Development of an Integrated Extreme Wind, Wave, Current, and Water Level Climatology to Support Standards-Based Design of Offshore Wind Projects

Technology Assessment and Research Project #672

FINAL REPORT

06 February 2014

TA&R Project #672 Extreme Metocean Climatology for Offshore Wind

Executive Summary

This report describes the methodology and results of a two-year study funded by the Technology Assessment and Research (TA&R) Program of the U.S. Bureau of Environmental Safety and Enforcement (BSEE). The primary goal of this study was to develop and apply methodologies for creating an extreme event climatology that characterizes standards-based design parameters for extreme winds, waves, currents, and water levels for the offshore Mid-Atlantic region at event return periods appropriate to the acceptable risk for safe operation and survival of the various different components of offshore wind projects, including the turbine, tower, foundation substructures, and accessory platforms.

The results presented herein are intended to assist BSEE regulators and Certified Verification Agents in their review of the Design Basis for offshore wind project plans in the Mid-Atlantic offshore Wind Energy Areas off New York, New Jersey, Delaware, Maryland, Virginia, and northeastern North Carolina. It also will be of interest to the designers of wind turbines and foundation substructures, and to the developers, financers, and insurers of any offshore wind project to be sited on the Mid-Atlantic Outer Continental Shelf.

The full report is divided into five chapters. Chapter 1 provides the standards-based context for selection of extreme event return periods, and the meteorological and oceanographic (metocean) parameters that the standards specify for various Design Load Cases (DLCs) and associated structural load modeling. This chapter describes the relationship between fundamental metocean parameters, as customarily produced by physical measurements and numerical models, and derived metocean parameters that the standards specify for each DLC, as summarized below.

The fundamental wind parameter is the 10-minute average wind speed at the meteorological “surface” elevation of 10 meters above sea level (U10). The fundamental wave parameter is the significant wave height (HS) for an assumed 3-hour sea state duration.

All metocean parameters specified by the standards for direct application in a given DLC are derived from one of the two fundamental metocean parameters defined above, typically by applying a multiplier. Thus, a “reference” 10-minute mean wind speed at turbine hub height (VREF) is estimated by deriving a U10 multiplier from the assumed vertical profile of wind speed. This then becomes the basis for estimating “extreme” or “reduced” 3-second gust speeds by applying a VREF multiplier, which is specified in the applicable standard. Likewise, various estimates for both individual waves and the sea state as a whole, including “extreme,” “severe” and “reduced,” are derived from HS multipliers, also specified in the applicable standard.

Chapter 2 describes the methodology our study used to estimate the fundamental metocean parameters for the two different types of extreme storm populations that occur in our study area: hurricanes (tropical cyclones) and nor’easters (extratropical cyclones). Section 2.1 describes the methodology and results for estimating the fundamental metocean parameters of nor’easters. Section 2.2 describes the methodology and results for estimating the fundamental metocean parameters of hurricanes. Section 2.3 examines the relationship between these two different storm populations and how this relationship affects the extreme probability distribution of fundamental wind and wave parameters throughout our study region.

Final Report 1 06 Feb 2014 TA&R Project #672 Extreme Metocean Climatology for Offshore Wind

For a given threshold U10 or HS, the number of hurricane events exceeding a given threshold is substantially less than the number of nor’easter events exceeding that same threshold over any given measurement or modeling period. For nor’easters, a 20- to 30-year historical sample of wind or wave data provides a sufficient number of events to accurately fit an extreme probability model to the high tail end of the sample distribution, such that the model can be reliably used to extrapolate design events having a 50- or 100-year return period.

Although the National Data Buoy Center (NDBC) has several long-lived offshore measurement stations in our study area with record lengths exceeding 20 years, these all contain gaps that have missed major nor’easter events. Therefore, our study evaluated two 20-year hindcast databases: the Wave Information Studies (WIS) database developed by the U.S. Army Corps of Engineers for design of shore and harbor protection measures, and the Wavewatch III database developed by the National Centers for Environmental Prediction (NCEP), which is the operational wave forecast system used by the National Weather Service.

By comparison with nor’easters, there are far fewer hurricane events in a 20- or 30-year sample, which introduces much more uncertainty in extrapolating the 50- or 100-year design event. Therefore, our study adopted the synthetic hurricane modeling approach used by the American Society of Civil Engineers for coastal building design. This enables the Monte Carlo simulation of thousands of synthetic storms, such that the high tail end of the sample distribution would be more reliably represented by this much larger number of storm events.

Chapter 3 describes the derivation of specific wind and wave design parameters from the fundamental wind and wave parameters estimated in Chapter 2. Measured wind and wave data from a variety of platforms are used to validate the derivation multipliers that are published in the standards. Where measurements depart from the standards-based multipliers, alternative multipliers are recommended. This section also describes how standards-based vertical profiles of wind speed (i.e., wind shear) compare with measured profiles as published in peer-reviewed literature. Finally, this section describes how wave breaking alters the probability distribution of individual wave heights in extreme sea states, and the effect this may have on various DLCs.

Chapter 4 describes the methodology and results for estimating extreme water levels, surface current speeds, and current profiles. These are governed primarily by the same fundamental wind and wave parameters estimated in Chapter 2, but also are influenced by astronomical tide. Although there remain large uncertainties in the characterization of wind-driven currents and underwater current profiles, the overturning moment contributions by wind loads on the wind turbine rotor and wave loads on the foundation substructure are so much greater that this uncertainty is likely to have only modest impact on the design of offshore wind facilities.

Six appendices are included with this report and can be downloaded as separate PDFs.

Final Report 2 06 Feb 2014 TA&R Project #672 Extreme Metocean Climatology for Offshore Wind

Chapter 1. Standards-Based Context

For limit state design and ultimate strength analysis, the IEC 61400-3 offshore wind turbine design standard specifies an extreme event return period of 50 years. For hurricane-prone areas such as the Mid-Atlantic region, the American Bureau of Shipping Guide for Building and Classing Bottom-founded Offshore Wind Turbine Installations (ABS-BOWTI) recommends a return period of 100 years. Both standards are otherwise identical in their specification of the Design Load Case (DLC) 6.x and 7.x series, which consider that the turbine has been shut down and is parked (idling) with power available from the utility grid to maintain or adjust turbine yaw (DLC 6.x) or idling with electrical fault (DLC 7.x).

A complete table of the IEC 61400-3 Design Load Cases is included as Appendix A.

Note that the 10-minute mean wind speed at hub height is referred to as the “reference” wind speed and must be vertically extrapolated from a modeled wind speed elevation, which is usually 10 m above sea level (ASL), or the elevation of a measured wind speed, which in our study area can range from 5 m ASL at 3-meter discus buoys operated by the National Data Buoy Center (NDBC), up to 45 m ASL on fixed platforms that are part of NDBC’s Coastal and Marine Automated Network (C-MAN). The IEC 61400-3 standard and the ABS BOWTI standard both specify a default shear profile described by a Power Law with a Power Law exponent of 0.11. Section 3.1 of this report evaluates the suitability of this default specification by comparing it with measured hurricane shear profiles published in the peer-reviewed literature.

Both standards specify DLCs for combined wind and wave loading by assuming that for a given design storm, the peak 3-second gust and the maximum individual wave height would not occur at the same instant at a given turbine location. They therefore specify two combined DLCs:  Extreme wind with reduced wave (e.g., DLC 6.1b): the peak 3-second gust is combined with a “reduced” individual wave that is lower than the maximum individual wave  Extreme wave with reduced wind (e.g., DLC 6.1c): the maximum individual wave height is combined with a “reduced” 3-second gust that is lower than the peak gust

Both standards also use the same multipliers to derive extreme and reduced design parameters:  Extreme wind speed = 1.4 times the 10-minute mean wind speed at hub height Reduced wind speed = 1.1 times the 10-minute mean wind speed at hub height  Extreme wave height = 1.86 times the 3-hour significant wave height Reduced wave height = 1.3 times the 3-hour significant wave height

Sections 3.2 and 3.3 of this report examine the suitability of the above multipliers in our study area by comparing with measured winds and waves during extreme Mid-Atlantic storms. These are followed by Section 3.4 which explores whether maximum individual wave heights and individual wave height probability distributions are limited by depth-induced wave breaking. Finally, Section 3.5 compares our modeled results with hourly measurements of winds and waves during the most extreme hurricanes and nor’easters in the Mid-Atlantic record, to address the question of whether or not Extreme-Reduced combinations are appropriately conservative.

