Poster Session
Tuesday, November 14, 2023 | 5:00 - 7:00 p.m. Pacific Time
|
The General Poster Session is open to all topics relevant to the remediation of complex sites. Examples include characterization studies, contaminant fate and transport, geochemical and microbiological contaminant controls over contaminant mobility, and conceptual site modeling, among others. Session Organizers: Hilary Emerson, Pacific Northwest National Laboratory (PNNL); Katie Muller (PNNL), and Nik Qafoku (PNNL). |
|
|
Deep Learning-Based Water Seepage Monitoring in F-Area 3 Basin Cap Duani Rojas Aris, Florida International University ► View poster |
Detecting water seepage in the F-Area 3 Basin Cap traditionally involves Electrical Resistivity Tomography (ERT) and subsequent manual analysis of conductivity data for identifying anomalies. This study aims to automate this analysis process using deep learning techniques such as Autoencoders and Long Short-Term Memory (LSTM) layers. The proposed model learns the inherent patterns of normal conductivity values by mapping them to a lower-dimensional space and then back to their original dimensions. This encoding enables the model to grasp the distribution and features of the standard data effectively. When confronted with data that diverges from the norm, the model's reconstruction from lower-dimensional to original space exhibits discrepancies, as the model lacks familiarity with such deviations. The disparity between input and output across spatial points signals the presence of anomalies. This information aids in localizing anomalies spatially, as points proximate to or constituting anomalies display larger input-output differences compared to normal points. The extent of anomaly severity correlates with the extent of the input-output mismatch, indicating the anomaly's magnitude. To incorporate environmental context, LSTM layers process antecedent rainfall and temperature data concerning the ERT measurements. These LSTM outputs fine-tune the Autoencoder's lower-dimensional representation, effectively compensating for weather effects and minimizing errors between model outputs and actual normal data. The resultant system harnesses ERT data and historical weather information to autonomously identify abnormal conductivity values, pinpoint their spatial coordinates within the 3D domain, and quantify anomaly severity. By delegating the analysis to the deep learning model, this approach streamlines decision-making processes, empowering stakeholders to make informed choices based on algorithmic outcomes. Coauthors: Aris Duani Rojas (Florida International University), Timothy Johnson (Pacific Northwest National Laboratory), Himanshu Upadhyay (Florida International University), Jayesh Soni (Florida International University), Leonel Lagos (Florida International University) |
Optimisation of Cu, Au, and Ree Biorecovery by Shewanella Oneidensis Natalie Byrd, University of Manchester ► VIEW POSTER |
Cu, Au and Rare Earth Elements (REESs) are all industrially important metals that can be present in effluents from mining, waste electric and electronic equipment (WEEE) and industrial processes. Future demands for these metals may challenge supply, making green-recovery from unconventional resources (e.g. wastes) increasingly essential. Microbially mediated metal biorecovery offers a potential bioremediation solution whilst promoting the circular economy by producing high-value metal nanoparticles (NPs). The model metal-reducing bacterium, Shewanella oneidensis can bioreduce Cu(II)(aq) to Cu(0)(s) and Au(III)(aq) to Au(0)(s), and deposit intracellular Cu/Au NPs. However, mechanisms for these reactions are still not well understood. In contrast to Cu and Au, REEs such as Ce, La, and Dy are not typically redox active and their biorecovery is expected to occur via biosorption/bioaccumulation. Again, literature regarding REE biorecovery, including with S. oneidensis, is scant and further work is necessary to understand mechanisms and challenges (e.g. toxicity). In this work, we aim to gain robust mechanistic understanding of Cu, Au and REE biorecovery which is essential for their optimisation and to develop feasible and industrially scalable technologies. Here, we present various studies that explored (bio)recovery of Cu, Au and REE using S. oneidensis with full characterisation of solution geochemistry and metal NPs by IC, ICP-MS, TEM (with SAED and EELS), XPS and XRD. Using