Opening Remarks
 

1:05 - 1:25 p.m.

From Realizations to Relevance: Ensemble Refinement and Composite Plume Mapping for Contaminant Fate and Transport Modeling

Corey Wallace, RSI EnTech, LLC

Groundwater flow and uranium transport at this former processing site are shaped by complex hydrogeologic and human influences, including past milling operations, variable recharge, and extensive gravel pit mining. The site covers a large area, but monitoring is limited in both space and time. This results in a poorly defined contaminant plume, making ensemble modeling essential for future site characterization. To address the challenges of simulating contaminant fate and transport under uncertainty, we developed an ensemble-based groundwater model using MODFLOW-USG and USG-Transport, calibrated with PEST++-IES. This approach uses multiple realizations to represent a wide range of plausible system behaviors. However, spatial variability in residual error can lead to divergent predictions, even when individual models are statistically “well calibrated”. To improve consistency and relevance, we introduce the Locations of Truth methodology, a spatially-focused ensemble refinement technique. By identifying observation points that exert strong influence on system behavior, we retain only those realizations that meet accuracy targets at these key locations. This ensures that selected models used in the predictive ensemble reflect observed conditions where they matter most to flow and transport parameter calibration. We also introduce the Parameters of Truth filter, which removes realizations with parameter values that conflict with the site conceptual model, further enhancing physical credibility. To summarize contaminant behavior across the refined ensemble, we developed the Super Plume methodology. This process uses normalized concentration fields to create ensemble mean and variance maps, providing a clear picture of expected plume behavior and associated uncertainty. Metrics such as total contaminant mass, footprint area, centroid movement, and volume-weighted concentration are computed for each realization and the ensemble mean. These innovations support stronger decision making by reducing predictive bias, improving validation, and identifying where data are most valuable.

1:25 - 1:45 p.m.

Applying Iterative Ensemble Smoother Modeling and Surface Geophysics for Dewatering Design at a Uranium Mill Tailings Disposal Site

Peter Schillig, RSI EnTech, LLC

The Rifle, Colorado, Disposal Site is a 71-acre engineered disposal cell that was constructed in 1996 to encapsulate uranium mill tailings from the nearby Rifle Processing sites. Since 2012, pore fluids within the tailings have been accumulating at a rate exceeding the removal capability of the existing leachate collection system. Pore fluid levels were approaching the top elevation of the HDPE liner located at the toe of the cell, necessitating the design and implementation of additional dewatering capability.

As a precursor to the design of a dewatering strategy, a stochastic water balance model was constructed to provide a prior distribution of estimated rates for sources and sinks of water within the disposal cell. These estimates were used to assign boundary conditions and groundwater flux calibration targets as part of the configuration of an ensemble of 223 numerical transient groundwater flow models. Data from two dewatering standpipes represent the only available data for hydraulic head calibration targets following completion of the disposal cell. The groundwater flow model ensemble was created using MODFLOW-USG and simultaneously calibrated with an iterative ensemble smoother (IES) implemented in PEST++. The posterior parameter distributions for the 223-member calibrated ensemble were retained and used to probabilistically forecast the potential effects from each dewatering strategy. Feasible drilling locations identified through analysis of electrical resistivity tomography and magnetometer data were applied to the forecast design simulations.

Significant data limitations for configuring and calibrating a numerical groundwater flow model illustrates the importance of uncertainty in the calibrated parameters when forecasting outcomes. The relatively recent advancement of IES modeling allowed for each remedial design to be evaluated with multiple calibrated groundwater flow models as opposed to a single, deterministic model. With IES, the uncertainty in the calibration parameters were carried forward and estimated during the evaluation of the predictive dewatering designs. 

Coauthors: Ronald Kent (Drummond Carpenter); Katie Mclain and Al Laase (RSI EnTech, LLC)

1:45 - 2:05 p.m.

Advancing In Situ Mining Simulations: A JupyterLab Framework for Reactive Transport Modeling and Environmental Impact Analysis

Glenn Hammond, Pacific Northwest National Laboratory

Reactive transport modeling serves as an essential tool for simulating the in situ leaching of critical minerals and the subsequent restoration processes. This presentation introduces an open-source JupyterLab framework designed to facilitate PFLOTRAN reactive transport simulations of in situ mining operations, alongside evaluating their environmental impacts. The framework empowers users to incorporate well locations and pumping rates throughout the mining history and experiment with varying well placements, pumping strategies, and lixiviant formulations. This tool aims to enhance users' understanding of subsurface reactive flow and transport dynamics associated with in situ mining practices.

Coauthor: Katherine Muller (PNNL)

2:05 - 2:25 p.m.

