# On grid independent solution for pulsatile flow

 Register Blogs Members List Search Today's Posts Mark Forums Read

 June 7, 2005, 14:01 On grid independent solution for pulsatile flow #1 David Guest   Posts: n/a Hi all, I am modelling a 2d pulsatile flow in a conical tube for an undergraduate project. This tube is about 15 cm long with a radius varying between 6.5 mm and 12.5 mm. I use a velocity inlet and a pressure outlet. I am trying to acheive an grid independent solution. To do so, I compare a grid having 200 000 cells vs. one having 800 000 cells. I have set-up monitors for various quantities including the area-weighted velocity magnitude of points lying on the centerline (symmetry axis) of the tube. When comparing the 2 cases I get satisfying results. My problem is when monitoring the integral of wall shear stress on the wall of the tube. I get vary different values (30% error) when taking the cycle time-average. My question is does this mean that my grid is not fine enough? Can someone give an order of magnitude as to how many cells I should be using to obtain a grid-independent solution? From what I had read I tought 200 000 cells would be more than enough for a laminar case... Any comments would be much appreciated. Thanking you in advance for your time. David

 June 8, 2005, 09:54 Re: On grid independent solution for pulsatile flo #3 David Guest   Posts: n/a Hi Aaron, Thank you so much for your comments. They really give me a good explanation. One other question: At some point you mention :"Don't make any changes to the grids (other than grid size)[...]". Does that mean that grid adaption should be avoided. In my case I used it because some of the cells near the wall did not have a good aspect ratio near the wall (both for the 200k and 800k grids). Is this a potential source of more trouble that need to be? Best Regards, David

 June 8, 2005, 12:50 Re: On grid independent solution for pulsatile flo #4 Aaron Guest   Posts: n/a David, Refining the grid through the adaption step in FLUENT is a very good method for locally optimizing cell sizes to achieve realistic solution values that are *expected* based on the solver, models, and BCs that have been applied. However, if there is a change made to one of the refined grids but not to the others, *will* you be able to ascertain whether any change in solution was due to the refinement or was due to the adaption? The answer is probably not because it may be difficult to determine where the change in solution came from. Thus, it would be best to conduct the grid analysis only grids that have the same kind of refinement. If you believe that grid adaption is necessary, then do that on the grid for which you have the most confidence in the solution. Then export that grid from FLUENT back into GAMBIT so you can see what it looks like and also have the ability to make changes to it. Using the adapted grid as a template, refine according to your chosen refinement factor being sure to maintain a grid that has a similar adaption appearance as the originally adapted grid. This may require some creative GAMBIT work, but it should be possible if you define edges to be the length of the refined area and then force the mesh to scale appropriately along that length and into the heart of the computional domain, i.e. toward the center. This can be done by applying the appropriate "size function" to the edge and meshing inward. Maintaining the cells as quadrilaterals may be difficult, but I would urge you to try. If you must go to tri-shaped cells, then make sure all of your grids are tri-shaped. Or if you can keep them as quads, again do that for all grids. One way to apply different sizes to different regions is to subdivide the computational domain into multiple faces. Use edge meshing to obtain the desired structure of the mesh, then mesh the interiors afterward. Hope this helped. Aaron

 June 8, 2005, 16:20 Re: On grid independent solution for pulsatile flo #5 David Guest   Posts: n/a Many thanks Aaron, Your comments are very much appreciated! Best Regards, David

 Thread Tools Display Modes Linear Mode

 Posting Rules You may not post new threads You may not post replies You may not post attachments You may not edit your posts BB code is On Smilies are On [IMG] code is On HTML code is OffTrackbacks are On Pingbacks are On Refbacks are On Forum Rules

 Similar Threads Thread Thread Starter Forum Replies Last Post holg FLUENT 0 July 13, 2009 17:10 CFX Begineer CFX 0 October 27, 2002 11:54 Art Stretton Phoenics 5 April 2, 2002 05:59 Seb Main CFD Forum 13 May 22, 2001 13:37 Hesham M. Aly FLUENT 2 October 5, 2000 08:24

All times are GMT -4. The time now is 23:49.