TurbSim User’s Guide

v2.00.00
B. J. Jonkman
National Renewable Energy Laboratory

Revised June 1, 2016 for TurbSim v2.00.00

DRAFT VERSION

NREL is a national laboratory of the U.S. Department of Energy

Office of Energy Efficiency & Renewable Energy
Operated by the Alliance for Sustainable Energy, LLC
This report is available at no cost from the National Renewable Energy
Laboratory (NREL) at www.nrel.gov/publications.

Technical Report
NREL/TP-xxxx-xxxxx
October 2014

Contract No. DE-AC36-08GO28308

 TurbSim User’s Guide
v2.00.00
B. J. Jonkman
National Renewable Energy Laboratory
Prepared under Task No(s). WER8.2102, WA093101,
WE110331, and WE143F01

NREL is a national laboratory of the U.S. Department of Energy

Office of Energy Efficiency & Renewable Energy
Operated by the Alliance for Sustainable Energy, LLC
This report is available at no cost from the National Renewable Energy
Laboratory (NREL) at www.nrel.gov/publications.

National Renewable Energy Laboratory Technical Report

15013 Denver West Parkway NREL/TP-xxxx-xxxxx
Golden, CO 80401 October 2014
303-275-3000 • www.nrel.gov
Contract No. DE-AC36-08GO28308
 NOTICE

This report was prepared as an account of work sponsored by an agency of the United States government.
Neither the United States government nor any agency thereof, nor any of their employees, makes any warranty,
express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of
any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately
owned rights. Reference herein to any specific commercial product, process, or service by trade name,
trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation,
or favoring by the United States government or any agency thereof. The views and opinions of authors
expressed herein do not necessarily state or reflect those of the United States government or any agency thereof.

This report is available at no cost from the National Renewable Energy

Laboratory (NREL) at www.nrel.gov/publications.

Available electronically at http://www.osti.gov/scitech

Available for a processing fee to U.S. Department of Energy

and its contractors, in paper, from:

U.S. Department of Energy

Office of Scientific and Technical Information
P.O. Box 62
Oak Ridge, TN 37831-0062
phone: 865.576.8401
fax: 865.576.5728
email: mailto:reports@adonis.osti.gov

Available for sale to the public, in paper, from:

U.S. Department of Commerce

National Technical Information Service
5285 Port Royal Road
Springfield, VA 22161
phone: 800.553.6847
fax: 703.605.6900
email: orders@ntis.fedworld.gov
online ordering: http://www.ntis.gov/help/ordermethods.aspx

Cover Photos: (left to right) photo by Pat Corkery, NREL 16416, photo from SunEdison, NREL 17423, photo by Pat Corkery, NREL
16560, photo by Dennis Schroeder, NREL 17613, photo by Dean Armstrong, NREL 17436, photo by Pat Corkery, NREL 17721.

NREL prints on paper that contains recycled content.

Acknowledgments
TurbSim was written by Bonnie Jonkman at the National Wind Technology Center (NWTC).

Analysis of coherent events was performed by Neil Kelley, Bonnie Jonkman, and George Scott
of the National Wind Technology Center, and Professor Jan Bialasiewicz, and Lisa Redmond of
the University of Colorado at Denver.

The turbulence modeling scaling parameters for the GP_LLJ and NWTCUP spectral models
were developed by Neil Kelley and Bonnie Jonkman. Neil Kelley developed scaling parameters
for the National Renewable Energy Laboratory wind farm models.

Levi Kilcher of the National Wind Technology Center added the NREL/UW Tidal Channel
(TIDAL) spectral model to extend TurbSim’s use for water turbulence.

Yi Guo of the National Wind Technology Center added the API model for hurricane modeling.

iii

This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications.
List of Acronyms
ART Advanced Research Turbine
BLAS Basic Linear Algebra Subprograms
CoRA Colorado Research Associates
CTKE coherent turbulent kinetic energy
CXML Compaq Extended Math Library
DNS direct numerical simulation
ETM Extreme Turbulence Model
EWM Extreme Wind Model
FF full field
FFT Fast Fourier Transform
FFTPACK FFT Package
HH hub height
IEC International Electrotechnical Commission
IFFT Inverse Fast Fourier Transform
LAPACK Linear Algebra Package
LES large-eddy simulation
LIST Long-Term Inflow and Structural Testing
LLLJP Lamar Low-Level Jet Project
MHK marine and hydrokinetic
NCAR National Center for Atmospheric Research
NREL National Renewable Energy Laboratory
NTM Normal Turbulence Model
NWTC National Wind Technology Center
pRNG pseudorandom number generator
SODAR sonic detection and ranging
TI turbulence intensity
TKE turbulent kinetic energy

iv

This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications.
Table of Contents
Acknowledgments ..................................................................................................................................... iii
List of Acronyms ........................................................................................................................................ iv
Introduction ................................................................................................................................................. 1
History .......................................................................................................................................................... 1
Retrieving Files from the Archive .............................................................................................................. 2
Distributed Files .......................................................................................................................................... 3
Certification Test ......................................................................................................................................... 4
Compiling TurbSim ..................................................................................................................................... 4
Using TurbSim ............................................................................................................................................. 4
Running the software.............................................................................................................................. 4
Coordinate Systems ................................................................................................................................ 5
Input Files .................................................................................................................................................... 6
Primary Input File .................................................................................................................................. 6
Input File for User-Defined Time Series .............................................................................................. 31
Input File for User-Defined Spectra ..................................................................................................... 32
Input file for User-Defined Profiles ..................................................................................................... 33
Output Files ............................................................................................................................................... 34
Summary Files ...................................................................................................................................... 34
Hub-Height Binary Files ...................................................................................................................... 35
Hub-Height Formatted Files ................................................................................................................. 35
Hub-Height AeroDyn Formatted Files ................................................................................................. 35
Full-Field TurbSim Binary Files .......................................................................................................... 36
Full-Field Bladed-Style Binary Files.................................................................................................... 36
Tower Data Binary Files ...................................................................................................................... 37
Full-Field Formatted Files .................................................................................................................... 37
Coherent Turbulence Time-Step Files.................................................................................................. 37
Spectral Models ......................................................................................................................................... 40
IECKAI: The IEC Kaimal Model......................................................................................................... 40
IECVKM: The IEC Von Karman Isotropic Model .............................................................................. 41
SMOOTH: The Risø Smooth-Terrain Model....................................................................................... 41
NWTCUP: The NREL National Wind Technology Center Model ...................................................... 44
GP_LLJ: The NREL Great Plains Low-Level Jet Model..................................................................... 45
WF_UPW: The NREL Wind Farm, Upwind Model ............................................................................ 46
WF_14D: The NREL Wind Farm, Downwind Model (14 Rotor Diameters) ...................................... 47
WF_07D: The NREL Wind Farm, Downwind Model (7 Rotor Diameters) ........................................ 48
TIDAL: The NREL/UW Tidal Channel Model ................................................................................... 49
TIMESR: Time Series Input................................................................................................................. 50
USRINP: User-Input Spectra ............................................................................................................... 50
USRVKM: von Karman Model with User-Defined Scaling ................................................................ 50
API: API Spectrum for Hurricane Winds ............................................................................................. 51
Spatial Coherence Models ....................................................................................................................... 51
GENERAL: A general spatial coherence model .................................................................................. 52
IEC: IEC Coherence Model.................................................................................................................. 52
NONE: Identity Coherence .................................................................................................................. 53
API: API Longitudinal Coherence ....................................................................................................... 53
Velocity and Direction Profiles ................................................................................................................ 54
Power-Law Wind Profile...................................................................................................................... 55
v

This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications.
 Logarithmic Wind Profile..................................................................................................................... 55
Logarithmic Water Profile.................................................................................................................... 56
IEC Wind Profile .................................................................................................................................. 56
Low-Level Jet Wind Profile ................................................................................................................. 56
API Wind Profile .................................................................................................................................. 57
User-Defined Velocity Profiles ............................................................................................................ 57
Coherent Structures ................................................................................................................................. 58
Adding and Scaling the Coherent Structures........................................................................................ 59
Using Coherent Turbulence Time-Step Files with AeroDyn v13 ........................................................ 60
Suggestions for Generating Coherent Turbulent Structures ................................................................. 61
Warnings .................................................................................................................................................... 62
Limitations ................................................................................................................................................. 62
Possible Future Enhancements .............................................................................................................. 62
Caveats....................................................................................................................................................... 63
Feedback .................................................................................................................................................... 63
Appendix A: Sample TurbSim Input Files .............................................................................................. 67
Appendix B: TurbSim Quick-Start Guidelines for IEC Turbulence ...................................................... 70
Appendix C: Flow Charts ......................................................................................................................... 73
Appendix D: Full-Field TurbSim Binary File Format ............................................................................. 81
Appendix E: Full-Field Bladed-Style Binary File Format ...................................................................... 83
Appendix F: Tower Data Binary File Format .......................................................................................... 85
Appendix G: Velocity Spectra Comparison Plots .................................................................................. 87
Appendix H: Sample AeroDyn v13 Coherent Turbulence Parameter Input File ................................. 93

vi

This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications.
List of Figures
Figure 1. TurbSim simulation method: a transformation from the frequency domain to time domain
producing wind output compatible with InflowWind; optional coherent structures are written
to a separate file and superimposed in AeroDyn v13 (they require a full-field background
wind file) .................................................................................................................................. 1
Figure 2. Example TurbSim command line output ....................................................................................... 5
Figure 3. Coordinates of a TurbSim wind field with 15° horizontal and 8° vertical mean flow angles ....... 6
Figure 4. Example of tower points below a rectangular grid ........................................................................ 8
Figure 5. Example grid and rotor placements: the circles pictured here are the rotor diameters assumed by
TurbSim; the actual rotor diameter(s) will be smaller than pictured ..................................... 12
Figure 6. Example of TurbSim grids as implemented in InflowWind: (a) The inertial frame coordinate
systems and planes “marching” along positive X, regardless of flow angles, (b) wind field
with both flow angles 0°, (c) the same wind field with VFlowAng = 8° and HFlowAng = 15°14
Figure 7. Longitudinal wind-speed standard deviation and TI for IEC turbulence categories as functions
of the mean hub-height wind speed, Vhub ............................................................................... 17
Figure 8. Coherent turbulent kinetic energy (CTKE) of an example simulation using KHTEST: the
coherent structure placement in the middle of the time series is shown by the red lines....... 17
Figure 9. Default jet wind speed for URef calculation: error bars indicate the range of random variate, N;
dotted lines mark the tenth and ninetieth percentiles ............................................................. 20
Figure 10. Default jet height, ZJetMax, without random variates (the random variation range is
approximately ±50 m) ............................................................................................................ 20
Figure 11. Diabatic friction velocity, u*0 , normalized by URef and calculated using RefHt = 80 m and
Z0 = 0.01 m ............................................................................................................................ 23
Figure 12. Default UStar as a function of diabatic friction velocity, u*0 (left: SMOOTH, GP_LLJ, and
NWTCUP models, right: wind farm models) ........................................................................ 23
Figure 13. Default u-component coherence parameters, IncDec1, ( au left, bu right) as a function of wind
speed, using RICH_NO = 0 and HubHt = 80 m (IEC values are plotted for comparison) .... 27
Figure 14. Default v-component coherence parameters, IncDec2, ( av left, bv right) as a function of wind
speed, using RICH_NO = 0 and HubHt = 80 m (IEC does not define v-component coherence
parameters) ............................................................................................................................. 28
Figure 15. Default w-component coherence parameters, IncDec3, ( aw left, bw right) as a function of
wind speed, using RICH_NO = 0 and HubHt = 80 m (IEC does not define w-component
coherence parameters)............................................................................................................ 29
Figure 16. Coherent structure scaling (looking downwind): CTLy is the tower location, CTLz is the hub
(height) location, and DistScl determines the size (m) relative to the rotor disk ................... 31
Figure 17. Example TurbSim FF and HH wind files as implemented in AeroDyn. ................................... 39
Figure 18. SMOOTH-model stable/neutral turbulence as a function of RICH_NO: left: standard deviation
normalized by UStar, right: Relationships between components’ standard deviations ......... 42
Figure 19. NWTCUP-model stable/neutral turbulence as a function of RICH_NO: left: standard deviation
normalized by UStar, right: relationships between components’ standard deviations .......... 44
Figure 20. GPLLJ-model stable/neutral turbulence as a function of local stability and shear velocities ... 46
Figure 21. WF_UPW-model stable/neutral turbulence as a function of RICH_NO: left: standard deviation
normalized by UStar, right: relationships between components’ standard deviations .......... 47
Figure 22. WF_07D- and WF_14D-model stable/neutral turbulence as a function of RICH_NO: left:
standard deviation normalized by UStar, right: ratios of standard deviations ....................... 48

vii

This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications.
Figure 23. Example wind-speed profiles generated in TurbSim for the GP_LLJ turbulence model using a
90-m hub-height wind speed of 12 m/s, and RICH_NO = 0.05............................................. 55
Figure 24. Example jet wind profiles with a 12 m/s wind speed at 80 m and RICH_NO = 0.05 ............... 57
Figure 25. Coherent structures in a 10-minute period from the NWTC LIST dataset: the time series shows
the 3-s mean CTKE (solid black line) crossing the dashed threshold line, which indicates the
location of the coherent structures (indicated in blue) ........................................................... 58
Figure 26. Example time series from the NWTCUP model: the black line indicates the background wind
file; the green shows the addition of events in a coherent time-step file (“.cts”) ................... 59
Figure A-1. Sample Primary TurbSim input file (Note: figure is continued from previous page.) ............ 68
Figure A-2. Sample input file for User-Defined Time Series ..................................................................... 68
Figure A-3. Sample input file for User-Defined Spectra ............................................................................ 69
Figure A-4. Sample input file for User-Defined Profiles ............................................................................ 69
Figure B-1. Sample TurbSim input file for IEC turbulence: parameters shown in blue should be changed
based on the turbine configuration; parameters shown in red should be changed for each load
case and simulation. (Note: figure is continued from previous page.) ................................... 72
Figure C-1. Overview of the TurbSim simulation method; blue lines indicate processes influenced by
input-file parameters; black lines indicate internal variables and processes .......................... 73
Figure C-2. Parameters in the Runtime Options section of the input file ................................................... 74
Figure C-3. Parameters in the Turbine/Model Specifications section of the TurbSim input file................ 75
Figure C-4. Parameters in the Meteorological Boundary Conditions section of the TurbSim input file (for
IECKAI and IECVKM models only) ..................................................................................... 76
Figure C-5. Parameters in the Meteorological Boundary Conditions section of the TurbSim input file (for
models other than IECKAI and IECVKM) ............................................................................ 77
Figure C-6. Parameters in the Non-IEC Meteorological Boundary Conditions section of the TurbSim
input file ................................................................................................................................. 78
Figure C-7. Default input values for the for the Meteorological Boundary Conditions and Non-IEC
Meteorological Boundary Conditions sections of the TurbSim input file ............................. 79
Figure C-8. Parameters for coherent structures and the Coherent Turbulence Scaling Parameters section of
the TurbSim input file; the SMOOTH model uses the GP_LLJ scaling ................................ 80
Figure G-1. Neutral velocity spectra for the 8 spectral models available in TurbSim, using a 15 m/s wind
speed at 80 m; IECKAI and IECVKM use NTM category “B” and 61400-1 3rd ed. scaling;
the non-IEC models use RICH_NO = 0 and UStar = “default”............................................. 87
Figure G-2. Neutral velocity spectra for the 8 spectral models available in TurbSim, using a 15 m/s wind
speed at 80 m; IECKAI and IECVKM use NTM category “B” and 61400-1 3rd ed. scaling;
the non-IEC models use RICH_NO = 0 and UStar = 1.1 m/s ............................................... 88
Figure G-3. Stable velocity spectra using a 15 m/s wind speed at 80 m; the non-IEC models use
RICH_NO = 0.05 and UStar = “default”; The IEC models, which are neutral
(RICH_NO = 0), were added for reference; they use NTM category “B” and 61400-1 3rd ed.
scaling .................................................................................................................................... 89
Figure G-4. Stable velocity spectra using a 15 m/s wind speed at 80 m; the non-IEC models use
RICH_NO = 0.05 and UStar = 1.1 m/s; the IEC models, which are neutral (RICH_NO = 0),
were added for reference; they use NTM category “B” and 61400-1 3rd ed. scaling............. 90
Figure G-5. Unstable velocity spectra using a 15 m/s wind speed at 80 m; the non-IEC models use
RICH_NO = -0.05 and UStar = “default”; the IEC models, which are neutral
(RICH_NO = 0), were added for reference; they use NTM category “B” and 61400-1 3rd ed.
scaling .................................................................................................................................... 91

viii

This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications.
Figure G-6. Unstable velocity spectra using a 15 m/s wind speed at 80 m; the non-IEC models use
RICH_NO = -0.05 and UStar = 1.1 m/s; the IEC models, which are neutral (RICH_NO = 0),
were added for reference; they use NTM category “B” and 61400-1 3rd ed. scaling............. 92

