ABSTRACT. Neste trabalho propõe-se uma abordagem que utiliza uma associação de dois métodos: primal-dual barreira logarítmica e o da função Lagrangiana barreira logarítmica modificada para a resolução do problema de Fluxo de Potência Ótimo Reativo. Na abordagem proposta, as restrições de desigualdade são transformadas em igualdade ao introduzir variáveis de folga e excesso, as quais são tratadas pela função barreira logarítmica ou pela função barreira logarítmica modificada, e as demais restrições são tratadas pelos multiplicadores de Lagrange. Testes numéricos, utilizando os sistemas elétricos IEEE 14, 30 e 57 barras indicam que o método é eficiente e robusto.

ABSTRACT. A Abordagem Hıbrida(AH) consiste no calculo da direção de busca do Metodo primal-dual de Pontos Interiores (MPI) via o Metodo dos Gradientes Conjugados Precondicionado (MGCP) em duas fases, nas iterações iniciais usa o precondicionador Fatoração Controlada de Cholesky (FCC) e depois de um criterio de troca de fase o precondicionador Separador (PS) e usado. No FCC os parâmetros que controlam o preenchimento e a correção das falhas que ocorrem na diagonal sao modificados para reduzir o numero de reinıcios da fatoração durante a construção deste precondicionador. Propoe-se uma modificação nos parâmetros do precondicionador FCC com o objetivo de aprimorar a eficiência e robustez da AH. O calculo dos novos parâmetros é feito considerando a relação que existe entre as componentes da FCC obtida antes e depois da falha na diagonal. Experimentos numericos que resolveram problemas de programação linear de grande porte corroboraram que esta abordagem e ́ competitiva.

ABSTRACT. Con el objetivo de mejorar el desempeno del precondicionador Separador aplicado al sistema de Ecuaciones Normales en el metodo primal-dual de Puntos Interiores se propone una nueva ordenacion de las columnas de la matriz de restricciones de un problema de programacion lineal. Esta propuesta da énfasis al calculo de una base dispersa para el precondicionador Separador, adicionalmente se demuestra que con la ordenacion propuesta el numero de condicion de la matriz precondicionada es limitada uniformemente por una cantidad que únicamente depende de los datos del problema y no de la iteracion del metodo de Puntos Interiores. Los experimentos numericos se realizaron usando un precondicionamiento hıbrido de dos fases, en la primera fase se uso el precondicionador Factorizacion Controlada de Cholesky y en la segunda fase el precondicionador Separador. La implementacion de esta propuesta en problemas de gran tamano corroboro su robustez y eficiencia computacional.

ABSTRACT. Neste trabalho métodos iterativos precondicionados são utilizados para solução dos sistemas lineares dos métodos de pontos interiores com o objetivo final de resolver problemas de otimização linear de grande porte. Durante as iterações dos métodos de pontos interiores, a matriz de coeficientes se torna mal condicionada, ocasionando instabilidade numérica e dificuldades em encontrar a solução, mesmo utilizando métodos iterativos. Assim, a escolha de um precondicionador eficiente é essencial para o sucesso da abordagem. O trabalho propõe um novo critério de ordenamento das colunas para construção do precondicionador separador que preserva a estrutura esparsa da matriz de coeficientes original. Os resultados teóricos mostram que quando o ordenamento proposto é adotado, a matriz precondicionada tem o número de condição limitado. Estudos de caso mostram que a ideia é competitiva com métodos diretos pois o número de condição do sistema precondicionado é muito melhor do que o original.

ABSTRACT. Bike sharing systems allow people to take and later return a bicycle at one of many stations scattered around the city. Users naturally imbalance the system by creating demands in a highly asymmetric pattern. In order for bike sharing systems to meet the fluctuating demand for bicycles and parking docks, inventory optimization among stations is crucial. In this paper, we model a subset of stations of the bike sharing system in Mexico City (ECOBICI) using discrete-event simulation and we evaluate different heuristics for the initial allocation of bicycles. In addition, we compare those heuristics with a simulation-optimization method to assess the optimal number of bicycles that should be placed at each station during the rush-hour period to enable the system to comply with ECOBICI’s operational requirement. The results show that simulation-optimization method outperforms other heuristics.

ABSTRACT. En este trabajo se analiza un modelo EOQ de inventario para artículos con patrón de demanda dependiente tanto del precio de venta como del tiempo. Más concretamente, se considera que la tasa de demanda combina multiplicativamente los efectos de una función exponencial del precio unitario de venta y el patrón de demanda potencial. Además, se admite la posibilidad de escasez o rotura del stock. El objetivo es determinar el tamaño óptimo del lote, la longitud óptima del ciclo y el precio óptimo de venta que maximizan los beneficios por unidad de tiempo. Se desarrolla un procedimiento para determinar la solución óptima del problema en función de las diferentes combinaciones de valores de los parámetros del modelo.

ABSTRACT. Dust suppression of hauling roads in open pit mines is done by periodically spraying water from a water truck. The objective of this article is to present and compare two methods for locating water depots along the road network so that penalty costs for the lack of humidity in roads and routing costs are minimized. Because the demands are located on the arcs of the network and the arcs require service more than once in a time horizon, this problem belongs to the periodic capacitated arc routing domain. We compare two methods for finding the initial depot location. We then use an exchange algorithm to modify the initial location and an adaptive large neighborhood search algorithm to modify the initial routing of vehicles. This method is the first one used for depot location in periodic arc routing problems.

ABSTRACT. In the increasingly competitive distribution companies environment, it is necessary to efficiently manage the inventory levels, allowing an adequate satisfaction of the products’ requirements of customers while minimizing the financial investment needed. This problem has been studied in the literature by different authors and it has naturally caused the development of a set of models for optimal inventory management. In this paper, a model of optimal inventory management is presented, which determines the quantities and optimal times order for an inventory system of large size, ensuring that the expected lost sales are minimized to a certain level of financial investment and a given number of annual replenishment orders. Given the algebraic nature of the resulting nonlinear programming model, the optimal inventory management is solved by applying an algorithmic strategy based on augmented Lagrangian method of Hestenes-Powell-Rockafellar, obtaining good results.

ABSTRACT. Finding an optimal sequence for the harvest planning teams in the forest industry, is a really complex task, due to the high number of combinations that exists. In this talk we will present a new integer linear programming model with binary decision variables to solve this problem. Due to the high complexity posed by this problem, two methods that simplify this task are studied: one that generates and selects harvesting patterns which are added at once to a simplified harvesting model and one based in the Dantzig-Wolfe decomposition principle and column generation. We will show a relational database and a web platform to create flexible instances allowing us to compare and analyze the proposed solution strategies. The results show that increasing the number of units and harvesting teams, can generate time differences among the proposed strategies of up to 600% being the one based in Dantzig-Wolfe decomposition and column generation the most efficient one.

ABSTRACT. Neste trabalho propomos dois algoritmos exatos para resolver o problema do Conjunto Dominante Mínimo Conexo, ou Minimum Connected Dominating Set Problem (MCDSP), baseados em uma abordagem de branch-and-cut com uma estratégia de ramificação denominada de branch combinatório. Esses algoritmos são adaptações de um método existente de decomposição de Benders que será usado para comparação com os aqui propostos. Resultados computacionais para um grande conjunto de instâncias da literatura são apresentados e mostram que o algoritmo de branch-and-cut que propomos gera melhores resultados, em termos de tempo de execução, para grande parte das instâncias.

ABSTRACT. En este trabajo se aborda bajo un enfoque de programación lineal entera el problema de coloreo de aristas por etiquetado total. Se propone una formulación para este problema y familias de desigualdades válidas a partir de las cuales desarrollamos un algoritmo Branch-and-Cut. Con el objetivo de mejorar la eficiencia del algoritmo, desarrollamos heurísticas iniciales y primales. Finalmente, presentamos experiencia computacional que evidencia la buena performance del algoritmo.

ABSTRACT. The adjacent vertex distinguishable proper edge coloring problem is the problem of finding the minimum number of colors required for a proper edge coloring such that every pair of adjacent vertices is distinguishable. In this paper we propose an integer linear programming approach to solve the problem. We study the underlying polyhedron, develop a Branch and Cut algorithm and present computational results that show a very good performance of the algorithm on random graphs.

ABSTRACT. Uma das áreas de aplicação da Pesquisa Operacional que, nas últimas cinco décadas, apresentou um crescente número de publicações é o Planejamento Florestal. Começando na década de 60,com 4 artigos precursores,e tendo um importante marco no trabalho de Johnson e Scheurman em 1977. Porém,foi apenas no final da década de 80 que o Planejamento Florestal passou a ser tratado em seus diferentes níveis hierárquicos: Estratégico, Tático e Operacional. Destes três, o nível Operacional é o que apresenta maior complexidade e menor prazo para planejamento, necessitando de ferramentas robustas e ágeis para otimizá-lo. Uma das etapas que pode consumir muito tempo no desenvolvimento de ferramentas de otimização é a coleta dos dados do cenário florestal, devido à escassez destes dados na rede. Assim, o presente trabalho pretende apresentar uma ferramenta que crie cenários florestais virtuais, com base em dados e conhecimentos de cenários reais, com agilidade e, assim, auxiliar no desenvolvimento e comparação de diferentes métodos de otimização para o Planejamento Florestal em seu nível Operacional.

ABSTRACT. De acuerdo con el Alto Comisionado de las Naciones Unidas para los Refugiados el número de desplazados por la fuerza a 2014 fue de 59.5 millones a nivel mundial; en Colombia se reportaron más de 6 millones de desplazados por la fuerza a finales de 2014. La sociedad colombiana ha sido severamente afectadas por el conflicto armado que ha existido en el país durante las últimas décadas, este ha originado diversas problemáticas como el desplazamiento forzado hacia las ciudades capitales lo cual ha demandado la intervención del Estado para mitigar el sufrimiento de las víctimas. Para el Municipio de Medellín existe la norma regulatoria y clasificatoria para este tipo de asentamientos la cual es el Acuerdo 48 de 2014(POT) aprobado por el Concejo Municipal. Utilizando la simulación se pretende establecer las posibilidades de nuevos asentamiento de desplazados en las diferentes comunas de Medellín, utilizando información del POT e información catastral para determinar resultados de como las personas se establecen en sitios específicos de la ciudad y proponer políticas para mitigar el impacto en la sociedad de dichos asentamientos.

ABSTRACT. Industrialization is one of the causes of climate change, species extinction and other environmental impact all over the world. As the development of mankind can’t be stopped, production in Eco-Industrial Parks (EIP) is an improvement. Several companies are connected in an EIP with the goal of reducing the global impact (e.g. use of water or energy, waste generation) through material sharing. How to configure the most sustainable EIP? How to quantify environmental, social and economic dimensions of sustainability? This work presents a design of an EIP in Valdivia, Chile, where industries use water from Calle-Calle river. The design is modeled as a multi-objective optimization problem taking into account the three aspects of sustainability at once. The problem is solved using different methods: Weighting, Epsilon-Constraint and Goal Programming. The relative importance of each objective function is studied with MACBETH, a theoretical tool oriented to multi-criteria decision-making.

