-
Notifications
You must be signed in to change notification settings - Fork 139
Frequently Asked Questions (FAQ)
Is Bladed supported?
The DNV team have created Bladed models of v1.0 and v1.1 of the fixed-bottom 15-MW reference turbine here. There is also a model on the VolturnUS-S floating platform here.
The model was defined based on the IEA Task 37 report, as well as data from the OpenFAST and HAWC2 models. The ROSCO controller parameters were re-tuned to work with the Bladed model, and a compiled copy of the ROSCO 32-bit controller v2.3.0 is provided for convenience.
The download package includes the Bladed model, a verification report, and instructions for use.
Is OrcaFlex supported?
The Orcina team have created a model of version 1.1.3 of the 15-MW reference turbine, atop the VolturnUS-S platform, using OrcaFlex. Modelling of the semi-submersible foundation has been supported by OrcaWave, which has been used to perform a diffraction analysis on the UMaine VolturnUS-S semi-submersible. The model files are publicly available at this link: see the example named 'K03 15MW semi-sub FOWT'. The included PDF document provides further details about the model itself, where the turbine controller is supported through a 64-bit version of NREL's ROSCO DLL. This is implemented through a Python external function which acts as a wrapper written in the Bladed style. It is possible to view this model without having access to an OrcaFlex license. This can be achieved by downloading the demo version of OrcaFlex, available here.
Also available is a validation report, comparing the aeroelastic response of the 15MW rotor and tower, calculated by OrcaFlex, against that predicted by OpenFAST running BeamDyn. The study considers the structural dynamics of the turbine blade oscillating freely under gravity and the aeroelastic response of the rotor-nacelle assembly (RNA) & tower to steady, stepped, skewed & turbulent wind.
Is fatigue accounted for
No detailed fatigue load analysis was performed in the report. We also do not have plans or the resources to add that at this time. As a conservative baseline, we expect the fatigue performance to be fair, but certainly not optimized, whether looking at the blades, tower, monopile, or other components.
Have you considered uprating?
We do not have plans to uprate the turbine at this time. We understand that having a single rotor design for both fixed-bottom and floating applications means it is likely suboptimal for both. Reference turbines are meant to be a starting point for improvement, redesign, and the addition of new technologies. We know the specific power is on the low side and the bigger blades lead to a heavy nacelle and require a larger substructure. We were trying to be a little forward looking with our choice of 240m and 15 MW, expecting specific power to decrease over time for offshore machines. If uprated, the drivetrain, generator, and bed plate would all have to be redesigned, as they were designed specifically for torque outputs of the current rotor. The tower would likely need to be modified too as the maximum thrust and RNA mass change, but this would be more simple.
Can the hub height be lowered?
With a rotor radius of 120m, the hub height of 150m was chosen to have a very conservative blade tip to wave clearance. We do not think there would be any major technical issues with reducing the hub height by 10m. This could have some positive effects for floating applications. However, we would not reduce it beyond 140m, otherwise it will approach some worst case extreme wave heights.
Can I use BeamDyn? And why isn't blade mass matching?
The OpenFAST model files are configured to use ElastoDyn by default. To use BeamDyn, set DT = 0.0005 and CompElast = 2 in the FST-file. Note however that the original ElastoDyn blade files were generated using WISDEM, whereas the BeamDyn files were generated at a subsequent stage using the numerical code SONATA. This mismatch of tools and timeline has caused a mismatch in blade elastic properties. This is visible in the blade mass, which is equal to 68.5 t in ElastoDyn and equal to 66.9 t in BeamDyn. The stiffness and inertia matrices K and I are also different. A more rigorous approach has been followed in the design of the newer 22MW reference wind turbine.
Is a 10m diameter too aggressive?
Current "XL monopiles" coming out of European factories are at about 8-8.5m in diameter. The IEA Wind 10-MW reference turbine uses a 9m diameter monopile. We were trying to aim for technology level that is both reasonable and also 5+ years out. That is a tough line to walk and ultimately a judgment call was made in setting the max OD for the monopile and tower to be 10m.
What is the precise rotor diameter?
There was an inconsistency in the stated hub diameter in the original report, YAML-definition, and WISDEM files (7.94m) versus the hub diameter in the OpenFAST and HAWC2 files (3.0m). This was due to a late design change to better accommodate the blade root chord/diameter size that did not fully propagate to all input files. With the hub diameter at 7.94m and the blade unchanged at 117.15m in length (arc length measurement that includes the prebend), the precise rotor diameter is 242.24m. This is also called out in Issue #51.
Is there additional soil or geotech data available?
