Offshore Micrositing - Meeting The Challenge V. Barth; DEWI GmbH, Oldenburg English Introduction Offshore wind is increasingly gaining importance in the wind energy sector. While countries like the UK and Denmark already have years of experience with offshore installations, the German market has only gathered momentum recently, marked by the start of operation of the first German offshore wind farm Alpha Ventus in 2010. Of the more than 80 areas in the North Sea that have been identified by the German authorities suitable for offshore wind farms, two are currently under construction, 22 are in an advanced development state and have already been approved by BSH (Bundesamt für Seeschifffahrt und Hydrographie/ Federal Maritime and Hydrographic Agency), while still more than 50 are in the development pipeline. In the Baltic Sea where the first purely commercial wind farm Baltic 1 went into operation in 2011, about 20 areas are available of which three projects are already approved by BSH. This situation has clearly been reflected in the micrositing requests that have reached DEWI. While the first assessments about five years ago were more explorative in nature and covered a lot of different areas, the latest requests have often been recalculations after layout refinements or assessments of new turbine types, indicating quite concrete planning steps. Until now, more than 60% of the approved offshore wind farms in the North Sea have been investigated by DEWI, covering about 55% of the planned capacity of these sites. For the Baltic Sea, the respective figures are 10% of all parks and 17% of capacity. Additional sites have been investigated in the UK, Sweden, France and South Korea. The Offshore Challenge From an engineer s or project planner s perspective, building wind farms off the coast in waters more than 30 m deep is a huge challenge compared to onshore sites: only few turbines are suitable for rough offshore conditions, foundations must withstand higher wind speeds and the additional loads from waves and sea currents. Also, the logistics for construction and maintenance are a lot more difficult to manage, and wind farms are larger. In the end, investment sums and associated risks are typically much higher, making financing a difficult issue as well. At first glance, the only piece of the puzzle that remains unchanged appears to be the wind resource assessment: take measured wind data, use some model tool to do the transfer to the wind farm site, and use the resulting wind statistics together with the turbine s power curve to calculate DEWI MAGAZIN NO. 40, FEBRUARY 2012 55
the expected energy yield. Given the fact that the offshore terrain is flat and there are virtually no spatial changes in roughness, an energy yield assessment should be even easier than for onshore sites. On second thought it soon becomes clear that this is too simple a picture. It starts with recognizing that erecting a met mast offshore is a much more demanding task than onshore, so that one cannot easily have a met mast at or near the site. Secondly, the wind atlas modelling method, which is standard for onshore sites, is no longer applicable when the wind farm site is far away from the met mast. And since offshore wind farms are typically much larger than onshore sites, the wind conditions within these wind farms tend to be different from those onshore, so that the usual models used to calculate energy yields need to be modified to capture these effects. Micrositing Methods Applied at DEWI It should have become clear by now that simply transferring the onshore micrositing concepts to offshore sites is quite difficult and would at least introduce large errors into the calculation. What does DEWI do to overcome these problems in order to arrive at reasonable and accurate results? Below, the central elements are described in more detail. Offshore meteorological data Wind data for reasonable wind resource and energy yield assessments need to be measured close to hub height and at several altitudes in order to obtain wind profile information. For that reason, the numerous measurements from oceanographic buoys and usual meteorological stations at the coast cannot be used, as these data are mostly taken at 10 m altitude only. However, erecting an offshore met mast of up to 100 m height requires the installation and maintenance of an offshore platform, which is difficult and an expensive upfront investment. For that reason, data from offshore masts are rare and usually quite difficult to obtain from the owners. Fortunately, the German government initiated the installation of three offshore platforms (Fino1 and Fino3 in the North Sea, Fino2 in the Baltic Sea) which, among other data, provide raw wind data that are publicly available from BSH. DEWI is responsible for the met mast setup and the ongoing measurements at the Fino1 platform, and has thus long experience with this data set. For micrositing purposes, numerous efforts are undertaken to further improve the raw data, including filling of data gaps, long-term correction and a correction of effects introduced by the mast structure, which can be quite large due to the massive structure of the offshore mast. The resulting data set, which is used for wind resource assessments, covers the 6 years period from September 2003 to August 2009. Later data are affected by the wake impact of Alpha Ventus; its correction is an ongoing research