Utilizing Matrix Completion for Simulation and Optimization of Water Distribution Networks

Simulation of water distribution networks (WDNs) constitutes a key element for the planning and management of water supply systems. The literature presents different formulations of heads-flows equations that differ in terms of dimensionality, computational cost, and solution accuracy. Whereas this problem has been the subject of active research in the past, in the last decades a state of stagnation was reached and no new formulations were introduced. In this study, we propose a novel formulation that utilizes a matrix completion technique to construct a reduced-size nonlinear system of equations that guarantees both mass and energy conservation. The advantages of the proposed method are demonstrated in simulation and optimization settings. In the former, the method demonstrates improved scalability and accuracy as compared with other widely known formulations. In the latter, the new formulation leads to compact optimization problems that result in better performance in terms of the objective function, run time, and rate of successful optimization runs.

Automated summary

Simulation and optimization of water distribution networks have long been a focus of research, with different formulations of heads-flows equations presenting varying degrees of dimensionality, cost, and accuracy. This study introduces a novel approach using matrix completion to construct a reduced-size nonlinear system that ensures conservation of mass and energy. The method demonstrates improved scalability, accuracy, and performance in simulation and optimization scenarios.