You are here

Mid-Atlantic Offshore Wind Integration & Transmission Build Blocks


Updated: 2013-04-24

This page exists primarily for internal documentation for Mid-Atlantic Offshore Wind Integration & Transmission (MAOWIT) project, funded by the US Department of Energy (contract # DE-EE0005366). Other users are welcome to read what is here and download the data but no support is offered for these products. Questions and comments should be directed to our informational mailing address


The main project deliverable for the Mid-Atlantic Offshore Wind Integration & Transmission (MAOWIT) wind resource team during 2012 Quarter 4 was wind power forecasts for build blocks along the Atlantic Wind Connection (AWC). Blaise Sheridan, University of Delaware (UD) analyzed bathymetry and competing uses along the AWC and provided the wind forecasting team (Mike Dvorak, Sailor's Energy and Cristina Archer, UD) with geographic files of build-blocks for 5 different buildout scenarios of 8, 28, 40, 55, and 78 GW of installed offshore wind power capacity. For the PJM interconnection, these capacities correspond to 4, 12, 18, 24, and 34% penetration, respectively, based on 35% capacity factor assumption and the average 2010 PJM load of 79.6 GW.

Power forecast download and revision history

  • r5 (download CSV) - 2013-03-31: Same as r4 below but with the addition of the BOEM Wind Energy Areas (WEAs) and a more intuitive renumbering of the sub-blocks. The WEAs for New Jersey, Maryland, Delaware, and Virginia were carved out of the existing sub-blocks in r4, to explicitly evaluate the impact of development in these WEAs. Each sub-block is now numbered in a X.Y fashion, where X is the build-order and Y is the serial number of the block in that build-out.
  • r4 - 2013-03-08: Sub-block CSV files for Jan, Apr, Jul, and Oct 2010 contain the wind farm mean 90m wind speed for each sub-block. The WRF mean wind speed for the forecast is the mean of all WRF points contained within the sub-block. The WRF forecast power is the mean of the power calculated at each WRF point, with 5000 kW being maximum power output at a single turbine (i.e. (sum i=1..n, powerInKW(speed_i))/n ). Buoy 44020 (Nantucket Sound) was added to the list of buoys to fill in missing "actual" wind speed data. See the README file for more information.
  • r3 - 2013-01-04: Sub-block CSV files for Jan, Apr, Jul, and Oct 2010 contain fully developed wind farm power output for each sub-block, based on the 10Dx10D spacing and 90% array loss factor, as defined below.
  • r2 - 2012-12-31: Sub-block CSV files for Jan, Apr, Jul, and Oct 2010 contain fully developed wind farm power output for each sub-block, based on the 5Dx10D spacing and 90% array loss factor, as defined below.
  • r1 - 2012-12-29: Sub-block CSV files Jan, Apr, Jul, and Oct 2010 only contained the average turbine power for a single turbine in each block.

Buildout scenarios

Four buildout scenarios were defined over nine different blocks (1-9), located along the AWC. Nine different blocks were built out over four different build scenarios(1-4). Each sub-block is labeled as [block_ID].[build_order]. For example, sub-block 8.3 in Figure 1 would correspond to the third buildout of the eighth block. Nine build areas were identified by Sheridan, et al. (2012) based on bathymetric and use-conflict analysis.

Figure 1: Mean 90-m wind resource from 2006-2010 (Dvorak, et al., 2012), MAOWIT bulild blocks (Sheridan, et al., 2012), and NDBC observations used for "actuals" (buoy 44017 off Long Island, New York and 44020 in Nantucket Sound are not shown). Depth contours of 30, 50, and 200-m correspond to monopile, muti-leg, and floating-turbine foundations. Click on image for full-size map.

Table 1: Total wind power capacity for each build scenario.
Build order Turbines (count) Capacity (MW) Cumulative capacity (MW)
1 1638 8190 8190
2 3999 19995 28185
3 2353 11765 39950
4 2916 14580 54530
5 4650 23250 77780


Table 2: Complete description of each build block and build order corresponding to the sub-blocks in Figure 1.
Build order Block ID Turbines (count) Capacity (MW)
1 1 903 4515
1 2 245 1225
1 3 203 1015
1 4 287 1435

2 1 286 1430
2 2 481 2405
2 3 525 2625
2 4 441 2205
2 5 492 2460
2 6 626 3130
2 7 540 2700
2 8 608 3040

3 1 599 2995
3 2 655 3275
3 3 642 3210
3 4 457 2285
4 1 417 2085

4 2 549 2745
4 3 591 2955
4 4 539 2695
4 5 408 2040
4 6 412 2060

5 1 647 3235
5 2 662 3310
5 3 659 3295
5 4 659 3295
5 5 721 3605
5 6 609 3045
5 7 693 3465



Power field calculation

The power forecast is the average power output for the entire lease sub-block for that 10-minute period in MW. All of the WRF grid points covered by the sub-block are included in the average. The observed power is the nearest NDBC buoy or tower (i.e. the Chesapeake Lighthouse (CHLVW) at 43 m height) scaled up to 90-m using log-law with z0=2E-4 m, similar to Dvorak, et al. (2010). All nine NDBC buoys and towers used are shown in Figure 1.

