Hydraulic modeling and evolutionary optimization for enhanced real-time decision support of combined sewer overflows

Zimmer, Andrea

Permalink

https://hdl.handle.net/2142/46814 Copy

Description

Title

Hydraulic modeling and evolutionary optimization for enhanced real-time decision support of combined sewer overflows

Author(s)

Zimmer, Andrea

Issue Date

2014-01-16T18:17:00Z

Director of Research (if dissertation) or Advisor (if thesis)

Minsker, Barbara S.

Doctoral Committee Chair(s)

Minsker, Barbara S.

Committee Member(s)

• Schmidt, Arthur R.
• Ostfeld, Avi
• Cai, Ximing
• Valocchi, Albert J.

Department of Study

Civil and Environmental Engineering

Discipline

Civil Engineering

Degree Granting Institution

University of Illinois at Urbana-Champaign

Degree Name

Ph.D.

Degree Level

Dissertation

Keyword(s)

• Real-Time Control
• Combined Sewer
• Optimization
• Hydraulic Model

Abstract

Operational strategies to mitigate Combined Sewer Overflows (CSOs) in older urban areas may be enhanced through real-time decision support provided to sewer operators. During severe rainfall events, real-time hydraulic simulations coupled with control algorithms can explore a large number of potential changes to control procedures at short time intervals to provide dynamic feedback and optimization. Calculations for water level and flow based on dynamic wave modeling may not complete calculations for the entire sewer system in the required real-time decision intervals, especially when multiple simulations must be tested within an optimization routine. In an effort to reduce computation time at each decision interval, this work couples a computationally-efficient sewer hydraulic model with variations to genetic algorithm (GA) optimization. An offline approximation of the Saint Venant equations is proposed for CSO prediction. Pre-calculated tabulations of the mass and momentum equations are utilized to allow online interpolation, which result in faster hydraulic computations when compared to standard industry software while preserving high levels of accuracy. This model also extends backwater profiles, which traditionally account for only subcritical, open channel flows, to include pressurized flows that are key for CSO conditions. Incorporation of the Darcy-Weisbach equation eliminates potential lookup table discontinuities between open channel and pressurized conditions. Iteration errors caused by graphical depiction of supercritical flows are eliminated by allowing the solution to proceed separately upstream and downstream from the governing critical water surface. Dynamic adjustment of operating strategies can potentially reduce CSOs beyond the mitigation offered by management routines that remain static despite variations in sewer water levels. A suite of model predictive control (MPC) genetic algorithms are developed and tested offline to explore their value for reducing CSOs during real-time use in a deep-tunnel sewer system. MPC approaches include the micro-GA, the probability-based compact GA, and domain-specific GA methods that reduce the number of decision variable values analyzed within the hydraulic model, thus reducing algorithm search space. Minimum fitness and constraint values achieved by all new GA approaches, as well as computational times required to reach the minimum values, are compared to large population sizes with long convergence times. Since stationary management practices may not account for varying costs at CSO locations and electricity rate changes in the summer and winter, the sensitivity of the results is evaluated for variable seasonal and diurnal CSO penalty costs and electricity-related system maintenance costs, as well as different sluice gate constraint levels. Optimization results for a subset of the Chicago combined sewer system indicate that genetic algorithm variations with a coarse decision variable representation, eventually transitioning to the entire range of decision variable values, are best suited to address the CSO control problem. Although diversity-enhancing algorithms evaluate a larger search space and exhibit shorter convergence times, these representations do not reach minimum fitness and constraint values. The most efficient GA types are used to test CSO sensitivity to energy costs, CSO penalties, and pressurization constraint values. The results show that CSO volumes are highly dependent on the tunnel pressurization constraint, with reductions of 13% to 77% possible with less conservative operational strategies. Maintaining low risk of excessive tunnel pressurization therefore requires strategies other than sluice gate and pump manipulation to significantly reduce CSOs, regardless of the energy costs or CSO penalties. One alternative strategy is to replace small-diameter conduits that may be restricting flows into the deep tunnel system. When compared to CSO reduction through real-time optimization, conduit replacement proves expensive due to excavation costs and requires a prohibitively long time to recover construction expenses. Lastly, the geographical scope of the optimization is extended beyond the initial sewer system to evaluate the possibilities of reducing CSOs at larger spatial scales. The results indicate that pump and dropshaft CSOs can be reduced by 14 percent with optimization over a larger spatial extent, without excessive pressurization.

Graduation Semester

2013-12

Permalink

http://hdl.handle.net/2142/46814

Copyright and License Information

Copyright 2013 Andrea Zimmer

Owning Collections