ABSTRACT. The classical job shop scheduling problem has existed for over 60 years. In this paper is summarized the recent mathematical models developed for addressing the job-shop scheduling problem under environmental constraints. It analyzes the mathematical models that capture the main environmental characteristics of the related processes and the solution methods developed so far. In this paper, the investigated problems are defined and classified, modeling gaps are highlighted and future research developments are proposed

ABSTRACT. Air cargo transportation represents an important source of income for airlines. Cargo can be transported via freight planes or in the belly aircrafts for passengers (\belly cargo"). One of the main difficulties faced by the belly cargo mode is the uncertainty on the available capacity: the actual availability for cargo is only known after the passengers have boarded and had their luggage loaded. We propose a multistage stochastic programming model whose goal is to find the optimal allocation of cargo to the passengers network in order to maximize prot. We solve the resulting problem using decomposition techniques. In addition, we develop a policy to use the solution in real instances using data from a major commercial airline. The policy achieved higher expected profits, better level of service, and a decrease in the variability of the results when compared against its deterministic counterpart.

ABSTRACT. The salmon industry is a major economic activity in Chile. At the fattening stage, the fish are farmed in big floating cages. These farms are located along the southern coast of Chile.

At the time of harvest, the fish must be transported to the gathering centers. These centers contain big cages where fish stock is kept to be sent to the slaughter plants. Each gathering center is located near a slaughter plant, where the fish are killed and cleaned. Then are loaded on refrigerated trucks to be sent to the processing plants. These plants represent the end of the process chain.

A big challenge for the salmon companies is the planning of the maritime and terrestrial logistics, combined with the stock planning at the gathering centers. In this work, we show two approaches for this problem. The first one consists of an integer programming model. The second one is based on an heuristical approach. The results on the application of them to a salmon farming company are shown.

ABSTRACT. Shippers at two points, A and B, trade products with an imbalance: the flow from A to B is greater than the flow from B to A. We postulate that the shipper at A faces an implicit negotiation with the shipper at B regarding which party must bear the cost of a vehicle traveling between the two points. Following the Nash bargaining model, we deduce that the ratio of the A-to-B freight rate to the B-to-A rate changes linearly with the ratio of the flow from B to A to the flow from A to B. We test this result for the Transatlantic, Transpacific and Europe-Asia maritime routes using quarterly data for the 2003-2012 period and validate this hypothesis for two of these routes.

ABSTRACT. Park-and-rides (P&Rs) are facilities where users park their cars to connect to transit modes. An important aspect when designing P&R services is the location of the P&R facility. The optimal selection of P&R location can be defined as following: Given a demand of trips and given a set of candidate locations for P&Rs, we select the best subset of a given size that maximizes expected demand captured by the P&Rs, where the demand can be estimated with a Multinomial Logit function. In this work we propose a column generation method for this problem, and we present a case study for P&R facility locations in NYC, covering 28 counties in NY, NJ, and CT. The evaluation seeks to find five P&R facilities that maximize demand for destinations in Manhattan among 59 candidates (21 in Queens, 13 in Bronx, 13 in Brooklyn, and 12 in Staten Island). Our proposed algorithm is able to solve this large-sized instance and provide results consistent with those obtained by previous studies.

ABSTRACT. En este trabajo se elaboró un modelo de simulación mediante Dinámica de Sistemas sobre la producción, almacenamiento, consumo, exportación e importación de gasolina y diésel en Venezuela, utilizando el software Vensim. El modelo posee una estructura genérica para el combustible, el parque automotor, el precio, el contrabando, la escasez, el subsidio y otra para relacionar el presupuesto para el subsidio con el ingreso y gasto. El modelo fue simulado en el período 2005-2014. Fueron planteados algunos escenarios para el período 2015-2021, con ajuste de precios, variaciones en la elasticidad precio de la demanda y racionamiento de gasolina de 91 octanos. Los resultados indican que deben iniciarse ajustes de precio en el corto plazo para evitar la escasez. Si se utiliza racionamiento, los precios tendrían que ajustarse en años posteriores. El modelo permite proponer o evaluar políticas de Estado asociadas con los precios del combustible, que pueden representar ahorro para el mismo.

ABSTRACT. El análisis de seguridad nuclear, se apoya en herramientas de cálculo realistas teniendo en cuenta las incertidumbres asociadas, como Best Estimate Plus Uncertainties. El objetivo es estudiar la respuesta de los sistemas nucleares en operación y accidente teniendo en cuenta las incertidumbres a través de modelos de simulación. En este contexto, la técnica de generación de números aleatorios utilizada en los modelos de simulación permitirá resultados más robustos en el análisis de sensibilidad posterior. El Análisis de Sensibilidad Global, permite evaluar la importancia de un gran número de incertidumbres pero con un gran coste computacional. Este se puede reducir utilizando Meta-Modelos, los cuales permiten calcular los índices de Sobol, con un adecuado equilibrio entre precisión y esfuerzo computacional. En esta ponencia se presenta los resultados del análisis de sensibilidad aplicado en la simulación de un escenario de accidente LBLOCA en un PWR, utilizando Meta-Modelos.

ABSTRACT. We are interested in analyzing decisions making processes related to track dispatching in open-pit mining systems. A truck cycle in a mining operation include loading, traveling to an unloading site (crusher or dumpsite), unloading, traveling to shovel, queuing and spotting. Every time a truck is unload, a decision must be made on where to assign it next. Existing methods available to solve this problem lack the capacity of estimating the impact of present decisions on future ones, and there are unable to generate accurate solutions. We propose three dispatching methods based on online simulation, that have less deviations from operational targets and ore production rates, when compared to existing methods.

ABSTRACT. Desenvolveremos um modelo que simula o comportamento dos frangos em um aviário. Os parâmetros de entrada para o modelo serão: tamanho do aviário, número de frangos, quantidade e posição dos comedouros e bebedouros e temperatura média do aviário. Os parâmetros internos são: fome, sede, sensação térmica e posição de cada frango, e conforto e tamanho dos frangos. Usaremos dinâmica de Langevin para descrever o movimento dos frangos. As forças envolvidas na simulação serão dadas pelas interações entre os frangos e as interações dos frangos com o ambiente. As interações entre os frangos serão de curto alcance. Estas serão para evitar que dois frangos ocupem a mesma área em um mesmo instante. As interações com o ambiente serão causadas pela fome, sede e sensação térmica. Estes parâmetros também nos darão o conforto dos frangos. Essa simulação facilitará a obtenção de dados de aviários, além de poder ser usada para criar funções de conforto.

ABSTRACT. Sport directly or indirectly affects the lives of millions of people around the world. For professional federations, the quality of the schedules strongly affects their financial turnover (media revenue, sales at stadiums, ticket sales etc.). For non-professional tournaments, it affects the distribution of games, tournament length, travelling distances, associated costs etc. Scheduling sport competitions is a hard combinatorial optimisation problem studied at length by researchers, at least for some problem variants. In this position paper, we argue that there are still several challenges in this field that are worthy of attention. On one side, ‘traditional’ sport scheduling problems (SSPs) still present open research questions that should be investigated. In addition, we argue that there has been a skewed focus towards the SSPs of interest for professional sport federations, while other SSPs have received less attention despite their theoretical and practical interest.

ABSTRACT. We study scheduling problems when jobs can be split and a setup is required before processing each part, to minimize the weighted sum of completion times. Using a simple splitting strategy and a reduction to an orders scheduling problem we derive a 2-approximation algorithm for the case with uniform weights and setups, improving upon previous work. We extend this idea to the general identical machine case and conclude by designing a constant factor approximation algorithm when machines are unrelated.

ABSTRACT. Se propone un modelo de optimización basado en programación entera-mixta para diseñar ciclos de intercambio voluntario de riñones. La literatura científica aborda este contexto como el problema de encontrar un paciente receptor para cada uno de los donantes que se postulan en el programa, y quienes en retorno esperan que un familiar o amigo reciba otro riñón sano. El modelamiento matemático se realiza sobre un grafo incompleto y dirigido. El objetivo es encontrar el conjunto de ciclos en el grafo que maximiza el número de nodos que hacen parte de los ciclos con restricciones de longitud de los mismos. Esta propuesta incluye optimizar la secuencia en la que se realizan las etapas del ciclo y considerar costos adicionales por el riesgo asociado al orden en que se realizan los intercambios. Se presentan resultados preliminares de diferentes formulaciones matemáticas propuestas para este problema y se comparan analizando el tamaño máximo de instancias que se logran resolver.

ABSTRACT. Muchos tratamientos se han enfocado solo a controlar los niveles de glucosa en la sangre de un paciente con diabetes en ayunas, sin embargo, estudios recientes han relacionado los altos niveles de glucosa que se generan después de comer al riesgo de complicaciones diabéticas, por lo que resulta ya insuficiente tratar a una persona con este padecimiento teniendo en cuenta solo su estado en ayuno. En este proyecto se optimiza el índice glucémico a través de la sugerencia de los alimentos a ingerir en una dieta de 1800 calorías, tras la ingesta de los alimentos (posprandial), tomando en cuenta su aporte energético, así como su contenido en grasas, proteínas y principalmente los carbohidratos. El modelo de optimización se basa en la minimización del índice glucémico sin considerar los costos de los alimentos. A partir del índice glucémico obtenido se calcula la carga glucémica para finalmente dar una propuesta de dieta.

ABSTRACT. Some of end-stage renal disease patients have relatives/friends who are willing to donate a kidney to them, but the kidneys may not be compatible. Since the same case arises for many other patients, the idea of a kidney exchange pool with those incompatible patient-donor pairs consists of finding incompatible donor pairs matchings so as to maximize the number of exchanges. Given requirements by law or by surgery rooms availability, we may find exchanges either in the form of cycles and/or chains. For this NP-hard problem, we compared the most recent integer programming formulations for unrestricted cardinality chains and restricted cardinality cycles. The formulations were tested in a set of instances from literature and a real-world set from a hospital in Mexico.

ABSTRACT. We study the pricing strategies of firms belonging to a vertical distribution structure (marketing channel or supply chain) where a base and an optional contingent products are sold. Optional contingent products are characterized by unilateral demand interdependencies. That is, the base product can be used independently of a contingent product. On the other hand, the contingent product's purchase is conditional on the possession of the base product. The problem is inspired from a pricing controversy in the e-book industry. We characterize the conditions under which the retailer practices a loss-leader pricing strategy, as well as when optional contingency mitigates the double-marginalization problem. We compare two business models, namely, wholesale pricing and revenue sharing.

