A method is described for a posteriori optimization of computational grids where grid non-orthogonality and grid overlap are minimized using an iterative optimization procedure. Desirable properties such as smoothness are at the same time maximized using the described technique. In this sense an optimal computational grid can be obtained, irrespective of the method used to generate the original grid. Thus, the user may generate a computational grid with whatever method he chooses. Then, if the grid is unacceptable for computations, it can be optimized in an a posteriori fashion. If the original grid has regions of overlap (non-positive Jacobian), the described method is capable of ‘unravelling’ the grid and making it useful for computations. Grid points on the boundary of the domain remain fixed, thereby allowing the correct resolution of regions of interest. The iterative optimization procedure is extremely fast due to the use of exact line searching and a conjugate direction method. The method can be applied to the optimization of two-or three-dimensional grids. Example grid optimization cases are shown for two-dimensional grids. The formulation of the method for three-dimensional grids is also given.
