Stochastic Modeling of Combined Sewer Flows

Modeling of sewer flow and quality is essential for real-time control (RTC) of sewer system and minimization of combined sewer overflows (CSOs). In this paper, the Box-Jenkins type transfer function is applied to the daily rainfall and sewer flow data to incorporate the stochastic nature of the rainfall and flow series. Two system identification techniques, different from the traditional Box-Jenkins', approach, are developed to identify sewer flow models. Both techniques are illustrated through their application of modeling sewer flow with actual time series of flow and rainfall from a combined sewer system. The onestep ahead predicted flow values are found to be in good agreement with the measured values.