various electron donors and deletion mutants, we showed bioreduction of Cu and Au by S. oneidensis was most efficient with H2 as an electron donor and was reliant on hydrogenase enzymes. Secondly, we found biosorption capacity of REE (Ce, La, Dy, Nd) was concentration dependant and metals were primarily associated with phosphate on/in S. oneidensis cells. Insights from these studies will be used to enhance metal biorecovery/metal-NP production (e.g. by using bioengineering), and inform future scale-up studies towards real-world application. Coauthors: Jinxin Xie (University of Manchester), Christopher Egan Morriss (University of Manchester), Elliott Nunn (University of Manchester), Jennifer S. Cavet (University of Manchester), Fabio Parmeggiani (Politecnico di Milano), Richard L. Kimber (University of Manchester), Jeffery A. Gralnick (University of Minnesota), Jonathan R. Lloyd (University of Manchester) |
Impact of Cr(VI) as a Co-Contaminant on the Sorption and Desorption of U(VI) in Hanford Sediments Under Mildly Alkaline Oxic Conditions Mariah Doughman, Florida International University ► VIEW POSTER |
Uranium (U) wastes, generated during plutonium production at the United States Department of Energy Hanford Site, have been historically released to the subsurface resulting in groundwater plumes. Column and batch experiments were conducted to investigate the impact of Cr(VI) on U(VI) attenuation mechanisms in quartz, plagioclase feldspar, and carbonate dominated sediment (2 mm diameter particle size). Experiments were performed under slightly alkaline conditions in the presence of major groundwater components (Ca2+, Mg2+, Na+, K+, carbonate, chloride, and sulfate, pH: 7.930.04) with U(VI) (168 mol/L) alone and with Cr(VI) at a U:Cr molar ratio of 1:10. Preliminary column results show a slight decrease in U(VI) sorption in the presence of Cr(VI), with the U(VI) distribution coefficients (Kd) changing from 1.5 L/kg to 1.2 L/kg. Tailing on the desorption limb of column experiments was indicative of nonequilibrium sorption conditions. This was further evident from the increase in concentration of U(VI) observed after stop flow events. Ten more pore volumes were required for U(VI) to fully desorb in the presence of Cr(VI) compared to when it was present alone. Previous batch experiments under the same conditions resulted in a similar magnitude of Kd for U(VI) of 0.80.2 alone and 1.40.3 L/kg in the presence of Cr(VI). Overall, U(VI) remains mobile under these conditions with a slight retardation of sorption in the presence of Cr(VI). The mobility of U(VI) under natural site conditions should be considered while assessing the feasibility of passive remediation strategies including monitored natural attenuation (MNA) after active remediation of the Hanford Site is completed. Coauthors: Yelena Katsenovich (Florida International University), Ravi Gudavalli (Florida International University), Kevin O'Shea (Florida International University), Hilary P. Emerson (Pacific Northwest National Laboratory), James Szecsody (Pacific Northwest National Laboratory), Kenneth Carroll (New Mexico State University), Nikolla Qafoku (Pacific Northwest National Laboratory) |
Enhancing Groundwater Contamination Monitoring at the Hanford Site with Sequential Deep Neural Network Hardik Gohel, University of Houston - Victoria ► VIEW POSTER |
The U.S. Department of Energy's Office of Environmental Management is responsible for an extensive groundwater and soil remediation effort at the Hanford Site in Washington State, where decommissioned nuclear production reactors, laboratories, and chemical reprocessing plants have caused groundwater contamination. Traditional physical and statistical models have provided valuable insights into contaminant behavior at the site. However, modeling large, commingled plumes containing carcinogens like hexavalent chromium remains challenging. Recently, machine learning models, particularly artificial neural networks, have shown promise in complementing existing methods. Their ability to model sequential data, adapt to various datatypes, and fine-tune parameters makes them appealing. This study investigated a Long Short-Term Memory-based framework to predict hexavalent chromium concentrations at the Hanford Site. The dataset comprised 2912 measurements from 121 wells collected between 2000 and 2008. The model's performance was evaluated using standard metrics such as mean squared error, root mean squared error, and coefficient of determination. The optimized model achieved impressive results with a mean squared error of 921.0, root mean squared error of 30.35, and coefficient of determination score of 0.94. These findings suggest that the implementation of this neural network architecture could prove beneficial in assisting existing modeling efforts for contaminants of concern not only at the Hanford Site but also at similar facilities. Coauthors: Hilary Emerson (Pacific Northwest National Laboratory) , Danial I. Kaplan (University of Georgia) |
