Heat treatment can be defined as a combination of heating and cooling operations applied to a metal or alloy in solid state. It is an important manufacturing process, which controls the mechanical properties of metals, therefore contributes to the product quality. Computerized Heat Treatment Planning System for Quenching and Tempering (CHT-q/t), a windows based stand alone software, is developed to assist the heat treatment process design. The goal of CHT-q/t is to predict the temperature profile of load in batch as well as continuous furnace during heating, quenching and tempering of steel, then to predict the mechanical properties as Quenched & Tempered, and finally to optimize the heat treatment process design. The thesis reviews existing heat treating simulation software and identifies the industrial need of a software tool which integrates part load and furnace model with heat treating process. The thesis discusses cooling curve of specimen and Time Temperature Transformation (TTT) diagram to determine the microstructure evolution and subsequently the mechanical properties of steel after quenching. An extensive database has been developed to support the various function modules. The thesis focuses mainly in the TTT and quenchant database development, property prediction after quenching and tempering and the implementation of software. The properties determined in the thesis are hardness, ultimate tensile strength, yield strength, toughness and percentage elongation. Hardness has been predicted by the use of some well known analytical equations and the TTT database, finally regression analysis has been used to give the value as a function of carbon percentage and volume fraction of martensite. The other mechanical properties are calculated based on a relation of hardness and volume fraction of martensite. Various case studies were performed to show the application of CHT-bf and CHT-cf at Bodycote Thermal Processing, Worcester & Waterbury. The objective behind the case studies was to study the effect of change in load arrangement, production rate and cycle time on the heat treated parts and finally to give recommendations in order to save energy and improve productivity and quality.

Steel is an important engineering material used in a variety of applications due to its mechanical properties and durability. With increasing demand for higher performance, complex structures, and the need for cost reduction within manufacturing processes, there are numerous challenges with traditional steel design options and production methods with manufacturing cost being the most significant. In this research, this challenge is addressed by developing a micro-genetic algorithm to minimize the manufacturing cost while designing steel with the desired mechanical properties. The algorithm was integrated with machine learning models to predict the mechanical properties and microstructure for the generated alloys based on their chemical compositions and heat treatment conditions. Through this, it was demonstrated that new steel alloys with specific mechanical property targets could be generated at an optimal cost. The research’s contribution lies in the development of a different approach to optimize steel production that combines the advantages of machine learning and evolutionary algorithms while increasing the number of input parameters. Additionally, it uses a small dataset illustrating that it can be used in applications where data is lacking. This approach has significant implications for the steel industry as it provides a more efficient way to design and produce new steel alloys. It also contributes to the overall material science field by demonstrating its ability in a material’s design and optimization. Overall, the proposed framework highlights the potential of utilizing machine learning and evolutionary algorithms in material design and optimization. / Thesis / Master of Applied Science (MASc) / This research aims to develop an AI-based functional integrated with a heuristic algorithm that optimizes parameters to meet desired mechanical properties and cost for steels. The developed algorithm generates new alloys which meet desired mechanical property targets by considering alloy composition and heat treatment condition inputs. Used in combination with machine learning models for the mechanical property and microstructure prediction of new alloys, the algorithm successfully demonstrates its ability to meet specified targets while achieving cost savings. The approach presented has significant implications for the steel industry as it offers a quick method of optimizing steel production, which can reduce overall costs and improve efficiency. The integration of machine learning within the algorithm offers a different way of designing new steel alloys which has the potential to improve manufactured products by ultimately improving their performance and quality.

