Efficient preconditioned iterative methods for hydraulic simulation of large scale water distribution networks

Abstract

In this paper, we consider the use of an efficient null space algorithm for hydraulic analysis that employs preconditioned conjugate gradient (PCG) methods for solving the Newton linear equations. Since large water network models are inherently badly conditioned, a Jacobian regularization is employed to improve the condition number to some degree, this resulting in an inexact Newton method whose analyses is presented. Based on this analysis, constraint preconditioners are used to improve the condition number further for more efficient use of CG solvers. Operational networks are used to study the computational properties of the various approaches.