We have no additional geo-tech or soil information for the 15-MW reference monopile. It is a reference turbine at a generic reference site, so there are no detailed p-y curves or other soil data. If it helps, the intent was to design the 15-MW monopile for a generic site along the U.S. East Coast. Some users may have access to soil data from that region.
I get a different blade mass value than the report?
The composite material properties that were used initially in the blade design were borrowed from the older NREL 5-MW Reference Wind Turbine description. These have been updated with better estimates for modern blade production techniques in the wind turbine yaml-ontology file and also reflected in the sectional properties of the model input decks. The biggest change is an increase in the carbon fiber density, resulting in heavier blades than the original report (65t vs 73t). Ideally the blades would be re-optimized around these updated material properties, but that will have to wait until another time.
I get a different rated thrust value than the report?
In the monopile report, Figure 3-1b (Steady State Performance) shows a rated rotor thrust value of ~2,800 kN, but Table ES-2 states that the design C_T Design = 0.799, which corresponds to a rated rotor thrust of 2,483 kN. Further simulation supported this lower value and is captured in the Excel tabular data. The higher thrust value is considered a report error.
Furthermore, the reference wind turbine definition has seen many bug fixes, as well as improvements to the underlying computational models, since the original report. We consider simulations using the latest model files to be "truth". We also can no longer recover some of the results shown in the report due to these changes.
I get different performance than the Excel tabular data?
The Rotor Performance tab in the Excel sheet is auto-generated from WISDEM, which uses idealized steady-state conditions and maximizes the power production at each velocity subject to constraints on min or max rpm, torque, etc. There is no true controller involved and full aeroelastic coupling is not present. So, if you simulate similar power curve performance tables in OpenFAST or HAWC2 or another time-domain code, you will arrive at different values for pitch, torque, power, thrust, etc.
Where can I find detailed blade properties?
The detailed description of blade composite layer geometry, thicknesses, and material properties are in the “Ontology” files. These files are compliant with our WindIO project. The documentation will be helpful to understanding coordinate reference systems along the curved blade. If you are just interested in post-processed mass and elastic properties (6x6 matrices), then the tables in the Excel "tabular" files will be the most helpful to you. In these files, the y-axis points to the trailing edge and the x-axis points to the suction side. So, inertia/stiffness M/K_44 (xx) is edge motion, M/K_55 (yy) is flap motion, and M/K_66 (zz) is torsion.
Are FEM or high-resolution CAD models of the blade available?
We have a couple of ongoing efforts to create models of the blade that can support high-fidelity analysis. This includes Ansys FEM models and CAD models of sufficient quality for CFD analysis. These efforts are starting up in late 2020 and will go through extensive testing before being made publicly available here.
The Class I average velocity and the metocean average velocity in the report do not match?
A Class I site average velocity is defined as 10m/s, but the report describes a generic U.S. East Coast offshore site with an average velocity of 8.65m/s. Which velocity was used for the AEP and loads calculations? The answer is that the 8.65m/s site was only described to find the significant wave height and period. Otherwise, the standard Class IB conditions (10m/s average velocity, Weibull shape parameter of 2) were applied to the AEP calculation. The DLC loads simulations parameterized the wind speeds.
Why am I seeing floating tower resonance?
The tower designed for the UMaine Volturn-S semisubmersible reported a first natural frequency around 0.5Hz, which kept it safely away from rotor 3P excitation. The set of modeling tools used to estimate that frequency and drive the tower design was found to have in error. In simulation in both OpenFAST and HAWC2, the first natural frequency of the floating tower is closer to 0.4Hz, which leads to some resonance excitation from the blade 3P mode. For now, this is a known issue. Future design updates will address this shortcoming.
Are the CFRP properties wrong?
The properties of the carbon fiber laminate adopted in the spar caps of the blades of the IEA15 RWT are unrealistic. The density of the laminate is too low, resulting in unrealistically low fiber volume and weight fractions, and low bill of material (BOM) costs. Also, modern blades do not adopt infused CFRP, rather prepreg or pultruded CFRP. The source of the error comes from Table 5 in this report. The error is also present in the BOM of the SNL 100 design in Table 27 of the report about the blade cost model in WISDEM. New CFRP props specified within the Big Adaptive Rotor project will be adopted in the next RWT design.
What are the values of structural damping?
There was some confusion/bug in the past about these values. For the tower, there is significant uncertainty about how to ascertain the true value as it is usually estimated experimentally. We understand 1% to be a realistic number for the isolated tower (ElastoDyn requires the input for the isolated tower).
