Print Email Facebook Twitter Model order reduction of nonlinear magnetic problems Title Model order reduction of nonlinear magnetic problems Author Singh, A.S. Contributor Tiso, P. (mentor) Faculty Mechanical, Maritime and Materials Engineering Department Precision and Microsystems Engineering Programme Engineering Mechanics Date 2015-08-24 Abstract Computational numerical methods are important tools in science and technology today. Numerical simulations of nonlinear problems are usually computationally expensive. Model order reduction is a technique to reduce computation time of mathematical numerical models. Model order reduction of nonlinear problems is still in its developing state. In this thesis, model order reduction techniques are implemented to a special class of nonlinear problems, where the non-linearity is localized in only some parts of the domain. The implementation of model order reduction techniques for this work can be divided into three parts. The first part consists of identification of the part of the domain, that is acting nonlinearly, and then condensing the rest of the domain on this nonlinear part. This method is called nonlinear condensation. The second part consists of finding appropriate reduction basis for the conventional Galerkin projection method. Reduction basis formed using static modes, magnetic modes, modal derivatives and proper orthogonal decomposed snapshots are used. The third part is called hyper reduction and is based on the idea of approximating the nonlinear terms of the analysis, by computing it over a subset of elements rather than the complete domain. To compensate for the energy of the remaining elements, elemental contribution terms from the sampling subset are multiplied by corresponding weights. Hence the name of this method: the Energy-Conserving mesh sampling and weighting (ECSW) hyper reduction method. This thesis concentrates on computational magnetics, but the discussed methods can be applied to any other problem involving localized non-linearities. These methods have been found to yield moderately accurate to very accurate results, while providing a speed-up of an order of magnitude of one to two. Subject MORnonlinearmagneticsfemecswnonlinear condensation To reference this document use: http://resolver.tudelft.nl/uuid:4344e0fa-3d50-4fe0-a4c1-fc935b354f90 Part of collection Student theses Document type master thesis Rights (c) 2015 Singh, A.S. Files PDF final_report.pdf 8.55 MB Close viewer /islandora/object/uuid:4344e0fa-3d50-4fe0-a4c1-fc935b354f90/datastream/OBJ/view