A neuro-fuzzy knowledge-based multi-criterion decision model for constructability analysis and improvement of construction technologies
Abstract
Selecting the most appropriate construction technology for a given project scenario is one of the most effective approaches for constructability improvement. Experienced construction engineers and managers are frequently involved in applying their constructability knowledge to technology selection during the early phase of a project's lifecycle in order to achieve cost effective construction. Due to natural attrition and other causes, the constructability knowledge of experienced personnel is diminishing in many construction firms. Moreover, because of the complex nature of construction operations, it is extremely difficult for construction experts to express their constructability knowledge while considering many variable attributes simultaneously. A methodology for automated constructability analysis and knowledge acquisition has yet to be developed. Without such a methodology, timely selection of the most appropriate construction technology and accumulation of knowledge for technology improvement will remain difficult. This research is the first work on both quantifying the conventional descriptive definition of constructability and on exploring the learning ability of neuro-fuzzy networks for automatic constructability knowledge acquisition. The developed methodology differentiates itself from the traditional constructability analysis and improvement approaches in two aspects: 1) the quantitative definition of constructability is adopted instead of the traditional descriptive definition, so that constructability can be measured, estimated, and improved; (2) the self-learning techniques for constructability knowledge acquisition are adopted instead of the traditional manual human-input approaches, so that the automation of constructability knowledge acquisition and accumulation becomes possible. With this generic methodology, construction firms can develop their own decision support systems to analyze and solve specific constructability problems according to their specialized fields. The result of this research is a tool for continuous constructability improvement of construction projects and technologies. A prototype computer implementation of the proposed methodology named COnstrUction techNology SELectOR (COUNSELOR) is developed for constructability analysis and improvement of concrete formwork technologies. The COUNSELOR system has demonstrated its abilities to quantify constructability according to the constructor's characteristics, detect potential constructability problems before the construction phase, and propose solutions for constructability improvement. The result of the research indicates a promising solution for barriers encountered in the implementation of conventional constructability approaches.
Degree
Ph.D.
Advisors
Skibniewski, Purdue University.
Subject Area
Civil engineering|Management|Systems design
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.