Performance of Constraint-Handling Techniques Applied to a Genetic Algorithm-Based Watershed Management Model

Date of Award

Summer 2004

Document Type

Thesis

Degree Name

Master of Science (MS)

Department

Civil & Environmental Engineering

Program/Concentration

Civil Engineering

Committee Director

Laura J. Harrell

Committee Director

Duc T. Nguyen

Committee Member

Jaewan Yoon

Call Number for Print

Special Collections LD4331.E54 Y87 2004

Abstract

Wet detention ponds are a commonly used structural best management practice to reduce nonpoint source pollutant loading into receiving water bodies. Design of these ponds is typically carried out individually to meet a target total suspended solids (TSS) removal level. An improvement to this approach is to generate cost-effective pond configurations that meet system-wide targets for removal of pollutant loadings, corresponding to a specific build-out land use plan. The cost-effectiveness can be further improved through appropriate land use allocation planning conducted simultaneously with the design of detention ponds. Harrell and Ranjithan (2003) presented an evolutionary algorithm- (EA-) based modeling approach to address this problem. The model includes constraints on the system-wide land use distribution, which were handled using penalty functions. Most penalty function implementations, which are incorporated into the fitness criterion to degrade the fitness of solutions that violate constraints, require the user to specify and fine-tune penalty factors that determine the amount of penalty assigned as a function of the level of constraint violation. Alternative techniques have also been reported in the literature, including a multi-objective optimization technique that treats the constraints in single objective problems as additional objectives (Coello, 2000a) and the stochastic ranking procedure (Runarsson and Yao, 2000), both of which have been shown to perform efficiently for other engineering applications. This research investigates the performance of three penalty function implementations and two alternative techniques for constraint handling in a watershed management problem. The alternative techniques include the multi-objective technique proposed by Coello (2000a), and a proposed stochastic selection technique. The various methods are implemented in the EA-based watershed management model and applied to a case study involving the City Lake watershed in North Carolina. The performance of the EA, including the ability to consistently find high-quality solutions that satisfy the constraints, and the fine-tuning and computational effort required, is investigated for each technique. A comparison of the perforn1ances is made to help provide guidelines for the most promising constraint-handling techniques for EA-based watershed management.

Rights

In Copyright. URI: http://rightsstatements.org/vocab/InC/1.0/ This Item is protected by copyright and/or related rights. You are free to use this Item in any way that is permitted by the copyright and related rights legislation that applies to your use. For other uses you need to obtain permission from the rights-holder(s).

DOI

10.25777/z48g-pj57

This document is currently not available here.

Share

COinS