A degradação da qualidade das águas dos mananciais superficiais prejudica a sua utilização para abastecimento público. O monitoramento do fitoplâncton torna-se importante para detectar essas alterações nas águas que chegam as estações de tratamento de água. Água de má qualidade pode apresentar, entre outros aspectos, odor, sabor e excesso de organismos fitoplanctônicos que podem colmatar filtros, flotar em decantadores e produzir toxinas. As cianobactérias são organismos potenciais produtores de toxinas prejudiciais à saúde humana e animal. Nesse sentido, seu monitoramento é determinado na legislação brasileira. Este trabalho objetiva estabelecer parâmetros de garantia da qualidade analítica da contagem do fitoplâncton pelo método de Sedgwick-Rafter, através da participação em programa de comparação interlaboratorial e da validação desse método utilizando amostras não preservadas. Também pretende mostrar a importância do monitoramento do fitoplâncton em águas de abastecimento. Como forma de verificar o desempenho do Laboratório Central de Águas da Companhia Riograndense de Saneamento na análise de cianobactérias, foi avaliada a participação em Programa de Comparação Interlaboratorial, promovido pela Rede Metrológica do Rio Grande do Sul. O desempenho foi considerado satisfatório na contagem dos organismos, dentre os 30 laboratórios brasileiros participantes. No entanto, de forma geral, foram constatadas dificuldades da identificação dos organismos, sinalizando a importância de uma boa análise qualitativa. O método utilizado pelo Laboratório foi validado como exigência da norma ISO/IEC 17025, pois o método de Sedgwick-Rafter utilizado com modificação precisava ser validado. A validação foi realizada para amostras não preservadas, evidenciando não haver diferença dos resultados de análise entre amostras vivas e preservadas até três dias após a coleta, mesmo com amostra não refrigerada. O método foi considerado exato e robusto, além de terem sido calculados limite de detecção, limite de quantificação e incerteza de medição. Assegurar que o método é preciso e confiável reflete nos resultados de monitoramentos fitoplanctônicos. Assim, durante o monitoramento mensal, realizado durante um ano, em seis mananciais utilizados para abastecimento público, detectaram-se os períodos de floração de cianobactérias, bem como os gêneros presentes. Esses dados auxiliam na avaliação da qualidade da água e na tomada de decisões em relação à melhor maneira de realizar o seu tratamento. Portanto, este trabalho mostrou a necessidade da utilização de métodos confiáveis no monitoramento fitoplanctônico. A participação em comparações interlaboratoriais é fundamental para avaliar o desempenho dos laboratórios e um meio de buscar treinamento e melhoria contínua dos analistas e dos métodos utilizados. Também ficou clara a necessidade de aperfeiçoamento na identificação dos gêneros de cianobactérias, pois a identificação incorreta pode provocar erros na quantificação e fornecer resultados que não correspondem à realidade do manancial amostrado. Além disso, evidenciou-se a importância do monitoramento do fitoplâncton na avaliação da água para abastecimento público. A correta identificação dos organismos presentes determina as ações que devem ser tomadas em relação à freqüência de monitoramento ou ao processo de tratamento de água. / The water quality degradation of surface waters compromise its use for public supply. Monitoring phytoplankton is important to detect these changes in waters going to water treatment plants. Poor water quality may cause, among other things, odor, taste and excess phytoplankton organisms that can bridge filters, decanters float in decanters and produce toxins. Cyanobacteria are organisms potential producers of toxins harmful to human and animal health. In this sense, their monitoring is determined by Brazilian law. This work aims to establish parameters for quality assurance of analytical phytoplankton counting by Sedgwick-Rafter method, through participation in interlaboratory comparison program and the validation of this method using samples unpreserved. Also demonstrates the importance of phytoplankton monitoring in the water supply. As a way to check the performance of the Central Laboratory for Water and Sanitation Company Riograndense analysis of cyanobacteria, it was evaluated the participation in Interlaboratory Comparison Program, sponsored by Rede Metrológica of Rio Grande do Sul. The performance was satisfactory during counting organisms among the 30 Brazilian laboratories participating. However, in general, were found difficulties in identifying, indicating the importance of a good qualitative analysis. The method was validated by the Laboratory as a requirement of ISO / IEC 17025, because the Sedgwick-Rafter method used with modification needed to be validated. The validation was performed for samples unpreserved, showing no difference between the results of analysis and samples preserved until three days after collection, even with no refrigerated sample. The method was considered accurate and robust, and have their limit of detection, limit of quantification and uncertainty calculated. Ensure that the method is accurate and reliable monitoring reflects phytoplankton results. Thus, during the monthly monitoring was conducted over a year for six water sources used for public supply, there were periods of cyanobacteria bloom, and the genera present were detected. These data help to assess the water quality and in making decisions about the best way to accomplish water treatment. Therefore, this study showed the necessity of using reliable methods for phytoplankton monitoring. Participation in interlaboratory comparisons is essential in evaluating the performance of laboratories and a means of seeking continuous improvement and training of analysts and the methods used. Also there was a clear need for improvement in the identification cyanobacteria genera, because misidentification can cause errors in the measurement and provide results that do not correspond to the reality of the water supply. In addition, the study showed the importance of phytoplankton monitoring in the assessment of water for public supply. The correct organisms identification determines the actions to be taken in relation to frequency of monitoring or process water treatment.

