Life-cycle assessment of R.C. bridge components based on cluster analysis and stochastic process

This article proposes an approach to assess the life-cycle of reinforced concrete bridge components by applying a cluster algorithm and a stochastic model for damage evolution. The k-means algorithm is used to identify families of bridge components that deteriorate at similar rates. A measure of performance, i.e., silhouette width, supports the choice of the optimal number of clusters. Once the cluster model is defined, a gamma process is fitted to the data on the evolution of the conditions that belong to each family. By simulating the gamma process, the cumulative distribution function of time to failure is calculated for each cluster of components. The procedure applies to reinforced concrete bridge components in Switzerland, whose inspection and maintenance data is collected in the KUBA-DB database. This approach ensures that the expected service life of bridge components can be predicted with limited uncertainty.