A decision support system for infrastructure management

The goal of this research was to create a system for the management of infrastructure information that facilitates engineering analysis and decision making. To this end, this research developed a hybrid database schema, "Relational Object" technology. Prototyped in DAESSIM (the Decision, Analysis, and Engineering Support System for Infrastructure Management), Relational Object technology provides an effective tool for the management and integration of information used in engineering analyses. Designed for the risk analysis of infrastructure structures and fleets, DAESSIM may be customized to meet users' needs. That ability was demonstrated with PIMS (the Platform Information Management System), a version of DAESSIM. Details of the sample evaluation objects included in PIMS were examined in a subjective assessment of the structural risk of an oil platform due to storm waves, and an objective assessment of the reliability of a platform due to storm waves. Also developed was a sample 4-level screening system for integration with DAESSIM. System test levels progress from the "quick and dirty" to the expensive and complex. The screening system uses common sense and economics to move to another structure within a fleet, iterate a present type of analysis on the present structure using different assumptions, or decide whether to accept a "safe" result for a structure. The latter was analyzed in detail with a decision tree model. It was found to be both possible and necessary to calculate, rather than estimate, conditional probabilities for a multi-level screening system. The ability to skip from an i-level test to an i + 2 or i + 3 level test was found to be critical in the decision model, as was the need ta account for the time lag between screens. The assumption that screening system tests should decrease in conservatism with increasing complexity and cost was found to be incorrect in one case. Given the example data used for demonstrating the model, the most influential factors upon the decision whether or not to continue testing after receiving a Level One "safe" result were found to be failure cost, the quality of a Level Four test, and the number of years until the next screening.