Hi, I am setting up the simulation of a scroll compressor alike to the one in this picture:

http://www.affordair.com.au/images/whatisaircon/scroll%20compressor.JPG

I had already defined all the geometry and the important meshing into Gambit. Thinking in processing it by smooth and remesh, I made a continuous mesh by permitting a tiny gap in between the two spirals.

Since the movement of the upper part is orbital, there is a crank rotating on the frame-base spiral part, and the upper spiral is actuated by the crank. FIRST QUESTION: Is it possible to define the crank (as a solid continuous type defined as a BC Wall for instance) making it move the UpperSpiral?

SECOND QUESTION: can this be done by an UDF? Where can I find the clues to build the right one up?

Since this movement is decomposed into two angular rotations, by using the reference system on the BaseSpiral Omega_Crank = - Omega_UpperSpiral. I thought about the possibility of erasing the crank and using it as reference system, by giving the relative rotational speeds (same Omega on different axis) to every Spiral part. THIRD QUESTION: has anybody ever done something like this? Is it possible to use a multiple reference frame model? Any recommendation?

Any clues are really welcome.

Regards, Pedro

I dont see that any sort of reference frame model will be applicable due to the orbital motion.

Looks like you have to use DEFINE_CG_MOTION i.e. directly resolve the motion and fluid domain deformation.

Set up a 2D case first. You cannot have wall-wall contact so you need to leave a small enough clearance. This implies careful meshing and careful use of the dynamesh settings e.g. layering, smoothing and remeshing.

For your later 3D model, the 2.5D dynamesh settings might be useful.

Expect convergence difficulties. The DP solver might be required.

We solved a similar problem last year. Movement of the top spiral was set by UDF. The greatest problem which you will face is remeshing of grids.

Thanks a lot for your help, It is really useful.

I was trying to avoid programming UDF's because as I never have done it before I am kind of afraid to start doing it. 8)

I know about C++ programming (low-intermediate level) and I have the FLUENT user defined functions guide (fl62udf 592pages PDF). Do you think the information given there is enough to build up the function I need? Are there other publications for UDF learning?

Could you please send to me or post any sample of the UDF to have a reference in order to build up the one I need? (I am searching the forum for any samples anyway.)

Thanks again.

It's hard to find an UDF example because I've changed my work.

The information given in UDF manual is enough. You must correctly describe the top spiral movement and left a small gap between two spirals. It is important for remeshing process.

Dear Pedro,

I am also working similar to you and stuck up with the motion. Please let me know how are you approaching it? please share the UDF if you are able to develop the UDF and I will share if I am suceesful.

Hello, Today I tried to build the UDF up. This is what I did:

************************************************** *****

#include "udf.h"

DEFINE_CG_MOTION(up_scroll, dt, vel, omega, time, dtime) {

Message("\n entering movement of the scroll")

vel[0]=-34.90658556*0.0126051*sin(34.90658556*time);

vel[1]=34.90658556*0.0126051*Cos(34.90658556*time);

Message("\n EXITING THE UDF");

}

************************************************** ******

Since I got to rebuild my model I have not tested it yet. This is my first UDF 8)so do not espect miracles... I hope the orientation of the CG not to be changed since there are no omega rotational speed in use, and it is just a circular translation of the CG point.

Some data to understand it: 2000 rpm == 34.9065.... rad/s crank excentricity == 0.0126051 m

commentaries, corrections and advices are nicely welcome.

thanks again

Next is literally copied and pasted the file for the movement, tested and working.

************************************************** ****

#include "udf.h" DEFINE_CG_MOTION(upscroll, dt, vel, omega, time, dtime) { vel[0]=-34.90658556*0.0126051*sin(34.90658556*time); vel[1]=34.90658556*0.0126051*cos(34.90658556*time); }

************************************************** *****

I found many errors when compiling it because I did not know I should start Fluent from the "MS Visual C++ Studio Command Prompt"; Thank you very much SIVA. In case someone needs to do it,just have to type "fluent" on the command prompt.

Later on, when I reach the right model I will post to let you know how did I set it up.