CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   Main CFD Forum (http://www.cfd-online.com/Forums/main/)
-   -   Immersed boundary method (http://www.cfd-online.com/Forums/main/13549-immersed-boundary-method.html)

 sfar May 30, 2007 11:05

Immersed boundary method

Hi,

I intend to carry my research on Immersed Boundary Method. I have already read lots of articles on IB method. Now what I need actually is a piece of code illustrating implementation of the IB method. Do anyone know any source from where i can get some initial idea about IB coding? I will prefer finite volume discretisation of the governing equations. Thanks in advance.

Sfar

 jinwon May 30, 2007 15:19

Re: Immersed boundary method

http://www4.ncsu.edu/~zhilin/

Above is the webpage of author writing the book "immersed interface method". He released the book containing the concept of interface and moving boundary treatment.

I also concern about the ibm to treat the material interface in the compressible flow. Current approach is to apply the analytical solution for a mixed cell coexisting two-phase fluids. I am not sure it will be a proper way. If you have had such problems before, please consult me.

 sfar May 31, 2007 05:43

Re: Immersed boundary method

Thanks jinwon. Actually I was looking for a piece of code to start with. I need some illustration. It could be either 2D or 3D case.

 phil May 31, 2007 08:29

Re: Immersed boundary method

Hi Sfar,

One question : -what sort of IBM do you want to use ? Peskin method, penalization method, virtual method ...

You want a piece of code ... but what can of alogorithm to you want to use ?

regards,

 rt May 31, 2007 11:07

Re: Immersed boundary method

IB is new (but hot) research area, so few opensource are realesed based on my knowledge, u can try the following links:

http://www.cs.nyu.edu/~smyau/berkele...s/ti_ibgs.html

http://gfs.sourceforge.net/

http://seesar.lbl.gov/ANAG/chombo/

 phil May 31, 2007 11:48

Re: Immersed boundary method

IBM is new research ?

First Paper : 1972 C.S. Peskin JCP 10 : 252-271

IBM is an eternal new research area since 30 years !

sincerly

 rt May 31, 2007 13:42

Re: Immersed boundary method

a few old papers don't make a topic as old topic !, you can compare volume of publication after 1995 to old ones and then judge,

also do you knowe some books related to this (old !!!) concept which are publishad before 2006

 opaque May 31, 2007 15:18

Re: Immersed boundary method

Dear rt,

Without getting defensive, the concepts behind the immersed boundary method has been around for decades.. The SOLA-VOF type methods and its predecessors have modeled solids within fluids using a similar approach long before 1972.. You can check the history of CFD from Los Alamos National Lab, and see what they have done since the 1950's..You can also search on how the Flow3d commercial CFD code from Flow Science uses the same idea.

Calling it Immersed Boundary Method is just labeling (formalizing) an old, tested and proven idea.. Others called mesh superposition, general moving object, overlapping meshes, etc..

It has its advantages as well as its pitfalls..

Opaque..

 Junseok Kim May 31, 2007 22:40

simple code is here

http://math.konkuk.ac.kr/kwon/board....ter2007&page=1

 rt June 1, 2007 01:31

Re: Immersed boundary method

Opaque,

u r right there are a little old work but as i said previously presence of some old work is not reason for calling a method as "old method" (this method find considrable attraction and progress very recently),

sola vof is not immerseb bc () !!! but nasavof (2d and 3d) is and flow3d is extension of solavof to 3d with adding paratial cell support, in fact your mentioned works are only related to the one research group in Los Alamos and also that work is related to a very simple IB method. It is worth to notice that Los Alamos team did'nt continue this method and put it away, u can see Telluride project which is based on boddy fitted FVM.

especially recently several versions of immersed bc are presented, while a little portion is older that 1995, u can compare it with FVM, FEM , if you have background in IB and FVM or FEM u can see that FVM and FEM find sufficient maturity and their recent progress is small while this is not right for IB methods.

also theoriticaly IB are very attractive and efficient while practical application is very limited and little commercial packages uses IB (i know only Flow3D, althouth IB in flow3D is very simple and limmited only for handdling solid boundaries).

 phil June 1, 2007 03:24

Re: Immersed boundary method

rt : - Firstly IBM is just a name for a concept. Like Opaque explain to you, adjust a new force in NS equation is not a very 'new' concept! I invit you to read annual review of Ballaras and Piomelli (2004) about it. - Secondly since 1972 Peskin work to improve this method. It's certainly the 'father' of IBM. He want to mimic a porous and moving boundary!