ABSTRACT. El reciente aumento de la globalización, los diversos eventos ocurridos (ataques terroristas, catástrofes naturales, incendios, etc.) y las acciones tendientes a aumentar la productividad en las organizaciones, han causado que su vulnerabilidad a imprevistos aumente considerablemente. Para tratar de contrarrestar y reducir dicha situación ha surgido la Gestión de Riesgos en la Cadena de Suministro (SCRM – Supply Chain Risk Management) como un proceso que permita evaluar la situación actual del negocio e implementar actividades y/o estrategias para disminuir su vulnerabilidad. Sin embargo, dada la misma incertidumbre que gobierna las actividades entorno a la Cadena de Suministro, evaluar el resultado que conllevaría la implementación de dichas estrategias y del mismo proceso SCRM es difícil y en ocasiones recae sobre los gerentes o un grupo de expertos. Por tal motivo, se ha analizado la creación de una herramienta que permita evaluar de forma a priori la selección de las estrategias.

ABSTRACT. Conectar el desempeño productivo de la cadena de suministro con el rendimiento financiero de las compañías es de gran relevancia para la toma de decisiones, pero la literatura en este tema es escasa. Nosotros proponemos un modelo de Dinámica de Sistemas para estudiar el impacto de la gestión de la cadena de suministro en los resultados financieros de un negocio de cárnicos en Colombia. El modelo es utilizado para simular diferentes escenarios de demanda, permitiendo evaluar el impacto de diferentes políticas de inventario en el nivel de servicio, el capital de trabajo, la capacidad, ingresos, egresos, y el EVA generado. El modelo desarrollado permite comprender el efecto de los retardos en las decisiones operativas de la cadena, convirtiéndose en una herramienta gerencial de alto valor para revisar políticas y orientar proyectos de mejora que impacten positivamente el EVA y las finanzas del negocio.

ABSTRACT. Este trabajo usa un enfoque biobjetivo para planificar los flujos en una red de suministro con pérdidas de material tratando de, simultáneamente, minimizar los costes y las pérdidas de material. Para ello se va a usar un enfoque de Análisis por Envoltura de Datos (DEA) considerando una red de procesos formada por los diferentes nodos y arcos de la cadena de suministro. Se supone que están disponibles observaciones de los flujos de material en una serie de periodos anteriores, lo cuales se van a usar para definir la tecnología DEA, esto es, los puntos de operación factibles. El modelo biobjetivo propuesto busca dentro de ese conjunto de puntos de operación factibles aquel que da lugar a menores costes y menores pérdidas de material.

ABSTRACT. Abordamos el problema de optimizar una red de nodos para la simulación de fluidos; particularmente, en la etapa de preprocesamiento de la simulación se diseña la red; cuya mala calidad de sus celdas poligonales incluye obtener malos resultados de convergencia, precisión y estabilidad en la simulación. Sin embargo, obtener una red de buena calidad requiere bastante tiempo de cómputo y gran capacidad de memoria. Además, diseñar una red de buena calidad implica minimizar el sesgo equiángulo. Esta investigación tiene como objetivo diseñar una metaheurística aplicando el procedimiento de búsqueda avariciosa, aleatoria y adaptativa, GRASP por sus siglas en inglés, que contenga un evaluador del sesgo y que mida la desviación de los ángulos internos de una celda poligonal irregular con respecto a una celda poligonal regular; de tal forma que el evaluador pueda ser una herramienta para explorar las celdas y elija las mejores de la red para minimizar el sesgo equiángulo en un tiempo razonable.

ABSTRACT. A diario nos enfrentamos a diferentes dinámicas que ponen a prueba nuestra mente. Daniel Kahnenam reconoce dos estructuras mentales asociadas a la toma de decisiones: la rápida, donde nuestra experiencia es la que nos guía, y la lenta, donde las acciones son meditadas y exhaustivamente racionales. Sin embargo, disponemos de un “presupuesto limitado" que podemos asignar a ciertas actividades y si se excede, se fracasa. En esta investigación se busca determinar mediante el uso de información de Electro Encéfalo Gramas (EEG) el momento en que nos encontramos en cada uno de estos "Sistemas". Utilizando herramientas de investigación operativa se busca caracterizar funciones y patrones matemáticos, entre los sensores del encefalograma, que nos permita discernir en qué Sistema se está operando en cada momento y su precisión en la predicción.

ABSTRACT. This article presents an evolutionary approach for the optimization of a computer player for video games. We develop an artificial intelligence for playing the pinball game and then optimize it applying a parallel evolutionary algorithm. Our proposal is compared against an approach which uses backtracking and local search, and against human players. This evolutionary computer player is able to improve between 43% and 541% over the results obtained by human players and also improves over backtracking/local search results.

ABSTRACT. We introduce a new combinatorial optimization problem referred to as socio-technical team formation problem (STFP), which arises from the formation of cooperatives teams by considering social and technical constraints between individuals. We initially prove the NP-hardness of STFP. Then, we present a binary integer linear programming formulation for the underlying problem and conduct a polyhedral study by deriving some facet-inducing inequalities related to its polytope. Three exact approaches based on the proposed formulation are tested on a set of 120 randomly generated instances with known optimal value. The experiments attest the efficiency of our implementations.

ABSTRACT. O objetivo deste artigo é estabelecer o posicionamento contábil de empresas de metalurgia e siderurgia listadas na BM&FBovespa utilizando a teoria dos jogos. São usados quatro lotes de indicadores econômico-financeiros: liquidez, endividamento, rentabilidade e atividade. Para tal, as empresas representam as estratégias do jogador I e os indicadores econômico-financeiros representam as estratégias do jogador II. Para determinar os posicionamentos foram aplicados três modelos, os quais pertencem a família dos jogos multicriteriais. Para o primeiro modelo foram gerados os rankings R1 e R2, para o segundo, os rankings R3 e R4 e para o terceiro modelo, o ranking R5, totalizando cinco rankings. Os resultados mostram que os rankings obtidos refletem os dados coletados, ou seja, observa-se a adequação dos modelos aplicados. Conclui-se que é possível aplicar os modelos de Teoria dos Jogos para estabelecer posicionamentos de empresas

ABSTRACT. Diverse problems in the area of optimal resource management, exploration of natural reserves, pension fund management, and environmental protection can naturally be formulated under a unified framework, as stochastic control problems of a specific type. We show that solutions to these problems are characterized in terms of fixed-point equations for martingales, where the martingale fixed-point process represents a marginal price of the resource remaining in the reserve. We address the existence of the martingale fixed point and elaborate on efficient solutions of our control problems.

ABSTRACT. In many sectors of the economy, oil is directly present in the production chain. In other sectors, it is present in the production chain as the necessary energy to manufacture goods. This way the crude oil market is related to the macroeconomic indicators such as inflation index, industrial production indicators, exchange rate and capital market index. Many researches aim at establishing the stochastic process that can represent the movements of macroeconomic indicators through the oil price returns or variations that have been done in recent years. The purpose of this work is to study the relationship between crude oil prices and selected macroeconomic indicators of the Brazilian economy. To do that this work carried out cointegration and causality tests, from VAR estimations, and impulse response analysis. The data used in this study is monthly Brazilian macroeconomic indicators and the Brent oil type price in US$. The period of the sample used is from January 2002 to October 2015.

ABSTRACT. Los problemas de optimización combinatoria binaria son difíciles conforme crece el tamaño de los casos y un algoritmo exacto tardaría mucho en resolverlos. Por ello, es necesario desarrollar estrategias heurísticas que permitan encontrar una buena solución factible. En el presente trabajo se muestra un modelo probabilístico basado en la teoría de campo medio, el cual involucra la función objetivo y sus restricciones. Modelo que satisface las condiciones de Karush-Kuhn-Tucker mediante la relajación de un problema binario. Se presenta un algoritmo basado en dicho modelo, el cual obtiene cotas superiores e inferiores a partir de una solución factible. Los problemas de prueba seleccionados fueron KP, QKP y MKP. Se usaron casos reportados en la literatura y otros creados mediante el generador de Pisinger. Los resultados muestran que la diferencia entre las cotas es pequeña en ciertos casos, inclusive en tamaños de grandes dimensiones.

ABSTRACT. This research attempts to solve the problem of finding a “good” bassline to accompany a given melody and a given drumbeat following the structure of the alternative rock genre. The solution method is made up of different steps which include: determining the key of the song by using a matching algorithm that compares the notes in the melody with the notes that made up each key, fitting chords to the melody by constructing a graph of possible chord progressions using candidate chords selected through a matching algorithm and finding the shortest path in it, improvising basslines using these chords as guidelines through the use of stochastic diffusion search and simulated annealing algorithms, resulting in a combination of exploration and intensification techniques. Solutions must satisfy various constraints regarding their relationship with the drumbeat while maximizing the quality of the overall sound, measured as the bass/melody relationship.

ABSTRACT. This work presents a study of different opposite learning strategies for solving combinatorial problems using ant based algorithms. We propose a two phases approach: A prior opposite learning phase for discarding regions of the search space. In the second phase, this opposite knowledge is used by the selected ant based algorithm for solving the original problem. The objective is to improve the search process of the ant algorithm maintaining its original design. We evaluate different strategies in terms of how opposite solutions can be constructed. The results obtained show that the opposition-based search help for improving the quality of the solutions found by these ant algorithms. Moreover, the best performance has been obtained using the Soft Opposite Learning strategy.

ABSTRACT. Neste trabalho apresentamos um método heurístico, baseado nas ideias de Restauração Inexata, para solução de problemas MINLP. No algoritmo incorporamos a estratégia Feasibility Pump na fase de restauração, onde controlamos a viabilidade do iterando, e na fase de minimização buscamos melhorar o valor da função objetivo resolvendo um MILP com as restrições não lineares relaxadas e o valor do gradiente no ponto restaurado como vetor de custos. Além disso, adicionamos uma restrição ao problema que exige decréscimo no valor da função objetivo. O método foi implementado no MATLAB, onde utilizamos o algoritmo de Programação Quadrática Sequencial, incorporado na rotina fmincon, para resolução dos Problemas Não Lineares e fizemos a interface com o Cplex para resolver os MILPs. Testes realizados com problemas da biblioteca MINLPlib sugerem que o algoritmo é eficiente na localização de minimizadores locais, encontrando a melhor solução conhecida para boa parte dos problemas testados.

ABSTRACT. This article proposes the design and implementation of a parallel evolutionary algorithm for public transport optimization on Garzon Avenue, Montevideo, Uruguay. This is an interesting complex urban scenario, due to the number of crossings, streets, and traffic lights in the zone. We introduce an evolutionary algorithm to efficiently synchronize traffic lights and improve the average speed of buses and other vehicles. The experimental analysis compares the numerical results of the evolutionary algorithm against a baseline scenario that models the current reality. The results show that the proposed evolutionary algorithm achieves better quality of service improving up to 15.3% average bus speed and 24.8% the average speed of other vehicles.

