A management model for closed-loop supply chains of reusable articles

This PhD dissertation is framed in the emergent fields of Reverse Logistics and ClosedLoop Supply Chain (CLSC) management. This subarea of supply chain management has gained researchers and practitioners' attention over the last 15 years to become a fully recognized subdiscipline of the Operations Management field. More specifically, among all the activities that are included within the CLSC area, the focus of this dissertation is centered in direct reuse aspects. The main contribution of this dissertation to current knowledge is twofold. First, a framework for the so-called reuse CLSC is developed. This conceptual model is grounded in a set of six case studies conducted by the author in real industrial settings. The model has also been contrasted with existing literature and with academic and professional experts on the topic as well. The framework encompasses four building blocks. In the first block, a typology for reusable articles is put forward, distinguishing between Returnable Transport Items (RTI), Reusable Packaging Materials (RPM), and Reusable Products (RP). In the second block, the common characteristics that render reuse CLSC difficult to manage from a logistical standpoint are identified, namely: fleet shrinkage, significant investment and limited visibility. In the third block, the main problems arising in the management of reuse CLSC are analyzed, such as: (1) define fleet size dimension, (2) control cycle time and promote articles rotation, (3) control return rate and prevent shrinkage, (4) define purchase policies for new articles, (5) plan and control reconditioning activities, and (6) balance inventory between depots. Finally, in the fourth block some solutions to those issues are developed. Firstly, problems (2) and (3) are addressed through the comparative analysis of alternative strategies for controlling cycle time and return rate. Secondly, a methodology for calculating the required fleet size is elaborated (problem (1)). This methodology is valid for different configurations of the physical flows in the reuse CLSC. Likewise, some directions are pointed out for further development of a similar method for defining purchase policies for new articles (problem (4)). The second main contribution of this dissertation is embedded in the solutions part (block 4) of the conceptual framework and comprises a two-level decision problem integrating two mixed integer linear programming (MILP) models that have been formulated and solved to optimality using AIMMS as modeling language, CPLEX as solver and Excel spreadsheet for data introduction and output presentation. The results obtained are analyzed in order to measure in a client-supplier system the economic impact of two alternative control strategies (recovery policies) in the context of reuse. In addition, the models support decision-making regarding the selection of the appropriate recovery policy against the characteristics of demand pattern and the structure of the relevant costs in the system. The triangulation of methods used in this thesis has enabled to address the same research topic with different approaches and thus, the robustness of the results obtained is strengthened.

Compatibilidad del principio de óptimo de Bellman con los métodos secuenciales de optimización de redes de tuberías

