HomeMy WebLinkAboutDesktop Wind Resource Study - Dillon Platform - Cook Inlet - Jan 2024 - REF Grant 7015023 V3 Energy, LLC Page | 1
Technical Memorandum
To: Mike Salze�, Homer Electric Associa�on
From: Douglas Vaught, P.E., V3 Energy, LLC, Anchorage, Alaska
Date: January 17, 2024
Subj: Desktop Wind Resource Characteriza�on of Dillon Pla�orm, Cook Inlet, Alaska
Introduc�on
This technical memorandum presents the an�cipated wind resource, as predicted by Global Wind Atlas
modeling, at Dillon Pla�orm in Cook Inlet, Alaska in support of Homer Electric Associa�on’s award from
Alaska Energy Authority for Renewable Energy Fund Round 15 for Cook Inlet Oil Platform Wind Project.
Project Descrip�on
Homer Electric Associa�on (HEA), through its genera�on subsidiary Alaska Electric & Energy Coopera�ve
(AEEC), is inves�ga�ng the poten�al for construc�on of approximately 30 MW of wind energy genera�on
capacity on or near the Kenai Peninsula. As part of that effort, the Cook Inlet Oil Platform Wind Project
will evaluate construc�on of three-to-four offshore wind turbines on the A, C, Baker, and Dillon offshore
oil pla�orms in Cook Inlet near Nikiski, Alaska. The turbines would connect electrically to the electrical
grid at HEA’s Bernice Lake Substa�on in East Forelands area of Nikiski.1
Project Sites
Although the four oil pla�orms noted above are presented as of equal interest in AEEC’s Renewable
Energy Fund (REF) Round 15 applica�on, HEA has indicated that Dillon Pla�orm is primary interest with
the other three pla�orms secondary interest for wind power development. This priority is logical with
reference to Figure 1 as Dillon Pla�orm is nearest Nikiski and the Bernice Lake Substa�on.
Per HEA communica�ons with V3 Energy, LLC, la�tude and longitude of the pla�orms are as follows
(note that coordinates are datam NAD83):
• Dillon Pla�orm (60.734670, -151.515210)
• Baker Pla�orm (60.828651, -151.485942)
• MGS-A Pla�orm (60.795843, -151.495668)
• MGS-C Pla�orm (60.764040, -151.502160)
1 From REF Round 15 Application Summaries, AEA website
V3 Energy, LLC Page | 2
Figure 1: Baker, A, C, and Dillon Oil Platforms in Cook Inlet, Google Earth image
Global Wind Atlas
The Global Wind Atlas (GWA) is a free, web-based applica�on developed to help policymakers, planners,
and investors iden�fy high-wind areas for wind power genera�on virtually anywhere in the world, and
then perform preliminary calcula�ons. The GWA facilitates online queries and provides freely
downloadable datasets based on the latest input data and modeling methodologies.
The current itera�on of the GWA (version 3.3) is the product of a partnership between the Department
of Wind Energy at the Technical University of Denmark (DTU Wind Energy) and the World Bank Group
(consis�ng of The World Bank and the Interna�onal Finance Corpora�on). Work on GWA 2.0 and GWA
3.0 was primarily funded by the Energy Sector Management Assistance Program (ESMAP), a mul�-donor
trust fund administered by The World Bank and supported by thirteen official bilateral donors. It is part
of the global ESMAP ini�a�ve on renewable energy resource mapping that includes biomass, small
hydropower, solar energy, and wind energy. GWA 3.0 builds on an ongoing commitment from DTU Wind
Energy to disseminate data and science on wind resources to the interna�onal community.
Purpose
The GWA primarily supports wind power development during the explora�on and preliminary wind
resource assessment phases prior to the installa�on of meteorology measurement sta�ons on site. It
Kenai
Nikiski
V3 Energy, LLC Page | 3
also serves as a useful tool for governments to beter understand their wind resource poten�al at
provincial and local levels.
Objec�ves
The objec�ves of the GWA are to:
• provide wind resource data accoun�ng for high-resolu�on effects.
• use microscale modeling to capture small-scale wind speed variability (crucial for beter
es�mates of total wind resource).
• use a unified methodology over the en�re globe and update the Global Wind Atlas as
methodologies develop.
• ensure transparency about the methodology used.
• support the verifica�on of the results in the long-term by coupling to measurement data and
campaigns.
