Exponential growth in computer technology, both in terms of individual CPUs and parallel technologies over the past decades has triggered rapid progress in large scale simulations. However, despite these achievements it has become clear that many conventional state-of-the-art techniques are ill-equipped to tackle problems that inherently involve multiple scales in configuration space. Our difficulty is that conventional ("time driven" or "time stepped") techniques update all parts of simulation space (fields, particles) synchronously, i.e. at time intervals assumed to be the same throughout the global computation domain or at best varying on a sub-domain basis (in adaptive mesh refinement algorithms). Using a serial electrostatic model, it was recently shown that discrete event techniques can lead to more than two orders of magnitude speedup compared to the time-stepped approach. In this research, the focus is on the extension of this technique to parallel architectures, using parallel discrete event simulation. Previous research in parallel discrete event simulations of scientific phenomena has been limited This thesis outlines a technique for converting a time-stepped simulation in the scientific domain into an equivalent parallel discrete event model. As a candidate simulation, an electromagnetic hybrid plasma simulation is considered. The experiments and analysis show the trade-offs on performance by varying the following factors: the simulations model characteristics (e.g. lookahead), applications load balancing, and accuracy of simulation results. The experiments are performed on a high performance cluster, using a conservative synchronization mechanism. Initial performance results are encouraging, demonstrating very good parallel speedup for large-scale model configurations containing tens of thousands of cells. Overheads for inter-processor communication remain a challenge for smaller computations.

The intent of this paper is to present the reader with a simple comparison of two systems of vertical transportation. Vertical transportation is a a relatively new field and is the subject of much interest in today's world. As buildings get taller and real estate becomes more expensive, the need to find a quick, efficient system with a small footprint becomes important. By performing a simulation and subjecting the two systems under study to similar traffic conditions, one can determine the effectiveness of one system relative to the other. Additionally, we look at the effects of changing various system attributes to gain a better understanding of the primary drivers of average travel time.

Cette thèse s’inscrit au carrefour du monde de la modélisation et simulation de systèmes d’une part et du monde de l’ingénierie logicielle d’autre part. Elle vise à faire bénéficier un formalisme de spécification de systèmes à évènements discrets (DEVS) des apports de l’ingénierie dirigée par les modèles (IDM) avec l’une de ses incarnations les plus populaires : MDA (Model Driven Architecture). Le formalisme DEVS de par son adaptabilité et son extensibilité permet l’expression et la simulation de modèles dans des domaines très variés, mais l’existence de plusieurs plateformes dédiées à ce langage nuit fortement à l’interopérabilité de ces modèles. Ces difficultés, si elles ne sont pas nouvelles, représentent cependant un défi d’autant plus important que les modèles considérés sont complexes (i.e composés en général de nombreux sous modèles et interagissant fortement entre eux). L’objectif de la thèse est de proposer une réponse à la problématique de l’interopérabilité des modèles DEVS, vis-À-Vis d’autres formalismes voisins de DEVS et également vis-À-Vis des différents simulateurs existants. Le cœur de notre travail est constitué par MetaDEVS, méta-Modèle offrant une représentation des modèles DEVS indépendante des plateformes. MetaDEVS est également le nom donné à l’approche globale qui vise à fournir des passerelles génériques entre différents formalismes et DEVS («Model-To-Model»). Cette approche montre également comment, à partir de modèles DEVS spécifiés selon MetaDEVS, du code orienté-Objet, simulable, peut être automatiquement généré («Model-To-Text»).Les formalismes choisis pour faire l’objet d’une transformation vers DEVS sont BasicDEVS, un petit formalisme pédagogique créé pour l’occasion, ainsi que les automates à états finis (FSM). La plateforme de destination choisie pour la génération de code est la plateforme éducative PyDEVS, compatible avec la plateforme DEVSimPy, utilisée par les chercheurs du projet TIC de l’Università di Corsica. / This thesis takes place at the intersection between the world of modeling and simulation, and the world of software engineering. Its provides a contribution to a discrete-Event specification formalism (DEVS) using techniques of Model-Driven Engineering, with one of its most popular incarnations : MDA (Model Driven Architecture). The DEVS formalism, thanks to its adaptability and its extensibility, is able to express and simulate models in various domains. However, the existence of many dedicated platforms damages the interoperability of those models. Those difficulties, even if they are not new, are a challenge which is all the greater as the studied models are complex (i.e. usually composed of several submodels with a strong interaction).The main purpose of this thesis is to tackle the problem of the DEVS models interoperability, with respect to other formalisms close to DEVS, and also with respect to the different existing simulators. The core of our work is constituted by MetaDEVS, a metamodel that offers a platform-Independent representation of DEVS models. MetaDEVS is also the name given to the global approach which aims to provide generic bridges between different formalisms and DEVS (“Model-To-Model”). This approach also shows how, starting from DEVS models specified with MetaDEVS, object-Oriented code can be automatically generated (“Model-To-Text”).The formalisms chosen to be transformed into DEVS are BasicDEVS, a small pedagogical formalism create for our needs, and the finite state machines (FSM). The chosen target platform for the code generation is the educative framework PyDEVS, compliant with the DEVSimPy framework used by the researchers of the TIC project of the University of Corsica.

