Multiscale topology optimization

Κωστόπουλος, Χρήστος; Kostopoulos, Christos

dc.contributor.author	Κωστόπουλος, Χρήστος	el
dc.contributor.author	Kostopoulos, Christos	en
dc.date.accessioned	2022-11-18T11:48:24Z
dc.date.available	2022-11-18T11:48:24Z
dc.identifier.uri	https://dspace.lib.ntua.gr/xmlui/handle/123456789/56178
dc.identifier.uri	http://dx.doi.org/10.26240/heal.ntua.23876
dc.rights	Default License
dc.subject	Topology Optimization	en
dc.subject	Multiscale Analysis	en
dc.subject	Finite Element Method	en
dc.subject	Composite materials	en
dc.subject	Homogenisation	en
dc.title	Multiscale topology optimization	en
heal.type	bachelorThesis
heal.classification	Topology Optimization	en
heal.classification	Multiscale Analysis	en
heal.classification	Finite Element	en
heal.access	free
heal.recordProvider	ntua	el
heal.publicationDate	2022-07-11
heal.abstract	An efficient multiscale material and topology optimization scheme for fiber-reinforced structures is proposed. The goal was to find the structural topology and set of fiber orientations that minimized a fiber-reinforced structure’s compliance using a multiscale framework. Effective properties of the material were calculated using homogenization. The design domain was discretized in 4-node plane stress finite elements. The topology optimization problem was described by assigning a density variable and an orientation variable to each element using the solid isotropic material with penalization method. The fiber orientation was simulated by rotating the homogenized elasticity tensor appropriately. The homogenization took place only once, and it was not included in the optimization iterations. A convolution filter was applied to both densities and orientations, aiming to avoid checkboard patterns, orientation discontinuities and convergence to local minima. The optimization problem was solved for densities and orientations concurrently using the gradient-based Method of Moving Asymptotes (MMA). The filter was applied adaptively to both densities and orientations in order to avoid dependency from initial conditions. Before the termination of optimization iterations elimination of intermediate densities was performed. The algorithm was applied to a carbon nanotube-reinforced structure and it proved to be stable and efficient, yielding checkerboard-free structures with smooth distribution of orientations. To conclude, we compared the results obtained by optimizing the topology and orientations to simple topology optimization. The proposed algorithm yielded superior and much more efficient results.	en
heal.advisorName	Παπαδόπουλος, Βησσαρίων	el
heal.committeeMemberName	Παπαδόπουλος, Βησσαρίων	el
heal.committeeMemberName	Τριανταφύλλου, Σάββας	el
heal.committeeMemberName	Νερατζάκη, Μαρία	el
heal.academicPublisher	Εθνικό Μετσόβιο Πολυτεχνείο. Σχολή Πολιτικών Μηχανικών	el
heal.academicPublisherID	ntua
heal.fullTextAvailability	false