We are looking for the way to use results of a structure modeling by softs like Abaqus or Nastran in a CFD soft, like Fire or Fluent, and also the other way. We aim at studying a seal in a fluid with a fluid-structure modeling, by iterating the CFD->FEM->CFD calculations.

The problem is to get the pressure from the CFD to the FEM and make calculate the FEM soft, then get the displacement and make the CFD calculate with it.

There seems to be just a few studies done on that subject. Any help would be welcomed. Even just an idea of who made such study or the name of someone able to answer our questions.

Have a look at

http://www.cd.co.uk/news/html/nlett.htm

Coupled Fluid/Structure Interactions - STAR-CD links to stress analysis codes.

Intersting problem. Since the problem seems to be strongly coupled, you may want to treat it as such. Or you may choose the approach you have suggested and look for a way to couple these solutions externally.

CFD Research Corporation (http://www.cfdrc.com) provides you with solutions for both the above approaches.

CFD-FEMSTRESS and CFD-ACE+ provide users with a fully coupled way of handling fluid-strucutural interaction problems. CFD-ACE+ is the advanced multi-purpose CFD code from CFD Research Corporation and CFD-FEMSTRESS is the FEM software that works with it (or as stand-alone for that matter). The combination has been used by our customers and by us to study problems such flow through mechanical heart valves, flow through valves in compressors, inflation of airbags.

For the other altenative, CFD Research Corporation has developed MDICE (Multi-DIsiplinary Computing Environment). This is a way let independent codes (CFD, FEM or other) to communicate.

If you (or someone else) need more information on CFD-FEMSTRESS or MDICE, please contact me directly at rmm@cfdrc.com. You are also welcome to visit CFD Reserch website at http://www.cfdrc.com, and can also contact us from there.

I worked under problem - aeroelasticity of turbomachines where are needs for calculations of unsteady forces from flow to cascdes. I did some codes for calculations of unsteady transonic flows in cascades (flows and pressures) then result was applied for structural analysis. By way CFD method was FEM too.

You can see article on the English www.univer.omsk.su/~rtf/AI11.TEX

Also I should say - of course FEM codes is beatiful thing but more important for application is correct mathematical approach for real( your ) problem.

There have been LOTS of studies on this very subject. Try a literature search for CFD/Aeroelasticity to find relevant fluid-structure coupling methods. We have a code here that was developed in house called STARS which seemlessy couples the FEM structures to a CFD model for an aeroelastic analysis. You should be able to find some relevant technical papers by Gupta, K.K. (primary developer) and also Arena, A.S. on our particular methodology to solve the coupled problem.

I anticipate that that you will have some difficulty in coupling two separate off-the-shelf commercial packages unless you have source code to make modifications and interface between the two. You should probably look into getting one package that has both modules and solves the coupled problem already (see previous posts for recommendations). Otherwise, you will be stuck running a single iteration with each separate code and writing some sort of batch code to pipe the outputs back and forth. This will be VERY slow. Ideally you would like to only initialize the problem once and do all of the iterations internally.

Recently Prof. F. Grosjean and his student, S. Sauvage, of ENSIETA in France have validated an iterative approach to modeling fluid/structure interaction using Abaqus and FLUENT. The study was presented at the 1998 Abaqus Users' Conference and a summary was published in Fall/Winter 1998 Fluent Newsletter. You can view this summary on-line at http://www.fluent.com/news/fall98/abaqus.htm.

Fluent will be introducing a coupled fluid/structure interaction capability in its FIDAP CFD software in May 1999. A simultaneous solution for the fluid flow field, stress/deflection in the solid and shape/position of the domain will be computed. The advantage obviously is the elimination of the data transfer between codes.

Eric Grald, Fluent Inc.

(1). The problem of a seal is different from that of a vibrating airfoil. The deformation of the seal and the change in the gap size could be important factors in determining the structure side and the fluid side calculations. (2). Unless it is a very thin seal, the seal will reach a static shape and the flow around and through the gap will also reach a steady-state condition. (3). So, I think, from practical point of view, a couple of iterations should be enough to derive the final shape of the seal. (4). On the other hand, a vibrating airfoil may require direct coupling formulation between the structure and the fluid because of the transient nature of the problem.

I've been working on this problem for a squeeze flow application where the pressure field interacts and deforms the bounding surface. I've written a hybrid program in Matlab that mixes a finite volume technique to determine the pressure field and a spectral method to calculate the deformation in the bounding surface. It's basically an iterative process around equilibrium conditions in the coupled set of equations. I myself feel the need to move over to a commercial package, especially as the problem becomes more complex using viscoelastic fluid mediums. I'd be interested in hearing any success you have. later, Phil

Have a look at

www.pallas.de/cispar

There you find a description of the COCOLIB library which allows coupling between CFD and FEA codes.

COCOLIB provides with the organisation between the codes - including the data transfert - but not the needed datas' transformation from one mesh's nodes to the other one. I mean, in the case of Finite Volumes coupled to Finite Elements method, no node correspond. This seems to be actually the main problem in my case. Rémi