Comparative Analysis of Service Life Prediction Methods

Abstract

In this chapter, a comparative analysis of the different proposed models is performed. First the predictive ability of each model is evaluated through the comparison between the values obtained by each model and the data collected in field work, corresponding to the real degradation condition of the façades analysed. In a second step of the analysis, the sensitivity and specificity of each model is evaluated based on ROC curves, to analyse the accuracy of each model to correctly classify the case studies that may have reached the end of their service life. Once the predictive accuracy and the classification capabilities of each model are evaluated, a multi-criteria analysis is performed. This multi-criteria analysis encompasses different criteria that should be taken into account by different stakeholders to choose the best model for a specific application. A sensitivity analysis is performed in order to analyse the best model for distinct decision-making profiles. Based on these analyses, some recommendations are made, referring the advantages and limitations of the models analysed and thus allowing a more rational and informed selection of a service life prediction model as far as the purpose of the model and the profile of the planner and the perspective of the user are concerned.