| A two-dimensional telescopic grid refinement (TGR) technique was evaluated for verification through a variety of synthetic-data model sets. The TGR technique employed bilinear interpolation of head from a coarse-grid model to provide boundary condition values for a finer-grid sub-domain model. The technique was employed on a simple model, on a series of models with perturbations of a single element of the simple model, and on a complex model involving all of the perturbations to the simple model. Analysis of residuals between the coarse-grid model results, finer-grid sub-domain model results, and results from a finer-grid full-domain model allow for evaluation of the benefits of the procedure and identification of case-particular potential sources of error due to the use of a coarse grid in the sub-domain and use of local boundary values derived from the coarse-grid model. Significant improvement to the solution due to the use of TGR is limited to the node of the well cell. Stress to the local boundaries generates no substantial error with the use of a time-dependent local boundary specification. Potential sources of error include the use of a coarse grid for a non-linear flow equation, irregular and steep gradients, and a heterogeneous conductivity field. A simple analysis of calculated drawdown, accuracy, improvement to the solution, and computational demand versus grid spacing indicates that the benefits of TGR diminish when computational time and memory are of concern. |
