Orientador: João Bosco Ribeiro do Val / Tese (doutorado) - Universidade Estadual de Campinas, Faculdade de Engenharia Eletrica e de Computação / Made available in DSpace on 2018-08-07T09:08:47Z (GMT). No. of bitstreams: 1 Arruda_EdilsonFernandesde_D.pdf: 7031418 bytes, checksum: 5dc90e2d3823b0e6bcf659159d183007 (MD5) Previous issue date: 2006 / Resumo: Apresenta-se um modelo de controle por intervenções para o problema de produção e estoque de vários itens, com diversos estágios de produção. Este problema pode ser solucionado via programação dinâmica discreta (PD) por um operador de custo descontado. Para contornar a dificuldade de obtenção da solução ótima via PD ao se considerar um número razoável de classes de itens e suas etapas de produção, esta tese desenvolve-se em duas linhas. A primeira delas consiste em tomar uma noção de estabilidade estocástica no sentido Foster-Lyapunov para caracterizar a família de soluções candidatas a ótima, originando uma classe de políticas que geram um subconjunto de estados que são recorrentes positivos. Dessa forma, é possível propor políticas sub-ótimas que sejam estáveis, e cuja consideração de otimalidade possa ser desenvolvida apenas no subconjunto de estados recorrentes, simplificando a tarefa da PD e focando nos estados mais freqüentados no longo prazo. A segunda linha de abordagem consiste em desenvolver técnicas de PD aproximada para o problema, através de uma arquitetura de aproximação fixa aplicada a um subconjunto amostra do espaço de estados. Um avanço analítico é alcançado por observar como uma arquitetura de aproximação pode capturar adequadamente a função valor do problema, vista como uma projeção da função valor na arquitetura. Condições para que um algoritmo de PD aproximada convirja para essa projeção são obtidas. Essas condições são independentes da arquitetura utilizada. Um algoritmo derivado dessa análise é proposto, a partir do monitoramento da variação de passos sucessivos / Abstract: We propose an intervention control model for a multi-product, multi-stage, single machine production and storage problem. The optimal policy is obtained by means of discrete dynamic programming (DP), through a discounted cost contraction mapping. In order to overcome the difficulty of obtaining the optimal solution for problems with a reasonable number of products and production stages, we take two different approaches. The first one consists in using a notion of stochastic stability in the Foster-Lyapunov sense to characterize the candidate policies, thus originating a class of policies that induce a subset of positive recurrent states. Therefore, one can propose suboptimal policies that are stable and seek optimality only in the subset of recurrent states, in such a way that simplifies the DP task and focuses on the states which are visited more frequently in the long run. The second approach consists in developing approximate dynamic programming techniques for the problem, by means of a fixed approximation architecture applied to a sample subset of the state space. A novel result is obtained by observing how an approximation architecture can adequately capture the value function of the problem, which is viewed as a projection of the value function into the architecture. We obtain conditions for an approximate DP algorithm to converge to this projection. These conditions are architecture independent. An algorithm derived from this analysis is proposed that monitors the variation between successive iterates / Doutorado / Automação e Controle / Doutor em Engenharia Elétrica

