< Back to previous page

Project

Design of Real-Time Control Strategies for Combined Sewer Networks

Real-time control (RTC) of sewer systems is generally recognised as a means for the ecologically and economically efficient upgrading of existing urban drainage systems. Many new, ever more complex control algorithms have been developed by the scientific community over the years. Their performance is frequently only tested for the single case study for which they have been designed. It thus remains unclear whether any of these newly developed algorithms hold potential for implementation into another system. In addition, a comparison between different control strategies is not usually documented for previous studies. To overcome this problem, it is sometimes advocated to use benchmark systems for an objective comparison between different control strategies. Such comparisons would, however, not necessarily allow for extrapolation of the results to other case studies. To help wastewater operators circumvent the numerous difficulties and pitfalls of the implementation of an RTC strategy to sewer systems, guidelines and simple, yet effective, tools for demonstration and preliminary screening are provided by the scientific community. In spite of these efforts, it appears that many wastewater companies are not yet convinced to embark on the seemingly tedious and time-consuming task of implementing RTC, which still bears the risk of failure due to limited experience of the modeller, manually introduced implementation errors or insufficient transferability of the applied control strategy. These are major reasons for which the practical implementation of RTC strategies in full-scale systems today is still met with reluctance by wastewater operators.

The objective of this dissertation is to contribute to overcoming these problems by automating the major steps required for the design of RTC strategies: model simplification, control location choice and control algorithm implementation.

A coarse preliminary systems analysis using simplified models is generally considered the first step of systems analysis in the course of an RTC study. It allows for the analysis of a large number of model scenarios in a short period of time even when simulating long evaluation periods. Both the automated reduction of detailed hydrodynamic sewer models to skeletonised models and the computer-aided simplification to calibrated conceptual models are covered in the context of this thesis. These two processes make use of a highly detailed, hydrodynamic sewer model as a basis to provide the simplified models with data for parametrisation and, if necessary, calibration. The resulting models prove to be highly accurate and reliable. Because the reduced hydrodynamic models still contain major trunk sewers, they could form a basis for analyses that require a high spatial resolution and high accuracy, as e.g. for flood modelling or for the offline optimisation of rule-based RTC strategies. The conceptual models created here only provide simulation output for a limited number of locations, but at higher speeds. They hence appear more appropriate for application as a prediction model in the scope of model-based predictive control (MPC).

The use of simplified models for the preliminary analysis of RTC scenarios, however, is afflicted by an inherent problem: the modeller’s decisions taken during the simplification of the detailed models may have a significant influence on the modelled performance of the RTC strategy. This is especially true for the selection of control locations when building conceptual models. Sensitivity analysis for both detailed and simplified models reveals that location choice has more influence on the performance of an RTC strategy than the parametrisation of the control algorithm or simulated monitoring inaccuracies. More complex sensitivity analysis methods thereby deliver the same results as simple screening methods.

 

The computer-aided methodology developed here for the design of RTC strategies thus makes use of the fully detailed hydrodynamic models as the basis for the choice of suitable control locations. These are defined based on a systematic and reproducible approach, rather than to entrust this task exclusively to the modeller’s experience as appears to be common practice for documented scientific case studies as well as real-life implementations. Next to the use of only existing control locations (e.g. pumping stations) or locations with limited discharge capacity (as is frequently done when relying on the use of simplified modelling approaches), a methodology has also been developed to identify control locations that allow for activation of the maximum storage volume in the system. These different control location choices are then combined with generically applicable control concepts, such as the frequently applied rule-based ‘equal filling degree’ control. This type of control is chosen because the application to a preliminary test case proved highly favourable in terms of efficiency and the potential to be extended into an integrated control of the sewer system and wastewater treatment plant. In addition to these automatically created scenarios, the application of MPC is evaluated here to be able to benchmark the results of the automatically created control algorithms.

The methodology was tested for five sewer networks in Flanders, Belgium, and proved applicable to all five test cases. With a modelled total combined sewer overflow volume reduction of 20 % to 50 %, the created strategies showed good to very good performance for the tested catchments. While volume-based control location selection showed the best overall results, no ‘best’ control algorithm was found. Also the use of MPC did not yield significantly better results than rule-based control. The results show that the automated design of RTC strategies for sewer systems is both feasible and desirable. In its current form, the methodology for the design of RTC strategies for sewer systems has been successfully applied in the course of the analysis of 10 real-life case studies at Aquafin and will form an integral part of future projects for the roll-out of RTC for sewer systems throughout Flanders.

Date:28 Sep 2009 →  10 Oct 2019
Keywords:RTC (real-time control) of sewer systems
Disciplines:Construction engineering, Earthquake engineering, Geotechnical and environmental engineering, Water engineering, Wind engineering
Project type:PhD project