Genetic algorithms are well known for being expensive optimization tools, especially if the cost for the analysis of each individual design is high. In the past few years, significant effort has been put forth in addressing the high computational cost GAs. The research conducted in the first part of this thesis continues this effort by implementing new multiple elitist and variable elitist selection schemes for the creation of successive populations in the genetic search process. The new selection schemes allow the GA to take advantage of a greater amount of important genetic information that may be contained in the parent designs, information that is not utilized when using a traditional elitist method selection scheme. By varying the amount of information that may be passed to successive generations from the parent population, the explorative and exploitative characteristics of the GA can be adjusted throughout the genetic search also. The new schemes provided slight reductions in the computational cost of the GA and produced many designs with good fitness' in the final population, while maintaining a high level of reliability. Genetic algorithms can be easily adapted to many different optimization problems also. This capability is demonstrated by modifying the basic GA, which utilizes a single chromosome string, to include a second string so that composite laminates comprised of multiple materials can be studied with greater efficiently. By using two strings, only minor adjustments to the basic GA were required. The modified GA was used to simultaneously minimize the cost and weight of a simply supported composite plate under different combinations of axial loading. Two materials were used, with one significantly stronger, but more expensive than the other. The optimization formulation was implemented by using convex combinations of cost and weight objective functions into a single value for laminate fitness, and thus required no additional modifications to the GA. To obtain a Pareto-optimal set of designs, the influence of cost and weight on the overall fitness of a laminate configuration was adjusted from one extreme to the other by adjusting the scale factors accordingly. The modified GA provided a simple yet reliable means of designing high performance composite laminates at costs lower than laminates comprised of one material. / Master of Science

Stacking sequence design is a combinatorial problem when the fiber orientations in each layer are restricted to certain angles. In addition, there often exist many optimal or nearoptimal designs for the stacking sequence of a composite pressure vessel under different loading conditions. Genetic algorithms are quite well suited for finding the optimal designs for such a combinatorial problem. In this thesis, a genetic algorithm code is developed in Matlab, optimizing the stacking sequence of a composite pressure vessel subjected to internal and external pressures, axial load and body force due to rotation. For testing of the code and identification of the effects of optimization parameters, a problem, whose optimum solution is obvious, is defined and the optimum design is tried to be found by using the developed code. The results have shown that the code was quite successful in finding the best design. Afterwards, the code is used for the optimization of the stacking sequence of a composite pressure vessel under different loading conditions. Again the code has proven its reliability in finding the optimal designs. The developed genetic algorithm optimization code also has the infrastructure to be easily adapted to the solution of different combinatorial problems.

L’assemblage des matériaux composites par collage présente des avantages incontestés par rapport à d’autres méthodes telles que le boulonnage ou le rivetage. Cependant, la principale difficulté que rencontrent les concepteurs est celle de la prévision du niveau et du mode de rupture de ces collages. Dans le cas des composites stratifiés, un des facteurs influents sur le comportement des joints collés est la séquence d'empilement, mais les travaux présentés dans la littérature ne séparent pas les effets globaux (modification des rigidités de membrane et de flexion) et les effets locaux (orientation des plis au contact de la colle). La présente étude s'intéresse à la caractérisation de ces effets dans le cas de joints de type simple recouvrement de stratifiés carbone/epoxy. Pour isoler les effets locaux, des séquences d'empilement spécifiques quasi isotropes quasi homogènes sont utilisées. A propriétés de raideur globale égales, des différences de résistance de plus de 30% sont observées selon les séquences considérées. Les essais réalisés avec un stratifié symétrique anisotrope en flexion montrent également que la raideur en flexion joue un rôle important dans le comportement des joints. Les modèles analytiques utilisés prédisent les effets globaux avec une bonne précision mais sont inappropriés lorsque des phénomènes locaux se produisent. Une approche par éléments finis permet de prendre en compte ces phénomènes, en modélisant explicitement les plis au contact de la colle et en rendant possible le décollement interlaminaire de ces plis à l'aide d'un modèle de zone cohésive. Cette modélisation est mise en œuvre pour réaliser une étude paramétrique de la géométrie du joint et pour produire une enveloppe de rupture en fonction de la direction de sollicitation. / Adhesive bonding of composite materials has undeniable advantages over other methods such as bolting or riveting. However, one of the difficulties encountered by designers is the prediction of the failure level and failure mode of these adhesively bonded assemblies. In the case of composite laminates, one of the factors acting on the bonded joint behaviour is the stacking sequence, but works presented in the literature do not separate global effects (membrane and bending stiffness modification) and local effect (ply orientation near the adhesive layer). This study deals with the characterization of such effects in the case of single lap joints of carbon/epoxy laminates. In order to isolate the local effects, specific quasi isotropic quasi homogeneous stacking sequences are used. When stiffness properties are maintained constant, strength variations of more than 30 % are observed. Tests performed with a symmetric laminate with bending anisotropy show that the bending stiffness plays also an important role in the joint behaviour. Closed form models are able to predict global effects with a good accuracy but are inappropriate when local effects occur. The use of finite element models can account for these phenomena, by explicitly modelling the laminate plies near the adhesive layer and introducing delamination between these plies with a cohesive zone model. This model is used to perform a parametric study of the joint geometry and to produce a failure envelope according to the orientation of the loading.

