Parameter Optimization in Groundwater using Proper Orthogonal Decomposition as a Reduced Modeling Technique

Abstract : This paper deals with different approaches of applying Proper Orthogonal Decomposition in the field of groundwater flow, specifically the Richards equation, which is a convection-diffusion partial differential equation governing the behaviour of unsaturated fluid flow through a porous medium. The motivation for this research is the need to reduce computational complexity in inverse modelling studies, where a significant number of simulations are needed to determine suitable model parameters. Three different methods of implementing Proper Orthogonal Decomposition are explored. The first method is the Petrov-Galerkin method, a method well suited to speeding up linear problems. The second method is a "Hybrid" method, and proposes a linearization of all non-linear functions, building upon the Petrov-Galerkin approach. As such, it is suitable for use in the non-saturated groundwater zone. The third method combines the use of kriging and Proper Orthogonal to create a non-intrusive model for comparison purposes. With these three methods, the suitability of Proper Orthogonal as a reduced modelling method for unsaturated groundwater flow is shown.
Document type :
Conference papers
Liste complète des métadonnées

https://hal-emse.ccsd.cnrs.fr/emse-00740968
Contributor : Florent Breuil <>
Submitted on : Thursday, October 11, 2012 - 2:46:25 PM
Last modification on : Tuesday, January 22, 2019 - 2:06:14 PM

Identifiers

  • HAL Id : emse-00740968, version 1

Citation

John N. Wise, G. Venter, Mireille Batton-Hubert, Eric Touboul. Parameter Optimization in Groundwater using Proper Orthogonal Decomposition as a Reduced Modeling Technique. 5th International Conference from Scientific Computing to Computational Engineering (5th IC-SCCE), Jul 2012, Athène, Greece. ⟨emse-00740968⟩

Share

Metrics

Record views

157