A Possibility-based Approach for the Assessment of the Performance of Civil Engineering Works

To best design and manage civil works, it is of main importance to assess their performance. Performance is defined as the ability of a civil work to perform the functions for which it has been designed. In the dam domain, the assessment of the performance is one of the key items of the risk management; indeed a dam failure represents a major risk (low occurrence and high consequences). In the building domain the determination of strategies of design and maintenance is highly concerned with the performance assessment which allows best fitting the user waits and best managing the associated costs. The performance threshold of a work, that is to say the limit under which a work no more fulfils its required functions, integrates at the same time technical, economical and social criteria; only the technical aspects are taken into account here. The objectives of use of the performance assessment may differ among the considered application domains; however the approach of the performance assessment may be all in all similar. The objectives and the performance assessment approach are presented in this paper for two application domains: (1) dams and (2) buildings with special focus on building components. Moreover, the similarities and differences of these two applications are analysed. This comparative analysis aims at underlining the approach items that can be transposed to other applications in the civil engineering domain. The assessment of the performance of civil engineering works implies the development of models. Usually, various types of data are available considering these works: visual observations, laboratory experimentations, in-situ tests, monitoring data, simulations, probabilities, statistics, and expert judgements. In our approach, we aim at considering all these types of data in the performance models. The common approach for the two applications comprises four phases: (1) collection and analysis of the quality of rough data, (2) modelling of rough data that allows their use as input data of the performance model, (3) unification and aggregation of rough data and (4) extraction of a performance value. The analysis of the similarities and differences of the methodology used in the two applications indicates that this methodology should be able to be transposed to other applications in the civil engineering domain. Indeed, the four phases are similar for the both applications and various methods and tools are available to answer to specific issues of each applicative domain