A NEW AND EFFICIENT SPATIAL SUB-DISCRETIZATON METHODOLOGY FOR CONCURRENT MULTI-SCALE GROUNDWATER MODELING

Multi-scale simulations in groundwater often necessitate the use of successive telescopic discretizations, each designed for a given spatial scale. The use of telescopic grids or meshes resulting from such a discretization procedure requires that successive simulations in different scales be carried out, starting with the one covering the largest spatial scale. Using this procedure, boundary conditions for successively smaller-scale models must be derived from larger-scale models. This type of simulation is computationally demanding, labor-intensive, and time-consuming as it is necessary to perform several sequential simulations, to transfer information from larger-scale models to smaller-scale models, to inspect results from all the simulations, and to provide quality control for all the models.

To overcome these problems, a new methodology for sub-discretization has been developed. This methodology is relatively flexible, allowing a given three-dimensional hexahedral block or element to be sub-discretized with infinite combinations of number of subdivisions in three dimensions. Each block can be successively sub-discretized to achieve the desired level of local resolution, thereby allowing several models with different scales to be combined into a single model. With the proposed methodology, for finite-element simulators, no new element types with varying combination of nodes are required. Because of its flexibility, the methodology is applicable to both static and dynamically adaptive grids.

The newly developed methodology of sub-discretization will be presented along with application examples, results, and discussions.