MODFLOW-USG gives you the flexibility to focus grid resolution where it counts, using a variety of grid geometries. As mentioned in an earlier blog post, Quad-Based grids are an attractive option due to their simplicity and rectangular-shaped geometry (using shapes similar to those used in finite difference grids). This blog article provides a preview of the Quad-Based grids that are currently being developed in Visual MODFLOW Flex along with a few examples for various applications.
Generating Quad-Based Grids in Visual MODFLOW Flex
When it comes to generating an Unstructured grid there is no “one size fits all” solution. Each model is unique and requires various adjustments to the grid in order to achieve an accurate, efficient, and robust numerical model. Depending on the type of model you are developing, you may need to refine your model grid to accommodate one or more of the following:
• Boundary features (surface water features, drains, wells, etc.)
• Local model(s) inside a regional-scale model (well permitting or contaminant transport)
• Geological features (fault plane or a surface)
• Geological pockets/lenses or riverbed infiltration (vertical sub-layering)
• Other features: karst channels/conduits, underground mine workings, 3D ore bodies, etc.
For this reason, VMOD Flex offers full flexibility when creating Quad-Based Grids. Similar to the Voronoi Grid Generator, the process starts with your conceptual model objects: the boundary conditions from the conceptual model are automatically included in the Quad-Based grid generator in the form of points, polylines, and polygons. Each of these objects can have its own level of refinement; for example constant heads and general heads around the model edges can be refined by a factor of 2, rivers can be refined by a factor of 4, and wells screens can be refined by a factor of 8. Additional point, polygon, or polylines shapes may be added to the grid generator to further customize the grid to your needs. In VMOD Flex, a single Quad-Based grid can consist of any combination of Quadtree/Octree and Nested grids, following a layer-dependent approach to discretization. An example of the Quad-Based grid generator is shown below.
Figure 1: Quad-Based Grid Generator in Visual MODFLOW Flex
Quad-Based grids are generated in an iterative approach. You can start with simple refinement options, generate a MODFLOW-USG numerical model, and run and evaluate the results. This grid can be adjusted to incorporate additional refinement or smoothing options. Alternatively, you can create multiple Quad-Based grids within the VMOD Flex environment. As you progressively build the Quad-Based grid, the 3D Viewer provides a real-time preview of the grid geometry allowing you to make adjustments to the smoothing and refinement “on-the-fly”. Once the grid is built, it can be displayed in various views (plan, cross-section, and 3D), and correlated to the conceptual model and raw data.
Sample Applications of Quad-Based Grids
The following examples highlight various applications of Quad-Based Grids.
This example is based on the Biscayne Aquifer model as discussed in the MODFLOW-USG Users’ Manual. The starting grid size was 97 rows x 56 columns corresponding to approximately 800m uniform cell dimensions (in a single layer). Quad-Based grids were generated from the shapes in the conceptual model (canals represented as polylines and extraction wells represented as points). Grid cells were refined by a level of 3 around the canals and by a level of 4 around the wells, as shown below.
Figure 2: Biscayne Aquifer: shapefiles of canals and wells (shown on left)
and resulting “smoothed” quadtree grid (shown on the right)
The resulting grid is similar to the one illustrated in Figure 37 of the MODFLOW-USG Users’ Manual. Note that in the example shown above the refinement levels and the locations of the wells and canals do not exactly match the data set used in the MODFLOW-USG Users’ Manual.
Dewatering and Regional Water Balance for Mines
When doing a regional water balance or dewatering model for an open-pit mine site, it is often necessary to refine the grid around the pit area in order to leverage high-resolution block model data and also to have focused grid resolution around the dewatering wells. In this example, you can see nested grids with smoothing around the pit region, and refinement around the rivers/constant head boundaries along the model edges. A sample of this grid is shown below.
Figure 3: Nested grid around dewatering wells at a mine site,
with coarser refinement around linear boundary conditions
Improving the Accuracy and Efficiency of MODFLOW-2000/2005 Models
This example illustrates a simple way to update and improve an existing MODFLOW-2000/2005 model. This may be desirable in a number of scenarios: the model is not converging, is suffering from long run times, or you need to incorporate new data into the model, and achieve higher accuracy around areas of interest. A regional-scale MODFLOW-2000 model used a variably spaced-grid around the well field.
Figure 4: Quad-Based Grid Generator in Visual MODFLOW Flex
The model was imported into VMOD Flex and a smoothed nested grid was defined in place of the variably-spaced grid, with twice the level of grid resolution around the well field (as shown in the figure below). Stay tuned to our blog for more details on this process along with some benchmark examples.
Figure 5: Nested grid (with smoothing) in place of a variably-spaced grid
The grid flexibility that comes with MODFLOW-USG allows you to focus grid resolution “where it counts”, both horizontally and vertically. Visual MODFLOW Flex will allow you to fully leverage unstructured grids through the use of Quad-Based grids. The Quad-Based grid generator will be available later in 2014, giving you the capability to build and tailor your grid to meet your specific needs.
MODFLOW–USG Version 1: An Unstructured Grid Version of MODFLOW for Simulating Groundwater Flow and Tightly Coupled Processes Using a Control Volume Finite-Difference Formulation. USGS Techniques and Methods 6–A45 http://water.usgs.gov/ogw/mfusg/