Estimating inflow to a combined sewer overflow structure with storage tank in real time: Evaluation of different approaches

Abstract

The performance assessment of storage tanks and combined sewer overflow (CSO) structures in sewer systems requires knowledge of the total inflow from the catchment during rainfall events. Many structures are, however, only equipped with sensors to measure water level and/or outflows. Based on the geometry of the tank, expressed as a level-storage relationship, inflow can be calculated from these data using a simple conceptual storage model. This paper compares a deterministic and a Bayesian approach for estimating the inflow to a CSO structure from measurements of outflows and water level. The Bayesian approach clearly outperforms the deterministic estimation which is very sensitive to measurement errors. Although computationally more demanding, the use of a simple linear storage model allows the online application of the Bayesian approach to repeatedly estimate inflow in short time intervals of a few minutes. The method could thus be used as an online software sensor for inflow to storage structures in sewer systems.