"Mapping is a way of visualizing parts of the world and maps are largely diagrammatic and two dimensional. There is usually a one-to-one correspondence between places in the world and places on the map, but while there are limitless aspects to the world, the cartographer can only select a few to map" Daniel Dorling, 1996 Map generalization is the process by which small scale maps are derived from large scale maps. This requires the application of operations such as simplification, selection, displacement and amalgamation to map features subsequent to scale reduction. The work is concerned with the problem of effective rendering of large scale datasets on small display devices by developing appropriate map generalization techniques for generating schematic maps. With the advent of high-end miniature technology and large scale digital geographic data products it is essential to devise proper methodologies and techniques for the automated generation of schematic maps specifically tailored for mobile GIS applications. Schematic maps are diagrammatic representation based on linear abstractions of networks. Transportation networks are the key candidates for applying schematization to help ease the interpretation of information by the process of cartographic abstraction. This study looks at how simulated annealing optimisation technique can be successfully applied for automated generation of schematic maps from large scale digital geographic datasets tailored specifically for mobile GIS applications. The software developed makes use of a simulated annealing based schematic map generator algorithm to generate route maps from OSCAR® dataset corresponding to a series of user defined start and end points. The generated schematic route maps are displayed and tested on mobile handheld devices shows promising results for mobile GIS applications. This work concentrates on the automatic generation of schematic maps, which, in the context of mobile mapping, are seen as being a particularly useful means of displaying routes for way finding type and utility network applications.

Esse trabalho tem dois objetivos principais: o primeiro deles consiste em apresentar o estado da arte sobre processos de otimização combinatorial dando uma ênfase especial ao método Simulated Annealing (SA). São apresentados seu histórico, funcionalidades, algoritmo genérico e propostas de paralelização presentes na literatura. Além disso, é apresentado o algoritmo de geração de redes aleatórias contínuas, algoritmo, esse, projetado por pesquisadores do Instituto de Física da UFRGS que utiliza o método SA para gerar redes que atendam certas restrições. O segundo objetivo consiste empropor a paralelização desse algoritmo visando diminuir significativamente o tempo de geração de cada rede, que com o algoritmo seqüencial chega a demorar mais de um mês. Nessa etapa foi utilizada uma adaptação de um dos métodos propostos pela literatura juntamente com a técnica de divisão de domínio. Os resultados obtidos mostraram-se satisfatórios tanto em relação à qualidade numérica quanto à diminuição do tempo de processamento. Além disso, discute-se no trabalho a genericidade da proposta de paralelização a outros problemas baseados em SA. / This work has two main goals: the first one is to present the state of the art on combinatorial optimization processes, with a special emphasis to the Simulated Annealing (SA) method. The work presents its history, features, generic algorithm and proposed parallelization present in the literature. Moreover, the algorithm to generate random networks continued is presented. This algorithm was designed by researchers of the UFRGS Physics Institute and it uses the SA method. The second goal of this work is to propose a parallelization for this algorithm in order to decrease significantly the generation time of each network, that with the sequential algorithm reaches more than months. To do that was used an adaptation of one of the methods proposed by literature together with the domain partitioning technical. The results were satisfactory in terms of the numerical quality and in the decrease of the processing time. In addition, this work discusses the genericity of the proposed parallelization to other problems based on SA.