Orientador: Sergio Persio Ravagnani / Tese (doutorado) - Universidade Estadual de Campinas, Faculdade de Engenharia Quimica / Made available in DSpace on 2018-08-07T03:49:14Z (GMT). No. of bitstreams: 1 Trochmann_JoseLuizLino_D.pdf: 1070584 bytes, checksum: 3407aee7ad6d88d9de0a1326aaf3d29d (MD5) Previous issue date: 2006 / Resumo: Neste trabalho de tese foi realizado um estudo do potencial de predição de propriedades de transporte em matrizes poliméricas de poli - imidas, utilizando a simulação dinâmica molecular de gases simples como Oxigênio, Nitrogênio e Dióxido de Carbono. A propriedade de transporte de interesse prático, a permeabilidade de uma membrana polimérica a um dado penetrante, envolve a determinação de propriedades de ordem cinética e termodinâmica, respectivamente a determinação do coeficiente de difusão e da solubilidade deste penetrante na matriz polimérica. Atenção especial foi conferida à propriedade cinética, pela predição do coeficiente de autodifusão dos penetrantes. Num procedimento experimental clássico é de vital importância para significância das conclusões derivadas dos experimentos, o uso de amostras de membranas poliméricas adequadamente preparadas quanto à composição química, estrutura física e morfologia. Analogamente, quando se utiliza a simulação molecular para a predição de propriedades, tais como o coeficiente de autodifusão, também é de fundamental relevância para os resultados obtidos, a qualidade dos modelos moleculares das matrizes poliméricas, que serão usados como base. Assim para a preparação de modelos moleculares com o adequado empacotamento, um procedimento para a obtenção de modelos bem equilibrados foi desenvolvido neste trabalho. Os modelos moleculares desenvolvidos foram usados para a obtenção dos valores de massa específica em função da temperatura, e comparados aos valores experimentais disponíveis e quando necessário a, valores preditos por meio da expressão de massa específica em função da temperatura, acima e abaixo da temperatura de transição. A capacidade do modelo molecular desenvolvido em predizer a massa especifica e temperatura de transição vítrea foi usada como critério para a validação da adequação do empacotamento proposto para o referido modelo molecular da matriz polimérica. Os modelos validados de empacotamento, células amorfas, foram utilizados para o cálculo do coeficiente de autodifusão dos gases acima mencionados, através do da simulação dinâmica molecular. A comparação dos coeficientes de autodifusão obtidos das poli-imidas aromáticas e éster imidas, BAAF, 6FDA-ODA, PMDA-ODA e BA-20DA, para os gases O2, N2 e CO2, com os dados experimentais, permitiu concluir a adequação das células amorfas e do esquema de simulação dinâmica molecular para a predição do coeficiente de autodifusão.. A versão preditiva de Vrentas e Duda, baseada na teoria do volume livre, foi utilizada para a predição dos coeficientes de autodifusão da água e do etanol para as poli-imidas acima. , Estes valores, quando comparados com os valores obtidos através da simulação dinâmica molecular mostram a validade de ambas as teorias para a predição da cinética de difusão de penetrantes em matrizes poliméricas complexas / Abstract: In this thesis a study of the predictive potential of the molecular dynamic simulation was performed for transport properties of light gases in polyimide matrix. From de practical point of view permeability is the property of most interest, and involves kinetics as well as thermodynamics properties, diffusion coefficient and solubility of the penetrants molecule in the bulk polymeric matrix, this work will be focus in the former. As important as is in as experimental work, a well prepared polymeric membrane is essential for the significance of the draw conclusions. Therefore a special attention was take in the preparation of the bulk molecular polymeric model, the so called amorphous cell, in order to obtain well-equilibrated molecular packing models for the polyimide matrixes. The amorphous cells were prepared throughout thermodynamic transforms, using one or more of the statistical ensembles and cell specific volume obtained as a function of temperature, this data was compared against the experimental data available, and when necessary to data obtained via predictive methods. The molecular packing model ability to predict the glass transition temperature was used as criteria to validate de amorphous cell, to be used in the molecular dynamic' simulations allow the matrix to be locally flexible and coupled to the classic molecular dynamics simulation. The resulting self diffusion coefficients for the polyimide, BAAF, 6FDA-ODA, PMDA­ODA and BA-20DA for the gases O2, N2 e CO2 were compared to the experimental data. The lack of quality experimental diffusion data available for polyimide membranes for larger penetrants as water and ethanol, showed up as a good opportunity to assess the predictive capability of the molecular dynamic simulation for self diffusion coefficients, considering the relevant technological relevance of polyimide membranes for pervaporation process. The data of self diffusion coefficient produced by the predictive version of free-volume theory after Vrentas and Duda, was compared with the data produced via coupled molecular dynamic simulation for the water and ethanol penetrants, showing the relevance of both theories for the prediction of penetrants kinetic in complex polymeric matrixes / Doutorado / Ciencia e Tecnologia de Materiais / Doutor em Engenharia Química

