Convergence of deterministic and stochastic approaches in optimal remediation design of a contaminated aquifer

The comparison between two series of optimal remediation designs using deterministic and stochastic approaches showed a number of converging features. Limited sampling measurements in a supposed contaminated aquifer formed the hydraulic conductivity field and the initial concentration distribution used in the optimization process. The deterministic and stochastic approaches employed a single simulation–optimization method and a multiple realization approach, respectively. For both approaches, the optimization model made use of a genetic algorithm. In the deterministic approach, the total cost, extraction rate, and the number of wells used increase when the design must satisfy the intensified concentration constraint. Growing the stack size in the stochastic approach also brings about same effects. In particular, the change in the selection frequency of the used extraction wells, with increasing stack size, for the stochastic approach can indicate the locations of required additional wells in the deterministic approach due to the intensified constraints. These converging features between the two approaches reveal that a deterministic optimization approach with controlled constraints is achievable enough to design reliable remediation strategies, and the results of a stochastic optimization approach are readily available to real contaminated sites.     

Remediation of Heterogeneous Aquifers Subject to Uncertainty

Optimal cost pump-and-treat ground water remediation designs for containment of a contaminated aquifer are often developed using deterministic ground water models to predict ground water flow. Uncertainty in hydraulic conductivity fields used in these models results in remediation designs that are unreliable. The degree to which uncertainty contributes to the reliability of remediation designs as measured by the characterization of the uncertainty is shown to differ depending upon the geologic environments of the models. This conclusion is drawn from the optimal design costs for multiple deterministic models generated to represent the uncertainty of four distinct models with different geologic environments. A multi scenario approach that includes uncertainty into the remediation design called the deterministic method for optimization subject to uncertainty (DMOU) is applied to these distinct models. It is found that the DMOU is a method for determining a remediation design subject to uncertainty that requires minimal postprocessing efforts. Preprocessing, however, is required for the application of the DMOU to unique problems. In the ground water remediation design problems, the orientation of geologic facies with respect to the orientation of flow patterns, pumping well locations, and constraint locations are shown to affect the preprocessing, the solutions to the DMOU problems, and the computational efficiency of the DMOU approach. The results of the DMOU are compared to the results of a statistical analysis of the effects of the uncertainty on remediation designs. This comparison validates the efficacy of the DMOU and illustrates the computational advantages to using the DMOU over statistical measures.     

A stochastic optimization framework for the restoration of an over-exploited aquifer

ABSTRACT

This study investigates the impact of hydraulic conductivity uncertainty on the sustainable management of the aquifer of Lake Karla, Greece, using the stochastic optimization approach. The lack of surface water resources in combination with the sharp increase in irrigation needs in the basin over the last 30 years have led to an unprecedented degradation of the aquifer. In addition, the lack of data regarding hydraulic conductivity in a heterogeneous aquifer leads to hydrogeologic uncertainty. This uncertainty has to be taken into consideration when developing the optimization procedure in order to achieve the aquifer’s sustainable management. Multiple Monte Carlo realizations of this spatially-distributed parameter are generated and groundwater flow is simulated for each one of them. The main goal of the sustainable management of the ‘depleted’ aquifer of Lake Karla is two-fold: to determine the optimum volume of renewable groundwater that can be extracted, while, at the same time, restoring its water table to a historic high level. A stochastic optimization problem is therefore formulated, based on the application of the optimization method for each of the aquifer’s multiple stochastic realizations in a future period. In order to carry out this stochastic optimization procedure, a modelling system consisting of a series of interlinked models was developed. The results show that the proposed stochastic optimization framework can be a very useful tool for estimating the impact of hydraulic conductivity uncertainty on the management strategies of a depleted aquifer restoration. They also prove that the optimization process is affected more by hydraulic conductivity uncertainty than the simulation process.