Toolset for Evaluating Energy Flexibility— Environmental Tradeoffs for Hydropower Hongfei Hou, Pacific Northwest National Laboratory ► VIEW POSTER |
Among the challenges of natural resource management in a period of a rapidly changing climate is creating environmental measures in Federal Energy Regulatory Commission hydropower licenses that can support the needs of the grid and environment today and into the future. Variable renewable energy (VRE) sources such as solar and wind, can facilitate decarbonization of the power sector but require support from flexible generation sources to respond to anticipated and unexpected grid needs when there are rapid changes in VRE power output. Hydropower is well-suited to provide this flexibility, but the quick generation ramp-ups and -downs needed to support VREs cause rapid water level fluctuations that can strand fish, erode shorelines, create hazardous boating conditions, etc. The Energy Flexibility Environmental Tradeoffs Toolset is a comprehensive software solution specifically created to tackle these challenges. Sponsored by the U.S. Department of Energy's Water Power Technologies Office, this toolset empowers users to effectively manage, analyze, and visualize environmental data, while also facilitating simulation capabilities. It leverages a framework linking power systems and environmental outcomes through flow decisions, and can be used to create a bundled product of environmental and power systems information, methodologies/procedures, datasets, and a software tool that can enable participants in hydropower regulatory proceedings to evaluate tradeoffs between hydropower energy flexibility and environmental impacts. This enables the design of environmental flow measures that provide robust environmental protections while supporting power system outcomes and objectives. This toolset is utilized to benchmark environmental, power system, and economic endpoints, enabling hydropower stakeholders to weigh energy-environment tradeoffs in the design of operational regimes or environmental flow requirements, and providing an updated stakeholder tool in light of new operational needs for hydropower. Coauthors: Vishvas Chalishazar (Pacific Northwest National Laboratory), Brenda Pracheil (Pacific Northwest National Laboratory), Quentin Ploussard (Argonne National Laboratory), Thomas Veselka (Argonne National Laboratory), Mucun Sun (Idaho National Laboratory), Thushara De Silva (National Renewable Energy Laboratory), Paul Matson (Oak Ridge National Laboratory), Bryan Bozeman (Oak Ridge National Laboratory), Shaun Carney (RTI Consulting), Florian Kluibenschaedl (RTI Consulting) |
A Methodology to Address Varying Background in Scanning Surveys Jan Irvahn, Pacific Northwest National Laboratory ► VIEW POSTER |
Varying background radiation levels have posed challenges for scanning surveys. Implicit data processing takes place in real time while surveyors notice gradual changes in audible background throughout the scanning activity when surveys are conducted with vigilance. However, when surveys are conducted without vigilance, surveyors do not respond to audio output in real time but only after surveys are completed, by post-processing the collected rate meter data. Methods are therefore required for the purposes of analyzing these continuously collected survey data. Coauthors: Jennifer Huckett (Pacific Northwest National Laboratory), Dan Fortin (Pacific Northwest National Laboratory), Lisa Newburn (Pacific Northwest National Laboratory), Debbie Fagan (Pacific Northwest National Laboratory) |
Creating Python Tools for Processing Spectral Induced Polarization Data Aaron Jimenez, Colorado School of Mines ► VIEW POSTER |