Submitted by Maurílio Figueiredo (maurilioafigueiredo@yahoo.com.br) on 2014-06-30T19:54:20Z No. of bitstreams: 2 license_rdf: 22190 bytes, checksum: 19e8a2b57ef43c09f4d7071d2153c97d (MD5) DISSERTAÇÃO_MetaheurísticasBuscaLocal.pdf: 5158364 bytes, checksum: 9ff1b731ffa05dadf988172a590814c1 (MD5) / Approved for entry into archive by Gracilene Carvalho (gracilene@sisbin.ufop.br) on 2014-07-07T13:15:13Z (GMT) No. of bitstreams: 2 license_rdf: 22190 bytes, checksum: 19e8a2b57ef43c09f4d7071d2153c97d (MD5) DISSERTAÇÃO_MetaheurísticasBuscaLocal.pdf: 5158364 bytes, checksum: 9ff1b731ffa05dadf988172a590814c1 (MD5) / Made available in DSpace on 2014-07-07T13:15:13Z (GMT). No. of bitstreams: 2 license_rdf: 22190 bytes, checksum: 19e8a2b57ef43c09f4d7071d2153c97d (MD5) DISSERTAÇÃO_MetaheurísticasBuscaLocal.pdf: 5158364 bytes, checksum: 9ff1b731ffa05dadf988172a590814c1 (MD5) Previous issue date: 2014 / Este trabalho apresenta uma proposta e a avaliação computacional de quatro métodos de busca local estocástica para o problema de sequenciamento de tarefas em máquinas paralelas não relacionadas com tempo de preparação dependente da sequência (UPMSP - unrelated parallel machine scheduling problem with sequence dependent setup times). As quatro abordagens metaheurísticas que são analisadas para o UPMSP baseam-se em: Simulated Annealing (SA), Iterated Local Search (ILS), Late Acceptance Hill Climbing (LAHC) e Step Counting Hill Climbing (SCHC). A estrutura das vizinhancas, bem como os parâmetros dos algoritmos, foram amplamente testados e analisados, sendo possível verificar como os parâmetros afetam o comportamento de cada algoritmo implementado e pesquisar os melhores parâmetros. As comparações dos resultados obtidos foram realizadas com os resultados apresentados por Vallada e Ruiz (2011), proponentes do conjunto de 50 instâncias consideradas e, mais recentemente, por Haddad (2012). O método que obteve o melhor resultado nessas 50 instâncias foi testado para todas as 1.000 instâncias grandes, apresentadas por Vallada e Ruiz (2011), melhorando em 96,6% (966 instâncias) a melhor solução conhecida encontrada por esses últimos autores. __________________________________________________________________________________________ / ABSTRACT: This paper presents a proposal and a computational review of four methods of stochastic local search to the unrelated parallel machine scheduling problem with sequence dependent setup times (UPMSP). The four metaheuristics approaches that are analyzed for the UPMSP are based in: Simulated Annealing (SA), Iterated Local Search (ILS), Late Acceptance Hill Climbing (LAHC) and Step Counting Hill Climbing (SCHC). The structure of neighborhoods, as well as the parameters of the algorithms, were widely tested and analyzed, being possible verify how the parameters affect the behavior of each algorithm implemented and search the best parameters. The comparisons of the results were accomplished with the results presented by Vallada and Ruiz (2011), who proposed the set of 50 instances considered, and, more recently, by Haddad (2012). The method that got the best result in these 50 instances was tested for every 1.000 large instances, presented by Vallada and Ruiz (2011), improving in 96.6% (966 instances) the best known solution found by this last authors

La recherche en optimisation est un secteur crucial pour les constructeurs aéronautiques. La performance des appareils est un élément déterminant dans la compétition commerciale qui oppose les principaux manufacturiers du marché. L'incorporation de plus en plus massive des matériaux composites dans les avions de ligne dans les années 2000 illustre le désir des constructeurs de réduire la masse de leurs appareils pour en diminuer la consommation de kérosène. Parallèlement, la sécurité est devenue au fil des années une préoccupation majeure pour l'ensemble des acteurs. Cependant, l'emploi massif de matériaux composites, dont les propriétés physiques sont très intéressantes pour les constructeurs mais qui sont conçus avec une marge de tolérance pour des raisons de coût, induit des variations indésirables dans la structure, des incertitudes. Outre ces matériaux, d'autres éléments non prévisibles sont susceptibles de perturber la structure de l'appareil. Le modèle d'un avion en avant-projet est toujours amené à évoluer pour répondre aux évolutions des exigences du constructeur, mais des études de faisabilité doivent être menées avant que la structure ne soit totalement définie, afin de s'assurer