[ivan_CFD]

Hi everyone,


I would like to request help with an issue when using edgeGrading in blockMesh. I attach some pictures to illustrate it.


I have a mesh consisting of two blocks, one of them is rectangular and the other one is trapezoidal (see blocks.jpg). There is only one cell in the z direction. The rectangular block has a uniform spacing in the vertical direction (y), whereas the trapezoidal block has a grading along the vertical direction.


To avoid jumps in cell size, I want the vertical cell size at each side of the interface between the two blocks to be the same. In other words, I want the first cell size of the trapezoidal block to be equal to the last cell size of the rectangular block (in the y direction). To achieve this, I use edgeGrading on the trapezoidal block, computing the respective expansion ratio that is needed at the left (inlet) and right (outlet) vertical edges of the block.



The problem is the following: at the left and right edges of the trapezoidal block, where the grading is imposed in the blockMeshDict (also in attachment), the grading is working exactly as expected and the cell size is matching (see mesh_inlet.jpg and mesh_outlet.jpg). On the other hand, in the vertical mesh divisions inside the block this is not being respected. This implies that for instance, the first horizontal grid line of the trapezoidal block is not straight but has a small curvature, causing a jump in cell size at the interface of the two blocks (see mesh_middle.jpg). This jump is maximum at the middle of the block and decreases towards the left and right vertical edges, where the grading is imposed. The longer is the domain, the more this problem accentuates.



I would expect blockMesh to join the distribution of points at the left and right vertical edges with straight lines, but for some reason this is not the case. It is not an optical artifact as can be confirmed by looking at the y coordinate of the points.



I have not been able to find another thread related to this problem. Only the following one https://www.cfd-online.com/Forums/op...s-geomety.html in which a similar problem is reported.



I have tried blockMesh with OF versions 3 to 6 with the same result.


Many thanks in advance.


Ivan