Introduction

Model flow of water and light nonaqueous phase liquid (LNAPL), and aqueous phase transport of up to five species in ground water with multiple pumping and/or injection wells with MARS. MARS is a finite-element model that allows accurate representation of highly irregular material and physical boundaries in a heterogeneous and anisotropic media.

MARS 2-D/3-D Flow

• Initial conditions and free oil volume are estimated internally from the monitoring well fluid level data.
• Rectangular 2-D prism or isoparametric quadrilateral elements to accurately model irregular domain and material boundaries, hydraulic, and physical boundaries.
• Oil and water recovery rates vs. time are computed.
• Areal distribution of residual hydrocarbon is computed and used to estimate transient contaminant loading to ground water if transport simulations are performed.
• Interactive finite-element mesh generator: rectangular/isoparametric quadrilateral mesh for areal simulations.
• Spatially-variable water recharge, injection or LNAPL leakage.
• Model multiple pumping and/or injection wells and trenches.
• Model specified head and flux boundary conditions.
• Simulates fractured media or granular porous media based on the dual porosity approach.

Included with MARS is the finite-element model, BIOF&T 3-D, that allows:

• Transient 2-D or 3-D multicomponent aqueous phase transport in groundwater aquifers. This feature enables computationally efficient simulations with a model that gives due regard to the dimensionality of the problem and is hydrogeologically defensible.
• Temporal and spatial variations in the source (i.e., residual dense or light nonaqueous phase liquids), and, given the initial conditions, changes in loading to ground water are computed and updated internally for aqueous phase transport.
• Model specified concentration, mass flux and source/sink boundary conditions.
• Convection, dispersion, diffusion, adsorption, desorption, and microbial processes based on oxygen-limited, first-order or Monod-type biodegradation kinetics as well as anaerobic sequential degradation involving multiple daughter products. This allows real-world modeling not accomplished in similar biodegradation packages.
• Computationally-efficient matrix solution by conjugate gradient method with preconditioning.

MARS 2-D/3-D Input

• Mesh discretization data
• Initial conditions for flow: water and oil pressure distribution
• Boundary conditions for flow: specified head boundaries, flux boundaries, and sources and sinks
• Soil hydraulic properties: van Genuchten parameters, hydraulic conductivity distribution, and porosity
• Initial conditions for transport
• Species concentration
• Boundary conditions for transport: specified concentration, specified mass flux, and spatial distribution of contaminant loading
• Dispersivities
• Mass transfer rate coefficient between oil and water phase
• Distribution coefficient
• Bulk density
• Diffusion coefficient for species
• Biodegradation parameters for each species
• Fraction of the mobile phase (needed for fractured media simulations only)
• Mass transfer coefficient between mobile and immobile phase (needed for fractured media simulations only)

MARS 2-D/3-D Output

Flow

• Spatial distribution of fluid pressure with time
• Spatial distribution of fluid saturation with time
• Fluid velocity distribution with time
• Fluid pumping/injection rates and volume vs. time

Transport

For each species:

• Spatial distribution of concentration with time
• Mass dissolved in water vs. time
• Mass remaining in NAPL phase vs. time
• Mass adsorbed on the solid phase vs. time

MARS 3-D with Transport

• MARS 3-D comes with BIOF&T 3-D and all features included in BIOF&T 3-D.

MARS 2-D/3-D Technical Information

MARS (Multiphase Areal Remediation Simulator) can be used to model recovery and migration of light nonaqueous phase liquids in unconfined heterogeneous, anisotropic aquifers. MARS writes input flow files for the BIOF&T model which simulates multispecies dissolved phase transport in heterogeneous, anisotropic, fractured media, or unfractured granular porous media.

Ground-water contamination from hydrocarbon spills/leaks is a serious environmental problem. Nonaqueous phase liquids (NAPL) are immiscible fluids that have insignificant solubility in water. NAPLs in the subsurface migrate under the influence of capillary, gravity, and buoyancy forces as a separate phase. Light NAPLs (LNAPLs) float and migrate on top of the water table posing a continuous source of contamination to the ground water. Due to water table fluctuations, some of the NAPL gets trapped in the unsaturated and saturated zones. NAPL trapped in the soil and ground water acts as a continuing source of ground-water contamination resulting in expensive restoration of these aquifers.