Made available in DSpace on 2016-08-17T16:54:31Z (GMT). No. of bitstreams: 1 TESE_Alex Oliveira Barradas Filho.pdf: 3027425 bytes, checksum: 63f427132994ab4f584101b97cb4cc80 (MD5) Previous issue date: 2015-01-30 / Alternative fuels have the potential to replace gradually the petroleum derivatives, and the biodiesel, that is a biofuel obtained from transesterification of triglycerides, is pointed as a substitute for mineral diesel. The present work focus on the optimization and application of artificial neural networks (ANNs) on the prediction of viscosity, iodine value, induction period, cetane number, specific gravity and cold filter plugging point of biodiesel, which are properties inherent to the composition. The input variables were the percentage of 13 fatty acid methyl esters (FAMEs) more common in biodiesels and, once the transesterification does not modify the fatty esters profile of the raw materials, the ANN method allowed the prediction of the six properties, even before the transesterification, after synthesis of the biodiesel or during the storage. Therefore, this method can be useful as a tool to evaluate the potential of raw materials to produce a biodiesel with good quality and to reach improvements on official methods. The optimization process of ANN occurred in three steps: test of algorithms for adjusting weights, test of stopping condition and test of activation functions, and the physicochemical properties were treated independently. For the set of test samples, which simulates real samples, the application of the optimized ANNs provided results with root mean squared errors (RMSE) of 0.55 mm²/s, 3.49 g/100g, 0.89 h, 2.06, 2.89 kg/m³ and 2.61 °C for viscosity, iodine value, induction period, cetane number, specific gravity and cold filter plugging point, respectively, what ensures the feasibility of the proposed method. A comparison between the proposed method and linear methods from literature, both based on the biodiesel composition indicate that our ANN model is much more adequate to the problem addressed. / Na busca por combustíveis alternativos que possam substituir gradualmente os derivados de petróleo, o biodiesel é apontado como um substituto para o diesel mineral e é definido como um biocombustível obtido a partir da transesterificação de triglicerídeos. O presente trabalho tem como objetivo a otimização e aplicação de redes neurais artificias (ANNs) na predição de viscosidade, índice de iodo, período de indução, número de cetano, massa específica e ponto de entupimento de filtro a frio (PEFF) de biodiesel, propriedades inerentes à composição. As variáveis de entrada foram os percentuais de 13 ésteres metílicos de ácidos graxos (FAMEs) mais comuns em biodieseis e, como a transesterificação não altera o perfil de ésteres de ácidos graxos da matéria-prima, o método ANN permitiu a predição das seis propriedades, seja antes da transesterificação, após a síntese de biodiesel ou durante o armazenamento. Portanto, este método pode ser útil como uma ferramenta para avaliar o potencial de matérias-primas para produzir um biodiesel com boa qualidade e para alcançar melhorias relativas aos métodos oficiais. O processo de otimização da ANN ocorreu em três etapas: teste dos algoritmos para ajuste de pesos, teste das condições de parada e teste das funções de ativação, e as propriedades físico-químicas foram tratadas de forma independentes. Para o conjunto de amostras de teste, que simula as amostras reais, a aplicação das ANN otimizadas forneceu resultados com a raiz do erro médio quadrático (RMSE) de 0,55 mm²/s, 3,49 g/100g, 0,89 h, 2,06, 2,89 kg/m³ e 2,61 °C para viscosidade, índice de iodo, período de indução, número de cetano, massa específica e PEFF, respectivamente, o que assegura a viabilidade do método proposto. Uma comparação entre o método proposto e métodos lineares, ambos com base na composição de biodiesel, indica que o modelo de ANN é mais adequado para o problema abordado.

Les solvants occupent une place prépondérante dans l’industrie chimique et se retrouvent au cœur de nombreuses applications telles que la formulation de produits phytosanitaires, d’encres ou de peintures, le nettoyage industriel ou les procédés d’extraction, de synthèse ou de séparation. L’épuisement des ressources pétrolières, le durcissement de la réglementation, et une prise de conscience collective motivent le développement d’alternatives à l’utilisation de solvants pétrochimiques. En effet, environ 45% des émissions de composés organiques volatils (COVs) en France proviennent de l’utilisation des solvants, qui, pour la plupart, présentent une empreinte environnementale et sanitaire peu favorable. Le panorama des solvants industriels amorce inévitablement une mutation, qui nécessite la recherche de solvants plus respectueux de l’environnement et des utilisateurs, au regard de leurs propriétés et de leur mode de production. Outre les liquides ioniques, les fluides supercritiques et les solvants fluorés qualifiés de solvants verts, les biosolvants sont apparus comme une solution alternative capable de répondre à un grand nombre de spécifications requises dans diverses applications. L’élaboration de biosolvants s’accompagne d’un changement de matière première, au profit de ressources renouvelables issues de la biomasse. Parmi les molécules plateforme biosourcées utilisées pour la synthèse de bioproduits, le furfural, obtenu par déshydratation des sucres contenus dans les rafles de maïs, a été sélectionné dans le cadre de cette étude visant à développer de nouveaux biosolvants, en collaboration avec la société Rhodia-Solvay (projet InBioSynSolv). Ainsi, afin de substituer des solvants conventionnels utilisés pour formuler des actifs phytosanitaires ou pour le nettoyage industriel, deux méthodologies, différentes de l’approche essais et erreurs, ont été étudiées. La première méthodologie, prédictive, se base sur la prédiction des propriétés avant la synthèse des molécules. La formulation inverse est, quant à elle, une méthodologie innovante qui permet de concevoir des molécules de biosolvants grâce à un laboratoire virtuel; les étapes de génération de structures moléculaires et de prédiction des propriétés, sont intégrées à un outil informatique d’aide au design moléculaire (CAMD) qui propose des solutions répondant aux spécifications visées. Dans un premier temps, ces méthodologies ont conduit à identifier un pool de molécules candidates dérivées du