ABSTRACT. Los operadores del sistema de transporte público de la ciudad de Santiago deciden los horarios de salida de sus buses cumpliendo con las restricciones operativas que imponen las autoridades de Transantiago, en particular, respecto a frecuencias como capacidades ofrecidas, dentro de ciertos rangos por tramo horario. Como la operación basada en frecuencias podría resultar en tiempos de espera muy elevados e incertidumbre para el usuario, y por lo tanto, en un nivel de servicio deficiente para los pasajeros, Transantiago definió como política de operación, que algunos de los servicios nocturnos operen con horarios predeterminados y sean coordinados. En este trabajo, se propone un modelo de programación entera mixta para definir horarios (timetables) en las diferentes estaciones de transferencia. Estos horarios consideran la opción de retener algunos buses en ciertas estaciones, de forma tal de poder realizar las transferencias y sincronizar mejor los recorridos de diferentes líneas.

ABSTRACT. Rapid transit network design is highly dependent on the future system usage. These spatially distributed systems are vulnerable to disruptions: during daily operations dierent incidents may occur. Despite the unpredictable nature of them, efective mitigation methods from an engineering perspective should be designed. In this paper, we present several risk averse measures for risk reduction in the rapid transit network design problem based on a set of finite scenarios to represent the disruptions’ uncertainty. As a counter-parts of the typical risk neutral strategy, some measures that are presented are aiming to minimizing the impacts of the worst scenario in the network operation, and another additionally, takes into account dierent risk reduction profiles. Some computational experience is presented.

ABSTRACT. In this paper, we evaluate the effectiveness of a non-linear tariffs contract to achieve perfect coordination in a supply chain in the presence of inventory decisions and price sensitive final demands. In the simplest model, we focus on a distribution channel formed by a single wholesaler and a single retailer who acts independently and maximize their own benefits under complete information. We show that when the wholesaler can establish a contract implementing non-linear tariffs with the retailer, we can easily reproduce the benefits of vertical integration. The strategy is not only simpler than alternative approaches found in the literature, but it is also easily extended to more general settings as the one presented in the second part of this article, where the solution arises as a Nash equilibrium.

ABSTRACT. We build on existing work on product positioning and service development, and present a new method for service positioning that considers both tangible and intangible attributes. The new service positioning method is based on a product positioning method proposed in the literature which uses the neural network methodology of Kohonen's self-organizing map. The method incorporates the concept of rings of influence, where a firm evaluates individual consumers and decides on the intensity to pursue a consumer, based on the probability that this consumer will purchase a competing product/service. The method has several advantages over earlier work. First, no limitations are imposed on the number of competing services and second, the method can position multiple services in multiple market segments. We present a number of examples of applying the method, demonstrating the computational efficiency of the method.

ABSTRACT. Este paper analiza el problema que enfrenta una empresa que debe determinar el precio y la composición óptima de un grupo de paquetes de productos/servicios que desea ofrecer en múltiples segmentos de mercado. El problema es enfrentado como un problema de programación no lineal en variables mixtas, el que enfrentamos mediante el uso de Metaheurísticas. Diseñamos un método de dos etapas: (1) búsqueda de una solución de partida determinando la composición de los paquetes a través del uso del método propuesto por Bitran y Ferrer(2007), y fijando el precio de estos paquetes mediante una optimización por gradiente conjugado y (2) mejoramiento a través del uso de diversas Metaheurísticas: Tabu Search, Simulating Annealing y Genetic Algorithm. Los resultados obtenidos muestran que se generan beneficios económicos adicionales mediante nuestro enfoque. En nuestro ejemplo de trabajo, vemos que el aumento de la utilidad es de cerca de un 18%.

ABSTRACT. Estimating customers' learning from their experiences has attracted researchers interests in recent years due to its importance for making investment decisions to improve service quality. In this paper, we propose a novel and general Bayesian hierarchical learning framework for estimating customers' spill-over learning. Here we define "spill-over" learning as customers' learning from their previous experiences of similar yet not identical service encounters. Specifically, we apply our model to a one-year shipping/sales historical data provided by a world-leading third party logistics company. We estimate customers' learning from adverse shipping experiences with this logistics company by modeling customers' purchasing decisions, such as shipping frequency and third-party logistics company choice. By studying customers reactions to adverse experiences, we are able to attach a dollar value to providing favorable and unfavorable service experiences to clients.

ABSTRACT. We present a two-stage MIP modeling method to determine an optimal supply chain network for converting food waste into energy and other value products. Stage-1 groups the waste sources into multiple compact clusters each served by a single pre-treatment facility. Stage-2 uses this information as input and finds the optimum routing of delivery vehicles within each cluster. The result is an approximation to the true optimum solution, which may be computationally intractable when working with too many waste sources, but we find that the accuracy is good. We present an empirical application to a large metropolitan area, Shenzhen, China. Currently the food waste is sent to landfills and part of its oil component is recycled and marketed as cooking oil by an illegal sector. Due to health safety concerns, the City plans to develop a centralized food waste-to-energy system. Our results show that an optimally designed system can be economically viable and also provides environmental benefits.

ABSTRACT. El presente estudio explora la viabilidad de aplicación de un método de apoyo a la decisión en el ámbito de la gobernanza ambiental y la democracia participativa para el manejo y conservación del capital natural y la biodiversidad en un caso de estudio en la Reserva de la Biósfera Sierra del Abra Tanchipa en México, como parte de los esfuerzos institucionales y de Organizaciones no Gubernamentales para promover el empoderamiento de los actores. Se resalta la efectividad de los mecanismos de participación deliberativos para la definición de criterios de decisión y el logro de consensos grupales.

ABSTRACT. Empirical research has shown that airside operations form a significant percentage of overall airport related emissions. Among those operations taxiing is one of the most emission-intensive processes, closely related to the initiating pushback process – that may determine the taxiing duration. Possible approaches for this problem are simulation and optimization models for an effective management of pushbacks as resources at the airport. Airside operations at major airports involve a complex interplay of many operations and parties and therefore need to be planned and optimized in a coordinated fashion. Yet, existing approaches have not been applied in a comprehensive planning environment for airside operations. In this work, we develop an algorithm-based relocation approach for pushback vehicles that enables an effective minimization of waiting times and emissions during the taxiing process. These algorithms are evaluated in a realistic simulation environment of a real-world case.

ABSTRACT. We introduce an efficient approach to evaluate counterparty risk and we compute the Credit Valuation Adjustement for derivatives having early exercise features. The computation model is based on a Dynamic Programming representation of the both the vulnerable and the risk-free Financial instruments, coupled with spectral interpolation of the value function. The approach is very flexible and can account for wrong-way risk and various models for the underlying risk factor's dynamics. Numerical experiments are presented to illustrate the versatility of the method and its efficiency with respect to the approaches that are currently used in the industry.

ABSTRACT. The copper spot price is the determining guideline parameter for trading of this commodity in the international market and the derivatives market are crucial in the formation of the spot market price. This work aims to study the relationship between copper spot price and the copper futures contracts traded on the LME in London, seeking to establish a stochastic relationship that allows to obtain this metal price expectations in the short term through the future price. For the construction of more adequately price forecast models, copper parices time series in the spot and futures markets are studied.re. The construction of these prediction models took into account other hypothesis such as non-normality and the conditional heteroskedasticity of price and return time series. The data used in this study comprise time series of daily copper prices traded in the spot and the futures markets, for three months ahead, spaning from of September 2006 to March 2016 in US$ per ton.

ABSTRACT. The valuation of commodities contingent claims depends on the drift of the process (futures) but also on the volatility (options). We compare the empirical performance of a model with constant volatility (CV) with one with a stochastic volatility specification (SV)for oil, copper and gold. First, the CV model is a better alternative for futures but for options errors may be large. Second, the longer the option maturity, the less relevant are the differences in pricing errors. Third, the higher complexity of the SV model is reflected in larger execution times. Fourth, the results of the SV model, applied for the first time to gold and copper, strongly suggest the presence of unspanned stochastic volatility components in all three commodities. Choosing the best model to implement in a real situation depends on the objectives pursued and on the tradeoffs between effort and precision. The results presented are not only new, but also relevant from a practitioner point of view.

ABSTRACT. A saúde pública está vivenciando um grande problema atualmente, a proliferação do mosquito Aedes Aegypti, transmissor da dengue, além de ter sido associado a outras doenças, entre elas a febre Chikungunya e a febre pelo vírus Zika. Este estudo propor um modelo multicritério para priorização dos bairros de uma cidade quanto à atenção no combate à proliferação do vetor. Para a aplicação, foi utilizada a metodologia PROMETHEE GDDS para obter uma ordenação final para os bairros que represente o ponto de vista dos diversos decisores envolvidos. Portanto, ao diagnosticar quais os níveis de prioridades existentes, é possível destinar de maneira mais eficaz os recursos públicos.

ABSTRACT. A crescente demanda por cursos profissionalizantes no Brasil impulsionou o surgimento de cursos técnicos, em diversas áreas, para atender a curto e médio prazo a escassez de mão de obra qualificada para as indústrias no país. O atual cenário político-econômico brasileiro, culminou na redução orçamentária para a educação subsidiada por programas federais, entre eles, o Programa Nacional de Acesso ao Ensino Técnico e Emprego (PRONATEC). Desta forma existe a necessidade de reestruturação tanto de gastos quanto de cursos a serem ofertados pelas instituições credenciadas para atender a demanda do programa. Este artigo tem como objetivo identificar quais as melhores alternativas, para determinação de quais cursos técnicos, devem ser mantidos em uma rede de escolas técnicas profissionalizante. O problema decisório é estruturado com a abordagem Value Focused Thinking (VFT) e resolvido com o método Analytic Hierarchy Process (AHP). com abordagem Ratings.

ABSTRACT. A busca por soluções tecnológicas que oportunizem economia de recursos humanos e de materiais é fundamental para o desenvolvimento sustentável da sociedade, das indústrias e para preservação do meio ambiente. O atual cenário político-econômico brasileiro, a crise hídrica e intervenções equivocadas do governo no setor elétrico resultaram num significativo aumento dos custos com eletricidade para toda a população brasileira, e para as indústrias. Este artigo tem como objetivo apresentar um estudo de caso para identificar quais as melhores alternativas para minimização dos custos em energia elétrica nas indústrias. O problema decisório é estruturado com a abordagem Value Focused Thinking (VFT) e resolvido com o método Simple Multiattribute Rating Technique (SMART). Dois cenários foram gerados para uma melhor Análise de Sensibilidade.