Induced polarization (IP) is a geophysical technique that measures the polarization response of a porous medium when an electric field is induced. IP measurements are sensitive to physical, chemical, and biological properties of interest, which allow them to be used for environmental applications, including contaminant monitoring and characterization of the subsurface. The polarization response has been found to be frequency dependent; therefore, spectral IP (SIP) can be collected to give further insight into subsurface properties. The SIP response is usually governed by the polarization of the electron double layer at the mineral-fluid interface, but at frequencies above 100Hz, a different mechanism, termed Maxwell-Wagner polarization, is observed. SIP collects two data types: phase angle (rad), and complex impedance (Ω). PNNL researchers are conducting novel experiments where SIP data is collected and have developed individualized codes to view the data, but a standardized, single-platform way to review it has not been developed. Coauthor: Judy Robinson (Pacific Northwest National Laboratory) |
Re-oxidation Behavior of Technetium-99 and Uranium Immobilized by Zero Valent and Sulfur Modified Iron Reductant Yelena Katsenovich, Florida International University ► VIEW POSTER |
Technetium-99 (99Tc) is a concern for the Hanford Site due to its high solubility, toxicity, and mobility in the environment, and historical release to the subsurface. Previous studies have shown that formation of reduced Tc(IV) precipitates decreases mobilization but reoxidation may lead to remobilization. Here, the reoxidation of 99Tc was investigated under sequential anaerobic (Phase 1) for 28-30 days then aerobic (Phase 2) conditions for 40 days using different reductants, including 1% zero valent iron (ZVI), or 1% sulfur modified iron (SMI) totaling 1000 mg in 100 mL. Sediment samples from the Hanford Site and relevant synthetic solutions [either synthetic groundwater (GW) or pore water (PW)] were used in batch experiments, where 99Tc (as pertechnetate), nitrate, and uranium (VI) species were reduced under anaerobic conditions and then were subjected to reoxidation under aerobic conditions. The targeted initial concentrations in simulant solutions were 150 mg/L U and 10 µg/L 99Tc for the PW and 420 µg/L of 99Tc in addition to 124 mg/L NO3- for the GW. For GW samples, the molar ratio of Tc/Fe was 0.0023. For PW samples, the molar ratios of Tc/Fe and U/Fe were 0.000056 and 0.035, respectively. The objective was to investigate contaminant behavior, specifically Tc(VII), when commingled with NO3- and U(VI). Coauthors: Hilary Emerson (Pacific Northwest National Laboratory), Jim Szecsody (Pacific Northwest National Laboratory), Nik Qafoku (Pacific Northwest National Laboratory), Leonel Lagos (Florida International University) |
Unraveling the Complex Solution Chemistry of Aluminum in Hanford Site Nuclear Tank Waste Ashley Kennedy, Pacific Northwest National Laboratory ► VIEW POSTER |
Hanford tank waste is characterized by extremes in alkalinity and low water activity, with chemical phenomena driven far from equilibrium by ionizing radiation. The Interfacial Dynamics in Radioactive Environments and Materials (IDREAM) Energy Frontier Research Center is mastering fundamental phenomena in solutions and at interfaces in these complex chemical environments. One of the primary chemical mysteries investigated by IDREAM in the solubility, nucleation, crystallization, and aggregation behavior of aluminum. Aluminum, present in the waste from chemical decladding of fuels, occurs in solution as the tetrahedrally coordination aluminate anion and crystallizes in an unpredictable manner as octahedrally coordinated aluminum hydroxide polymorphs. Here, IDREAM is using a multifaceted characterization approach to investigate the role of counter-ions, hydrogen bonding, and radiation effects in aluminate speciation in solution and nucleation mechanisms across spatial and temporal scales. Radiation is leveraged both as a variable in the system and to preferentially induce transient species in these materials under laboratory timescales. Ultimately, IDREAM seeks to determine the mechanisms required for a predictive understanding of aluminum hydroxide crystallization in Hanford tank waste. Coauthors: Ashley Kennedy, Emily Nienhuis (Pacific Northwest National Laboratory), Carolyn Pearce (Pacific Northwest National Laboratory), Micah Prange, Linda Young, Jay LaVerne, Thom Orlando, Greg Kimmel, Lili Liu, Sebastian Mergelsberg, Lixin Lu, Shuai Li, Shawn Riechers, Larry Anovitz, Xin Zhang, Zheming Wang, Xiaosong Li, Aurora Clark, Kevin Rosso (Pacific Northwest National Laboratory), Greg Schenter (Pacific Northwest National Laboratory) |