List of Tables
Table 1. Files in the TurbSim Archive .......................................................................................................... 3
Table 2. Definitions of TurbSim Wind-Component Coordinate Systems .................................................... 6
Table 3. Valid ScaleIEC Values ................................................................................................................. 10
Table 4. Valid TurbSim Turbulence (Spectral) Models.............................................................................. 15
Table 5. Valid IEC Turbulence Models ...................................................................................................... 18
Table 6. Valid WindProfileType Values ..................................................................................................... 19
Table 7. Default Inputs for WindProfileType.............................................................................................. 19
Table 8. Default Inputs for Meteorological Boundary Conditions ............................................................. 21
Table 9. Default Inputs for SCMod1, SCMod2, and SCMod3 .................................................................... 26
Table 10. Valid CTEventFile Entries .......................................................................................................... 30
Table 11. Parameters in Hub-Height Binary and Formatted Files .............................................................. 36
Table 12. Format of Hub-Height AeroDyn Files ........................................................................................ 36
Table 13. Coefficients for the API (Frøya) Coherence Model.................................................................... 54
Table 14. Coherent Structure Peak CTKE Scaling ..................................................................................... 60
Table D-1. Full-Field TurbSim Binary File Header Format ....................................................................... 81
Table D-2. FF TurbSim Binary File Grid Format ....................................................................................... 82
Table E-1. Full-Field Bladed-Style Binary File Header Format ................................................................. 83
Table E-2. Format of Grid Velocities in Full-Field Bladed-Style Binary File Format ............................... 84
Table F-1: Format of Header in TurbSim Binary Tower-Data File ............................................................ 85
Table F-2: Format of Grid Velocities in TurbSim Binary Tower-Data File ............................................... 85

ix

This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications.
Introduction
TurbSim is a stochastic, full-field, turbulent-wind simulator. It uses a statistical model (as
opposed to a physics-based model) to numerically simulate time series of three-component wind-
speed vectors at points in a two-dimensional vertical rectangular grid that is fixed in space.
TurbSim output can be used as input into InflowWind-based [1] codes such as FAST [2] or
MSC.ADAMS® [3]. InflowWind uses Taylor’s frozen turbulence hypothesis to obtain local
wind speeds, interpolating the TurbSim-generated fields in both time and space.

Spectra of velocity components and spatial coherence are defined in the frequency domain, and
an inverse Fourier transform produces time series. The underlying theory behind this method of
simulating time series assumes a stationary process. To simulate non-stationary components,
TurbSim—used with AeroDyn v13 [4]—can superimpose coherent turbulent structures onto the
time series it generates. The basic simulation method is summarized in Figure 1.

Figure 1. TurbSim simulation method: a transformation from the frequency domain to time
domain producing wind output compatible with InflowWind; optional coherent structures are
written to a separate file and superimposed in AeroDyn v13 (they require a full-field background
wind file)

History
In 1988, Paul Veers of Sandia National Laboratories wrote a program called SNLWIND [5] that
could generate full-field turbulent wind for the streamwise (u) component only. In 1992, Neil
Kelley of the National Renewable Energy Laboratory (NREL) added several spectral models to
SNLWIND and modified it to generate the v and w components [6]. SNLWIND-3D was the
result. During the next five years NREL researchers modified the program further, including
adding the Kaimal and von Karman spectral models specified by the International
Electrotechnical Commission (IEC). Gary Desroachers modified it to run on many different
platforms by including C-preprocessor directives for conditional compilation. In 1997, Marshall

1
Buhl added the ability to generate binary files that are compatible with Garrad Hassan’s “GH
Bladed” turbine design code [7].

Both SNLWIND and SNLWIND-3D were written in FORTRAN 77 and required recompilation
for different grid densities and run lengths. This made using and supporting the programs
difficult.

In January 2000, Neil Kelley removed all of the spectral models except the two defined by the
IEC, and Marshall Buhl modernized the remaining code by rewriting all but the Fast Fourier
Transform (FFT) routines in Fortran 95 and eliminating the C-preprocessor directives. Buhl’s
changes included using dynamic-memory allocation for the big arrays, which eliminated the
need to recompile for different grid sizes and run lengths. He modified the input and output file
format, streamlined processes, and added the ability to generate hub-height files in AeroDyn
format. Because the changes were substantial, Buhl renamed the program SNwind
(Sandia/NREL Wind) [8].

In 2003, NREL researchers updated the code to add results from the Lamar Low-Level Jet
Project (LLLJP) and from the National Wind Technology Center (NWTC) Long-Term Inflow
and Structural Testing (LIST) project. Bonnie Jonkman also added the spectral models from
SNLWIND-3D that were removed from SNwind, and she replaced the FORTRAN 77 FFT
routines with more modern routines from the Compaq Extended Math Library (CXML).
Jonkman made changes to the Cholesky factorization algorithm, which sped up the code and
allowed for a significant reduction in the memory required to run the program. She eliminated
the requirement that the grid be an even number of points in each direction, and allowed the grid
height to be different from its width. After these enhancements were in place, the code was
modified to generate coherent structures with realistic temporal and spatial scaling, and coherent
turbulence time-step files became another output option. The code was then renamed TurbSim
(turbulence simulator) because of its ability to generate coherent turbulence.

In 2005, Bonnie Jonkman replaced the CXML FFT routines with routines from FFTPACK [9] so
that TurbSim could be compiled on the Intel® Visual Fortran compiler. This made the code run
much faster and also made it more portable. In 2009, Jonkman updated algorithms in the code to
enable users to create much larger grids than were possible in earlier versions.

In 2012, Levi Kilcher added a new spectral model (TIDAL) to extend TurbSim’s use for marine
and hydrokinetic (MHK) energy. This model simulates turbulence in the water, and is not
intended to be used for wind simulation.

Neil Kelley has written a companion document, Overview of the TurbSim Stochastic Inflow
Turbulence Simulator [10], which discusses the development of TurbSim and includes some of
the theory behind that development.

Retrieving Files from the Archive

The TurbSim archive can be downloaded from the NREL Web server page at
http://nwtc.nrel.gov. The downloaded file will have a name like “TurbSim_v150.exe.” Create a
TurbSim folder somewhere on your file system and put this file there. You can double click on it
from Windows Explorer or type “TurbSim_v150” (or the exact file name) at a command prompt,

2
using the TurbSim folder as the current directory. Running this executable file creates some files
and folders. Please see Marshall Buhl’s paper Installing NWTC CAE Tools on PCs Running
Windows® [11] for information on how to set up TurbSim to run in any folder.

To be able to generate coherent structures with TurbSim, users will also need to download the
coherent structures archive from NREL’s Web server page. The file is named
“TSM_structures.exe.” Create a folder on your file system and put this file there. Execute the
program by double clicking on it or by typing “TSM_structures” at a command prompt with the
folder you created as the current directory. When executed, this archive creates the files and
folders used to define coherent structures. It is necessary to type the name of the path to these
folders in TurbSim and AeroDyn v13 input files.

Distributed Files
The archive contains the TurbSim executable program for both 32- and 64-bit Windows®
platforms. See Table 1 for a complete list of the files included in the TurbSim archive.

Table 1. Files in the TurbSim Archive

File(s) Description
ArcFiles.txt The list of files that are written to the archive
Archive.bat The batch file that creates the archive
ChangeLog.txt The list of changes to TurbSim
Disclaimer.txt The software disclaimer
RunTurbSim.pl A sample Perl script used to run TurbSim, using a different seed each time
TurbSim.exe The TurbSim program for 32-bit Windows® platforms
TurbSim64.exe The TurbSim program for 64-bit Windows® platforms
TurbSim.inp A sample input file
TurbSim_Hydro.inp A sample input file with parameters set for the TIDAL model (for MHK use only)
TurbSim.pdf The user’s guide in PDF format
TurbSim_AD.ctp A sample AeroDyn coherent turbulence parameter input file
TurbSimOverview.pdf Kelley & Jonkman’s overview of TurbSim, in PDF format [9]
Source\*.* The Fortran source code for TurbSim
Files used to run and manage the certification tests and MATLAB® scripts for
Test\*.*
reading TurbSim data
Test\EventData\*.* Files used to run the certification tests with coherent turbulence
Test\TstFiles\*.* NREL results for the certification tests

3
Certification Test
Before using TurbSim for the first time, run the certification testing program. It is a batch file
called “CertTest.bat” and is located in the “Test” folder. To test the installation, edit
“CertTest.bat” and set the environment variables found near the top of the file to settings that are
compatible with your system. You probably will have to change only the “Editor” variable. Then
open a command window, go to the Test folder, and type “CertTest” or—if you have
MATLAB® [12] installed on your computer and would like to see plots of the data—type
“CertTest MATLAB.”

When the certification testing program is run, TurbSim executes several times. The test
procedure compares the new results to those stored in the “Test\TstFiles” folder, and it writes the
differences between the output files to a file called “CertTest.out.” If you have specified the
“MATLAB” option, MATLAB opens and plots many results. It might be necessary to close the
MATLAB program before the test procedure can continue. Before finishing, the test procedure
automatically opens the “CertTest.out” file with the editor you specified with the “Editor”
variable. Scan through the file; the only differences should be the date and time stamps in the
headers of the files and the CPU time in the summary files. If you recompiled TurbSim with
another compiler, some slight differences could appear in the last digit of many of the numbers.

Compiling TurbSim
It should not be necessary to compile TurbSim unless you want to make changes to the code or
want to run TurbSim using a different operating system. The archive contains Fortran code
specific to TurbSim. It also contains the Fortran FFTPACK version 4.1 [9], LAPACK version
3.0 [13], BLAS [14], [15], and RanLux [16] routines that TurbSim uses. Users must also
download the NWTC Subroutine Library version 2.04.00, [17], which TurbSim uses. It can be
found under Miscellaneous Software on the NREL Web server page at http://nwtc.nrel.gov/.

The code has been written primarily for the Intel® Visual Fortran compiler. To port TurbSim to
another platform or compiler, it might be necessary to make changes in the NWTC Subroutine
Library’s SysVF.f90 file and possibly the BLAS LSAME() function. If you have access to an
optimized BLAS library, you are encouraged to link your code with it instead of using the
reference BLAS routines included in the TurbSim archive.

Using TurbSim
Running the software
To begin using TurbSim, a text input file is required. Sample input files—which can be
modified—are contained in the TurbSim archive and in Appendix A. A quick-start guide for
using the most basic turbulence is included in Appendix B.

To run TurbSim enter “turbsim [/h] [<RootName.ext>]” at a command prompt; /h and

<RootName.ext> are optional. The /h switch generates a help message, and <RootName.ext> is
the name of the TurbSim input file. Following are two examples:

4
• turbsim
This starts TurbSim and opens the input file “turbsim.inp.” It is equivalent to entering
“turbsim turbsim.inp.”

• turbsim myroot.tsm
This starts TurbSim and opens the input file “myroot.tsm.”
All output files have the specified root file name and different extensions.

Figure 2. Example TurbSim command line output

Coordinate Systems
Wind components are defined in two separate coordinate systems as described in Table 2 and
pictured in Figure 3. TurbSim computes winds in a coordinate system aligned with the direction
of the mean velocity vector at each point in space. The velocities are rotated to the inertial
reference frame coordinate system before they are written to output files.

5
 Table 2. Definitions of TurbSim Wind-
Component Coordinate Systems

Inertial Reference Frame

Along positive X (nominally
U
downwind)
Along positive Y (to the left
V
when looking along X)
Up, along positive Z (opposite
W
gravity)

Aligned with the Mean Wind

u Streamwise (longitudinal)
v Transverse (crosswise)
w Vertical Figure 3. Coordinates of a TurbSim wind field with
15° horizontal and 8° vertical mean flow angles

Input Files
TurbSim reads text input files to set the parameters required for the program to execute. Most of
the parameters are contained in the primary TurbSim input file; certain options may require
parameters from secondary input files. All of these files are described below.

TurbSim assumes that parameters are located on specific lines in its input file(s), so do not add or
remove lines from the sample input files included in the archive. None of the input parameters
are case sensitive.

Each line in the input file that contains a parameter also contains its name, a description, and the
units for that parameter. The names of the parameters are provided for reference, but TurbSim
does not read those names from the input file. Note that other programs or scripts that generate
TurbSim input files, however, can and do use these parameter names.

Primary Input File

The parameters in TurbSim’s primary input file are divided into several sections (discussed
below). The text of an example TurbSim input file is included in Appendix A of this guide.

Runtime Options
The Runtime Options section initializes the pseudorandom number generator (pRNG) and tells
TurbSim what type of output to generate. Appendix C contains a flow chart showing the function
of the input parameters from this section. Users can choose any combination of output types
listed in this section, but at least one output file must be generated to successfully run the code.
The Output Files section of this guide provides more complete descriptions of these file types.

6
Echo: Echo input data to <RootName>.ech [-]
This input parameter is used for debugging purposes. When “true”, the program will echo the
variables to a file with a “.ech” extension each time the program reads a line from the input file.
When “false”, the “.ech” file is not created.

