New CFD user trying to learn the mesh/convergence ropes
 

I have only been using ANSYS CFX for less than one month and have really been impressed in the ease of the interface. The one thing I am not happy with (and it should be no surprise to me either) is the intricacies of the modeling techniques and calculation method selection. Much more intense than I ever thought. I am a very seasoned structural FEA user, so I understand the importance of good modelling to achieve good results. Just because I understand it doesn't mean I can't complain about it! :)

My Problem:
I am running a transient flow past an obstruction (circular cylinders with teeth protruding from the side). The fluid is water. The flow is 1m/s and my reynolds number is high enough for steady vortex shedding. I have used surface inflation to take my mesh size down to a Y+ value of less than 2 at the boundary layer. I have reduced my mesh size in the wake. Picture of mesh is attached.

When I run an adaptive timestep transient analysis, I can't get my coefficient loops to drop down to 2-5 until my timestep is VERY small (<0.000025 secs). Am I chasing my tail by making my mesh really small then my timesteps need to be really small too?

I have wind tunnel test data showing a Cd of this obstruction to be approximately 1.6-1.7. My results come up with 3.0+. This is only one piece of a larger puzzle. I need to validate the cylinder/tooth before I can move forward.

For such a simple model I have close to 700,000 elements. Each coefficent loop takes 20secs to complete, which by my calculations tells me I can expect a solution sometime next week... :( How on earth am I going to run the full model, not just this tiny little piece?

Any advice on mesh size, modeling strategy, solver, etc. would be greatly appreciated.

Thanks,
Shannon



http://i491.photobucket.com/albums/r...ord_Page_1.jpg



http://i491.photobucket.com/albums/r...ord_Page_3.jpg

http://i491.photobucket.com/albums/r...ord_Page_4.jpg


http://i491.photobucket.com/albums/r...ord_Page_2.jpg