LONDON — Several European countries have defined targets to install and operate offshore wind energy as part of their renewable energy goals. According to these targets, more than 40 GW offshore wind power is expected to be installed across the region by the year 2020.
With an average turbine size of about 5-10 MW, between 4000 and 8000 wind turbines need to be transported and installed, operated and maintained. When other international developments are also considered, these numbers are much higher. This means that worldwide the required effort for operation and maintenance (O&M) of offshore wind farms will be enormous. And as a result the control and optimisation of O&M during the lifetime of these offshore wind turbines is essential for an economic exploitation.
The Need for O&M Cost Modeling
At present O&M costs of offshore wind farms contribute substantially (by €2-4 cents/kWh) to the life cycle costs, so it may be profitable to check whether the O&M costs can be reduced so that the cost per kWh can be reduced over the operational life of a wind turbine. Accurate cost estimations for future O&M of offshore wind farms can be used as input by both operators and original equipment manufacturers (OEMs), for example for deciding on new O&M contracts, making reservations for future O&M budgets, optimizing O&M at the end of a warranty period and periodically for the optimization of accessibility.
The key to efficient operation and maintenance of wind farms is a full understanding of the processes and cost drivers related to O&M. Most operators collect data related to their wind farm in separate documents, files and databases. Yet they do not necessarily have the tools or the working processes available to make maximum use of this data. As a result, some operators lack a full understanding of their wind farm and thus are unable to develop O&M strategies that reduce operational costs and ultimately increase revenues.
Making estimates and optimization of O&M costs and downtime is not an easy task: the amount of preventive, corrective and condition-based maintenance needs to be determined. Many parameters play a role and need to be quantified.
The most important are the number of failures which lead to a shutdown, the number of failures that can be repaired during regular visits, the repair strategy (varying from simple resets to expensive replacements with crane vessels), weather constraints, and costs for labor, spares and equipment. These parameters can be quantified using the operational data and experiences generated by the wind farm from day one of operation.
Most operators are currently not able to collect these kinds of data (failure data, repair data, logistics data, costs, SCADA data, data from load measurement campaigns, and data from condition monitoring) in a structured manner which allows semi-automated data analysis for estimating the total O&M costs.
Operation & Maintenance Cost Estimator Approach
In order to improve the situation, the Operation & Maintenance Cost Estimator (OMCE) approach was developed with two main parts: the OMCE Building Blocks (BBs) to process operational data, and the OMCE-Calculator to assess future O&M costs through maintenance strategy analysis.
The OMCE requires feedback of operational data from a specific wind farm under consideration, such as O&M data, data from measurement campaigns, and data from condition monitoring programs. Data about failures, repair actions, vessel usage, spare parts and weather conditions are analyzed to estimate the effort for unplanned corrective maintenance. Data from condition monitoring systems and load measurements are analyzed to estimate the effort for condition-based maintenance.
For this purpose four BBs have been specified, each covering a specific data set. Their main objective is to process operational data in such a way that useful information is obtained, and can be used for performance monitoring and as input for the OMCE-Calculator. For the processing of wind farm data by two of these BBs, a format is required to link the different maintenance actions to a single event.
Meanwhile, the OMCE-Calculator is a tool for the assessment of the expected O&M effort and associated costs for the coming period, where among others the information provided by the OMCE BBs is taken into account. Three types of maintenance should be considered to assess the O&M effort: unplanned corrective maintenance, condition-based maintenance and calendar-based maintenance. The structure of the OMCE approach is depicted in Figure 1.
Since operators and OEMs are the owners of the operational data, ECN and partner RWE defined a research project in the context of the Dutch Far and Large Offshore Wind (FLOW) program. The project goal is to apply the OMCE baseline model to an actual offshore wind farm, and assess the contribution in terms of cost reductions. Within this research project operational data is supplied by RWE as input for further development of the OMCE approach.
Structuring and Analysis of Wind Farm Data
To derive useful information from operational data which can be used as input for O&M cost modeling, it is important that data from sources related to O&M are collected in a structured manner. In fact, structuring raw O&M data is a key challenge.
Wind farm operators collect O&M data in different sources which are frequently stored at independent locations. Additionally, the format between data sources is often different, which makes them unsuitable for automated data processing (for example, data is reported per turbine, chronologically, per month, tri-monthly, etc). For example, data is reported on a per-turbine, chronological, per month, and tri-monthly basis. In addition, it is not always clear how different alarms, maintenance actions, downtimes and such are linked with each other. Therefore, within the OMCE approach the Event List was introduced.
An Event List can be best visualized as a list showing events per turbine in chronological order. It should include all fields that are relevant for further processing in order to obtain information about the failure behavior of components and use of equipment and spare parts. Within the context of the OMCE, an “event” is considered as a (sequence of) maintenance action(s) to prevent or correct turbine malfunctions. The total duration of an event is often longer than the sum of the individual maintenance actions. Maintenance actions can be remote resets, visits with technicians only, or the replacement of large components.