ABSTRACT. In different industries where the demands of end customers for faster and more reliable deliveries, better quality and lower prices increases, success strongly relies on supplier development. In this article, the TODIM multicriteria method is used to propose a ranking of suppliers for development in a real context defined by the Mexican aerospace industry. Given that poor information on available input data is observed, an extension of the TODIM multicriteria method is used to solve the problem. It utilizes preference thresholds to model imprecision and Monte Carlo simulation to deal with uncertainty. Outcomes of the standard TODIM and the extension are compared, which reveals the effect of both imprecision and uncertainty.

ABSTRACT. We present here an uncertain version of the classical minimum spanning tree problem. We are given a graph G=(V,E) embedded in a d-dimensional space. For each vertex v, we consider a lp-ball (neighborhood) centered at v and with a given radius. The neighborhood (ball) models the uncertainty in the vertex location. The goal is then to find a minimum spanning tree of the graph whose vertices (that must be determined) belong to each of the balls (each neighborhood contributes exact one node to the tree), the edges are labeled by E, and the edges-weights are the (unknown) lq-norm distances between the (also unknown) nodes. Hence, several decision have to be taken simultaneously: (1) the optimal location of the nodes of the tree; (2) the lq-norm distances between those points; and (3) the minimum spanning tree for the new graph. The problem is known to be NP-hard even for polyhedral norms (l1). We present different Mixed Integer Non Linear Programming (MINLP) formulations for the problem and different decomposition strategies to solve it efficiently. Some preliminary computational experiment run in Gurobi are reported. Finally, we present different versions of the problem that allows a similar treatment, e.g., if the distances between nodes are measured as the maximum distance between balls, or the case when the radius of the neighborhoods also vary (contributing to the overall cost of the tree).

ABSTRACT. We consider the problem of finding an optimum schedule for restoring a transportation network destroyed or damaged as a result of an extreme event (disaster). The extreme event destroyed connectivity between vertices which needs to be restored as soon as possible. For each pair of vertices, a due date for restoring the connectivity of the pair is given. The restoration speed, defined by available resources, is constant. The restoration crews can access any part of the network at any time. It is required to find an optimal schedule of restoration activities that minimizes the maximum lateness of vertex pairs. We discuss complexity of the problem and its structural properties, a mixed-integer linear programming formulation, and present a branch-and-bound exact algorithm and results of computational experiments.

ABSTRACT. We propose an approach to design min-cost routes for a vehicle moving a commodity between customers, allowing two characteristics uncommon in the pickup-and-delivery literature. One characteristic is that a customer is allowed to be visited several times. The other characteristic is that customers may be used as intermediate locations to temporarily collect and deliver product. We describe a math-heuristic approach that iteratively applies a constructive phase and an improvement phase. The constructive phase represents each customer by a set of nodes, decomposes each customer demand into partial demands associated to its nodes, and solves a one-commodity pickup-and-delivery TSP with a variable neighbourhood search. The improvement phase is a branch-and-cut procedure to optimize some partial routes of a given initial solution. Exhaustive computational results on benchmark instances demostraste the good performance of the approaches when solving instances with up to 500 customers.

ABSTRACT. We present mixed integer linear optimization models with a polynomial number of variables and constraints for combinatorial optimization problems in graphs: optimum elementary cycles, optimum elementary paths even in a graph with negative cycles, and optimum trees (whose Steiner tree problem) problems. When we have to solve small and median size combinatorial optimization problems using commercial codes of mixed integer linear optimization as CPLEX, XPRESS, GUROBI, etc., the use of branch and cut techniques needs a hard work of implementation. For the traveling salesman problem (TSP) and the Steiner tree problem (STP) in graphs, one uses mathematical models with an exponential number of constraints to avoid subtours, these constraints are generated step by step. We present 0-1 mixed linear optimization problems using nonsimultaneous flows for avoiding subtours. For the Steiner tree problem in graphs, we present interesting numerical results of linear relaxations of these models.

ABSTRACT. This research addresses a real scheduling problem for chemotherapy patients. The problem presented is divided in two. The first one refers to how to allocate treatment sessions in time and the second part is related to the daily operation of the Chemotherapy unit, how daily schedules are established for every patient, and how the available capacity (treatment chairs and hours of nursing care) is allocated. We consider the requirement for available lab hours to prepare the medicine for patients as an additional complexity. The first one is based on previous research conducted by Sauré et al. (2012), which implements the scheduling policy for chemotherapy, suggested by authors for radiotherapy treatments. The result of this first stage is input for the second stage, which consists in solving a problem of parallel machine programming with additional capacity constraints. This stage is addressed by generating treatment patterns.

ABSTRACT. La optimización de la producción de concentrados de plaquetas sanguíneas es importante para conseguir una mejor utilización de los recursos médicos. La gestión de su inventario es complicada debido a la corta caducidad, cinco días, a que sólo se produce en días laborables y se transfunde todos los días, y a que la demanda es incierta y no admite demora. Esto suele generar un exceso de producción, lo cual conlleva unas altas tasas de caducidad y un aumento de los costes económicos, dado que se trata de un producto caro. Se presenta aquí un modelo de producción de concentrados de plaquetas que toma como referencia la práctica del Centro Vasco de Transfusión y Tejidos Humanos (CVTTH). El modelo sigue una política FIFO de suministro y estima la producción y caducidades diarias con métodos estadísticos y probabilísticos. El resultado de su aplicación en el CVTTH es una reducción del 70 % de las caducidades y una disminución de un día de la edad de las plaquetas transfundidas.

ABSTRACT. Logistic problems faced by a medical emergencies services provider are usually split in vehicle location, dispatching and relocation decisions. The first one defines the system kick-off, that is, it identifies where the vehicles must stay waiting for a call, in order to satisfy or pursue given performance indicators. The second one selects the vehicle that must be sent to provide the service; and finally, once a vehicle leaves its position, the need to relocate the vehicles is verified and the required movements must be determined. This work develops a vehicle relocation mathematical model that fits the operational conditions of a provider of these services in Cali, Colombia. It adopts recent approaches that include a preparedness measure that is used as a minimum threshold. Our proposal optimizes this measure considering type of service and type of vehicle, among other company policies. It was validated with company’s real data. Improvements and computational results are discussed.

ABSTRACT. O conceito de Gerenciamento da Cadeia de Suprimentos, que já é amplamente conhecido e aplicado por empresas do setor privado, também pode ser adaptado ao setor público, no entanto a sistemática de contratação e os entraves do sistema licitatório são barreiras para seu efetivo uso. As atividades de transporte e administração de estoques representam um alto custo para qualquer empresa, inclusive para um almoxarifado público, e as ferramentas de roteirização de veículos são fundamentais na redução destes custos. Este artigo se propõe a aplicar uma heurística de resolução do Problema de Roteirização de Veículos em um almoxarifado do sistema público de saúde na cidade de Joinville, no Brasil, propondo assim uma rota alternativa de distribuição física. As rotas atuais e propostas pela heurística são comparadas a fim de encontrar melhora. O resultado foi satisfatório, e mostrou uma redução de 18,3% da distância percorrida no mês.

ABSTRACT. The disaster management has to contribute in to minimize the adverse effects of disasters. One of the main purposes is reduce the suffering of the population. This can be achieved by delivering relief items efficiently and effectively. This is a relevant task in the response phase, when the population is more vulnerable. But, it is not easy. The preparation phase plays a key role because it supports and facilitates the response. In this work, we focus in the preparation phase and we develop a methodology to the distribution network design in a strategic level, with the objective of deliver relief to the victims of a disaster efficiently. The main decisions are where to localize aid centers, which type of centers and how to deliver the relief items. We propose a mixed integer-programming model. The mathematical model is a tool to support decision-making in the preparation phase, and after a disaster could be useful to re-allocate resources. The proposed methodology is applied in Chile.

ABSTRACT. Supply Network Design (SND) is a strategic option to develop the supply network to get better performance in terms of productivity, competitiveness and sustainability. The present study presents a robustness paradigm on SND problems. This research is oriented to proposing a methodological approach for designing the supply network, considering perturbation parameters, robustness requirements, and performance features were qualitatively and quantitatively identified and the impact of the perturbation parameters on the robustness requirements and the performance features which have been determined. The methodology was validated on milk supply network, especially in storage and refrigeration process, where the primary milk distribution network was made by SND and robustness parameters jointly. The SND problem was attended the location of cooling tanks, assignment of producers for the tanks, and calculations of system capacity. The goal of this research is to identify the most robust and cost efficient configuration.

ABSTRACT. The supply chain (SC) is formed by a large number of participants, including production units, suppliers and customers. Its successful operation depends on the agreement and coordination of the involved facilities. Therefore, it is necessary to consider aspects such as the harvesting and distribution of raw materials, products and by-products, so as to ensure that products are produced and distributed correctly. In particular, the forest SC has a large number of participants, which by its heterogeneity, poses interesting integration challenges to be faced. In the Argentinean context, the forest industry is one of the most important topics of interest, as defined by MINCyT (the local Ministry of Science and Technology). It turns out to be a sector of high economic attractiveness that has great potential for growth, mainly due to the possibilities of increasing and improving productivity through the application of various types of technologies. Moreover, the continuous increase in energy requirements and the need of new sources of environmentally friendly energy has driven the development of bioenergy, in particular second generation biofuels, those produced from raw materials that do not compete with the food sector. Considering these matters and the large amount of wastes generated by the forest industry, biofuel production is an attractive option for adding value to the forest SC. Taking into account the characteristics of the involved industries in supply chain, the scale of production of various factories is a critical topic when selecting the location and sizing of the production plants. The installation of large size plants reduces the fixed and operative costs, but it increases the transportation costs due to the movements of high volume of raw materials, residues and products. Therefore, there is a tradeoff between industry scale and transportation issues. In this way, the option of smaller size plants located near to raw material sites seems to be more attractive. Another key element to be considered is the integration among the involved plants. In the forest industry there are several relations among the different production plants, as well as an important competition for the use of raw materials, residues and byproduct. In this sense, grouping plants forming clusters can be a good option to improve resource sharing and its use, and thus reducing costs. This is a particular feature of this industry, which states several alternative integration scenarios and that it is worth to be studied. Mathematical programming represents a useful tool for designing and planning the operations involved in the SC. The aim of this work is to present a mixed integer linear programming model for the optimal design and strategic planning of a forest SC. The model considers different production plants (sawmills, woodboards, pellets, and biorefineries) with the possibility of being installed in different places. Raw materials will be acquired from various sites with different species of trees and qualities. Each factory will produce diverse types of products to be distributed in several areas of consumption. Different decisions such as facilities location and size, material flows, alternative uses of the materials, etc. are effectively considered and the optimal integration of the SC is attained. Thus, the simultaneous optimization allows analyzing and assessing the different involved tradeoffs and the network performance in a comprehensive way. Installation of clusters is compared with the individual factories. A detailed analysis of has on the transportation and production cost, and residues, by-products and products production and distribution, as well as the effect of the production scale in the chain, is presented.

