This dissertation addresses gaps in the interaction between End-of-Life (EoL) product acquisition systems and disassembly sequencing. The research focuses on two remanufacturing research problems; 1) modeling uncertain EoL product quality, quantity, and timing in regards to EoL product acquisition and disassembly sequencing and 2) designing EoL product acquisition schemes considering EoL product uncertainty. The main research objectives within these areas are; analyzing, predicting, and controlling EoL product uncertainty, and incorporating EoL product uncertainty into operational and strategic level decisions. This research addresses these objectives by researching a methodology to determine optimal or near-optimal partial disassembly sequences using infeasible sequences while considering EoL product quality uncertainty. Consumer incentives are integrated into the methodology to study the effect of EoL product take-back incentives, but it also allows for the study of EoL product quantity uncertainty. EoL product age distributions are key to integrating the disassembly sequence method with EoL product acquisition management, acting both as an indicator of quality and as a basis for determining return quantity when considering incentives. At a broader level, this research makes it possible to study the impact of EoL product quality, and to an extent quantity, uncertainty resulting from strategic level (acquisition scheme) decisions, on operational (disassembly sequencing) decisions. This research is motivated by the rising importance of value recovery and sustainability to manufacturers. Extended Producer Responsibility (EPR) and Product Stewardship (PS) policies are, globally, changing the way products are treated during their use-life and EoL. Each new policy places a greater responsibility on consumers and manufacturers to address the EoL of a product. Manufacturers, in particular, may have to fulfill these obligations by such means as contracting 3rd parties for EoL recovery or performing recovery in-house. The significance of this research is linked to the growing presence of remanufacturing and recovery in the US and global economy, either via profitable ventures or environmental regulations. Remanufacturing, in particular, was surveyed by the US International Trade Commission in 2011-2012, where it was determined that remanufacturing grew by 15% to $43 billion, supported 180,000 full-time jobs from 2009-2011, and is continuing to grow. A partial disassembly sequence, multi-objective genetic algorithm (GA) is used a solution procedure to address the problem of determining the optimal or near-optimal partial disassembly sequence considering a continuous age distribution of EoL or available consumer products, with and without a consumer take-back incentive. The multi-objective GA, novel to the presented approach, relies on infeasible sequences to converge to optimal or near-optimal disassembly sequences. It is verified with a discrete economic and environmental impact case prior to incorporating EoL product age distributions. Considering the age distribution of acquired EoL products allows for decisions to be made based not only on expected profit, but also on profit variance and profit probability per EoL product, which was not observed in previous literature. As such, the research presented here within provides three contributions to disassembly and EoL product acquisition research: 1) integrating EoL product age distributions into partial disassembly sequencing objective functions, 2) accounting for partial disassembly sequence expected profit, profit variation, and profit probability as compared to disassembly sequencing methods that have, historically, only considered expected profit, and 3) studying the impact of EoL product age distributions and consumer take-back incentives on optimal or near-optimal partial disassembly sequences. Overall, this doctoral research contributes to the body of knowledge in value recovery, reverse logistics, and disassembly research fields, and is intended to be used, in the future, to develop and design efficient EoL product acquisition systems and disassembly operations. / Ph. D.

La compréhension et la prédiction des relations structure-fonction de protéines par des approches in sillico représentent aujourd'hui un challenge. Malgré le développement récent de méthodes algorithmiques pour l'étude du mouvement et des interactions moléculaires, la flexibilité de macromolécules reste largement hors de portée des outils actuels de modélisation moléculaire. L'objectif de cette thèse est de développer une nouvelle approche basée sur des algorithmes de planification de mouvement issus de la robotique pour mieux traiter la flexibilité moléculaire dans l'étude des interactions protéiques. Nous avons étendu un algorithme récent d'exploration par échantillonnage aléatoire, ML-RRT pour le désassemblage d'objets articulés complexes. Cet algorithme repose sur la décomposition des paramètres de configuration en deux sous-ensembles actifs et passifs, qui sont traités de manière découplée. Les extensions proposées permettent de considérer plusieurs degrés de mobilité pour la partie passive, qui peut être poussée ou attirée par la partie active. Cet outil algorithmique a été appliqué avec succès pour l'étude des changements conformationnels de protéines induits lors de la diffusion d'un ligand. A partir de cette extension, nous avons développé une nouvelle méthode pour la résolution simultanée du séquençage et des mouvements de désassemblage entre plusieurs objets. La méthode, nommée Iterative-ML-RRT, calcule non seulement les trajectoires permettant d'extraire toutes les pièces d'un objet complexe assemblé, mais également l'ordre permettant le désassemblage. L'approche est générale et a été appliquée pour l'étude du processus de dissociation de complexes macromoléculaires en introduisant une fonction d'évaluation basée sur l'énergie d'interaction. Les résultats présentés dans cette thèse montrent non seulement l'efficacité mais aussi la généralité des algorithmes proposés. / Understanding and predicting structure-function relationships in proteins with fully in silico approaches remain today a great challenge. Despite recent developments of computational methods for studying molecular motions and interactions, dealing with macromolecular flexibility largely remains out of reach of the existing molecular modeling tools. The aim of this thesis is to develop a novel approach based on motion planning algorithms originating from robotics to better deal with macromolecular flexibility in protein interaction studies. We have extended a recent sampling-based algorithm, ML-RRT, for (dis)-assembly path planning of complex articulated objects. This algorithm is based on a partition of the configuration parameters into active and passive subsets, which are then treated in a decoupled manner. The presented extensions permit to consider different levels of mobility for the passive parts that can be pushed or pulled by the motion of active parts. This algorithmic tool is successfully applied to study protein conformational changes induced by the diffusion of a ligand inside it. Building on the extension of ML-RRT, we have developed a novel method for simultaneously (dis)assembly sequencing and path planning. The new method, called Iterative-ML-RRT, computes not only the paths for extracting all the parts from a complex assembled object, but also the preferred order that the disassembly process has to follow. We have applied this general approach for studying disassembly pathways of macromolecular complexes considering a scoring function based on the interaction energy. The results described in this thesis prove not only the efficacy but also the generality of the proposed algorithms

No description available.