Final Report 3 06 Feb 2014 TA&R Project #672 Extreme Metocean Climatology for Offshore Wind

Chapter 2. Fundamental Wind and Wave Design Parameters

Section 2.1 Nor’easter 50- and 100-Year Wind and Wave Parameters

This section describes how our study estimated the fundamental parameters of 10-minute mean wind speed at the sea surface (i.e. at an elevation of 10 m ASL) and significant wave height (average height of the highest one-third waves in a sea state) for nor’easters (i.e., extratropical cyclones) in our study region. This work was performed by Jeff Hanson and Mike Forte of the U.S. Army Corps of Engineers (USACE) Field Research Facility (FRF) at Duck, North Carolina. Their full report for this task is included as Appendix B and summarized in this section.

FRF researchers first performed a rigorous validation of wind and wave hindcasts, comparing peak mean wind speeds and peak significant wave heights measured during extratropical storm events at four long-term measurement stations within our study region, as mapped below.

Figure 1. Validation stations used to compare WIS and NCEP hindcasts of extratropical (nor’easter) winds and waves with measurements on a storm-by-storm basis.

Final Report 4 06 Feb 2014 TA&R Project #672 Extreme Metocean Climatology for Offshore Wind

The results of this hindcast comparison indicated that the Oceanweather kinematically adjusted winds used by WIS are more accurate than the NCEP Climate Forecast System Re-analysis (CFSR) winds used by Wavewatch III, which have an average positive bias of ~ 2 m/s. When CFSR winds match a given storm, the Wavewatch III hindcast is superior, as expected from a third-generation wave model compared with the WIS second-generation wave model. This advantage was more than offset by the generally poorer quality of CFSR winds, such that the WIS wave hindcast was found to be slightly superior to the NCEP Wavewatch III wave hindcast.

The FRF team then evaluated four long-term probability models: Generalized Extreme Value (GEV) applied to storm events identified as annual maxima and the Weibull Distribution, Generalized Pareto Distribution (GPD), and Empirical Simulation Technique (EST) applied to storm events identified by peak-over-threshold (POT) analysis, first removing tropical storms and hurricanes from the WIS hindcast database.

Example fits of the four long-term probability models are given in Figures 2 and 3, for two validation stations off Virginia, WIS #63197 located 12 n.mi. offshore near the Chesapeake Light Tower, and WIS #63255, located 64 n.mi. offshore, near the continental shelf edge. NCEP63197 11 WIS

Extratropical (POT) Events = 31 Extratropical (AMS) Events = 30 10 Generalized Extreme Value (GEV) Generalized Pareto Distribution (GPD) 9 Weibull )m( thgieH evaW thgieH )m( Empirical Simulation Technique (EST)

8 EST 90% Confidence Interval

7

6

5

Figure 2. Return period fits of four probability models to peak significant wave height events identified at WIS grid point #63197, adjacent to the Chesapeake Light Tower. Events were4 identified either as annual maxima series (AMS) or by peak over threshold (POT) analysis. The GEV was fit to AMS events, while the other three probability models were fit to POT events.

3 1 2 Final Report 10 5 1006 Feb 2014 Return Period (years) TA&R Project #672 Extreme Metocean Climatology for Offshore Wind

WIS

NCEP63255 11

10

Figure 3. Return period fits of four probability models to peak significant wave height events9 identified at WIS grid point #63255, near the continental shelf edge. Events were identified either as annual maxima series (AMS) or by peak over threshold (POT) analysis.

)m( thgieH evaW thgieH )m( The GEV was fit to AMS events, while the other three probability models were fit to POT events.

Note8 that at both locations, the EST probability distribution is most conservative at the target return periods of 50- and 100-years. The GEV and GPD are somewhat less conservative, and the Weibull distribution is least conservative, underestimating the 100-year significant wave height by 0.5 m near the Chesapeake Light Tower and by 1.0 m near the continental shelf edge. 7 The EST was selected because of its good fit to the data and relative lack of sensitivity to the threshold value used to identify storm events in the POT analysis. EST estimates of 100-year mean surface wind speed and 100-year significant wave height are mapped in Figures 4 and 5. The6 grid points with the two lowest and two highestExtratropical events are indicated (POT) on each Events map. Full = 31 numerical results for 50- and 100-year return periodsExtratropical are mapped in Appendix(AMS) C.Events = 30

Accurate5 hindcasting of storm peak mean wind speedsGeneralized and storm peak Extreme significant Valuewave heights (GEV) is critical to having high confidence in the EST estimates. As described in Section 5.2, a WIS re-analysis is now underway, and this will have a Generalizedone-hour rather than Pareto three-hour Distribution time step, (GPD) which will better resolve the storm peaks in meanWeibull wind speed and significant wave height. The WIS4 re-analysis also will use a third-generation wave model, greatly improving the accuracy of any future extreme event analysis of this new hindcastEmpirical database (seeSimulation Section 5.2 forTechnique details). (EST) EST 90% Confidence Interval 3 1 2 Final Report 10 6 1006 Feb 2014 Return Period (years) TA&R Project #672 Extreme Metocean Climatology for Offshore Wind

Figure 4. Map of extratropical cyclone (nor’easter) 100-year peak mean wind speeds at WIS grid points superimposed on map of six Mid-Atlantic offshore wind energy areas (WEAs).

Final Report 7 06 Feb 2014 TA&R Project #672 Extreme Metocean Climatology for Offshore Wind

Figure 5. Map of extratropical cyclone (nor’easter) 100-year peak significant wave heights at WIS grid points superimposed on map of six Mid-Atlantic offshore wind energy areas (WEAs).

Final Report 8 06 Feb 2014 TA&R Project #672 Extreme Metocean Climatology for Offshore Wind

Section 2.2 Hurricane 50- and 100-Year Wind and Wave Parameters

This section describes how our study applied synthetic hurricane modeling to estimate the fundamental parameters of 10-minute mean wind speed at the sea surface (i.e. at an elevation of 10 m ASL) and significant wave height (average height of the highest one-third waves) for hurricanes in our study region. This work was performed by Peter Vickery and Sudhan Banik of Applied Risk Associates in Raleigh, North Carolina.

To appreciate why synthetic hurricane modeling is needed to accurately estimate extreme event return period statistics at the uppermost tail end of the statistical distribution, it is instructive to view one year of significant wave height time series. This is shown for 1999 in Figure 6 at two of the WIS hindcast validation stations shown in Figure 1 for NDBC Station 44009 off Delaware, and NDBC Station 44014 located far offshore Virginia. The horizontal magenta dashed lines represent the significant wave height (Hs) threshold for the peak-over-threshold analysis conducted by the USACE.

The year 1999 is show because it was an active year for hurricanes, with large wave events produced by Hurricanes Dennis in August, Floyd in September, and Irene in October. Not surprisingly, hurricane waves are more prevalent in the extreme wave climate at the more southerly buoy located farther offshore. At both locations, however, extratropical storm waves are much more common throughout the year, with hurricane wave events being comparatively rare even during this active Mid-Atlantic hurricane season. This illustrates why synthetic hurricane modeling is required to reliably estimate the 50- and 100-year return periods for relatively rare tropical storm systems.

Final Report 9 06 Feb 2014

TA&R Project #672 Extreme Metocean Climatology for Offshore Wind

Figure 6. NDBC-measured vs WIS-hindcast Hs events in 1999 at Buoys 44009 and 44014.

Final Report 10 06 Feb 2014 TA&R Project #672 Extreme Metocean Climatology for Offshore Wind

ARA first completed historical hurricane reconstructions for 1985 Gloria, 1991 Bob, 1954 Carol, 1960 Donna, and the Great Atlantic Hurricane of 1944. This exercise had two purposes: A. to validate the synthetic hurricane model (SHM) as compared with measurements, where measurements existed; and B. to fill in measurement gaps so that a suitably long period (~100 years) historical record can be reconstructed, which will serve as a “reality check” on the wind speed and wave height return period curves produced from the SHM stochastic simulation of ~100,000 years.

Figures 7 and 8 present example comparisons of simulated and measured wind speeds, directions and central surface pressure reductions for two of the above hurricanes at two validation stations.