Collaboration-Driven Characterisation to Expedite a Remedial Strategy Vicky Newling, Quintessa Limited ► VIEW POSTER |
In traditional site characterisation and remediation programmes, specialists are introduced in sequence at the point of demand, through the client supply chain. This creates reliance upon the skills and judgement of preceding specialists from related disciplines to provide the data required to perform assessments and to make underpinned decisions. Commonly, this approach results in the identification of data gaps and uncertainties as each new specialist is brought on-line. This weakens the decision-making process, drives further cycles of data gathering and results in lengthened programmes, increases in costs and creates data management challenges. |
Suite of Groundwater Simulation Datasets for Deconvoluting Average and Hydrogeologic Layer-Specific Concentrations Amber Nguyen, University of Texas, Austin ► VIEW POSTER |
The data provided by groundwater samples collected from wells provides important characterization and performance assessment information for environmental remediation of aquifers. However, there are challenges in interpreting “point” data relative to the three-dimensional (3D) context of groundwater contaminant plumes, particularly when wells are not designed as monitoring wells with shorter well screens for targeted zones. Older wells or wells designed as injection or extraction wells can have relatively long (> 10 ft) screens that potentially span multiple subsurface formation materials, with water entering (and leaving) the borehole at different rates correlated with transmissivity. Traditional samples taken from such a well are an averaged mixture of water from the different geological units, making it difficult to interpret the 3D extent of contamination and contaminant mass flux from particular zones. A recent paper described equations and the data needed to better interpret groundwater concentrations for long-screened wells. The work here used the FloPy Python package to construct a set of MODFLOW and MT3DMS models for groundwater flow and contaminant transport. This set of simulations provided data sets with known inputs and known outputs that could be used to test the published equations to verify their suitability and assess any limitations. Findings of this study will allow for a better understanding of what data should be collected and how to interpret this data, particularly for long-screen wells. Coauthor: Christian Johnson (Pacific Northwest National Laboratory) |
Upper Tolerance Limits for Radiological Decision Making Moses Obiri, Pacific Northwest National Laboratory ► VIEW POSTER |
Cleanup decisions based on dose assessment criteria typically utilize comparisons between observed mean concentrations and action limits based on a risk model or background value (NUREG-1575, NUREG-1757, NUREG-1505). However, there are situations where decisions should be based on comparing the upper tail of a distribution, rather than a mean, to an action limit. Examples include: 1) in decommissioning, when the site is classified as a MARSSIM Class III area (no prior knowledge that contamination is present) (NUREG-1575), 2) in consequence management, when determining whether an area is contaminated, and 3) in the event of a fuel transportation accident when it is important to verify that no radioactivity has been released. In these cases, as well as others, the relevant determination is whether some high percentage of an area or population of individuals is below an action limit. Coauthors: Jennifer Huckett (Pacific Northwest National Laboratory), Debbie Fagan (Pacific Northwest National Laboratory), Lisa Newburn (Pacific Northwest National Laboratory) |
Determination of Radiological Stressors in Petroleum Drilling, Bitumen Exploitation, and Coal Mining Sites in Southern Nigeria Olalekan Olatunji, Keele University, United Kingdom ► VIEW POSTER |
In-situ exposure measurement and activity concentration of naturally occurring radioactive materials in different environmental media were determined in vicinity of petroleum drilling, bitumen and coal mining site in southern Nigeria using an advanced gamma scout survey meter and sodium iodide detector NaI (TI) gamma-ray spectroscopy techniques. Coauthors: Sharon George (Keele University), Ralf Halama (Keele University) |