de la viabilité du modèle. Des éléments non pris en compte dans la structure, comme les câbles, peuvent également avoir une influence non négligeable sur le comportement global de l'appareil. Ces incertitudes ont un impact non négligeable sur la stabilité de la structure en vol. Des études ont commencé à incorporer cet aspect incertain dans les processus d'optimisation, mais généralement en adaptant les algorithmes existants et sans exploiter la nature incertaine des problèmes. Afin de tenir compte de l'aspect incertain, on se propose de représenter ces incertitudes par des variables aléatoires et d'exploiter des outils théoriques développés dans d'autres domaines, notamment les outils des mathématiques financières. / Research in optimization is a fundamental field for aircraft manufacturers. The performance of airplanes is a crucial element in the commercial competition that pit main aircraft manufacturers against each other. Since 2000, composite materials have been more and more used in aircraft design. This shows the manufacturers' desire to reduce the airplane weight to diminish kerosene consumption. At the same time, safety has become a major concern for all the parties involved. But the use of composite materials, which are designed with a margin of tolerance on the physical properties to cut cost, causes unwanted variations in the structure. Other factors in the airplane design could also disrupt the overall structure of the final product. An airplane model is intended to be modified according to the manufacturer's wishes and their evolution, but feasibility studies must be carried out before the structural design is complete, in order to make sure that the model is viable. Elements which are not modeled in the structure, e.g. cables, can affect the overall behavior of the airplane. These uncertainties have a non-negligible influence on the stability of the structure during flights. Some studies have started to take into account these uncertainties in the optimization process, but they usually consist in adapting existing deterministic algorithms, without regard for the inherent uncertainty of the problem. In order to take uncertainty into account, we propose to represent these uncertainties by random variables and to use theoretical tools that are used in other domains, such as financial mathematics.

Esse trabalho tem dois objetivos principais: o primeiro deles consiste em apresentar o estado da arte sobre processos de otimização combinatorial dando uma ênfase especial ao método Simulated Annealing (SA). São apresentados seu histórico, funcionalidades, algoritmo genérico e propostas de paralelização presentes na literatura. Além disso, é apresentado o algoritmo de geração de redes aleatórias contínuas, algoritmo, esse, projetado por pesquisadores do Instituto de Física da UFRGS que utiliza o método SA para gerar redes que atendam certas restrições. O segundo objetivo consiste empropor a paralelização desse algoritmo visando diminuir significativamente o tempo de geração de cada rede, que com o algoritmo seqüencial chega a demorar mais de um mês. Nessa etapa foi utilizada uma adaptação de um dos métodos propostos pela literatura juntamente com a técnica de divisão de domínio. Os resultados obtidos mostraram-se satisfatórios tanto em relação à qualidade numérica quanto à diminuição do tempo de processamento. Além disso, discute-se no trabalho a genericidade da proposta de paralelização a outros problemas baseados em SA. / This work has two main goals: the first one is to present the state of the art on combinatorial optimization processes, with a special emphasis to the Simulated Annealing (SA) method. The work presents its history, features, generic algorithm and proposed parallelization present in the literature. Moreover, the algorithm to generate random networks continued is presented. This algorithm was designed by researchers of the UFRGS Physics Institute and it uses the SA method. The second goal of this work is to propose a parallelization for this algorithm in order to decrease significantly the generation time of each network, that with the sequential algorithm reaches more than months. To do that was used an adaptation of one of the methods proposed by literature together with the domain partitioning technical. The results were satisfactory in terms of the numerical quality and in the decrease of the processing time. In addition, this work discusses the genericity of the proposed parallelization to other problems based on SA.