RandSeed1: The First Random Seed [-]

This input parameter is used in conjunction with the next parameter, RandSeed2; it tells TurbSim
how to initialize the pRNG. This random seed must be an integer between –2147483648 and
2147483647 (inclusive).

The random numbers generated by the pRNG are used to create random phases (one per
frequency per grid point per wind component) for the velocity time series. When the pRNG is
initialized in the same way (i.e., RandSeed1 and RandSeed2 are not changed), the user can
reproduce the same random phases between runs, which is useful in comparing the effects of
changes to other input parameters. Random numbers also are used to generate some default input
values and the superimposed coherent structures for the non-IEC spectral models.

RandSeed2: The Second Random Seed or pRNG to Use [-]

This input parameter indicates which of three available pRNGs to use. This value is a second
random seed or the strings “RNSNLW” or “RanLux.” Using RanLux is recommended because
initial tests show that it seems to be the best behaved of the three generators.

If RandSeed2 is a random seed, it must be an integer between –2147483648 and 2147483647

(inclusive). TurbSim then uses the two seeds to initialize the intrinsic pRNG—which uses two
separate congruential generators together to produce a period of about 1018. This intrinsic pRNG
is based on an algorithm developed by Pierre L’ecuyer [18] and is identical to the one found in
SNwind.

If RandSeed2 is the string “RNSNLW,” TurbSim generates random numbers using the algorithm
found in SNLWIND and SNLWIND-3D. It is initialized with only one seed.

If RandSeed2 is the string “RanLux,” TurbSim uses Lüscher’s level 3 “Luxury Pseudorandom
Numbers” [19], [20]. This pRNG is based on a subtract-and-borrow algorithm with a period on
the order of about 10171 and is modified by throwing numbers away to destroy correlations. This
pRNG is initialized with only one seed.

WrBHHTP: Write Binary Hub-Height Turbulence Parameters? [T/F]

This parameter must be either “true” or “false.” Setting this output option to “true” generates a
binary file with a “.bin” extension. The file contains time series of wind data and turbulence
parameters from the center grid point at hub height (HH).

WrFHHTP: Write Formatted Hub-Height Turbulence Parameters? [T/F]

This “true” or “false” parameter is similar to the previous parameter, WrBHHTP. When set to
“true,” it generates a file containing time series of wind data and turbulence parameters from the
center grid point at hub height. This file is formatted (i.e., human-readable text) and has a “.dat”
extension.

7
WrADHH: Write AeroDyn Hub-Height Files? [T/F]
This “true” or “false” parameter provides an option to generate time series in the AeroDyn v13
hub-height format; this format is called “uniform wind” in InflowWind. These files have an
“.hh” extension.

WrADFF: Write AeroDyn Full-Field Files? [T/F]

This parameter must be either “true” or “false.” It provides the option to generate binary, full-
field (FF) time series in a format designed to be read by InflowWind (formerly AeroDyn). This
format is unique to TurbSim and provides an encoding scheme designed to give maximum
resolution. It contains all the information necessary to decode the time series in one file, but it is
not compatible with GH Bladed. These files have a “.bts” extension.

WrBLFF: Write Bladed-Style Full-Field Files? [T/F]

Like the WrADFF parameter, the WrBLFF “true” or “false” parameter also provides the option
to generate binary, full-field (FF) time series in a format designed to be read by InflowWind
(formerly AeroDyn). This format is compatible with GH Bladed, but in some cases the encoding
scheme provides less resolution than the files generated by WrADFF. These files have a “.wnd”
extension.

WrADTWR: Generate AeroDyn Tower Points? [T/F]

The WrADTWR “true” or “false” parameter determines whether TurbSim generates binary tower
time series, which contain points in a line at the tower centerline from the bottom of the
rectangular grid to the ground. Figure 4 shows an example of the location of tower points. If a
user chooses to output FF time series in the format unique to TurbSim using WrADFF (“.bts”
files), these tower points are added to the FF binary output file. Otherwise, a separate tower-
points binary file is created using an encoding scheme similar to the Bladed-style FF file format
used in the “.wnd” files (generated with parameter WRBLFF). This extra binary file for the tower
points has a “.twr” extension.

WrFMTFF: Write Formatted Full-Field Files? [T/F]

This “true” or “false” output option determines whether TurbSim generates FF time series in
SNLWIND-3D human-readable format. One file is generated for each wind component, and the
three files have extensions “.u,” “.v,” and “.w,” respectively. Please note that InflowWind cannot
read these text files.

WrACT: Write Coherent Turbulence Files? [T/F]

This “true” or “false” parameter indicates whether coherent
turbulence should be generated, creating time-step files in AeroDyn
v13 format. The coherent turbulence feature is not available with the
IEC or TIDAL spectral models. For the other spectral models, the
feature is available only when the gradient Richardson number (input
parameter RICH_NO) is greater than –0.05. The mean wind speed at
the top of the coherent structure also must be greater than the mean
wind speed at the bottom of the coherent structure. The Coherent Figure 4. Example of
tower points below a
rectangular grid

8
Turbulence Scaling Parameters section of the input file discusses how to set the coherent
structure location.

The coherent turbulence time-step files, which have a “.cts” extension, are intended to be
superimposed on background FF turbulence files. As a result, TurbSim also creates binary FF
time series (WrBLFF or WrADFF) when a coherent turbulence time-step file is requested. If no
FF time series format has been specified, TurbSim creates a GH Bladed-style binary FF file
(WrBLFF “.wnd” file). For more information on using these “.cts” files, see the Using Coherent
Turbulence Time-Step Files with AeroDyn v13 section in this guide.

Clockwise: Does the Turbine Rotate Clockwise? [T/F]

This true” or “false” parameter is a flag to indicate whether the turbine rotates in a clockwise
direction when looking downwind. This feature determines the order in which the horizontal grid
points of the Bladed-style FF files are written (the parameter WrBLFF must be “true”). Because
InflowWind also reads the Bladed-style FF files based on the direction of rotation, this flag does
not affect the results when used with InflowWind. This parameter probably is useful only for
comparing FF results between older versions of Bladed and InflowWind-based simulators.

ScaleIEC: Scale IEC Turbulence to Exact Standard Deviations? [0, 1, or 2]

The ScaleIEC parameter is a switch to tell how to scale the time-domain velocity output of the
IEC spectral models and is applicable to only the IECKAI and IECVKM spectral models. For
numerical reasons, the turbulence intensity (TI) of the IEC spectral models—without this
scaling—usually is slightly less than the specified value. Increasing the time series length and/or
decreasing the size of the time step results in values closer to the specified TI. Different random
seeds produce a Gaussian distribution of TI in the longitudinal wind component, due to the
spatial coherence. To get the exact specified value of TI, the time series are multiplied by a
scaling factor determined by the ratio of the target to the actual calculated standard deviation.

When ScaleIEC is set to “0,” no scaling takes place in the time domain. The result is the
variation in TI discussed above. When the ScaleIEC switch has a value of “1,” the time series at
each simulated point use the same scaling factor with a different factor for each wind component.
Those three scaling factors (one each for u, v, and w) are determined so that the standard
deviations in wind speed (and thus TI) at the hub point are the exact value specified for the
AnalysisTime-length time series that is generated. The TI at the other simulated points will vary.
When ScaleIEC is “2,” the time series at each simulated point in space is scaled independently
(i.e., each point and each component has its own scaling factor) so that the TI is the exact
specified value at each point. This scaling method alters the coherence between points. Table 3
summarizes the valid input values.

9
 Table 3. Valid ScaleIEC Values

Input Value Description

0 No scaling: time series will remain as generated.
Scaling by HH value: all time series will be modified, using the same scaling
1 factor for each point (each component has separate scale). The hub point will
have the exact specified TI; other points will not.
Independent scaling: all time series will be modified independently; scaling
2
factors vary by point and component. Each point will have the exact specified TI.

Turbine/Model Specifications
The Turbine/Model Specifications section of the TurbSim input file determines the size and
shape of the grid where time series is generated. It also determines the time/frequency content of
the resulting time series and sets the mean flow angles. Appendix C contains a flow chart
showing the function of the input parameters from this section.

NumGrid_Z: Number of Vertical Grid Points [-]

This input parameter is the number of grid points to generate in the vertical direction. It must be
an integer greater than 1. Unlike SNwind, which accepted only even numbers, TurbSim allows
both even and odd grid-point sizes. TurbSim always generates a point at the hub, regardless of
where the other grid points are located. (Note that this “extra” hub point is not contained the in
binary FF files generated when WrBLFF or WrADFF are set.)

NumGrid_Y: Number of Horizontal Grid Points [-]

This parameter indicates the number of grid points in the horizontal direction, and it must be an
integer greater than 1. If NumGrid_Y is an odd number, points fall along the undeflected tower
centerline.

TimeStep: Time Step [s]

The TimeStep parameter is the time step in seconds (i.e., ∆t ). It is set to 0.05 seconds in the
sample input files, and that value is recommended for most simulations. The time step
determines the maximum frequency, f max , used in the computation of the inverse FFT:

f max = 1 (1)
∆t
AnalysisTime: Length of Analysis [s]
The AnalysisTime parameter is the length in seconds of the data to be analyzed (i.e., tmax). This
number dictates the frequencies which are used to generate the output time series. The following
equations relate AnalysisTime to the frequency, f, and the number of frequencies, NumFreq:

1
∆f = (2)
AnalysisTime

10
 AnalysisTime
NumFreq = (3)
TimeStep

It is recommended that AnalysisTime be at least 600 seconds. To speed up the inverse FFT
computations, TurbSim might add a few extra time steps to ensure that the number of analysis
time steps is a product of small prime numbers. Extra time steps also are added if the length of
the output time series is less than the AnalysisTime (see the discussion of the UsableTime
parameter below).

UsableTime: Usable Time Series Length [s]

This parameter is the usable length (in seconds) of the data to output. This number differs
slightly from the actual amount of data that TurbSim outputs. Because InflowWind requires that
there be data both upwind and downwind of the tower in case the turbine is yawed, it mandates
that there be enough data in the FF files to shift the data to ensure that the turbine resides entirely
within the wind-data domain. When UsableTime is input as a number, TurbSim adds the amount
of time equal to the grid width divided by the mean HH wind speed, u hub , to the requested
amount of usable time:

GridWidth
OutputTime
= UsableTime + (4)
u hub

The analysis time must be at least as large as the output time:

AnalysisTime ≥ OutputTime (5)

If necessary, TurbSim increases AnalysisTime to satisfy this relationship.

To output all of the AnalysisTime time series without adding the additional GridWidth u hub
output time, enter the string “ALL”. When UsableTime is “ALL”, the output file is periodic in
time (the period is equal to the AnalysisTime), which allows the wind file to be repeated for the
extra amount of time required for startup transitions and the small amount of time required in
case the turbine is yawed.

HubHt: Turbine Hub Height [m]

The HubHt parameter is hub height of the turbine for which the inflow is being generated.
TurbSim uses the metric system so enter the value in meters. This parameter is used as a
reference height for determining the grid location.

GridHeight: Height of the Grid [m]

This parameter is the distance (in meters) between the top and bottom of the grid. The top of the
grid is assumed to be aligned with the top of the rotor disk (see Figure 5), and because all points
of the grid must be above ground level, 12 GridHeight < HubHt .

When choosing a value for GridHeight, keep in mind that InflowWind does not allow any part of
the blade—including all system displacements—to lie outside the FF grid. The grid height must

11
be large enough to encompass the entire rotor disk of FF files. See the parameter GridWidth for
further discussion.

GridWidth: Width of the Grid [m]

This parameter is the width of the grid in meters. The rotor is assumed to be centered
horizontally on the grid. If you are generating FF files for FAST simulations, the grid width—
like the height—must be large enough to ensure that no part of the blade lies outside the grid,
even when the system is displaced.

TurbSim assumes that the diameter of the rotor disk is the smaller of the GridHeight and
GridWidth values. Because InflowWind must interpolate within the grid for any point at which it
needs wind speeds (i.e., InflowWind cannot extrapolate), GridHeight and GridWidth should be
larger than the rotor diameter. In fact, AeroDyn warns users if the grid width and height are not
at least 10% larger than the rotor diameter. For turbines that move a lot during simulation (e.g.,
floating wind turbines), the grid might have to be even larger.

As pictured in Figure 5, the hub is in the horizontal center of the grid, and the turbine hub height
plus assumed rotor radius determines the top of the grid.

Height = Width Height > Width Height < Width

Figure 5. Example grid and rotor placements: the circles pictured here are the rotor diameters
assumed by TurbSim; the actual rotor diameter(s) will be smaller than pictured

VFlowAng: Mean Vertical Flow Angle [°]

This parameter is the mean vertical angle of the wind, which is constant across the entire grid.
Enter the angle in degrees, and do not exceed 45° in magnitude. A positive value means that the
wind is blowing uphill; a negative value indicates that the wind is blowing downhill. See
HFlowAng and Figure 6 for more details.

HFlowAng: Mean Horizontal Flow Angle [°]

This parameter is the mean horizontal (crosswise) angle of the wind in degrees. In all cases
except the GP_LLJ model, the horizontal flow angle is constant across the entire grid. For the
GP_LLJ model, which introduces direction shear with height, HFlowAng is the horizontal angle
at hub height.

12
The mean flow angles VFlowAng and HFlowAng are used to rotate the wind from its alignment
with the mean flow to the inertial reference frame. Users should be cautious, however, because
InflowWind—in its implementation of Taylor’s frozen turbulence hypothesis—marches FF grids
through the turbine along the positive X axis (or Propagation Direction) at the mean hub-height
wind speed, without regard to the flow angles used in TurbSim (see Figure 6). This could give
strange results if the mean flow angles are not small (for example, if HFlowAng = 180°, the grids
move through the turbine in the opposite direction the wind is blowing). We recommend setting
the propagation direction in InflowWind or using a yaw error in the turbine simulation rather
than using the HFlowAng parameter and using only small angles (e.g., less than 10°) for
VFlowAng.

13
 Figure 6. Example of TurbSim grids as implemented in InflowWind: (a) The inertial frame
coordinate systems and planes “marching” along positive X, regardless of flow angles, (b) wind
field with both flow angles 0°, (c) the same wind field with VFlowAng = 8° and HFlowAng = 15°

14
Meteorological Boundary Conditions
The Meteorological Boundary Conditions section of the TurbSim input file sets the spectral
model to simulate, determines the mean wind speeds, and sets the boundary conditions for the
spectral models defined in the IEC standards. Appendix C contains flow charts showing the
function of the input parameters from this section.

TurbModel: Turbulence Model [-]

The TurbModel parameter tells TurbSim which spectral model it should use. Enter the
six-character input value of the desired spectral model. Valid values are found in Table 4. For
more information on these models, see the Spectral Models section in this document.

UserFile: Name of the User Spectra or Time-Series Input File [-]

The UserFile parameter tells TurbSim the name of the input file that contains user-defined
spectra inputs or user time-series input data. If UserFile contains a relative path, that path will be
considered relative to the primary TurbSim input file. UserFile is unused if TurbModel is not
TIMESR or USRINP. See sections “Input File for User-Defined Spectra” and “Input File for
User-Defined Time Series” (below) for details on the formats of these input files.