En el proceso de cálculo de redes de tuberías se maneja un conjunto de variables con unas características muy peculiares, ya que son discretas y estandarizadas. Por lo tanto su evolución se produce por escalones (la presión nominal, el diámetro y el costo de los tubos). Por otro lado la presión de diseño de la red es una función directa de la presión de cabecera. En el proceso de optimización mediante programación dinámica la presión de cabecera se va reduciendo gradualmente en cada secuencia del proceso, haciendo que evolucione a la par la presión de diseño, lo que genera a su vez saltos discriminados en la presión nominal de los tramos, y con ello en su costo y en su gradiente de cambio. En esta tesis doctoral se analiza si estos cambios discriminados que se producen en el gradiente de cambio de algunos tramos en el curso de una secuencia, ocasionados por la evolución de la presión de cabecera de la red, generan interferencias que alteran el proceso secuencial de la programación dinámica. La modificación del gradiente de cambio durante el transcurso de una secuencia se conoce con el nombre de mutación, la cual puede ser activa cuando involucra a un tramo optimo modificando las condiciones de la transacción o pasiva si no crea afección alguna. En el análisis realizado se distingue entre la mutación del gradiente de cambio de los tramos óptimos (que puede generarse exclusivamente en el conjunto de los trayectos que los albergan), y entre los efectos que el cambio de timbraje produce en el resto de los tramos de la red (incluso los situados aguas abajo de los nudos con holgura de presión nula) sobre el mecanismo iterativo, estudiando la compatibilidad de este fenómeno con el principio de óptimo de Bellman. En el proceso de investigación llevado a cabo se destaca la fortaleza que da al proceso secuencial del método Granados el hecho de que el gradiente de cambio siempre sea creciente en el avance hacia el óptimo, es decir que el costo marginal de la reducción de las pérdidas de carga de la red que se consigue en una iteración siempre sea más caro que el de la iteración precedente. Asimismo, en el estudio realizado se revisan los condicionantes impuestos al proceso de optimización, incluyendo algunos que hasta ahora no se han tenido en cuenta en los estudios de investigación, pero que están totalmente integrados en la ingeniería práctica, como es la disposición telescópica de las redes (reordenación de los diámetros de mayor a menor de cabeza a cola de la red), y la disposición de un único diámetro por tramo, en lugar de que estén compartidos por dos diámetros contiguos (con sus salvedades en caso de tramos de gran longitud, o en otras situaciones muy específicas). Finalmente se incluye un capítulo con las conclusiones, aportaciones y recomendaciones, las cuales se consideran de gran utilidad para la ingeniería práctica, entre las que se destaca la perfección del método secuencial, la escasa transcendencia de las mutaciones del gradiente de cambio y la forma en que pueden obviarse, la inocuidad de las mutaciones pasivas y el cumplimiento del principio de Bellman en todo el proceso de optimización. The sizing process of a water distribution network is based on several variables, being some of them special, as they are discrete and their values are standardized: pipe pressure rating, pipe diameter and pipe cost. On another note, the sizing process is directly related with the pressure at the network head. Given that during the optimization by means of the Granados’ Method (based on dynamic programming) the pressure at the network head is being gradually reduced, a jump from one pipe pressure rating to another may arise during the sequential process, leading to changes on the pipe cost and on the gradient change (unitary cost for reducing the head losses). This chain of changes may, in turn, affect the sequential process diverting it from an optimal policies path. This thesis analyses how the abovementioned alterations could influence the results of the dynamic programming algorithm, that is to say the compatibility with the Bellman’s Principle of Optimality, which states that the sequence has to follow a route of optimal policies, and that past decisions should not influence the remaining ones. The modification of the gradient change is known as mutation. Mutations are active when they affect the optimal link (the one which was selected to be changed during iteration) or passive when they do not alter the selection of the optimal link. The thesis analysed the potential mutations processes along the network, both on the optimal paths and also on the rest of the network, and its influence on the final results. Moreover, the investigation analysed the practical restrictions of the sizing process that are fully integrated in the applied engineering, but not always taken into account by the optimization tools. As the telescopic distribution of the diameters (i.e. larger diameters are placed at the network head) and the use of a unique diameter per link (with the exception of very large links, where two consecutive diameters may be placed). Conclusions regarding robustness of the dynamic programming algorithm are given. The sequence of the Granados Method is quite robust and it has been shown capable to auto-correct the mutations that could arise during the optimization process, and to achieve an optimal distribution even when the Bellman’s Principle of Optimality is not fully accomplished. The fact that the gradient change is always increasing during the optimization (that is to say, the marginal cost of reducing head losses is always increasing), provides robustness to the algorithm, as looping are avoided in the optimization sequence. Additionally, insight into the causes of the mutation process is provided and practical rules to avoid it are given, improving the current definition and utilization of the Granados’ Method.