Submitted by Felipe Alves Pires null (fealvespires@hotmail.com) on 2017-05-15T17:24:32Z No. of bitstreams: 1 Dissertação_Felipe_final.pdf: 9171100 bytes, checksum: b1b4cdbe6480b62edce35f61d888f5b9 (MD5) / Approved for entry into archive by Luiz Galeffi (luizgaleffi@gmail.com) on 2017-05-16T14:44:07Z (GMT) No. of bitstreams: 1 pires_fa_me_bauru.pdf: 9171100 bytes, checksum: b1b4cdbe6480b62edce35f61d888f5b9 (MD5) / Made available in DSpace on 2017-05-16T14:44:07Z (GMT). No. of bitstreams: 1 pires_fa_me_bauru.pdf: 9171100 bytes, checksum: b1b4cdbe6480b62edce35f61d888f5b9 (MD5) Previous issue date: 2017-03-28 / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) / Devido à crescente demanda por sistemas de comunicação, monitoramento territorial e previsão climática, satélites foram lançados por agências espaciais e aproximadamente 2,271 orbitam a Terra. Com o objetivo de minimizar os custos de lançamento, estruturas espaciais devem possuir forma de sistemas do tipo treliça devido à considerável redução de peso por serem montadas em uma forma triangular e com materiais leves, por exemplo, alumínio. Como resultado, alguns problemas como flexibilidade e baixo amortecimento estrutural podem aparecer. Diante disso, um método de controle deve ser aplicado para manter os níveis de vibração em pequenas escalas. Neste trabalho, um método de controle passivo é aplicado a uma treliça espacial utilizando o método dos elementos finitos (MEF). MEF é uma das ferramentas computacionais mais utilizadas para analisar estruturas sujeitas a vibrações, tornando possível prever as respostas em frequência (FRFs) do sistema. Este trabalho utiliza a técnica de otimização \textit{simulated annealing} para otimizar a norma $\mathcal{H}_{2}$ para os seis primeiros modos de vibrar do sistema de maneira a encontrar as coordenadas de cada nó da treliça que minimizem o valor de norma $\mathcal{H}_{2}$. Nota-se que, com a minimização da norma $\mathcal{H}_{2}$, novas coordenadas das juntas podem ser obtidas e novas FRFs podem ser estudadas. Algumas amplitudes de vibração das FRFs diminuíram, porém algumas amplitudes aumentaram. Verifica-se que houve regiões que o controle aplicado foi efetivo e outras que a técnica não mostrou eficiência. O método estatístico simulação de Monte Carlo também é aplicado para estudar as incertezas associadas ao sistema. Para isto, histogramas são utilizados para verificação das distribuições das frequências naturais. É pertinente a necessidade de avaliar se as distribuições de frequências apresentam um comportamento de curvas normais e para isso alguns parâmetros estatísticos, como \textit{skewness}, \textit{kurtosis} e teste de qui-quadrado, são utilizados. Observa-se que, as incertezas relacionadas à geometria da estrutura afetam mais o sistema do que as incertezas ligadas às propriedades do material. / Due to the increasing demand for monitoring for telecommunication systems, weather and territorial monitoring, satellites have been launched by space agencies and 2,271 still orbit Earth. In order to minimize rocket launch costs, space structures should have the shape of trusses because of the significant weight reduction for being assembled in a triangular form and with elements made of light materials (e.g. aluminum). As a result of it, some problems such as flexibility and low structural damping may arise. For this reason, a control method must be applied to maintain the requirements of vibration levels in these types of structures. This work applies a passive vibration control to a space truss utilizing the Finite Element Method (FEM). FEM is one of the most used computational tools to analyze structures under vibrations and is also used to predict the system’s frequency response functions (FRFs). This work utilizes the optimization technique simulated annealing in order to optimize the norm H2 of the first six vibration modes of the system so that it’s possible to find the coordinates of each node of the space truss that minimize the value of H2. It has been noticed that with the minimization of the H2 norm, new coordinates of the joints can be obtained so that new FRFs can be studied. Some amplitudes of vibration in the FRFs diminished, however, some of them rose. It is verified that there are regions where the control method applied was effective and others where the control technique was not effective. The statistical method Monte Carlo simulation is also applied to study the uncertainties associated to the system. For this study, histograms are utilized to verify the natural frequencies distributions. It is pertinent the need to evaluate whether the frequencies distributions present the behavior of normal curves and to do this, some statistical parameters such as skewness, kurtosis and chi-square test are used. It has been noticed that uncertainties related to the geometry of the structure affect the system more than uncertainties related to the material properties.

