Determining power transformers' sequence of maintenance and repair in power grids

Utilities need to ensure high reliable power grids that are cost-efficient as well. Periodic replacements are costly. So utilities try to extend the life-span of their power transformers. However, the older power transformers the higher the effort for maintenance and repair. Thus it is necessary to provide just the right maintenance and repair capacity to ensure a certain reliability of supply. Hence, short maintenance and repair capacities need to be efficiently distributed over a fleet of power transformers. An efficient distribution of maintenance and repair capacities means that - according to some selected criteria - one power transformer would be maintained or repaired before another. The paper introduces two criteria in order to assess the status of power transformers: condition classes and default risk. Power transformers are initially assorted in one of the following four condition classes: "not aged, not worn out", "aged", "worn out (not noticeably)" and "worn out (noticeably)". Between the classes a sequence of maintenance and repair is introduced. Within these classes the default risk is used in order to define a sequence of maintenance and repair. A power transformer's default risk is defined as product of importance (impact of the outage on the power grid and on the power transformer) and wearout. In order to determine the wearout of a power transformer one need first to identify transformer's components. Then the components' function and wearout need to be identified (by construction plan) and assessed by inspection, respectively. The paper deals with the inspection of a very important power transformer component, namely the insulating system. The inspection is performed by dissolved gas analysis and oil characteristics. A new interpretation method for each of them is used in order to asses the condition of the insulation system. The interpretation methods use some concepts to overcome certain limits during mapping between feature and condition space. The maintenance and repair concept will finally be validated with practical data that originate from a homogeneous and a heterogeneous group of power transformers.