Numerical Modeling of Aquifer/Well Hydraulics and Transport for Data Analysis and Targeted Extraction

Miguel Valencia, Pacific Northwest National Laboratory

The representation of wells in numerical models of groundwater flow and contaminant transport rarely accounts for well construction (e.g., filter pack, well diameter, bentonite seals, etc.), head losses, and vertical flow and transport within the well itself. Simplified representations of wells do not allow for simulation of complex aquifer/well interaction including vertical flow in the filter pack, movement of solute up or down the well under ambient flow conditions, or mixing that occurs when sampling in long-screened wells (LSWs). Here, we present and demonstrate an approach that leverages numerical tools based on MODFLOW 6 and its new unstructured grid and solute-transport capabilities, the FloPy Python package for MODFLOW model setup, and mathematical frameworks from our recent work on analytical models. We implement an efficient, asymmetric model for flow and transport and account for head losses associated with vertical flow in the well. This approach will allow us to (1) investigate the  implications of borehole flow on samples collected in LSWs, (2) evaluate the effectiveness of different sampling techniques (e.g., low-flow, snap sampling, etc.) given realistic well construction and aquifer/well hydraulics, (3) assess approaches for retrofitting LSWs for discrete-zone isolation (e.g., packers or well liners), and (4) design new wells for targeted remediation of specific aquifer layers. In future work, we are integrating PEST++ for rigorous parameter estimation and model calibration.

Coauthors: Rebecka Iveson and Rob D. Mackley (PNNL); Christian Langevin (SSP&A); Frederick D. Day-Lewis (PNNL)

Open Discussion
 

BREAK
 

3:15 - 3:35 p.m.

Development of a Numerical Transport Model to Support Groundwater Remedy Selection at a Former Uranium Mill Site 

Keaton Belli, Geosyntec Consultants

Removal of the tailings impoundment at the United States Department of Energy Moab Uranium Mill Tailings Remedial Action (UMTRA) Project is expected to be completed in 2029, at which time a final remedy to address uranium and ammonium in site groundwater will be implemented. Under Title 1 of the Uranium Mill Tailings Radiation Control Act, natural flushing (i.e., monitored natural attenuation) may be selected as the groundwater remedy if it can be demonstrated using a numerical fate and transport model that the remedy will achieve groundwater standards within 100 years. To support selection of a groundwater remedy at the site, a numerical fate and transport model was developed using historical and recent characterization data, including column leaching tests, a transducer study, and several field efforts lead by the Network of National Laboratories for Environmental Management and Stewardship. The site is considered complex due to multiple source areas, vertically stratified groundwater due to brine formation at depth, discharge to the Colorado River, endangered species habitats in river backchannels, and ongoing implementation of multiple interim measures to achieve protection of human health and environment, including groundwater extraction and freshwater injection. The model simulated variable density flow and reactive transport of uranium and ammonium and was calibrated to a 15-year period with monthly timesteps for stresses, including site precipitation, evapotranspiration measured from remote sensing satellites, extraction and injection, and Colorado River stage. The calibration dataset included water levels, specific conductivity, uranium and ammonium concentrations, contaminant mass extracted, and plume evolution. Predictive simulations were used to estimate the time to achieve groundwater standards using passive and active remediation technologies. Geosyntec will discuss the key features of the model, how uranium geochemistry was incorporated into the transport model, and present results of predictive modeling scenarios to support identification of a remedy for site groundwater.

Coauthors: Chelsea Bokman and Jennifer Nyman (Geosyntec Consultants); Liz Moran (U.S. Department of Energy, Office of Environmental Management); Ken Pill (North Wind Portage)

3:35 - 3:55 p.m.

A Geostatistical Assessment of the Extent of Radioisotope Contamination and Potential Transport Under Building 324 in the Hanford 300 Area

Moses Obiri, Pacific Northwest National Laboratory

In October 1986, approximately 510 liters of radioactive liquid waste containing cesium-137 (¹³⁷Cs) and strontium-90 (⁹⁰Sr) spilled within the B-Cell of Building 324 at the U.S. Department of Energy’s Hanford Site. Some of this waste entered the subsurface through a floor sump, posing long-term contamination risks. Characterizing the three-dimensional extent of this contamination is critical for safe and effective remediation, particularly given the complex transport dynamics in the deep, heterogeneous vadose zone.

This study presents a geostatistical framework to estimate zones with high radiation levels (exceeding 1 R/hr) beneath Building 324. Using data from over 2,200 measurements across 45 boreholes, we modeled spatial autocorrelation and anisotropy through variogram analysis and moment of inertia techniques. These inputs fed an indicator kriging model, producing a 3D probability map of threshold exceedance. Confidence intervals were included to quantify uncertainty, identifying zones where contamination exceeds the threshold with at least 5% probability at 95% confidence.

To evaluate potential contaminant migration, 3D flow, and transport simulations were conducted using 50 sequential Gaussian simulation realizations and the expected-value (E-type) plume distributions under steady-state conditions. A linear sorption model was applied, assuming relatively mobile solutes. Simulations encompassed both vadose zone and aquifer transport, assessing the ability of the existing monitoring well network to detect contaminants.

Results indicate that under current conditions, the strongly sorbing nature of ¹³⁷Cs and ⁹⁰Sr limits their mobility, suggesting minimal transport through the vadose zone or into the aquifer. However, if more mobile analogs were present, the model suggests that several wells in the existing groundwater monitoring network could detect contaminants migrating from the original spill. These insights help guide future remediation and monitoring strategies at the Hanford Site.