Influencia de las restricciones medioambientales en el sector eléctrico

El objetivo de esta tesis es la caracterización de la generación térmica representativa de la existente en la realidad, para posteriormente proceder a su modelización y simulación integrándolas en una red eléctrica tipo y llevar a cabo estudios de optimización multiobjetivo económico medioambiental. Para ello, en primera instancia se analiza el contexto energético y eléctrico actual, y más concretamente el peninsular, en el que habiendo desaparecido las centrales de fuelóleo, sólo quedan ciclos combinados y centrales de carbón de distinto rango. Seguidamente se lleva a cabo un análisis de los principales impactos medioambientales de las centrales eléctricas basadas en combustión, representados sobre todo por sus emisiones de CO2, SO2 y NOx, de las medidas de control y mitigación de las mismas y de la normativa que les aplica. A continuación, a partir de las características de los combustibles y de la información de los consumos específicos, se caracterizan los grupos térmicos frente a las funciones relevantes que definen su comportamiento energético, económico y medioambiental, en términos de funciones de salida horarias dependiendo de la carga. Se tiene en cuenta la posibilidad de desnitrificación y desulfuración. Dado que las funciones objetivo son múltiples, y que están en conflicto unas con otras, se ha optado por usar métodos multiobjetivo que son capaces de identificar el contorno de puntos óptimos o frente de Pareto, en los que tomando una solución no existe otra que lo mejore en alguna de las funciones objetivo sin empeorarlo en otra. Se analizaron varios métodos de optimización multiobjetivo y se seleccionó el de las ε constraint, capaz de encontrar frentes no convexos y cuya optimalidad estricta se puede comprobar. Se integró una representación equilibrada de centrales de antracita, hulla nacional e importada, lignito y ciclos combinados en la red tipo IEEE-57, en la que se puede trabajar con siete centrales sin distorsionar demasiado las potencias nominales reales de los grupos, y se programó en Matlab la resolución de flujos óptimos de carga en alterna con el método multiobjetivo integrado. Se identifican los frentes de Pareto de las combinaciones de coste y cada uno de los tres tipos de emisión, y también el de los cuatro objetivos juntos, obteniendo los resultados de costes óptimos del sistema para todo el rango de emisiones. Se valora cuánto le cuesta al sistema reducir una tonelada adicional de cualquier tipo de emisión a base de desplazarse a combinaciones de generación más limpias. Los puntos encontrados aseguran que bajo unas determinadas emisiones no pueden ser mejorados económicamente, o que atendiendo a ese coste no se puede reducir más allá el sistema en lo relativo a emisiones. También se indica cómo usar los frentes de Pareto para trazar estrategias óptimas de producción ante cambios horarios de carga. ABSTRACT The aim of this thesis is the characterization of electrical generation based on combustion processes representative of the actual power plants, for the latter modelling and simulation of an electrical grid and the development of economic- environmental multiobjective optimization studies. In this line, the first step taken is the analysis of the current energetic and electrical framework, focused on the peninsular one, where the fuel power plants have been shut down, and the only ones remaining are coal units of different types and combined cycle. Then it is carried out an analysis of the main environmental impacts of the thermal power plants, represented basically by the emissions of CO2, SO2 y NOx, their control and reduction measures and the applicable regulations. Next, based on the combustibles properties and the information about the units heat rates, the different power plants are characterized in relation to the outstanding functions that define their energy, economic and environmental behaviour, in terms of hourly output functions depending on their load. Optional denitrification and desulfurization is considered. Given that there are multiple objectives, and that they go in conflictive directions, it has been decided the use of multiobjective techniques, that have the ability of identifying the optimal points set, which is called the Pareto front, where taken a solution there will be no other point that can beat the former in an objective without worsening it in another objective. Several multiobjective optimization methods were analysed and pondered, selecting the ε constraint technique, which is able to find no convex fronts and it is opened to be tested to prove the strict Pareto optimality of the obtained solutions. A balanced representation of the thermal power plants, formed by anthracite, lignite, bituminous national and imported coals and combined cycle, was integrated in the IEEE-57 network case. This system was selected because it deals with a total power that will admit seven units without distorting significantly the actual size of the power plants. Next, an AC optimal power flow with the multiobjective method implemented in the routines was programmed. The Pareto fronts of the combination of operative costs with each of the three emissions functions were found, and also the front of all of them together. The optimal production costs of the system for all the emissions range were obtained. It is also evaluated the cost of reducing an additional emission ton of any of the emissions when the optimal production mix is displaced towards cleaner points. The obtained solutions assure that under a determined level of emissions they cannot be improved economically or, in the other way, at a determined cost it cannot be found points of lesser emissions. The Pareto fronts are also applied for the search of optimal strategic paths to follow the hourly load changes.