MARS consists of:

1. The MARS flow module simulates recovery and migration of water and LNAPL in unconfined aquifers following an LNAPL spill or leakage at a facility. It can also simulate NAPL recovery with skimmers and trenches, and optimize the number, location, and recovery rates for water and oil.

2. MARS writes input files for BIOF&T, a transport model that simulates decoupled 2-D or 3-D multispecies aqueous phase transport from the free and residual NAPLs.

The MARS flow module invokes an assumption of near-equilibrium conditions in the vertical direction. This reduces the nonlinearity in the constitutive model and transforms a 3-D problem into a 2-D areal problem, thereby drastically reducing computational time for the simulation.

MARS gives the initial distribution of NAPL specific volume in the domain for BIOF&T which models the aqueous phase transport, and computes and updates the temporal and spatial variation in the source during the simulation.

This software is accompanied by a user-friendly pre-processor, mesh editor and post-processor. The pre-processor and mesh editor can be used to create input data files for MARS. They include modules for: mesh generation; allocating heterogeneous and anisotropic soil properties to zones; defining fixed head, flux, source/sink boundary conditions for water and oil phases; and allocating spatially-variable recharge in the domain. Two-dimensional rectangular or isoparametric quadrilateral elements are permissible to accurately model irregular domain and material boundaries.

Required input for flow analyses consists of initial Zaw, Zao distribution, soil hydraulic properties, fluid properties, time integration parameters, boundary conditions and mesh parameters. The van Genuchten constitutive model, along with fluid scaling parameters, is used to compute water and oil phase volumes.

MARS output includes a list of the input parameters, initial and boundary conditions, and the mesh connectivity. It also includes simulated water and oil phase pressures, water and oil phase velocities at each node, total volume of water and oil versus time, and water and oil recovery/injection rates for each sink/source location versus time. Volume of free oil and residual oil and their spatial distributions are also printed versus time. Flow simulations can be performed in stages. MARS creates an auxiliary file at the end of the current stage that can be used to define initial conditions for the next stage.

MARS 2-D/3-D Input Parameters

Estimation of Soil Properties

Soil properties needed for a MARS flow simulation are: saturated hydraulic conductivity in principal flow directions, anisotropy angle of the main principal flow direction in the areal plane with the x-direction of the model domain, soil porosity, irreducible water saturation, and van Genuchten retention parameters. SOILPARA, 1995, a proprietary computer model, provides an easy-to-use tool for estimating soil hydraulic parameters from soil texture based on: 1) the public domain model RETC developed by M. Th. van Genuchten et al., 1991, 2) the work of Shirazi and Boersma, 1984 and Campbell, 1985, and 3) a selection of USDA-recommended typical parameter values for various texture classes available in the SOILPARA database are included in the MARS document.

Fluid Properties

Fluid properties required by MARS are specific gravity, oil to water dynamic viscosity ratio, and fluid scaling parameters. Methods to estimate these parameters are included in the MARS document.

Creating Input Data Files

The sequence of the input parameters and their definitions has been furnished in Appendix D of the MARS document. This section explains the procedure for spatial discretization and mesh generation, defining initial conditions, boundary conditions, and the maximum permissible array dimensions.

Spatial Discretization and Mesh Generation

The MARS modules allow use of rectangular and isoparametric elements. The element size and shape can be changed to obtain mesh refinement that is necessary to obtain accurate results.

Initial Conditions for Flow

Initial head distribution in the domain for water and oil can be specified by:

• bilinear interpolation with heads defined on the left and right boundaries
• a non-uniform head distribution defined by fluid levels in the monitoring wells

Boundary Conditions

Specified pressure head (type-1) boundary conditions can be defined at selected nodes versus time.

Type-2 (specified flux) and source/sink boundary conditions can be defined by specifying the volumetric rate [L3 T -1] versus time for respective nodes. For a type-2 boundary condition, when flux [L T -1] is known at a node, the user should multiply flux with the area represented by the node in a plane perpendicular to the flux.

MARS 2-D/3-D Windows Interface

What is the MARS pre-processor?

