as part of an experimental thesis that led to the implementation of a Decision Support System. The objective of the work was to implement a tool capable of evaluating-in relation to the choices concerning the morphology of the building, the construction technologies, the materials and the design of the architectural elements-the levels of maintenance quality implemented in the various phases of the project, from the first phases, in which few relevant decisions are made, to the executive phase characterized by a multiplicity of choices. The aim was to construct a tool in which the reliability of the evaluations was related to the quantity and quality of the data that feeds the decision-making process, but which is also able to evaluate preliminary decisions based on the elements of choice that characterize the first phases of the project. The conceptual model has been defined through the construction and implementation of a Bayesian Network or a graphical system of probabilistic inference able to represent the set of stochastic variables and their conditional dependencies through the use of a direct acyclic graph. Through the interrogation of the network it is therefore possible to evaluate through the expression of a synthetic index, a real overall rating of the different aspects that contribute to define the maintenance quality. The use of Bayesian Networks, in the light of the analyses carried out on an experimental basis-exemplified here on the case study of ING Groupe headquarters-for the ability to control a multitude of factors linked to the durability of materials, the morphology of systems and ease of intervention, seems capable of generating useful, effective and expandable tools to support the design decision-making process.

No abstract

Contemporary architecture is characterized 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.