Disertační práce začíná rešerší výpočtů pro návrh sendvičových nosníků, desek a složitějších konstrukcí, kde zaujímá významnou roli MKP. Dále jsou popsány optimalizační metody pro ujasnění široké oblasti matematického programování a základních principů topologické optimalizace až po její implementaci na kompozitní konstrukce jinými autory. Pro názornost jsou zmíněny jak analytické, tak i numerické přístupy k optimalizaci sendvičů, kde numerické přístupy umožňují řešit daleko širší oblast úkolů. Cíl disertační práce je stanoven jako implementace zautomatizovaného algoritmu pro optimalizaci za účelem vylepšení návrhového procesu sendvičů s ohledem na napjatost a únosnost. Cíle je dosaženo prostřednictvím vlastní implementace gradientní optimalizace založené na principech topologické optimalizace, známé jako diskrétní optimalizace materiálu (Discrete Material Optimization - DMO) a jejích variant, které pomáhají najít optimální vrstvení. Přístup k materiálové interpolaci a interpolaci poruchový omezujících podmínek je vyvinut a naprogramován v pythonu za použití teorie smykových deformací prvního řádu (First Order Shear Deformation Theory - FSDT) pro vyhodnocení napětí na elementech na základě zatížení daného MKP řešičem Nastran. Gradientní optimalizér hledá nejlepší materiály pro každou vrstvu potahu sendviče a jádra z definovaných kandidátů. Program je odzkoušený na příkladech různé složitosti od nosníku tvořeného jedním elementem, kde je skutečné optimum známé, až po praktickou úlohu sendvičové kuchyňky z dopravního letadla. Výsledky ukázaly, že algoritmus je schopen dosáhnout diskrétního řešení bez (významného) narušení omezujících podmínek a může tedy být prakticky využit ke zefektivnění koncepčního návrhu sendvičů.

