Grid Method for Project: Geotrans CAU Site Refinement



 

The method of Design is as follows:

 

2D Template

Create a 300 meter spaced quad grid, refine the corrective action unit  (CAU) site areas and rotate 5 degrees. The  point distribution used to represent the Bechtel Geologic model consists of a rectangle with the corners NW (542909.12, 4132671.15),  NE (563530.36, 4130867.03), SE (561856.97, 4111740.08), SW (541235.73, 4113544.2) with 300 meter spacing. The entire set of points is a rectangle translated 5 degrees off the x-axis.
There are two areas of refinement; Scotch is represented with the x,y polygon with corners at (554833., 4124400.),  (556626., 4124243.),
(554990., 4126193.), (556783., 4126036.). Serena has corners at (548714., 4126742.), (550507., 4126585.), (548897., 4128834.), (550690.,
4128677.). Each  CAU  site is refined so that each has elements of edge size 75 meters, surrounded by a transitional buffer with elements of size 150. The remaining elements have edge size of 300 meters.
/scratch/rigel3/tam/nts_geotrans_9904/points_cau.in

This image shows a close up of the two CAU sites, Serena on the left and Scotch on the right. Spacing for the grid is 300 meters, 150 meters around the sites and 75 meters within the sites.

Resample Surfaces

The input surface files for the Bechtel model are created using the CAU triangulated template, a ray-shooting technique, and contoured surface
files from NTS. To create each new surface, the CAU template is positioned above a HSU surface and rays are projected through each x,y point on
to the HSU surface to find the elevation for each point. This is done with each HSU surface to create the x,y,z coordinate points for each
surface. The points are then connected into triangular elements, creating a Triangular Irregular Network (TIN) for each surface of the model.
Each surface is represented by a TIN sheet of 5275 x,y,z coordinates and the connectivity for 10282 triangles.
/scratch/rigel3/tam/nts_geotrans_9904/mksurf.in
 

This image illustrates the ray-shooting technique. The CAU point distribution (black points) is used to shoot rays through a HSU surface, surface braq is shown here colored by elevation. A ray is shot through each CAU ray point, on to the braq surface to find the braq elevation coinciding with the CAU ray point. The x,y,z points are then connected into a TIN sheet

The image below shows the new surface representing braq colored with site and buffer values.

Stack-to-Tet

The three-dimensional model is developed by effectively stacking all the contoured surfaces and populating the volume between the layers with CAU attributes. The TIN sheets are stacked from lowest elevation to highest elevation. The volumes between the surfaces are converted to prism elements (6 nodes, 2 triangle faces, 4 quadrilateral faces) with vertical connections between adjacent layers. Each prism is converted to three tetrahedra so that the final representation is in the form of a 3D tetrahedral mesh.
In preparation of a computational grid, thick units are subdivided vertically to provide a more gradual transition to the thinner layers, and to keep the horizontal edges in proportion to the vertical edges. The interfaces are buffered by a lower and upper surface of a distance of 15 meters to capture the unit interfaces with voronoi cells after the computational grid is created. A minimum unit thickness of 14 meters is chosen so that large aspect ratio tetrahedral and triangular elements are avoided. The grid is checked to ensure there are no holes, and that the geometry correctly represents the Bechtel Geologic model.
/scratch/fourier1/tam/geotrans_trilayers_refv3/tet.in

This image shows the tetrahedral mesh. A region has been cut out to better show the unit volume elements. The refinement within Serena site can be seen.

Connect Delaunay

Once a tetrahedral mesh is formed, the points are copied to a clean CMO. Duplicate points are dudded with filter/1,0,/.01 The points are then connected into a Delaunay mesh. Numerical solution techniques for flow and transport calculations with finite volume and integrated finite difference methods place geometric  constraints on the quality of a mesh. To optimize the mesh reconnection algorithms enhance the quality while preserving the geometry. Reconnection can be done without adding points by allowing connections to flip. In the final mesh, points may be added or removed as needed to create a Delaunay grid with positive coupling coefficients.
/scratch/fourier1/tam/geotrans_trilayers_refv3/connect.in

Final FEHMN input files

The Fehm flow and transport code uses finite volume control volumes for solution of flow and transport equations. Part of the grid generation process is to calculate the Voronoi control volume associated with each node in the grid and the area of the polygonal faces of the Voronoi control volumes. In addition to control volumes, lists are created containing the surface area of each node on the surface of the grid, for use in scaling constant flux boundary conditions. Node sets for each material property are also written, including outside, top and bottom locations for each node.
The final Delaunay computational grid for three-dimensional calculations contains 138680 nodes and 839289 tetrahedral elements and accurately represents the geometry of the Bechtal Model.
/scratch/fourier1/tam/geotrans_trilayers_refv3/fehm.in
 

Recolor with Fault - Unit relationship

To find the elements that are located in fault regions, a new routine developed by Bogdan A Pathak was used. This routine provides the capability to find the intersection of a grid containing one-dimensional and two-dimensional elements with a grid containing two-dimensional and tetrahedral elements and note which elements (if any) intersected. By using the fence diagram grid and stratigraphic grid as input to intersectelements, a grid that has fault information, as well as the original stratigraphy, is produced as output.
Once the elements that are within any of the three faults are found, the material colors are rewritten. The material units not in a fault region are kept with the values 1 to 23. For each unit, if the points fall within a fault region, the color is the unit value + 50. So that a fault in material unit 2 has a value of 52.
/scratch/fourier1/tam/geotrans_trilayers_refv3_flt/fault_zone.in
/scratch/fourier1/tam/geotrans_trilayers_refv3_flt/ck_toppts.in