A Decision Support Software Package

 

Note: This Decision Support System is undergoing a technical review within the EPA's Environmental Technology Verification (ETV) Program. Results of the review are due out at the end of Summer 1999. For more information on the ETV program, visit their web site at: http://www.epa.gov/etv/


GroundwaterFX is a very comprehensive groundwater analysis package. It is one of the most sophisticated flow and transport simulation tools on the market, providing the following features:

Utilizes flow and transport models in a probabilistic framework to account for uncertainty in contaminant movement.

Decision support tool to optimize the number and placement of monitor wells to delineate the nature and extent of a contaminant plume.

Utilizes optimization theory to minimize the number and cost of monitor wells and boreholes for sampling locations.

Simplifies the analysis of natural attenuation potential, thereby saving significant monies by avoiding costly and time-consuming pump and treat options for groundwater remediation

GroundwaterFX is an outgrowth of the DOE code, the Groundwater Analysis and Network Design Tool (GANDT), developed at Sandia National Laboratories. GroundwaterFX is a decision support tool intended to provide defensible information regarding the necessary number and location of monitor wells to delineate the nature and extent of a contaminant plume beneath a waste site. The GroundwaterFX methodology comprises several steps. First, analytical and numerical models based on the physics of flow and transport are employed in a stochastic modeling scheme (i.e., Monte Carlo simulations). Many runs are made to quantify the uncertainty in the nature and extent of contamination. Second, the computer-generated output is compared to known concentrations within the plume, and the output is conditioned to these data. Next, a probability field describing expected plume distributions is generated from the stochastic output. Finally, an optimization methodology based on Operations Research methods is employed to determine the location(s) of the next monitor well(s) or to determine whether their is adequate numbers of wells. This technology is intended to guide monitor well placement during characterization by providing near real-time or overnight feedback to the site operator.

In addition to the post-conditioning of water quality data to the transport results, GroundwaterFX has the ability to simulate spatial variability through the use of geostatistical simulations on the hydraulic conductivity variable. GroundwaterFX has process models to simulate either aqueous or vapor-phase (e.g., volatile organics) movement of contaminants through the unsaturated zone. GroundwaterFX also has an automated grid generation function for ease of use when applying the 3-D finite difference solver for the groundwater flow and transport model. The user interface is quite friendly and has real-time visualization capabilities for viewing probabilistic depictions of plumes and network design. The code has been developed for PC Windows (i.e., Windows 3.1, Windows95, and WindowsNT), and the Macintosh (i.e., 68K and PowerPC) platforms. GroundwaterFX is being applied to Sandia waste sites and to Uranium Mill Tailings Remedial Action (UMTRA) Project sites. For UMTRA, GroundwaterFX is also being used to evaluate the potential for natural attenuation. The results of these analyses, as well as the basic methodology description, are currently being documented in Sandia reports. Future work on the GroundwaterFX methodology may include: optimization of pump and treat analyses; the added capability to perform 3-D visualization; enhancing the optimization algorithms for remedial strategies; biodegradation analysis; and adding in cost and risk modules to optimize performance and endpoint indicators of remedial measures. Utilization of the GroundwaterFX methodology could save millions of dollars in unnecessary monitor well installations and groundwater monitoring activities.

Natural attenuation may be the most cost-effective remedy for a contaminated waste site, because the main emphasis is monitoring, not cleanup. Processes that could lead to natural attenuation as a potential alternative include:

Flushing (e.g., discharge of contaminated groundwater to a stream, with subsequent dilution in the stream)

Dilution (e.g., areal recharge or infiltration of water into the aquifer)

Degradation (e.g., biological, radioactive decay, or chemical process change in concentrations)

GroundwaterFX currently has the capabilities to simulate flushing, dilution, radioactive decay, and will have the added functionality to evaluate biodegradation in the near future.

In conventional flow and transport simulations a calibration procedure is manually performed to "match" observed site data (e.g., water quality data) with the model results. This process is quite tedious and time consuming. In GroundwaterFX we have built in the capability to condition, or honor, the data through use of statistical analysis techniques. The effect is essentially a built-in calibration method. This feature is extremely important when the user is interested in probabilistic (i.e., Monte Carlo) analyses, where many simulations are performed to get a statistical representation of the uncertainty in model results. The user has the ability to choose from a Chi-square test for acceptance of a run or a root mean square analysis. In order to apply a Chi-square method the residuals must be normally distributed. GroundwaterFX has the ability to display real-time graphics in order to evaluate normality, giving the user great flexibility in deciding conditioning criteria. The following graphics illustrate some of the graphical constructs in GroundwaterFX.

 

GroundwaterFX has the ability to display many types of data, as a result of probabilistic analyses. Types of 2-D visualization graphics include hydraulic conductivity distributions and contaminant distribution plots. Contaminant distributions can be visually displayed for all 5 user-specified time steps, and for each of the ten possible vertical layers in a numerical simulation. The types of contaminant plots available are:

Plume concentrations for each individual Monte Carlo simulation, with user-specified color mapping of the contaminant concentrations, and an optional player mode that cycles the transient output in an "animation" type format;

Average plume distribution from all Monte Carlo runs;

Standard deviation plot;

Variance plot;

Coefficient of variation plot;

Probability plots, whereby the user can type in or use a slider bar to specify a contaminant concentration, and the resulting plot displays the probability of exceeding that concentration.

An example of the transient output capabilities for an average plume distribution is shown in the next several graphics. This example is from a preliminary Sandia/DOE analysis of the Riverton, WY UMTRA site, intended to evaluate the potential for natural attenuation as a remedy within the next 100 year period.

The following figure represents average plume concentrations 35 years after contaminant releases began. The tailings source was removed some years later, making this a pulsed duration source term.

The following graphic represents the contaminant plume70 years after the initial release.

The following graphic represents the contaminant plume90 years after the initial release.

The preceding example shows some of the potential that the GroundwaterFX package has for analyzing the potential for a site to utilize natural attenuation as the preferred remedy on a site. GroundwaterFX is also capable of analyzing the optimal well network to demonstrate compliance of the site for natural attenuation, because monitoring becomes the key to demonstration of success.

DecisionFX plans to release the GroundwaterFX package for commercial sale at the end of 1999. Pricing will be reasonable for an environmental analysis package (probably around $1000). We hope to offer the package free-of-charge to environmental regulators, in the hope that they will evaluate GroundwaterFX for appropriateness of use. DecisionFX would appreciate an email inquiry from prospective buyers of this product to help in marketing, packaging, training, and support considerations. If you express interest, we will keep you apprised of the pending release and planned training activities. email address: info@decisionfx.com

An example of the application of the GroundwaterFX methodology and code is given in the Case Studies section of this web site.

 



Return to Software Products page


Return to Main DecisionFX Homepage