Discretization Error Estimation and Mesh Optimization for Plasma Edge Transport Simulations of Nuclear Fusion Tokamaks

Riverola Calavia, Javier (2016). Discretization Error Estimation and Mesh Optimization for Plasma Edge Transport Simulations of Nuclear Fusion Tokamaks. Tesis (Master), E.T.S.I. Industriales (UPM).

Descripción

Título: Discretization Error Estimation and Mesh Optimization for Plasma Edge Transport Simulations of Nuclear Fusion Tokamaks
Autor/es:
  • Riverola Calavia, Javier
Director/es:
  • Jiménez Varas, Gonzalo
Tipo de Documento: Tesis (Master)
Título del máster: Ingeniería Industrial
Fecha: 2016
Materias:
Escuela: E.T.S.I. Industriales (UPM)
Departamento: Ingeniería Energética
Licencias Creative Commons: Reconocimiento - Sin obra derivada - No comercial

Texto completo

[img]
Vista Previa
Pdf - Se necesita un visor de ficheros PDF, como GSview, Xpdf o Adobe Acrobat Reader
Descargar (2MB) | Vista Previa

Resumen

In nuclear fusion, tokamak devices are the most advanced technology of magnetic confinement, in which the divertor configuration allows to control plasma-edge motion and power and particle exhaust. In this sense, computational simulations are an indispensable tool to study and characterize the plasma-edge physics. A key aspect is the selection of a suitable mesh that discretizes the spatial domain and allows to capture accurately the steep gradients of state variables and complex behavior of the plasma-edge. In the first part of this work, different discretization error estimation methodologies are collected from the field of CFD to study their applicability to a plasma-edge model. It is found that Richardson extrapolation is the most suitable discretization error estimation strategy in terms of accuracy, computational cost and implementation cost. Richardson Extrapolation methodology is thus applied to the plasma-edge problem in order to estimate the distribution of discretization error of the ion density and ion temperature over the 2D simplified domain, so-called slab case. The mesh is defined by an exponential refinement toward the divertor targets.The analysis of the observed order of accuracy reveales that several regions present non-asymptotic behavior. In order to fix this problem, the discretization errors are converted into uncertainty errors by the GCI method. Results showed that the highest relative errors for the ion density are concentrated at the targets in the scrape-off layer. For the ion temperature, maximum relative errors are located near the X-point. In the second stage of the thesis, the goal of finding a mesh adaptation method for reducing the discretization errors of the plasma state variables is tackled. First, it is concluded that the best error sensor candidates for guiding the adaptation process are feature-based error indicators and Richardson-based error indicators. An adaptation with two different approaches of the classical undivided difference detector to structured grids is proposed. After the application to the plasma-edge model, it is demonstrated that these feature error sensors do not provide sufficient accuracy for driving a mesh adaptation process to reduce the discretization errors. The final contribution of this work is the development of a mesh optimization method aiming relative discretization error reduction applied to the plasma-edge model. A cost functional based on Richardson extrapolation as error adaptive sensor is presented. Two optimization strategies are tested: (i) the steepest descent method without line search and (ii) BFGS with line search satisfying Wolfe conditions. After an application test, a relative discretization error reduction from 15% to 5% is achieved between the maximum values of both configurations.

Más información

ID de Registro: 44743
Identificador DC: http://oa.upm.es/44743/
Identificador OAI: oai:oa.upm.es:44743
Depositado por: Biblioteca ETSI Industriales
Depositado el: 22 Feb 2017 07:41
Ultima Modificación: 22 Feb 2017 07:42
  • Open Access
  • Open Access
  • Sherpa-Romeo
    Compruebe si la revista anglosajona en la que ha publicado un artículo permite también su publicación en abierto.
  • Dulcinea
    Compruebe si la revista española en la que ha publicado un artículo permite también su publicación en abierto.
  • Recolecta
  • e-ciencia
  • Observatorio I+D+i UPM
  • OpenCourseWare UPM