Problem description Electrical power grids serve to transport and distribute electrical power with high reliability and availability at acceptable costs and risks. These grids play a crucial though preferably invisible role in supplying sufficient power in a convenient form. Today’s society has become increasingly dependent on the availability of power, and has become a more and more demanding "client", putting strong pressure on the reliability, availability and cost efficiency of supply. Once the functionality of a grid is designed and the grid is constructed, it is taken into operation and expected to stay in operation for several decades. From then on the ways to control grid quality (reliability, availability, costs and risks) are operation and maintenance. The quality of the grid may be measured in terms of quality of supply (grid performance), condition (ability to perform) or costs (to ensure quality and control risks). The grid functionality may be endangered by capacity or quality limitations. In that case the grid operator needs to come up with either operational measures (maintenance, revision, load control, process improvements) or investments (replacement, extension). For making substantiated decisions it is important to know the condition of the grid and its components. Condition information is crucial to make the expected performance quantifiable, and tomake risks and costs predictable and controllable. Without condition information risks and costs may either be accepted, at the possible expense of reliability or availability, or prevented at the expense of additional safety margins and costs. Specifically, condition assessment may contribute significantly to make maintenance effective, efficient and timely, it may allow to postpone investments in a justified way and to permit controlled overloading. Moreover, it enables to justify the asset management policy to stakeholders such as clients, shareholders and regulator. One of the key components in the grid, in terms of both reliability and investment, is the power transformer, which allows for power transmission and distribution at the required voltage level. The reliability of transformers is a prime concern to grid operators. The ultimate aim of the present study is to develop an integral transformer life time model. This model will predict the transformer reliability based on relevant degradation mechanisms. These degradation mechanisms can occur in the transformer subcomponents, i.e. tank, bushings, tapchanger, core, oil and windings. Further, the transformer life time model must be applicable to individual power transformers and to power transformer populations. Technical reliability A technical reliability model aims at supporting the asset management process by providing reliability information from a technical perspective. In most cases, technical reliability refers to the technical condition and the way it changes over time, catalysed by operational and environmental parameters. One contribution to the technical reliability, for example, is the quality of electrical insulation. The technical reliability model predicts the technical condition of a component or system in terms of the probability that a component or system can perform its designated function. The method used to predict the reliability may depend on the topology, the life cycle stage of the component or system, the available level of information, and the type and level of the required output. Several methods may be used to predict future technical reliability. Two basic options are distinguished, one based on statistical data analysis and one based on the understanding of degradation mechanisms. The concept of "quality parameters" is introduced as the link between a degradation mechanism and a degradation model to predict its corresponding technical reliability. Parameters describing the condition of the systemare defined as quality parameters (QP). Three different types of quality parameters (QP) are distinguished: Direct explicit QP: is a measurable quantity that is directly linked to the degradation process, and is a direct measure of the degree of degradation. Indirect explicit QP: is a measurable quantity as well, but it is not directly linked to a specific degradation mechanism, and may originate from various sources. As a result it requires additional processing to link it to a specific mechanism. Implicit QP: is a comparative value of how well the component works in normal operating scenarios, but the result in any specific situation cannot by verified by measurement. Three distinctive methods for predicting the QP, the rate of degradation, and the level of degradation, in an increasing order of linkage with the degradation process, are: Expert judgement model An expert judgement model is a set of knowledge rules based on expert judgement. With these rules a classification of the degradation level and rate can be derived. The classification is made by defining trigger levels based on the value of the QP. Regression model With a regression model observed correlations and trends are employed to match key properties of an asset. The available data is matched to a suitable mathematical relationship. Physical model By understanding the physics behind the degradation process, QPs can be defined that are representative for the process. A physical model relies on accurate input data, which may be hard to obtain. Having defined explicit QPs, it allows tracing the progress of degradation. The results can even be fed back