topic at DEWI. From September 2009 on, data are available from the Fino3 platform. DEWI is not involved in these measurements but obtains these data via the BSH server. The same holds for the Fino2 data in the Baltic Sea. In both cases, numerous mast-specific corrections are applied to remove mast effects and to fill data gaps. Mesoscale modelling Once there is a proper data set from the selected met mast, the next issue is to obtain data valid for the actual wind farm site, which can be several dozens of kilometres away given the sparseness of available met masts. However, over larger distances the standard wind atlas methods (as applied e.g. in the well-known WASP model) become increasingly less precise, as these are built on the assumption that the geostrophic wind conditions are the same at the met mast and at the wind farm site. This is no longer the case when larger distances are involved. For such cases, other approaches are needed. The common method is to use mesoscale models of the atmosphere. This class of models is designed to describe the atmospheric conditions over areas of several hundred square kilometres by integrating the hydrodynamic equations of motion. It is widely used as regional weather prediction models. Well-known examples are the MM5 model, which is used by DEWI, or the WRF model. At the boundaries, these models need to be supplied with external data. While weather prediction models are driven with recent weather observations to predict the weather for the next few days, DEWI supplies large scale global model data like NCEP reanalysis data, which are available on a grid with a resolution of 2.5 x2.5 (160 km x 280 km in the North Sea Area). The mesoscale model is then used to simulate the wind conditions over the complete model area on a fine grid with a typical grid length in the order of few kilometres. It is also possible to use a mixture of long-term wind measurements and localized model results to drive the mesoscale model. Such an approach, which requires huge efforts in collecting data and local modelling, has recently been undertaken for the Baltic Sea (by Risø-DTU) and the North Sea (by the Norsewind research project). A quick comparison with these outcomes largely confirms DEWI s calculation results. If there is only one met mast available, DEWI uses the MM5 results at the locations of the met mast and the wind farm site to transfer the time series from the met mast to the site by scaling the wind speed but leaving the wind direction untouched. This gives good results for the average wind speed, but ignores possible changes in the wind direction distribution. Even though this is recognized as an important issue when designing the wind farm layout, there is no way to validate the consistency of the wind direction changes as indicated by MM5, since there is only one single measurement location. The simple wind speed scaling is thus the best thing that can be done. Only recently, this situation has changed in the North Sea as sufficiently long-term of Fino3 became available. These measurements confirm the wind direction data from our MM5 simulations (see also article on page 67 in this issue). Encouraged by this, DEWI is currently working towards a method that makes both the wind speed from Fino1 and Fino3 and the wind direction information provided by MM5 available for the site assessments. This will probably make wind resource assessments in the North Sea even more precise. 56 DEWI MAGAZIN NO. 40, FEBRUARY 2012
Werbung Windforce 12 1/1 s/w oder 4c book your booth now! HOCHTIEF Germany s first offshore trade fair 2012 in Bremen The Windforce 2012 German Offshore Conference & International Fair is aimed at all aspects of the offshore industry: developers, producers, suppliers, shipyard industry and other marine services, logistics, services, training and qualification. Don t miss out! 15% discount for all WAB members Information and booking for exhibitors: www.windforce2012.com
Fig. 1: Planned offshore wind farms in the North Sea (Source: Federal Maritime and Hydrographic Agency) Energy yield calculations and wind farm modelling Once the wind conditions at the wind farm site are known from measurements and mesoscale model results, the energy production can be estimated. For offshore sites, potential difficulties may arise from the available power curve. As many of the offshore turbines on the market today have hardly left the prototype status, power curve measurements of dedicated offshore turbines are rare and most of the power curves are based on theoretical model calculations by the manufacturers. Since these models cannot be validated by DEWI, the resulting power curves are considered to be quite uncertain from a technical point of view. In few cases power curve measurements exist already, which generally have reduced uncertainties compared to the theoretically calculated curves. However, one has to bear in mind that there are differences in the atmospheric conditions at the onshore sites where the measurements have usually been made, and at the offshore sites where the turbines will eventually operate. These differences comprise mainly the turbulence conditions and the vertical wind profile and are particularly important given the large rotor sizes of offshore turbines. Therefore, additional uncertainties are introduced by applying power curves measured onshore for offshore sites. Apart from that, offshore