La complexité des nouveaux systèmes ne cesse de grandir, en terme d’intégration de multiples technologies, de nombre de composants, ainsi qu’en terme de performances attendues. De plus, leur phase de conception doit également garantir la tenue de délais stricts pour un coût maîtrisé. Beaucoup de ces systèmes doivent notamment fournir des performances élevées en terme de Sûreté de Fonctionnement (SdF). Les analyses de SdF à mener dans ces cas doivent concilier temps et qualité d’étude, être réalisées à des niveaux de description parfois peu élevés et être capables de traiter de multiples technologies. C’est pourquoi nous avons cherché durant cette thèse à décrire une méthode d’analyse de la SdF des systèmes complexes, intégrée aux méthodes d’Ingénierie Système (IS) utilisées, que nous avons nommé MéDISIS. Les travaux de cette thèse s’attachent donc à rendre plus efficaces les analyses de SdF au cours du processus d’IS, en créant des outils et méthodes intégrés aux outils actuels de conception et proposant un support efficace pour les études SdF. Les méthodes d’Ingénierie Système Basée sur les Modèles sont les plus efficaces pour la création de tels systèmes. Nous plaçons donc ces travaux dans un référentiel utilisant un modèle central, que nous supposons écrit en SysML. Nous avons alors approfondi 3 axes de recherches : Utiliser le langage SysML comme point d’ancrage du processus d’IS intégrant les analyses de SdF, Extraire des diagrammes SysML fonctionnels les informations nécessaires aux études de risques et Exploiter les représentations SysML pour les études formelles de SdF. Dans ce contexte, nous illustrons l’emploi de MéDISIS sur la problématique du projet CAPTHOM, dont l’objectif est de bâtir une solution innovante pour la détection de présence humaine. Cette recherche nous conduit à la réalisation d’un simulateur de réseau de capteurs capable d’évaluer la réponse à des scénarios de stimulation et d’optimiser le placement des capteurs. / The design processes of new innovative systems must be able to master the increasing complexity brought by the need of integrating multiple technologies, utilized by even more components with high awaited performances. Moreover, stakeholders impose strict delays and costs on their development phase. Many of those systems are intended to possess strong reliability performances. This kind of system must include in their design phase complicated reliability analysis, within a limited time and with limited resources. Therefore, we tackle in this thesis the construction of a methodology, named MéDISIS, for integrating reliability analysis in the design processes using Model-Based System Engineering techniques. The main goal of this work is to make the reliability analysis more efficient during the engineering process, by creating methods and tools using or being included in the current design softwares utilised by system engineers. As we assume that SysML is the best language for Model-Based System Engineering, we developed 3 axes: Using SysML as the language used for the central model of all developments. Extracting the needed information on the dysfunctional behaviour, from functional SysML description studied with classic FMEA-like techniques. Setting translations between SysML descriptions and formal models for reliability studies. Finally, the use of MéDISIS is exemplified on the CAPTHOM project, aiming at developing a new human presence sensor. This study leads us to create a simulation and optimisation software for sensor networks design.