It's true when you say 'theoriticaly IB are very attractive and efficient while practical application is very limited and little commercial packages uses IB '. But commercial application is not scientific work !

Sfar : you can read the Mohd-Yusof paper on the CTR internet WEB : http://www.stanford.edu/group/ctr/ResBriefs/ARB97.html

It's good paper for a 'how to'.

Regards,

 rt June 1, 2007 06:21

Re: Immersed boundary method

phil,

>I invit you to read annual review of Ballaras and Piomelli (2004)

could u please send to me (rohtav@gmail.com) an e-copy of this paper (what is full address)

>But commercial application is not scientific work !

aim of CFD is help to solve real problems, so if a method be interesting, naturally move into commersical packages (especially old methods !!! ), don't forget that commercial CFD codes has R & D section and has some scientist in their developing team and also are sponser of some scientific projects.

finally it is hard for me to accept that IB are old because if i accept this i should conclude that IBs are not as efficient and accurate as claimed in literature because if they are "old" and efficient they should absorb attraction and sholud be included in classical books, commercial packages and find more popularity.

 phil June 1, 2007 09:19

Re: Immersed boundary method

rt : I'm sorry but i wrote a great mistake. The paper is : Annual Review of Fluid Mechanics Vol. 37: 239-261 (Volume publication date January 2005) (doi:10.1146/annurev.fluid.37.061903.175743)

'IMMERSED BOUNDARY METHODS' by Rajat Mittal, Gianluca Iaccarino

Piomelli and Balaras 's paper subject is around the wall layer Model for LES ... Sorry for the mistake. Unfortunately, i don't try the first paper and the second don't help us.

For me IBM is "efficient and accurate" research area and old. There are lot of problem when you want to use it. So it's for me a 'eternal' new research area. Finaly i have a question : What do you think about LES or URANS ? is it new ?

BR

 rt June 1, 2007 10:29

Re: Immersed boundary method

>What do you think about LES or URANS ? is it new ?

sorry, i don't have sufficent background in turbulence field.

>There are lot of problem when you want to use it. So it's >for me a 'eternal' new research area.

"lot of problem" ! IB is name for a family of methods, could you count some of these problems which cover vast maturity of these methods.

 rt June 1, 2007 10:54

Re: Immersed boundary method

>The paper is : Annual Review of Fluid Mechanics Vol. 37: 239-261 (Volume publication date January 2005) (doi:10.1146/annurev.fluid.37.061903.175743)

good review, it is good for judge, look at its bibliography, about 30% of citations are older that 1998, this shows that most of works are not at least 10 years old, for a 2005 review this shows that method is not old and still dones not acheive maturity.

i know that some families of IBs are old but most of them which are under investigation and progress now are new (especially Cartesian type methods).

 sfar June 1, 2007 11:33

Re: Immersed boundary method

Thank a lot Junseok. You simply made my day! My thanks also goes to rt for providing some useful links. I decide to stick to Peskin's IB method . After gaining some control over the method , I'll think about some new ideas relating to IBM. I am interested in the finite volume discretisation of the governing equations.

Good day to all!

Sfar

 opaque June 1, 2007 17:26

Re: Immersed boundary method

Dear rt,

Welcome to the world of paper publishing... How many papers solving the Navier Stokes equations refer to the basic works of fluid mechanics? Most recent papers refer the equations to known or easily accessible books.. That does not make the Navier Stokes equations any newer.. What about the finite element? not new by any means and still a lot of research on it

Do not worry about the age of a method, but to how useful it is for particular types of applications, and how it can be improved.