Table 4. Valid TurbSim Turbulence (Spectral) Models

TurbModel Input Value Description

API API model for hurricane winds

GP_LLJ NREL Great Plains low-level jet
IECKAI IEC Kaimal
IECVKM IEC von Karman

NONE Steady winds (used only for testing)

NWTCUP NREL National Wind Technology Center
SMOOTH Risø smooth terrain
TIDAL Tidal channel turbulence model (water)

TIMESR User defined using time-series data

USRINP User defined using velocity spectra
USRVKM von Karman model with user defined scaling profiles

WF_07D NREL wind farm: 7 rotor-diameters downwind

WF_14D NREL wind farm: 14 rotor-diameters downwind
WF_UPW NREL wind farm: upwind

IECstandard: IEC Standard [-]

This input parameter tells TurbSim which IEC standard to use. Enter “1” to use the scaling from
the IEC 61400-1 [21] standard or enter “2” or “3” to use the scaling from the IEC 61400-2 (small

15
wind turbine) [22] or -3 (offshore wind turbine) [23] standards. To use the scaling parameters
from the second edition of the IEC 61400-1 standard [24], follow the input with the string
“-ED2” (i.e., “1-ED2”). Likewise, to use the scaling parameters from IEC 61400-1, 3rd ed. [21],
input the string “1-ED3”. If the 61400-1 edition number is not specified, TurbSim uses the
scaling from the third edition of IEC 61400-1 for the Kaimal model and scaling from the
standard’s second edition for the von Karman model (which is not defined in the newer edition).
This input parameter is used only if the spectral model is IECKAI or IECVKM.

IECturbc: IEC Turbulence [%]

The IECturbc parameter tells TurbSim what turbulence intensity you want to use with the IEC
Kaimal or von Karman spectral models. Input values of “A,” “B,” or “C” correspond to the
standard IEC categories of turbulence characteristics, with “A” being the most turbulent. Figure
7 contains the relationship between wind speed and standard deviation for the standard IEC
categories and turbulence types. You can also specify the TI in percent instead of choosing the
turbulence categories. In this case, the standard deviation of the longitudinal wind speed, σ 1 , is
calculated using the following equation:

IECturbc
σ1 = u hub (6)
100
If you use the NWTCUP spectral model and enter the string “KHTEST” for the IECturbc
parameter, TurbSim creates a test wind field that can be used to see the effects of a KH billow.
With this test function, TurbSim overrides the inputs for Richardson number (0.02); power-law
coefficient (0.3); and billow type, size, and location. An LES-type billow centered on the rotor
disk is scaled so that the billow achieves a bandwidth of at least 25 Hz and so that the expected
maximum coherent turbulent kinetic energy (CTKE), defined as

( u′w′) + ( u′v′) + ( v′w′)

2 2 2
CTKE
= 1
2 (7)

is at least 30 m2/s2. This billow lasts at least half of the usable length of the output time series,
and starts a quarter of the way through the time series. An example of KHTEST is presented in
Figure 8.

The IECturbc parameter is not used for any other spectral model.

16
 Figure 7. Longitudinal wind-speed standard deviation and TI for IEC turbulence categories as
functions of the mean hub-height wind speed, Vhub

Figure 8. Coherent turbulent kinetic energy (CTKE) of an example simulation using KHTEST:
the coherent structure placement in the middle of the time series is shown by the red lines

IEC_WindType: IEC Turbulence Model [-]

This parameter indicates which IEC wind model will be used. Valid entries, which are found in
Table 5, include the Normal Turbulence Model (NTM), Extreme Turbulence Model (ETM), and
Extreme Wind Speed Model (EWM) using the 10-minute average wind speed with a recurrence
period of 1 year or 50 years. Note that the EWM scaling parameters in TurbSim are valid only
for 10-minute simulations. The definitions of these models and of the wind turbine classes can be
found in the IEC 61400-1 standard (3rd ed.) [21]. If the IECturbc parameter was specified as a
percentage instead of as a standard turbulence category, the wind model must be “NTM.” This
input is used only with the IEC spectral models.

17
 Table 5. Valid IEC Turbulence Models

IEC_WindType Description
NTM Normal Turbulence Model
1ETM Class I Extreme Turbulence Model
2ETM Class II Extreme Turbulence Model
3ETM Class III Extreme Turbulence Model
1EWM1 Class I turbulent Extreme Wind Speed Model, 1-yr recurrence
2EWM1 Class II turbulent Extreme Wind Speed Model, 1-yr recurrence
3EWM1 Class III turbulent Extreme Wind Speed Model, 1-yr recurrence
1EWM50 Class I turbulent Extreme Wind Speed Model, 50-yr recurrence
2EWM50 Class II turbulent Extreme Wind Speed Model, 50-yr recurrence
3EWM50 Class III turbulent Extreme Wind Speed Model, 50-yr recurrence

ETMc: Extreme Turbulence Model Parameter c [m/s]

The ETMc input parameter is the value of the variable c in the equation for the longitudinal
component standard deviation, σ 1 , in the ETM (see Eq. 19 in section 6.3.2.3 of IEC 61400-1 3rd
ed. [21]):

 V  u  
=σ 1 c I ref  0.072  ave + 3   hub − 4  + 10  (8)

  c   c 




The values for the variables I ref and Vave —defined respectively as the expected value of
turbulence intensity and 20% of the reference wind-speed average—are determined by the wind
turbine class. Enter a value for c in meters per second, or enter “default” for TurbSim to use
c = 2 m/s, as defined in the standard. This parameter is used only with the Extreme Turbulence
Model (i.e., when IEC_WindType = xETM).

WindProfileType: Type of Wind Profile [-]

The WindProfileType parameter tells TurbSim how to calculate the mean wind profile. Valid
entries are found in Table 6. Users can enter the string “default” here for TurbSim to pick a wind
profile based on the spectral model. The default values are listed in Table 7. Please see the
Velocity and Direction Profiles section of this guide for more details about the different wind
profile types. The “H2L” velocity profile must always be used with, and only with, the TIDAL
spectral model.

18
 Table 6. Valid WindProfileType Values

WindProfileType Description
PL Power-law wind profile
Diabatic (logarithmic) wind profile; not valid with KHTEST or with EWM or ETM
LOG
wind types
Logarithmic velocity profile for MHK models (TIDAL). Not valid with other values of
H2L
TurbModel.
JET Low-level jet wind profile, valid only with GP_LLJ model
IEC Power-law profile on the rotor disk; logarithmic profile elsewhere
API API (Frøya) wind profile, based on 1-hr mean wind speed at 10 m above sea level
User-defined velocity profile, values are read from a table in ProfileFile and
USR
interpolated.
TS Valid only with the “TIMESR” TurbModel
Default Uses a default. See Table 7.

Table 7. Default Inputs for WindProfileType

TurbModel Default WindProfileType
GP_LLJ JET
TIDAL H2L
API API
TIMESR TS when time series for more than one point are entered; PL otherwise
All other models IEC

ProfileFile: Name of the file containing input profiles [-]

The ProfileFile parameter tells TurbSim the name of the input file that contains user-defined
profiles for wind speed and direction when WindProfileType is “USR” and for standard deviation
and length-scale profiles when TurbModel is “USRVKM”. If ProfileFile contains a relative path,
that path will be considered relative to the primary TurbSim input file. See the “Input file for
User-Defined Profiles” section of this document for further details.

RefHt: Reference Height [m]

The RefHt parameter specifies the height (in meters) of the corresponding reference wind speed
(parameter URef). This parameter enables users to specify the mean wind speed at a height other
than the hub height. TurbSim uses this reference height and wind speed with the wind profile
type to calculate the HH mean wind speed. The reference height also is used with URef and the
surface roughness (parameter Z0) to compute default input values for parameters UStar and ZI.

URef: Reference Wind Speed [m/s]

The URef parameter is the mean streamwise wind speed at the reference height. It is the mean
value over the entire AnalysisTime length of the simulation of the u-component wind speed;
however, when the API spectral model is used, URef must be the 1-hr mean wind speed
regardless of the value of AnalysisTime. URef must be a positive value in units of meters per

19
second. This value is not used for any of the extreme wind speed models (EWMs). If you are
using the “USR” or “TS” WindProfileType, the URef input is ignored and is instead calculated at
the RefHt using the profile data contained in ProfileFile.

If you use “JET” for the WindProfileType parameter, you can enter the string “default” here for
TurbSim to calculate a default wind speed in two steps: (1) TurbSim calculates the maximum
speed of the jet wind profile, u JetMax , based on the jet height, ZJetMax, and a random variate
(shown in Figure 9) then (2) it calculates the wind speed at RefHt using u JetMax along with
parameters ZJetMax, RICH_NO, and UStar. The calculations of the low-level jet wind speed
profile are discussed further in the Velocity and Direction Profiles section of this guide.

If you are using the “TIMESR” turbulence model, you can also enter “default” for URef. Doing
so will set URef equal to the mean wind speed calculated at the RefPtID point (specified in the
input file for user-defined time series). It will also override RefHt to be the height of the RefPtID
point.

ZJetMax: Height of the Jet [m]

The ZJetMax parameter is the height in meters of the low-level jet. Enter the approximate height
at which the low-level jet wind profile reaches its maximum wind speed, or enter the string
“default” to have TurbSim calculate a jet height. The default height is a function of parameters
RICH_NO and Ustar with a random component based on LLLJP measurements. The default
height—without the random component—is plotted in Figure 10. ZJetMax, which must be a
value between 70 m and 490 m, is used to calculate the mean wind speed and direction profiles.
It is used only when WindProfileType is “JET.”

Figure 9. Default jet wind speed for URef calculation: Figure 10. Default jet height, ZJetMax,
error bars indicate the range of random variate, N; without random variates (the random
dotted lines mark the tenth and ninetieth percentiles variation range is approximately ±50 m)

20
PLExp: Power-Law Exponent [-]
The PLExp parameter is used to compute the mean u-component wind speeds across the rotor
disk when WindProfileType is “IEC” or “PL.” It is the exponent used to define the power-law
wind profile,
PLExp
 z 
u ( z ) = u hub   (9)
 HubHt 
where z is the height above ground level. The exponent can be positive, negative, or zero (for no
shear). Enter the string “default” to have TurbSim use a default value based on the specified
spectral model, as shown in Table 8. If KHTEST is specified for parameter IECturbc, the PLExp
parameter is overwritten to 0.3.

Z0: Surface Roughness Length [m]

The surface roughness length, Z0, is the last parameter in this section. This length—a measure of
the roughness of the surface terrain—is the extrapolated height at which the mean wind speed
becomes zero in a neutral atmosphere, assuming a logarithmic vertical wind profile:

ln ( z Z0 )
u ( z ) = URef (10)
ln ( RefHt Z0 )

Enter the length in meters, or enter the string “default” to have TurbSim use a default value
based on the specified spectral model. The default values are listed in Table 8. This parameter is
not used for the TIDAL spectral model.

Table 8. Default Inputs for Meteorological Boundary Conditions

TurbModel PLExp Z0 (m)

0.11 for EWM

IECKAI, IECVKM, API 0.14 for offshore (61400-3) NTM, 0.03
0.2 otherwise
TIDAL, TIMESR 0.143 0.03
SMOOTH 0.143 0.01
GP_LLJ 0.143 0.005
0.08-0.15, increasing with RICH_NO,
NWTCUP 0.021
0.3 for KHTEST option
WF_UPW same as NWTCUP 0.018
WF_07D 0.13-0.18, increasing with RICH_NO 0.064
WF_14D same as WF_07D 0.233

21
Non-IEC Meteorological Boundary Conditions
If you have specified either the Kaimal or von Karman spectral model, TurbSim does not use the
values in this section of the input file. The other (non-IEC) spectral models may require the
additional meteorological boundary conditions contained in this section. All of the inputs in this
section, with the exception of the gradient Richardson number, can be replaced with the string
“default.” Appendix C contains flow charts showing the function of the input parameters from
this section and how the default values are chosen.

Latitude: Site Latitude [°]

The first parameter in this section is the site latitude in degrees. The latitude is used only to
calculate a Coriolis term in the default mixing layer depth (parameter ZI). The magnitude of the
latitude must be between 5° and 90°; the default value is 45°.

RICH_NO: Gradient Richardson Number [-]

The RICH_NO parameter is the turbine-layer vertical stability given by the dimensionless
gradient Richardson number, which is defined as

g ∂θ
RICH _ NO = θ ∂z2 (11)
 ∂u 
 
 ∂z 

In this equation, g is the gravitational acceleration, z is the height above ground, and u is the
wind speed. The variable θ represents potential temperature, which is calculated using the mean
absolute air temperature, T, and atmospheric pressure, p:
0.286
 1000 
θ =T  (12)
 p 

The RICH_NO parameter is used to calculate the velocity spectra and the JET and LOG wind
profiles, scale coherent structures, and determine default values for many input parameters.

Enter zero for neutral conditions, a negative value for unstable conditions, or a positive number
for stable atmospheric conditions. The GP_LLJ and NWTCUP models limit this input to
−1 ≤ RICH _ NO ≤ 1 . If “KHTEST” is specified for parameter IECturbc, the RICH_NO
parameter is overwritten to 0.02. The RICH_NO parameter does not accept the value “default.”
The RICH_NO parameter is not used for the TIDAL spectral model.

UStar: Rotor-Disk Average Friction Velocity [m/s]

The parameter UStar is the friction or shear velocity, u* , averaged over the rotor disk:

UStar = u*
np
(13)
= 1
np ∑
i =1
u 'w'
i

22
where the prime quantities indicate the fluctuating (zero-mean) longitudinal (u) and vertical (w)
wind components at np measurement points on the rotor disk. The GP_LLJ model, which scales
the velocity spectra with local friction velocities ( u* values varying with height), assumes that
UStar is the average friction velocity of three points on the u* profile: one at the hub, one at the
top of the rotor, and one at the bottom of the rotor.

UStar is used to scale the velocity spectra of non-IEC spectral models, to scale the JET and H2L
mean velocity profiles, and to calculate the default values of many input parameters. Enter UStar
in units of meters per second or enter “default” to have TurbSim calculate an appropriate value.
The default value for the TIDAL model is UStar = 0.05 URef. For non-hydro spectral models
(i.e., all but TIDAL), the default value is calculated using the diabatic u*0 (near the surface),
which is predicted by Panofsky and Dutton’s modified logarithmic profile [25] using

0.4 URef
u*0 = (14)
(
ln RefHt
Z0 ) − Ψ M ( RICH _ NO )

where Ψ M is a function that depends on the RICH_NO stability parameter. The relationship
between RICH_NO and u*0 , normalized by URef, at RefHt = 80 m is plotted in Figure 11. The
relationship between u*0 and the default Ustar is shown in Figure 12.

If “default” is entered for the reference wind speed, Uref, the string “default” cannot be entered
for the UStar parameter, because the default values for the two parameters are interdependent.

