Amy B. Chan Hilton

Optimal groundwater monitoring design using an ant colony optimization paradigm

Abstract Groundwater long-term monitoring (LTM) is required to assess the performance of groundwater remediation and human being health risk at post-closure sites where groundwater contaminants are still present. The large number of sampling locations can make the LTM costly, especially since LTM may be required over several decades. An optimization algorithm based on the ant colony optimization (ACO) paradigm is developed to minimize the overall data loss due to fewer sampling locations for a given number of monitoring wells. The ACO method is inspired by the ability of an ant colony to identify the shortest route between their nest and a food source. The developed ACO-LTM algorithm is applied to a field site with an existing 30-well LTM network. When compared to the results identified through complete enumeration, the ACO-LTM solutions are globally optimal for the cases with 21 to 27 remaining wells. Results from the developed ACO-LTM algorithm provide a proof-of-concept for the application of the general ACO analogy to the groundwater LTM sampling location optimization problem. A major contribution of this work is the successful development of an efficient and effective stochastic search algorithm for solving the LTM optimization problem based on the ACO paradigm.

Optimizing Groundwater Remediation Design under Uncertainty

There always exists some degree of uncertainty associated with groundwater problems, often in determining the aquifer parameter values. Therefore finding an optimal remediation strategy based on a deterministic description of the system may not yield an optimal and feasible design. This work develops a genetic algorithm (GA) approach that takes into account the uncertainty of hydraulic conductivity values when determining the best remediation design possible. During the GA optimization, the heterogeneous hydraulic conductivity field realization varies between generations and on-going performance is measured. A policys fitness is based on its performance over multiple generations. Therefore the most fit policy should provide a robust solution since this policy would be a good design over a range of aquifer realizations. Results of this approach applied to a hypothetical contaminated aquifer remediated by a pump-and-treat system indicate that a set of non-dominated policies can be generated by this modified GA. Additionally, this work shows that using a deterministic description of the aquifer, either homogeneous or heterogeneous, can result in significant under-design, with poor reliability.

Multi-Objective Groundwater Remediation Design under Uncertainty Using Robust Genetic Algorithms

Given the inherent uncertainty in groundwater management problems uncertainty in determining aquifer parameter values, identifying an optimal remediation strategy based on a deterministic description of the system may not yield an optimal and feasible design. This work builds on the robust genetic algorithm (GA) developed by Chan Hilton and Culver. The robust GA is a simulation-optimization approach which combines a GA with a contaminant fate and transport simulation model and a spatially correlated random field generator to identify tradeoffs between design cost and reliability while considering uncertainty of hydraulic conductivity values. This work evaluates the application of the robust GA to two formulations of a groundwater remediation design problem. In this problem, the objectives are to minimize the cost of the remediation design while satisfying water quality constraints and indirectly maximizing the reliability of the designs. This is done by identifying the location and pumping rates of a set of extraction well used for pump-and-treat remediation. The results show that the robust GA can successfully identify cost-effective and reliable designs in a computationally efficient manner. Future work involving the robust GA and planned modifications also are discussed in this paper.

Analysis of Federal and State Policies and Environmental Issues for Bioethanol Production Facilities

The purpose of this work was to investigate incentives and barriers to fuel ethanol production from biomass in the U.S. during the past decade (2000-2010). In particular, we examine the results of policies and economic conditions during this period by way of cellulosic ethanol activity in four selected states with the potential to produce different types of feedstocks (i.e., sugar, starch, and cellulosic crops) for ethanol production (Florida, California, Hawaii, and Iowa). Two of the four states, Iowa and California, currently have commercial ethanol production facilities in operation using corn feedstocks. While several companies have proposed commercial scale facilities in Florida and Hawaii, none are operating to date. Federal and state policies and incentives, potential for feedstock production and conversion to ethanol and associated potential environmental impacts, and environmental regulatory conditions among the states were investigated. Additionally, an analysis of proposed and operational ethanol production facilities provided evidence that a combination of these policies and incentives along with the ability to address environmental issues and regulatory environment and positive economic conditions all impact ethanol production. The 2000-2010 decade saw the rise of the promise of cellulosic ethanol. Federal and state policies were enacted to increase ethanol production. Since the initial push for development, expansion of cellulosic ethanol production has not happened as quickly as predicted. Government and private funding supported the development of ethanol production facilities, which peaked and then declined by the end of the decade. Although there are technical issues that remain to be solved to more efficiently convert cellulosic material to ethanol while reducing environmental impacts, the largest barriers to increasing ethanol production appear to be related to government policies, economics, and logistical issues. The numerous federal and state policies do not effectively give investors confidence to commit to the construction and long-term operation of facilities under current economic conditions. Additional changes in policy and breakthroughs in technology and logistics will be required to address these hurdles to increases in ethanol production in the U.S. in the next decade.