A degradação da qualidade das águas dos mananciais superficiais prejudica a sua utilização para abastecimento público. O monitoramento do fitoplâncton torna-se importante para detectar essas alterações nas águas que chegam as estações de tratamento de água. Água de má qualidade pode apresentar, entre outros aspectos, odor, sabor e excesso de organismos fitoplanctônicos que podem colmatar filtros, flotar em decantadores e produzir toxinas. As cianobactérias são organismos potenciais produtores de toxinas prejudiciais à saúde humana e animal. Nesse sentido, seu monitoramento é determinado na legislação brasileira. Este trabalho objetiva estabelecer parâmetros de garantia da qualidade analítica da contagem do fitoplâncton pelo método de Sedgwick-Rafter, através da participação em programa de comparação interlaboratorial e da validação desse método utilizando amostras não preservadas. Também pretende mostrar a importância do monitoramento do fitoplâncton em águas de abastecimento. Como forma de verificar o desempenho do Laboratório Central de Águas da Companhia Riograndense de Saneamento na análise de cianobactérias, foi avaliada a participação em Programa de Comparação Interlaboratorial, promovido pela Rede Metrológica do Rio Grande do Sul. O desempenho foi considerado satisfatório na contagem dos organismos, dentre os 30 laboratórios brasileiros participantes. No entanto, de forma geral, foram constatadas dificuldades da identificação dos organismos, sinalizando a importância de uma boa análise qualitativa. O método utilizado pelo Laboratório foi validado como exigência da norma ISO/IEC 17025, pois o método de Sedgwick-Rafter utilizado com modificação precisava ser validado. A validação foi realizada para amostras não preservadas, evidenciando não haver diferença dos resultados de análise entre amostras vivas e preservadas até três dias após a coleta, mesmo com amostra não refrigerada. O método foi considerado exato e robusto, além de terem sido calculados limite de detecção, limite de quantificação e incerteza de medição. Assegurar que o método é preciso e confiável reflete nos resultados de monitoramentos fitoplanctônicos. Assim, durante o monitoramento mensal, realizado durante um ano, em seis mananciais utilizados para abastecimento público, detectaram-se os períodos de floração de cianobactérias, bem como os gêneros presentes. Esses dados auxiliam na avaliação da qualidade da água e na tomada de decisões em relação à melhor maneira de realizar o seu tratamento. Portanto, este trabalho mostrou a necessidade da utilização de métodos confiáveis no monitoramento fitoplanctônico. A participação em comparações interlaboratoriais é fundamental para avaliar o desempenho dos laboratórios e um meio de buscar treinamento e melhoria contínua dos analistas e dos métodos utilizados. Também ficou clara a necessidade de aperfeiçoamento na identificação dos gêneros de cianobactérias, pois a identificação incorreta pode provocar erros na quantificação e fornecer resultados que não correspondem à realidade do manancial amostrado. Além disso, evidenciou-se a importância do monitoramento do fitoplâncton na avaliação da água para abastecimento público. A correta identificação dos organismos presentes determina as ações que devem ser tomadas em relação à freqüência de monitoramento ou ao processo de tratamento de água. / The water quality degradation of surface waters compromise its use for public supply. Monitoring phytoplankton is important to detect these changes in waters going to water treatment plants. Poor water quality may cause, among other things, odor, taste and excess phytoplankton organisms that can bridge filters, decanters float in decanters and produce toxins. Cyanobacteria are organisms potential producers of toxins harmful to human and animal health. In this sense, their monitoring is determined by Brazilian law. This work aims to establish parameters for quality assurance of analytical phytoplankton counting by Sedgwick-Rafter method, through participation in interlaboratory comparison program and the validation of this method using samples unpreserved. Also demonstrates the importance of phytoplankton monitoring in the water supply. As a way to check the performance of the Central Laboratory for Water and Sanitation Company Riograndense analysis of cyanobacteria, it was evaluated the participation in Interlaboratory Comparison Program, sponsored by Rede Metrológica of Rio Grande do Sul. The performance was satisfactory during counting organisms among the 30 Brazilian laboratories participating. However, in general, were found difficulties in identifying, indicating the importance of a good qualitative analysis. The method was validated by the Laboratory as a requirement of ISO / IEC 17025, because the Sedgwick-Rafter method used with modification needed to be validated. The validation was performed for samples unpreserved, showing no difference between the results of analysis and samples preserved until three days after collection, even with no refrigerated sample. The method was considered accurate and robust, and have their limit of detection, limit of quantification and uncertainty calculated. Ensure that the method is accurate and reliable monitoring reflects phytoplankton results. Thus, during the monthly monitoring was conducted over a year for six water sources used for public supply, there were periods of cyanobacteria bloom, and the genera present were detected. These data help to assess the water quality and in making decisions about the best way to accomplish water treatment. Therefore, this study showed the necessity of using reliable methods for phytoplankton monitoring. Participation in interlaboratory comparisons is essential in evaluating the performance of laboratories and a means of seeking continuous improvement and training of analysts and the methods used. Also there was a clear need for improvement in the identification cyanobacteria genera, because misidentification can cause errors in the measurement and provide results that do not correspond to the reality of the water supply. In addition, the study showed the importance of phytoplankton monitoring in the assessment of water for public supply. The correct organisms identification determines the actions to be taken in relation to frequency of monitoring or process water treatment.