La simulation à événements discrets (SED) est une technique dans laquelle le simulateur joue une histoire suivant une chronologie d'événements, chaque événement se produisant en des points discrets de la ligne continue du temps. Lors de l'implémentation, un événement peut être représenté par un message et une heure d'occurrence. Le type du message n'est lié qu'au modèle et donc sans conséquences pour le simulateur. En revanche, les variables de temps ont un rôle critique dans le simulateur, pour construire la chronologie des événements, dans R+. Or ces variables sont souvent représentées pas des types de données produisant des approximations, tels que les nombres flottants. Cette approximation des valeurs du temps dans la simulation peut altérer la ligne de temps et conduire à des résultats incorrects. Par ailleurs, il est courant de collecter des données à partir de systèmes réels afin de prédire des phénomènes futurs, comme les prévisions météorologiques. Les résultats de cette collecte, à l'aide d'instruments et procédures de mesures, incluent une quantification d'incertitude, habituellement présentée sous forme d'intervalles. Or répondre à une question requiert parfois l'évaluation des résultats pour toutes les valeurs comprises dans l'intervalle d'incertitude. Cette thèse propose des types de données pour une gestion sans erreur du temps en SED, y compris pour des valeurs irrationnelles et périodiques. De plus, nous proposons une méthode pour obtenir tous les résultats possibles d'une simulation soumise à des événements dont l'heure d'occurrence comporte une quantification d'incertitude. / Discrete-Event Simulation (DES) is a technique in which the simulation engine plays a history following a chronology of events. The technique is called “discrete-event” because the processing of each event of the chronology takes place at discrete points of a continuous time-line. In computer implementations, an event could be represented by a message, and a time occurrence. The message data type is usually defined as part of the model and the simulator algorithms do not operate with them. Opposite is the case of time variables; simulator has to interact actively with them for reproducing the chronology of events over R+, which is usually represented by approximated data types as floating-point. The approximation of time values in the simulation can affect the time-line preventing the generation of correct results. In addition, it is common to collect data from real systems to predict future phenomena, for example for weather forecasting. When collecting data using metrological instruments and procedures, the measurement results include uncertainty quantifications, usually defined as intervals. However, sometimes, answering a question requires evaluating the results of all values in the uncertainty interval. This thesis proposes data types for handling representation of time properly in DES, including irrational and periodic time values. Moreover, we propose a method for obtaining every possible simulation result of DES models when feeding them events with uncertainty quantification on their time component.

A proposta do trabalho é criar um modelo de simulação híbrida capaz de representar um sistema logístico. Partindo de uma investigação bibliográfica conceitual das metodologias de simulação de eventos discretos e da metodologia de simulação contínua (System Dynamics) capaz de revelar suas características principais e seus potenciais de aplicabilidade, é possível delinear as possibilidades de contribuição de ambas as metodologias para o desenvolvimento de modelos híbridos (metodologias discreta + contínua) de simulação. O modelo híbrido é desenvolvido sob a plataforma AnyLogic, e a integração adequada de ambas as metodologias no modelo criado é capaz de expandir o espectro de compreensão do sistema, com a possibilidade de integração de aspectos físicos e dimensionais a aspectos políticos e comportamentais do sistema em estudo, apresentando-se como ferramenta poderosa para sobrevivência às exigentes demandas concorrenciais. / This study aims at creating a hybrid simulation model able to properly represent a logistics system. Departing from a conceptual literature review about the discrete-event and the continuous (System Dynamics) simulation methodologies able to reveals its main features and potential of applicability, it is possible to define the possibilities of developing hybrid (discrete + continuous) simulation models. The proper integration of both methodologies in a hybrid model allows the expansion of the comprehension spectrum of the system, with the possibility of integrating the physical and dimensional aspects to political and behavior patterns, revealing the hybrid methodology as a powerful tool to succeed in the highly demanding business world.

