Stochastic System Identification of Sewer‐Flow Models

Abstract

Modeling of sewer flow and quality is essential for real‐time control of sewer systems and minimization of combined sewer overflows (CSO). Studies have shown that CSOs contribute substantially to the overall pollution loads discharged into surface‐water bodies. A modeling strategy based on system identification analysis of single‐input, single‐output stochastic processes is presented in this paper. An application example is illustrated using the flow and rainfall time series observed in the collection system discharging to the treatment plant of Fusina (Venice, Italy). The advantages of this type of modeling strategy, compared with a traditional deterministic model, are the relative simplicity of the model, its requirement for a minimal amount of investigation to describe the physical system, and the possibility of continuous updates of the model as the system data base expands. Furthermore, stochastic models are able to reflect truly the dynamic features of the system under investigation and allow the prediction of its future behavior with a specified degree of confidence.