Life-cycle sustainability design of post-tensioned box-girder bridge obtained by metamodel-assisted optimization and decision-making under uncertainty

Resumo

Currently, there is a trend towards sustainability, especially in developed countries, where the concerns of society about environmental degradation and social problems have increased. Following this trend, the construction sector is one of the most influential sectors due to its high economic, environmental, and social impacts. At the same time, there is an increase in the demand for transport, which drives a need to develop and maintain the necessary infrastructure for this purpose. Taking all these factors into account, bridges become a key structure and therefore assessment of sustainability throughout their whole life-cycle is essential. The main objective of this thesis is to propose a methodology that allows assessment of the sustainability of a bridge under uncertain initial conditions (subjectivity of the decision-maker or variability of initial parameters) and optimization of the design to obtain a robust optimal bridge. To this end, an extensive bibliographic review of all the works that perform assessments of the sustainability of bridges through the valuation of criteria related to their main pillars (economic, environmental, or social) has been carried out. In this review, it has been observed that the most comprehensive way to evaluate the environmental and social pillars is through the use of life-cycle impact assessment methods. These methods allow sustainability assessment to be performed for the whole life-cycle of the bridge. This process provides valuable information to decision-makers for the assessment and selection of the most sustainable bridge. However, the decision-makers' subjective assessments of the relative importance of the criteria influence the final assessment of sustainability. For this reason, it is necessary to create a methodology that reduces the associated uncertainty and seeks robust solutions according to the opinion of decision-makers. In addition, for bridges, the design and decision-making are conditioned by the initially defined parameters. This leads to solutions that may be sensitive to small changes in these initial conditions. A robust optimal design makes it possible to obtain optimal solutions and structurally stable designs under variations of the initial conditions as well as sustainable designs that are not influenced by the preferences of the stakeholders who are part of the decision-making process. Thus, obtaining a robust optimal design becomes a probabilistic optimization process that has a high computational cost. For this reason, the use of metamodels has been integrated into the proposed methodology. Specifically, Latin hypercube sampling is used for the definition of the initial sample and a kriging model is used for the definition of the mathematical approximation. In this way, kriging-based heuristic optimization reduces the computational cost by more than 90% with respect to conventional heuristic optimization while obtaining very similar results. This thesis provides, first of all, an extensive bibliographic review of both the criteria used for the assessment of sustainability of bridges and the different methods of life-cycle impact assessment to obtain a complete profile of the environmental and social pillars. Subsequently, a methodology is defined for the full assessment of sustainability, using life-cycle impact assessment methods. Likewise, an approach is proposed that makes it possible to obtain structures with little influence from the structural parameters, as well as from the preferences of the different decision-makers regarding the sustainability criteria. The methodology provided in this thesis is applicable to any other type of structure.