Prediction of sediment build-up from analysis of physical network data

The localization of solid deposits in sewerage networks is basically due to clearly identified structural causes: shallow slope, loops, geometry of pipes, special structures, etc. Our purpose is to analyse the major role played by the structure of the network itself in localisation and development of the deposits it conveys.

The conceptual model proposed determines the theoretical risk of sediment build-up in each section of the sewerage network. It is a set of simple parametrized expert rules (and/or, if/then) based on the fact that silting-up is mainly a conjunction of several causes, and therefore occurs in preferential locations. These rules are then implemented by processing each section in three separate and successive stages: assessment of the different criteria used to estimate the vulnerability of the network, i.e. its likelihood of trapping solids at specific locations; assessment of the different criteria used to estimate hazard, i.e. the probability that there are effectively some solids to be trapped; combination of vulnerability and hazard rankings used logical operators, thereby classifying theoretical sediment build-up risk as very high, high, average, low, very low. The model's performance compared to measurements are: 67% accurate forecasts (zero theory-measurement deviation), 25% average forecasts (1 or 2 classes deviation), 8% incorrect forecasts (3 or 4 classes deviation).