Segmentação de imagens pela transformada imagem-floresta com faixa de restrição geodésica

Vários métodos tradicionais de segmentação de imagens, como a transformada de watershed de marcado- res e métodos de conexidade fuzzy (Relative Fuzzy Connectedness- RFC, Iterative Relative Fuzzy Connected- ness - IRFC), podem ser implementados de modo eficiente utilizando o método em grafos da Transformada Imagem-Floresta (Image Foresting Transform - IFT). No entanto, a carência de termos de regularização de fronteira em sua formulação fazem com que a borda do objeto segmentado possa ser altamente irregular. Um modo de contornar isto é por meio do uso de restrições de forma do objeto, que favoreçam formas mais regulares, como na recente restrição de convexidade geodésica em estrela (Geodesic Star Convexity - GSC). Neste trabalho, apresentamos uma nova restrição de forma, chamada de Faixa de Restrição Geodésica (Geodesic Band Constraint - GBC), que pode ser incorporada eficientemente em uma sub-classe do fra- mework de corte em grafos generalizado (Generalized Graph Cut - GGC), que inclui métodos pela IFT. É apresentada uma prova da otimalidade do novo algoritmo em termos de um mínimo global de uma função de energia sujeita às novas restrições de borda. A faixa de restrição geodésica nos ajuda a regularizar a borda dos objetos, consequentemente melhorando a segmentação de objetos com formas mais regulares, mantendo o baixo custo computacional da IFT. A GBC pode também ser usada conjuntamente com um mapa de custos pré estabelecido, baseado em um modelo de forma, de modo a direcionar a segmentação a seguir uma dada forma desejada, com grau de liberdade de escala e demais deformações controladas por um parâmetro único. Essa nova restrição também pode ser combinada com a GSC e com as restrições de polaridade de borda sem custo adicional. O método é demonstrado em imagens naturais, sintéticas e médicas, sendo estas provenientes de tomografias computadorizadas e de ressonância magnética.

Real time optimization in chemical process: evaluation of strategies, improvements and industrial application.

The increasing economic competition drives the industry to implement tools that improve their processes efficiencies. The process automation is one of these tools, and the Real Time Optimization (RTO) is an automation methodology that considers economic aspects to update the process control in accordance with market prices and disturbances. Basically, RTO uses a steady-state phenomenological model to predict the process behavior, and then, optimizes an economic objective function subject to this model. Although largely implemented in industry, there is not a general agreement about the benefits of implementing RTO due to some limitations discussed in the present work: structural plant/model mismatch, identifiability issues and low frequency of set points update. Some alternative RTO approaches have been proposed in literature to handle the problem of structural plant/model mismatch. However, there is not a sensible comparison evaluating the scope and limitations of these RTO approaches under different aspects. For this reason, the classical two-step method is compared to more recently derivative-based methods (Modifier Adaptation, Integrated System Optimization and Parameter estimation, and Sufficient Conditions of Feasibility and Optimality) using a Monte Carlo methodology. The results of this comparison show that the classical RTO method is consistent, providing a model flexible enough to represent the process topology, a parameter estimation method appropriate to handle measurement noise characteristics and a method to improve the sample information quality. At each iteration, the RTO methodology updates some key parameter of the model, where it is possible to observe identifiability issues caused by lack of measurements and measurement noise, resulting in bad prediction ability. Therefore, four different parameter estimation approaches (Rotational Discrimination, Automatic Selection and Parameter estimation, Reparametrization via Differential Geometry and classical nonlinear Least Square) are evaluated with respect to their prediction accuracy, robustness and speed. The results show that the Rotational Discrimination method is the most suitable to be implemented in a RTO framework, since it requires less a priori information, it is simple to be implemented and avoid the overfitting caused by the Least Square method. The third RTO drawback discussed in the present thesis is the low frequency of set points update, this problem increases the period in which the process operates at suboptimum conditions. An alternative to handle this problem is proposed in this thesis, by integrating the classic RTO and Self-Optimizing control (SOC) using a new Model Predictive Control strategy. The new approach demonstrates that it is possible to reduce the problem of low frequency of set points updates, improving the economic performance. Finally, the practical aspects of the RTO implementation are carried out in an industrial case study, a Vapor Recompression Distillation (VRD) process located in Paulínea refinery from Petrobras. The conclusions of this study suggest that the model parameters are successfully estimated by the Rotational Discrimination method; the RTO is able to improve the process profit in about 3%, equivalent to 2 million dollars per year; and the integration of SOC and RTO may be an interesting control alternative for the VRD process.

