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

 October 3, 2006, 23:04 Gradient calculation #1 Harry Guest   Posts: n/a To calculate 3-D gradient at centroid of a cell, which neighboring cells should be involved? To explain my question in clarity , I divide the neighboring cells of the considered cell into two groups. one group is the cell which has a common face with the considered cell. one group is the cell which only has a common line/point with the cosidered cell. Should I include both two group for gradiet calculation, Or the first group is enough. I found my code based on unstructured grids just considered the neigbouring cells with common face. Now I want to use gradient for interpolation of value at a arbitry point in a cell. I doubt the existing method of gradient calculation is not good enough. I would appreciate your idea. Regards, Harry

 October 7, 2006, 03:02 Re: Gradient calculation #2 Sunny Guest   Posts: n/a Hi Harry, I have used the first group for gradient calculation, that is only the neighbors with will my cell shares common faces. Using this gradient value for interpolation, I was able to obtain good results for unstructured grids. I dont think the value of the gradient should depend on values of cells with common edges or vertices, as you dont have any flux input through these cells. -Sunny

 October 7, 2006, 03:07 Re: Gradient calculation (typo) #3 Sunny Guest   Posts: n/a ....only the neighbors with WHICH my cell shares common faces....

 October 7, 2006, 04:19 Re: Gradient calculation (typo) #4 Harry Guest   Posts: n/a Thanks, Sunny. I orginally want to acquire gas properties (velocity, porosity, pressure) at an arbitary point in a cell using their gradients. I mean gradient is used for interpolation calculation, not for the solution of fluid flow. The related results are dissapoint as this method has a bad boundedness (some value is beyong physical meaning) and accuracy, especially for big-size cell. Now I completely give it up.