furfural et susceptibles de jouer le rôle de solvant pour les applications envisagées. Dans un deuxième temps, la faisabilité des filières de leur production a été étudiée, depuis la molécule plateforme jusqu’à l’utilisation du biosolvant au sein d’une formulation. Pour cela, les molécules candidates ont été obtenues selon différentes voies de synthèse, que l’on a caractérisées à l’aide de la détermination d’indicateurs verts. Une démarche d’éco conception a également contribué à la mise en place d’une approche multi critère intégrant les aspects techniques, environnementaux et socio- économiques. Enfin, la production d’échantillons a permis de vérifier expérimentalement les propriétés recherchées, et de valider l’intérêt des méthodologies de substitution de solvants utilisées, en termes de gain de temps et d’efficacité. Celles-ci pourront être généralisées au développement de différents bioproduits pour accompagner les évolutions des marchés auxquelles doit faire face l’industrie chimique. / The solvents play a significant role in the chemical industry and are at the heart of many applications such as the formulation of pesticides, inks or paints, industrial cleaning or extraction processes, synthesis and separation. The depletion of fossil resources, stricter regulations and collective awareness incite the development of alternatives to the use of petrochemical solvents. In fact, about 45% of emissions of volatile organic compounds (VOCs) come from the use of solvents, most of which have a very unfavorable environmental and health impact. The panorama of industrial solvents inevitably initiates a change, which requires the search for more eco friendly solvents in terms of their properties and their mode of production. In addition to the ionic liquids, supercritical fluids and fluorinated solvents, called green solvents, biosolvents emerged as an alternative capable of meeting a large number of specifications required in various applications. Developing biosolvents is accompanied by a change in raw material, from petroleum to renewable resources from biomass. Among the biobased platform molecules used for the synthesis of bioproducts, furfural, obtained by dehydration of sugars in corn cobs, was selected as part of this study to develop new biosolvents in collaboration with Rhodia-Solvay (InBioSynSolv project). Thus, to replace conventional solvents used in phytosanitary formulations or for industrial cleaning, two methodologies different from the tests and error approach, were studied. The first methodology, predictive, is based on the properties prediction before the synthesis of the molecules. The inverse formulation is, in turn, an innovative methodology to design molecules of biosolvents through a virtual laboratory. Stages of generation of molecular structures and properties prediction are integrated in a computer-aided molecular design tool (CAMD) providing solutions that meet the outlined specifications. First, these methodologies have led to identify a pool of candidate molecules derived from furfural that may act as a solvent for the intended applications. In a second step, the feasibility of their production chains has been studied from the molecule platform to the use of the biosolvent in a formulation. For this, the candidate molecules were obtained by different synthetic routes, which were characterized using the determination of green indicators. An eco-design approach has also contributed to take into account different criteria including technical, environmental and socio-economic aspects. Finally, with the production of samples, properties were experimentally verified, to validate the interest of solvents substitution methodologies in terms of time savings and efficiency. These could be generalized to the development of various bioproducts to make possible innovation in the chemical industry.

In this master’s thesis are discussed static properties of network traffic trace. There are also addressed the possibility of a predication with a focus on neural networks. Specifically, therefore recurrent neural networks. Training data were downloaded from freely accessible on the internet link. This is the captured packej of traffic of LAN network in 2001. They are not the most actual, but it is possible to use them to achieve the objective results of the work. Input data needed to be processed into acceptable form. In the Visual Studio 2005 was created program to aggregate the intensities of these data. The best combining appeared after 100 ms. This was achieved by the input vector, which was divided according to the needs of network training and testing part. The various types of networks operate with the same input data, thereby to make more objective results. In practical terms, it was necessary to verify the two principles. Principle of training and the principle of generalization. The first of the nominated designs require stoking training and verification training by using gradient and mean square error. The second one represents unknown designs application on neural network. It was monitored the response of network to these input data. It can be said that the best model seemed the Layer recurrent neural network (LRN). So, it was a solution developed in this direction, followed by searching the appropriate option of recurrent network and optimal configuration. Found a variant of topology is 10-10-1. It was used the Matlab 7.6, with an extension of Neural Network toolbox 6. The results are processed in the form of graphs and the final appreciation. All successful models and network topologies are on the enclosed CD. However, Neural Network toolbox reported some problems when importing networks. In creating this work wasn’t import of network functions practically used. The network can be imported, but the majority appear to be non-trannin. Unsuccessful models of networks are not presented in this master’s thesis, because it would be make a deterioration of clarity and orientation.