Esse trabalho tem dois objetivos principais: o primeiro deles consiste em apresentar o estado da arte sobre processos de otimização combinatorial dando uma ênfase especial ao método Simulated Annealing (SA). São apresentados seu histórico, funcionalidades, algoritmo genérico e propostas de paralelização presentes na literatura. Além disso, é apresentado o algoritmo de geração de redes aleatórias contínuas, algoritmo, esse, projetado por pesquisadores do Instituto de Física da UFRGS que utiliza o método SA para gerar redes que atendam certas restrições. O segundo objetivo consiste empropor a paralelização desse algoritmo visando diminuir significativamente o tempo de geração de cada rede, que com o algoritmo seqüencial chega a demorar mais de um mês. Nessa etapa foi utilizada uma adaptação de um dos métodos propostos pela literatura juntamente com a técnica de divisão de domínio. Os resultados obtidos mostraram-se satisfatórios tanto em relação à qualidade numérica quanto à diminuição do tempo de processamento. Além disso, discute-se no trabalho a genericidade da proposta de paralelização a outros problemas baseados em SA. / This work has two main goals: the first one is to present the state of the art on combinatorial optimization processes, with a special emphasis to the Simulated Annealing (SA) method. The work presents its history, features, generic algorithm and proposed parallelization present in the literature. Moreover, the algorithm to generate random networks continued is presented. This algorithm was designed by researchers of the UFRGS Physics Institute and it uses the SA method. The second goal of this work is to propose a parallelization for this algorithm in order to decrease significantly the generation time of each network, that with the sequential algorithm reaches more than months. To do that was used an adaptation of one of the methods proposed by literature together with the domain partitioning technical. The results were satisfactory in terms of the numerical quality and in the decrease of the processing time. In addition, this work discusses the genericity of the proposed parallelization to other problems based on SA.

O polímero Poli(3-hidroxibutirato) - PHB - é sintetizado e consumido por microorganismos, o que o torna um material muito interessante e adequado aos problemas ambientais atuais. Porém, sua baixa tenacidade e estabilidade térmica são algumas das principais desvantagens em relação aos principais polímeros de engenharia. A mistura com polímeros modificadores de impacto ou a incorporação de aditivos como plastificantes são algumas das possibilidades para melhorar certas características desse polímero. Contudo, uma razão fundamental para a fragilidade marcante desse material é o processo de envelhecimento que ele sofre mesmo durante o armazenamento, caracterizado por mudanças na estrutura cristalina que diminuem a capacidade de dissipação de energia. Esse trabalho investigou esse fenômeno e seu efeito sobre o PHB puro, a blenda com o Poli(etileno-co-acrilato de metila-co-metacrilato de glicidila) - P(EMAGMA) - e o PHB com o plastificante trietilenoglicol bis(2-etil-hexanoato) TEG(EH). A partir desse ponto, condições de recozimento foram definidas como tentativa de corrigir os problemas conseqüentes do envelhecimento. As amostras recozidas foram caracterizadas e comparadas com amostras envelhecidas, por meio de ensaios de calorimetria exploratória diferencial (DSC), espalhamento de raios-x de baixo ângulo (SAXS), propriedades mecânicas, técnicas de microscopia e biodegradação. Os resultados mostraram que as alterações na morfologia lamelar causadas pelo recozimento melhoram significativamente as propriedades mecânicas das composições testadas e ajudam a prevenir um novo envelhecimento. A composição PHB/Plastificante se mostrou menos susceptível a esse processo de fragilização. Alterações na biodegradabilidade dos materiais também foram observadas, considerando mudanças nas morfologias esferulítica e lamelar. / High biodegradability makes Poly(3-hydroxybutyrate) - PHB - interesting from an environmental point of view. However, this material presents some properties below the required for a large-scale use, especially its low toughness. Blending with polymers with higher impact strength or incorporate additives