The correct usage of the Global Wind Atlas dataset is for aggrega�on, upscaling analysis and energy
integra�on modeling for energy planners and policy makers. It is not advisable to solely use the data and
tools for wind farm site selec�on.
Limita�ons of GWA
The GWA uses two major modeling components that can introduce uncertainty into the calcula�ons.
These components are mesoscale modeling and microscale modeling.
Uncertain�es associated with the mesoscale modeling include representa�veness of the large scale
forcing and sampling, model grid size, descrip�on of the surface characteris�cs, model spin-up,
simula�on �me and modeling domain size.
Uncertain�es associated with the microscale modeling include the orographic flow model within WAsP 2,
the surface descrip�on, and departures from the reference wind profile. Concerning the orographic flow
model, the model performs well when the surrounding terrain is sufficiently gentle and smooth to
ensure mostly atached flows. With the global coverage of the GWA, we use the BZ-model in areas
beyond its recommended opera�onal envelope.
The GWA website allows users to see where the flow modeling is likely to be increasingly uncertain, by
adding a RIX layer to the set of maps. The RIX layer represents ruggedness index and is an objec�ve
measure of the steepness or ruggedness of the terrain. Large RIX values will lead to large errors in the
flow modeling, most likely leading to an overes�ma�on of mean wind speeds on ridges and hilltops. We
therefore recommend users to inspect the RIX of their region of interest.3
Model of Cook Inlet-wide Wind Resource
The GWA model of the Cook Inlet wind resource at the 100-meter level (refer to Figure 2) indicates
rela�vely high wind speeds over water compared to surrounding lower eleva�on land masses. The
2 Wind energy industry-standard software - WAsP
3 Global Wind Atlas (About menu; this and preceding sections)
V3 Energy, LLC Page | 4
model predicts highest over-water wind speeds in Upper Turnagain Arm and lower Cook Inlet with more
moderate, but s�ll robust, wind speeds in central Cook Inlet.
Figure 2: The GWA model of 100-meter level wind speeds in Cook Inlet (Dillon Platform location noted with blue pin)
Dillon Pla�orm Modeled Wind Resource
Pinning the loca�on of Dillon Pla�orm within the GWA so�ware creates a 3 km x 3 km (9 km2) box
around the pla�orm with data presented for the 10% windiest area, or 90th percen�le probable wind,
within the box. This is helpful in complex terrain where wind speeds vary considerably over short
distances. Over a large expanse of water, this probability feature is insigh�ul primarily because it
demonstrates a rela�ve lack of wind speed variability over longer distances (refer to Figure 3 and Figure
4). At the 100-meter level at Dillon Pla�orm, GWA predicts a 7.74 m/s mean wind speed with extremely
low variability within the 9 km2 box.
V3 Energy, LLC Page | 5
Figure 3: Dillon Platform, GWA-modeled wind speed in 9 km2 boxed area, 100-meter level
Figure 4: GWA-modeled distribution of wind power density and wind speed within 9 km2 boxed area of Dillon Platform
The GWA models prevailing winds at Dillon Pla�orm as primarily north-northeasterly with secondary
south-southwest winds (refer to Figure 5). This matches the secondary intercardinal orienta�on of Cook
Inlet, especially when considering the long fetch of the Susitna River to the north-northeast instead of
Knik Arm. The modeled wind power rose though indicates that power-producing winds are
overwhelmingly northerly and north-northeasterly (refer to Figure 5).
Figure 5: GWA-modeled wind frequency and power roses at Dillon Platform
V3 Energy, LLC Page | 6
Modeled monthly wind speed variability (refer to Figure 6) predicts higher winds during the cold
weather months compared to summer, which is expected and generally aligns with residen�al electric
load demand. Modeled hourly, or diurnal, wind speed variability predicts highest winds in the early
morning hours, peaking at 6:00 a.m. (refer to Figure 6), which is less aligned with load demand.
Figure 6: GWA-modeled monthly and hourly wind speed variability at Dillon Platform
Wind Power Class
NREL’s seven wind power classes are based on wind power density (WPD), calculated with the following
equa�on: 𝑃𝑃𝐴𝐴=1/2𝜌𝜌𝑣𝑣3
Classifica�on by wind power density or mean wind speed is not a design standard for wind turbine
manufacturers, but rather a useful and intui�ve method to assess and compare sites (refer to Table 1).
The reader is cau�oned to note that wind turbine energy yield versus wind class is a non-linear
rela�onship.