Please do not call something new, just because it has been resurrected by some who found it useful again..

Cartesian methods, new? Tell that to the Flow3d developers. Actively work on, maybe, but hardly new.

Have a nice weekend.

Opaque

 rt June 2, 2007 01:34

Re: Immersed boundary method

>Welcome to the world of paper publishing...

hope u r go in too ... and then continue disscusion.

>Cartesian methods, new? Tell that to the Flow3d developers. Actively work on, maybe, but hardly new.

flow3d use only one type of these methods which is not undergone considerable improvment, but there are some new works on that method, e.g. see works due to P. Colella and co-workers, Berger & Aftosmis works,...

but Cartesian and new, probably u r not familiar with its state-of-the-art since most of recent work on IB are related to Cartesian types, do you see works of Udaykumar et. al. or fedkiw et. al. and ... if u have look in JCP on this subject see various kind of IBs which are presented very recently with new idea.

> Flow3D, ... Flow3D,...

Why Flow3D?, do you know any other code which is more reliable, i don't include flow3D in scientific community, in my opinion its developing team is (small and) weak (exclude Prof. Hirt from them), and its progrees is slow its developer don't have considerable contribution in CFD community recently (but only have some application oriented works), it only cover various applications but its computational kernel is weak, e.g. for multi-scale applications it is has serious problem.

instead, you can rely on Cart3D code which is an IB code with strong developping team and based on state-of-the-art method and is efficient in practical applications (phil: see that IB don't have essentially difficulty in practice and this is not reason for limiting generality of IBs !)

Finally thanks from u and phil to contributing in this discussion, such discussion is very interesting for me (and probably for readership of this forum, i feel that CFD-online forum converted into communication forum not discussion forum).

>Have a nice weekend. and you.

 phil June 2, 2007 02:11

Re: Immersed boundary method

rt: yes it's an excellent paper about this method, and in the introduction the authors say' Since 1972 ...'. Joke :)

I know that IBM are "not very" difficult to implement, I use and improved it during my PhD. So probably, you can believe me if I say that I have nothing agains't.

Since the beginning, lot of papers try to improve the order of this method, to suppress spurious oscillations, ... But ,for me, nothing new in the concept since several years. Maybe Mohd-Yusof (1997) is the last author to propose a new IBM definition.

Good week and BR.

 rt June 2, 2007 02:42

Re: Immersed boundary method

>I use and improved it during my PhD

good, so u r master in this field, i like to have more diss. about these methods with an expert (there are some doubt about some claims in this field for me that u may have idea about them)

i use IB too. i use method based on recent work of collela, or aftosmis (Cartesian FVM with vof-like treatement of solid buondaries coupled with octree method).

- but what kind of IB u used (Cartesian or ...)?

- what kind of interface is under your study (static solid boundary, or interface of two-phase flow, or moving solid boundary, ... )

what is basis of your method (which paper)?

but some qs.:

recently some Cart. IBs use vof-like basen on FVM and some Cart. IBs use FDM and are interpolation based.

what kind is useful for modeling static solid boundary from your experience?

- among interpolation based methods, most of works use 1D-interpolation (1D- taylor seris) to find value of field or its gradient on off-grid points, and then extend it to higher dimension, e.g. work of Udaykumar and Fedkiw, and 2nd convergence slope is claimed, i have some doubt about claimed accuracy because extension to higher dim. needs naturally multi-dimensional taylor expansion, what is your idea?

have them 2nd order accuracy near solid boundary (not globally)?

in fact, i think that if we use cartesian grid with stair boundary and refine grid global 2nd order accuracy is acheived naturally and i guess that they acheive these accuracy,

in clear: we have accuracy porportional with A * O(h^2) "A" is larger for stair boundary grid and smaller for boddy-fitted grid, and probably those works (IB) get "A" in the order of stair boundary grid not boddy fitted, what is your opinion?

(althouth some multi-dimensional talor based methods are presented too)

All times are GMT -4. The time now is 13:35.