Orientadores: Secundino Soares Filho, Marinho Gomes de Andrade Filho / Tese (doutorado) - Universidade Estadual de Campinas, Faculdade de Engenharia Eletrica e de Computação / Made available in DSpace on 2018-08-15T04:56:04Z (GMT). No. of bitstreams: 1 Siqueira_ThaisGamade_D.pdf: 1878251 bytes, checksum: c4c0b7a2af163305b181cf917f725354 (MD5) Previous issue date: 2009 / Resumo: O planejamento da operação energética (POE) tem como objetivo determinar metas de geração para cada usina do sistema elétrico a cada estágio (mês) do período de planejamento (anos) que minimizam o custo total de operação, atendendo a demanda e respeitando as restrições operativas das usinas. A Programação Dinâmica (PD) é a ferramenta mais utilizada na resolução do POE devido, entre outras coisas, a sua capacidade de representar a natureza estocástica das vazões afluentes futuras. Entretanto, a PD é restrita pela "maldição da dimensionalidade", pois o esforço computacional cresce exponencialmente com o aumento do número de usinas consideradas. Para poder utilizar a PD no POE de sistemas de grande porte, como o brasileiro, com centenas de usinas, uma abordagem adotada baseia-se na agregação do sistema hidrelétrico em reservatórios equivalentes, reduzindo a dimensão do sistema - quatro reservatórios equivalentes no caso do sistema brasileiro. Este trabalho visa elaborar uma análise comparativa detalhada de diversas políticas operativas para o POE baseadas em PD com o objetivo de estimar o benefício do seu uso em sistemas equivalentes. Para melhor comparar as diferentes políticas são considerados somente sistemas formados por uma única usina, evitando-se assim o efeito das simplificações e limitações dos modelos equivalentes. Alem disso, são comparadas diferentes modelagens para as distribuições de probabilidades das vazões na PD. Os desempenhos das diferentes políticas são avaliados por simulação no histórico de vazões e os resultados, considerando usinas hidrelétricas localizadas em diferentes bacias do sistema brasileiro, indicam que as diferenças não são expressivas / Abstract: The medium term hydrothermal scheduling problem aims to determine, for each stage (month) of the planning period (years), the amount of generation at each hydro and thermal plant that attends the load demand and minimizes the expected operation cost in the planning period. Dynamic programming has been the most applied technique to solve the long term hydrothermal scheduling problem because it adequately copes with the uncertainty of inflows. Although efficient in the treatment of river inflows as random variables described by probability distributions, the dynamic programming technique is limited by the so-called "curse of dimensionality" since its computational burden increases exponentially with the number of hydro plants. In order to overcome this difficulty, allowing the use of dynamic programming in medium term hydrothermal scheduling, one common manipulation adopted is to represent the hydro system by an aggregate model, as it is the case in the Brazilian power system. This work evaluates the influence of inflow modeling in the performance of dynamic programming for medium term hydrothermal scheduling. Different models have been progressively considered in order to evaluate the benefits of increasing sophistication of inflow modelling. Features and sensitivities of different models are discussed. Numerical results for hydrothermal test systems composed by a single hydro plant located in several Brazilian rivers were obtained by simulation and indicated that the differences among the policies are not significant / Doutorado / Energia Eletrica / Doutor em Engenharia Elétrica

Orientador: Alberto Luiz Francato / Dissertação (mestrado) - Universidade Estadual de Campinas, Faculdade de Engenharia Civil, Arquitetura e Urbanismo / Made available in DSpace on 2018-08-23T22:42:20Z (GMT). No. of bitstreams: 1 Deantoni_VictordeBarros_M.pdf: 2817178 bytes, checksum: 1c05a50aa818606a82c052b88f62c14d (MD5) Previous issue date: 2013 / Resumo: O planejamento da operação do setor elétrico brasileiro é realizado com base em modelos que por meio de otimização determinam a geração de energia de fontes térmicas e hidráulicas. Utilizando a técnica de otimização estocástica robusta possibilita-se a análise com um conjunto de cenários históricos, com o objetivo de determinar a operação do primeiro intervalo de tempo do horizonte de planejamento, e assim obter uma solução que não necessariamente seja a melhor para um determinado cenário, mas sim uma solução que seja interessante para qualquer um dos cenários que possam ocorrer. O objetivo deste trabalho foi criar uma nova versão do modelo SolverSIN com um módulo de otimização estocástica robusta, chamado de SolverSINR, esse novo modelo permite o planejamento da operação considerando um conjunto de cenários históricos. As séries históricas são obtidas do relatório do programa mensal de operação, que é um arquivo de saída do NEWAVE. Para organização dessas séries foi desenvolvido um aplicativo chamado SHENA. O novo modelo apresentou resultados coerentes em comparação com o modelo SolverSIN determinístico e mostrou valores viáveis para a operação de reservatórios. A ferramenta permite a otimização assumindo a hipótese de repetição de uma série histórica. O modelo SolverSINR vem a contribuir como mais uma alternativa de avaliação crítica às políticas operacionais do SIN implementadas pelos modelos oficiais do SEB. Durante a avaliação de resultados notou-se que uma restrição de geração hidráulica mínima para o subsistema Nordeste causava infactibilidade em alguns cenários, adotou-se uma variável de folga para solucionar esse problema. Destaca-se que houve factibilidade nos procedimentos de operação em tempo de processamento compatível com otimização estocástica de processos em equipamentos usais para tal tarefa / Abstract: The operation planning of the interconnected power generation Brazilian system is carried out based on models through deterministic optimization power generation from thermal and hydraulic sources. Using the robust stochastic optimization technique enables the analysis grounded in a set of historical scenarios, in order to determine the operation of the first time interval of the planning horizon, and thus obtain a solution that is not necessarily the best for a selected scenario, but a solution that is interesting for any of the scenarios that might happen. The objective of this work was to create a new version of the model SolverSIN with a robust stochastic optimization module, called SolverSINR , this new model allows for the operation planning considering a set of historical scenarios . The time series is obtained from the monthly report called "pmo.dat" that is an output file from NEWAVE model. For organizing these series was developed an application called SHENA. The new model showed consistent results in comparison with the deterministic model (SolverSIN) and showed feasible values for reservoir operation. The tool allows the optimization under the assumption of repetition of a historical series. The model SolverSINR comes to contribute as an alternative assessment to critical operational policies implemented by SIN official models of SEB. During the evaluation of results was noted that a restriction of minimum hydraulic generation on Northeast subsystem caused infeasible answer in some scenarios, we adopted a penalty variable to solve this problem. It is noteworthy that there was feasibility in operating procedures in processing time compatible with stochastic optimization of processes in usual equipment for this task / Mestrado / Recursos Hidricos, Energeticos e Ambientais / Mestre em Engenharia Civil