Class WPD (W/m2) (50 m) Mean Speed (m/s) (50 m) Rating
1 <200 <5.6 Poor
2 200 – 300 5.6 – 6.4 Marginal
3 300 – 400 6.4 – 7.0 Fair
4 400 – 500 7.0 – 7.5 Good
5 500 – 600 7.5 – 8.0 Excellent
6 600 – 800 8.0 – 8.8 Outstanding
7 >800 >8.8 Superb
Vertical extrapolation based on 1/7 power law (0.14 wind shear exponent)
Mean speed based on Rayleigh distribution; Weibull k = 2.0
Table 1: NREL classification of wind power density, at 50-meter level
Dillon Pla�orm
For the loca�on of Dillon Pla�orm in Cook Inlet, at the 50-meter level, the GWA predicts a 7.11 m/s
mean wind speed and a 601 W/m2 mean WPD. With reference to Table 1, this equates to a Class 4
(good) wind resource by wind speed and a Class 5 (excellent) to Class 6 (outstanding) wind resource by
WPD. This difference can be explained by the colder than standard condi�ons mean air temperature of
Cook Inlet and perhaps also by a possible non-Rayleigh distribu�on of wind speeds (where Weibull k ≠
2.0).
V3 Energy, LLC Page | 7
IEC 61400-1 Classifica�on
Design standards for wind power are published by the Interna�onal Electrotechnical Commission (IEC),
based in Geneva, Switzerland, with regional centers in North and South America, Africa, and Asia. The
IEC was founded in 1906 and is the world’s leading organiza�on for all electrical, electronic, and related
technologies. Their mission: achieve worldwide use of IEC International Standards and Conformity
Assessment systems to ensure the safety, efficiency, reliability, and interoperability of electrical, electronic
and information technologies, to enhance international trade, facilitate broad electricity access and
enable a more sustainable world.
IEC’s Technical Commitee (TC) 88 covers wind genera�on via the 61400 series. The standard of interest
is IEC 61400-1, Wind energy genera�on systems – Part 1: Design requirements. Although a new 4th
edi�on of IEC 61400-1 was published in 2019, most wind turbines of interest to HEA were designed
under 3rd edi�on (2005) standards and hence it will be the focus of this review.
Six parameters or analyses comprise IEC 61400-1 wind classifica�on as listed below, with the simplified
classifica�on (that references only extreme wind and turbulence intensity) in Table 2. They are:
• Extreme wind
• Turbulence intensity
• Wind shear
• Wake turbulence
• Flow inclina�on
• Wind distribu�on
Table 2: IEC 61400-1, 3rd edition, simplified wind classification
Wind Class I II III S
Vref (m/s) 50.0 42.5 37.5 Values specified
by the designer A (TIref) 0.16
B (TIref) 0.14
C (TIref) 0.12
Extreme Wind
The classifica�on of extreme wind is by Vref, the reference wind speed; the highest measured or probable
10-minute average wind speed in a 50-year return period. This can be calculated probabilis�cally with a
periodic maxima analysis (by Gumbel distribu�on), the method of independent storms (also a Gumbel
distribu�on), and EWTS II.4 Table 3 includes Vref as presented in Table 2, plus Ve50, the maximum
measured or probable 3-second average or gust wind speed in a 50-year return period. Ve50 is defined as
Vref x 1.4 (for equivalent height), though the later is not part of the IEC 61400-1, 3rd edi�on simplified
classifica�on noted in Table 2.
Table 3: IEC 61400-1, 3rd edition, extreme wind classes
Wind Class I II III S
Vref (m/s) 50.0 42.5 37.5 Designer spec.
Ve50 (m/s) 70.0 59.5 52.5
4 EWTS II (European Wind Turbine Standards II) does not consider peak winds, rather only mean wind speed and
Weibull k value.
V3 Energy, LLC Page | 8
Dillon Pla�orm
The GWA only predicts Vref or Ve50 qualita�vely, hence numerical informa�on for IEC classifica�on must
be derived from wind measurement with a met tower or Lidar. Cook Inlet is exposed to occasional
tropical-origin North Pacific storms, though the energy of the storms typically diminishes further up the
inlet compared to the mouth. These storms can yield high extreme wind speeds, even at loca�ons with
moderate mean winds.
Besides mean wind speed, calcula�ng extreme wind probability from met tower or Lidar data is the
primary objec�ve of the wind data collec�on effort as the extreme wind classifica�on (I, II, III, S)
determines wind turbine model op�ons and subsequent wind turbine capacity factor (which equates to
annual energy produc�on). That noted, the GWA graphically predicts Class II-C (or 2C) winds at Dillon
Pla�orm (see Figure 7).
