Maintenance strategies for large offshore wind farms

Abstract

Which equipment is needed and how shall tasks be scheduled in order to implement the economically most efficient operation and maintenance strategy for large offshore wind farms? This is the question motivating this research project. Considering production losses due to turbine downtime as well as local geographical and weather conditions, an efficient operation and maintenance (O&M) solution shall be achieved for two reference sites at the UK east coast. For this purpose, a Matlab-based tool has been developed, consisting of the following five main modules: Weather, Failures, Resources, Strategy and Cost. The "Weather" module is able to generate future sea states and wind speeds based on historical data. It uses a finite state Markov chain in discrete time to model significant wave heights. Wind speeds are then generated according to their conditional probability distribution at the corresponding wave height. In order to validate the weather module, several time series were generated and compared with existing data. For comparison, the mean values, standard errors, linear correlations and cumulative distribution functions for persistence of operational weather windows were chosen, both for synthetic and observed wind speed and wave height time series. Both reference sites in the UK North Sea were considered for validation. Failure rates are the basis for the "Failure" module. As an input, data gathered from onshore reliability investigations are used, which can be updated once more detailed data is available for offshore turbines. The outcomes of this module are turbine-failures occurring at a certain time. Within the “Resources” module, it is defined which equipment and personnel is available for O&M activities. The equipment is specified by its technical characteristics, e.g., the maximum transportable personnel and the operational wave height boundary. Another key parameter is the "Strategy". The main goal of this module is to take the decision whether to perform an operation or not. Within this thesis, one specific strategy has been used, but references are made to possible modifications in the according paragraphs. The measurement of economic performance is done in the "Cost" module. Here, production losses are quantified by combining the wind speed during a failure with the linearized power curve of the turbine and the local buyback price system. Therefore, the worth of additional or better maintenance equipment can be seen directly as an increase in availability and a decrease of production losses. Results show how sensitive availability and therefore production losses change with respect to changes in the maintenance fleet, reliability characteristics of components and distance to shore. Major improvements of availability were achieved by applying maintenance vessels with a higher operational wave height boundary. An increase of this constraint from one to 1.8 m significant wave height raised the availability by up to 30 percent, leading to a much better economic performance. The influence of the weather forecast accuracy on the number of maintenance vessel and crane deployments is also stated, showing a significant increase of deployments if the weather forecast is only accurate for short times. An improvement of component-reliability, modeled as a 50% decreased annual failure rate, could save up to 440 k€ of yearly production losses for each modeled wind turbine. Higher transit times, due to a greater distance to shore, strongly decrease the wind park availability.