A knowledge-based framework for construction methods selection

The objectives of this thesis are to investigate, formulate, and structure the problem of methods selection, and apply a Knowledge-Based Expert System (KBES) approach. A complete conceptual KBES framework for the methods selection problem is proposed and selected aspects of it were implemented using NExpert Object. Defined hierarchically, a conceptual method frame consists of the following attributes: design element, construction strategy, construction resources, and construction process model. The roles of the KBES control strategy are to first specify a method and then rank it versus others. In so doing, the control strategy is applied at two levels: A preliminary feasibility level, and a detailed feasibility level. The former is used to reduce the number of available methods and rank them for processing by the latter. The preliminary feasibility part constitutes declarative knowledge with high level premises. The detailed feasibility level, develops the attributes of the method. This component contains empirical, analytical, and procedural knowledge that draws on the civil engineering knowledge domains of design, analysis and construction. Because the notion of a frame is a useful way of identifying the attributes of a construction method, a conceptual frame is used throughout to demonstrate the build-up of the method attributes through preliminary, then detailed feasibility. An expert system called CMSA (Construction Methods Selection Assistant) was developed to implement a subset of the proposed solution approach with Cut-and-Cover tunnelling as the problem domain. CMSA, as designed, constitutes a methods selection shell that can be applied to other domains. It entails a solution paradigm of Suggest, Design, Predict, and Analyze operators. CMSA incorporates previous experience (shallow knowledge) as well as algorithmic procedures (deep knowledge). Key elements central to CMSA knowledge base include risk, design technical feasibility, resources compatibility, cost and time performance measures, and regulatory constraints. Allowance is made for modelling project context variables. A range of geotechnical conditions were treated for the example problem domain. The KBES framework proposed for the methods selection problem shows promise for tackling this ill-structured problem, helping to organize site experience, and contributing to productivity improvement.