Ett transformatorfel kan få stora ekonomiska konsekvenser och påverka ett elkraftsystems driftsäkerhet. Genom att övervaka en transformators tillstånd kan haverier och avbrott undvikas. Tillståndsövervakning möjliggör även att förebyggande underhåll kan planeras effektivt. Det här arbetet har undersökt tillgängliga diagnostikmetoder och övervakningstekniker för tillståndsbedömning av transformatorer.  Tillgänglig felstatistik för transformatorer har studerats ingående för att förstå vilka fel som kan uppstå i en transformator och i vilken grad de förekommer. Eftersom felen har olika konsekvenser för transformatorns fortsatta drift har dessa undersökts. Befintliga metoder för att upptäcka fel har dokumenterats. Detta har utförts genom att studera tillgänglig litteratur och via intervjuer med företag som tillverkar övervakningsutrustning för transformatorer. Övervakningsmetoderna har jämförts med avseende på kostnad, livslängd och noggrannhet. Utifrån dessa analyser har en rekommendation tagits fram över vilka övervakningsmetoder som är mest lämpade för Svenska Kraftnät. Det finns verktyg för att kvantifiera resultaten från övervakningsmetoder till lättolkad information. Ett sådant är hälsoindexet, som komprimerar data från olika mätningar, till ett enskilt värde som representerar en komponents tillstånd. En fallstudie har utförts där tre transformatorers tillstånd under perioden 2002–2015 har undersökts med ett befintligt hälsoindex. Resultatet visar att hälsoindex kan utnyttjas för att indikera transformatorernas tillstånd. Nackdelen med hälsoindexet som användes är att indexet är fortsatt högt oavsett kvaliteten på isolationsmaterialet. Därför föreslås en alternativ modell för en hälsoindexberäkning, i vilken pappersåldringen beaktas i större utsträckning. / A transformer fault could affect the reliability of a power system and have serious economic consequences. Therefore there is an increasing demand among electrical utilities to assess the actual condition of their transformers and hence detect faults before they occur. Condition monitoring has the potential to reduce operating costs and improve the reliability of operation. The aim of this master thesis has been to investigate available diagnostic measurement techniques that can assess the condition of power transformers. Transformer failure statistics has been studied in order to understand what different faults that can occur and to which extent they appear. The consequences of different faults have been investigated. Available measurement techniques were coupled with what faults they can detect. The diagnostic measurement techniques have also been compared in regard to cost, lifetime and accuracy. The comparison resulted in a recommendation over which measurement techniques that are most suitable for the Swedish transmission system operator, Svenska Kraftnät. There is a need among electrical utilities of economic and technical justifications for optimal investment and maintenance decisions. One available asset management technique that enables this is the health index. A health index is a tool that combines available measurement data from operating observations, field inspections and laboratory testing into an objective and quantitative index. The index provides information about the overall condition of an asset. A case study has been conducted where an available health index was used to estimate the condition of three transformers. The results indicate that the health index presents an efficient way to represent the condition of transformers. The disadvantage of the used index is that it implies a good condition even when the quality of the transformer insulation paper is poor. An alternative health index is therefore proposed where the paper insulation quality has a larger impact on the index.

Transformer ageing is inevitable and it is a challenge for utilities to manage a large fleet of ageing transformers. This means the need for monitoring transformer condition. One of the most widely used methods is oil sampling and testing. Databases of oil test records hence manifest as a great source of information for facilitating transformer ageing assessment and asset management. In this work, databases from three UK utilities including about 4,600 transformers and 65,000 oil test entries were processed, cleaned and analysed. The procedures used could help asset managers in how to approach databases, such as the need for addressing oil contamination, measurement procedure change and oil treatment discontinuities. An early degradation phenomenon was detected in multiple databases/utilities, which was investigated and found to be caused by the adoption of hydrotreatment oil refining technique in the late 1980s. Asset managers may need to monitor more frequently the affected units and restructure long term plans. The work subsequently focused on population analyses which indicated higher voltage transformers (275 kV and 400 kV) are tested more frequently and for more parameters compared with lower voltage units (33 kV and 132 kV). Acidity is the parameter that shows the highest correlation with transformer in-service age. In addition, the influence of the length of oil test records on population ageing trends was studied. It is found that it is possible to have a representative population ageing trend even with a short period (e.g. two years) of oil test results if the transformer age profile is representative of the whole transformer population. Leading from population analyses, seasonal influence on moisture was investigated which implies the importance of incorporating oil sampling temperature for better interpretation of moisture as well as indirectly breakdown voltage records. A condition mismatch between dielectric dissipation factor and resistivity was also discovered which could mean the need for revising the current IEC 60422 oil maintenance guide. Finally, insulation condition ranking was performed through principal component analysis (PCA) and analytic hierarchy process (AHP). These two techniques were demonstrated to be not just capable alternatives to traditional empirical formula but also allow fast, objective interpretation in PCA case, as well as flexible and comprehensive (objective and subjective incorporations) analysis in AHP case.