Figure 7. Comparison of measured observations (Obs) and modeled simulations (Sim) of mean wind speed, gust speed, wind direction and central pressure reduction at two validation stations during 1985 Hurricane Gloria: CHLV2 off Virginia Beach and Buoy 44009 off Delaware.

Final Report 11 06 Feb 2014 TA&R Project #672 Extreme Metocean Climatology for Offshore Wind

Figure 8. Comparison of measured observations (Obs) and modeled simulations (Sim) of mean wind speed, gust speed, wind direction and central pressure reduction at two validation stations during 1991 Hurricane Bob: CHLV2 off Virginia Beach and Buoy 44009 off Delaware.

Final Report 12 06 Feb 2014 TA&R Project #672 Extreme Metocean Climatology for Offshore Wind

ARA then tested its implementation of the SWAN (Simulating WAves Nearshore) model, by comparing modeled and measured waves at validation stations throughout our study area. The nested SWAN grid for Station 44009 off Delaware is shown in Figure 9, and the SHM-SWAN modeled winds and waves are compared with measurements at that station in Figure 10.

Figure 9. Nested grid used for wave generation by the SWAN model, as driven by SHM reconstructed wind fields and track for 1999 Hurricane Floyd.

Note that the SWAN model generates waves from the modeled hurricane traveling over still water. Thus the background wave field due to pre-existing winds and swell from other storms is not modeled, which is why in Figure 10, the observed significant wave heights before the storm exceed the simulated significant wave heights.

Also note the relatively good agreement between simulated and observed peak wave heights at Station 44009 despite the relatively poor agreement between simulated and observed peak winds. This is because most of the wave energy generated by a hurricane is propagated as swell that disperses ahead of the storm, with the direction of wave travel roughly aligned with the storm track. Local wind-driven seas only contribute a small amount of wave energy to the storm peak.

Final Report 13 06 Feb 2014 TA&R Project #672 Extreme Metocean Climatology for Offshore Wind

Figure 10. Comparison of simulated and observed mean wind speeds and significant wave heights at Buoy 44009 off Delaware for 1999 Hurricane Floyd.

Note that SWAN model generates waves from the modeled hurricane traveling over still water. Thus the background wave field due to pre-existing winds and swell from other storms is not modeled, which is why observed wave heights before the storm exceed modeled wave heights.

Final Report 14 06 Feb 2014 TA&R Project #672 Extreme Metocean Climatology for Offshore Wind

In order to use a synthetic hurricane model (SHM) for simulating storm wave conditions in the Mid-Atlantic offshore wind energy areas, the model needed to reproduce the along-track characteristics of hurricanes well offshore in the Atlantic Ocean, not just at landfall. This is because extreme wave conditions on the outer continental shelf are determined by hurricane characteristics even when the eye of the storm is up to a thousand kilometers distant. To date, SHM input parameters have been developed only for land-falling hurricanes, in order to support coastal building design standards for extreme winds (e.g., ASCE-7).

Development of Atlantic Ocean basin-wide SHM input parameters was carried out by ARA in the first year of this study, by fitting a cumulative probability distribution (CPD) for all tropical storms and hurricanes passing within a 250 km radius of the model grid points. CPDs were developed for track heading, forward translation speed, and storm central pressure. Appendix D details the methodology and results for fitting the CPDs of these input parameters for the ARA across the synthetic hurricane model domain used in this study.

Once the oceanic input parameter database was established, the SHM modeled approximately 100,000 tropical storms and hurricanes, the exact number depending on the grid point being evaluated. This number had to be greatly reduced in order to simulate waves. A first-pass subset for SWAN modeling was created by retaining only those storms that were of hurricane intensity north of 35°N and west of 70°W. From this subset, 150 hurricanes from each of four different Saffir-Simpson intensity categories (1, 2, 3, and 4) were chosen randomly and then weighted based on the proportion of each different category in the full set. To this number were added all Category 5 hurricanes, bringing the final subset number to approximately 650 storms.

Return-period curves were obtained by rank ordering the SHM wind or wave outputs and assigning Poisson arrival probabilities (and associated return periods) to the rank ordered-data. As a “reality check” SHM wind return period curves also were developed at four validation stations, simulating historical hurricanes for which wind measurements exist. Because offshore wind measurements are available only since the mid-1980s, the SHM also simulated significant historical hurricanes occurring before this time (such as 1964 Hurricane Donna, for example), in order to see where the historical storm population falls along the return period curves.

The largest numbers of historical measurements occur at two stations on automated light towers located off Virginia Beach (Chesapeake Light Tower, station CHLV2) and off Cape Hatteras (Diamond Shoals Light Tower, station DSLN7). The return period curves for 3-second gust at these two stations are given in Figures 11 and 12. To verify that the subset selected for wave modeling well represents the larger, full simulation, the return period curve for the subset is plotted as a thin blue line on top of the thicker magenta line that represents the full simulation.

Final Report 15 06 Feb 2014 TA&R Project #672 Extreme Metocean Climatology for Offshore Wind

Figure 11. CHLV2 hurricane return period curve for 3-second gust wind speed at 10m ASL.

Looking first at the gust return period curve for CHLV2, there are several features to note:

 At return periods greater than ten years, there are only 11 storms in the historical hurricane population. Any return period trend line extrapolated from this small sample would be well below the trend line of the synthetic hurricane subset used for wave modeling, which has 173 storms with return periods greater than ten years.

 The SHM return period curve for CHLV2 estimates the 100-year 3-second gust speed at 10 m above sea level to be 110 mph (49 m/s). Using our study’s recommended hurricane wind shear profile (see Section 3.1 for details), this translates to 133 mph (59 m/s) at a turbine rotor hub height of 100 m above sea level. By comparison, the IEC 61400-3 gust wind speed standard for a type-certified Class I turbine is 157 mph (70 m/s) at hub height. Thus there is a 1.18 safety margin between our estimated 100-year gust wind speed at CHLV2 and the upper bound for which an IEC Class I turbine is designed.

 Much of this safety margin arguably would be consumed by uncertainties in the SHM gust wind speed simulation, which is reflected by the wide degree of scatter in agreement between measured and modeled gust wind speeds on either side of the historical storm trend line. This scatter is explored later in this section.

 Based on the return period of storm intensity (in terms of central pressure reduction, which is considered accurate in the historical hurricane database from 1900 onward), 1960 Hurricane Donna was a 100-year event at CHLV2. There are no well documented offshore wind speed measurements off Virginia during this storm.

Final Report 16 06 Feb 2014 TA&R Project #672 Extreme Metocean Climatology for Offshore Wind

Figure 12. DSLN7 hurricane return period curve for 3-second gust wind speed at 10m ASL.

Now turning to the gust return period curve for DSLN7, the following features are noteworthy:

 Somewhat surprising for this more southerly station is that there are only 11 historical hurricanes with return periods greater than ten years, which is the same number as at CHLV2. The synthetic hurricane subset used for wave modeling is more in line with expectations, having 273 storms with return periods greater than ten years.

 Although the number of historical storms is small, their trend is closely aligned with the SHM trend line. Moreover, the agreement between modeled and measured gust speeds for historical hurricanes is much better at DSLN7 than at CHLV2.

 The SHM return period curve for DSLN7 estimates the 100-year 3-second gust speed at 10 m above sea level to be 132 mph (59 m/s). Using our study’s recommended hurricane wind shear profile (see Section 3.1 for details), this translates to 159 mph (71 m/s) at a turbine rotor hub height of 100 m above sea level, which just exceeds the IEC 61400-3 gust wind speed standard for a type-certified Class I turbine.

As already mentioned, there is quite a bit of scatter in the agreement between SHM modeled gust wind speeds and measured gust wind speeds at CHLV2. This linear regression relationship is plotted in Figure 13.

Final Report 17 06 Feb 2014 TA&R Project #672 Extreme Metocean Climatology for Offshore Wind

Figure 13. CHLV2 modeled vs measured 3-second gust wind speeds at 10m ASL.

By comparison, the agreement between modeled and measured gust wind speeds is much better at DSLN7, as shown in Figure 14. This may be due to the fact that peak gusts at Diamond Shoals Light Tower are experienced before the hurricane core has been affected by traveling over the Outer Banks of North Carolina. Once a hurricane is over land and very shallow water, the SHM is more erratic in how well it simulates the wind field, accounting for the much wider scatter in SHM agreement with measurements at Chesapeake Light Tower, where the wind field already has been substantially affected by interactions with land and very shallow water.