Editor Z.W. Kundzewicz; Guest editor S. Weijs     

Prediction of Remediation of a Heterogeneous Aquifer: A Case Study

Contaminant plumes whose characteristic length is smaller than the horizontal integral scale of the hydraulic conductivity, K, are abundant in shallow, phreatic aquifers. In such cases, the aquifer can be regarded as layered, with K being only a function of the vertical coordinate. The heterogeneity of K has a critical role upon the efficiency of remediation of such sites, for example, by Pump and Treat schemes. The expected efficiency is a random variable, with uncertainty. Quantifying this uncertainty can be of great importance to decision making. In this study, we focus on a case study in the coastal aquifer of Israel and compare two different approaches for constructing realizations of K: continuous and indicator. We observe a significant difference between the constructed realizations, which results in a considerable difference in the predicted remediation efficiency and its uncertainty. Furthermore, we study the effect of conditioning the realizations by a rather limited number of K data points. We find that the conditioning results in a major reduction of the uncertainty. In addition, we compare the results of the transport model to a simplified semi‐analytical solution that is based on assuming radial flow. We find a good agreement with the three‐dimensional numerical model. This result illustrates that the simplified solution can be used for prediction of the remediation efficiency when the flow at the plume vicinity can be regarded as radial.     

A comparative study of Monte Carlo simple genetic algorithm and noisy genetic algorithm for cost-effective sampling network design under uncertainty

This study evaluates and compares two methodologies, Monte Carlo simple genetic algorithm (MCSGA) and noisy genetic algorithm (NGA), for cost-effective sampling network design in the presence of uncertainties in the hydraulic conductivity (K) field. Both methodologies couple a genetic algorithm (GA) with a numerical flow and transport simulator and a global plume estimator to identify the optimal sampling network for contaminant plume monitoring. The MCSGA approach yields one optimal design each for a large number of realizations generated to represent the uncertain K-field. A composite design is developed on the basis of those potential monitoring wells that are most frequently selected by the individual designs for different K-field realizations. The NGA approach relies on a much smaller sample of K-field realizations and incorporates the average of objective functions associated with all K-field realizations directly into the GA operators, leading to a single optimal design. The efficacy of the MCSGA-based composite design and the NGA-based optimal design is assessed by applying them to 1000 realizations of the K-field and evaluating the relative errors of global mass and higher moments between the plume interpolated from a sampling network and that output by the transport model without any interpolation. For the synthetic application examined in this study, the optimal sampling network obtained using NGA achieves a potential cost savings of 45% while keeping the global mass and higher moment estimation errors comparable to those errors obtained using MCSGA. The results of this study indicate that NGA can be used as a useful surrogate of MCSGA for cost-effective sampling network design under uncertainty. Compared with MCSGA, NGA reduces the optimization runtime by a factor of 6.5.     

Cost Optimization of DNAPL Remediation at Dover Air Force Base Site

This study investigates stochastic optimization of dense nonaqueous phase liquid (DNAPL) remediation design at Dover Air Force Base Area 5 using emulsified vegetable oil (EVO) injection. The Stochastic Cost Optimization Toolkit (SCOToolkit) is used for the study, which couples semianalytical DNAPL source depletion and transport models with parameter estimation, error propagation, and stochastic optimization modules that can consider multiple sources and remediation strategies. Model parameters are calibrated to field data conditions on prior estimates of parameters and their uncertainty. Monte Carlo simulations are then performed to identify optimal remediation decisions that minimize the expected net present value (NPV) cleanup cost while maintaining concentrations at compliance wells under the maximum contaminant level (MCL). The results show that annual operating costs could be reduced by approximately 50% by implementing the identified optimal remediation strategy. We also show that recalibration and reoptimization after 50 years using additional monitoring data could lead to a further 60% reduction in annual operating cost increases the reliability of the proposed remediation actions.     

Data-Worth Assessment for a Three-Dimensional Optimal Design in Nonlinear Groundwater Systems

Groundwater model predictions are often uncertain due to inherent uncertainties in model input data. Monitored field data are commonly used to assess the performance of a model and reduce its prediction uncertainty. Given the high cost of data collection, it is imperative to identify the minimum number of required observation wells and to define the optimal locations of sampling points in space and depth. This study proposes a design methodology to optimize the number and location of additional observation wells that will effectively measure multiple hydrogeological parameters at different depths. For this purpose, we incorporated Bayesian model averaging and genetic algorithms into a linear data-worth analysis in order to conduct a three-dimensional location search for new sampling locations. We evaluated the methodology by applying it along a heterogeneous coastal aquifer with limited hydrogeological data that is experiencing salt water intrusion (SWI). The aim of the model was to identify the best locations for sampling head and salinity data, while reducing uncertainty when predicting multiple variables of SWI. The resulting optimal locations for new observation wells varied with the defined design constraints. The optimal design (OD) depended on the ratio of the start-up cost of the monitoring program and the installation cost of the first observation well. The proposed methodology can contribute toward reducing the uncertainties associated with predicting multiple variables in a groundwater system.     