Figure 11. Diabatic friction velocity, u*0 , Figure 12. Default UStar as a function of diabatic friction
normalized by URef and calculated velocity, u*0 (left: SMOOTH, GP_LLJ, and NWTCUP
using RefHt = 80 m and Z0 = 0.01 m models, right: wind farm models)

ZI: Depth of the Mixing Layer [m]

The parameter ZI is the depth of the mixing layer (in meters). This parameter scales the velocity
spectra in unstable atmospheric conditions and is not used for stable atmospheric conditions. The
default mixing layer depth is calculated using

23
  400URef
, UStar < u*0

ZI = 
log
 10 Z0
RefHt
( ) (15)
 UStar
, UStar ≥ u*0
12.0ω sin ( π Latitude )
 180

where
= ω 7.292116 × 10−5 rad/s is the Earth’s angular speed of rotation. This equation combines
the work of Dutton et al. [26] with the ESDU [27]. This parameter is not used for the TIDAL,
TIMESR, or USRINP spectral models.

PC_UW: Average u′w′ Reynolds Stress at the Hub [m2/s2]

The PC_UW parameter is the desired average u′w′ Reynolds stress (in m2/s2) at the simulated
hub point. It is used in conjunction with the next two inputs, parameters PC_UV and PC_VW, to
create some correlation between the wind-speed components.

TurbSim modifies the v- and w-component wind speeds (for non-IEC models only) by
computing a linear combination of the time series of the three independent wind-speed
components to obtain the mean Reynolds stresses PC_UW, PC_UV, and PC_VW at the hub
point. The linear combinations are computed for each point, j, using the equations

u ′j , correlated = u ′j ,independent
v′j , correlated= a uv u ′j ,independent + v′j ,independent + a vw w′j , independent (16)
w′j , correlated a uwu ′j ,independent + w′j ,independent
=

The three α variables are coefficients chosen to generate the desired Reynolds stresses for the
correlated wind components at the hub:

′ ,correlated whub
PC _ UW = uhub ′ ,correlated
′ ,correlated vhub
PC _ UV = uhub ′ ,correlated (17)
′ ,correlated whub
PC _ VW = vhub ′ ,correlated

Because this method affects the frequency domain somewhat, we have placed the following limit
on the coefficients: α ≤ 1 . This limit can cause the actual hub Reynolds stresses to differ from
the desired values.

Enter the string “default” for TurbSim to compute an appropriate Reynolds stress for PC_UW.
The default value for the TIDAL model is PC _ UW = −UStar 2 (1 − HubHt RefHt ) The default
value for the SMOOTH model is the same as that for the WF_UPW and WF_07D models:
PC _ UW = −UStar 2 . The default value for the WF_14D model has the same magnitude as the
SMOOTH model, but is positive 1% of the time (randomly). The magnitudes of the defaults for
the NWTCUP and GP_LLJ models are functions of UStar, RICH_NO, height, mean hub-height
wind speed, and shear across the rotor disk. The signs of the defaults are determined randomly,
with the probability that PC_UW is negative increasing with the magnitude of the default. Users

24
can also enter the string “none” to set α uw = 0 and disable the correlation between the u and w
components. The USRINP and TIMESR models use “none” for their default values.

PC_UV: Average u′v′ Reynolds Stress at the Hub [m2/s2]

The PC_UV parameter is the desired average u′v′ Reynolds stress (in m2/s2) at the simulated hub
point. It is used in conjunction with the parameters PC_UW and PC_VW to create cross-
component correlation. See the discussion after parameter PC_UW for details of the correlation.

To set α uv = 0 and disable the correlation between the u and v components, enter the string
“none.” Users also can enter the string “default” if you would like TurbSim to compute a default
value for PC_UV. The magnitudes of the defaults for site-specific models (GP_LLJ, NWTCUP,
WF_UPW, WF_07D, and WF_14D) are functions of UStar, RICH_NO, height, mean hub-height
wind speed, and shear across the rotor disk. The signs of the defaults are determined randomly.
The default for the SMOOTH and TIDAL models is “none.”

PC_VW: Average v′w′ Reynolds Stress at the Hub [m2/s2]

The PC_VW parameter is the desired average v′w′ Reynolds stress (in m2/s2) at the simulated hub
point. It is used in conjunction with the parameters PC_UW and PC_UV to create cross-
component correlation. See the discussion after parameter PC_UW for details.

Users can enter the string “none” to set α vw = 0 and disable the correlation between the v and w
components. To have TurbSim compute a default value for PC_VW, enter the string “default.”
The magnitudes of the defaults for site-specific models are functions of UStar, RICH_NO,
height, mean hub-height wind speed, and shear across the rotor disk. The signs of the defaults are
determined randomly. The default for the SMOOTH and TIDAL models is “none.”

Spatial Coherence Parameters

This section of the input file tells TurbSim how the spatial coherence should be modeled. See the
“Spatial Coherence Models” section of this document for details on the four available spatial
coherence models.

SCMod1: Spatial Coherence Model for the u-Component Velocity [-]

The SCMod1 parameter tells TurbSim what coherence model to use for the u-component wind
speed. Valid values are “GENERAL”, “IEC”, “API”, “NONE”, or “default”. The default values
are listed in Table 9.

SCMod2: Spatial Coherence Model for the v-Component Velocity [-]

The SCMod2 parameter tells TurbSim what coherence model to use for the v-component wind
speed. Valid values are “GENERAL”, “IEC”, “NONE”, or “default”. The default values are
listed in Table 9.

SCMod3: Spatial Coherence Model for the w-Component Wind Speed [-]
The SCMod3 parameter tells TurbSim what coherence model to use for the w-component wind
speed. Valid values are “GENERAL”, “IEC”, “NONE”, or “default”. The default values are
listed in Table 9.

25
 Table 9. Default Inputs for SCMod1, SCMod2, and SCMod3

TurbModel SCMod1 SCMod2 SCMod3

IECKAI, IECVKM,
IEC None None
USRVKM
API API None None
USRINP General None None
All other models General General General

IncDec1: Spatial Coherence for the u-Component Wind Speed [-, m-1]
The IncDec1 parameter defines the spatial coherence decrement, a, and offset parameter, b, for
the u-component wind speed (K = u) when SCMod1 is GENERAL or IEC. These two values are
used to define the degree of spatial coherence between points on the grid using the definition

 CohExp 2 
 r   f r
Cohi , j = 
exp −aK    u  + ( bK r ) 2
 (18)

  zm    


where r is the distance between points i and j, f is the cyclic frequency, CohExp is the coherence
exponent input parameter (which is 0 for the IEC coherence model), zm is the mean height of
points i and j, and u is the mean wind speed, defined differently for the two spatial coherence
models. Please see the Spatial Coherence Models section of this document for more information.

The IncDec1 decrement, a, must be a positive number. Users can enter “default” or both the a
and b coherence parameters in quotation marks on the same line. For example, “10.0 0.1E-02”
uses a coherence decrement of a = 10.0 and an offset parameter of b = 0.1E-02 m-1. Omitting the
quotation marks around the two input parameters causes TurbSim to use b = 0.

The default a parameter for the u-component is au = u hub for the SMOOTH, USRINP, TIMESR,
and TIDAL models. For the API and IEC models, the default is au = 8.8 for IEC 61400-1 2nd ed.
and au = 12 for IEC 61400-1 3rd ed. The other non-IEC models base this default value on
measured vertical coherence spectra from their respective datasets. The default a parameter for
these models is a function of HubHt and RICH_NO parameters, as well as the mean hub-height
wind speed.

The default b parameter for the u-component is bu = 0 for the SMOOTH, USRINP, TIMESR,
and TIDAL models. For the API and IEC models, the default is

0.12
bu = (19)
2.45 min ( 30m, HubHt )

for IEC 61400-1 2nd ed. and

26
 0.12
bu = (20)
5.67 min ( 60m, HubHt )

for IEC 61400-1 3rd ed. (the function min ( ) denotes the minimum of the two values). The other
non-IEC models calculate the default b parameter as a function of mean hub-height wind speed.
The GP_LLJ and NWTCUP models also use the RICH_NO parameter to calculate the default b.
Figure 13 shows the default parameters for neutral conditions (i.e., RICH_NO = 0) using a value
of 80 m for the HubHt parameter.

Figure 13. Default u-component coherence parameters, IncDec1, ( au left, bu right) as a function
of wind speed, using RICH_NO = 0 and HubHt = 80 m (IEC values are plotted for comparison)

IncDec2: Spatial Coherence for the v-Component Wind Speed [-, m-1]
The IncDec2 parameter defines the spatial coherence decrement, a, and offset parameter, b, for
the v-component wind speed using the coherence definition of Eq. (18) (with K = v). Users can
enter “default” for TurbSim to pick appropriate values for both a and b, or enter both a and b
parameters in one set of quotation marks on the same line. See the discussion for IncDec1
(above) for further details.

The default a parameter for the v-component is av = 0.75 u hub for the SMOOTH, USRINP,
TIMESR, and TIDAL models. For the API and IEC models, the default av is a very large
number (effectively making this the identity coherence model). The other non-IEC models
calculate the default value as a function of HubHt and RICH_NO parameters, as well as the mean
hub-height wind speed.

The default b parameter for the v-component is bv = 0 for the SMOOTH, USRINP, TIMESR,
TIDAL, API, and IEC models. The other non-IEC models calculate the default b parameter as a
function of mean hub-height wind speed. The GP_LLJ and NWTCUP models also use the

27
RICH_NO parameter to calculate the default b. Figure 14 shows the default parameters for
neutral conditions using an 80-m HubHt.

Figure 14. Default v-component coherence parameters, IncDec2, ( av left, bv right) as a function
of wind speed, using RICH_NO = 0 and HubHt = 80 m (IEC does not define v-component
coherence parameters)

IncDec3: Spatial Coherence for the w-Component Wind Speed [-, m-1]
The IncDec3 parameter defines the spatial coherence decrement, a, and offset parameter, b, for
the w-component wind speed using the coherence definition of Eq. (18) (with K = w). Users can
enter “default” for TurbSim to pick appropriate values for both a and b, or enter both a and b
parameters in one set of quotes on the same line. See the discussion for IncDec1 for further
details.

The default a parameter for the w-component is aw = 0.75 u hub for the SMOOTH, USRINP,
TIMESR, and TIDAL models and aw = 0.4 au (using the default au , not the value entered in
IncDec1) for the three wind farm models (WF_UPW, WF_07D, and WF_14D). The GP_LLJ
and NWTCUP models calculate the default value as a function of the HubHt and RICH_NO
parameters, as well as the mean hub-height wind speed. For the API and IEC models, the default
aw is a very large number (effectively making this the identity coherence model).

The default b parameter for the w-component is bw = 0 for the SMOOTH, USRINP, TIMESR,
TIDAL, API, and IEC models and bw = 10 bu for the three wind farm models. The GP_LLJ and
NWTCUP models calculate the default b parameter as a function of mean hub-height wind speed
and the RICH_NO parameter. Figure 15 shows the default parameters for neutral conditions
using an 80-m HubHt.

28
 Figure 15. Default w-component coherence parameters, IncDec3, ( aw left, bw right) as a
function of wind speed, using RICH_NO = 0 and HubHt = 80 m (IEC does not define w-
component coherence parameters)

CohExp: Coherence Exponent [-]

The CohExp parameter is the exponent in the general coherence definition of Eq. (18). The same
value of CohExp is used for all three wind components (if they are using the GENERAL
coherence model); enter a non-negative number or “default” to use the default value of 0.

Coherent Turbulence Scaling Parameters

The coherent turbulence scaling parameters found in this section are used with non-IEC
atmospheric (non-hydro) spectral models when the gradient Richardson number (RICH_NO) is
greater than –0.05 and the option to create coherent turbulence time-step files has been selected
(i.e., WrACT = true). Appendix C contains a flow chart showing the functions of the input
parameters from this section.

TurbSim uses empirical values to calculate when and how coherent events—pieces (sections in
time) of a Kelvin-Helmholtz (KH) billow simulated using either direct numerical simulation
(DNS) or large-eddy simulation (LES)—should be added to the background turbulence. It
creates a coherent turbulence time-step file that AeroDyn v13 can read. The super-positioning of
coherent events on the background turbulence occurs in AeroDyn v13 in the inertial reference
frame coordinate system that AeroDyn v13 uses. The Coherent Structures section of this
document discusses this topic further.

CTEventPath: Name of Coherent Turbulence Events path [-]

The CTEventPath parameter is the name of the path that contains the coherent event definition
files. Use quotation marks around the path name. This directory should contain files named
“Events.les,” “Events.dns,” and “Events.xtm” as well as one or more files named
“Eventxxxxx.dat” (with digits replacing the xxxxx). These event definition files and the associated
binary data files that AeroDyn v13 reads are provided in the coherent structure archive on the
TurbSim Web site (in folder “EventData”).

29
CTEventFile: Type of Coherent Events [-]
This parameter tells TurbSim which type of coherent event files to use. Valid entries are found in
Table 10. In each individual simulation, all events are of the same type (either all LES or all
DNS). TurbSim automatically uses LES events when KHTEST is specified for parameter
IECturbc.

Table 10. Valid CTEventFile Entries

Input Value Description

DNS Reads DNS event files (CTEventPath\Events.dns)

LES Reads LES event files (CTEventPath\Events.les)
Random Randomly chooses between LES and DNS (equal probability)

Randomize: Randomize Size and Location of KH Billow Pieces? [T/F]

Set the Randomize parameter to “true” to randomize the size and location of the coherent
structures in the rotor disk or to “false” to specify these values yourself. A value of “true”
overrides the next three input parameters (DistScl, CTLy, and CTLz). Instead, the coherent
structures are centered laterally and randomly chosen to cover either (1) the full rotor disk (75%
of the time), (2) only the lower half of the disk (12.5% of the time), or (3) only the upper half
(12.5% of the time).

DistScl: Disturbance Scale [-]

The DistScl parameter is the disturbance scale, which determines the size of the coherent event
data set relative to the rotor disk. It is the ratio of the height of the coherent dataset to the
(assumed) rotor diameter. A value of 1.0 makes the coherent structures the height of the rotor
disk; 0.5 makes them half the height of the rotor disk. If parameter IECturbc is KHTEST,
TurbSim overrides DistScl with a value of 1.0. When Randomize is “true,” the value of this input
is overridden (as discussed above).

CTLy: Lateral Location of Coherent Turbulence [-]

This parameter laterally positions the coherent structures from the KH billow on the rotor disk.
CTLy is the fractional location of the tower centerline from the right to left side (looking
downwind) of the coherent event dataset. A value of 0.5 puts the tower centerline in the center of
the billow. The coherent structures are periodic in the lateral direction so they cover the grid
horizontally, regardless of the location of the tower centerline. Figure 16 shows coherent
structure scaling with CTLy on the abscissa. If parameter IECturbc is KHTEST, TurbSim
overrides CTLy with a value of 0.5. When Randomize is “true,” the value of this input is
overridden (see the discussion of Randomize).