The event list is located within the OMCE concept between the raw wind farm data and the BBs, as shown in Figure 1 (above). The data does not have to be stored in the event list directly, but the event list could be extracted from different databases. This requires that the data in these databases be correlated. The following requirements are relevant when constructing an event list suitable for data analysis: ability to combine data from various data sources; ability to make relations between event and corresponding maintenance action(s); set out events per turbine in a chronological order; contain sufficient detail to determine input parameters; and integration with works management systems, for example SAP or ultimo.
Data analysis requires tools
Once operational data is available in a structured format such as that proposed by the Event List, it is suitable for further processing in order to obtain insight and knowledge about the wind farm’s performance, and also to obtain required input parameters for the various cost modeling tools.
Within the context of the OMCE approach, four Building Blocks may be defined as data processing tools. These are: BB Operation and Maintenance, BB Logistics, BB Usage & Load Monitoring and BB Health Monitoring.
The Operation & Maintenance BB has a twofold objective. Firstly, it is designed to generate information about the failure behavior of the components, in order to assess the adequacy of the maintenance strategy. Secondly, it generates updated figures on failure rates (and failure modes) and repair actions on components to be used as input for the OMCE-Calculator.
Information on failure behavior can be obtained from the structured data in the Event List by means of a ranking analysis to show the contribution of each system or component to the overall number of failures. Next, a trend analysis can be performed to determine the corresponding failure frequencies. The trend analysis is typically performed based on a cumulative sum control (CUSUM) plot of observed failures.
To process the structured data with the goal of obtaining information on equipment and spare parts used, the Logistics BB also has a twofold objective similar to that of the Operations & Maintenance BB.
Here the BB should, firstly, generate information about the use of logistical aspects such as equipment, personnel, spare parts and consumables, in order to assess the adequacy of vessel and spare part usage. Secondly, it should generate updated figures on the logistical aspects of an event, including accessibility, repair times, number of visits, delivery time of spares and so forth, to be used as input for the OMCE-Calculator.
The logistics aspects can be derived from the structured data in the Event List by an analysis of equipment, crew size, spare parts, repair time and so on for the different types of recorded events, for example corrective, condition-based.
The Usage & Load Monitoring and Health Monitoring BBs both use the raw wind farm data as a direct input to generate information on the remaining lifetime of various wind turbine components.
Currently ECN is developing the Usage & Load Monitoring BB, while the Health Monitoring BB is seen as a third party tool, such as drivetrain condition monitoring systems or inspection results.
O&M Cost Modeling
Within the context of the OMCE approach, the OMCE-Calculator tool, which will calculate the expected O&M effort and associated costs (amongst others in terms of downtime, costs, and revenue losses), is also being developed. The tool has been designed to model the three following types of maintenance: unplanned corrective maintenance; condition-based maintenance and calendar-based (preventive) maintenance.
The OMCE-Calculator is a time-domain simulation program which is designed to assist operators of wind farms to determine the optimal O&M strategy during the operational phase of a wind farm. The estimates of the future O&M costs also include a quantification of uncertainties in costs and downtime. Once the O&M aspects have been analyzed for the ‘as-is’ or baseline situation with the tool, the operator may analyze different O&M scenarios (by changing the input parameters) and select the most cost-effective one.
The tool is best used with operational data, to be analyzed as indicated by the connection through the BBs. It can also be used with generic data in order to use its features designed to optimize the maintenance strategy. Data required includes failure rates, expected time-to-failures, preventive maintenance, repair strategies, wind and wave statistics, costs, lead time of vessels and spare parts and so on. After a number of simulations are executed, the output is presented in the form of plots and tables with costs, downtimes, cost drivers and uncertainties.
To demonstrate how such a tool can be used in practice, the sensitivity of a maintenance scenario to a change in the number of workboats was studied by varying the number of available vessels for corrective maintenance from one to six. The objective of this study was to minimize the total costs due to O&M, which is defined as the sum of the direct O&M costs (repair costs made) and indirect O&M costs (revenues lost due to downtime). A graph is depicted in Figure 2, above, which shows the average total O&M costs and the two individual cost components.
We see that the trend of revenue losses versus the number of available workboats decreases, since less waiting time for resources is required. The direct maintenance costs increase since the workboats are assumed to be operator-owned and each vessel will lead to additional fixed and variable costs. The total O&M costs are at minimum when four workboats are available at the wind farm.
Future Research on Modeling
In ECN’s view, it is important that wind farm operators and OEMs structure their operational data collection processes and apply data analysis tools to generate both general insight in wind farm behavior and inputs for cost modeling tools.
Cost modeling tools can be applied to determine the future requirements for O&M effort with better accuracy and to optimize the maintenance strategy to reduce O&M costs.
René van de Pieterman is a researcher in wind energy systems at ECN.