A proposta do trabalho é criar um modelo de simulação híbrida capaz de representar um sistema logístico. Partindo de uma investigação bibliográfica conceitual das metodologias de simulação de eventos discretos e da metodologia de simulação contínua (System Dynamics) capaz de revelar suas características principais e seus potenciais de aplicabilidade, é possível delinear as possibilidades de contribuição de ambas as metodologias para o desenvolvimento de modelos híbridos (metodologias discreta + contínua) de simulação. O modelo híbrido é desenvolvido sob a plataforma AnyLogic, e a integração adequada de ambas as metodologias no modelo criado é capaz de expandir o espectro de compreensão do sistema, com a possibilidade de integração de aspectos físicos e dimensionais a aspectos políticos e comportamentais do sistema em estudo, apresentando-se como ferramenta poderosa para sobrevivência às exigentes demandas concorrenciais. / This study aims at creating a hybrid simulation model able to properly represent a logistics system. Departing from a conceptual literature review about the discrete-event and the continuous (System Dynamics) simulation methodologies able to reveals its main features and potential of applicability, it is possible to define the possibilities of developing hybrid (discrete + continuous) simulation models. The proper integration of both methodologies in a hybrid model allows the expansion of the comprehension spectrum of the system, with the possibility of integrating the physical and dimensional aspects to political and behavior patterns, revealing the hybrid methodology as a powerful tool to succeed in the highly demanding business world.

Submitted by Cláudia Bueno (claudiamoura18@gmail.com) on 2016-08-15T20:00:03Z No. of bitstreams: 2 Dissertacao - Joao Ricardo Braga de Paiva - 2016.pdf: 14227997 bytes, checksum: 3c191bf387f29137782a51d5673851e6 (MD5) license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) / Rejected by Luciana Ferreira (lucgeral@gmail.com), reason: Não consta o título em português, só em inglês e na citação o nome dele não está como quer ser citado, de acordo com o currículo lattes: Nome em citações bibliográficas PAIVA, J. R. B. Detalhe, não sei se no formulário de metadados ele preencheu como você colocou. on 2016-08-17T11:36:19Z (GMT) / Submitted by Cláudia Bueno (claudiamoura18@gmail.com) on 2016-08-17T16:45:46Z No. of bitstreams: 2 Dissertacao - Joao Ricardo Braga de Paiva - 2016.pdf: 14723579 bytes, checksum: ed8f1d6ddc6fd84e86378bbe63427ee7 (MD5) license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) / Approved for entry into archive by Cláudia Bueno (claudiamoura18@gmail.com) on 2016-08-17T16:47:20Z (GMT) No. of bitstreams: 2 Dissertacao - Joao Ricardo Braga de Paiva - 2016.pdf: 14723579 bytes, checksum: ed8f1d6ddc6fd84e86378bbe63427ee7 (MD5) license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) / Made available in DSpace on 2016-08-17T16:47:20Z (GMT). No. of bitstreams: 2 Dissertacao - Joao Ricardo Braga de Paiva - 2016.pdf: 14723579 bytes, checksum: ed8f1d6ddc6fd84e86378bbe63427ee7 (MD5) license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) Previous issue date: 2016-04-27 / Outro / This work presents metric for complexity calculation in discrete event systems. The proposed methodology uses the dynamics of the system taking into account their active connections, mapped in relationship matrix. The complexity metric is applied to three distinct systems. The results show the relationship between the complexity and specic characteristics of each system. The proposed method combines knowledge about the system, contributing to behavior and performance analysis. / Este trabalho apresenta métrica para cálculo de complexidade em sistema a eventos discretos. A metodologia proposta utiliza a dinâmica do sistema levando em consideração suas conexões ativas, mapeadas na matriz de relacionamento. A métrica de complexidade é aplicada a três sistemas distintos. Os resultados obtidos apresentam a relação entre a complexidade e as características específicas de cada sistema. O método proposto agrega conhecimento sobre o sistema, contribuindo para análises de comportamento e desempenho.

