CFD残差不收敛的问题(国外英语资料).docVIP

CFD残差不收敛的问题(国外英语资料) CFD: residual non convergence problem Source: soup Benche log Always convergent CFD program, can not be a good thing. Most of the flow is unsteady. Constant is an approximation. Usually we do constant calculations. When there is a great deal of instability in the fluid, the reasonable result should be that the constant calculation can not converge. Those software that always converges, such as NUMECA, must have a man-made, non physical object to stabilize the iteration. In this case, nine times out of ten, the result of convergence is very different from the actual situation. If someone with such a result to do the project, that is to get into trouble. Id like to read the article about Denton in this years IGTI meeting. It summed up a lot of experience in how to use the CFD. To tell you the truth, I dont dare to say anything about these problems, but Ive met each other. Sometimes there are hardware problems besides software. I used to compute the same problem with a single machine and a parallel machine. The single computer showed some of the same things that the landlord saw, or even no convergence, but there was no problem with the parallel machine! You dont have many grids, and the graphics arent very complicated. Try using a structured mesh to see if there is a problem with the boundary conditions. Using this software requires a lot of experience and is the same as experiment, especially at the beginning. When the continuity does not converge, do a few aspects of the inspection (of course, some specific problems, but also specific analysis): 1, grid problem: in the relatively complex geometric models tend to block painting grid, when drawing grid block adjacent grid size difference too, this time there will be continuity equation does not converge or residual, between adjacent block grid difference not too big to control in about 1.2, at the same time grid near the wall also has some influence on the continuity, when using the boundary layer