ABSTRACT. En este trabajo se analiza un problema donde una compañía tiene que surtir la demanda solicitada por minoristas de un producto específico. Para hacer esto, cuenta con una flotilla de vehículos, cada uno de ellos asignado a un repartidor. El problema recae en que la compañía debe hacer el diseño territorial para que el repartidor surta los pedidos de los clientes contenidos en cada distrito. El objetivo de la compañía es minimizar la dispersión de los distritos para que los despachadores no tengan que hacer viajes por toda la ciudad. Además, no debe perder de vista que los distritos tienen que estar lo mejor balanceados posibles para que los repartidores hagan un esfuerzo similar al surtir los pedidos. Una vez que los repartidores conocen a los minoristas que van a surtir, ellos son libres de diseñar el ruteo. Como se procura que los clientes estén satisfechos con el servicio, el criterio para hacer el ruteo será orientado en los minoristas. Es decir, cada despachador intentará minimizar la latencia; esto es, la posición promedio que tienen los minoristas en el ruteo. El repartidor debe tomar en cuenta algunas consideraciones al momento de hacer su ruteo: la capacidad del vehículo (aunque puede hacer múltiples viajes al centro de distribución para resurtir), la duración del día laboral, debe cubrir toda la demanda del minorista, que visite a todos los clientes una sola vez y que no haya ciclos en la ruta. Es necesario resaltar que las rutas que hagan los repartidores pueden ser muy costosas dado el criterio utilizado para construirlas. La compañía tiene un presupuesto limitado para gastar en el ruteo de los minoristas. En el caso en que el ruteo exceda el presupuesto, la compañía deberá desechar el diseño territorial realizado porque es muy costoso. Entonces, deberá elegir otra configuración de minoristas-repartidores tal que después de hacer los ruteos basados en la latencia se cumpla con esa restricción de presupuesto. Esta situación puede ser modelada como un problema de programación binivel donde el problema del nivel superior le corresponde a la compañía y el problema del nivel inferior a los repartidores. Este último puede ser visto como un problema con múltiples seguidores independientes ya que cada repartidor tiene que atender a los minoristas de su distrito únicamente. Resumiendo lo anterior, en el nivel superior se toma la decisión de diseñar los distritos, es decir, la asignación de minoristas a repartidores; mientras que en el nivel inferior cada minorista debe elegir la mejor forma de hacer el ruteo de todos los minoristas en su distrito.

ABSTRACT. In every beginning of an academic period, higher education institutions need to allocate classes in the classrooms available on their physical structures. This allocation problem is known in the literature as University Timetabling. At the University Center of Belo Horizonte (UniBH), this activity is done manually, being very expensive. A mathematical model oriented to minimize the total of rooms used per class in the institution was formulated considering its particularities. The proposed model sought to automate the classes allocation process, aiming to improve the use of the institution’s infrastructure. In order to validate the model, tests were performed with real data from previous semesters and the results gave viable solutions to the problem with acceptable computational costs. The comparison between the results of the allocation performed manually and the computational allocation was encouraging. Finally, based on the results, a future model improvement will be discussed.

ABSTRACT. A instituição Serviço Nacional de Aprendizagem Industrial (SENAI) tem a missão de promover a educação profissional para elevar a competitividade da indústria brasileira. O Programa Nacional de Acesso ao Ensino Técnico e Emprego, iniciado em 2011, foi implementado no SENAI originando alguns problemas como: maior número de alunos matriculados, alterações no perfil dos alunos, problemas relativos à qualidade do ensino profissional e evasão de alunos; relegando a segundo plano o atendimento à indústria e inovação. O objetivo do trabalho é aplicar uma Multimetodologia de Pesquisa Operacional Soft para entender/estruturar a situação problemática e gerar ações para melhoria do nível de qualidade de um processo de educação profissional assim como para a minimização da evasão escolar. A multimetodologia proposta é uma combinação dos métodos de estruturação Soft Systems Methodology (SSM), e Strategic Options Development and Analysis (SODA).

ABSTRACT. In the literature there are many methods and tools that can be helpful in strategic management of universities. Some of them are related to the aspect of sustainability, in terms of balancing the level of fulfillment of different objectives, often conflicting, which must be considered when building strategies. These tools include portfolio analysis methods. However, their use is often intuitive and detached from the quantitative aspects of management, such as, for instance, cost. This article presents a proposal of the modification of the portfolio methods application through the use of one of discrete optimization problems, namely - the multiple knapsack problem. The proposal is applied to a selected university. The case study is presented and discussed.

ABSTRACT. Most of the courses at university level includes practical sessions that are taught by teaching assistants (TA), which are also students. When courses are scheduled, practical sessions are usually scheduled as part of the courses, so the day and time when they are taught are known by the time the students register their courses. Since TAs have also to attend their own courses, they can apply for teaching only in courses that match their own schedules rather than for courses they are good at. We propose a framework in which practical sessions are scheduled after course registration has occurred, and present a multiobjective optimization model that schedules these practical sessions and simultaneously selects the TAs for each course. We discuss the advantages the using this approach and present results of its application.

ABSTRACT. Pseudoinverses are ubiquitous tools for handling over- and under-determined systems of equations. For computational efficiency, sparse pseudoinverses are desirable. Recently, sparse left and right pseudoinverses were introduced, using 1-norm minimization and linear programming. We introduce several new sparse pseudoinverses by developing linear and semi-definite programming relaxations of the well-known Moore-Penrose properties.

ABSTRACT. In a recent paper, it was proven that the subadditive dual for mixed-integer conic program is a strong dual under a strict feasibility requirement. In this paper, we generalize this result by exploring alternative sufficient conditions when strict feasibility is not present. In particular, we prove that in the case of essential strict feasibility and in the case of bounded feasible region, the subadditive dual has zero duality gap. Moreover, we show that the dual is solvable when essential strict feasibility holds while the solvability of the dual problem cannot be guaranteed for bounded feasible regions. Finally, we discuss a relationship between strong duality and cut generating functions.

ABSTRACT. Ellipsoid bounds for strictly convex quadratic integer programs have been proposed in the literature. These bounds are obtained by convex quadratic underestimators of the objective function of the problem that can be minimized over the integers by simply rounding its continuous minimizer componentwise. We initially propose in this work an alternative way of constructing an underestimator for the same problem based on the geometric intuition of the methodology given by the ellipsoidal sublevel sets of the quadratic functions. We also show how the ellipsoid bound technique of Buchheim, Hübner, and Schöbel can be extended to more general problems where the objective function is convex (not necessarily strictly convex), and box constraints are introduced. The quality of the proposed bounds is evaluated experimentally and compared to related methodologies.

ABSTRACT. Uncertainty is present in many stages of the conservation planning process due to limited biodiversity data or knowledge about potential threats. However, predicting future spatial patterns of threats remains challenging. We propose a network interdiction and fortification model that identifies critical sites to protect, while mitigating the impact of uncertain adverse events that could harm species survival. The modeling approach includes an absorbing discrete Markov chain (DMC) to describe individuals’ movements among regions. The population’s life expectancy is measured by the DMC’s weighted expected hitting time (WEHT). Adverse events (e.g., habitat destruction) affect the transition probabilities, deteriorating the WEHT. We solve the resulting mathematical program using large-scale integer programming formulations, first-order approximation methods, and supervalid inequalities to strengthen the mathematical formulations.

ABSTRACT. La programación matemática potencia el análisis y toma de decisiones basadas en información en las cadenas de abastecimiento a nivel global. En particular, las decisiones estratégicas y tácticas son semiestructuradas o no estructuradas, dificultando el trabajo de modelación al no existir procedimientos para ello: aún con aproximaciones diversas a esta representación no hay un claro referente para el usuario, quien suele estar formado en términos prácticos y de negocio pero no analíticos o técnicos de la suficiente profundidad. Así mismo, para la representación algebraica como insumo a los lenguajes de modelación disponibles, existen barreras importantes de lectura e interpretación. Este trabajo presenta un modelo descriptivo y una serie de procedimientos y métricas tendientes a brindar usabilidad a los procesos de modelación y dar la posibilidad de generación asistida de modelos matemáticos en estos ámbitos de decisión estratégica y táctica en las cadenas de abastecimiento.

ABSTRACT. El trabajo aquí desarrollado es enmarcado en el problema de planificación a mediano plazo en la industria de contenedores de vidrio, se decide el color a asignar a cada horno, el tiempo y el orden de procesamiento. El estudio es llevado a cabo en una planta de fabricación de envases de vidrio con capacidad de 54.000 toneladas de inventario paletizado. El caso se aborda desde la programación lineal buscando disminuir los costos: de inventario en bodega propia y rentada, e instalacion en hornos por cambio de colores. El principal resultado es la disminución del tiempo de toma de decision, el cual actualmente son las horas hombre contenidas en un mes de una persona tiempo completo en un cargo de nivel medio alto, quien se dedica a realizar combinaciones posibles de longitudes de tiempo de producción diferentes de 5 colores principales en 4 hornos para el horizonte de tiempo (12 meses siguientes) y evaluar su factibilidad. La instancia real se ejecuta en menos de un minuto.

ABSTRACT. En este trabajo se proponen modelos y algoritmos para abordar un modelo de optimización combinatorial vinculado a la asignación de pallets en cámaras de congelado con sistema de gestión drive-in considerando reorganización de pallets. La problemática operativa se presenta en dos etapas, en primer lugar en la colocación del pallet, en donde se debe determinar cuál es la mejor posición dentro de la cámara, considerando el plazo de estadía y, en segundo lugar, en la recuperación del pallet, se debe considerar que típicamente la recuperación de un pallet provoca movimientos de pallets adicionales, estos movimientos también deben ser determinados. Los modelos y algoritmos desarrollados son aplicados a la empresa Frunar de la zona central de Chile que cuenta con 14 cámaras de congelados con una capacidad de 3.222 Pallets de 1,90 metros. Se presentan resultados de la experimentación de los modelos implementados directamente en solvers comerciales y de algoritmos propios.

ABSTRACT. En este trabajo se evaluó el desempeño del algoritmo de enrutamiento para tráfico multipunto Mejor Contribución (BC), para su operación en una red óptica WDM (Wavelength Division Multiplexing) con capacidad total de bifurcación y con conversión de longitud de onda. Se ejecutó un conjunto de experimentos mediante simulación, calculando en todos ellos la probabilidad de bloqueo de sesiones en la red (probabilidad de no conectar a todos los nodos del grupo multipunto), bajo distintas condiciones de carga de tráfico y usando diferentes topologías de red. Los resultados obtenidos demuestran que BC presenta un desempeño superior al obtenido cuando se usa un algoritmo de ruta más corta con poda (SP), para generar el árbol de distribución multipunto, en al menos una orden de magnitud en todas las condiciones evaluadas. Esto implica que la red puede soportar más tráfico multipunto sin cambiar su equipamiento o infraestructura.