Rede de Petri é uma poderosa ferramenta de modelagem gráfica e matemática bastante aplicada no desenvolvimento de projetos de sistemas de eventos discretos. Porém a sua aplicação na análise e interpretação de tais sistemas torna-se inviável por resultar em modelos grandes, com muitos elementos gráficos. Para solucionar tal problema, vários pesquisadores têm concentrado esforços no desenvolvimento de novas extensões e métodos de síntese de rede de Petri, para reduzir o tamanho dos modelos e assim facilitar a sua aplicação e análise de sistemas grandes e complexos. Rede de Petri Virtual é uma nova extensão de rede de Petri que possibilita a modelagem de tais sistemas de forma modular. Cada elemento do sistema é representado por um módulo e a comunicação entre eles é feita através dos nós virtuais. Esse trabalho formaliza a definição de rede de Petri virtual, desenvolve algoritmo e procedimento de junção dos módulos para gerar o modelo final, uma rede de Petri ordinária que representa o sistema completo. / Petri net is a powerful graphical and mathematical modeling tool commonly used to project Discrete Events Systems. The increasing complexity of such systems does not allow the use of Petri net tools due to the large size of the models (many graphical elements), which is difficult to understand and analyze. Due to this fact, researchers have been made efforts to the development of new synthesis methods for Petri nets, in order to reduce the models size and become easier its use and the analysis task of the systems properties. Virtual Petri net is a new kind (extension) of Petri net, combining its best in control and representation of discrete events systems to the best of modular modeling. This king of Petri net allows to build up models of complex systems from the modules that represent its elements, linked by the virtual nodes. The present work formalizes the definition of Virtual Petri Net. Also develops an algorithm to its use in the discrete events systems and presents a procedure to link and assemble the modules in the whole model. The modular modeling makes easier the understanding and graphical visualization of the system, keeping the final model the same features of the common Petri nets.

An important consideration in discrete event dynamic systems control theory is the selection of a suitable modeling formalism that can capture the complex characteristics of the system and the capability to automatically synthesize a controller based on the system model. Net condition event systems are well suited for modeling complex discrete event dynamic systems owing to their input and output structure, which effectively captures the behavior of the physical devices to be monitored and/or controlled. To date, net condition event systems control models have not been extensively applied to highly automated manufacturing systems and there are few guidelines on how to automatically generate Programmable Logic Controller programming languages from net condition event systems models. This research automatically converted net condition event systems control models into Programmable Logic Controller programming language and evaluated the applicability of the proposed methodology in highly automated manufacturing systems using HAS-200 as a test bed.

Rede de Petri é uma poderosa ferramenta de modelagem gráfica e matemática bastante aplicada no desenvolvimento de projetos de sistemas de eventos discretos. Porém a sua aplicação na análise e interpretação de tais sistemas torna-se inviável por resultar em modelos grandes, com muitos elementos gráficos. Para solucionar tal problema, vários pesquisadores têm concentrado esforços no desenvolvimento de novas extensões e métodos de síntese de rede de Petri, para reduzir o tamanho dos modelos e assim facilitar a sua aplicação e análise de sistemas grandes e complexos. Rede de Petri Virtual é uma nova extensão de rede de Petri que possibilita a modelagem de tais sistemas de forma modular. Cada elemento do sistema é representado por um módulo e a comunicação entre eles é feita através dos nós virtuais. Esse trabalho formaliza a definição de rede de Petri virtual, desenvolve algoritmo e procedimento de junção dos módulos para gerar o modelo final, uma rede de Petri ordinária que representa o sistema completo. / Petri net is a powerful graphical and mathematical modeling tool commonly used to project Discrete Events Systems. The increasing complexity of such systems does not allow the use of Petri net tools due to the large size of the models (many graphical elements), which is difficult to understand and analyze. Due to this fact, researchers have been made efforts to the development of new synthesis methods for Petri nets, in order to reduce the models size and become easier its use and the analysis task of the systems properties. Virtual Petri net is a new kind (extension) of Petri net, combining its best in control and representation of discrete events systems to the best of modular modeling. This king of Petri net allows to build up models of complex systems from the modules that represent its elements, linked by the virtual nodes. The present work formalizes the definition of Virtual Petri Net. Also develops an algorithm to its use in the discrete events systems and presents a procedure to link and assemble the modules in the whole model. The modular modeling makes easier the understanding and graphical visualization of the system, keeping the final model the same features of the common Petri nets.