Sonar Surveys/Nurb-Surfaces/Adaptive, Body-Fitted Meshing for Large-Scale Modeling of Multiple Solution Mines

An adaptive meshing technique is detailed which is suitable for large scale finite element modeling of solution mines. The resulting meshes of "bricks" are fully compatible (portable), that is no hanging nodes result, requiring multi-point constraints to satisfy continuity. The procedure begins with sonar surveys and a corrective process is undertaken to make the resulting geometries suitable for meshing. Oftentimes the sonar surveys have irregularities, not to mention excessive detail which makes meshing cumbersome and time-consuming. Solution mines used for waste disposal are often incomplete since the sonar doesn't penetrate the waste. Solid modeling techniques are used to ‘plug’ holes in incomplete surveys. Non-uniform rational b-splines are subsequently fit to the surfaces of the pre-processed solution mines and a subsequent discretization is made to create a suitably-coarse description while respecting the mine geometry and volume. The voxel characteristic at the cavern wall is removed, resulting in a mesh, composed entirely of bricks, accurately representing the cavern geometry. 1. INTRODUCTION Sonar surveys of solution-mined caverns, typically situated in salt domes, once read into a visualization application, may have obvious irregularities, may be incomplete, or may have unnecessarily high resolution for meshing purposes. In order to prepare these geometries for meshing, we shall pre-process these, first by fitting a b-spline surface to each, then discretize each to a resolution that both respects the geometry and admits an efficient interrogation for mesh generation purposes. Once pre-processed a solid model is constructed using the discretized caverns. This model defines not only the modeling domain, but also includes geometries identifying locations for refining the mesh (adapting) and also parses the domain to isolate areas of particular importance for post-processing. Certain post-processing techniques may become excessively slow for interaction and interrogation purposes, if these techniques are extended throughout the modeling domain. For salt domes hosting multiple caverns, interest has increased in assessing the influence of caverns on each other, particularly when considering that cavern volumes generally increase over time. This is due largely to difficulties in maintaining a fully saturated brine in caverns that are frequently filled and emptied. This ongoing dissolution generally overcomes the natural tendency for the caverns to close due to creep. Adaptive meshing is vital when multiple caverns are considered in a single simulation exercise. This allows suitably ‘small’ elements for cavern walls while simultaneously allowing increasingly larger elements beyond.