sites are relatively easy to handle. In contrast to onshore sites, topography is no issue and roughness is uniform and low, so that the free wind conditions can be assumed to be identical over the size of common offshore wind farms. The actual difficulties thus arise not from the site, but from the wind farm itself. Typical offshore wind farms are large (about 80 turbines) and equipped with huge turbines (rotor diameter 120 m and more). Within such wind farms, turbine wakes are the main reason for efficiency losses. For the micrositing consultant, the task is to capture these effects as precisely as possible. The most comprehensive way of doing this is to run a computational fluid dynamics (CFD) model of the wind flow through the wind farm. Numerous models with more or less similar capabilities and quality are available on the market. DEWI operates a model developed in-house, based on the Phoenics CFD core. The unrivalled precision of CFD models, however, comes at a high computational cost. The long runtime of a single CFD calculation makes these models less suited for many practical tasks. In particular comparing different turbine types on the same layout or optimizing the layout for maximum energy yield are tasks that will normally not be accomplished by using a CFD model. Fortunately, more cost efficient alternatives do exist: The method that is commonly applied by DEWI is using the linear wake model by Jensen (also known as PARK model). Its computational runtime is extremely small and it is known to deliver reasonably good results for smaller wind farms. To make this model usable also for larger wind farms, DEWI uses a non-standard parameter setting that has been calibrated against DEWI s CFD model. This calibration is designed to give good agreement between the results of the CFD model and the PARK model for the overall farm energy yield of various types of farm layouts, but is of course not completely exact for every single turbine location. 58 DEWI MAGAZIN NO. 40, FEBRUARY 2012
Fig. 2: Planned offshore wind farms in the Baltic Sea (Source: Federal Maritime and Hydrographic Agency) Additional services Besides the classical energy yield calculations, DEWI offers a number of additional services. One common service is layout optimization, where an optimization model is run in order to determine a turbine layout that produces the maximum energy possible at the site with a given turbine type. Helicopter corridors and exclusion zones for cables or pipelines can be defined as side conditions. Such calculations are performed using the calibrated PARK model and the optimisation routines of the GH WindFarmer software. Another important aspect is the calculation of atmospheric load parameters, in particular turbulence intensity and extreme wind assessments. These are offered by DEWI as core elements of its Design Basis package. Ambient turbulence is determined in a straightforward way from the standard deviation of the wind speed measurements. The actual determination of the additional turbulence created by the wind turbines themselves is calculated according to the well-known model by Sten Frandsen, as requested by the IEC 61400-1, using a model developed in-house. For extreme winds we usually apply the Lieblein method to extrapo late the storms found in the measured time series to time horizons of 50 or 100 years as requested by the IEC 61400-1 or other standards. While these tasks are offered on a regular basis, DEWI is always open to offer its knowledge, experience and creativity for other services and tasks beyond these well-established items. Similarly, we are not restricted to use data by the Fino platforms, and customers who have different data sets or planning a wind farm elsewhere on the globe are most welcome to contact us. Outlook The ongoing rapid development of the offshore sector promises further additional need for offshore micrositing services. About 60% of the German offshore sites are still in an early stage of development and will require closer investigation in the years to come. Similarly, the recent French call for tender will stimulate intense planning efforts in 2012 and 2013; and equipping the huge Round 3 areas in the UK will require huge investments, which call for proper assessment of wind resources and energy yield. Other developments will also offer new opportunities for the future of the offshore micrositing business. The tremendous improvement of LIDAR technology in recent years will probably make high-quality offshore wind measurements much simpler so that more data will become available. This will all the more be true if recent new developments of LIDAR buoys ( floating LIDAR ) actually pass the reality test in terms of accuracy and availability. Likewise, the increasing number of offshore wind turbines offers lots of data from nacelle anemometers, which could also be a source of information. Until now, however, nacelle anemometers still cannot compete with the quality of mast-based cup anemometer data. Similarly, also the modelling world is constantly evolving and new concepts are being developed. Recently, a number of new developments such as linearized CFD models have entered the field. DEWI is also constantly monitoring or is even involved in research activities that address hot topics like wake modelling or turbine loads in order to adopt up-todate technologies and new insights. DEWI MAGAZIN NO. 40, FEBRUARY 2012 59