Optimal design-for-control of self-cleaning water distribution networks using a convex multi-start algorithm

Area of Science:

• Environmental Engineering
• Water Resource Management
• Optimization Theory

Background:

• Discolouration in water distribution networks (WDNs) poses a significant water quality challenge.
• Current WDN control strategies primarily focus on pressure and leakage management, neglecting self-cleaning velocities.
• Maximizing self-cleaning capacity (SCC) is essential for maintaining water quality and preventing pipe discolouration.

Purpose of the Study:

• To develop an optimal design-for-control strategy for WDNs to maximize self-cleaning capacity (SCC).
• To jointly optimize the placement and operational settings of pressure control and automatic flushing valves.
• To address the challenges of solving a complex nonconvex mixed integer nonlinear programming (MINLP) problem.

Main Methods:

• Formulation of a nonconvex MINLP optimization problem for WDN control.
• Development of a heuristic algorithm combining convex relaxations, randomization, and a multi-start strategy.
• Evaluation of the algorithm on diverse case study networks, including a large-scale operational UK network.

Main Results:

• The proposed convex multi-start algorithm demonstrated superior robustness and solution quality compared to a genetic algorithm.
• The algorithm successfully computed feasible solutions for all design-for-control experiments across various network complexities.
• The multi-start strategy proved to be a fast and scalable method for solving the nonlinear SCC control problem.

Conclusions:

• The developed method effectively optimizes WDNs for enhanced self-cleaning capacity, improving water quality.
• This approach extends the control capabilities of dynamically adaptive networks for better water management.
• The findings offer a practical and efficient solution for improving water quality in existing and new WDNs.