CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   Main CFD Forum (http://www.cfd-online.com/Forums/main/)
-   -   Immersed boundary method for Master Thesis (http://www.cfd-online.com/Forums/main/92874-immersed-boundary-method-master-thesis.html)

CharlieTan84 September 28, 2011 04:37

Immersed boundary method for Master Thesis
 
Dear people

Soon I would like to start on a Master Thesis about the implementation of Immersed Boundary Method on Turbulent Flows. I am more interested in aerospace applications. What do you think about this kind of Master Thesis? Is IBM applicable on aerospace or is it only good for Biofluidmechanics?

I know that my question is quite rough but any kind of comment would make me happy.

Thank you very much and have a nice day.

CharlieTan84 October 1, 2011 08:19

No idea from anyone?

leflix October 1, 2011 10:32

Hi,

Theoriticaly IBM are able to tackle all types of complex geometries problems, even in aerospace. IBM are very promising and attractive and also very elegant and smart concept. Bur for practical applications IBM are very limited. How many commercial packages use it?
Despite the fact that theoriticaly these methods can handle every kind of geometries, each time these methods have to demonstrated their capabilities in papers, authors solve the flow around a circle cylinder or a sphere. And one has to admit that in terms of complex geometries circle cylinder is the ground zero !!!


With all the respect we do have for Peskin the father of IBM, I also read on this forum that since 1972 he's still trying to improve this method. A huge quantity of papers have been published too just to try to improve this concept. So it does mean than these methods are largely perfectible. At the best second order can be achieved but in most of the cases it is first order.

I even don't know if there are commercial packages based on this concept. Most of them are based on body fitted approach. This would indicate that IBM didn't succeed to be attractive enough for commercial codes market. It is meaningfull....

To conclude, if you want to solve the flow around a square or a circle cylinder use IBM. If you want to solve the flow around an whole aircraft use body fitted approach.

arjun October 2, 2011 21:17

Quote:

Originally Posted by leflix (Post 326349)
Hi,

Theoriticaly IBM are able to tackle all types of complex geometries problems, even in aerospace. IBM are very promising and attractive and also very elegant and smart concept. Bur for practical applications IBM are very limited. How many commercial packages use it?
Despite the fact that theoriticaly these methods can handle every kind of geometries, each time these methods have to demonstrated their capabilities in papers, authors solve the flow around a circle cylinder or a sphere. And one has to admit that in terms of complex geometries circle cylinder is the ground zero !!!


With all the respect we do have for Peskin the father of IBM, I also read on this forum that since 1972 he's still trying to improve this method. A huge quantity of papers have been published too just to try to improve this concept. So it does mean than these methods are largely perfectible. At the best second order can be achieved but in most of the cases it is first order.

I even don't know if there are commercial packages based on this concept. Most of them are based on body fitted approach. This would indicate that IBM didn't succeed to be attractive enough for commercial codes market. It is meaningfull....

To conclude, if you want to solve the flow around a square or a circle cylinder use IBM. If you want to solve the flow around an whole aircraft use body fitted approach.


I disagree with lot of what you said. I think you are not well informed.


About the IBM and aerodymics part, it is difficult to very difficult to use IBM for aerodynamics, for the reasons that most of such simulations are trying to get drag and lift forces. Since most of the IBM formulations can not represent boundary surface as good as body fitted grid, they are at disadvantage. It does not mean that it is not used for aerodynamics, some people (including me) use it.

leflix October 3, 2011 11:13

I do not see the disagreement between you and me. I'm not a proIBM or even IBM defender. I do think that body fitted approach is much better.
If you read carrefully what I wrote, I said theoreticaly IBM can tackle all type of problems, but pratical applications show rapidly the limits.

arjun October 3, 2011 22:30

Quote:

Originally Posted by leflix (Post 326509)
I do not see the disagreement between you and me. I'm not a proIBM or even IBM defender. I do think that body fitted approach is much better.
If you read carrefully what I wrote, I said theoreticaly IBM can tackle all type of problems, but pratical applications show rapidly the limits.

Well there are lots of disagreements between what we two are saying.