This dissertation addresses the modeling and solution of mixed-integer linear multistage stochastic programming problems involving both endogenous and exogenous uncertain parameters. We propose a composite scenario tree that captures both types of uncertainty, and we exploit its unique structure to derive new theoretical properties that can drastically reduce the number of non-anticipativity constraints (NACs). Since the reduced model is often still intractable, we discuss two special solution approaches. The first is a sequential scenario decomposition heuristic in which we sequentially solve endogenous MILP subproblems to determine the binary investment decisions, fix these decisions to satisfy the first-period and exogenous NACs, and then solve the resulting model to obtain a feasible solution. The second approach is Lagrangean decomposition. We present numerical results for a process network planning problem and an oilfield development planning problem. The results clearly demonstrate the efficiency of the special solution methods over solving the reduced model directly. To further generalize this work, we also propose a graph-theory algorithm for non-anticipativity constraint reduction in problems with arbitrary scenario sets. Finally, in a break from the rest of the thesis, we present the basics of stochastic programming for non-expert users.

Master of Science / Department of Industrial & Manufacturing Systems Engineering / David H. Ben-Arieh / In present competitive market, there is growing concern for ascertaining and fulfilling the individual customer’s wants and needs. Therefore, the focus of manufacturing has been shifting from mass production to mass customization, which requires the manufacturers to introduce an increasing number of products with shorter life span and at a lower cost. Also, another challenge is to manage the variety of products in an environment where demands are stochastic and the lead times to fulfill those demands are short. The focus of this thesis is to develop and investigate platform based production strategies, as opposed to producing each product independently, which would ensure the economic production of the broader specialized products with small final assembly time and under demand uncertainty. The thesis proposes three different platform based production models. The first model considers the economic production of products based on a single platform and with forecasted demands of the products. The model is formulated as a general optimization problem that considers the minimization of total production costs. The second model is the extension of the first model and considers the production of products based on multiple platforms and considers the minimization of total production costs and the setup costs of having multiple platforms. The third model is also an extension of the first model and considers the demands of the products to be stochastic in nature. The model considers the minimization of total production costs and shortage costs of lost demands and holding cost of surplus platforms under demand uncertainties. The problem is modeled as a two stage stochastic programming with recourse. As only the small instances of the models could be solved exactly in a reasonable time, various heuristics are developed by combining the genetic evolutionary search approaches and some operations research techniques to solve the realistic size problems. The various production models are validated and the performances of the various heuristics tailored for the applications are investigated by applying these solution approaches on a case of cordless drills.