Evaluating Geostatistical Realizations of Contaminant Plumes with Supervised Machine Learning Based on Well Performance Data Rohan Shanbhag, Florida International University ► VIEW POSTER |
The distributions of subsurface contamination are crucial for deciding remediation targets, designing remediation systems [e.g., pump and treat (P&T) networks], managing sites, and estimating remediation costs. However, the estimation of contamination distributions often suffers from great uncertainty due to limited groundwater sampling data. Traditionally, geostatistical simulations or other interpolation methods are used to generate possible plume distributions. Then, numerical models are used to calibrate/evaluate these distributions along with other conceptual model factors through inverse modeling or sensitivity analysis. This study explores an alternative data-driven approach that uses supervised machine learning models to correlate well performance data collected during P&T operations with geostatistical plume distributions, and then selects the best possible distribution. The Random Forest (RF) method is used to build regression-based machine learning models that predict which plume distribution better explains the spatial variability of well performance. Summary statistics on the well performance data and geostatistical concentration data are used as predictors and targets of the RF model. To overcome the limited number of wells in the original dataset for training and fitting the machine learning model, a bootstrapping technique is also applied. The CCl4 plume data in the 200 West Area P&T of Hanford site is used as a case study to evaluate this method. RF models are constructed and trained for each of 100 geostatistical CCl4 plume realizations for 200 West Area P&T, and corresponding goodness-of-fit measures are generated and compared to identify which realizations consistently perform better. This study demonstrates a potential new data-driven approach to evaluate multiple geostatistical realizations of contaminant plumes using actual extraction well data. Coauthors: Xuehang Song (Pacific Northwest National Laboratory), Mark Rockhold (Pacific Northwest National Laboratory), Marinko Karanovic, Matt Tonkin, Inci Demirkanli (Pacific Northwest National Laboratory), Rob Mackley (Pacific Northwest National Laboratory) |
Functionality of Coal Fly Ash in Mortars After Rare Earth Element Recovery Rahima Tufail, Wayne State University ► VIEW POSTER |
Rare earth elements (REEs) are a set of 17 elements (lanthanides plus scandium and yttrium) that are not found concentrated in large ore deposits in nature. There has been huge growth in the amount of technology that uses REEs due to their diverse properties, but REEs are difficult to mine and expensive to separate. The global supply of REEs is largely controlled by a single country (China; >95%), and the lack of a domestic supply has been identified as a vulnerability to U.S. economic security. To alleviate this vulnerability, economical extraction methods from traditionally overlooked domestic feedstocks, such coal fly ash, are being explored. Coal fly ash is a waste product produced after coal is burned for power generation. A traditional use of fly ash is as an additive to concrete aggregate. However, it is unknown whether acidified fly ash, after the recovery of REEs, may be suitable for use in concrete. This study focuses on characterizing acidified fly ash after REE extraction, with the goal of determining how much ash may be added to mortar without eliminating the functionality of the fly ash. The particle size and roughness of fly ash particles will be studied using scanning electron microscopy imaging before and after exposing the ash to an acid leaching process for the recovery of REEs. The waste ash from the leaching process will then be added to mortars in various fractions. The mortar functionality will be tested by measuring the strength of the samples. A future study will include mechanical testing of concrete cores made using the mortars with the acidified fly ash. Coauthors: Timothy Dittrich (Wayne State University), Matthew Allen (Wayne State University), Sarah Brownlee (Wayne State University), Mohammed Dardona (Wayne State University), Sai Praneeth (Wayne State University), Dimitrios Kakaris (Wayne State University), Chandra Tummala (Wayne State University), Preetom Kishore Roy (Wayne State University) |
|