Pipe Roughness Identification of Water Distribution Networks: A Tensor Method

TL;DR

This paper introduces a tensor-based numerical method to improve the accuracy and robustness of pipe roughness identification in water distribution networks, addressing the challenges of solving nonlinear algebraic equations in real-world scenarios.

Contribution

It extends the Newton-Raphson method with second-order derivatives, providing a more efficient and robust solution technique for pipe roughness identification.

Findings

Enhanced method outperforms conventional Newton-Raphson in simulations

Mathematical findings enable compact formulation and improved robustness

Algorithm demonstrates increased efficiency in identifying pipe roughness

Abstract

The identification of pipe roughnesses in a water distribution network is formulated as nonlinear system of algebraic equations which turns out to be demanding to solve under real-world circumstances. This paper proposes an enhanced technique to numerically solve this identification problem, extending the conventional Newton-Raphson approach with second-order derivatives in the determination of the search direction. Enabled through some interesting mathematical findings, the resulting formulation can be represented compactly and thus facilitates the development of an efficient and more robust solving-technique. Algorithms on the basis of this more enhanced solving method are then compared to a customized Newton-Raphson approach in simulation examples.