The one major hurricane that is well modeled at CHLV2 is 2003 Hurricane Isabel, which came more directly from the southeast, such that its eye remained over shelf waters right up to the time that its peak winds reached CHLV2. This supports the above hypothesis.

Final Report 18 06 Feb 2014 TA&R Project #672 Extreme Metocean Climatology for Offshore Wind

Figure 14. DSLN7 modeled vs measured 3-second gust wind speeds at 10m ASL.

Combined wind and wave loading is required to assess IEC 61400-3 Design Load Cases (DLCs) 6.1 and 6.2. Due to the importance of these DLCs, ARA archived the significant wave height at the time of peak wind speed for each simulated storm at a given grid point and likewise archived the wind speed at the time of peak significant wave height for that same storm. At CHLV2, for example, Figure 15 shows the return period curve for peak mean wind speed with a scatter plot of the associated significant wave heights for all storms used to build the curve. Likewise, Figure 16 shows the return period curve for peak significant wave height with a scatter plot of the associated gust wind speed for all storms used to build the curve.

Final Report 19 06 Feb 2014 TA&R Project #672 Extreme Metocean Climatology for Offshore Wind

Figure 15. CHLV2 hurricane return period curve for peak mean wind speed (left axis), co-plotted with the associated significant wave height (right axis) at the time of peak wind.

Figure 16. CHLV2 hurricane return period curve for peak mean wind speed (left axis), co-plotted with the associated significant wave height (right axis) at the time of peak wind.

Final Report 20 06 Feb 2014 TA&R Project #672 Extreme Metocean Climatology for Offshore Wind

Note that at return periods of less than ten years, it is quite possible to have significant wave heights of 4 to 6 m at mean surface wind speeds of 20 m/s, which would translate to 100 m hub height wind speeds of 24 m/s. This is below the 25 m/s cut-out speed of wind turbines, meaning that unless turbines are manually taken off line, they could be operating while experiencing fairly extreme wave loads, quite possibly including breaking waves (see Section 3.4 for exploration of breaking waves). Based on wind and wave directions in advance of a typical hurricane track toward the Mid-Atlantic from the Bahamas, there is a high probability that large waves could arrive “edge-on” to the plane of the turbine rotor, greatly reducing the aerodynamic damping of wave-induced response of the turbine-tower-substructure while the turbine is operating. This is illustrated in Figure 17 for 2011 Hurricane Irene at NDBC Station 44014 off Virginia and for other recent hurricanes in the full report. This could be an important fatigue loading case to be included in the metocean design basis for Mid-Atlantic offshore wind projects.

TURBINE CUT OUT SPEED -- based on 5m anemometer mast on buoy and 100 m rotor hub height

LARGE WIND – WAVE MISALIGNMENT

Figure 17. Time series of 2011 Hurricane Irene wind speeds and atmospheric pressure (top), significant wave height (middle) and wind & wave directions (bottom) at NDBC Station 44014.

Final Report 21 06 Feb 2014 TA&R Project #672 Extreme Metocean Climatology for Offshore Wind

Section 2.3 Wind and Wave Parameters for Combined Storm Population

The Mid-Atlantic outer continental shelf is subject to two different storm populations, which are different in their genesis and behavior: hurricanes, which can occur in June through November, but are most common in August through October, and nor’easters, which typically are much larger, slower moving extratropical systems that can occur at any time of the year, but are most common in October through May.

For the design of Mid-Atlantic offshore wind facilities subject to both of these storm types, the metocean basis of design should use a “mixed” rather than a “commingled” analysis of peak mean wind speeds and peak significant wave heights. Commingled estimates include all extreme values exceeding the threshold being considered, regardless of storm type. Mixed estimates are derived as the product of the underlying independent probability distributions of each storm type, which is more appropriate from the perspective of both statistics (sampling and probability theory) and storm physics (causal mechanism and general characteristics of storm size and forward speed). When each type of storm has a similar event magnitude at a given return period, a commingled analysis yields non-conservative results when compared with a mixed analysis.

As shown in Figure a, below, if the event magnitude for one storm type is much greater than that for the other storm type, then simply using the higher of the two may be appropriate. For example, Figure 18 shows wind speed return period estimates at La Guardia Airport. At the longest return period of 500 years, thunderstorm wind speeds are nearly 10 knots greater than non-thunderstorm (e.g., synoptic scale) winds, and the 500-year thunderstorm wind speed is within a fraction of a knot of the 500-year mixed wind speed.

Figure 18. Wind speed return periods for LaGuardia Airport obtained using a statistically “mixed” distribution (M), thunderstorm wind speeds alone (T), non-thunderstorm wind speeds alone (NT), and a commingled population of both storm types (C).

Final Report 22 06 Feb 2014 TA&R Project #672 Extreme Metocean Climatology for Offshore Wind

To compare the peak wind and wave conditions of the two major types of Mid-Atlantic storms, 21 WIS grid points were subject to both the USACE extratropical storm hindcast analysis and the ARA synthetic hurricane model analysis. These 21 points are mapped in Figure 19, and it’s important to note that all the WIS grid points occur near the western (shoreward) border of the Mid-Atlantic offshore wind energy areas (WEAs) in mean lower low water (MLLW) depths of 18 to 25 m. This is useful for comparing the relative wind and wave intensities of these two storm populations, but for properly developing the metocean design basis for these different WEAs, more grid points are needed along their eastern (seaward) borders – this recommendation is addressed in Section 5.4, at the end of this summary.

To see how these two different storm systems compare in cross-shelf trends of winds and waves, we analyzed five more WIS grid points in an east-west (onshore-offshore) transect just below the northern border of the North Carolina WEA off Kitty Hawk. There are a total of six grid points in this transect; its western end point is in 22 m MLLW depth, and its eastern end point in 38 m MLLW depth (see map in Figure 19).

The north-south trend comparing hurricanes and extratropical storms for mean surface wind speed at 100-year return periods is presented in Figure 20. This comparison indicates that the surface mean wind speeds of hurricanes are at least 5 m/s faster than those of nor’easters having the same return period, even as far north as New York. This difference increases farther south, exceeding ~10 m/s off Virginia and approaching ~15 m/s off Kitty Hawk, NC.

Note that in all cases, the 10-minute mean surface wind speed at storm peak is less than 40 m/s. Using our study’s recommended hurricane wind shear profile (see Section 3.1 for details), the x- axis maximum of 40 m/s at 10 m above sea level translates to 48 m/s at a height of 100 m above sea level. This is below the reference 10-minute mean wind speed upper limit of 50 m/s specified at turbine rotor hub height for Class I type-certified turbines by IEC 61400-3. Given the uncertainty in hurricane modeled wind speeds shown in Figure 9 for offshore Virginia, it will be important for future studies to particularly focus on this WEA to determine whether or not Class I turbines are suitably designed for this hurricane wind climate, or whether a Type S certification would be required.

The west-east (onshore-offshore) trend comparing hurricanes and extratropical storms for mean surface wind speed at 100-year return periods is presented in Figure 21, for the cross-shelf transect in the Kitty Hawk, NC WEA. As with the north-south trend, hurricane wind speeds dominate the extreme wind climate, being in the range of 12 to 14 m/s faster than nor’easter winds for 100-year events and 10 to 11 m/s faster for 50-year events.

Figure 21 also shows that for both storm types there is relatively little change in peak wind speed farther offshore. For nor’easters, this is a consequence of the very large size and relatively slow motion of extratropical storm systems, such that their maximum winds are experienced over large regions of the continental shelf. Although hurricanes are much smaller and faster moving storms, the large number of simulated storms afforded by the synthetic modeling approach indicates that with enough different storm tracks in the simulated database, high hurricane winds can be experienced over equally large regions of the continental shelf.

Final Report 23 06 Feb 2014 TA&R Project #672 Extreme Metocean Climatology for Offshore Wind

Figure 19. Map of WIS grid points subject to comparative analysis of hurricanes and nor’easters superimposed on map of six Mid-Atlantic offshore wind energy areas (WEAs).

Final Report 24 06 Feb 2014 TA&R Project #672 Extreme Metocean Climatology for Offshore Wind

Figure 20. North-south trends comparing the mean surface wind speeds of hurricanes and nor’easters at 100-year return period for six Mid-Atlantic WEAs, based on the grid points mapped in Figure 19. Note that 40 m/s at 10m ASL translates to 48 m/s at 100 m ASL.

