Mixed Poisson Deterioration Model for Bridge Management

In this study, the authors develop a model of the generating process on peeling/falling (P/F) of protective covering concrete as a counting process. The model is able to consider a time-independent factor (random event) and a time-dependent factor (deterioration event) of P/F process. In detail, the random event is presented by the homogeneous Poisson process, the deterioration event is presented by the non-homogeneous Poisson process where arrival rates are changed with time. Then, a mixed Poisson deterioration process is formulated by both the homogeneous and the non-homogeneous Poisson process. Lastly, the appropriateness of the proposed model is discussed empirically, by applying it to actual inspection data. By estimating the mixed Poisson deterioration model, it is possible to predict the generating process on P/L of protective covering concrete. The proposed model is able to represent three things at the same time. Firstly, the number of P/F concrete increases if an inspection interval increases. Secondly, the model formulates the deterioration event that the number of P/F concrete increases with accelerating speed if an inspection interval increases, which is unable to be presented by homogeneous Poisson model. Lastly, the model stands for the random event that is generated even if the service life is small, which is unable to be presented by non-homogeneous Poisson model. Moreover, the model makes possible to classify observed data stochastically by the generating factor which is the random event or the deterioration event.