Optimization of groundwater-monitoring networks for identification of the distribution of a contaminant plume

A new methodology is proposed to optimize monitoring networks for identification of the extent of contaminant plumes. The optimal locations for monitoring wells are determined as the points where maximal decreases are expected in the quantified uncertainty about contaminant existence after well installation. In this study, hydraulic conductivity is considered to be the factor that causes uncertainty. The successive random addition (SRA) method is used to generate random fields of hydraulic conductivity. The expected value of information criterion for the existence of a contaminant plume is evaluated based on how much the uncertainty of plume distribution reduces with increases in the size of the monitoring network. The minimum array of monitoring wells that yields the maximum information is selected as the optimal monitoring network. In order to quantify the uncertainty of the plume distribution, the probability map of contaminant existence is made for all generated contaminant plume realizations on the domain field. The uncertainty is defined as the sum of the areas where the probability of contaminant existence or nonexistence is uncertain. Results of numerical experiments for determination of optimal monitoring networks in heterogeneous conductivity fields are presented.     

Data-worth analysis for multiobjective optimal design of pump-and-treat remediation systems

The design and the management of pump-and-treat (PAT) remediation systems for contaminated aquifers under uncertain hydrogeological settings and parameters often involve decisions that trade off cost optimality against reliability. Both design objectives can be improved by planning site characterization programs that reduce subsurface parameter uncertainty. However, the cost for subsurface investigation often weighs heavily upon the budget of the remedial action and must thus be taken into account in the trade-off analysis. In this paper, we develop a stochastic data-worth framework with the purpose of estimating the economic opportunity of subsurface investigation programs. Since the spatial distribution of hydraulic conductivity is most often the major source of uncertainty, we focus on the direct sampling of hydraulic conductivity at prescribed locations of the aquifer. The data worth of hydraulic conductivity measurements is estimated from the reduction of the overall management cost ensuing from the reduction in parameter uncertainty obtained from sampling. The overall cost is estimated as the expected value of the cost of installing and operating the PAT system plus penalties incurred due to violations of cleanup goals and constraints. The crucial point of the data-worth framework is represented by the so-called pre-posterior analysis. Here, the tradeoff between decreasing overall costs and increasing site-investigation budgets is assessed to determine a management strategy proposed on the basis of the information available at the start of remediation. The goal of the pre-posterior analysis is to indicate whether the proposed management strategy should be implemented as is, or re-designed on the basis of additional data collected with a particular site-investigation program. The study indicates that the value of information is ultimately related to the estimates of cleanup target violations and decision makers’ degree of risk-aversion.     

Maximizing Net Extraction Using an Injection‐Extraction Well Pair in a Coastal Aquifer

In this study, we examine the maximum net extraction rate from the novel arrangement of an injection‐extraction well pair in a coastal aquifer, where fresh groundwater is reinjected through the injection well located between the interface toe and extraction well. Complex potential theory is employed to derive a new analytical solution for the maximum net extraction rate and corresponding stagnation‐point locations and recirculation ratio, assuming steady‐state, sharp‐interface conditions. The injection‐extraction well‐pair system outperforms a traditional single extraction well in terms of net extraction rate for a broad range of well placement and pumping rates, which is up to 50% higher for an aquifer with a thickness of 20 m, hydraulic conductivity of 10 m/d, and fresh water influx of 0.24 m²/d. Sensitivity analyses show that for a given fresh water discharge from an inland aquifer, a larger maximum net extraction is expected in cases with a smaller hydraulic conductivity or a smaller aquifer thickness, notwithstanding physical limits to drawdown at the pumping well that are not considered here. For an extraction well with a fixed location, the optimal net extraction rate linearly increases with the distance between the injection well and the sea, and the corresponding injection rate and recirculation ratio also increase. The analytical analysis in this study provides initial guidance for the design of well‐pair systems in coastal aquifers, and is therefore an extension beyond previous applications of analytical solutions of coastal pumping that apply only to extraction or injection wells.     

