CFD Online Discussion Forums

CFD Online Discussion Forums (
-   Main CFD Forum (
-   -   Is Mesh Accessary for Analysis (

Selina Tracy March 22, 2001 04:30

Is Mesh Accessary for Analysis
Dear Scholars,

My question may not make any sense to you cause it depends on geometry and flow. But I would like to know if there are any papers that indicate the difference in terms of the accuracy of results; hexa vs. tet meshes, if the number of nodes are the same.


John C. Chien March 22, 2001 13:58

Re: Is Mesh Accessary for Analysis
(1). The solution we are looking for in CFD is normally called the "mesh independent solution". (2). In other words, the solution obtained remains the same regardless of the meshes used. (3). So, you are free to use any kind of mesh arrangement or types, as long as you have the mesh independent solution. (4). For this reason, CFD is not for everybody. It is expensive, time consuming, and requires expert knowledge of the problem being solved. (5). If someone is placing the time limit on the CFD problem solving, he is likely going to get the wrong answer. And the rest of that is history. In other words, if someone is asking you for the CFD solution as soon as possible, he is saying that all he needs is a solution and he does not care whether it is right or wrong. In this case, you would put in fewer mesh points or cells, and cut down the number of iterations, and argue that the mesh used has the best quality, scheme selected has the higher order accuracy, and perhaps the algorithm is the advanced coupled method. (6). Anyway, if you change the mesh size and the solution also changes, then you don't have the mesh independent solution yet. The consequence of that can be nothing or another multi-million dollar failure.

clifford bradford March 24, 2001 23:23

Re: Is Mesh Accessary for Analysis
You have to realise that it is dependent on the CFD scheme. some schemes are specially developed for arbtrary meshes while some require structured mesh also it's usually difficult to meet the requirement of number of nodes being the same for the same geometry between tets ahnd bricks. you may wirsh to peruse the Journal of Computational Physics though. you can also try any of the other big CFD type journals you'll have to look hard though.

All times are GMT -4. The time now is 23:34.