Un problème auquel sont confrontés de nombreux réseaux de distribution, en particulier dans la vente de détail, est celui du choix entre une installation locale de stockage de produits ou une installation commune pour tous les centres du réseau. Le stockage commun permet des économies d’échelle. En revanche, il peut occasionner des coûts de transport importants. Quant au stockage local, il est coûteux du point de vue de l’investissement mais bon marché en utilisation. Pour profiter des avantages des deux techniques tout en limitant leurs inconvénients, on envisage l’installation de stocks locaux pouvant aussi être utilisés par les autres entreprises. Dans ce cas, le problème qui se pose alors est double. Il s’agit d’une part pour chaque entreprise, de déterminer sa propre quantité à commander, en fonction de la demande estimée à chaque point de vente. Il s’agit d’autre part d’organiser les échanges de produits entre les entreprises en fonction des demandes locales réelles. Les travaux de recherche présentés dans ce mémoire de thèse proposent des mécanismes de coordination qui agissent d’une façon collective et non plus d’une façon individuelle, pour améliorer à la fois le rendement au niveau de chaque centre de distribution et la satisfaction des clients.. Notre principal défi dans cette thèse est de proposer une approche analytique innovante issue de la théorie des jeux pour traiter des problèmes de gestion de stock visant à réduire globalement les stocks dans les réseaux de distribution toute en garantissant un bon niveau de service. / A problem facing many distribution networks, especially in retail, is the choice between a local product storage facility or a shared installation for all network centers. The common storage enables economies of scale. However, may cause significant transport costs. As for local storage, it is costly in terms of investment but cheap in use. To enjoy the benefits of both techniques while minimizing their drawbacks, it is envisaged the installation of local stocks can also be used by other companies. In this case, the problem that then arises is twofold. This is one for each company to determine its own order quantity, depending on the demand estimated at each point of sale. It is secondly to organize the exchange of products between companies based on actual local demands. The research presented in this thesis propose coordination mechanisms that act in a collective manner and not in an individual manner, to improve both the yield at each distribution center and customer satisfaction .. Our main challenge in this thesis is to propose an innovative analytical approach end of game theory to address inventory management problems to reduce overall inventory in the distribution networks throughout ensuring a good level of service.

A large part of the foreign exchange market concerns the trading of FX swaps. While entering a position in a FX swap does not cost any money, banks earn money on FX swaps when their customers cross the bid/ask spread, creating a perceived transaction costs for the swaps. To hedge the risks of their customer positions, banks enter new positions in FX swaps with other banks, crossing the same bid/ask spread. Traditional hedging methods does not take perceived transaction costs into account when determining hedge positions, resulting in greater portfolio losses than necessary for the banks. Therefore, the topic of hedging while taking transaction costs into account could be of great value. When valuing FX swaps and estimating risk factors in a FX environment, term structures need to be estimated for pricing the instruments. The estimation of term structures can be done using several ap- proaches, among them bootstrapping and interpolating the curve or parameterizing the curve, assuming it to be described by a functional form. These traditional methods of term structure measurement has the downside of being unstable and fluctuating greatly over time because of different local optimas each day, or result in very large pricing errors due to certain instruments needing to be excluded from the term structure measurement. These attributes result in capturing extra, unnecessary volatility in the curves which does not model the true risk, consequently estimating the risk factors wrongly when risk management and hedging needs to be done. The estimation of good quality term structures which are stable over time and result in low pricing errors are therefore of great interest to study. In this thesis, a FX swap portfolio is hedged using a Stochastic Programming (SP) model developed by Blomvall and Hagenbj ̈ork (2020). For the valuation of FX swaps in the portfolio and the generation of risk factors for the model, term structures were estimated using a multiple yield curve framework of Blomvall and Ndengo (2013), which penalizes pricing errors and use regularization functions to produce smooth curves. For both the term structure measurement method and the hedging method, a critical part affecting the per- formance of the methods lies in choosing good parameter values, which is what has been the main purpose of this study. The results show that good quality term structures can be estimated using the multiple yield curve frame- work if good parameter choices are made. The resulting curves fulfill the criteria of being stable over time while also keeping the price errors out-of-sample small. A portfolio hedged using a SP-model with certain chosen parameter values and also using the good quality term structures estimated is shown to eliminate a great deal of risk compared to an unhedged portfolio. When compared with a traditional hedging model called the Boxes model, the SP-model gains value from taking perceived transaction costs into account and thus manages to hedge the risks less costly than the Boxes model does.