Bayesian Chance‐Constrained Hydraulic Barrier Design under Geological Structure Uncertainty

The groundwater community has widely recognized geological structure uncertainty as a major source of model structure uncertainty. Previous studies in aquifer remediation design, however, rarely discuss the impact of geological structure uncertainty. This study combines chance‐constrained (CC) programming with Bayesian model averaging (BMA) as a BMA‐CC framework to assess the impact of geological structure uncertainty in remediation design. To pursue this goal, the BMA‐CC method is compared with traditional CC programming that only considers model parameter uncertainty. The BMA‐CC method is employed to design a hydraulic barrier to protect public supply wells of the Government St. pump station from salt water intrusion in the “1500‐foot” sand and the “1700‐foot” sand of the Baton Rouge area, southeastern Louisiana. To address geological structure uncertainty, three groundwater models based on three different hydrostratigraphic architectures are developed. The results show that using traditional CC programming overestimates design reliability. The results also show that at least five additional connector wells are needed to achieve more than 90% design reliability level. The total amount of injected water from the connector wells is higher than the total pumpage of the protected public supply wells. While reducing the injection rate can be achieved by reducing the reliability level, the study finds that the hydraulic barrier design to protect the Government St. pump station may not be economically attractive.     

Optimizing remediation of an unconfined aquifer using a hybrid algorithm

We present a novel hybrid algorithm, integrating a genetic algorithm (GA) and constrained differential dynamic programming (CDDP), to achieve remediation planning for an unconfined aquifer. The objective function includes both fixed and dynamic operation costs. GA determines the primary structure of the proposed algorithm, and a chromosome therein implemented by a series of binary digits represents a potential network design. The time-varying optimal operation cost associated with the network design is computed by the CDDP, in which is embedded a numerical transport model. Several computational approaches, including a chromosome bookkeeping procedure, are implemented to alleviate computational loading. Additionally, case studies that involve fixed and time-varying operating costs for confined and unconfined aquifers, respectively, are discussed to elucidate the effectiveness of the proposed algorithm. Simulation results indicate that the fixed costs markedly affect the optimal design, including the number and locations of the wells. Furthermore, the solution obtained using the confined approximation for an unconfined aquifer may be infeasible, as determined by an unconfined simulation.     

Optimal design of pumping tests in leaky aquifers for stream depletion analysis

We analyze the optimal design of a pumping test for estimating hydrogeologic parameters that are subsequently used to predict stream depletion caused by groundwater pumping in a leaky aquifer. A global optimization method is used to identify the test’s optimal duration and the number and locations of observation wells. The objective is to minimize predictive uncertainty (variance) of the estimated stream depletion, which depends on the sensitivities of depletion and drawdown to relevant hydrogeologic parameters. The sensitivities are computed analytically from the solutions of Zlotnik and Tartakovsky [Zlotnik, V.A., Tartakovsky, D.M., 2008. Stream depletion by groundwater pumping in leaky aquifers. ASCE Journal of Hydrologic Engineering 13, 43–50] and the results are presented in a dimensionless form, facilitating their use for planning of pumping test at a variety of sites with similar hydrogeological settings. We show that stream depletion is generally very sensitive to aquitard’s leakage coefficient and stream-bed’s conductance. The optimal number of observation wells is two, their optimal locations are one close to the stream and the other close to the pumping well. We also provide guidelines on the test’s optimal duration and demonstrate that under certain conditions estimation of aquitard’s leakage coefficient and stream-bed’s conductance requires unrealistic test duration and/or signal-to-noise ratio.     