Made available in DSpace on 2014-12-17T14:07:19Z (GMT). No. of bitstreams: 1 PedroLSF_DISSERT.pdf: 5571248 bytes, checksum: d237019829d37099df0ac47168668556 (MD5) Previous issue date: 2014-07-21 / Conselho Nacional de Desenvolvimento Cient?fico e Tecnol?gico / Composite laminates with plies in different directions finely dispersed are classified as homogenized. The expected benefits of homogenization include increased mechanical strength, toughness and resistance to delamination. The objective of this study was to evaluate the effect of stacking sequence on the tensile strength of laminates. Composite plates were fabricated using unidirectional layers of carbon/epoxy prepreg with configurations [903/303/-303]S and [90/30/-30]3S. Specimens were subjected to tensile and open hole tension (OHT) tests. According to the experimental results, the mean values of strength for the homogenized laminates [90/30/-30]3S were 140% and 120% greater for tensile and OHT tests, respectively, as compared to laminates with configuration [903/303/-303]S. The increase in tensile strength for more homogenized laminates was associated with the increment in interlaminar interfaces, which requires more energy to produce delamination, and the more complicated crack propagation through plies with different orientations. OHT strength was not affected by the presence of the hole due to the predominance of the interlaminar shear stress in relation to the stress concentration produced by the hole / Laminados formados por l?minas com diferentes orienta??es, finamente dispersas, s?o classificados como homogeneizados. Os benef?cios esperados da homogeneiza??o incluem aumento de resist?ncia mec?nica, da tenacidade e da resist?ncia ? delamina??o. O objetivo deste trabalho foi avaliar o efeito da sequ?ncia de empilhamento na resist?ncia ? tra??o de laminados quase-isotr?picos. Placas de comp?sitos foram fabricadas com l?minas unidirecionais de ep?xi/carbono com configura??es [903/303/-303]S e [90/30/-30]3S. Corpos de prova foram submetidos a ensaios de tra??o e de open hole tension (OHT). De acordo com os resultados experimentais, os valores m?dios de resist?ncia para os laminados mais homogeneizados [90/30/-30]3S foram incrementados em 140% e 120% nos ensaios de resist?ncia ? tra??o e OHT, respectivamente, em rela??o aos laminados com configura??o [903/303/-303]S. O aumento de resist?ncia ? tra??o para laminados mais homogeneizados foi relacionado com o maior n?mero de interfaces interlaminares, que aumenta a quantidade de energia necess?ria para produzir delamina??o, e a maior dificuldade de propaga??o de trincas atrav?s de l?minas com orienta??es diferentes. A resist?ncia medida nos ensaios de OHT n?o foi afetada pela presen?a do furo devido ? predomin?ncia da influ?ncia da tens?o de cisalhamento interlaminar em rela??o ao efeito de concentra??o de tens?o provocado pelo furo

Mn+1AXn-Phasen sind thermodynamisch stabile nanolaminierte Ternärcarbide und -nitride, die sowohl metallische als auch keramische Eigenschaften aufweisen. Der Buchstabe M steht für ein frühes Übergangsmetall, der Buchstabe A für ein A-Element aus den Gruppen IIIA – VIA und X für Kohlenstoff und/oder Stickstoff. Die M-Atome bilden Oktaederschichten mit X-Atomen in ihren Zentren. Der Index n beschreibt die Dicke der Mn+1Xn-Lage, die zwischen zwei hexagonalen A-Schichten eingebettet ist. Die außergewöhnlichen Eigenschaften dieser Materialien haben ihren Ursprung in ihrer Mikrostruktur. Um einen Einblick auf atomarer Ebene zu gewinnen wird die Messmethode der gestörten γ-γ-Winkelkorrelation (PAC) angewendet. Die radioaktiven Sonden 111In/111Cd und 181Hf/181Ta werden durch Ionenimplantation und/oder durch Neutronenaktivierung in das Wirtsmaterial eingebracht, um den elektrischen Feldgradienten (EFG) zu messen, der am Gitterpatz des Sondenatoms herrscht. Das erste Ziel der Arbeit ist die Suche nach optimalen Ausheilparametern, mit denen ein möglichst hoher Anteil der Sonden die gleiche lokale Umgebung spürt. Der nächste Schritt ist die Bestimmung des Gitterplatzes der Sonden in der MAX-Struktur. Als Ergebnis kann festgestellt werden, dass 111In in den In- und Al-basierten MAX-Phasen fast ausschließlich den A-Platz besetzt, während 181Hf in Hf2InC auf dem M-Platz eingebaut wird. Als überraschendes Ergebnis zeigt diese Arbeit, dass die PAC-Methode bei Phasen mit gleichen Konstituenten, aber unterschiedlicher Mn+1Xn-Schichtdicke sensitiv auf die Änderung der Stapelfolge ist. Die Experimente werden mit umfangreichen Rechnungen auf Basis der Dichtefunktionaltheorie (DFT) verglichen, die hier erstmalig für nahezu alle Mitglieder der Familie der MAX-Verbindungen durchgeführt wurden. Die DFT-Rechnungen reproduzieren die gemessenen EFGs mit hoher quantitativer Genauigkeit und stützen die Hypothese, dass sich die Sonden auf den prognostizierten Gitterplätzen befinden.