As computer systems are built with aggressively scaled and unreliable technologies, some implementations rely on function specialization with reconfigurable computing to increase performance by exploiting parallelism, with possible energy gains. However, the use of reconfigurable devices in general purpose computing also brings extra reliability challenges at the system level. Solutions to cope with that are generally accompanied with the addition of excessive area, performance and power overheads to the overall system. These overheads could be reduced if a more extensive analysis was performed to evaluate the best fault tolerance strategy to balance the tradeoff between reliability and the mentioned aspects. In this context, this work present a comprehensive analysis of architectural design that includes the use of reliability modeling and takes into consideration aspects such as area, performance, and power. The analysis aims to assist the design of reliability-aware reconfigurable architectures by giving some indications about what kind of redundancy should be used in order to increase reliability. In the proposed analysis, we show that communication among functional units is critical to the overall reliability of reconfigurable architectures. Therefore, where most of the reliability investments should be made. Moreover, the analysis also demonstrate that there is a threshold in the amount of redundancy that can be added in order to increase reliability. This limit is determined by the fact that adding redundancy increases area overhead. This overhead influences reliability until overcomes the reliability gains. Therefore, even disregarding area cost, the gains in reliability will cease or even decrease. To provide a more extended evaluation, a fault tolerance approach was proposed to cope with permanent faults. The LOwER-FaT strategy is a mechanism embedded in a run-time reconfiguration mechanism that automatically selects the fault-free resources without adding extra time overhead to the configuration generation mechanism. The fault-tolerant strategy takes advantage of the on-line transparent configuration generation mechanism to transparently avoid faulty functional units and interconnects. Moreover, the strategy does not require the addition of spare resources. All the resources are used to accelerate execution, and only in case of fault, a resource is replaced by a working one, with a performance penalty caused by the reduction in the amount of resources. In spite of that, experimental results showed a mean performance degradation of 14% on overall performance under 20% fault rate. Moreover, reliability results indicated gains of around six orders of magnitude when the fault tolerance strategy was in place.

As computer systems are built with aggressively scaled and unreliable technologies, some implementations rely on function specialization with reconfigurable computing to increase performance by exploiting parallelism, with possible energy gains. However, the use of reconfigurable devices in general purpose computing also brings extra reliability challenges at the system level. Solutions to cope with that are generally accompanied with the addition of excessive area, performance and power overheads to the overall system. These overheads could be reduced if a more extensive analysis was performed to evaluate the best fault tolerance strategy to balance the tradeoff between reliability and the mentioned aspects. In this context, this work present a comprehensive analysis of architectural design that includes the use of reliability modeling and takes into consideration aspects such as area, performance, and power. The analysis aims to assist the design of reliability-aware reconfigurable architectures by giving some indications about what kind of redundancy should be used in order to increase reliability. In the proposed analysis, we show that communication among functional units is critical to the overall reliability of reconfigurable architectures. Therefore, where most of the reliability investments should be made. Moreover, the analysis also demonstrate that there is a threshold in the amount of redundancy that can be added in order to increase reliability. This limit is determined by the fact that adding redundancy increases area overhead. This overhead influences reliability until overcomes the reliability gains. Therefore, even disregarding area cost, the gains in reliability will cease or even decrease. To provide a more extended evaluation, a fault tolerance approach was proposed to cope with permanent faults. The LOwER-FaT strategy is a mechanism embedded in a run-time reconfiguration mechanism that automatically selects the fault-free resources without adding extra time overhead to the configuration generation mechanism. The fault-tolerant strategy takes advantage of the on-line transparent configuration generation mechanism to transparently avoid faulty functional units and interconnects. Moreover, the strategy does not require the addition of spare resources. All the resources are used to accelerate execution, and only in case of fault, a resource is replaced by a working one, with a performance penalty caused by the reduction in the amount of resources. In spite of that, experimental results showed a mean performance degradation of 14% on overall performance under 20% fault rate. Moreover, reliability results indicated gains of around six orders of magnitude when the fault tolerance strategy was in place.