#1 " I do think that body fitted approach is much better"

'Better' is based on situation and type of simulation required. We used almost all of commercial softwares including starccm,fluent, CFX etc and decided to dump them and now using only IBM based solvers those are inhouse codes.
If they were any better for us, we would be using them. Here are the problems with them:
a) Too slow for mesh sizes we want to run simulations at. (Amost 50 times slower than our IBM code).
b) can not handle or require too much resources to run very large simulations. (We did a calculation involving 3 billion case, try that with Fluent or StarCCM+)

#2 "I said theoreticaly IBM can tackle all type of problems, but pratical applications show rapidly the limits"

It is other way round for body fitted solvers too. Try a case with screw type compressor where two screws touch each other and rotate. Or try two complicated bodies hitting each other with body fitted approach and you will get the feeling at only premitive calculations that involve no grid movement or simple grid movements could be done with body fitted solvers.
With IBM I rotate and move things all the time, with no fuss and difficulty.

So no we are not saying the same thing.

leflix October 4, 2011 04:04

Dear Arjun,

After your first post I thought you were pro-body fitted so that's why I said that for me there were no disagreements
But after your second post I understood you are pro-IBM.
So it seems that there is a contradiction between what you say in your last two posts.

Quote:

Since most of the IBM formulations can not represent boundary surface as good as body fitted grid, they are at disadvantage.
Quote:


If they were any better for us, we would be using them. Here are the problems with them:
a) Too slow for mesh sizes we want to run simulations at. (Amost 50 times slower than our IBM code).
b) can not handle or require too much resources to run very large simulations. (We did a calculation involving 3 billion case, try that with Fluent or StarCCM+)


Or try two complicated bodies hitting each other with body fitted approach and you will get the feeling at only premitive calculations that involve no grid movement or simple grid movements could be done with body fitted solvers.
With IBM I rotate and move things all the time, with no fuss and difficulty.
So in one hand you say that IBM has some disadvantages compared to body fitted.
In the next post you say that IBM is much better, less time consuming, better for moving grids...

choose your side and be more clear.

I'm eagger to see some simulations with screws compressor rotating with IBM.
In which parallel universe this has been performed? ;-)

Can you simulate the flow over a whole aircraft landing with the landing gear, wheels, wings and reactors with IBM ?

arjun October 4, 2011 05:06

Quote:

Originally Posted by leflix (Post 326585)

choose your side and be more clear.


I am in side that gets my work done for my company. There is no such thing as pro IBM or pro body fitted. If body fitted solvers get the work done then use them and if IBM gets your work done then use them. There are lots of things where body fitted solvers are dis-advantage and there are places where IBM solvers have disadvantages. Trick is to know what you want to do and which will get the work done.

Plus I was responding to your points that every time someone uses IBM it is for simple geometry like cylinders etc. Which is not true because we (my company) has been using it for complicated geometries.

If you try to search around you will find that lot of good work is done with them.

PS: I have written both types of solvers and I am not partial to any of them. I like and enjoy them both.

VincentD October 8, 2011 04:22

Quote:

Dear people

Soon I would like to start on a Master Thesis about the implementation of Immersed Boundary Method on Turbulent Flows. I am more interested in aerospace applications. What do you think about this kind of Master Thesis? Is IBM applicable on aerospace or is it only good for Biofluidmechanics?

I know that my question is quite rough but any kind of comment would make me happy.

Thank you very much and have a nice day.
Hello there,

I have done a master thesis on a Immersed Boundary Method. First of all I would like to point out that there are different ways to immerse your boundaries. You should spend some time checking the different possibilities. Some are better for complex media, while others are better for more simple geometries.

I do not see the direct link of a Immersed Boundary Method to turbulent flows. It's a different way to apply boundaries to your flow profile, it's not a turbulent model. What kind of benefits do you expect to obtain from using a IBM in your research?

I used the IBM to determine flow profiles inside realistic porous media. It proved very succesfull. This was due to the complex geometry (which is more difficult to apply in other methods) and the large amount of forcing done by the IBM. This high forcing (and low packed bed Reynolds number) helped a great deal in terms of simulation time.

I'm no expert on aerospace problems, but I would say the benefit from using a IBM should come from complex structure. I think it is more important in those cases to pick the correct turbulence model.

Good luck and Enjoy!

Vincent

javad.rahro January 14, 2015 06:56

Thesis
 
Hi.
I have found a thesis in your subject. I think it will help you.

http://www.math.rug.nl/~veldman/Scri...erTechWisk.pdf

lovecraft22 January 14, 2015 15:01

Some papers on the validation of PowerFLOW that uses something similar to immersed boundaries:

http://exa.com/web_support_2012/exa_tech_pubs.html


All times are GMT -4. The time now is 22:56.