Service-life assessment of building components: application of evidence theory
This paper deals with the assessment of the service life of in-service building components subjected to known environmental and usage conditions. This assessment is complex because of two primary features. First, the assessment has to be carried within a multiscale context: a geometric scale that ranges from the material or elemental to building scale; a range in the complexity of the degradation (phenomena that varies from a single phenomenon to the consideration of several degradation scenarios); a range of possible performance requirements, from one function to several; and consideration, as well, to the time over which the process is carried out that may span from the design stage to that of management and repair. Second, this assessment must also take into consideration the availability and features of service-life data that by nature is heterogeneous, imprecise, uncertain, and incomplete. In this context, a comprehensive methodology is developed using all available data on service life derived from existing methods of service-life assessment of materials, elements or building components. Such data may, for example, be extracted from fundamental studies on durability, accelerated short-term exposure tests, statistical methods, factorial methods, feedback from practice, or expert opinion or other sources. The main stages of this methodology are: (i) identification of all possible degradation scenarios provided by failure mode and effects analysis (FMEA); (ii) collection of all available service-life data (SL-data) associated with these degradation scenarios, transformation of this data into a fuzzy-set format, and assessment of its quality; (iii) processing of unification of data and aggregation of data; and (iv) assessment of the service life of building components. The case study of a window unit allows for: (i) service-life assessment of a building component to be processed by unification of data and aggregation of data and (ii) a conclusion to be deduced.