Na presente pesquisa foi elaborada uma metodologia investigativa inerente aos efluentes de estaÃÃes de tratamento de esgotos no estado do CearÃ. Para esse estudo foram considerados padrÃes DBO (Demanda BioquÃmica de OxigÃnio), DQO (Demanda QuÃmica de OxigÃnio), SST (SÃlidos Suspensos Totais), CF (Coliformes Fecais), Temperatura, Compostos Nitrogenados (AmÃnia, FÃsforo, Nitrato) e pH, como os parÃmetros mais representativos, por serem usuais no monitoramento de ETEâs, conforme os patamares estabelecidos pelo CONAMA (Conselho Nacional do Meio Ambiente) e por legislaÃÃes estaduais SEMACE (SuperintendÃncia do Meio Ambiente do Estado do CearÃ). O principal objetivo foi elaborar uma metodologia investigativa à qualidade do tratamento de esgotos, bem como analisar a variabilidade das amostras efluentes, a confiabilidade, a estabilidade das estaÃÃes, classificar os efluentes quanto Ãs normas do Programa de PoluiÃÃo de Bacias HidrogrÃfica (PRODES) e analisar o impacto do polimento de efluentes utilizando microalgas do tipo Spirulina platensis. A metodologia foi a abordagem e a experimentaÃÃo, qualitativa e quantitativa de parÃmetros, compreendendo avaliaÃÃo da eficiÃncia, confiabilidade e estabilidade, que contribuiram para a aproximaÃÃo das sistemÃticas de projeto com alteraÃÃes das variÃveis referentes Ãs ETEâs. A avaliaÃÃo Ãs normas mostrou que as ETEs estudadas atingiram desempenhos irregulares, considerando os padrÃes de eficiÃncia e nÃveis de confianÃa rigorosos. TambÃm, foi verificado que, o mÃximo padrÃo atingido pelas EstaÃÃes cearenses se mostrou ser o nÃvel F, considerado de patamar mediano. Constatou-se ainda que, o polimento com microalgas promoveu um abatimento de 51,5%, 39,4% e 38,7% da carga de DQO, DBO e SST, respectivamente, com relaÃÃo ao efluente do esgoto tratado convencionalmente / In the present research was an elaborate investigative methodology inherent in effluent from sewage treatment plants in the State of CearÃ. For this study were considered patterns BOD (biochemical oxygen demand), COD (chemical oxygen demand), TSS (Total suspended solids), CF (faecal coliforms), temperature, Nitrogen Compounds (Ammonia, phosphorus, nitrate) and pH, as the most representative parameters, as usual in the monitoring of ETE's, as the levels established by CONAMA (National Environment Council) and by State laws SEMACE (supervision of the Environment of the State of CearÃ). The main objective is to elaborate a methodology research to quality of sewage treatment, as well as analyze the variability of the effluent samples, reliability, stability, sort the waste stations as Pollution program standards of Hydrographic Basins (PRODES) and analyze the impact of effluent polishing using microalgae like Spirulina platensis. The methodology was the approach and experimentation, qualitative and quantitative information, including evaluation of the efficiency, reliability and stability, which contributed to the rapprochement of systematic project with changes of variables concerning ETE ' s. The evaluation showed that the standards ETEs studied reached irregular performances, considering the standards of efficiency and strict confidence levels. Also, it was verified that the default maximum hit from Cearà Stations turned out to be the level F, considered median level. It was noted further that the polishing with microalgae promoted a reduction of 51.5%, 39.4% and 38.7% of the load of cod, BOD and TSS, respectively, with respect to the conventionally treated sewage effluent.

A degradação da qualidade das águas dos mananciais superficiais prejudica a sua utilização para abastecimento público. O monitoramento do fitoplâncton torna-se importante para detectar essas alterações nas águas que chegam as estações de tratamento de água. Água de má qualidade pode apresentar, entre outros aspectos, odor, sabor e excesso de organismos fitoplanctônicos que podem colmatar filtros, flotar em decantadores e produzir toxinas. As cianobactérias são organismos potenciais produtores de toxinas prejudiciais à saúde humana e animal. Nesse sentido, seu monitoramento é determinado na legislação brasileira. Este trabalho objetiva estabelecer parâmetros de garantia da qualidade analítica da contagem do fitoplâncton pelo método de Sedgwick-Rafter, através da participação em programa de comparação interlaboratorial e da validação desse método utilizando amostras não preservadas. Também pretende mostrar a importância do monitoramento do fitoplâncton em águas de abastecimento. Como forma de verificar o desempenho do Laboratório Central de Águas da Companhia Riograndense de Saneamento na análise de cianobactérias, foi avaliada a participação em Programa de Comparação Interlaboratorial, promovido pela Rede Metrológica do Rio Grande do Sul. O desempenho foi considerado satisfatório na contagem dos organismos, dentre os 30 laboratórios brasileiros participantes. No entanto, de forma geral, foram constatadas dificuldades da identificação dos organismos, sinalizando a importância de uma boa análise qualitativa. O método utilizado pelo Laboratório foi validado como exigência da norma ISO/IEC 17025, pois o método de Sedgwick-Rafter utilizado com modificação precisava ser validado. A validação foi realizada para amostras não preservadas, evidenciando não haver diferença dos resultados de análise entre amostras vivas e preservadas até três dias após a coleta, mesmo com amostra não refrigerada. O método foi considerado exato e robusto, além de terem sido calculados limite de detecção, limite de quantificação e incerteza de medição. Assegurar que o método é preciso e confiável reflete nos resultados de monitoramentos fitoplanctônicos. Assim, durante o monitoramento mensal, realizado durante um ano, em seis mananciais utilizados para abastecimento público, detectaram-se os períodos de floração de cianobactérias, bem como os gêneros presentes. Esses dados auxiliam na avaliação da qualidade da água e na tomada de decisões em relação à melhor maneira de realizar o seu tratamento. Portanto, este trabalho mostrou a necessidade da utilização de métodos confiáveis no monitoramento fitoplanctônico. A participação em comparações interlaboratoriais é fundamental para avaliar o desempenho dos laboratórios e um meio de buscar treinamento e melhoria contínua dos analistas e dos