MODFLOWT is an enhanced version of the USGS MODFLOW model which includes packages to simulate advective-dispersive contaminant transport. Fully three-dimensional, MODFLOWT simulates transport of one or more miscible species subject to adsorption and decay through advection and dispersion. MODFLOWT performs groundwater simulations utilizing transient transport with steady-state flow, transient flow, or successive periods of steady-state flow. Groundwater flow data sets created for the original MODFLOW program function without alteration in MODFLOWT; thus extension of modeling projects to simulate contaminant transport is much easier using MODFLOWT than any other commercially-available model. MODFLOWT is thoroughly tested and has been bench-marked against other transport codes including MT3D, SWIFT and FTWORK. A comprehensive and pragmatic approach to contaminant transport has been incorporated in MODFLOWT which allows for three distinct directional dispersivity values, multiple chemicals and a rigorous treatment of the hydrodynamic dispersion tensor.
MODFLOWT KEY FEATURES
MODFLOWT uses an implicit finite-difference discretization scheme for the numerical solution of the partial-differential equations for solute transport. Both central-differencing and upstream-weighting options are available for solution of the advective terms. In addition, fully implicit (backward-in-time) or Crank-Nicholson (central-in-time) approximation options for the time derivative are included in MODFLOWT. Solvers for flow include Strongly Implicit Procedure (SIP), Preconditioned Conjugate Gradient (PCG2), Slice Successive Overrelaxation (SSOR), and the new ORTHOMIN solver (OMN). For transport, the available solvers are Slice Successive Overrelaxation and ORTHOMIN.
Data set preparation for MODFLOWT is simple using Groundwater Vistas. The Windows-based graphical interface features full support for creation of all packages to simulate both groundwater flow and contaminant transport using MODFLOWT. Because it is compatible with ModelCad from Geraghty & Miller and can import MODFLOW data sets, conversion of existing modeling projects to simulate contaminant transport using MODFLOWT is straightforward. In addition, other popular preprocessors like ModelGIS, Visual MODFLOW, and ModelCad for Windows have the ability to create MODFLOWT data sets.
MODFLOWT allows for full expansion of the cross-product terms of the dispersion tensor or "lumping" of these terms on the main diagonal. Full expansion of these terms is appropriate in scenarios where the principal direction of groundwater flow is not parallel with the grid. In these cases, lumping of the terms may overestimate transverse dispersion and cause "fattening" of the plume. Full expansion of these terms in MODFLOWT minimizes this effect. In addition, MODFLOWT can use either backward (implicit) or central (implicit-explicit) differencing techniques of the time derivative. This allows for larger time steps (and shorter execution times) without sacrificing numerical stability or accuracy.
How do execution times of MODFLOWT compare with other transport models?
One of the principal advantages of MODFLOWT is its execution speed compared to other popular transport models. In benchmark tests against MT3D96, MODFLOWT was shown to range from two to 20 times faster depending on the nature of the problem. This is particularly true for simulations with complex geology comprised of variable-layer elevation and thicknesses. The new ORTHOMIN solver in MODFLOWT offers significant advantages over the traditional Slice Successive Overrelaxation (SSOR) solver without sacrificing numerical stability or accuracy.
What are cross-products and why are they important?
In simulations where the principal direction of groundwater flow is not parallel with the finite-difference grid, the dispersion tensor will have non-zero off-diagonal components. Traditionally, these components are approximated in the numerical solution by "lumping" them along the main diagonal (Dxx, Dyy, Dzz) of the dispersion tensor. MODFLOWT allows for both "lumping" and full inclusion of the cross products during solution of the dispersion term in the finite-difference equation. For some scenarios, "lumping" the cross products can greatly overestimate transverse dispersion, unrealistically "fattening" the contaminant plume. In the example, full inclusion of the cross products has minimized this effect.
How is time-stepping implemented in MODFLOWT?
Each flow stress period is subdivided into solute transport time steps. In MODFLOWT, the time step length can be limited such that the maximum local Courant number criteria is satisfied. However, since MODFLOWT can use the backward (fully implicit) or central (implicit-explicit) differencing scheme for the numerical evaluation of the time derivative, larger time steps can be used without sacrificing numerical stability or accuracy. This allows for much shorter execution times. In the example, MODFLOWT matched observed data closely while utilizing a much larger transport time step than that used in the MT3D simulations.