Preliminary probabilistic prediction of ice/snow accretion on stay cables based on meteorological variables

The scope of the present paper is to present a framework for assessment of the probability of occurrence of ice/snow accretion on bridge cables. The framework utilizes Bayesian Probabilistic Networks and the methodology is illustrated with an example of the cable-stayed Øresund Bridge. The case study focuses on the ice/snow accretion due to the in-cloud icing or precipitation icing mechanisms and includes probabilistic assessments of the meteorological variables influencing the ice/snow accretion on the stay cables. Different probability distribution functions are utilized for the representation of the meteorological variables and evaluated both by goodness-of-fit test and qualitatively. Conditional probability curves are developed
to predict the amount of ice accretion given a set of meteorological conditions using the Gaussian Kernel Smoothing method. The fitted probability distribution functions for the meteorological variables and the conditional ice accretion curves are implemented in a Bayesian Probabilistic Network and the annual average number of ice/snow accretion occurrences is estimated.