métodos utilizados. Também ficou clara a necessidade de aperfeiçoamento na identificação dos gêneros de cianobactérias, pois a identificação incorreta pode provocar erros na quantificação e fornecer resultados que não correspondem à realidade do manancial amostrado. Além disso, evidenciou-se a importância do monitoramento do fitoplâncton na avaliação da água para abastecimento público. A correta identificação dos organismos presentes determina as ações que devem ser tomadas em relação à freqüência de monitoramento ou ao processo de tratamento de água. / The water quality degradation of surface waters compromise its use for public supply. Monitoring phytoplankton is important to detect these changes in waters going to water treatment plants. Poor water quality may cause, among other things, odor, taste and excess phytoplankton organisms that can bridge filters, decanters float in decanters and produce toxins. Cyanobacteria are organisms potential producers of toxins harmful to human and animal health. In this sense, their monitoring is determined by Brazilian law. This work aims to establish parameters for quality assurance of analytical phytoplankton counting by Sedgwick-Rafter method, through participation in interlaboratory comparison program and the validation of this method using samples unpreserved. Also demonstrates the importance of phytoplankton monitoring in the water supply. As a way to check the performance of the Central Laboratory for Water and Sanitation Company Riograndense analysis of cyanobacteria, it was evaluated the participation in Interlaboratory Comparison Program, sponsored by Rede Metrológica of Rio Grande do Sul. The performance was satisfactory during counting organisms among the 30 Brazilian laboratories participating. However, in general, were found difficulties in identifying, indicating the importance of a good qualitative analysis. The method was validated by the Laboratory as a requirement of ISO / IEC 17025, because the Sedgwick-Rafter method used with modification needed to be validated. The validation was performed for samples unpreserved, showing no difference between the results of analysis and samples preserved until three days after collection, even with no refrigerated sample. The method was considered accurate and robust, and have their limit of detection, limit of quantification and uncertainty calculated. Ensure that the method is accurate and reliable monitoring reflects phytoplankton results. Thus, during the monthly monitoring was conducted over a year for six water sources used for public supply, there were periods of cyanobacteria bloom, and the genera present were detected. These data help to assess the water quality and in making decisions about the best way to accomplish water treatment. Therefore, this study showed the necessity of using reliable methods for phytoplankton monitoring. Participation in interlaboratory comparisons is essential in evaluating the performance of laboratories and a means of seeking continuous improvement and training of analysts and the methods used. Also there was a clear need for improvement in the identification cyanobacteria genera, because misidentification can cause errors in the measurement and provide results that do not correspond to the reality of the water supply. In addition, the study showed the importance of phytoplankton monitoring in the assessment of water for public supply. The correct organisms identification determines the actions to be taken in relation to frequency of monitoring or process water treatment.

As computer systems are built with aggressively scaled and unreliable technologies, some implementations rely on function specialization with reconfigurable computing to increase performance by exploiting parallelism, with possible energy gains. However, the use of reconfigurable devices in general purpose computing also brings extra reliability challenges at the system level. Solutions to cope with that are generally accompanied with the addition of excessive area, performance and power overheads to the overall system. These overheads could be reduced if a more extensive analysis was performed to evaluate the best fault tolerance strategy to balance the tradeoff between reliability and the mentioned aspects. In this context, this work present a comprehensive analysis of architectural design that includes the use of reliability modeling and takes into consideration aspects such as area, performance, and power. The analysis aims to assist the design of reliability-aware reconfigurable architectures by giving some indications about what kind of redundancy should be used in order to increase reliability. In the proposed analysis, we show that communication among functional units is critical to the overall reliability of reconfigurable architectures. Therefore, where most of the reliability investments should be made. Moreover, the analysis also demonstrate that there is a threshold in the amount of redundancy that can be added in order to increase reliability. This limit is determined by the fact that adding redundancy increases area overhead. This overhead influences reliability until overcomes the reliability gains. Therefore, even disregarding area cost, the gains in reliability will cease or even decrease. To provide a more extended evaluation, a fault tolerance approach was proposed to cope with permanent faults. The LOwER-FaT strategy is a mechanism embedded in a run-time reconfiguration mechanism that automatically selects the fault-free resources without adding extra time overhead to the configuration generation mechanism. The fault-tolerant strategy takes advantage of the on-line transparent configuration generation mechanism to transparently avoid faulty functional units and interconnects. Moreover, the strategy does not require the addition of spare resources. All the resources are used to accelerate execution, and only in case of fault, a resource is replaced by a working one, with a performance penalty caused by the reduction in the amount of resources. In spite of that, experimental results showed a mean performance degradation of 14% on overall performance under 20% fault rate. Moreover, reliability results indicated gains of around six orders of magnitude when the fault tolerance strategy was in place.