For the blades, the models on GitHub now use log-dec structural damping of 3% in flap and edge and 6.787 % in torsion. This leads to damping ratios of 0.48% in flap and edge, and 1% in torsion. Lab tests of much smaller blades and some industrial collaborations suggest that these values might be still a little high, but not crazy. Note that BeamDyn uses Rayleigh damping, which is computed like this:
import numpy as np
delta = np.array([0.03, 0.03, 0.06787]) # logarithmic decrement, natural log of the ratio of the amplitudes of any two successive peaks. 3% flap and edge, 6% torsion
zeta = 1. / np.sqrt(1.+(2.*np.pi / delta)**2.) # damping ratio, dimensionless measure describing how oscillations in a system decay after a disturbance
omega = np.array([0.508286, 0.694685, 4.084712])*2*np.pi # Frequency (rad/s), flap/edge/torsion
mu1 = 2*zeta[0]/omega[0]
mu2 = 2*zeta[1]/omega[1]
mu3 = 2*zeta[2]/omega[2]
mu = np.array([mu1, mu2, mu3, mu2, mu1, mu3])
A journal paper describing aeroelastic studies on the IEA15 (and other turbines) is in preparation.
Is it me or do the nacelle mass properties seem inconsistent?
The original report and model files had a number of inconsistencies that were cleaned up and corrected in v1.1 (see the Release Notes). The model files always see the greatest use and testing and should be treated as the most trusted. Second, the tabular documentation is auto-generated by WISDEM, but not necessarily updated with regularity. At this point, we do not recommend using the original report data. Note that the definition of what constitutes the "nacelle" differs from model to model. For instance, OpenFAST considers components above the yaw drive whereas WISDEM includes the yaw system.
Why do the simulated VolturnUS-S properties differ from the report?
If you are using WISDEM or OpenFAST, currently neither model support structural members with rectangular cross sections, which are found on the VolturnUS-S. Therefore, the models do not perfectly capture the original design intent. If the geometry isn't the same as the original report, then the center of mass, displacement, and other properties will also be off. The values used in the OpenFAST model for ballast are those necessary to obtain equilibrium with an imperfect geometry. At the same time, the original design did not include details on the internal structure or scantling of the VolturnUS-S, so as we fill in that detail over time, some of the properties may eventually deviate from the published values.
Why are the tower properties for the monopile model a little jumpy?
The thickness in the tower models are piecewise constant to reflect manufacturing with rolled steel plates stacked on top of one another. Also, there is a discrete jump in the thickness from 15.000 to 15.001 m in the original tower design, but in the yaml and in OpenFAST/HAWC2, the "tower" body is modeled from 15m upwards. To prevent these models having a discrete jump in thickness in the bottom part of the tower, the properties at the base are changed slightly. Please see this Issue and Pull Request for more details.
Does the monopile diameter-to-thickness ratio seems unrealistic?
The values of ~250 currently in the 15-MW design are a reflection of our design process that had to neglect fatigue loads (FLS) due to limitations in schedule and budget and focus on ultimate load cases (ULS) and modal analysis only. The beam models and buckling checks used prefer large diameters with thin walls for stability and nailing a target natural frequency. If FLS were included, we would expect to see lower D:t ratios. We acknowledge that D:t ratios in the range of 80-160 are more realistic. Should there be a major design revision to the 15-MW reference turbine, the monopile would be updated accordingly.
Fixed-bottom report
- Table 1-1 has a space-character printed instead of decimal for "Design Tip Speed Ratio". The correct values are 7.5 and 9.0. Same issue for Table 4-1 for the first three material properties. Correct values are E=2.0E11, G=7.93E10, rho=7.85E3 .
- Table 1-1 lists monopile parameters for the DTU 10 MW, which is an onshore turbine and thus does not have a monopile. The values for transition piece height, embedment depth, base diameter and monopile mass in the table are parameters from the IEA 10 MW monopile.
- Figure 3-1b (Steady State Performance) shows a rated rotor thrust value of ~2,800 kN, but the correct value should be 2,483 kN as reflected in Table ES-2 (C_T Design = 0.799).
- Figure 3-3 is non-dimensional and should not have the '[m]' included
- Table 5-1 is inaccurate and updated in the tabular Excel documentation
- The Rotor Thrust in the middle plot of Figure 3-1 is the summary of all loads on the shaft ('RotThrust' output in OpenFAST which includes gravitational and inertial loads), and is not the pure aerodynamic thrust from which the Thrust Coef is defined (‘RtAeroFxh’ output)in the bottom plot of Figure 3-1. See this thread on the NREL Forums for more info.
- In Figure 5-4b, the legend descriptions for items 7 and 8 should be swapped
- In Table 5-4, the line entry for "RMS line voltage" should be "RMS phase voltage"
Floating report
- Table 4 entry (4,2) for (Roll,Sway) should be equal to entry (2,4) for (Sway,Roll) = 1.010E+08
- On page 12, the density of salt water should be 1,025 kg/m^3, not 1,250 as written.