Premium
Autocalibration of a water distribution model for water quality parameters using GA
Author(s) -
Munavalli G.R.,
Kumar M.S. Mohan
Publication year - 2006
Publication title -
journal ‐ american water works association
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.466
H-Index - 74
eISSN - 1551-8833
pISSN - 0003-150X
DOI - 10.1002/j.1551-8833.2006.tb07758.x
Subject(s) - genetic algorithm , calibration , inverse problem , inverse , mathematical optimization , field (mathematics) , estimation theory , computer science , distribution (mathematics) , water quality , mathematics , biological system , algorithm , statistics , mathematical analysis , ecology , geometry , pure mathematics , biology
Water quality simulation models are required to predict spatio‐temporal variation of chlorine (Cl) residuals throughout a distribution system. However, the propagation and level of Cl within a distribution system are governed by the various combinations of complex bulk‐flow and pipe‐wall reaction kinetics. As a result, the reaction parameters involved in these kinetic expressions directly affect Cl residuals. Field determination of these parameters, particularly those related to wall reactions, is difficult, and calibration against field measurements is required. In this article an inverse problem for parameter estimation was created in terms of an unconstrained optimization problem that minimizes the sum of observed and computed Cl concentrations in a least‐square sense. The inverse problem was solved using a simulation‐optimization procedure consisting of a water quality simulation model and an improved genetic algorithm (GA) technique, advancements of which include niching, creep mutation, and elitism operations. Results show that the GA technique is efficient for a system with a number of unknown parameters.