Submitted by Pedro Barros (pedro.silvabarros@ufpe.br) on 2018-06-20T22:54:11Z No. of bitstreams: 2 license_rdf: 811 bytes, checksum: e39d27027a6cc9cb039ad269a5db8e34 (MD5) TESE Marilia Abílio Ramos.pdf: 6997571 bytes, checksum: 1514e881a0919bde7d2b45038eed3a91 (MD5) / Made available in DSpace on 2018-06-20T22:54:11Z (GMT). No. of bitstreams: 2 license_rdf: 811 bytes, checksum: e39d27027a6cc9cb039ad269a5db8e34 (MD5) TESE Marilia Abílio Ramos.pdf: 6997571 bytes, checksum: 1514e881a0919bde7d2b45038eed3a91 (MD5) Previous issue date: 2017-04-07 / ANP (Agência Nacional do Petróleo) / Petrobras / The oil industry has grown in recent decades in terms of quantity of facilities and process complexity. However, human and material losses still occur due to major accidents at the facility. The analysis of these accidents reveals that many involve human failures that, if prevented, could avoid such accidents. These failures, in turn, can be identified, modeled and quantified through Human Reliability Analysis (HRA), which forms a basis for prioritization and development of safeguards for preventing or reducing the frequency of accidents. The most advanced and reliable HRA methods have been developed and applied in nuclear power plant operations, while the petroleum industry has usually applied Quantitative Risk Analysis (QRA) focusing on process safety in terms of technical aspects of the operation and equipment. This thesis demonstrates that the use of HRA in oil refining and petrochemical operations allows the identification and analysis of factors that can influence the behavior of operators as well as the potential human errors that can contribute to the occurrence of an accident. Existing HRA methodologies, however, were mainly developed for the nuclear industry. Thus, they may not reflect the specificities of refining and petrochemical plants regarding the interaction of the operators with the plant, the failure modes of the operators and the factors that influence their actions. Thus, this thesis presents an HRA methodology developed specifically for use in this industry, HERO - Human Error in Refinery Operations HRA Methodology. The Phoenix HRA methodology was used as a basis, which has three layers i) a crew response tree (CRT), which models the interaction between the crew and the plant; ii) a human response model, modeled through fault trees, that identifies the possible crew failures modes (CFMs); and (iii) "contextual factors" known as performance influencing factors (PIFs), modeled through Bayesian networks. In addition to building on such a structure, HERO's development relied on interviews with HRA specialists, visitations to a refinery and its control room, and analysis of past oil refineries accidents - four accidents were analyzed in detail. The methodology developed maintains the three-layer structure and has a guideline flowchart for the construction of the CRT, in order to model the team-plant interactions in oil refining and petrochemical operations; it also features CFMs and PIFs developed specifically for this industry, with definitions that make them easily relatable by an analyst. Finally, the methodology was applied to three potential accidental scenarios of refinery operations. In one of these scenarios, it was combined with a QRA to illustrate how an HRA can be applied to a traditional QRA and to demonstrate the influence of PIFs and of human error probability on the final risk. The use of this methodology for HRA of refineries and petrochemical plants operations can enhance this industry safety and allow for solid riskbased decisions. / A indústria de petróleo teve grande crescimento nas últimas décadas em termos de quantidade de instalações e complexidade de processo. No entanto, perdas humanas e materiais ainda ocorrem devido a acidentes graves nas instalações. A análise desses acidentes revela que muitos envolvem falhas humanas que poderiam ser prevenidas de forma a evitar tais acidentes. Estas falhas, por sua vez, podem ser identificadas, modeladas e quantificadas através da Análise de Confiabilidade Humana (ACH), que forma uma base para priorização e desenvolvimento de salvaguardas na prevenção ou redução da frequência de acidentes. Os métodos de ACH mais avançados e confiáveis têm sido desenvolvidos e aplicados nas operações de controle de plantas nucleares; já a indústria de petróleo tem usualmente aplicado a Análise Quantitativa de Risco (AQR) com foco na segurança de processo em termos técnicos da operação e equipamentos. Esta tese demonstra que o uso da ACH em operações de refino e petroquímica possibilita a identificação e análise dos fatores que podem influenciar o comportamento do operador bem como as potenciais falhas humanas que podem contribuir para a ocorrência de um