Quantum Simulation and Optimization of Water Distribution Networks

TL;DR

This paper explores the feasibility of using quantum computing algorithms to improve water distribution network modeling, testing hybrid and full quantum approaches on small networks to enhance computational efficiency.

Contribution

It introduces quantum algorithms as potential replacements or enhancements for classical methods in water network modeling, demonstrating initial feasibility with small networks.

Findings

Hybrid VQLS improved modeling on small networks

Quantum annealing used for optimal design in small networks

Quantum approaches show promise for future large-scale applications

Abstract

To compute models for Water Distribution Networks (WDN), a large system of non-linear equations needs to be solved. The hallmark algorithm for computing these models is the Newton-Raphson Global Gradient Algorithm (NR-GGA), which solves these systems iteratively. Even so, large networks can take multiple days to model and the complexity of networks is only expected to increase in the future. It is therefore important to explore different algorithms using innovative technologies, to improve the tractability of modelling large networks. Quantum computing is such an innovative technology that is still in its early stages of development, paired with a different computational paradigm. In this research we have determined the feasibility of using quantum computing algorithms as a subroutine of NR-GGA and alternatively for replacing NR-GGA with a quantum algorithm in its entirety. Calculations were run on emulators of gate-based quantum computers and using simulated annealing while models were tested on small 2-loop networks of 7 nodes. To improve NR-GGA, three different quantum subroutines were used: the hybrid Variational Quantum Linear Solver (VQLS) showed the best results on this small dataset. To replace NR-GGA by a quantum algorithm in its entirety, we also used a Quantum Annealing-based approach for simulating higher-order polynomials to determine the Least-Cost Design (LCD) of a small 3 node network.