Water Table Uncertainties due to Uncertainties in Structure and Properties of an Unconfined Aquifer

We consider two sources of geology‐related uncertainty in making predictions of the steady‐state water table elevation for an unconfined aquifer. That is the uncertainty in the depth to base of the aquifer and in the hydraulic conductivity distribution within the aquifer. Stochastic approaches to hydrological modeling commonly use geostatistical techniques to account for hydraulic conductivity uncertainty within the aquifer. In the absence of well data allowing derivation of a relationship between geophysical and hydrological parameters, the use of geophysical data is often limited to constraining the structural boundaries. If we recover the base of an unconfined aquifer from an analysis of geophysical data, then the associated uncertainties are a consequence of the geophysical inversion process. In this study, we illustrate this by quantifying water table uncertainties for the unconfined aquifer formed by the paleochannel network around the Kintyre Uranium deposit in Western Australia. The focus of the Bayesian parametric bootstrap approach employed for the inversion of the available airborne electromagnetic data is the recovery of the base of the paleochannel network and the associated uncertainties. This allows us to then quantify the associated influences on the water table in a conceptualized groundwater usage scenario and compare the resulting uncertainties with uncertainties due to an uncertain hydraulic conductivity distribution within the aquifer. Our modeling shows that neither uncertainties in the depth to the base of the aquifer nor hydraulic conductivity uncertainties alone can capture the patterns of uncertainty in the water table that emerge when the two are combined.     

Evaluation of Combined Direct-Push Methods Used for Aquifer Model Generation

Most established methods to characterize aquifer structure and hydraulic conductivities of hydrostratigraphical units are not capable of delivering sufficient information in the spatial resolution that is desired for sophisticated numerical contaminant transport modeling and adapted remediation design. With hydraulic investigation methods based on the direct-push (DP) technology such as DP slug tests, DP injection logging, and the hydraulic profiling tool, it is possible to rapidly delineate hydrogeological structures and estimate their hydraulic conductivity in shallow unconsolidated aquifers without the need for wells. A combined application of these tools was used for the investigation of a contaminated German refinery site and for the setup of hydraulic aquifer models. The quality of DP investigation and the models was evaluated by comparisons of tracer transport simulations using these models and measured breakthroughs of two natural gradient tracer tests. Model scenarios considering the information of all tools together showed good reproduction of the measured breakthroughs, indicating the suitability of the approach and a minor impact of potential technical limitations. Using the DP slug tests alone yielded significantly higher deviations for the determined hydraulic conductivities compared to considering two or three of the tools. Realistic aquifer models developed on basis of such combined DP investigation approaches can help optimize remediation concepts or identify flow regimes for aquifers with a complex structure.     

Calibration of high-resolution geophysical data with tracer test measurements to improve hydrological predictions

Relationships between porosity and hydraulic conductivity tend to be strongly scale- and site-dependent and are thus very difficult to establish. As a result, hydraulic conductivity distributions inferred from geophysically derived porosity models must be calibrated using some measurement of aquifer response. This type of calibration is potentially very valuable as it may allow for transport predictions within the considered hydrological unit at locations where only geophysical measurements are available, thus reducing the number of well tests required and thereby the costs of management and remediation. Here, we explore this concept through a series of numerical experiments. Considering the case of porosity characterization in saturated heterogeneous aquifers using crosshole ground-penetrating radar and borehole porosity log data, we use tracer test measurements to calibrate a relationship between porosity and hydraulic conductivity that allows the best prediction of the observed hydrological behavior. To examine the validity and effectiveness of the obtained relationship, we examine its performance at alternate locations not used in the calibration procedure. Our results indicate that this methodology allows us to obtain remarkably reliable hydrological predictions throughout the considered hydrological unit based on the geophysical data only. This was also found to be the case when significant uncertainty was considered in the underlying relationship between porosity and hydraulic conductivity.     

Investigation of Small‐Scale Preferential Flow with a Forced‐Gradient Tracer Test

A new tracer experiment (referred to as MADE‐5) was conducted at the well‐known Macrodispersion Experiment (MADE) site to investigate the influence of small‐scale mass‐transfer and dispersion processes on well‐to‐well transport. The test was performed under dipole forced‐gradient flow conditions and concentrations were monitored in an extraction well and in two multilevel sampler (MLS) wells located at 6, 1.5, and 3.75 m from the source, respectively. The shape of the breakthrough curve (BTC) measured at the extraction well is strongly asymmetric showing a rapidly arriving peak and an extensive late‐time tail. The BTCs measured at seven different depths in the two MLSs are radically different from one another in terms of shape, arrival times, and magnitude of the concentration peaks. All of these characteristics indicate the presence of a complex network of preferential flow pathways controlling solute transport at the test site. Field‐experimental data were also used to evaluate two transport models: a stochastic advection‐dispersion model (ADM) based on conditional multivariate Gaussian realizations of the hydraulic conductivity field and a dual‐domain single‐rate (DDSR) mass‐transfer model based on a deterministic reconstruction of the aquifer heterogeneity. Unlike the stochastic ADM realizations, the DDSR accurately predicted the magnitude of the concentration peak and its arrival time (within a 1.5% error). For the multilevel BTCs between the injection and extraction wells, neither model reproduced the observed values, indicating that a high‐resolution characterization of the aquifer heterogeneity at the subdecimeter scale would be needed to fully capture 3D transport details.     

