and to understand if it is available in OpenFOAM:

- Do you need to study the primary atomization or can you assume an atomizer model?
- Do you need to account for evaporation or other forms of mass transfer?

Best,
Alberto

For these initial simulations I should be able to assume an atomizer model and no other forms of mass transfer. My primary objective is to first show the effect various changes to the injectors geometry may have on the spray geometry, and I think it may be reasonable to simplify the problem to get started.

I have been attempting to run OpenFOAM through an emulator with Ubuntu installed so I didn't have to completely switch over from Windows, but after running some of the tutorials it seems my computer will not be able to handle running it through the emulator. I think my next step will be to re-install everything on a standalone version of linux, and try to learn what I can from the tutorial cases.

-Jason

[alberto]

Originally Posted by JasonG

For these initial simulations I should be able to assume an atomizer model and no other forms of mass transfer. My primary objective is to first show the effect various changes to the injectors geometry may have on the spray geometry, and I think it may be reasonable to simplify the problem to get started.

So you already know how the liquid is converted in droplets, and you know how to represent this, right? If so, you can use the lagrangian approach in OpenFOAM, which has already some injector model in it.

If you need to understand how the geometry influences the formation of droplets, you cannot do that, and need VOF.

I have been attempting to run OpenFOAM through an emulator with Ubuntu installed so I didn't have to completely switch over from Windows, but after running some of the tutorials it seems my computer will not be able to handle running it through the emulator. I think my next step will be to re-install everything on a standalone version of linux, and try to learn what I can from the tutorial cases.

Well, running a CFD code in a virtual machine is going to be slow.
If you use a 64 bit machine, I would recommend you use openSUSE 11.1 (biased suggestion, I'm involved in openSUSE community ). You'll have linux, OpenFOAM will work like a charm on it, and the desktop will also be usable (small hint: pick GNOME, since KDE 4 is not mature yet in that release), with nice additions like flash, java and other small details working out of the box also on 64 bit machines. For help on openSUSE, join the #suse channel on Freenode, they're usually friendly and helpful.

Best,
Alberto

[JasonG]

Originally Posted by alberto

If you need to understand how the geometry influences the formation of droplets, you cannot do that, and need VOF.



Well, running a CFD code in a virtual machine is going to be slow.
If you use a 64 bit machine, I would recommend you use openSUSE 11.1 (biased suggestion, I'm involved in openSUSE community ). You'll have linux, OpenFOAM will work like a charm on it, and the desktop will also be usable (small hint: pick GNOME, since KDE 4 is not mature yet in that release), with nice additions like flash, java and other small details working out of the box also on 64 bit machines. For help on openSUSE, join the #suse channel on Freenode, they're usually friendly and helpful.

Best,
Alberto

I think I will look into the VOF approach. Is openSUSE 11.1 at all comparable to Ubuntu in regards to the user interface? The only reason I loaded Ubuntu was simply that I had a friend running it, and I know that when I do a full install it has a partition utility so I will not have to wipe my Windows install.


Thanks again for all the advice.

-Jason

[alberto]

Originally Posted by JasonG

Is openSUSE 11.1 at all comparable to Ubuntu in regards to the user interface? The only reason I loaded Ubuntu was simply that I had a friend running it, and I know that when I do a full install it has a partition utility so I will not have to wipe my Windows install.

First of all, if you prefer ubuntu, use it. It is really a question of taste, and mine was only a suggestion, and also biased, since I have been using openSUSE for years now.

However, yes, openSUSE has a graphical installer which is probably one of the best around in my opinion, and has surely more graphical configuration tools (YaST) than Ubuntu. You can of course re-partition your disk directly from the installer, and your Windows partition will be resized and made readable from Linux too.
You can see the steps of the installation of openSUSE 11.1 from the DVD (which I recommend, if you have a fast connection, since it brings you all the tools you need in one single medium) here:

http://en.opensuse.org/Installation/11.1_DVD_Install

The live-cd installation is similar. The installer follows almost the same steps, but it is started from inside the live system. You can see it here for the KDE cd, but of course it is similar also for the GNOME cd:

http://en.opensuse.org/Installation/11.1_Live_CD

Best,
Alberto

[JasonG]

Thanks! I will have to try to download the DVD when I get on campus with the better connection speeds.


-Jason

[JasonG]

I received some more clarification on my roles in this project, and I will only be modeling the atomization of a diesel fuel as it is injected with a known angle and initial velocity or pressure. I will need to develop a dynamic mesh to incorporate the changing volume of a cylinder, and can initially model the lower boundary as a flat surface. I am going to start by installing OpenSUSE and re-installing OpenFOAM, as I don't think the emulator will get me much further.

[alberto]

Hi, this is good news for you, I guess, since similar work has already been done in OpenFOAM, and you should be able to find references or contact who did it.

Good luck, and ask if you have problems :-)

[JasonG]

Originally Posted by alberto

Hi, this is good news for you, I guess, since similar work has already been done in OpenFOAM, and you should be able to find references or contact who did it.

Good luck, and ask if you have problems :-)

Thanks! I am just having a few issues getting a full install of linux. I only have an external CD drive for my laptop, and I am running into issues with being able to boot from it... but that topic is slightly outside the scope of this forum .

-Jason

[eugene]

Hi Jason,

I did some VOF modelling of primary diesel spray atomisation a few years ago. If you send me you e-mail address I can forward the pdfs of the papers we published on the subject to you.

Eugene

PS. Not sure if you can see my e-mail address. If not, just send me your address via the forum messaging service.