Stationarity and regularity of infinite collections of sets. Applications to infinitely constrained optimization

This article continues the investigation of stationarity and regularity properties of infinite collections of sets in a Banach space started in Kruger and López (J. Optim. Theory Appl. 154(2), 2012), and is mainly focused on the application of the stationarity criteria to infinitely constrained optimization problems. We consider several settings of optimization problems which involve (explicitly or implicitly) infinite collections of sets and deduce for them necessary conditions characterizing stationarity in terms of dual space elements—normals and/or subdifferentials.

Mathematical programming model for heat exchanger design through optimization of partial objectives

Mathematical programming can be used for the optimal design of shell-and-tube heat exchangers (STHEs). This paper proposes a mixed integer non-linear programming (MINLP) model for the design of STHEs, following rigorously the standards of the Tubular Exchanger Manufacturers Association (TEMA). Bell–Delaware Method is used for the shell-side calculations. This approach produces a large and non-convex model that cannot be solved to global optimality with the current state of the art solvers. Notwithstanding, it is proposed to perform a sequential optimization approach of partial objective targets through the division of the problem into sets of related equations that are easier to solve. For each one of these problems a heuristic objective function is selected based on the physical behavior of the problem. The global optimal solution of the original problem cannot be ensured even in the case in which each of the sub-problems is solved to global optimality, but at least a very good solution is always guaranteed. Three cases extracted from the literature were studied. The results showed that in all cases the values obtained using the proposed MINLP model containing multiple objective functions improved the values presented in the literature.

Rigorous design of distillation columns using surrogate models based on Kriging interpolation

The economic design of a distillation column or distillation sequences is a challenging problem that has been addressed by superstructure approaches. However, these methods have not been widely used because they lead to mixed-integer nonlinear programs that are hard to solve, and require complex initialization procedures. In this article, we propose to address this challenging problem by substituting the distillation columns by Kriging-based surrogate models generated via state of the art distillation models. We study different columns with increasing difficulty, and show that it is possible to get accurate Kriging-based surrogate models. The optimization strategy ensures that convergence to a local optimum is guaranteed for numerical noise-free models. For distillation columns (slightly noisy systems), Karush–Kuhn–Tucker optimality conditions cannot be tested directly on the actual model, but still we can guarantee a local minimum in a trust region of the surrogate model that contains the actual local minimum.

Robust solutions to multi-objective linear programs with uncertain data

In this paper we examine multi-objective linear programming problems in the face of data uncertainty both in the objective function and the constraints. First, we derive a formula for the radius of robust feasibility guaranteeing constraint feasibility for all possible scenarios within a specified uncertainty set under affine data parametrization. We then present numerically tractable optimality conditions for minmax robust weakly efficient solutions, i.e., the weakly efficient solutions of the robust counterpart. We also consider highly robust weakly efficient solutions, i.e., robust feasible solutions which are weakly efficient for any possible instance of the objective matrix within a specified uncertainty set, providing lower bounds for the radius of highly robust efficiency guaranteeing the existence of this type of solutions under affine and rank-1 objective data uncertainty. Finally, we provide numerically tractable optimality conditions for highly robust weakly efficient solutions.

Testing Independence in High Dimensions & Identifiability of Graphical Models

Thesis (Ph.D.)--University of Washington, 2016-06

Optimally distributed actuator placement and control for a slewing single-link flexible manipulator

A method is proposed for determining the optimal placement and controller design for multiple distributed actuators to reduce the vibrations of flexible structures. In particular, application of piezoceramic patches to a horizontally-slewing single-link flexible manipulator modeled using the assumed modes method is investigated. The optimization method uses simulated annealing and allows placement of any number of distributed actuators of unequal length, although piezoceramics of fixed equal lengths are used in the example. It also designs an linear-quadratic-regulator controller as part of the optimization procedure. The measures of performance used in the investigation to determine optimality are the total mass of the system and the time integral of the absolute value of the hub and tip position error. This study also varies the relative weightings for each of these performance measures to observe the effects on the controller designs and piezoceramic patch positions in the optimized solutions.