The Windows pre-processor for MARS is designed to help users create and edit input files for the MARS numerical model. The pre-processor works in concert with the mesh editor to allow users to assign boundary condition schedules, soil types, recharge zones, etc. to the finite-element mesh used in the MARS numerical model. The pre-processor contains all control parameters that determine model run options, initial conditions, monitoring well information, fluid properties, boundary schedule data and soil type definitions as well as serving as a binder for mesh editor files. The pre-processor also contains a module for writing input files for the MARS numerical model and for actually running the numerical model.

Using the MARS pre-processor

The MARS pre-processor runs under Windows 3.X, Windows 95 and Windows NT. The pre-processor uses the familiar tabbed notebook interface to allow quick editing of input files. The main program has two sets of tabs: one along the bottom which separates major sections of the interface, and, on some of the large notebook pages, tabs along the top that separate subsections to make the most use of available screen space. For example, clicking on the bottom tab "Boundary Schedules" takes the user to the boundary schedule notebook, a tabbed notebook for editing type 1 and type 2 boundary condition schedules.

The Tools selection on the main menu opens a tabbed notebook which includes Cue Cards, files used in the pre-processor, a numerical model runner, and a place to determine the location of the mesh editor. The files listed in the pre-processor are used to store variables and retrieve data and are generated automatically by the pre-processor and the mesh editor.

MARS data files

All the MARS variables for the numerical file are stored in ASCII text files that resemble Windows .ini files. These files are read and written by the MARS pre-processor and mesh editor. They can be interchanged in the pre-processor setup window. For example, a material property file used in an earlier project can be assigned to a new project and all of those soils will be available in the new project. A mesh file and all its associated files can be imported in the same manner. The data files can be edited with any ASCII text editor, although this is not necessary. This open architecture was designed for future expansions of all these numerical models or for third party development of graphical interface tools.

What is the Mesh Editor?

The mesh editor allows designing irregular quadrilateral meshes in two and three dimensions. Working with a numerical model pre-processor, the mesh editor provides a graphical interface for assigning properties to a mesh like initial concentrations of contaminants, soil properties, boundary conditions, etc.

Speed Buttons

Pan

Rotate

Editing

Moving Nodes

Nodes can be moved by holding down the Ctrl key (control) and the left mouse button and moving the cursor on the screen. Nodes move according to the dimension displayed on the screen so that two-dimensional meshes should be in the default X-Y view for node movement. Nodes can only be moved when the mesh editor is in its "editing" state.

DXF Import

Version 1.1 of the mesh editor introduced DXF import. This tool allows for .dxf files to be placed on a mesh. This way, site files in CAD programs can be exported to the mesh editor and then used to aid mesh refinement and adjustment.

Post-processor

The post-processor is a data parsing tool, graphing package and contour export tool for these numerical models. The post-processor is designed to be a user-friendly tool for quickly discerning model results. Users of these models can also review model text output files for a more detailed view of model results.

MARS 2-D/3-D Verification

Example: NAPL leak in an unconfined aquifer

This example is to test the accuracy of MARS to simulate a LNAPL leak into an unconfined aquifer. The domain is 40 m x 40 m, discretized with uniformly-spaced 21 rows and 21 columns (x = 2 m, y = 2 m). A leak occurred at the center of the domain (x = 20 m, y = 20 m) at a rate of 1 m 3/day for 20 days. Simulation was performed with an initial time step of 0.001 days, and the time incremental factor and maximum time step size were 1.03 and 0.2 days, respectively. The initial uniform piezometric head (Pao = Paw ) was 100 m through out the domain. All boundaries were no-flow boundaries for water and oil phases. The water head at the outer boundary was fixed at 100 m throughout the simulation.

The simulated specific oil volume distribution is shown in the following figure along with the corresponding results from MOFAT. There are some differences in these solutions due to the difference in the formulations of the MARS and MOFAT model. MARS is a vertically-integrated areal model while MOFAT is a 2-D vertical slice (Planar or radially-symmetric vertical section) model. Nevertheless, it can be seen in the following figure that both solutions agree reasonably well over the range of the simulation.

Comparison of MARS with MOFAT

MARS Requirements

(1,500 nodes) Windows 95/98/2000/NT and 16 MB RAM.

MARS Technical Information - Equations (pdf file)

Requires the Adobe Acrobat Reader to view pdf file.