such as plasticizers can be an alternative to enhance these properties. Even so, a fundamental reason for the remarkable brittleness of PHB is the intrinsic ageing phenomenon that this polymer undergoes through time even at room temperature, marked by changes in microstructure that ultimately restrict energy dissipation. In this work, this phenomenon was investigated for pure PHB, a blend PHB/Poly(ethylene-co-methyl acrylate-co- glycidyl methacrylate) PEMAGMA and PHB with plasticizer Tri(ethylene glycol) bis(2- ethylhexanoate) TEG(EH). As an attempt to overcome the drawbacks caused by ageing, annealing conditions were defined and the annealed samples were characterized and compared with aged ones, using differential scanning calorimetry (DSC), small angle x-ray scattering (SAXS), tensile testing, microscopy techniques and biodegradation. The results showed a significant improvement in mechanical properties and prevention of a possible re-aging. The composition PHB/Plasticizer proved to be less susceptible to ageing. Changes in the spherulitic and lamelar morphology have also affected the biodegradability of the samples.

A liga Cu-30%Zn, conhecida como latão 70/30, possui diversas aplicações comerciais, tais como cartuchos para munição e conectores elétricos. Na produção de chapas e barras dessa liga no formato e nas propriedades almejadas para a aplicação final, comumente utiliza-se de tratamento térmicos para restauração das propriedades mecânicas, chamados de recozimento. Em particular, o efeito de solutos em solução sólida é um fator determinante para o comportamento de ligas metálicas durante o recozimento e deve ganhar importância com a tendência industrial de utilização de matéria-prima reciclada, gerando adições inadvertidas de elementos químicos que estarão presentes como solutos em solução sólida. Este trabalho levanta dados referentes a alterações no comportamento do latão 70/30 diante do tratamento de recozimento causados por teores residuais de ferro. Com base na revisão da literatura, propôs-se a metodologia de tratamentos interrompidos em dilatômetro para o estudo do recozimento de três materiais: CuZn-1Fe (0,0126%Fe), CuZn-4Fe (0,0417%Fe) e CuZn-6Fe (0,0599%Fe). Duas taxas de aquecimento foram utilizadas: 10 ºC/min e 1000 ºC/min. As propriedades mecânicas foram avaliadas em termos de microdureza Vickers e a fração recristalizada determinada por metalografia quantitativa. As análises microestruturais mostraram que a recristalização inicia-se preferencialmente em bandas de cisalhamento. A análise conjugada dos resultados mostra a seguinte faixa de temperatura de recristalização: taxa de aquecimento de 10 ºC/min - 295ºC a 345ºC; taxa de 1000 ºC/min - 365ºC a 445ºC, sendo que o teor residual de ferro não causa alterações neste comportamento. A análise das amostras recozidas isotermicamente a 500ºC e a 600ºC mostram uma tendência à formação de uma microestrutura heterogênea, possivelmente pela evolução da microestrutura de recristalização, que dá origem a grãos grosseiros em sua etapa final. A amostra com 599 ppm de Fe não apresentou evolução do tamanho de grão a 500ºC. Resultados de MET não mostraram a formação de partículas dispersas, enquanto análises de DSC revelaram um pico prévio à recristalização, identificado como a formação de uma atmosfera de Cottrell, sendo esta atrasada pela presença de ferro em solução sólida. / The Cu-30%Zn alloy, known as 70/30 brass, has many commercial applications like ammunition cartridge and electronic connectors. To properly produce strips and bars of this material it is necessary to realize a heating treatment called annealing, which aims on recovering the mechanical properties of the material. The effect of solute atoms is a high important factor to determine the material behavior during annealing. This issue is getting even more importance to industrial applications, due to the increase of the recycle of waste products. These wastes often contaminate the molten metal with foreign elements which may be in solid solution after solidification. This work aims to assess changes in behavior of the 70/30 brass during annealing due to iron in residual content. Based on literature, it was proposed a methodology which takes advantages of interrupted annealing in dilatomer and isothermal annealing