ABSTRACT. In air cargo transportation, there is a high uncertainty in the demand to be carried in due to the following factors: i) There are a reduced number of clients who transport large volumes of cargo; ii) Reservations are made in short time windows, which makes difficult to correctly estimate the supply of freighters to be provided; iii) Cargo booked to travel, in several occasions only arrives partially; iv) There are no economic penalties to clients for not showing up with the cargo or to the airline for leaving booked cargo behind. All these factors cause an imbalance between the planned supply and the demand. In this work, we formulate a MIP model that optimize the schedule adjustment by taking into account the demand fluctuations at the moment of a flight departure and solved it using a hybrid metaheuristic. The results show that the proposed solution algorithm attained on average profits that are 10% higher and computational times 86% faster than those obtained by traditional solver.

ABSTRACT. O objetivo do estudo consiste em estabelecer um ranking das cooperativas agropecuárias com base nos indicadores econômico financeiros e a relação com atributos de governança corporativa. A amostra é composta de 25 cooperativas agropecuárias. A coleta dos dados foi realizada por meio dos relatórios de gestão e das demonstrações contábeis divulgadas em 2013. Os resultados indicaram o rol de cooperativas que apresentaram melhores posicionamentos com base nos indicadores estudados. Concluiu-se que a separação de propriedade e controle está relacionada com desempenhos econômico financeiros mais satisfatórios e que não existe uma relação entre o tamanho do conselho de administração e o número de cooperados com o ranking dos indicadores de desempenho econômico financeiro. Além disso, a participação de capital de terceiros, liquidez geral e retorno sobre o patrimônio líquido foram os indicadores que se mostraram preponderantes à separação de propriedade e controle das cooperativas estudadas.

ABSTRACT. Se presenta una aplicación a la optimización energética en redes de sensores inalámbricos móviles (WSN-m) a la intemperie. Se desarrolla un modelo matemático no lineal multiobjetivo para el consumo de energía y calidad de servicio para nodos móviles de una red en un predio ganadero/lechero En condiciones de terreno se busca una alta autonomía energética de la red con una calidad de obtención de datos que asegure la disponibilidad oportuna y en suficiente volumen para analizar variables físicas y biológicas de los animales y así optimizar el manejo del plantel Se identifica los factores que determinan el comportamiento del dispositivo móvil en cuanto a energía y calidad de recepción permitiendo desarrollar un modelo que representa el consumo en función de una serie de parámetros operacionales de las WSN-m. El modelo se transforma para permitir su análisis con herramientas estándar de PLE Se presentan los resultados de implementación del diseño de la red a un predio real

ABSTRACT. El análisis de conjuntos de distribuciones es uno de los problemas más frecuentes en “data mining”. Aquí, una distribución es un conjunto finito, ordenado, de números no negativos que suman 1. Lógicamente, se han usado distancias para comparar dos distribuciones. Pero, en ciertos casos, una distancia pequeña puede ocultar alteraciones importantes de una distribución para otra. En esas situaciones, el uso de índices de poder, de la teoría de juegos, ha permitido detectar alteraciones significativas, que podrían haber sido ignoradas. Esa técnica ha sido usada para estudiar cambios en la agricultura brasileña a lo largo de más de 30 años, con datos oficiales que cubren todo el País, en dos situaciones: a) datos anuales, desde 1975, para más de 50 productos; b) datos de censos, más esporádicos, sobre uso de la tierra y valor de la producción. En muchos casos, los índices de poder han permitido identificar cambios estructurales entre las distribuciones correspondientes a dos años.

ABSTRACT. El objetivo de este trabajo es analizar la relación causal entre los costos de la inseguridad y la pérdida de eficiencia productiva de los productores de limón en el Valle de Apatzingán, Michoacán, México. El método de estudio consistió en relacionar los costos de la inseguridad con el impacto de la eficiencia productiva a través del modelo paramétrico de Análisis de Frontera Estocástica (AFE). Los resultados muestran a los costos provocados por el delito de extorsión por parte del crimen organizado, como los detonantes de un 11% de la ineficiencia técnica del modelo; mientras que a manera de hallazgo el mecanismo de retención de la producción, generó un incremento de la eficiencia técnica del 12%. Finalmente, dichas variaciones, permitieron la estimación de manera estocástica, de una eficiencia técnica del 74%. Por lo que dichos hallazgos, pueden suponer, un incremento de la eficiencia técnica del 85%, si se logra eliminar la extorsión y se mantiene el mecanismo de retención

ABSTRACT. The evaluation of environmental impacts and the efficiency of farming in Chile are key factors in the export of products to the world. Blueberries orchards have the largest land area in Chile and the country is the main producer in the southern hemisphere. This paper aims to assess the eco-efficiency of five organic blueberry farms throughout three growing seasons with the joint use of Life-Cycle Assessment (LCA) and Data Envelopment Analysis (DEA). In LCA, a LCI for blueberries orchards was used; and in DEA, following the eco-efficiency definition different variables were taken into account, with two different returns to scale models, CCR and BCC. The results indicate that there is only one efficient orchard, and four deemed inefficient in all cases. For the inefficient orchards, the DEA models set benchmarks (best-practices) and targets that indicate dramatic reductions of plastic waste (polypropylene and polyethylene) and GHD emissions.

ABSTRACT. La investigación tiene como objetivo modelar el eslabón de abastecimiento de una cadena de suministro de biocombustibles contemplando el uso de biomasas de segunda generación producto del café, lo que permitirá una alternativa de energía renovable, y además abrirá otros campos de ingresos al sector cafetero colombiano. No obstante, hay que resaltar la complejidad del problema si se tiene en cuenta que el primer eslabón se compone de un conjunto de productores dispersos en una geografía amplia y con limitaciones en el acceso. Adicionalmente, el proceso se encuentra supeditado a las etapas de cosecha, motivo por el cual los periodos de abastecimiento requieren un planeamiento que esté en capacidad de surtir las necesidades de producción de la planta. Por tanto, se presenta un modelo de Programación Entera Mixta con el cual se pudo establecer la configuración del primer eslabón de la cadena que logra el equilibrio entre los beneficios económicos y el impacto ambiental generado.

ABSTRACT. En este trabajo resolvemos el diseño de una cadena de suministro sustentable (DSSC) para la industria minera. Para ello, proponemos un modelo de optimización multi-objetivo para el DSSC que considera las dimensiones económica y ambiental de la sustentabilidad y el rendimiento de las plantas de procesamiento. El problema incluye las decisiones sobre la ubicación de las instalaciones y asignación de flujo para una cadena compuesta de varios escalones, instalaciones, productos y flujos de reproceso. Se propone un algoritmo de Optimización de Enjambre de Partículas (PSO) híbrido multi-objetivo, que incorpora los conceptos de dominancia de Pareto, PSO continuo y PSO binario para resolver el problema. Basado en varias métricas se concluye que el enfoque propuesto e competitivo con el método AUGMECON, y, además, es mejor con respecto a las métricas relacionadas con la diversidad y la distribución de las soluciones en la frontera.

ABSTRACT. Este articulo muestra de manera sistemática la situación de la ciudad de Santa Marta (Colombia) en cuanto a la recuperación y aprovechamiento de botellas plásticas (PET), logrando identificar las estrategias existentes a nivel comunitario, industrial y gubernamental, de igual forma se establecen los puntos críticos de generación de este residuo en dicha ciudad. Basado en la información levantada se propone el desarrollo de una cadena de logística verde que se encargue de recuperar las botellas, para lo cual se recurre a un modelo de programación lineal entera mixta (PLEM), por medio de este se busca encontrar las mejores alternativas para el diseño y configuración de la red, y la localización de instalaciones. Por medio de este trabajo se presenta una alternativa para potenciar implementaciones de logística verde, asociadas a la disminución del impacto ambiental por uso de plásticos en ciudades pequeñas que no cuenten con esquemas de reciclaje eficientes.

ABSTRACT. El desafío de gobiernos, industrias y comunidades es considerar la complejidad de los residuos sólidos –RS para equilibrar las principales esferas de la sostenibilidad en las cadenas de suministro. Teniendo como base la recuperación de materia prima, el potencial de la Logística de Reversa- LR para la gestión de RS al interior de las empresas y la importancia de los aspectos sociales, se propuso la configuración de una red de LR para la gestión integral y sostenible de RS en pequeñas empresas, a partir de un modelo multiobjetivo en el que se consideraron objetivos económicos, ambientales y sociales. Además, se definieron varios niveles de sostenibilidad y se asignaron pesos cada nivel. Se consolidaron 75 soluciones diferentes del modelo y se procesaron mediante la técnica ELECTRE. Los resultados mostraron que los niveles de sostenibilidad fuerte presentan una compensación entre los objetivos y la importancia de la participación de la alta gerencia en este proceso.

ABSTRACT. The increasing interest in integrating intermittent renewable energy sources into the grid, like photovoltaic (PV) generation, presents several major challenges from the viewpoint of reliability and control. Despite its benefits, energy storage has not been fully utilized. Among the limiting factors is, besides the cost, the lack of appropriate control and management strategies, since now the user has several options. In this work we present a novel approach for designing energy control policies for a PV microgrid, based on stochastic programming and data clustering to deal with the uncertainty of the generation. The output of our procedure is a simple control policy that can be implemented in small microcomputers to manage energy flow, with the objective of minimizing the total cost of consumption. Our experimental results show that the resulting policy obtains costs that are very similar, in average, to the ones obtained by solving a multi-stage stochastic program.

ABSTRACT. We present a microgrid model consisting of a small community, or city, comprised of a photovoltaic panel (PV) and two types of storages: batteries and pumping. Demand will be deterministic, for example via contracts, but energy generated from the PV is stochastic. We frame the problem as a multistage stochastic programming problem in which the goal is to produce the schedule that minimizes average costs, while satisfying demand at every time period. We illustrate our methodology with an example of a oneweek operation of such system.

ABSTRACT. En este trabajo se propone un modelo estocástico exponencial en tiempo discreto, para estimar el comportamiento dinámico de precios spot de la energía eléctrica en mercados de generación mixta (hídrica y térmica) que está gobernado por los costos marginales del sistema de generación, las condiciones climáticas (fenómenos del niño y de la niña), las intervenciones de una Comisión de Regulación, la demanda diaria, las cantidades generadas por los agentes y el precio rezagado un periodo. Se desarrolla una metodología para capturar las variables inobservables como la tendencia funcional del logaritmo de los precios históricos y las elasticidades dinámicas asociadas a las cantidades ofertadas por los agentes. Los demás parámetros son estimados usando una función de verosimilitud. Se ilustran gráficamente algunos resultados de la modelación y la estimación, usando datos reales de un mercado de generación mixta.