Final Report 25 06 Feb 2014 TA&R Project #672 Extreme Metocean Climatology for Offshore Wind

Figure 21. East-west trends comparing the mean surface wind speeds of hurricanes and nor’easters at 50- and 100-year return periods for six grid points in a cross-shelf transect off Kitty Hawk, NC, as mapped in Figure 19.

The north-south trend comparing hurricanes and extratropical storms for significant wave height (Hs) at 100-year return periods is presented in Figure 22. Unlike the wind speed graph, the only consistent trend evident in this graph is off Virginia and Kitty Hawk, NC, where peak Hs values for hurricanes are 1 to 2 m higher than their extratropical counterparts. Farther north, hurricane and nor’easter Hs values swap supremacy back and forth but typically are within 0.5 m of one another, except off southern New Jersey, where hurricane Hs values are up to 1 m higher than their extratropical counterparts.

In the four southernmost New Jersey grid points, hurricane sea states are higher than nor’easter sea states with an Hs difference of about 1 m. This may be due to the fact that these four grid points are farther offshore and thus lie more directly in the path of wave groups emanating from hurricanes farther to the south. The highest hurricane sea states among the grid points mapped are off Virginia, where peak 100-year Hs values are 8.4 m.

The west-east (onshore-offshore) trend comparing hurricane and extratropical storm sea states at 100-year return periods is presented in Figure 23, for the cross-shelf transect in the Kitty Hawk, NC WEA. Unlike the west-east trend in wind speeds, which remains relatively constant, there is a dramatic increase in significant wave heights with increasing distance offshore. This trend is more pronounced for hurricanes than for nor’easters, with hurricane Hs values being about 1 m higher than nor’easter Hs values in 22 m depth nearer shore, with this difference growing to ~3 m higher in 38 m depth on the far offshore outer continental shelf.

Final Report 26 06 Feb 2014 TA&R Project #672 Extreme Metocean Climatology for Offshore Wind

Figure 22. North-south trends comparing the significant wave height (Hs) values of hurricanes and nor’easters at 100-year return period for six Mid-Atlantic WEAs, based on the grid points mapped in Figure 15. The highest 100-year hurricane Hs is 8.4 m off Virginia and the highest 100-year nor’easter Hs is 8.5 m off New Jersey.

Final Report 27 06 Feb 2014 TA&R Project #672 Extreme Metocean Climatology for Offshore Wind

Figure 23. East-west trends comparing the significant wave heights of hurricanes and nor’easters at 50- and 100-year return periods for six grid points in a cross-shelf transect off Kitty Hawk, NC, as mapped in Figure 15.

Our extensive Chapter 2 results for the two fundamental metocean design parameters of mean surface wind speed and significant wave height can be distilled into two main findings.

First, for the Virginia and Kitty Hawk WEAs, simply applying the much higher hurricane wind speeds and significant wave heights will yield an appropriately conservative metocean design basis, and a mixed analysis that includes nor’easters would not be needed for these two WEAs.

Second, for offshore WEAs farther north, between Delaware and New York, the use of hurricane winds is also likely to yield an appropriately conservative result, but only for extreme winds. By comparison, however, the peak significant wave height (Hs) values produced by nor’easters in this sub-region are of comparable magnitude to the Hs values produced by hurricanes having the same return period, and simply using the higher of the two (as can be done for wind speeds) will yield a non-conservative design basis for wave loading.

Appendix E tabulates the numerical wind and wave results for hurricanes and nor’easters at both 50- and 100-year return periods for all 26 WIS grid points in the Mid-Atlantic WEAs, as mapped in Figure 19. This can provide preliminary guidance until more robust studies become available.

Appendix F provides a listing and maps of additional grid points that have been submitted to the USACE and NCEP for new30-year wind and wave hindcasts that both groups are now doing. These will provide data along the outer edges of the WEAs, where grid points are needed.

Final Report 28 06 Feb 2014 TA&R Project #672 Extreme Metocean Climatology for Offshore Wind

Chapter 3. Derived Metocean Design Quantities

Section 3.1 Wind Shear Profile and Hub Height Wind Speeds

The 10-minute mean surface wind speeds resulting from the nor’easter hindcasts and synthetic hurricane modeling results summarized to this point are for an elevation of 10 m above sea level (ASL). By comparison, the IEC 61400-3 and ABS BOWTI standards specify a 10-minute mean “reference” wind speed at the center of the turbine rotor, which is referred to as the hub height elevation. To translate mean wind speeds at 10 m ASL to hub height, both standards specify a default shear profile described by a Power Law with a Power Law exponent of 0.11.

Given the dominance of hurricane winds over nor’easter winds, the extreme wind profiles specified for the metocean design basis of Mid-Atlantic offshore wind energy projects must be based on measured values in hurricanes, rather than the above standards, which are derived from European experience in the North Sea.

The most recent and complete summary of hurricane wind profiles in the coastal and marine boundary layers is a 2012 Wind and Structures paper (citation in full report) by Ian Giammanco and John Schroeder of Texas Tech University, and Mark Powell of the NOAA/AOML Hurricane Research Division. Composite over-water wind profiles were generated from 1,080 individual GPS sonde profiles consisting of nearly 430,000 observations.

Giammanco, et al, 2012 developed Logarithmic and Power Law profile models for each wind speed group in the marine boundary layer (MBL). Least-square fits were performed on the composite profiles in order to test the validity of both models. Least-squares regressions were applied to a surface layer from 20 to 160 m ASL, which represents the layer depth in which the wind direction was typically constant. Table 1 summarizes lists the best fit Power Law exponent they estimated for each MBL wind speed group.

Table 1. Power Law Least Squares Fits for 20-160 m Surface Layer

Final Report 29 06 Feb 2014 TA&R Project #672 Extreme Metocean Climatology for Offshore Wind

In order to correctly validate hindcast or synthetic modeled hurricane wind speeds, which are “surface values” at 10 m above sea level (U10), the modeled values must be translated to the measurement height of the validating station. Table 2 lists the multipliers that would be used based on the average Power Law coefficient of 0.081 shown in Table 1.

Table 2. Wind Speed Multipliers to Apply Hurricane Power Law Profile

Finally, in order to translate IEC 61400-3 reference wind speeds at hub height to equivalent surface values, Table 3 is provided for a 100 m hub height.

Table 3. Hurricane Surface Wind Speed Values Corresponding to Standards

Note that the above table uses the gust factor assumed by the IEC 61400-3 standard, which multiplies the 10-minute mean wind speed by 1.4 to obtain an estimate of the 3-second gust wind speed. The next section evaluates the suitability of this gust factor and evaluates factors to be used for other averaging periods, such as the 5-second gusts and 2-minute means measured by NDBC stations in their Coastal-Marine Automated Network (C-MAN) of fixed platforms, or the 1-minute sustained wind speed typically reported by NOAA’s National Hurricane Center.

Final Report 30 06 Feb 2014 TA&R Project #672 Extreme Metocean Climatology for Offshore Wind

Section 3.2 Extreme and Reduced Wind Speeds

Both the IEC 61400-3 and the ABS BOWTI standards specify Design Load Cases (DLCs) for combined wind and wave loading by assuming that for a given design storm, the peak 3-second gust and the maximum individual wave height would not occur at the same instant at a given turbine location. They therefore specify two combined DLCs:  Extreme wind with reduced wave (e.g., DLC 6.1b): the peak 3-second gust is combined with a “reduced” individual wave that is lower than the maximum individual wave  Extreme wave with reduced wind (e.g., DLC 6.1c): the maximum individual wave height is combined with a “reduced” 3-second gust that is lower than the peak gust

Both standards also use the same multipliers to derive extreme and reduced design wind speeds from the 10-minute mean “reference” wind speed:  Extreme wind speed = 1.4 times the reference wind speed at hub height Reduced wind speed = 1.1 times the reference wind speed at hub height

To evaluate the suitability of the above multipliers, the first step is to adjust measured mean wind speeds and measured gust wind speeds to have the same averaging periods as the mean and gust in the standards. The second step is to compare the adjusted measurements with the standards.

