Modeling the spatial behavior of infrastructure condition

• A new understanding of the spatial behavior of infrastructure condition is presented and a methodology for identifying this behavior is developed. Based on two deterioration mechanisms, causal and interactive, a nonstationary stochastic spatial model with a piecewise constant mean function is proposed. Based on this model, a methodology founded on nonparametric cluster analysis and dynamic programming is developed to identify the optimal spatial regions, referred to as fields, within which behavior is uniform and, consequently, condition can be estimated accurately. Validation using detailed distress data from three roadway facilities, each 15 km long, is presented. The value of the model is also demonstrated by comparing it against existing methods for characterizing infrastructure condition over space.

Nyckelord

infrastructure management

condition assessment

spatial modeling

stochastic processes

cluster analysis

Publikations- och innehållstyp

ref (ämneskategori)

art (ämneskategori)