Bayesian Networks as a Support to Decision-Making When Evaluating Maintenance Scenarios

Contemporary architecture is characterised by an ever-greater morphological, functional and technological complexity. This requires new skills to effectively implement a process capable of ensuring an uninterrupted link between the concept of a building and its construction. The research described here focuses on implementing a tool for evaluating levels of maintenance quality applied at various stages in a project. In methodological terms, the research was conducted with the intention of constructing a system capable of managing the large number of variables involved in the design of a work of architecture from the earliest phases. The proposed digital model was defined through the construction and implementation of a Bayesian Network. The duration and maintainability of building components—generally expressed in statistical forecasts as expected duration in relation to operating conditions and as estimated costs and times of maintenance—were jointly evaluated using a synthetic index to generate a true overall rating of the various aspects that play a part in defining the quality of maintenance. The effectiveness of this decision-making support was tested in evaluations of complex architectural projects. Specifically, this text presents the results of an analysis of different maintenance scenarios for the ING Group Headquarters.