For the first step, two publications were consulted to determine which factors to use for adjusting between different wind speed averaging periods and different gust speed averaging periods: Guidelines for Converting between Various Wind Averaging Periods in Tropical Cyclone Conditions, published in August 2010 by the World Meteorological Organization (WMO), and Recommended Practice for Constructing Fixed Offshore Platforms—Working Stress Design, 21st Edition, Supplement 3, published in October 2007 by the American Petroleum Institute (API RP 2A-WSD). Figure 24 compares the ratio formulations in these two publications.

As shown in this figure, API RP 2A-WSD is more conservative (estimates higher means at lesser averaging periods), and for averaging periods of one minute or longer, the API recommended formulation has greater ratios. Since API RP 2A-WSD is cited by the ABS BOWTI guide and other offshore wind standards, the API formulation was applied in our evaluation.

Buoy-based wind measurements (such as at our NDBC validation stations 44009 and 44014) were not used due to anemometer inflow accelerations caused by buoy motions in extreme storm sea states as well as intermittent shadowing of the buoy’s 5 m mast when in the troughs of comparably high or higher waves. The anemometer masts on fixed light tower C-MAN stations are considered to be much more representative of the offshore wind gust environment in extreme hurricanes and nor’easters.

The C-MAN wind data acquisition payload logs a 2-minute mean wind speed and a 5-second gust. The API formulation in Figure 24 yielded a 0.936 multiplier to translate the measured 2­ minute mean to the standards-specified 10-minute mean reference wind speed at hub height. The API formulation also yielded a 1.018 multiplier to adjust the measured 5-second gust to the standards-specified 3-second gust for the extreme and reduced DLC wind speeds at hub height.

Final Report 31 06 Feb 2014 TA&R Project #672 Extreme Metocean Climatology for Offshore Wind

Figure 24. Ratios of mean wind speed at given averaging period to hourly mean wind speed.

Once adjusted as described above, the measured gust factors were calculated for all C-MAN 2-minute mean wind speed measurements of 20 m/s or more. These are plotted for stations CHLV2 and DLSN7 in Figures 25 and 26, respectively.

Figure 25. Measured gust factors at Chesapeake Light Tower off Virginia.

Final Report 32 06 Feb 2014 TA&R Project #672 Extreme Metocean Climatology for Offshore Wind

Figure 26. Measured gust factors at Diamond Shoals Light Tower off North Carolina.

Out of 800 extreme wind measurements at Chesapeake Light Tower (C-MAN station CHLV2), only two had gust factors greater than the IEC 61400-3 standard of 1.4, while Diamond Shoals Light Tower (C-MAN station DSLN7) had just 5 out of 1,559 extreme wind measurements with greater gust factors. Thus 1.4 appears to be an appropriately conservative multiplier to translate the IEC 61400-3 reference mean wind speed to the DLC “extreme wind speed.”

As previously noted, the offshore wind standard DLCs also require a “reduced wind speed,” which is the 3-second gust value to be used with the “extreme wave height.” The reference wind speed multiplier to be used in this case is 1.1. As shown in Figures 21 and 22, however, this is significantly lower than the most common gust factor in extreme offshore wind speed measurements, which is 1.2. For Mid-Atlantic offshore wind design purposes, it therefore is recommended that the IEC 61400-3 reference mean wind speed be multiplied by 1.2 to obtain the DLC “reduced wind speed.”

Section 3.3 Extreme and Reduced Wave Heights

Both the IEC 61400-3 and the ABS BOWTI standards use the following multipliers to derive extreme and reduced design wave heights for combined wind-wave DLCs:  Extreme wave height = 1.86 times the 3-hour significant wave height Reduced wave height = 1.3 times the 3-hour significant wave height

To evaluate the suitability of the above multipliers, we surveyed the Atlantic Ocean data records posted by the Coastal Data Information Program (CDIP) of Scripps Institution of Oceanography, which archive the daily maximum crest-to-trough individual wave height. This maximum height was then divided by the significant wave height archived for the same measurement period.

Final Report 33 06 Feb 2014 TA&R Project #672 Extreme Metocean Climatology for Offshore Wind

The results of our maximum-to-significant wave height ratio analysis are shown in Figure 27. Out of 60 total extreme wave measurements, 18 exceeded the standards-specified ratio of 1.86, and five measurements had a ratio in the range of 2.1 to 2.2. Therefore, we recommend that 2.2 be used as a more appropriately conservative significant wave height multiplier to estimate the DLC “extreme wave height.”

Figure 27. Measured ratio of maximum to significant wave height in Mid-Atlantic storms.

Final Report 34 06 Feb 2014 TA&R Project #672 Extreme Metocean Climatology for Offshore Wind

Because the CDIP only archives the maximum individual wave height in a day, many more measurements would be available for ratio analysis if the maximum individual wave height could be archived for each significant wave height measurement. NDBC buoys do not archive any maximum wave height information, and adding an archive of the daily maximum individual wave height would be a useful improvement, and even better would be to archive the maximum individual wave height of each sea state measurement.

With such a small data set (only 60 measurements), it is difficult to evaluate the suitability of using 1.3 as a significant wave height multiplier to estimate the DLC “reduced wave height” to be combined with extreme wind gusts. Based on our limited sample, however, 1.3 appears to be much lower than the band between 1.6 and 1.8, which appears to be where most of the ratios of maximum-to-significant wave height occur in Figure 23. Our preliminary recommendation based on this admittedly limited data set is that 1.7 be used as a more appropriately conservative significant wave height multiplier to estimate the DLC “reduced wave height.”

This section concludes with an analysis of whether it is reasonable to combine the peak mean wind speeds plotted in Figures 16 and 17with the significant wave heights plotted in Figures EC­ 18 and 19 as a departure point for estimating the DLC 6.1 extreme-reduced combinations. The IEC 61400-3 and ABS BOWTI standards emerged from European offshore wind experience, where the extreme metocean events are large extratropical storms, where peak wind and peak wave conditions often occur within the same measurement record; this also is true for nor’easters in the Mid-Atlantic region.

The question is whether the 50- or 100-year wind and wave events can be considered as occurring at the same time in a single storm. Historically, some of the most intense hurricanes have a small radius to maximum wind, generating extreme wind speeds but not very high waves due to the limited fetch over which the winds blow (e.g.,1992 Hurricane Andrew), while less intense storms, with relatively low winds, blowing over a very large fetch, can produce huge waves (e.g., 2012 Hurricane Sandy).

The SHM simulations by ARA produced the full spectrum of synthetic storms, ranging from small-radius wind-makers to large-radius wave makers, including some hurricanes that produced both extreme winds and extreme waves. Our analysis wanted to determine where historic hurricanes fall on this spectrum.

First we plotted the time-series of combined wind and wave measurements for historical severe storms to identify the worst-case combination. We then applied the multipliers recommended above and superimposed those results on a joint wind-wave plot of synthetic hurricane modeling results for the same measurement station.

Figure 28 shows the time series for the two most extreme hurricanes where simultaneous wind and wave measurements exist at the Chesapeake Light Tower (CHLV2), namely 1985 Gloria and 2003 Isabel. The dotted circles outline the times during these storms when peak wind and wave conditions occurred simultaneously. These peak values were then adjusted using the multipliers recommended above to estimate extreme and reduced winds and waves for these historic storms, and compare them with the SHM values for both 50-year and 100-year return periods. The results are plotted in Figure 29 and suggest that the SHM peak combinations are realistic.

Final Report 35 06 Feb 2014 TA&R Project #672 Extreme Metocean Climatology for Offshore Wind

Figure 28. Time series of simultaneous wind and wave measurements at CHLV2 for 1985 Hurricane Gloria and 2003 Hurricane Isabel. The 10-minute mean wind speed at anemometer height has been translated to an assumed hub height of 100 m using a Power Law with a 0.081 exponent, as recommended in Section 3.1. The significant wave height has been translated from the 20-minute NDBC measurement period to a 3-hour significant wave height, as specified in the standards. Simultaneous peak wind and wave conditions are indicated by dotted circles.

Final Report 36 06 Feb 2014 TA&R Project #672 Extreme Metocean Climatology for Offshore Wind

Figure 29. DLC extreme and reduced wind-wave combinations at CHLV2 for 1985 Hurricane Gloria and 2003 Hurricane Isabel, with 50- and 100-year synthetic hurricane results. The hub height 10-minute mean wind speed was multiplied by 1.4 to estimate the extreme gust, and by 1.2 to estimate the reduced gust. The significant wave height was multiplied by 2.2 to estimate the extreme wave height, and by 1.7 to estimate the reduced wave height.

