The use of nested sampling for prediction of infrastructure degradation under uncertainty

Because of the competing demands for scarce resources (funds, manpower, etc) national road owners are required to monitor the condition and performance of infrastructure elements through an effective inspection and assessment regime as part of an overall asset management strategy, the primary aim being to keep the asset in service at minimum cost. A considerable amount of information is then already available through existing databases and other information sources. Various analyzes have been carried out to identify the different forms of deterioration affecting infrastructures, to investigate the parameters controlling their susceptibility to, and rate of, deterioration. This paper proposes such an approach by building a transition matrix directly from the condition scores. The Markov assumption is used stating that the condition of a facility at one inspection only depends on the condition at the previous inspection. With this assumption, the present score is the only one which is taken into account to determine the future of the facility. The objective is then to combine nested sampling with a Markov-based estimation of the condition rating of infrastructure elements to put some confidence bounds on Markov transition matrices, and ultimately on corresponding maintenance costs.