CTLz: Vertical Location of Coherent Turbulence [-]

This parameter positions the coherent structures vertically on the rotor disk. CTLz is the
fractional location of hub height from the bottom of the dataset. A value of 0.5 places the vertical
center of the billow at hub height. The structures are constant above and below the top and
bottom of the dataset. Figure 16 shows how the structures are scaled. If parameter IECturbc is

30
KHTEST, TurbSim overrides CTLz with a value of 0.5. When Randomize is “true,” the value of
this input also is overridden (see the discussion of Randomize).

Figure 16. Coherent structure scaling (looking downwind): CTLy is the tower location, CTLz
is the hub (height) location, and DistScl determines the size (m) relative to the rotor disk

CTStartTime: Minimum Start Time for Coherent Turbulence [s]

The CTStartTime parameter is used to determine where the first coherent structure will be placed
in the time-step file. TurbSim ensures that the first event in the coherent time-step file does not
occur before the time entered here (in seconds). This feature can be useful if you do not want a
turbine to encounter coherent structures during the startup transient of a simulation.

Input File for User-Defined Time Series

When using the “TIMESR” turbulence model, TurbSim requires additional inputs, which are
specified in the User-Defined Time Series Input File. The name of this file is specified in the
UsrFile input parameter in the primary TurbSim input file. An example of a User-Defined Time-
Series input file is shown in Figure A-2.

nComp: Number of Velocity Components in the File [1, 2, or 3]

This parameter is used to determine what columns are in the Time Series table at the end of this
input file. If nComp is “1”, the file contains only the U-component wind speeds. If nComp is “2”,
the file contains only the U- and V- components of the total wind speed. If nComp is “3”, the file
contains all three (U, V, W) wind components.

Currently if nComp is less than 3, TurbSim generates zeros for the missing components.

nPoints: Number of Time Series Points in the File [-]

This parameter tells TurbSim how many measurement locations are stored in this file. nPoints
must be at least 1.

31
RefPtID: Index of the Reference Point [-]
The RefPtID parameter is a number between 1 and nPoints. It is the index of the measurement
point whose phase angles will be used to correlate with the simulated points (i.e., it will be the
first point used to form Veers’ [28] transformation matrix, H).

Pointyi, Pointzi: Coordinate Locations of Time Series Points in the File [m, m]
This is a table listing the coordinates of the measurement locations contained in this file. These
coordinates must be defined in the inertial reference frame with origin at the ground level and the
undisplaced tower centerline. This table has two columns: the first contains the Y (lateral)
coordinate, and the second contains the Z (vertical) coordinate. There are nPoints rows of data in
the table, one for each measurement location. The locations must be entered in order of
increasing height.

Time Series
This table lists the time series values. TurbSim uses the raw time-series data entered in this file,
so make sure you have processed the data before using it here. It is recommended that all of
these time series be rotated in the direction of RefPtID, so that the horizontal and vertical
directions at RefPtID are 0, but that any other points contain a direction profile relative to the
reference point.

The first column in the table is the elapsed time. The remaining columns are wind speeds at the
given time. The first nComp columns after the time column are the wind components (in the
order U, V, W) for the first point entered in the coordinate location table. The next nComp are for
the second point in the coordinate location table, and so forth.

The columns must be in order of the positions listed in the coordinate location table, and the time
between each entry (Δt) must be constant.

Input File for User-Defined Spectra

When using the “USRINP” turbulence model, TurbSim requires additional inputs, which are
specified in the User-Defined Spectra Input File. The name of this file is specified in the UsrFile
input parameter in the primary TurbSim input file. An example User-Defined Spectra input file
is shown in Figure A-3.

NumUSRf: Number of frequencies [-]

The NumUSRf input determines how many rows TurbSim will read from the spectra table in this
file. If the table contains more than NumUSRf rows, the additional rows are ignored, which can
give unexpected results in the generated time series.

SpecScale1: Scaling factor for the input u-component spectrum [-]

SpecScale1 is a constant that TurbSim uses to multiply the u-component PSD column in the
Spectra table of this file to obtain the desired longitudinal-component velocity spectra in given a
simulation. This scaling factor can be used—for example—to modify the target standard
deviation for the u-component without changing all of the values in the u column of the Spectra
table.

32
SpecScale2: Scaling factor for the input v-component spectrum [-]
SpecScale2 is a constant that TurbSim uses to multiply the v-component PSD column in the
Spectra table of this file to obtain the desired lateral-component velocity spectra in given a
simulation. This scaling factor can be used—for example— to modify the target standard
deviation for the v-component without changing all of the values in the v column of the Spectra
table.

SpecScale3: Scaling factor for the input w-component spectrum [-]

SpecScale3 is a constant that TurbSim uses to multiply the w-component PSD column in the
Spectra table of this file to obtain the desired vertical-component velocity spectra in given a
simulation. This scaling factor can be used—for example— to modify the target standard
deviation for the w-component without changing all of the values in the w column of the Spectra
table.

Spectra
This table lists the spectra that will be used uniformly over the simulation (Y-Z) grid. There are
four columns in this table:

1. Frequency (Hz)

2. u-component PSD (m^2/s)

3. v-component PSD (m^2/s)

4. w-component PSD (m^2/s)

The frequencies must be unique, but TurbSim will sort them into increasing order if they are not
entered that way. The “Spectral Models” section of this document describes how these input
values are used in simulation.

Input file for User-Defined Profiles

When using the “USRVKM” turbulence model or the “USR” wind profile type, TurbSim
requires additional inputs, which are specified in the User-Defined Profiles Input File. The name
of this file is specified in the ProfileFile input parameter in the primary TurbSim input file. An
example User-Defined Profile input file is shown in Figure A-4.

NumUSRz: Number of heights [-]

The NumUSRz input determines how many rows TurbSim will read from the Profile table in this
file. If the table contains more than NumUSRz rows, the additional rows are ignored, which can
give unexpected results in the generated time series.

StdScale1: Scaling factor for the input standard deviation (“USRVKM” only) [-]
StdScale1 is a constant that TurbSim uses in the “ÚSRVKM” turbulence model to scale the
standard deviation column of the Profiles table for the u-component spectra. It is ignored with all
other turbulence models.

33
StdScale2: Scaling factor for the input standard deviation [-]
StdScale2 is a constant that TurbSim uses in the “ÚSRVKM” turbulence model to scale the
standard deviation column of the Profiles table for the v-component spectra. It is ignored with all
other turbulence models.

StdScale3: Scaling factor for the input standard deviation [-]

StdScale3 is a constant that TurbSim uses in the “ÚSRVKM” turbulence model to scale the
standard deviation column of the Profiles table for the w-component spectra. It is ignored with all
other turbulence models.

Profiles
This table lists the user-defined profiles that will be used in TurbSim when using the
“USRVKM” turbulence model or the “USR” wind profile type. The columns are

1. Height (m)

2. Wind speed (m/s)

3. Wind direction (degrees, measured counter-clockwise from above)

4. Standard deviation (m/s), read and used only when TurbModel = “USRVKM”

5. Length scale (m) , read and used only when TurbModel = “USRVKM”

TurbSim will read NumUSRz rows from this table. Note that the last two columns are read only
when TurbModel = “USRVKM”.

Output Files
TurbSim can generate several different sets of output files. They have the root name of the
TurbSim input file, and their extensions indicate what type of files they are. The Runtime
Options section in the primary input file (above) describes how to tell TurbSim which sets to
output.

Summary Files
TurbSim generates a summary file for all runs. This summary file is a text file with a “.sum”
extension. The first part of the file tells you what was specified in the input file. After that,
TurbSim prints out many statistics for the run. These statistics are calculated using the entire
AnalysisTime so if a shorter UsableTime was requested, the statistics of the output time series
could be different than what is displayed in the summary file. Also keep in mind that the
turbulence statistics are for the background turbulence only; they do not include effects of any
coherent structures generated in coherent turbulence time-step files. If a coherent turbulence
time-step file is generated, TurbSim prints the number of events and the total length of those
events in the summary file. If Bladed-style FF files or separate tower output are requested,
TurbSim adds another section that tells InflowWind how to convert the normalized data to
floating-point form.

34
Hub-Height Binary Files
The hub-height binary files are in a machine-readable form designed to be read by GenPro, a
postprocessor from the National Center for Atmospheric Research (NCAR). TurbSim gives these
files a “.bin” extension. At each time step, TurbSim writes the values of a series of parameters in
the binary file. The parameters are listed in Table 11 in the order in which they appear in the file.
Each value is stored as a 4-byte floating-point (real) number. A MATLAB® script for reading
these files is included in the TurbSim archive; it is named “Test\readHHbin.m.”

Hub-Height Formatted Files

The hub-height formatted files contain essentially the same information as the hub-height binary
files, but the parameters are written in columns in human-readable form. See Table 11 for the list
of parameters. These files have a “.dat” extension.

Hub-Height AeroDyn Formatted Files

These human-readable files are in a format compatible with InflowWind (formerly AeroDyn).
They have the “.hh” extension. See Table 12 for the file format; the InflowWind Manual [4]
contains a detailed description of the parameters. The horizontal wind speed and wind direction
are equivalent to the vector sum of the instantaneous U- and V-component time series from the
hub-point, and the vertical wind speed is the corresponding W-component time series. TurbSim
always sets the horizontal wind-shear, vertical linear wind-shear, and gust-speed parameters to
zero in the AeroDyn hub-height files. The vertical power-law wind-shear exponent is constant
for the entire time series. If the input wind-profile type (WindProfileType) is PL or IEC, the
value in the AeroDyn HH file is the PLExp parameter; if WindProfileType is JET or LOG, the
power law exponent is calculated based on the mean wind speeds at the top and bottom of the
rotor disk:

PLExp = ln 
( )
 u ztop 

 ztop 
ln  (21)

 u ( zbottom )   zbottom 
 

The column of plots on the right side of Figure 17 shows how InflowWind uses the information
in these HH files to produce wind speeds at any part of the volume surrounding the turbine.

35
 Table 11. Parameters in Hub-Height Binary and Table 12. Format of Hub-Height
Formatted Files AeroDyn Files

Column Description Column Description

Time Time
Time Time from start of the simulation HorSpd Horizontal wind speed
U U-component wind speed |vectorial U+V|
uh Horizontal wind speed |vectorial U+V| WndDir Wind direction

ut Total wind speed |vectorial U+V+W| VerSpd Vertical wind speed (W

component)
V V-component wind speed
HorShr Horizontal linear wind-
W W-component wind speed shear parameter
u' Fluctuating u-component wind speed VerShr Vertical power-law wind-
(the mean is removed) shear exponent
v' Fluctuating v-component wind speed LnVShr Vertical linear wind-
w' Fluctuating w-component wind speed shear parameter
u'w' u'w' Reynolds stress component GstSpd Gust speed (not sheared
by InflowWind)
u'v' u'v' Reynolds stress component
v'w' v'w' Reynolds stress component
TKE Turbulent kinetic energy
CTKE Coherent turbulent kinetic energy

Full-Field TurbSim Binary Files

The FF TurbSim binary files are designed to be read by InflowWind. They have a “.bts”
extension. (The column of plots on the left side of Figure 17 shows how InflowWind uses FF
data.) TurbSim normalizes the time-series data (in the inertial reference frame coordinate
system) and encodes them in 2-byte integers stored in these files. The first part of each file is a
header that provides information about the grid and tells InflowWind how to convert the integers
to floating-point values. The wind speeds for the NumGrid_Y × NumGrid_Z grids and the tower
points (if specified) follow that. See Appendix D in this document for the file format. A
MATLAB script for reading these files is included in the TurbSim archive; it is named
“Test\readTSgrid.m.”

This binary format has been designed so that InflowWind does not need to read any other file to
properly convert the data to floating-point form. (In contrast, the FF Bladed-style binary files
store scaling information in the summary file.) This format also provides the maximum
resolution possible in two-byte integers.

Full-Field Bladed-Style Binary Files

The FF Bladed-style binary files are designed to be read by both InflowWind and GH Bladed.
They have a “.wnd” extension. TurbSim normalizes the data (in the inertial reference frame
coordinate system) and encodes them in 2-byte integers. The first part of the file is a header that
provides information about the grid; the normalized wind speeds for the

36
NumGrid_Y × NumGrid_Z grid points follow that. See Appendix E in this guide for the file
format. (The column of plots on the left side of Figure 17 shows how InflowWind uses FF data.)

When generating these files, TurbSim adds a section to the end of the summary file that tells
InflowWind how to convert the data to floating-point form. To decode the data, InflowWind
must read both the summary file (with the “.sum” extension) and the binary FF file. TurbSim
uses a newer file format than the format SNwind used. In general, this updated format retains
more resolution in the normalized 2-byte integers than the previous encoding method did. A
MATLAB script that reads these files is included in the TurbSim archive; it is named
“Test\readBLgrid.m.”

Tower Data Binary Files

The tower data binary files are similar to the FF Bladed-style binary files, except they contain
data for points in a single line at the grid center—going from the bottom of the grid to the
ground—using the same vertical resolution as the rest of the grid (see Figure 4). These files have
a “.twr” extension. TurbSim normalizes the data (in the inertial reference frame coordinate
system) and encodes them in 2-byte integers. The first part of the file is a header that provides
information about the location of the tower points and size of the file; this header is followed by
the wind speeds. When generating these files, TurbSim adds a section to the end of the summary
file that indicates how to convert the data to floating-point form (this is the same section that is
generated for the FF Bladed-style “.wnd” binary files). See Appendix F in this guide for a more
complete description of this binary format.

If a user requests FF binary files in TurbSim format (WrADFF = “true”), the tower points are
normalized and stored as 2-byte integers along with the full-field grid data in the file with a
“.bts” extension. In that case, a separate file with the “.twr” extension is not generated.

Full-Field Formatted Files

The FF formatted files are the traditional SNLWIND-3D FF output. These three files are human
readable (text), but use five times more storage than the binary files. Early versions of AeroDyn
could read these files, but AeroDyn and InflowWind no longer support this format. There is one
file for each component, with “.u,” “.v,” and “.w” file extensions, respectively.

Each of the files begins with a header containing with some basic information about the
simulation, and blocks of data follow. The first line in each block includes the time and the hub-
height wind speed. Following that line is a table with the number of rows and columns being the
number of grid points specified in the input file. The tables contain the wind speeds for the
different grid points. Their orientation is as if you are looking upwind (i.e., Y increases from left
to right, and Z increases from bottom to top), and all of the velocities are in the inertial reference
frame coordinate system. A MATLAB script for reading these files is included in the TurbSim
archive; it is named “Test\loadFFtxt.m.”

Coherent Turbulence Time-Step Files

One of the unique features of TurbSim is its ability to add coherent turbulence events based on
data obtained from numerical simulations of a Kelvin-Helmholtz billow. The data comes from
two sources: a large-eddy simulation from NCAR and a direct numerical simulation from

37
Colorado Research Associates (CoRA), both of Boulder, Colorado. Because the grid size of the
coherent events is very large (roughly 92 x 92 points), these events are not added directly to the
background turbulence in TurbSim. Instead, we create coherent turbulence time-step files, which
have a “.cts” extension. These text files contain a header indicating how to scale the non-
dimensional coherent structures; the header is followed by the times and file numbers of the
subset of LES or DNS data that define the coherent events. AeroDyn v13 reads this file along
with the background wind file and adds the two wind fields together. This feature can be used
only in programs that use AeroDyn v12.57 through v13.*. See the Using Coherent Turbulence
Time-Step Files with AeroDyn v13 section of this document for more information.