No description available.

A growing use of sensors on the electric grid has opened the door to new methods of asset management: Distribution System Operators are now looking into conditionbased maintenance, as opposed to the traditional corrective or time-based methods. As an emerging field, the methodology must be constructed from the ground, with the little data available. The focus is put on cross-linked polyethylene medium voltage overhead lines (XLPE MVOHL)asthe asset to manage, and the aim of this work is to study under which conditions the use of sensors to improve maintenance scheduling on those lines is economically profitable. Solving this problem starts with a necessary review of key ageing mechanisms of XLPE MV overhead lines, followed by the identification of sensors which can monitor the quantities behind these mechanisms. Statistical models for the lifetime of electrical assets and economic models for the analysis of investments are also described. From this preliminary study, a condition-based maintenance methodology was devised using the concept of Health Index to gather data from multiple types of sensors into one unique indicator. Using existing literature, this health index is used to dynamically estimate the failure rate of the line. This failure rate is the key to condition-based maintenance scheduling: maintenance operations are triggered when the failure rate reaches a threshold. Selecting one ageing mechanism- electrical stress-, and one type of sensorpartial discharge inductive sensors-, a Python simulation was built (and is shared at the end of this thesis) allowing to compare the cost of predictive maintenance to the cost of corrective maintenance over several decades, with the key parameters clearly identified and analysed. Beyond the methodology in itself, the main result of the work is that the use of sensors is economically profitable in most of the studied conditions. This project also reveals the strong influence of some parameters on this profitability: condition monitoring is particularly justified for short-lived assets, with a narrow distribution of failures. The failure rate threshold must be set carefully as it has a major impact on the analysis: setting it too high leads to an unprofitable scenario. / Den ökande användningen av sensorer i elnätet har öppnat dörren för nya metoder för förvaltning av tillgångar: Distributionsnätsoperatörer tittar nu på tillståndsbaserat underhåll, i motsats till de traditionella korrigerande eller tidsbaserade metoderna. Eftersom det rör sig om ett nytt område måste metoden byggas upp från grunden, med de få data som finns tillgängliga. Fokus ligger på luftledningar av tvärbanden polyeten med medelhög spänning (XLPE MV OHL) som den tillgång som ska förvaltas, och syftet med detta arbete är att undersöka under vilka förhållanden det är ekonomiskt lönsamt att använda sensorer för att förbättra underhållsplaneringen på dessa ledningar. För att lösa detta problem börjar man med en nödvändig genomgång av de viktigaste åldringsmekanismerna för XLPE MV luftledningar, följt av identifiering av sensorer som kan övervaka de kvantiteter som ligger bakom dessa mekanismer. Statistiska modeller för livslängden för elektriska tillgångar och ekonomiska modeller för analys av investeringar beskrivs också. Utifrån denna preliminära studie utarbetades en metod för tillståndsbaserat underhåll med hjälp av begreppet hälsoindex för att samla data från flera olika typer av sensorer till en unik indikator. Med hjälp av befintlig litteratur används detta hälsoindex för att dynamiskt uppskatta felfrekvensen för ledningen. Denna felfrekvens är nyckeln till en tillståndsbaserad underhållsplanering: underhålls åtgärder utlöses när felfrekvensen når ett tröskelvärde. Genom att välja en åldringsmekanism- elektrisk belastning- och en typ av sensor- induktiva sensorer med partiell urladdning- byggdes en Python-simulering (som delas i slutet av denna avhandling) som gör det möjligt att jämföra kostnaden för förebyggande underhåll med kostnaden för korrigerande underhåll under flera decennier, med de viktigaste parametrarna tydligt identifierade och analyserade. Utöver själva metoden är det viktigaste resultatet av arbetet att användningen av sensorer är ekonomiskt lönsam under de flesta av de studerade förhållandena. Projektet visar också att vissa parametrar har ett starkt inflytande på denna lönsamhet: tillståndskontroll är särskilt motiverat för tillgångar med kort livslängd och en snäv fördelning av fel. Tröskelvärdet för felfrekvensen måste sättas med omsorg eftersom det har stor inverkan på analysen: om det sätts för högt leder det till ett olönsamt scenario.