An Algorithm for Groundwater Long-Term Monitoring Spatial Optimization by Analogy to Ant Colony Optimization for TSP

A methodology for optimizing groundwater long-term monitoring (LTM) is presented. Groundwater LTM is required to assess the performance of groundwater remediation and human being health risk at post-closure sites where groundwater contaminants are still present. Some monitoring wells in the existing LTM network may be redundant, making it possible to remove some of them without compromising data quality. An optimization algorithm based on the ant colony optimization (ACO) paradigm is developed to minimize the overall data loss by identifying a given number redundant sampling locations. The ACO method is inspired by the ability of ant colony to identify the shortest route between their nest and a food source. The algorithm searches for redundant wells from among sampling locations in the monitoring network and follows steps analogous to traveling salesman problem (TSP), which is a cardinal combinatorial problem successfully solved by ACO. Results from the developed ACO-LTM algorithm show global optima or near-optimal solutions were identified. A comparison of the ACOLTM results to those from complete enumeration indicates that the developed ACO-LTM algorithm is efficient and effective.

Development of Ant Colony Optimization for Long-Term Groundwater Monitoring

Groundwater remediation projects require long-term monitoring (LTM) to assess compliance of active remedial systems and post-closure sites where groundwater contamination is still present. LTM can be costly given the large number of sampling locations, frequency of monitoring, and number of constituents monitored at a given site. This work presents the development of a methodology to optimize a groundwater-monitoring network in order to maximize cost-effectiveness without compromising program and data quality. We propose method that combines ant colony optimization (ACO) with a genetic algorithm (GA). The ACO method is inspired by the fact that ants are able to find the shortest route between their nest and a food source. This is accomplished by using pheromone trails as a form of indirect communication. Ant colony simulation techniques are adapted to minimize the number of monitoring locations in the sampling network without significant loss of information.

Constraint-handling methods for optimal groundwater remediation design by genetic algorithms

Typically in optimal groundwater remediation design, the objective is to minimize the cost of remediation while meeting the water quality constraints by the end of the remediation period. Given that many common groundwater contaminants are hazardous at very low concentrations, even a small violation of the water quality may be the difference between reaching a hazardous or nonhazardous end point. Furthermore, the remediation costs increase dramatically as one attempts to remove the last units of concentration. This work compares two methods for constraint-handling, an additive penalty method and a multiplicative penalty, for use in optimal groundwater remediation design with a genetic algorithm. The multiplicative approach was found to be a more robust technique for finding cost-effective designs, while enforcing the water quality constraints.

Analysis of the Primal and Dual Problem for Long-Term Groundwater Monitoring Spatial Optimization

Long term monitoring (LTM) can be costly given the large number of sampling locations monitored at given site. Redundant monitoring wells in the existing LTM network make it possible to remove some of them while maintaining data reliability from the remaining wells. We can optimize a monitoring network design by maximizing cost-effectiveness without compromising data quality. Alternatively, decision makers may define their LTM goal depending on the budget; they may set a low number of monitoring wells and identify the best combination of sampling points among all remaining monitoring wells. The problem of LTM spatial optimization is a non-linear one. We formulate the LTM optimization problem in two ways: one is to minimize the number of remaining wells given constraint rules on data quality and estimation errors; the other is given the number of remaining wells, the objective is to determine the optimal combination of a reduced set of wells from among the original ground water monitoring network. Here we reverse the role of the number of remaining wells from objective function to constraint, and we call the former optimization formulation the primal problem and the latter the dual problem. An ant colony optimization (ACO) method is developed to solve these problems based on field data. The ACO method is inspired by the fact that ants are able to find the shortest route between their nest and a food source. Individual ants can contribute their own information by pheromones; and the shorter the path, the higher the density of pheromones. Increased pheromones will attract the ant colony to choose the shortest route. Two ACO algorithms for LTM optimization are developed in this work — one based on a binary combinatorial formulation and the other is analogous to the traveling salesman problem (TSP).

Dynamic Optimal Design of Groundwater Remediation Using Genetic Algorithms

The use of genetic algorithms for the dynamic optimal design of pump-and-treat groundwater remediation systems is demonstrated through two new dynamic formulations. In the first formulation in which the contaminant sorption was assumed to be in equilibrium, the lengths of management periods were decision variables. The second formulation assumed a pulsed pumping approach to remove a contaminant with mass-transfer-limited sorption. While the genetic algorithm could successfully solve these dynamic problems, only small percentage reductions in the overall remediation costs were achieved. However, the savings in the operational costs were more significant with the mass transfer-limited pulsed pumping example saving up to 10% compared to continuous pumping and the flexible-length management periods saving up to 3% compared to fixed-length periods. With the high costs of remediation, even a small percentage of savings in operational costs could be significant. For instance, the 3% savings with flexible management periods corresponded to a cost reduction of more than

Impact of Lipopolysaccharide Extraction on Bacterial Adhesion and Transport

26,000 that was achieved by allowing for variable length management periods, a relatively simple change within the GA algorithm. Dynamic pumping that is adapted over time to the unique site conditions may be an option to improve the cost-effectiveness of a remediation design, especially for mass-transfer limited sites.