38
 Figure 17. Example TurbSim FF and HH wind files as implemented in AeroDyn.
The bottom left plot shows the FF grids after marching GridWidth/4 meters (along positive X)
from the position in the upper left plot; HH wind files (right column) do not march through the
turbine. At each time step, the FF wind velocity at X = GridWidth/2, Y = 0, Z = HubHt is
identical to the HH wind velocity at X = 0, Y = 0, Z = HubHt. Because TurbSim sets the
horizontal shear to 0 in the HH files, the velocity does not change with either X or Y. Thus, the
wind velocities in the FF and HH files are identical at X = GridWidth/2, Y = 0, Z = HubHt
(where the X axis on the plots emerges from the wind volume).

39
Spectral Models
TurbSim uses a modified version of the Sandia method [5] to generate time series based on
spectral representation. Several different spectral models are available, including two IEC
models, the Risø smooth-terrain model, and several NREL site-specific models (NWTCUP,
GP_LLJ, WF_UPW, WF_07D, and WF_14D). This section describes the velocity spectra used
in each of the models and discusses the measurements used to develop scaling for the site-
specific models. Standard deviations, σ , have been calculated by integrating the velocity
spectra, S:
∞
σ 2 = ∫ S ( f )df (22)
0

Plots comparing the velocity spectra of the different models are presented in Appendix G.

IECKAI: The IEC Kaimal Model

The IEC Kaimal model is defined in IEC 61400-1 2nd ed. [24], and 3rd ed. [21] and assumes
neutral atmospheric stability (RICH_NO = 0). 1 The spectra for the three wind components,
K = u, v, w, are given by

4 σ K 2 LK u hub
SK ( f ) = (23)
( )
5/3
1 + 6 f LK u hub

where f is the cyclic frequency and LK is an integral scale parameter. The IEC 61400-1 standard
defines the integral scale parameter to be

8.10LU , K =u

LK =2.70LU , K =v (24)
0.66L , K = w
 U

where the turbulence scale parameter, ΛU , is

0.7 ⋅ min ( 30m, HubHt ) , Edition 2

ΛU =  (25)
0.7 ⋅ min ( 60m, HubHt ) , Edition 3

(Note that the function min ( x1 , x2 ) in Eq. (25) indicates the minimum of x1 and x2 .) The
relationships between the standard deviations are defined to be

σ v = 0.8σ u
(26)
σ w = 0.5σ u

1
This model differs slightly from the original neutral spectra defined by Kaimal.

40
The velocity spectra (and standard deviations) of the IECKAI model are assumed to be invariant
across the grid. In practice, a small amount of variation in the u-component standard deviation
occurs due to the spatial coherence model.

IECVKM: The IEC Von Karman Isotropic Model

This IEC model is defined in IEC 61400-1 2nd ed. [24] for isotropic turbulence and neutral
atmospheric stability. The velocity spectra for the wind components are given by

4 σ u 2 L u hub
Su ( f ) = (27)
(1 + 71( f L u ) )
5
2 6

hub

and

=SK ( f )
2 σ K 2 L u hub
(1 + 189 ( f L / u ) ) 2
(28)
(1 + 71( f L u ) ) 2
11
6 hub

hub

for K = v, w. In these equations, f is the cyclic frequency and L is an integral scale parameter. L
is defined using the turbulence scale parameter, ΛU , from Eq. (25):

L 3.5LU
= (29)

The IEC standard defines the relationship between the standard deviations of the components to
be

σ=
v σ=
w σu . (30)

The velocity spectra (and standard deviations) of the IECVKM model are invariant across the
grid. In practice, a small amount of variation in the u-component standard deviation occurs due
to the spatial coherence model.

SMOOTH: The Risø Smooth-Terrain Model

TurbSim also offers the Risø smooth-terrain model (SMOOTH), based on work by Højstrup et
al. [29] and Olesen et al. [30]. This spectral model has separate equations for stable/neutral and
for unstable flows. The SMOOTH model (as well as the site-specific models) defines the
velocity spectra using local height and wind speed; this contrasts with the IEC models which use
the wind speed and height of the hub to define the spectra at all points. The spectra from the
SMOOTH model also form the basis for the spectra for all the site-specific models.

For stable and neutral conditions ( RICH _ NO ≥ 0 ), the SMOOTH-model velocity spectra for
the three wind components, K, are given by

41
 2
 z   fE  3

s1, K    
S K ( f ) = UStar 2  u fM   fM  (31)
5
 fz  3

1.0 + s2, K  
 u fM 

where f is the cyclic frequency, UStar is the friction velocity input parameter, u is the mean
wind speed at height z, and φE and φM are functions of the stability parameter, RICH_NO. The
two scales, s1 and s2, are defined as follows for each component:

 79.0, 263.0 K = u

s1, K , s2, K =
= 13.0, 32.0 K v (32)
 3.5, 8.6 K = w


The theoretical standard deviations of the wind components in stable and neutral conditions are
plotted in Figure 18. These values are calculated assuming infinite, continuous spectra with no
spatial coherence or time-domain cross-component correlation (i.e., the input mean hub
Reynolds stresses, PC_UW, PC_UV, and PC_VW, are “none”). The standard deviations
theoretically are constant across the rotor disk (using the same assumptions); in practice,
however, they can appear to vary with height (depending on the input values used). This variance
should decrease with increased record length. The relationships between the components’
standard deviations are

Figure 18. SMOOTH-model stable/neutral turbulence as a function of RICH_NO: left: standard

deviation normalized by UStar, right: Relationships between components’ standard deviations

42
 σ v = 0.76σ u
(33)
σ w = 0.59σ u

For unstable flows, ( RICH _ NO < 0 ), the SMOOTH spectra are modeled as the sum of low-
and high-frequency spectral peaks:

S ( f ) Slow ( f ) + S high ( f
= ) (34)

These two peaks are defined for the three wind components as follows:

 
z (
1 − ZIz )
z 2
 0.5 ZI ( ZI ) 3 105 u 1+15
2

( ) 
Su ( f ) UStar 2  (35)
− L
= u
+ ZI

( ) ( )
5 5
 1 + 2.2 f ZI 3 
(1 + 15 ZIz ) 3
3 2
fz
 1 + 33 u 1+15 z 

u
( ZI ) 

 
z (
1 − ZIz )
z 2
( ) 17 u 1+ 2.8
2
 0.95 ZI ZI 3
( 
ZI )
Sv ( f ) UStar 2  (36)
−L
= u
+ 
( ) (1 + 9.5 )
5 5
 1+ 2 f ZI 
(1 + 2.8 ZIz )
3 3 2
fz 3
 u u (1+ 2.8 ZIz ) 
 

and

 1

( ) +( )  2
2 2
2 uz (1 − ZIz )
fz 2
 0.95 ZI ( ZI ) 3
2
 u
0.3 ZIz  
Sw ( f )
−L
UStar 
2 u
  +  (37)
( ) ( ) + ( 0.15) ( )
5 2 5
 1 + 2 f ZI 3  fz 2
 1 + 5.3 fuz
3

 u
 u  
 

where f is the cyclic frequency, UStar and ZI are input parameters, and u is the mean wind
speed at height z. L is the Monin-Obukhov length parameter, which is a function of RICH_NO
and HubHt.

The standard deviations of the wind components in unstable atmospheric conditions vary with
height, the mixing layer depth (ZI), and L. Their approximate values are determined from the
following equations:

  ZI  3
2
 z   z 
2 −23

σ u2 ≈ UStar 2  0.62   + 4.77 1 −   1 + 15   (38)
  −L   ZI   ZI  
 

 2
 ZI  3 
2
z   z  3
−2
2
σ ≈ UStar  0.71 
v
2
 + 2.68 1 −  1 + 2.8   (39)
  −L   ZI   ZI  


 2
 ZI  3  z  
2
2
σ ≈ UStar  0.71
w
2
 + 1.46  1 −   (40)
  −L   ZI  


43
NWTCUP: The NREL National Wind Technology Center Model
The NWTCUP model, based on measurements from the NWTC/LIST project, represents
turbulent inflow characteristics at the NWTC, downwind of a major mountain range. In this
project, three towers were installed 1.5 rotor diameters upwind of the 600-kW NWTC Advanced
Research Turbine (ART). The central tower contained three-axis sonic anemometers at 15 m,
37 m, and 58 m above ground level; cup anemometers and wind vanes were located at 3 m,
37 m, and 58 m; and temperature measurements were obtained at 3 m, 15 m, 37 m, and 58 m.
Two additional towers, which were located 21 m north and south of the central tower, contained
three-axis sonic anemometers at 37 m. Neil Kelley et al. discuss this project and the
instrumentation further [31].

The spectra for this model are based on the 40-Hz time series data collected by the five sonic
anemometers. The default spatial coherence parameters generated for this model are based on
vertical coherence measured by the sonic anemometers on the central tower.

For neutral and stable flows, the NWTCUP spectra are defined by adding scaled versions of the
SMOOTH-model spectra:
NumPeaksK
SK ( f ) = ∑ pi , K S K , SMOOTH ( Fi , K f ) (41)
i =1

where NumPeaksK = 2 for all wind components K = u, v, w and the function S K , SMOOTH is defined
in Eq. (31). All of the pi , K and Fi , K scaling factors are functions of RICH_NO. Figure 19 shows
the standard deviations for the three wind components and the ratios between the components’

Figure 19. NWTCUP-model stable/neutral turbulence as a function of RICH_NO: left: standard

deviation normalized by UStar, right: relationships between components’ standard deviations

44
standard deviations.

For unstable flows, the NWTCUP model modifies the SMOOTH-model low- and high-
frequency peaks from Eq. (35) through Eq. (37):

=S K ( f ) p1, K S K ,low, SMOOTH ( F1, K f ) + p2, K S K ,high, SMOOTH ( F2, K f ) (42)

The scaling factors p1,K , p2,K , F1,K , and F2,K , which are empirically derived from spectra
calculated using NWTC/LIST velocity measurements, are functions of the RICH_NO and UStar
parameters. The standard deviations are similar to those of the unstable SMOOTH-model, scaled
by the p1,K and p2,K terms.

GP_LLJ: The NREL Great Plains Low-Level Jet Model

The Great Plains model (GP_LLJ) is based on measurements from a 120-m tower and from an
acoustic wind profiler (SODAR [sonic detection and ranging]) obtained during the Lamar Low-
Level Jet Project in southeastern Colorado. The tower included three-axis sonic anemometers at
54 m, 67 m, 85 m, and 116 m above the ground; cup anemometers and direction vanes located at
3, 52, and 113 m; and temperature measurements obtained at 3 m, 52 m, 83 m, and 113 m. The
SODAR provided measurements of wind speed and direction at 10-m vertical increments from
20 m to 500 m. The spectra and spatial coherence parameters defined in this model are based on
20-Hz time-series data collected at the sonic anemometers. Please refer to Neil Kelley et al. [32]
for details of that project.

The GP_LLJ model defines vertical profiles of stability and of shear velocity (i.e., stability and
shear velocity are functions of height). The stability profile—related to RICH_NO—is a local
Monin-Obukhov stability parameter, z L , and the shear velocity profile is a local u* value. The
values used for these profiles are placed in the TurbSim summary file. Both of these profiles are
calculated based on height, wind speed, and RICH_NO. The shear velocity profile also relies on
UStar and u*0 , which is defined in Eq. (14).

The z L and u* profiles are used to scale the GP_LLJ velocity spectra (in contrast, the other
models use the UStar and RICH_NO parameters, which are averaged values). For stable and
neutral flows, the spectra are defined by adding peaks from the form of the SMOOTH-model
spectra:
NumPeaksK
u*2
SK ( f ) = ∑ pi , K S K , SMOOTH ( Fi , K f ) (43)
UStar 2 i =1

where the function S K , SMOOTH is defined in Eq. (31), using the local stability parameter, z
L , to
determine the values of functions φE and φM (instead of using RICH_NO as the SMOOTH
model does). The u and v components have two peaks (NumPeaksK = 2, K = u, v), and the w
component is modeled with only one peak (NumPeaksw = 1). The scaling factors pi , K and Fi , K
are functions of both z L and u* . The standard deviations for the three wind components are
plotted in Figure 20. The ratios between the components satisfy the following inequalities:

45
 Figure 20. GPLLJ-model stable/neutral turbulence as a function of local stability and shear
velocities
σv
0.70 ≤ ≤ 0.98 (44)
σu
and

σw
0.52 ≤ ≤ 0.71 (45)
σu
By design, most of the LLLJP data was collected in the stable atmosphere. As a result, there was
not enough data to create a model of the spectra in unstable flows. Instead, the GP_LLJ spectra
for unstable atmospheric conditions use the same equations as the SMOOTH model spectra in
Eq. (35) through Eq. (37). The one difference is that the GP_LLJ scales the spectra using the
local u* values instead of the UStar input parameter. The GP_LLJ spectra for unstable flows are
thus defined as

u*2
SK ( f ) = S K , SMOOTH ( f ) (46)
UStar 2

WF_UPW: The NREL Wind Farm, Upwind Model

The WF_UPW wind-farm model is based on measurements taken from a 50-m tower upwind of
a large wind plant in San Gorgonio Pass, California. The spectra were calculated using 50-Hz
wind-speed measurements from a three-axis sonic anemometer located 23 m above the ground.
The parameters for spatial coherence were calculated using measurements from 5-Hz cup
anemometers and direction vanes located at 5 m, 10 m, 20 m and 50 m above ground level.

46
Please refer to Kelley [6] for details of the model development and Kelley and Wright [33] for
further details on the measurements.

For neutral and stable flows, the WF_UPW spectra are defined by adding scaled versions of the
SMOOTH-model spectra, using Eq. (41). All of the wind components use two spectral peaks
(NumPeaksK = 2, K = u, v, w) and each of the scaling factors pi , K and Fi , K are functions of
RICH_NO. Figure 21 shows the standard deviations for the three wind components and the ratios
between the components’ standard deviations.

For unstable flows, the WF_UPW model modifies the SMOOTH-model low- and high-
frequency peaks, using Eq. (42). The scaling factors p1,K , p2,K , F1,K , and F2,K are functions of
the RICH_NO parameter. The resulting standard deviations are similar to those of the unstable
SMOOTH model, but scaled by the p1,K and p2,K terms.

WF_14D: The NREL Wind Farm, Downwind Model (14 Rotor

Diameters)
The WF_14D wind-farm model is based on measurements taken on a 50-m tower downwind of a
41-row wind plant in San Gorgonio Pass, California. The tower was approximately 14-rotor-
diameters downwind of the plant, which consisted of 23-m hub-height Micon 65/13 machines
with 16-m rotor diameters.

