Data assimilation in hydrodynamic models for system-wide soft sensing and sensor validation for urban drainage tunnels

Abstract

Tunnels are increasingly used worldwide to expand the capacity of urban drainage systems, but they are difficult to monitor with sensors alone. This study enables soft sensing of urban drainage tunnels by assimilating water level observations into an ensemble of hydrodynamic models. Ensemble-based data assimilation is suitable for non-linear models and provides useful uncertainty estimates. To limit the computational cost, our proposed scheme restricts the assimilation and ensemble implementation to the tunnel and represents the surrounding drainage system deterministically. We applied the scheme to a combined sewer overflow tunnel in Copenhagen, Denmark, with two sensors 3.4 km apart. The downstream observations were assimilated, while those upstream were used for validation. The scheme was tuned using a high-intensity event and validated with a low-intensity one. In a third event, the scheme was able to provide soft sensing as well as identify errors in the upstream sensor with high confidence.

Soft sensing of water depth in combined sewers using LSTM neural networks with missing observations

“…The CSO is part of a large combined drainage network that serves the westernmost area of the city and discharges in the local treatment plant (Figure 2,left). A detailed description of the study area is given by Palmitessa et al, 2020. The CSO chamber is designed to discharge excess inflows to a storage tunnel via a fixed weir set 0.99 m above the chamber invert (Figure 2, right).…”

Control theory-based data assimilation for open channel hydraulic models: tuning PID controllers using multi-objective optimization

“…This method is successfully applied for improved flood forecasting on large-scale domains (Madsen et al 2003;Neal et al 2007;Li et al 2014;Barthélémy et al 2017;Jafarzadegan et al 2021). Along with these applications, EnKF was also applied for improved urban flood forecasting and urban drainage system management (Lund et al 2019;Kim et al 2021;Palmitessa et al 2021). In most of the hydrological-hydraulic applications of DA, measured water levels from monitoring systems are used to update the model (Romanowicz et al 2006;Hostache et al 2010;Jean-Baptiste et al 2011;Rakovec et al 2012) or streamflow observations (Thirel et al 2010;Dumedah & Coulibaly 2014;Randrianasolo et al 2014;Sun et al 2015).…”

Probabilistic digital twin of water treatment facilities

“…Data assimilation is an advanced technique whereby real-time observations are assimilated with predictions from models established with historical data in order to achieve an improved estimate of the evolving states of a system [20,21]. With the continuous advances of pervasive sensing and monitoring of the environment [22,23], it can be foreseen that data assimilation will become even more popular going forward [24].…”

“…However, the ML updating can be very timeconsuming in some cases. For example, Palmitessa, et al [21] developed a LSTM model for the water level predictions of the pumping control inside a underground tunnel water system in Copenhagen, Demark. The training and testing of their LSTM model (based on minutes data over a 5-month period) required an extensive duration using a fast workstation.…”

“…In other words, 𝑋, 𝑌 and 𝑍 have 8, 64 and 512 elements, respectively. A 4 th -order Runge-Kutta solver with time step Δ𝑡 = 0.005 is employed to solve Equations ( 19)- (21). The model time unit (MTU), which is equal to 200Δ𝑡, is used in the following discussion.…”