IMPORTANCE AND SIMULATION OF HORIZONTAL ANISOTROPY IN GROUND-WATER SYSTEMS

Geologic media are often deposited and deformed in ways that result in hydraulic conductivity being enhanced and (or) diminished in some directions. For example, common horizontal layering of granular deposits can result in horizontal hydraulic conductivity being greater than vertical hydraulic conductivity, and lateral tension and compression can produce fractures and faults that result in enhanced and diminished horizontal hydraulic conductivity in selected directions. The resulting hydraulic-conductivity fields are said to be anisotropic. In such fields hydraulic conductivity needs to be represented by a full tensor to obtain accurate flows for all hydraulic gradient directions. Using a diagonal instead of a full tensor can result in substantial errors in simulated flow, both in terms of the amount and direction of flow. The magnitude of the error depends on the angle between the hydraulic gradient and the principle components of anisotropy. Capabilities for simulating full hydraulic-conductivity tensors in the structured-grid finite-difference models often used to evaluate ground-water systems have been limited. The work presented here describes how a method originally developed in the petroleum literature for determining the hydraulic-conductivity tensor terms needed for variable-direction horizontal anisotropy in finite-difference grids was adapted for use in the U.S. Geological Survey three-dimensional finite-difference ground-water model MODFLOW. The new capability allows anisotropy within model layers – that is, horizontal anisotropy – to be oriented in directions other than the directions of the model rows and columns. Unlike some alternatives, the method presented conserves mass even when the material is highly heterogeneous. Attributes of the method are discussed, including solver issues that were addressed using operator splitting and improved accuracy of simulated flows.