L’industrie microélectronique est un domaine compétitif, confronté de manière permanente à plusieurs défis. Pour évaluer les étapes de fabrication, des tests de qualité sont appliqués. Ces tests étant discontinus, une défaillance des équipements peut causer une dégradation de la qualité du produit. Des alarmes peuvent être déclenchées pour indiquer des problèmes. D’autre part, on dispose d’une grande quantité de données des équipements obtenues à partir de capteurs. Une gestion des alarmes, une interpolation de mesures de qualité et une réduction de données équipements sont nécessaires. Il s’agit dans notre travail à développer des méthodes génériques d’analyse multi-variée permettant d’agréger toutes les informations disponibles sur les équipements pour prédire la qualité de produit en prenant en compte la qualité des différentes étapes de fabrication. En se basant sur le principe de reconnaissance de formes, nous avons proposé une approche pour prédire le nombre de produits restant à produire avant les pertes de performance liée aux spécifications clients en fonction des indices de santé des équipement. Notre approche permet aussi d'isoler les équipements responsables de dégradation. En plus, une méthodologie à base de régression régularisée est développée pour prédire la qualité du produit tout en prenant en compte les relations de corrélations et de dépendance existantes dans le processus. Un modèle pour la gestion des alarmes est construit où des indices de criticité et de similarité sont proposés. Les données alarmes sont ensuite utilisées pour prédire le rejet de produits. Une application sur des données industrielles provenant de STMicroelectronics est fournie. / The microelectronics industry is a highly competitive field, constantly confronted with several challenges. To evaluate the manufacturing steps, quality tests are applied during and at the end of production. As these tests are discontinuous, a defect or failure of the equipment can cause a deterioration in the product quality and a loss in the manufacturing Yield. Alarms are setting off to indicate problems, but periodic alarms can be triggered resulting in alarm flows. On the other hand, a large quantity of data of the equipment obtained from sensors is available. Alarm management, interpolation of quality measurements and reduction of correlated equipment data are required. We aim in our work to develop generic methods of multi-variate analysis allowing to aggregate all the available information (equipment health indicators, alarms) to predict the product quality taking into account the quality of the various manufacturing steps. Based on the pattern recognition principle, data of the degradation trajectory are compared with health indices for failing equipment. The objective is to predict the remaining number of products before loss of the performance related to customer specifications, and the isolation of equipment responsible for degradation. In addition, regression- ased methods are used to predict the product quality while taking into account the existing correlation and the dependency relationships in the process. A model for the alarm management is constructed where criticality and similarity indices are proposed. Then, alarm data are used to predict the product scrap. An application to industrial data from STMicroelectronics is provided.