[pt] As companhias petrolíferas dedicam grande esforço para manter sua rentabilidade e melhorar sua eficiência, principalmente frente às incertezas presentes neste negócio. As empresas que pretendem manter a competitividade precisam planejar suas operações cada vez melhor e com maior segurança. Em face destas oportunidades e desafios, foi proposta no âmbito desta tese uma abordagem estocástica para o problema de planejamento operacional de refinarias. Neste sentido foi desenvolvido um modelo não-linear (NLP) de programação estocástica com dois estágios. O modelo proposto representa os processos de natureza não-linear presentes em uma refinaria, como as transformações químicas e o cálculo de qualidade dos derivados. Devido ao elevado nível de complexidade do problema NLP formulado, foram avaliados cinco métodos de solução associados aos principais solvers comerciais. Uma metodologia de geração de cenários e medidas de qualidade para árvore de cenários também foram definidas para representar adequadamente as incertezas presentes neste problema. A abordagem estocástica proposta neste trabalho foi avaliada considerando dados reais de uma refinaria brasileira. Os resultados finais desta pesquisa devem proporcionar avanços no processo de planejamento operacional de refinarias, explorando a técnica de programação não-linear (NLP) e os novos solvers disponíveis para problemas do tipo NLP. Pretende-se também gerar contribuições na área de programação estocástica, definindo medidas de qualidade para árvore de cenários que permitam uma melhor representação das incertezas e consequentemente um melhor uso da abordagem estocástica. / [en] Oil companies make a great effort to maintain profitability and improve efficiency, especially given the uncertainties present in this business. Companies that intend to remain competitive need to plan their operations better and with greater safety. In light of these opportunities and challenges, this thesis proposes a stochastic approach to the refinery operational planning problem. In this sense, a two-stage nonlinear stochastic programming model (NLP) developed. The proposed model is intended to adequately represent nonlinear processes encountered in a refinery, such as chemical transformations and calculations of the properties of the oil derivatives. Due to the high level of complexity of the NLP problem formulated, five solution methods associated with major commercial solvers were evaluated. A methodology for generating scenarios and quality measures for scenarios tree were also defined to properly represent the uncertainties present in this problem. The stochastic approach proposed in the present study was evaluated based on actual data from a Brazilian refinery. The final results of this research should provide advances in the processes of refinery operational planning exploiting the technique of nonlinear programming (NLP) and new solvers available for NLP-type problems. Another objective was to generate contributions in the field of stochastic programming by defining quality measures for scenario trees that allow a better representation of uncertainties and, consequently, better use of the stochastic approach.

The portfolio selection problem has been widely understood and practised for millennia, but it was rst formalised by Markowitz (1952) with the proposition of a risk-reward trade-o model. Since then, portfolio selection models have continued to evolve. The general consensus is that three objectives, to maximise the uncertain Rate Of Return (ROR), to maximise liquidity and to minimise risk, should be considered. It was found that there are opportunities for improvement within the existing portfolio selection models. This can be attributed to three gaps within the existing models. Generally, existing portfolio selection models are generic, especially in how they incorporate risk, they generally do not incorporate Socially Responsible Investing (SRI), and generally they are considered to be unvalidated. This dissertation set out to address these gaps and compare the real-world performance of generic and individualised portfolio selection models. A new method of accounting for risk was developed that consolidates the portfolio's market risk with the investor's nancial risk tolerance. Two portfolio selection models that incorporate individualised risk and SRI objectives were developed. These two models were called the risk-adjusted and social models, respectively. These individualised models were compared to an existing generic Markowitz model. These models were formulated using stochastic goal programming. A sample of 208 companies JSE Limited companies was selected and two independent datasets were extracted for these companies, a training (2010/01/01 { 2016/12/31) and testing (2017/01/01 { 2019/12/31) dataset. The models solved were in LINGO using the training dataset and tested on an unknown future by using the testing dataset. It was found that in the training period, the individualised risk-adjusted model outperformed the generic Markowitz model and the individualised social model. Furthermore, it was found that it would not be bene cial for an investor to be Socially Responsible (SR). Nevertheless, investors invest to achieve their ROR and SRI goals in the future, not in the present. Thus, it was necessary to evaluate how the portfolios selected by all three models would have performed in an unknown future. In the testing period, both the generic Markowitz model and the risk-adjusted models had dismal performance and were signi cantly outperformed by the South African market and unit trusts. Thus, these models are not useful or suitable for their intended purpose. On the contrary, the social model portfolios achieved high ROR values, were SR, and outperformed the market and the unit trusts. Thus, this model was useful and suitable for its intended purpose. The individualised social model signi cantly outperformed the other two models. Thus, it was concluded that an individualised approach that incorporates SRI outperforms a generic portfolio selection approach. Given its unparalleled performance and novel model formulation, the social model makes a contribution to the eld of portfolio selection. This dissertation also highlighted the importance of testing portfolio selection models on an unknown future and demonstrated the potentially horri c consequences of neglecting this analysis. / Dissertation (MEng (Industrial Engineering))--University of Pretoria 2021. / Industrial and Systems Engineering / MEng (Industrial Engineering) / Unrestricted

This master's thesis deals with the proposal of the cost-effective biomass and coal combustion concept for a real generation plant. The optimization of the current fuel basis exploitation and annual operation scheduling was the main goal. Applying the regression analysis and stochastic programming the mathematical model was constructed based on the operating data. The overall energy source model was implemented to the concept of the optimal operation scheduling considering the economic aspects. The fuel utilization and energy production planning are the main applications of this design.