Coauthors: Deb Fagan, Ben Jensen, Frederick Day-Lewis, Bryan He, and Mark Rockhold (PNNL)

3:55 - 4:15 p.m.

Applying State-of-the-Practice Tools to Determine the Presence and Transport of PFAS at a Former Air Force Base

Ryan Samuels, AECOM

Background/Objectives. Since the late 1940s, the US military has released property once used to support war efforts. Since the late 1980s, over 350 installations have been closed under a formal US Department of Defense program (BRAC). As a requirement of returning property to the public, there are requirements to investigate, manage, and mitigate environmental issues that resulted from military use. The former Reese Air Force Base in Lubbock, TX, was founded in 1941 and closed under BRAC in 1997, and legacy chlorinated solvent groundwater contamination was cleaned up by the military based on hydrogeological understanding and Texas state environmental requirements. With the discovery of per- and polyfluoroaklyl substances (PFAS) releases at this former airfield, the need arose to assess these emerging contaminants and an Affected Property Assessment (APA) was undertaken.

The objective of the APA investigation was to define the nature and extent of PFAS in soil, sediment, surface water, and groundwater at the former Reese AFB.  The project posed numerous challenges including 10+ release areas, highly heterogeneous geologic conditions, seasonal/irregular agricultural pumping, complex surface water/groundwater interactions, exceedances in private wells 2.5 miles (4 kilometers) downgradient of the former Base extent, and redistribution of PFAS via the historic Injection/Extraction well network.

Approach/Activities. To accomplish the APA, AECOM developed a first-of-its-kind 3-D numerical PFAS groundwater fate & transport (GW F&T) model.  The GW F&T model integrated PRISM® - a state-of-the-art correlation technique that constructs a predictive, 3-D geologic framework and more accurately defines subsurface heterogeneity – to provide a high-resolution, geologically-defensible basis for estimating measured hydraulic conductivity data and constraining groundwater flow estimates.

Stratigraphic information was transferred into the numerical model by means of 2-dimensional (2-D) grids. The model was vertically discretized into eight layers, each a constant 10 feet (ft) thick.  For each 10-ft horizontal slice, the stratigraphic team mapped the stratigraphic facies. These units were then identified by material identifications (IDs) that carried all the hydraulic parameters for that material. Each 2-D grid was then inserted into the proper elevation slot until the entire thickness of the aquifer was built.

Groundwater flow at former Reese AFB and the surrounding area was simulated using the United States Geological Survey’s (USGS) MODular Groundwater FLOW modeling code MODFLOW 2005 (Harbaugh, A.W., 2005) to simulate the physical flow system.

The calibrated flow model was then integrated with transport parameters derived from the APA investigation to simulate the solute transport of PFAS in groundwater at the former AFB.  The incorporation of PFAS fate and transport mechanisms in subsurface environments was based on recent research (Adamson and others, 2022; Newell et al., 2021a and 2021b). Fate and transport of PFAS was simulated using the modular three-dimensional transport model MT3DMS (Zheng and Wang, 1999).

Results/Lessons Learned. The model provided unparalleled understanding of preferential contaminant migration pathways and was used to refine mass flux estimates, guide optimal placement/screen depths of 150+ monitoring wells, and streamline investigations to address data gaps. Over the next 3 years, the model will serve as an invaluable tool for evaluating remedial alternatives, designing an optimized remedy, and effectively communicating remedial decision-making to all stakeholders.  

4:15 - 4:35 p.m.

Modernizing Soil Solution Thermodynamics by Returning to Nineteenth Century Concepts

Jacob Reynolds, Central Plateau Cleanup Company

In the late nineteenth century, two theories emerged to explain non-ideality in aqueous solutions: (1) the formation of ion pairs, and (2) the strong binding of water molecules by ions, effectively making them part of the solute rather than the solvent. These concepts were later overshadowed by electrostatic theories that worked well for very dilute solutions. Most current thermodynamic codes rely on models from the 1970s that amend electrostatic equations with empirically determined and ill-defined "specific interactions," necessitating extensive experimental datasets for multicomponent solutions found in soils and groundwater.

Since the 1970s, two significant developments have reinforced the earlier theories: (1) the Law of Matching Water Affinities (LMWA), which estimates ion-pair constants, and (2) spectroscopic evidence showing that many ions bind water strongly. These advances support the nineteenth-century views on non-ideality. Zavitsas’ new model incorporates these developments by adjusting the activities of water and ions to their mole fractions after accounting for ion-pairing and water bound to electrolytes (Zavitsas, 2019, J. Phys Chem. B, 123, 869-883). The solvation number in Zavitsas’ model can be directly measured using Dielectric Relaxation Spectroscopy (DRS).

This study demonstrates that Zavitsas’ model accurately calculates the water activity of aqueous NaF solutions and the solubility of villmenite (NaF ) in multicomponent systems using model parameters directly measured via DRS and the LMWA, without fitting the model to activity or solubility data. In contrast, the Pitzer model requires four to five empirically determined parameters to force-fit the same solubility data.