Assessing leakage detectability at geologic CO2 sequestration sites using the probabilistic collocation method

We present an efficient methodology for assessing leakage detectability at geologic carbon sequestration sites under parameter uncertainty. Uncertainty quantification (UQ) and risk assessment are integral and, in many countries, mandatory components of geologic carbon sequestration projects. A primary goal of risk assessment is to evaluate leakage potential from anthropogenic and natural features, which constitute one of the greatest threats to the integrity of carbon sequestration repositories. The backbone of our detectability assessment framework is the probability collocation method (PCM), an efficient, nonintrusive, uncertainty-quantification technique that can enable large-scale stochastic simulations that are based on results from only a small number of forward-model runs. The metric for detectability is expressed through an extended signal-to-noise ratio (SNR), which incorporates epistemic uncertainty associated with both reservoir and aquifer parameters. The spatially heterogeneous aquifer hydraulic conductivity is parameterized using Karhunen–Loève (KL) expansion. Our methodology is demonstrated numerically for generating probability maps of pressure anomalies and for calculating SNRs. Results indicate that the likelihood of detecting anomalies depends on the level of uncertainty and location of monitoring wells. A monitoring well located close to leaky locations may not always yield the strongest signal of leakage when the level of uncertainty is high. Therefore, our results highlight the need for closed-loop site characterization, monitoring network design, and leakage source detection.     

Stochastic Cost Optimization of Multistrategy DNAPL Site Remediation

This paper investigates numerical optimization of dense nonaqueous phase liquid (DNAPL) site remediation design considering effects of prediction and measurement uncertainty. Results are presented for a hypothetical problem involving remediation using thermal source reduction (TSR) and bioremediation with electron donor (ED) injection. Pump-and-treat is utilized as a backup measure if compliance criteria are not met. Remediation system design variables are optimized to minimize expected net present value (ENPV) cost. Adaptive criteria are assumed for real-time control of TSR and ED duration. Source zone dissolved concentration data enabled more reliable and lower cost operation of TSR than soil concentration data, but using both soil and dissolved data improved results sufficiently to more than offset the additional cost. Decisions to terminate remediation and monitoring or to initiate pump-and-treat are complicated by measurement noise. Simultaneous optimization of monitoring frequency, averaging period, and lookback periods to confirm decisions, in addition to remediation design variables, reduced ENPV cost. Results indicate that remediation design under conditions of uncertainty is affected by subtle interactions and tradeoffs between design variables, compliance rules, site characteristics, and uncertainty in model predictions and monitoring data. Optimized designs yielded cost savings of up to approximately 50% compared with a nonoptimized design based on common engineering practices. Significant improvements in accuracy and reductions in cost were achieved by recalibrating the model to data collected during remediation and re-optimizing design variables. Repeating this process periodically is advisable to minimize total costs and maximize reliability.     

Dynamic optimal ground water remediation including fixed and operation costs

In time-varying ground water remediation, the lack of an optimal control algorithm to simultaneously consider fixed costs and time-varying operating costs makes it nearly impossible to obtain an optimal solution. This study presents a novel algorithm that integrates a genetic algorithm (GA) and constrained differential dynamic programming (CDDP) to solve this time-varying ground water remediation problem. A GA can easily incorporate the fixed costs associated with the installation of wells. However, using a GA to solve for time-varying policies would dramatically increase the computational resources required. Therefore, the CDDP is used to handle the subproblems associated with time-varying operating costs. A hypothetical case study that incorporates fixed and time-varying operating costs is presented to demonstrate the effectiveness of the proposed algorithm. Simulation results indicate that the fixed costs can significantly influence the number and locations of wells, and a notable total cost savings can be realized by applying the novel algorithm herein.