Description
TitleAccessibility forecasting for offshore wind farm maintenance operations
Date Created2021
Other Date2021-05 (degree)
Extent1 online resource (vii, 40 pages)
DescriptionOffshore wind is a rapidly growing renewable energy source that is set to play an essential
role in the U.S. electricity system in the near future. Despite the promising potential, the
offshore wind industry still faces fundamental barriers pertaining to its typically high
operations & maintenance (O&M) expenditures, which contribute around 30% of offshore
wind’s cost of energy. Conducting O&M activities in offshore wind farms often entails
sending a crew and/or equipment to the turbine site when wind and sea conditions are
within safe operational modes, in order to ensure safe transfer of the personnel and success
of the assigned maintenance task. Therefore, accurate knowledge about the accessibility of
the offshore maintenance site plays a pivotal part in scheduling offshore maintenance. Our
analysis of real-world offshore data recorded at the New York/New Jersey (NY/NJ) Bight,
in proximity to at least three future offshore wind projects, reveals that a turbine site can
be inaccessible for 53% of its operational time, with up to six consecutive days of sustained
inaccess. A turbine requiring maintenance, if left unattended due to inaccess, can result in
significant production losses, as well as delayed maintenance activities, potentially turning
minor repairs into major failures (e.g., top-tower to down-tower repairs). It is thus
understandable that accurate accessibility forecasts enable wind farm operators to schedule
cost-optimal maintenance activities, thereby reducing downtime and preventing production
losses. Current practices for accessibility prediction rely on applying deterministic safety
thresholds on point forecasts of wind speed and wave height to identify accessibility
windows. Recent literature has focused on replacing those deterministic forecasts with
probabilistic formulations, which can provide the O&M operator with richer information
about accessibility states. Still, however, the large majority of those efforts adopt a
sequential paradigm, wherein wind speed and wave height are first forecast a priori, and
then accessibility predictions are derived by post-processing those forecasts. Instead, this
thesis proposes an alternative approach, which is called hereinafter the “Direct
Accessibility Model,” or in short “DAM”. Unlike the sequential paradigm for accessibility
forecasting, DAM directly aims at minimizing the accuracy loss of the accessibility
forecasts, rather than that of its constituent variables, i.e., wind speed and wave height.
Tested on real-world wind and wave data from the NY/NJ Bight, and with forecast horizons
up to three days ahead, DAM yields accessibility forecasts which are, on average, 2.73%
more accurate than prevalent benchmarks in the offshore wind literature and practice
NoteM.S.
NoteIncludes bibliographical references
Genretheses, ETD graduate
LanguageEnglish
CollectionSchool of Graduate Studies Electronic Theses and Dissertations
Organization NameRutgers, The State University of New Jersey
RightsThe author owns the copyright to this work.
Statistical Profile