The power is calculated for each sub-block with 10Dx10D spacing of a REpower 5M (D=126 m) as follows:

Sub-block power [MW] = blockAreaKm2/(10*126 m*10*126 m/1E6 m^2*km^2)/1000 kW/MW * forecastKw * arrayLossFactor
  = blockAreaKm2 * 0.62988 MW/km^2 * forecastKw * arrayLossFactor


where forecastKw is the average power output from the REpower 5M 5.0 MW turbine in kW (calculated with the average 90m wind speed for the entire sub-block) and the arrayLossFactor is set at 90%, similar to Dvorak, et al., 2012.

Algorithm to create build-block power forecasts and validation

Power forecasts with 10-min time resolution were generated for each MAOWIT bulid-block every 24-hr using WRF-ARW on the UD Mills Cluster. WRF-ARW was initialized using the 12Z NAM forecast and started at 16Z during daylight saving time (DST) and 17Z for no DST. These times correspond to local noon (12:00 LST) for the US East Coast. Forecasts were started every 24-hr and run for 48-hr, creating power forecasts that overlapped by 24-hr. The NAM forecast was used to update the WRF-ARW boundary conditions every hour from 0-36 hr and every 3-hr after the 36-hr forecast.

Offshore meteorological observations are sparse and generally taken at the surface, making model validation at the turbine hub height of 90-m infeasible. A simplifying assumption was made to take the nearest offshore observation to the build block, scale that observation up to the turbine hub height using the log-law, and use this calculated wind speed to determine the "observed power" by running this wind speed through the REpower 5M power curve. Forecasting error is likely increased using this methodology due to the spatial offset of the forecast and in-situ observations. The validity of this assumption should be explored in future research.

The general algorithm to create the 10-min power forecast and "observed power" is as follows:

  1. Obtain all NDBC buoy/tower data for 2010 to ensure complete coverage - All available NDBC CWIND data (10-min wind observations) was loaded into a WinDB database. Having all available 2010 CWIND observations ensured that we used the closest data available for comparison.
  2. Load the WinDB database forecasted winds and observations at all MAOWIT build locations - Forecasted WRF-ARW 90-m winds and NDBC observations were loaded into a WinDB geospatial database for Jan, Apr, Jul, and Oct 2010. To increase the performance of the WinDB database, WRF and observational winds were only uploaded in the vicinity of MAOWIT build blocks and buoy/tower locations. The map of the data mask is available at this link, which shows the MAOWIT build blocks and WRF-ARW grid points.
  3. Create simplified build-block polygons to increase the performance of geospatial joins - The build blocks were simplified to approximately the resolution of the WRF runs to increase the speed of geospatial queries. To maintain consistency of the build-block sizes with Sheridan, et al. (2012), the build-block surface areas calculated in that study were maintained and used for wind farm capacity calculations (Tables 1 and 2), rather than using the surface area of the simplified geometries.
  4. Create static database tables of wind power forecasts and observed power for each wind farm - Using a geospatial join to determine which WRF-ARW wind speed grid points should be included in the power average for each block, the average power at each wind farm sub-block was calculate every 10 min in WinDB. An example average wind speed calculation can be viewed at this link, which shows the WRF-ARW points used for Sub-block 1.1.
  5. Create dynamic views (as opposed to tables) of 10-min block power and observed power - The static tables created in the previous step were combined via a "view", which allows a temporary view of the data in the combined tables. By dynamically/temporarily creating these views, the project can easily and quickly change attributes like turbine spacing and wind array loss (defined above). Table dumps of these views were used to create the CSV files at the beginning of this page.


  • Dvorak, M.J., C.L. Archer, and M.Z. Jacobson. California offshore wind energy potential. Renewable Energy (2010), doi:10.1016/j.renene.2009.11.022. 2009..
  • Dvorak, M.J., Corcoran, B.A., Ten Hoeve, J.E., McIntyre, N.G., and Jacobson, M.Z.: US East Coast offshore wind energy resources and their relationship to peak-time electricity demand, Wind Energy, in press. Available online.
  • Sheridan, B., Baker, S. D., Pearre, N. S., Firestone, J., & Kempton, W. (2012). Calculating the offshore wind power resource: Robust assessment methods applied to the U.S. Atlantic coast. Renewable Energy, 43, 224-233. doi: 10.1016/j.renene.2011.11.029

Copyright © 2012-2014 Sailor's Energy