Final Report 37 06 Feb 2014 TA&R Project #672 Extreme Metocean Climatology for Offshore Wind

Another question is this: when the peak mean wind speed and peak significant wave height coincide in a hurricane, can the extreme gust within that peak mean wind environment and the extreme individual wave within that peak sea state occur at the same place at the same time?

While this appears to be true for extratropical storms such as nor’easters, there are no continuous measurements of moving 3-second gust averages and individual crest-to-trough wave heights to know if such a combination can occur in hurricanes and with what probability. This is an urgent measurement need, requiring a data acquisition strategy that does not discard gust measurements or individual wave measurements until both the wind time series and the sea surface elevation time series records can be synchronized and then analyzed to fully characterize the gust-wave combinations to see if the extreme-reduced DLC model holds true for hurricanes.

Section 3.4 Breaking Waves

A major concern for all Mid-Atlantic offshore wind energy areas is the occurrence of breaking waves, even in deep water under relatively modest storm conditions. For example, a video camera at the FINO-1 measurement platform in the North Sea captured the occurrence of what appears to be a plunging breaker impacting the platform substructure during an extratropical storm on 04 October 2009. The measured significant wave height at the time of the video was 5.2 to 5.3 m, and the maximum individual wave height measured during this same time series was just over 10 m high. The water depth at the FINO-1 platform is 28 m, giving a maximum wave height to water depth ratio of ~0.36. This is less than half the 0.78 ratio at which wave breaking is specified in Annex C of the IEC 61400-3 offshore wind standard.

In such water depths, it is no surprise when spilling breakers (“whitecapping”) occur for steep component wave trains whose periods are less than the spectral peak wave period, such that only longer-period wave trains can continue to absorb wind energy without over-steepening and breaking. When a sea state is saturated across all frequencies in the wave spectrum, it is said to be “fully developed,” as is often the case in extreme storms. When fully developed, the entire spectrum is in equilibrium with the local wind speed, such that the total energy input from the wind equals the total energy dissipated by breaking. What is surprising in these depths, however, is the occurrence of plunging breakers, which are thought to occur only in shallower water.

Hurricane sea states are known to contain steep wind-driven seas in the front, right quadrant of the storm, and as the hurricane’s forward speed accelerates off the Mid-Atlantic coast, it can overtake its forerunner swell. As shown in Figure 30, these steeper wind driven seas will cross the longer swell at wide angles, and individual wave superposition is almost certain to result in a greater probability of wave breaking than would occur in slow-moving nor’easters.

Dissipative spilling breakers generate wave loads that are quite similar to those of non-breaking waves. Plunging breakers, by comparison, generate impulsive impact loads as the near-vertical face of the toppling wave slams against any surface-piercing structure. Identifying breaking waves and mapping their sea state area coverage (which is proportional to their probability of striking a structure at any given point in the mapped area) will be critical to understanding the contribution of these high-loading events to the fatigue-history of Mid-Atlantic offshore wind turbine substructures and the tower-turbine they support.

Final Report 38 06 Feb 2014 TA&R Project #672 Extreme Metocean Climatology for Offshore Wind

Figure 30. Map characterizing the wave field in a hypothetical Northern Hemisphere hurricane moving northward. Source: Holthuijsen, L. H., M. D. Powell, and J. D. Pietrzak, 2012. Wind and waves in extreme hurricanes. Journal of Geophysical Research. 117, C09003.

In the above figure, blue hurricane symbol shows eye location associated with local wind-driven seas, whose crest alignment is indicated by curved, blue lines. The hurricane wind field at this time is indicated by dark grey bars, whose length is proportional to wind speed.

The red hurricane symbol shows an earlier eye location to the south. Curved, red, dashed lines indicate swell generated when the eye was at that southern location, initially outrunning the eye as the swell begins dispersing away from the storm. Then, as the hurricane’s forward speed accelerates, the hurricane overtakes these young, dispersing swell. Superposition of local wind sea crests crossing long swell crests at wide angles in front of the hurricane may create favorable conditions for plunging breakers to form, but this remains to be confirmed.

Final Report 39 06 Feb 2014 TA&R Project #672 Extreme Metocean Climatology for Offshore Wind

Section 3.5 Maximum Wave Crest Elevation

Another derived metocean parameter is the maximum wave crest elevation (Crmax), which must be superimposed on the extreme still-water level (see Section 4.1) in order to ensure that structures not designed to withstand hydrodynamic loading, notably the wind turbine blades, remain clear of the water. IEC 61400-3 specifies the minimum required clearance, referred to as the air gap, as 20% of the 50-year significant wave height, or one meter, whichever is greater. The value of Crmax is added to the extreme still water level, and the specified air gap is added on top of that to arrive at the required structure elevation relative to the chart datum.

In order to estimate Crmax, a multiplier is applied to the significant wave height at the required return period. Preliminary guidance for this multiplier is from the Cooperative Research on Extreme Seas and their impacT Joint Industry Project (CresT JIP), a summary of which has been published by BSEE under TA&R Project # 605 (Public Summary CresT JIP, October 2010).

Guidance from the CresT JIP is given in Figure 31, which plots the probability of different multipliers for the ratio of maximum individual wave height (Hmax) to significant wave height (Hs) and for the ratio of Crmax to Hs. The right graph of this figure shows that the widely accepted Forristall distribution estimates that our recommended Hmax to Hs value of 2.2 has a probability of exceedance of 10-6 in an extreme storm sea state – one in a million waves. The left figure shows that the corresponding ratio of Crmax to Hs at that probability level would be 1.6, which appears to represent an asymptotic limit due to wave breaking.

Figure 31. CresT JIP preliminary guidance on the ratio of Crmax to Hs (left graph) and the ratio of Hmax to Hs (right graph). Source: BSEE TA&R Project #605 report.

Final Report 40 06 Feb 2014 TA&R Project #672 Extreme Metocean Climatology for Offshore Wind

For the Chesapeake Light Tower (CHLV2), using the 50-year SHM significant wave height of 6.3 m as an example, the estimated value of Crmax would be 10.08 m. From the water level results presented in Section 4.1, the 50-year extreme still water level at CHLV2 is estimated to be 1.65 meters above mean sea level (MSL). The minimum elevation of any structure not designed to withstand hydrodynamic loading would thus be the sum of the following values: Integrated tide + storm surge: 1.65 m Maximum wave crest elevation: 10.08 m IEC 61400-3 air gap standard: 1.26 m (= 0.2*6.3m) Total required elevation: 12.99 m above MSL

Note that the probability plots in Figure 31 and example summation above are for a point, as would apply to the maximum crest elevation and maximum individual wave heights at a single wind turbine. For a larger structure such as an electric service platform, the increase in deck area increases the probability (relative to a single wind turbine) of the platform experiencing the same maximum crest elevation or maximum wave height in a given sea state.

Figure 32 depicts the probability of maximum crest elevation for various platform deck areas, with the deck area range for the electric service platform of the 504 MW Greater Gabbard offshore wind project in the UK indicated by a green band.

Figure 32. Ratio of Crmax to Hs as a function of the notional number of waves, T/Tz being the total length of the time period divided by the mean zero-crossing period of the sea state; thus, 1000 represents n encounter probability of 10-3 . Modified after: Forristall, G.Z., 2007. Wave crest heights and deck damage in Hurricanes Ivan, Katrina and Rita, Offshore Technology Conference Proceedings, OTC 18620

Final Report 41 06 Feb 2014 TA&R Project #672 Extreme Metocean Climatology for Offshore Wind

Chapter 4. Extreme Water Levels and Current Speeds

Section 4.1 Extreme Water Levels

This work was performed by Jeff Hanson and Mike Forte of the U.S. Army Corps of Engineers (USACE) Field Research Facility (FRF) at Duck, North Carolina. The FRF leveraged a FEMA Region III storm surge modeling effort to estimate extratropical storm surge levels, hurricane storm surge levels, and the joint-probability storm surge level from a “mixed” statistical analysis of both storm populations (refer back to Figure 14).