acidente. As metodologias de ACH existentes, no entanto, foram desenvolvidas para a indústria nuclear. Desta forma, elas não refletem as especificidades de refino e petroquímica no que se refere à interação dos operadores com a planta, aos modos de falha dos operadores e aos fatores que influenciam suas ações. Assim, esta tese apresenta uma metodologia de ACH desenvolvida especificamente para uso nessa indústria, a HERO - Human Error in Refinery Operations HRA Methodology. Como base, utilizou-se a Metodologia Phoenix, que possui três camadas i) uma árvore de resposta da equipe (crew response tree - CRT), que modela a interação da equipe com a planta; ii) um modelo de resposta humana, modelado através de árvores de falhas, que identifica os possíveis modos de falhas da equipe (crew failures modes - CFMs); e iii) os “fatores contextuais” conhecidos como fatores de desempenho ou performance influencing factors (PIFs), modelados através de redes Bayesianas. Além de basear-se em tal estrutura, o desenvolvimento da HERO apoiou-se em entrevistas com especialistas em ACH, visitas a uma refinaria e sua sala de controle e na análise de estudos de acidentes passados em refinarias – foram analisados em detalhe quatro acidentes. A metodologia desenvolvida mantém a estrutura de três camadas e possui um fluxograma-guia para construção da CRT, de forma a modelar as interações equipe-planta na operação de refino e petroquímicas; ela também apresenta CFMs e PIFs desenvolvidos especificamente para esta indústria, com definições que os tornam facilmente identificáveis por um analista. Por fim, a metodologia foi aplicada a três cenários acidentais de operações de refinaria. Em um destes cenários, ela foi conjugada a uma AQR de forma a ilustrar como uma ACH pode ser aplicada a uma tradicional AQR e para demonstrar a influência dos PIFs e da Probabilidade de Erro Humano no risco final. Espera-se que o uso da metodologia proposta nesta tese poderá aumentar a segurança em refinarias e petroquímicas e permitir sólidas decisões baseadas no risco.

A passagem do método das tensões admissíveis para o método dos estados limites na NBR 7190 - Projeto de Estruturas de Madeira, ocorrida em 1997, foi feita por meio de calibração determinística que teve como ponto central a resistência da madeira na compressão paralela às fibras. Um dos aspectos modificados foi o dimensionamento de peças fletidas no tocante à verificação das tensões normais devidas ao momento fletor, em que é utilizada a resistência à compressão paralela às fibras. Com a intenção de averiguar o grau de segurança do modelo de cálculo para esse caso, foram realizadas análises de confiabilidade para vigas fletidas de madeira. Foram coletados dados de 549 testes de flexão de vigas, obtendo-se valores relativos à ruptura, e também foram obtidas informações estatísticas a respeito como a média e o desvio-padrão da resistência à compressão paralela às fibras, das mesmas espécies usadas nas vigas. Nas análises de confiabilidade foram utilizadas 5 variáveis aleatórias, com 5 tipos de combinações diferentes, analisadas com e sem erro de modelo para cada um dos 16 grupos de resultados levantados, totalizando 2752 análises de confiabilidade. Os resultados das análises sem o erro de modelo mostram que a norma não atinge valores suficientes para o índice de confiabilidade e que, com a introdução do erro de modelo, os resultados são mais adequados. Também se verificou que o modelo adotado pela norma é muito conservador, no caso de peças de madeira de Pinus SP classificadas. / The transition of allowable stress to limit state design methods in \"NBR 7190 - Projeto de Estruturas de Madeira\", in 1997, was made considering the strength in compression parallel to the grain as central point of deterministic calibration. One of the aspects modified was the design of beams, related to tension and compression; in this case is used the strength in compression. A reliability analysis was made for timber beams to determine the security level of the theoretical model. Were collected data related to failure from 549 bending tests in beams, and statistical information about the mean and standard deviation of the strength in compression of the wood species used in the beams. In reliability analysis were used 5 random variables, with 5 different types of combinations, analyzed with and without model error for each of the 16 groups of results collected, resulting 2752 reliability analysis. The results of the analysis without the model error show that the standard doesnt achieve sufficient values for the reliability index and, with the introduction of model error, the results are more adequate. It was also verified that the theoretical model is very conservative in the case of graded members of Pine species.