ABSTRACT. Large levels of wind and solar power are being deployed in power systems all over the world. Motivated by the challenge of reliably operating power systems with significant intermittent renewable energy sources, we present a multistage adaptive robust optimization model for the unit commitment problem, in which a new type of dynamic uncertainty sets are proposed for capturing temporal and spatial correlations in wind and solar power. The model also considers energy storage devices, which are crucial for flexibility, and we employ a simple affine policy that determines how dispatch decisions adapt to the realization of available renewable power. We also propose a new algorithm that combines duality based reformulations and constraint generation, with which we can efficiently solve the problem for a 2736-bus system with 90 renewable units. We also show that the proposed model can lead to significant advantages when compared to other deterministic and robust models.

ABSTRACT. SIMUS permite resolver problemas con objetivos múltiples a través de un procedimiento basado en la programación lineal. Consiste en la resolución de diversos escenarios formulados como programas lineales. Para cada escenario el decisor debe seleccionar el criterio que será considerado como objetivo, los restantes constituirán el sistema de restricciones de cada programa lineal. En cada caso, si existe una solución factible óptima se registra en una matriz definida como Pareto eficiente. El procedimiento continúa hasta que todos los criterios definidos como objetivos hayan sido evaluados. Con los datos de la matriz de resultados eficientes, se obtienen dos ordenamientos que permiten al decisor comparar los resultados obtenidos. Para el primer ordenamiento se analiza verticalmente la matriz para determinar el valor asociado a cada proyecto y considerando todos los objetivos. El segundo ordenamiento analiza la matriz en forma horizontal usando relaciones de dominación.

ABSTRACT. Uno de los grandes desafíos a nivel mundial en la actualidad es enfrentar los niveles de obesidad infantil, en el presente estudio proponemos un primer modelo de programación matemática multiobjetivo enfocado en esta población y que permita minimizar los principales factores de riesgo para un paciente con obesidad, es decir, consumo de alimentos con alta carga glicémica y con alto aporte de colesterol. Adicionalmente, el modelo busca minimizar el costo total del menú para evitar que su aplicación real se limite económicamente. El modelo propuesto, recoge recomendaciones nutricionales de organizaciones expertas para asegura una dieta saludable. Se utilizó el método de escalamiento de la ε-restricción y se implementó en el software General Algebraic Modeling System (GAMS-CPLEX). Se lograron generar menús saludables para varios días y al comparar los menús generados versus los propuestos por nutricionistas, los primeros poseen una mayor adecuación nutricional.

ABSTRACT. The article considers transportation problems in which the cost parameters are fuzzy variables. Generative probability distributions are used to model uncertainty. The minimization of expected transport costs and minimization of the variance of these costs are used as optimization criteria. An illustrative example is given.

ABSTRACT. Los sistemas de transporte público son un componente fundamental en las ciudades, por su efecto en la calidad de vida y los niveles de contaminación. En este estudio se propone y resuelve un modelo de programación matemática biobjetivo para el diseño de redes de transporte público, que considere como objetivos relevantes el tiempo total de viaje y las emisiones de CO2. El modelo considera la selección de rutas y la asignación de frecuencias y de velocidades de operación de una red de autobuses urbanos con congestión. Se utilizó el método de la e-restricción, para encontrar un conjunto de soluciones eficientes para el conjunto de instancias, pequeñas y medianas, generadas. Los resultados obtenidos muestran, que es posible lograr importantes reducciones en el nivel de emisiones generadas sin afectar el tiempo de viaje. Además, se encuentra que la reducción de emisiones implica la disminución del tamaño de la flota, lo que implica una disminución en el costo económico.

ABSTRACT. One of the most important problems in Latin-American countries is the rural development. Most of the rural population in these countries lacks of basic services such as healthcare, education and transportation. This is partially due to the high investment and operational costs related to geographical dispersion, limited access to roads and abrupt topography. This work studies the problem of opening or merging basic facilities in rural areas. The problem is formulated as a multi-objective facility location problem that incorporates constraints related to the social network, road infrastructure and connectivity with larger urban centers. The model includes stakeholders from both rural and semi-urban areas and from the local government. The proposed methodology is employed to address a real scenario from a peripheral area of Colombia, where a sensitivity analysis is performed and discussed. The results may bring new insights and directions for future public policies and research lines.

ABSTRACT. The globalization of the world economy calls for integrated supply chains that can efficiently distribute products and services to global markets, urging market players to re-think their roles in the logistics process. Inland distribution is becoming a critical dimension of the efficiency in international trade. This imposes significant challenges to the role of ports, intermodal terminals and empty container depots as functional nodes in logistics networks. In this work, we analyze the empty containers handling operations at an Empty Container Depot located in Valparaíso, Chile. The aim of the analysis is to determine the current performance of operational policies related to the stacking of empty containers at the yard in order to derive recommendations for container stacking policies and layout re-design. For this, we have developed a detailed stochastic simulation model using SIMIO software to evaluate the current situation and identify potential areas for improvement. Preliminary results are discussed, as well as recommendations for further research.

ABSTRACT. The aim of the dynamic berth allocation problem is to determine the allocation and berthing time of each ship to each berth to minimize the total service time. We propose a dynamic programming based matheuristic approach. In the proposed method, an iterated enhanced dynasearch is used to derive a good upper bound. The lower bound is derived by the Lagrangian decomposition of the original problem. The performance of the proposed method is compared with that of the hybrid tabu search method with path relinking and POPMUSIC. The results show that the performance is significantly better than those of tabu search with path-relinking and POPMUSIC for large scale instances.

ABSTRACT. Normal railway operations are often influenced by unexpected situations that affect, for example, the rolling stock and the infrastructure and may cause delays and disruptions in the network. Under these scenarios, one of the key issues is to construct a new disposition timetable to recover from the disruption and to inform the passengers as fast as possible, usually within a few minutes. Due to its complexity, the overall problem is usually divided into three phases which are solved sequentially, namely timetable rescheduling, rolling stock rescheduling and crew rescheduling. In this paper we focus on the timetable rescheduling phase and propose two new MILP formulations. A tailored Branch and Cut algorithm is considered for each formulation, for which preliminary computational results are presented evaluating the quality of the LP relaxations as well as the overall computing times requried, aiming to evaluate the potential to be applied in practice.

ABSTRACT. Este trabajo analiza la programación de carga de buques al granel, en puertos marítimos, mediante la integración de la Simulación en un marco Metaheurístico. La secuenciación consiste en programar buques sobre máquinas de embarque llamadas Shiploaders para minimizar los costos de penalización que se incurren al incumplir el Laycan, que es una cláusula del contrato de fletamento. Este problema ha sido tratado mediante la utilización de Metaheurísticas. Sin embargo, la combinación de la Simulación con Metaheurísticas puede ser útil para obtener soluciones que hagan frente a la incertidumbre de procesos en el cargue de buques al granel. Se integró al software Flexsim 7.0.6 un algoritmo genético para comparar el desempeño de operadores de cruzamiento en la secuenciación de los buques en un puerto de carga granel permitiendo medir la penalización. Se evidenció que la Simheurística minimizó el incumplimiento del Laycan haciendo frente a la incertidumbre en la operación del puerto marítimo.

ABSTRACT. The oil industry has grown considerably in recent years, particularly in the offshore environment (in Brazil, for example) and technological developments are essential to improve their processes. This work is focused on the transportation of cargo to offshore units. The units are supported by a logistics system that is critical to ensure the mobility of products efficiently and on time, reducing the total logistics cost. The objective of this work is improve the operations involving the transportation of defining the departure times of supply vessels by modeling and solving it as a mixed integer programming problem. This problem is known as Berth Allocation Problem (BAP) and consists in choosing the best schedule to the offshore supply port operations. Two formulations paradigms were developed for the problem. The models proved to be very useful for realistic applications, reducing the planning time from day to a few hours reaching seconds in some cases.

ABSTRACT. Districting and Staffing decisions are part of Home Health Care (HHC) logistics management at the strategic and tactical level, respectively. Districting decisions consist of defining districts made up of several territorial basic units to ensure that the service is delivered to the patients’ location at the prescribed time, and to assigned balance workloads to medical staff. Staffing decisions involve determining the number of staff of the required qualification to meet estimated patient demands. Both logistics problems are critical in HHC and their impact goes beyond network design and customer service considerations. In this work we propose a two-stage approach to support such logistics decisions, based on the integration of two mixed-integer linear programming models. We evaluate the approach with a set of randomly generated instances based on HHC conditions in a developing country in Latin America. Our approach allows finding better logistics decisions and improvement opportunities.

ABSTRACT. There is a high correlation between the increase in life expectancy and the coverage of health services. In Latin America, the health sector faces a context of limited resources and increases in expenditure levels that affect the service. Under these conditions, Home Health Care (HHC) services arise to improve the coverage, efficiency, and patient’s recovery process, as well as to reduce hospitalization costs. One of the logistics decisions that impact the network design in the long-term, and therefore the service quality, is the Districting Problem (DP). The DP is a complex problem due to the conditions to model the system, to the requirements to deliver the service, and to its computational complexity to solve it. In this work, we present the use of a metaheuristic to solve the DP considering balance workloads, contiguity, and compactness. We evaluate the metaheuristic with a set of instances, and our preliminary results show that better districting configurations can be obtained.

ABSTRACT. En el presente trabajo se estudia la problemática de mejorar las rutas de transporte de pacientes a centros hospitalarios. La transportación se realiza en vehículos compartidos y se tratan los llamados traslados puerta a puerta. El problema se puede modelar como un DARP que es un problema de ruteo de vehículos tipo Dial-a-Ride, donde se busca minimizar los costos de transporte y al mismo tiempo los inconvenientes de los usuarios. Al tratar con personas con movilidad reducida o que acuden a un hospital a recibir algún tratamiento, se pretende que el viaje sea lo más corto posible. Se plantea un modelo biobjetivo y se presenta un caso de estudio de una ciudad del norte de México que necesita trasladar pacientes a una ciudad vecina a recibir tratamientos especializados. Se mostrará la problemática que se presenta y los resultados al aplicar modelos matemáticos para mejorar los procedimientos de transportación de pacientes.

ABSTRACT. In this study we develop a model highlighting many of the features of a healthcare system: a limited number of approved treatments of certain qualities, insurance schemes which reimburse costs of a standard service, and nonprofit service organizations competing with for-profit firms. We find that nonprofits (for-profits) tend to provide the standard (premium) service, and that nonprofits would thereby capture a larger share of the market. Furthermore, with such a market configuration all patients are served. In contrast, in a for-profit duopoly the provider of the standard quality would also, if allowed, charge a price premium, implying that a segment of the market is excluded.