To assess the FEMA model skill in forecasting the open-ocean storm surges relevant to offshore wind design, the FRF compared measured and modeled water level elevations for the gauging station that is operated at the end of the FRF pier for the three largest storm-surge producing events in the Mid-Atlantic region, namely 2003 Hurricane Isabel, 2006 Hurricane Ernesto, and the 2009 extratropical storm Nor’Ida. The modeled peak surge at the end of the FRF pier was in perfect agreement with the peak measured storm surge for Ernesto, ~0.3 m higher than measured for Isabel and ~0.4 m higher than measured for Nor’Ida. These modeled levels err on the high side and thus can be considered conservative. Moreover, as can be seen in the example summation for the Chesapeake Light Tower in Section 3.5, above, this error is quite small when compared with the maximum wave crest elevation, which must be superimposed on storm surge in order to design appropriate air gaps for offshore wind structures.

The FEMA Region III maps of the 100-year still-water extreme level (SWEL) for extratropical storms, hurricanes, and both storm types combined is given in Figure 33. The mapped values integrate the contributions of astronomical tide and storm surge into a single SWEL elevation.

Final Report 42 06 Feb 2014 TA&R Project #672 Extreme Metocean Climatology for Offshore Wind

Figure 33. Maps from the FEMA Region III storm surge study, showing still-water extreme level (SWEL), in feet above mean seal level (ft AMSL) for hurricanes, extratropical storms, and both storm types combined.

Final Report 43 06 Feb 2014 TA&R Project #672 Extreme Metocean Climatology for Offshore Wind

Section 4.2 Extreme Surface Current Speeds

For installations seaward of the surf zone, where there is no current induced by breaking waves, IEC 61400-3 and ABS BOWTI specify the surface current speed as a simple multiplier (0.1) times the 10-minute mean surface wind speed (i.e. at an elevation of 10 m ASL). Measurements from a variety of different surface-buoy and seabed-mounted acoustic Doppler current profilers (ADCPs) were evaluated in attempting to validate this multiplier.

One promising source of data was a seabed-mounted ADCP located in 32 m water depth off Wallops Island, Virginia. Unfortunately, however, there was no co-located anemometer, the nearest NDBC station being Buoy 44009 off the coast of Delaware. Much closer to NDBC Buoy 44009 was a data set collected by BOEM off the coast of Delaware and furnished to our project team for analysis. This proved to be most useful for analyzing the vertical profile of current speeds and this analysis is presented in Section 4.3

To complete this task, we analyzed ADCP data from a surface-buoy station operated by the Northeastern Regional Association of Coastal and Ocean Observing Systems (NERACOOS Station A0) in Massachusetts Bay in 65 m water depth. This buoy has both a down-looking ADCP and an anemometer mast. Figure 34 shows the relationship between the shallowest current measurement at 12 m below the surface and the mean wind speed. There is wide scatter in the wind speed multiplier values, which largely fall in the range between the design standard of 0.1 and 65% of this value (i.e., a multiplier of 0.065).

Figure 34. Current speed at 12 m below the sea surface in 65 m water depth as a function of mean surface wind speed from data measured at NERACOOS Station A01.

Final Report 44 06 Feb 2014 TA&R Project #672 Extreme Metocean Climatology for Offshore Wind

Section 4.3 Extreme Current Profile and Sub-Surface Current Speeds

There were two ADCP data sets with well-resolved profiles of sub-surface current speeds, and each of these is summarized separately, below. The first was from a surface-buoy at NDBC Station 41012, with a down-looking ADCP, located in 38 m depth off St. Augustine, Florida, measuring current profiles during Tropical Storm Beryl on 25 May 2012. Mean wind speeds during these measurements were relatively modest, in the range of 8 to 11 m/s. Significant wave height ranged from 1.4 to 2.2 m. The surface current speed was in the range of 0.9 to 1 m/s, tending to validate the IEC wind-speed multiplier of 0.1 to estimate the surface current speed. Five consecutive current profiles are plotted in Figure 35.

Figure 35. Time series of consecutive measured current profiles in 38 m water depth off the coast of Florida during Tropical Storm Beryl on 25 May 2012.

As shown in Figure 36, the current profiles in Tropical Storm Beryl comes close to fitting a 1/7 Power Law as specified in the IEC 61400-3 and ABS BOWTI standards, but this would underestimate the current speeds nearer the bottom. An exponent of 1/14, which is half of the standards-specified value, would be more appropriately conservative.

Final Report 45 06 Feb 2014 TA&R Project #672 Extreme Metocean Climatology for Offshore Wind

Figure 36. Power Law fits to measured current profiles in 38 m water depth off the coast of Florida during Tropical Storm Beryl on 25 May 2012.

The second data set for current profile characterization was from a seabed-mounted ADCP deployed by BOEM on a shoal off the coast of Delaware in 24 m water depth from the end of February to the end of May 2007, where a nor’easter on 06 May 2007 produced the strongest currents in this three-month record. By comparison with the Tropical Storm Beryl situation, these currents developed rapidly as the wind speed increased from 3-4 m/s early in the series to 16-17 m/s at the end of this series.

The measured current profiles during the nor’easter off Delaware (Figure 37) have a very different character than the Tropical Storm Beryl profiles off Florida. When wind speeds are still low (3 to 4 m/s), there is a strong shear gradient in the upper meter of the water column. As the wind speed grows to 11.5 m/s the surface shear layer deepens and the gradient lessens somewhat. By the times of the fastest two profiles, with wind speeds up to 16-17 m/s, the surface layer of fast water deepens further. The wind speed then decreased and the currents rapidly dissipated (not shown in figure).

Final Report 46 06 Feb 2014 TA&R Project #672 Extreme Metocean Climatology for Offshore Wind

It is conjectured that had the winds continued at high levels, perhaps the layer of fast water would have deepened further, and the profiles eventually would have looked like those measured during Tropical Storm Beryl. With such limited data, however, this is only a conjecture.

Figure 37. Time series of consecutive measured current profiles in 24 m water depth off the coast of Delaware during nor’easter on 06 May 2007.

Much more data is needed in order to better understand and characterize the development of extreme wind-driven currents in the Mid-Atlantic region, particularly during storms with mean wind speeds of 20 m/s or more.

Overall, it appears that applying the IEC 61400-3 and ABS BOWTI multiplier of 0.1 to the mean wind speed will provide an appropriately conservative estimate of the surface current speed, and then applying the standards-specified 1/7 Power Law to extrapolate the surface current to the seabed is reasonable, although may underestimate current speeds lower in the water column.

Final Report 47 06 Feb 2014 TA&R Project #672 Extreme Metocean Climatology for Offshore Wind

Despite the great uncertainty in how to best derive storm-driven surface current speed from the fundamental mean surface wind speed parameter, this does not have such a great impact on offshore wind design, because the hydrodynamic loading effects currents are much less than those due to waves. This is shown, for example, in Figure 38, which compares the different metocean loading components that contribute to the overturning moment of a monopile-based turbine in 15 m water depth at a hypothetical site off the southern coast of Massachusetts.

Source: BSEE TA&R Project #605 report.

Figure 38. Mud-line overturning moment (OTM) as a function of wind speed for a 5MW NREL reference wind turbine supported by a monopile foundation in 15 m depth at a study site off the south coast of Massachusetts. Magenta line with square symbols is the OTM contribution by current loading which is only slightly greater than the contribution by wind loading on the tower (green line). Source: BSEE TA&R Project #618 report.

Likewise, currents have less effect than waves on erosion and bottom scour, as can be seen by comparing Figures 39 and 40, which are seabed shear stress analyses made for this study by Larry Atkinson and Jose Blanco of Old Dominion University’s Center for Coastal Physical Oceanography. These are derived from a year’s worth of seabed-mounted ADCP measurements at a station designated COBY-5, located 15 miles off Wallops Island, Virginia, in a water depth of ~32 m. This ADCP yielded two records: 28-Jan-2006 through 25-Jul-2006 and 08-Oct-2006 through 23-Jun-2007. Despite this long record that included a strong nor’easter, there were no wind speed measurements at the COBY-5 station, and so it could not be used for our multiplier validation. Nevertheless, it well illustrates the relative scouring potential of winds and currents.

Final Report 48 06 Feb 2014 TA&R Project #672 Extreme Metocean Climatology for Offshore Wind

Figure 39. Bottom shear stress exceedance diagram for COBY-5 measurement record for currents only, without the sub-surface orbital velocity shear stress of waves. Heavy lavender line indicates tidal current shear stress in the absence of wind-driven currents.

Figure 40. Bottom shear stress exceedance diagram for COBY-5 measurement record, with wave-induced shear stresses included. The erosion potential is much greater with waves added.

Final Report 49 06 Feb 2014