Submitted by Biblioteca de Pós-Graduação em Geoquímica BGQ (bgq@ndc.uff.br) on 2018-03-09T16:23:29Z No. of bitstreams: 1 TESE DOUTORADO JEFERSON AZE: 5526411 bytes, checksum: f96f27e2cf68f512149ab6842518d6ff (MD5) / Made available in DSpace on 2018-03-09T16:23:31Z (GMT). No. of bitstreams: 1 TESE DOUTORADO JEFERSON AZE: 5526411 bytes, checksum: f96f27e2cf68f512149ab6842518d6ff (MD5) / Universidade Federal Fluminense. Instituto de Química. Programa de Pós-Graduação em Geoquímica, Niterói, RJ / Os objetivos gerais desta tese foram: (a) Selecionar indicadores socioeconômicos e ambientais e integrá-los através da estrutura metodológica FPCIR (Força-Motriz/ Pressão/ Condicionamento/ Impacto/ Resposta). (b) Desenvolver uma metodologia para reconstrução espacial e temporal dos setores censitários em uma nova unidade de planejamento urbano, denominada de Aglomerados dos Setores Censitários (ASC´s). Nesse sentido, foram reconstruídos os Setores Censitários das pesquisas populacionais de 1980 a 2000 (AZEVEDO et al., 2004). Ainda, com relação a essa unidade, cabe destacar que é capaz de resolver os problemas de comparação dos dados georeferenciados das pesquisas populacionais do IBGE e de outras Instituições. Na etapa seguinte, escolheu-se a Bacia Hidrográfica do Rio Imboassú, localizada no Município de São Gonçalo no Estado do Rio de Janeiro para testar essa metodologia. No processo de identificação dos setores censitários que compunham a bacia, deparou-se com o problema de que algumas dessas unidades espaciais tinham parte do seu território contido em outra bacia hidrográfica. Para equacionar essa situação, utilizou-se o conceito de Bacia Ambiental apresentado por Rutkowski (2004). Dentre os vários resultados obtidos na tese, cabe mencionar: O modelo FPCIR permitiu uma análise detalhada da interação entre as atividades humanas e o ambiente; A elaboração do histórico evolutivo do processo de ocupação para 1980 a 2000 e cálculo dos IQV’s-UFF para o mesmo período; Os valores médios para os ICH’s da área de estudo em 2000 eram muito próximos ao do município; O ISA/BA de 2000 mostrou que o valor do indicador diminuía, à medida que os ASCs estavam mais distantes da área central da Bacia; A respeito do melhor índice a ser utilizado, pouco se pode concluir. No máximo, a análise indica que tanto faz escolher IQV-UFF ou ICH; Estimou-se o custo de implantação ou realização dos serviços: (a) abastecimento por rede geral em 2005 (US$ 1.705.000) e em 2010 (US$ 1.927.000); (b) Sistema de Esgotamento Sanitário (Tratamento Secundário/terciário) em 2005 estaria na faixa de (US$ 15.251.200) e em 2010 (US$ 16.559.300); e, (c) Coleta de Lixo Domiciliar Diária em 2005 (US$ 2.596) e em 2010 (US$ 2.820); e, Os Poderes Públicos vêm desenvolvendo muito lentamente e de maneira desorganizada, poucas respostas políticas, socioeconômicas e ambientais aos problemas que ocorrem na área de estudo. / The general objectives of this theory were: (a) Select socioeconomic and environmental indicators and integrate them through the methodological structure DPSIR (Driving Forces- Pressure-State-Impact-Response). (b) Develop a methodology for spatial and temporal reconstruction of the census sectors in a new urban planning unity, named Census Sector Clusters (CSCs). In that sense, the Census Sectors were reconstructed from the population research from 1980 to 2000 (AZEVEDO et al., 2004). Still, regarding this unity, we must emphasize that it is capable to solve the problems of comparison of the geo-reference data from the population research from IBGE and from other institutions. In the following phase, the Environmental Bay of the Imboassú River was chosen to test this methodology. The Imboassú River is located in the Town of São Gonçalo in the State of Rio de Janeiro. In the process to identify the census sectors that composed the bay, we encountered a problem in that some of those spatial units had part of its contained territory in another hydrographic bay. To resolve this situation, we utilized the concept of Environmental Bay presented by Rutkowski (2004). Among the several results obtained in the theory, it is important to mention that the model DPSIR permitted a detailed analysis of the interaction between the human activities and the environment; the elaboration of the historical evolution of the occupational process for 1980 to 2000 and the calculation of the Quality of Life Index (IQVUFF) for the same period; the medium values for the ICH's from the studied area in 2000 were very near to those of the town’s; the ISA/BA of 2000, showed that the value of the indicator diminished as the ASC’s were more distant from that of the Bay’s central area. As to the best index to be utilized, little can be concluded. At most, the analysis indicates that the choice between IQV-UFF and ICH doesn’t matter. It estimated the cost of introduction or realization of the service: (a) supplying it through the general network in 2005 (US$ 1,705,000) and in 2010 (US$ 1,927,000); (b) Sanitation Treatment Systems (Secondary/Tertiary) in 2005 would be in the range of (US$ 15,251,200) and in 2010 (US$ 16,559,300); and, (c) Daily Trash Collection in 2005 (US$ 2,596) and in 2010 (US$ 2,820); and, Public Policies have been developing very slowly and disorganized, few political answers to the socioeconomic and environmental problems are given within the area of study.