for three different iron residual content: CuZn- 1Fe (0,0126%Fe), CuZn-4Fe (0,0417%Fe) and CuZn-6Fe (0,0599%Fe). Two heating rates were used: 10 ºC/min e 1000 ºC/min. The mechanical properties were evaluated by Vickers microhardness and the recrystallized fraction was determined by quantitative microscopy. The microstructural analysis showed that the recrystallization starts preferentially on shear bands. The results from different techniques showed that the recrystallization temperature range is 295ºC to 345ºC when the heating rate was 10 ºC/min and 365º to 445ºC for 1000 ºC/min. There were no evidences of changing the brass behavior during annealing with the residual iron content. The isothermal annealed samples demonstrated a tendency for the formation of a heterogeneous microstructure at both 500ºC or 600ºC, which can be due to the recrystallization microstructure evolution. In the final stage, bigger recrystallized grains are formed inside cold worked grains. Also, the material with 599 ppm Fe did not change its grain size during annealing at 500ºC. MET results did not demonstrate the occurrence of dispersed iron particles. DSC analysis showed a peak prior to recrystallization, which was identified as the formation of a Cottrell atmosphere in all samples. The iron solute seems to detain its occurrence.

Crude oil distillation is an energy intensive and environmentally challenging process. To decrease the large energy demand of crude oil distillation, heat integration is implemented. The system (i.e. distillation unit and heat exchanger network, HEN) needs to perform an energy-efficient separation in a broad range of scenarios (e.g. changes in product yields or product specifications), without compromising overall profit. Operational optimisation and revamp projects are frequently implemented to adapt an existing system to such diverse scenarios. The present work provides a new approach for optimising crude oil distillation systems. The scope of this methodology consists of: 1) finding the operating conditions for the distillation system that maximise net profit, while 2) proposing retrofit modifications for the HEN that allow a feasible operation. Artificial neural networks (ANN) are used to represent the distillation process. In the proposed modelling strategy, results of rigorous simulations provide the data used to train the ANN models. The resulting ANN models have the advantages of overcoming convergence problems presented by both rigorous and simplified models, of handling fewer variables and performing calculations in less time. The HEN models used in this work consist of a retrofit model and a simulation model. The HEN retrofit model employed by Chen (2008) is extended to include constraints on heat transfer areas, utility consumption; and to optimise stream split fractions. In addition, the segmented linear data used by Chen (2008) to calculate temperature-dependent heat capacities are replaced by models tailor-made for each stream. This allows a more flexible and accurate representation of these properties, compared to the approach of Chen (2008). The HEN simulation model of de Oliveira Filho et al. (2007) is modified and extended in this work to simulate simple unit operations and to consider heat exchangers specified in terms of heat loads. Distillation, HEN and economic models are implemented in a two-level optimisation framework. The first level consists of a simulated annealing algorithm that optimises the operating conditions of the distillation unit (e.g. flow rates of products and stripping steam, pump-around duties and temperature drops, furnace exit temperature) and HEN topology (i.e. number and location of heat exchangers and stream splitters). The second level solves a non-linear least squares problem that addresses the violation of HEN constraints. Different objective functions can be considered, such as maximising net profit or minimising total annualised costs. The case studies presented in this work show that ANN models are suitable for their implementation in optimisation methodologies for crude oil distillation systems. Results indicate that interactions between the distillation process and HEN are captured, and that significant economic improvements can be achieved with the proposed optimisation approach.