Figure 7: Dillon Platform, GWA-modeled IEC Class 2C fatigue loads at 100-meter level
Turbulence Intensity
The turbulence intensity (TI) is a dimensionless number defined by the standard devia�on (σ) of the
wind speed within each �me step (10 minutes for wind power analysis) divided by the mean wind speed
(V) over that �me step. 𝑇𝑇𝑇𝑇= 𝜎𝜎𝜎𝜎𝑉𝑉𝜎𝜎�
IEC 61400-1, 3rd ed., defines turbulence categories based on mean turbulence intensity at a fixed wind
speed of 15 m/s (see Table 4).
Table 4: IEC 61400-1, 3rd edition, turbulence categories
Turb. Category S A B C
TI at 15 m/s >0.16 0.14-0.16 0.12-0.14 <0.12
V3 Energy, LLC Page | 9
Dillon Pla�orm
As with Vref or Ve50, the GWA only predicts TI qualita�vely, hence numerical data for IEC classifica�on
must be obtained from wind measurement with a met tower or Lidar. Note however that with a very
long fetch of open water in the prevailing wind direc�ons, it is exceedingly unlikely that turbulence
intensity at Dillon Pla�orm hub height eleva�ons will be in the higher S, A, or B categories. As one would
an�cipate and as noted in Figure 7, the GWA predicts turbulence category C winds at Dillon Pla�orm.
Wind Shear
A wind shear, or power law, exponent α is calculated by the following equa�on where V = wind speed
and Z = height above ground level. α=0 would indicate no wind shear and α=0.2 would indicate very high
wind shear.5 𝑉𝑉(𝑧𝑧)=𝑉𝑉(ℎ𝑢𝑢𝑢𝑢)× �𝑍𝑍𝑍𝑍ℎ𝑢𝑢𝑢𝑢��𝛼𝛼
Dillon Pla�orm
The GWA predicts mean wind speed at five levels: 10, 50, 100, 150, and 200 meters. Using the previously
referenced 50- and 100-meter level mean winds speeds and calcula�ng for α using the wind shear
equa�on, a power law exponent of 0.122 is derived. Using predicated mean wind speed at the higher
heights of 100 meters and 150 meters and calcula�ng for α, a power law exponent of 0.100 is derived.
Either is low and highly desirable for wind power development.
Another method to assess wind shear is via roughness length, which is an alternate measure of the rate
which wind speed changes with height above ground level.6 The GWA-predicted roughness length for
Dillon Pla�orm is 0.0002 meters (refer to Figure 8), which translates to a terrain descrip�on of “calm,
open sea.” This too is indica�ve of low wind shear, which is highly desirable for wind turbine opera�ons.
Figure 8: Dillon Platform, GWA-predicted roughness length (in meters)
5 High wind shear results in considerable mechanical stress from torquing of the main rotor shaft as aerodynamic
lift will be higher at the top of the rotor swept area than at the bottom.
6 Windographer software Help.
V3 Energy, LLC Page | 10
Wake Turbulence
For comparison with the normal turbulence model, IEC 61400-1 suggests an effec�ve turbulence
intensity, which is an ideal turbulence independent on wind direc�on and expected to cause the same
fa�gue damage as variable turbulence in winds from all direc�ons. The effec�ve turbulence intensity
includes added turbulence from wakes of neighbor turbines.7
Dillon Pla�orm
Neither the GWA or wind resource measurement with a met tower or Lidar predict or measure wake
turbulence by themselves. Rather this is es�mated with reference to prospec�ve wind farm turbine array
layout designs. For HEA’s Cook Inlet project, presuming only one wind turbine per pla�orm and
considering that the pla�orms are fixed loca�ons, inter-turbine distances are presently known. The
Baker, A, C, and Dillon pla�orms are not exactly equidistant but average about 3,500 meters. For sake of
example, a General Electric Haliade 150-6 MW offshore wind turbine 8 has a rotor diameter of 150
meters, which, for a 3,500-meter inter-turbine distance, is 23 rotor diameters. This is well above the
minimum recommended spacing distance for wind turbines, whether perpendicular to or parallel with
the GWA -predicted prevailing winds (refer to Figure 5).9
Flow Inclina�on
A wind flow vector not exceeding 8 degrees from horizontal (plus or minus).