Prospective evaluation of a D-optimal designed population pharmacokinetic study

Recently, methods for computing D-optimal designs for population pharmacokinetic studies have become available. However there are few publications that have prospectively evaluated the benefits of D-optimality in population or single-subject settings. This study compared a population optimal design with an empirical design for estimating the base pharmacokinetic model for enoxaparin in a stratified randomized setting. The population pharmacokinetic D-optimal design for enoxaparin was estimated using the PFIM function (MATLAB version 6.0.0.88). The optimal design was based on a one-compartment model with lognormal between subject variability and proportional residual variability and consisted of a single design with three sampling windows (0-30 min, 1.5-5 hr and 11 - 12 hr post-dose) for all patients. The empirical design consisted of three sample time windows per patient from a total of nine windows that collectively represented the entire dose interval. Each patient was assigned to have one blood sample taken from three different windows. Windows for blood sampling times were also provided for the optimal design. Ninety six patients were recruited into the study who were currently receiving enoxaparin therapy. Patients were randomly assigned to either the optimal or empirical sampling design, stratified for body mass index. The exact times of blood samples and doses were recorded. Analysis was undertaken using NONMEM (version 5). The empirical design supported a one compartment linear model with additive residual error, while the optimal design supported a two compartment linear model with additive residual error as did the model derived from the full data set. A posterior predictive check was performed where the models arising from the empirical and optimal designs were used to predict into the full data set. This revealed the optimal'' design derived model was superior to the empirical design model in terms of precision and was similar to the model developed from the full dataset. This study suggests optimal design techniques may be useful, even when the optimized design was based on a model that was misspecified in terms of the structural and statistical models and when the implementation of the optimal designed study deviated from the nominal design.

Variance-balanced designs under interference for dependent observations

It is shown that variance-balanced designs can be obtained from Type I orthogonal arrays for many general models with two kinds of treatment effects, including ones for interference, with general dependence structures. These designs can be used to obtain optimal and efficient designs. Some examples and design comparisons are given. (C) 2002 Elsevier B.V. All rights reserved.

A framework for the study of genetic variation in migratory behaviour

Evolutionary change results from selection acting on genetic variation. For migration to be successful, many different aspects of an animal's physiology and behaviour need to function in a co-coordinated way. Changes in one migratory trait are therefore likely to be accompanied by changes in other migratory and life-history traits. At present, we have some knowledge of the pressures that operate at the various stages of migration, but we know very little about the extent of genetic variation in various aspects of the migratory syndrome. As a consequence, our ability to predict which species is capable of what kind of evolutionary change, and at which rate, is limited. Here, we review how our evolutionary understanding of migration may benefit from taking a quantitative-genetic approach and present a framework for studying the causes of phenotypic variation. We review past research, that has mainly studied single migratory traits in captive birds, and discuss how this work could be extended to study genetic variation in the wild and to account for genetic correlations and correlated selection. In the future, reaction-norm approaches may become very important, as they allow the study of genetic and environmental effects on phenotypic expression within a single framework, as well as of their interactions. We advocate making more use of repeated measurements on single individuals to study the causes of among-individual variation in the wild, as they are easier to obtain than data on relatives and can provide valuable information for identifying and selecting traits. This approach will be particularly informative if it involves systematic testing of individuals under different environmental conditions. We propose extending this research agenda by using optimality models to predict levels of variation and covariation among traits and constraints. This may help us to select traits in which we might expect genetic variation, and to identify the most informative environmental axes. We also recommend an expansion of the passerine model, as this model does not apply to birds, like geese, where cultural transmission of spatio-temporal information is an important determinant of migration patterns and their variation.

Designs for generalized linear models with several variables and model uncertainty

Standard factorial designs sometimes may be inadequate for experiments that aim to estimate a generalized linear model, for example, for describing a binary response in terms of several variables. A method is proposed for finding exact designs for such experiments that uses a criterion allowing for uncertainty in the link function, the linear predictor, or the model parameters, together with a design search. Designs are assessed and compared by simulation of the distribution of efficiencies relative to locally optimal designs over a space of possible models. Exact designs are investigated for two applications, and their advantages over factorial and central composite designs are demonstrated.