Risiko-gebaseerde besluitondersteuning in siviele ingenieurswese: 'n metodologiese benadering tot verbeterende inligtingversameling en benutting

[摘要] ENGLISH ABSTRACT:Decisions affecting construction projects are often characterised by uncertainty.One of the sources of this uncertainty is the unit costs used in detailcost estimates. Analytical techniques are available to model these uncertainties,but information is required to quantify it. Historical data is the preferred sourceof information, but due to unavailability and unreliability it hampers the applicationof the analytical techniques. Changes in the construction environmentnecessitate the improvement of data sources and the utilisation thereof in astructured, comprehensive and integrated manner.The purpose of this study is to determine how decision-making can be improvedby enhanced information capturing and usage of relevant data forimproved cost calculations. This is achieved by investigating the various componentsof an information system, viz. dataflow, data source, data application,and shown how an integrated development of each of these components withprocessmodelling, database development and quantitative risk analysis canlead to improved decision-making. This is achieved by the conceptual redevelopmentand development of information systems for decision support in twodifferent construction environments, viz. road and dam construction, respectively.During the research it was found that processmodelling can contribute to theimprovement of dataflow in the fragmented construction environment, but thatshortcomings exist in the processmodelling tools. A method, based on set andgraph theory, was developed to address it. It was also found that databasedevelopment provides the opportunity to improve the quality of data stored inelectronic format. A comprehensive database model to store all informationrelevant to construction costs, specifications and legal documents with therequired error checking mechanisms was developed. The benefit of modellingthe global uncertainty of project cost estimates with probabilistic techniquessuch as Monte Carlo simulation and the Limit State Cost Function, whilst includingthe effects of correlation, is also shown.The information gained from historical data-analysis, besides the applicationin the probabilistic techniques, was used to identify properties such as impact,variability and correlation. It was found in the study that a shortage of descriptiveinformation, whether it be at project or item level, is predominately thesource of variability in unit cost data after errors have been removed. In order toidentify the information that would contribute to the lowering of variability amethod was developed by using influence factors and unit cost correlationmechanisms to identify the most influential factors for data capturing. This led tothe development of an ideal data-application model in which all the relevantdata used in the different cost-estimation phases, viz. long term, conceptual anddetail, would be captured and used.Based on what was found, a generic information system development modelwas proposed that indicated the interaction of all the aspects that were investigated.Two of the model's foremost properties are its self-triggering process andincremental development (improvement). On top of this development model adevelopment process was designed that can be used for the structured, comprehensiveand integrated development/redevelopment of an informationsystem for cost calculation.

[发布机构] Stellenbosch University