Dillon Pla�orm
The GWA does not predict wind flow angle, or inclina�on, hence this informa�on for IEC classifica�on
must be derived from wind measurement with a met tower or Lidar. It is exceedingly unlikely however
that wind crossing a long fetch of water such as Cook Inlet will deviate significantly from 0° at the power-
producing wind speeds of 4 m/s to 25 m/s.
Wind Distribu�on
A wind speed, or histogram, where a Weibull func�on10 yields a unitless shape factor (k) of 2.0 (known
as a Rayleigh distribu�on) or less.
Dillon Pla�orm
The GWA does not predict wind speed distribu�on or histogram (Weibull shape factor), hence this
informa�on for IEC classifica�on must be derived from wind measurement with a met tower or Lidar. It is
highly unusual, however, to observe wind speed distribu�ons with Weibull k > 2.0 in loca�ons outside
mountainous areas where bi-modal wind variability 11 is occasionally observed.
7 The IEC 61400-1 turbine safety standard - WAsP.
8 Haliade 150-6MW Offshore Wind Turbine | GE Renewable Energy.
9 The Danish Wind Energy Association says: “As a rule of thumb, turbines in wind parks are usually spaced between
5- and 9-rotor diameters apart in the prevailing wind direction, and between 3- and 5-rotor diameters apart in the
direction perpendicular to the prevailing winds.” WAsP software help section.
10 Weibull distribution - Wikipedia.
11 Two wind speed peaks: a normal lower peak and a secondary high peak
V3 Energy, LLC Page | 11
Wind Turbine Energy Yield
Lidar data – the longer �me the beter so annual variability can be assessed with probability predic�ons
P90, P50, etc. 12 for annual energy produc�on – will enable HEA to es�mate annual energy produc�on
(AEP) with a high degree of confidence, but for ini�al modeling the GWA includes capacity factor
es�mates for representa�ve-type IEC Class I, II, and III turbines. Because the GWA predicts an IEC Class II-
C (or 2C) wind resource at Dillon Pla�orm (refer to Figure 7), the predicted IEC Class II wind turbine gross
capacity factor13 at Dillon Pla�orm is 0.45 (at the 150 meter level, a possible hub height of wind turbines
constructed on the pla�orm deck). To convert to (gross) AEP, mul�ply by 8,760 hours/year. So, for a 3
MW wind turbine at the reference hub height: 𝐴𝐴𝐴𝐴𝑃𝑃=3 𝑀𝑀𝑀𝑀 × 0.45 × 8,760 ℎ𝑟𝑟 𝑦𝑦𝑟𝑟= 11.8 𝐺𝐺𝑀𝑀ℎ/𝑦𝑦
Figure 9: Dillon Platform, GWA-predicted IEC Class II wind turbine capacity factor, 150-meter level
For reference, the GWA predicts a 0.41 gross capacity factor for an IEC Class I wind turbine at Dillon
Pla�orm (at 100 meters). For a 3 MW capacity wind turbine, this equates to 10.8 GWh/y AEP. For an IEC
Class III wind turbine, the GWA predicts a 0.48 gross capacity factor, which for a 3 MW capacity wind
turbine equates to a 12.6 GWh/y AEP.
Baker, A, and C Pla�orms
At Baker Pla�orm, the furthest north of the four pla�orms, the GWA predicts a 7.76 m/s mean wind
speed and 717 W/m2 WPD at the 100-meter level, which are only slightly higher than those predicted at
Dillon Pla�orm. Considering an IEC Class II wind turbine at the 150-meter level, the GWA predicts an
equivalent 0.45 gross capacity factor at Baker, A, and C Pla�orms as at Dillon Pla�orm. A wind turbine
located at Baker Pla�orm may have a slightly higher gross capacity factor and hence AEP, but the
12 Probability of Exceedance, see Terminology explained: P10, P50 and P90 (dnv.com) for explanation.
13 Mean power output divided by rated power.
V3 Energy, LLC Page | 12
difference would be likely be minimal and probably offset by higher development costs as Baker Pla�orm
is further offshore than Dillon Pla�orm and hence further away from Bernice Lake Substa�on, thus
requiring longer electrical distribu�on connec�on.
For planning purposes at this stage of the project, it is reasonable to assume equivalent wind turbine
AEP at Dillon, Baker, A, and C Pla�orms, and given their distance from each other, negligible wake effects
should all four be developed.