in to the physical model to improve the accuracy of the modelling. The overall fault probability follows from a combination of the three types of information. Any of these fault probabilities can be used as input for the integral asset management decision model. For the power transformer an overview is given of the different diagnosis and monitoring tools, and the corresponding quality parameters that link observable or deducible quantities to the degradation mechanism. Integral transformer technical reliability model Paper-oil insulation The power transformer is a component operating in a high voltage, high current, and consequently high power environment. Each aspect imposes its specific challenge on the transformer design. Paper-oil insulation is widely applied in power transformers. Oil has an intrinsic high insulating strength and at the same time serves as cooling medium by either passive or active flow. The paper prevents electric bridging by contaminants left behind and serves as a mechanical barrier between the windings and winding layers. The paper is a critical factor in paper-oil insulation. Bad paper quality leads to premature insulation degradation. The technical reliability of the paper is affected mainly by the load and ambient temperature of the transformer and concerns the modelling of three phenomena: the reliability of winding paper insulation, the temperature dependency of paper degradation and the winding hot-spot temperature. The predicted DP-value of the degradation mechanism model is in good agreement with the investigation results of a failed machine transformer. The used paper degradation model has two benefits: it describes the degradation process from a physical perspective and it is adjustable to different end-of-life thresholds. The trade off of this approach is the need of detailed process parameters. The paper degradation model is extended to include the impact of harmonic currents on the winding hot-spot temperature. Bushings and tap-changer Transformer reliability does not only depend on the condition of the winding insulation. The most relevant other subcomponents which may contribute to the overall transformer reliability are the bushings and the tap-changer. Unfortunately, for these components easily accessible direct QPs are harder to find than for paper insulation. Conceptual ideas to predict the reliability of bushings and tap-changers are discussed according to the degradation mechanism modelling principle. The reliability of a system depends on its subcomponents and their interaction. The interaction of subcomponent reliabilities on the overall reliability can be visualised by a fault tree analysis, a degradation mechanism tree analysis or by a reliability block diagram. An integral reliabilitymodel for power transformer is formed from combining two statistics based models for bushing and tap-changer failure with the paper degradation model. The results of the integration of the transformer subcomponent models confirm that for European load scenarios the performance of the tap-changer endangers the performance of the transformermore than the quality of winding paper insulation. Transformer population reliability model Modelling populations Typical high voltage grids may contain thousands of power transformers. The load of these transformers depends upon the number and character of the costumers connected. In heavy industry with a high power demand transformers are continuously loaded nominally, whereas in rural areas only a small fraction of the rated power needs to be delivered. A specification of all individual transformers reliabilities only does not provide a clear picture of the overall network condition. There is a need for defining properties that characterise the condition of a population of similar assets by means of a limited number of figures. Useful figures are the expected number of transformers still operational or the number of transformers expected to fail each year. Depending on past policy of installation, on present and future scenario of transformer loading, and on maintenance strategy, an increased replacement effort may be required within several decades. The projected failure rate from a population model predicts whether, and onwhat time scale, transformer failure is to be expected. For population analysis an alternative is provided to predict the population reliability froman individual perspective. Failure wave scenarios and replacement alternatives The transformer population reliability model is applied to two Dutch population data sets. For these populations the effects of failure wave scenarios and replacement alternatives are determined and compared. The simulations performed illustrate the importance of correct load scenarios and the history of the individual components. The future scenario analysis and replacement alternatives show that for moderate growth scenarios the expected replacement wave will probably not start earlier than the year 2050. This implies that the majority of transformers will be replaced due to transportation capacity issues. Further a preventive replacement of all transformers is only effective in increasing the reliability of the population as a whole if a differentiation can be made between highly loaded and moderately loaded transformers. If not, the fleet needs to be replaced completely before the failure wave will start.
|Qualification||Doctor of Philosophy|
|Award date||24 Jun 2010|
|Place of Publication||Eindhoven|
|Publication status||Published - 2010|