The spectra were calculated using 50-Hz wind-speed measurements from a three-axis sonic
anemometer located 23 m above the ground. The parameters for spatial coherence were
calculated using measurements from 5-Hz cup anemometers and direction vanes located at 5 m,
10 m, 20 m, and 50 m above ground. The development of this model is described by Kelley 6,

Figure 21. WF_UPW-model stable/neutral turbulence as a function of RICH_NO: left: standard

deviation normalized by UStar, right: relationships between components’ standard deviations

47
 and the measurements are discussed further in Kelley and Wright [33].

For neutral and stable flows, the WF_14D spectra are defined by adding scaled versions of the
SMOOTH-model spectra, also using Eq. (41). All wind components use two spectral peaks
(NumPeaksK = 2, K = u, v, w) and each of the scaling factors pi , K and Fi , K are functions of
RICH_NO. Figure 22 shows the standard deviations for the three wind components and the ratios
between the components’ standard deviations.

For unstable flows, the WF_14D model modifies the SMOOTH-model low- and high-frequency
peaks listed in Eq. (35) through Eq. (37):
NumPeaksK
=S K ( f ) p1, K S K ,low, SMOOTH ( F1, K f ) + ∑ pi , K S K ,high, SMOOTH ( Fi , K f ) (47)
i =2

The u- and w-component spectra have two peaks (NumPeaksK = 2, K = u, w). For the v-
component spectra, Kelley found a third peak (NumPeaksv = 3), which he attributed to wakes
from the wind turbines upstream. The scaling factors pi , K and Fi , K , i = 1, 2,..., NumPeaksK , are
functions of the RICH_NO parameter. The resulting standard deviations are similar to those of
the unstable SMOOTH-model, scaled by the pi , K terms.

WF_07D: The NREL Wind Farm, Downwind Model (7 Rotor Diameters)

The scaling for the WF_07D wind-farm model is based on measurements taken at row 37 of a
41-row wind plant in San Gorgonio Pass, California for the SERI Thin-Airfoil Blade
Atmospheric Performance Test [34]. The 16-Hz measurements were obtained from a three-axis

Figure 22. WF_07D- and WF_14D-model stable/neutral turbulence as a function of RICH_NO: left:
standard deviation normalized by UStar, right: ratios of standard deviations

48
sonic anemometer 23-m above the ground, on a tower approximately 7-rotor-diameters
downwind of a row of operating Micon 65/13 wind turbines.

These measurements were used to calculate the scaling for coherent structures and default input
parameters. The measurements used to form the scaling for the WF_07D model, however, were
not sufficient to develop spectral scaling or spatial coherence. As a result, the WF_07D model
uses the same equations for the velocity spectra and spatial coherence as the WF_14D model.

TIDAL: The NREL/UW Tidal Channel Model

The TIDAL model for water turbulence is based on measurements taken near Marrowstone
Island in Puget Sound Washington [35]. These measurements were taken in a tidally-mixed tidal
boundary layer 4.6 meters above the bottom in 18 meters of water. The spectral form is
essentially the same as the SMOOTH spectral model, but the spectral amplitude and shear are
scaled directly based on this tidal channel’s turbulent kinetic energy (TKE) and shear rather than
implicitly from atmospheric boundary layer theory. In particular, the form is:
−1
 ∂u 
s K 2 s1, K  
SK ( f ) =  ∂z 
5/3 (48)
 f 
1 + s2, K  
 ∂u ∂z 

where the empirically determined coefficients are (for frequency, f, in hertz and ∂u/∂z in
1/second):

 1.21, 4.30 K = u

s1, K , s2, K =
= 0.33, 0.50 K v (49)
 0.23, 0.26 K = w


The shear, ∂u/∂z, is calculated internally from the specified mean velocity profile. In the case of
a logarithmic velocity profile, the shear is proportional to u/z and this form is essentially the
same as the SMOOTH spectral model. The component-TKE levels, σK2 are determined based on
an exponential profile proportional to UStar2:

σ K 2 = UStar 2 µ K e− 2z RefHt (50)

where, 𝜇𝑢 = 4.5, 𝜇𝑣 = 2.25, 𝜇𝑤 = 0.9, are empirically determined coefficients from the
Marrowstone Island site. RefHt is the reference height input parameter at which the input
parameter Uref is specified. For simulating fully mixed tidally forced boundary layers, RefHt
should be approximately equal to the water depth, and Uref the surface velocity. A simple way to
match observed velocity and TKE (𝜎𝐾 2 ) profiles is to make minor adjustments to Uref and
RefHt.

The TurbSim archive includes a “TurbSim_Hydro.inp” sample input file., The parameters set in
“TurbSim_Hydro.inp” are more appropriate to water turbulence and tidal channels than the
“TurbSim.inp” file, which has default values appropriate for atmospheric turbulence. Note that

49
many of the values in the input file are not used for the TIDAL spectral model (e.g. Z0,
RICH_NO, ZI, and all of the Coherent Turbulence Scaling Parameters).

TIMESR: Time Series Input

The TIMESR model accepts time series input at several points in space. TurbSim calculates the
mean direction at each point (saving the angles for use with direction profiles), then rotates the
velocities each point so that they are aligned with the mean wind direction, i.e., we now have u,
v, and w components. The mean velocity is removed from each point, and the mean values are
saved for use with velocity profiles.

TurbSim then performs an FFT of these zero-mean time series and calculates the spectral
amplitudes and phase angles. The resulting spectra are linearly interpolated in frequency and
space 2 (using nearest-neighbor extrapolation) to obtain spectral amplitudes for the simulated
points. The phase angles of the simulated points are chosen from a random uniform distribution;
they are then correlated to the phase angles of the time series at the single point specified by
RefPtID (see Input File for User-Defined Time Series) using the specified coherence model. This
ensures that there is coherence between the simulated points and the user-input RefPtID point
(when coherence is requested). Coherence between the simulated points and other user-input
time series is not guaranteed.

USRINP: User-Input Spectra

This spectral model produces uniform spectra for each of the three velocity components. Three
spectra representing the u, v, and w velocity components are input in a separate input file (see
Input File for User-Defined Spectra). These spectra are scaled with the scaling factors listed at
the top of the user-defined spectra input file and are linearly interpolated in frequency (using
nearest neighbor extrapolation) to compute target spectra for the simulated points.

Su ( f ) SpecScale1 ⋅ Suinput ( f )
= (51)

Sv ( f ) SpecScale2 ⋅ Svinput ( f )
= (52)

S w ( f ) SpecScale3 ⋅ S winput ( f )
= (53)

USRVKM: von Karman Model with User-Defined Scaling

The von Karman model with user-defined scaling computes the velocity spectra for the wind
components using local values of standard deviation, length scale, and wind speed. The velocity
spectra for the wind components are given by:

2
Currently the interpolation in space is limited to Z (height). It is envisioned that future versions will also
interpolate in the lateral direction.

50
 ( )
2
4 StdScaleu ⋅ σ uinterp Linterp u
Su ( f ) = (54)
(1 + 71( f L ))
5
2 6

interp u

and

( )
2

(1 + 189 ( f L ))
2 StdScaleK ⋅ σ Kinterp Linterp u 2
=SK ( f ) /u (55)
(1 + 71( f L ))
11 interp
2 6

interp u

for K = v, w. In these equations, f is the cyclic frequency, u is the time-averaged local wind
speed, and StdScaleK are constants whose values are defined in the input file for user-defined
profiles. Linterp and σ Kinterp are the length scale and standard deviation values from the user-
defined profile input file; these values are linearly interpolated with height (and, if necessary,
extrapolated using the nearest height) to obtain local values for these equations.

API: API Spectrum for Hurricane Winds

The API spectral model implements the Frøya model spectral density for the longitudinal wind
component proposed by Andersen and Løvseth as documented by Det Norske Veritas (DNV)
[36]. The Frøya spectral model was developed for neutral conditions over water in the
Norwegian Sea. Use of the Frøya spectrum can therefore not necessarily be recommended in
regimes where stability effects are important. A frequency of 1/2400 Hz defines the lower bound
for the range of application of the spectrum.
2 0.45
U   z 
320  0   
Su ( f ) =  10   10 
5 (56)
1+ f (
n 3n
)
where 𝑛 = 0.468 and
2/3 -0.75
 z  U 
f = 172 f    0  (57)
 10   10 

𝑈0 (assumed to be URef) is the one-hour mean wind speed at a height of 10 meters above mean
sea level, and z is the local height above sea level.

Spatial Coherence Models

In general, spatial coherence between points i and j is defined as

Sij ( f )
Cohi , j ( f ) = (58)
Sii ( f ) S jj ( f )

51
where f is the frequency, Sii is the power spectral density as defined in the “Spectral Models”
section, and Sij is the cross-spectral density. This coherence adds correlation between the same
wind components at two spatially separated points (e.g., ui-uj correlation, not u-v correlation).

The four spatial coherence models that are implemented in TurbSim are described below.

GENERAL: A general spatial coherence model

TurbSim’s general coherence function for all three of the velocity components, K = u, v, w, is
defined as

 2 
( )  f r
CohExp
exp  −aK zr
2 
Cohi , jK =
   + ( bK r )  (59)
 m
 um  
 

where f is the frequency, r is the distance between points i and j, zm is the mean height of the two
points, and u m is the mean of the wind speeds of the two points (over the entire simulation). The
variables a and b are the input coherence decrement and offset parameter, respectively, which are
defined by the values of the IncDec1, IncDec2, and IncDec3 input parameters (for each of the
components). Their default values are discussed in the “Input Files” section of this document and
are plotted in Figure 13 through Figure 15.

This coherence model is based on the form suggested by Thresher et al. [37] and implemented in

( )
CohExp
r
the IEC coherence model. The zm term has been added to allow users to implement
Solari’s coherence definition [38]. Note that if b = 0 and CohExp = 0, this equation also becomes
the Davenport coherence model [39].

IEC: IEC Coherence Model

The IEC coherence function for all three of the wind components, K = u, v, w is implemented as

 2 
 fr 
exp  −aK
2 
Cohi , jK =
   ( K ) 
+ b r (60)
  u hub  
 
u
where f is the frequency, r is the distance between points i and j on the grid, and hub is the mean
hub-height wind speed. The variables a and b are the input coherence decrement and offset
parameter, respectively, which are defined by the values of the IncDec1, IncDec2, and IncDec3
input parameters. If CohExp = 0, the only difference between the general and IEC spatial
coherence models is the use of mean wind speeds.
To implement the coherence model defined in the IEC 61400-1 standards for the u-component,
define

bu = 0.12 (61)
Lc

52
where Lc is a coherence scale parameter. For IEC 61400-1 2nd ed. [24], the parameters a and Lc
are

au = 8.8
(62)
Lc = 2.45 min ( 30m, HubHt )

where the function min ( ) is the minimum of 30 meters and HubHt. For IEC 61400-1 3rd ed.
[21], the parameters are

au = 12
(63)
Lc = 5.67 min ( 60m, HubHt )

The IEC 61400-1 standard does not specify coherence for the v or w wind-speed components.

NONE: Identity Coherence

When using the identity spatial coherence model for velocity component K, the coherence
between points i and j is defined as

1 i = j
Cohi , jK =  (64)
0 i ≠ j

API: API Longitudinal Coherence

The API coherence model implements the Frøya coherence model for wind over water, and
applies only to the longitudinal (u) component of the velocity. The coherence between points i
and j at frequency f is defined as

 1 3 
Coh
=i , jU exp −
 U0 ∑ Ak2 

(65)
 k =1 
where 𝑈0 (assumed to be URef) is the one-hour mean wind speed at a height of 10 meters above
mean sea level, and coefficient 𝐴𝑘 is defined by
− pk
 zi z j 
rk
A=k α k ⋅ f ⋅ ∆ kqk ⋅  (66)
 H 
 
with reference height 𝐻 = 10 m, and the coefficients ∆𝑘 , 𝑞𝑘 , 𝑝𝑘 , 𝑟𝑘 , and 𝛼𝑘 given in Table 13.
Note that in TurbSim, ∆1 = 0.

53
 Table 13. Coefficients for the API (Frøya) Coherence Model

k ∆𝑘 𝑞𝑘 𝑝𝑘 𝑟𝑘 𝛼𝑘
1 �𝑥𝑗 − 𝑥𝑖 � 1.00 0.4 0.92 2.9
2 |𝑦𝑗 − 𝑦𝑖 | 1.00 0.4 0.92 45.0
3 |𝑧𝑗 − 𝑧𝑖 | 1.25 0.5 0.85 13.0

Velocity and Direction Profiles

TurbSim offers users a choice of mean wind (velocity) profiles. The velocity profiles determine
the mean u-component velocity at each height for the length of the simulation. By definition, the
mean v- and w-component velocities are zero. Wind-direction profiles determine the mean
horizontal wind direction at each height. A wind-direction profile is calculated with the low-level
jet wind-speed profile and with the user-defined velocity profiles, but direction profiles are not
calculated with the other velocity profiles.

For velocity profiles that use a reference height and wind speed, TurbSim uses the inputs URef
and RefHt as the reference point to calculate the mean velocity at HubHt, u hub . The velocities at
other heights then are calculated using u hub and HubHt as the reference point. Figure 23 shows
an example of four different types of mean velocity profiles that were generated using default
boundary conditions and RICH_NO = 0.05 with the GP_LLJ turbulence model. For each of the
velocity profiles plotted in the figure, URef = 12 m/s and RefHt = HubHt = 90 m.

54
 Figure 23. Example wind-speed profiles generated in TurbSim for the GP_LLJ
turbulence model using a 90-m hub-height wind speed of 12 m/s, and RICH_NO = 0.05

Power-Law Wind Profile

The power-law mean velocity profile uses the PLExp input parameter to calculate the average
wind speed at height z using the equation
PLExp
 z 
u ( z ) = u ( zref )   (67)
z
 ref 

where u ( z ) is the mean wind speed at z and zref is a reference height above ground where the

( ) is known.
mean wind speed u zref

Logarithmic Wind Profile

The diabatic (logarithmic) wind profile calculates the average wind speed at height z using the
equation

ln ( z Z0 ) −ψ m
u ( z ) = u ( zref ) ln (68)
( zref
Z0 ) −ψ m

where u ( z ) is the mean wind speed at z, zref is a reference height above ground where the mean
wind speed is known, and Z0 is the input surface roughness. The function ψ m varies with the

55
RICH_NO stability parameter. When RICH_NO = 0 (as is the case with the IEC spectral
models), ψ m = 0 .

Logarithmic Water Profile

The “water” logarithmic mean velocity profile calculates the average flow speed at height z using
the equation

UStar  z 
=u ( z) ln   + URef (69)
κ z
 ref 

where 𝜅 = 0.41 is von Karmon’s constant. To specify this type of mean velocity profile use
“H2L” (short for “H2O Log”) as the WindProfileType input parameter. This velocity profile
should always and only be used with the TIDAL spectral model. Note that zref should be far from
the inertial boundary layer. In general, zref should be greater than 10 meters and/or equal to the
water depth of the tidal channel.

IEC Wind Profile

The IEC wind profile was the only wind-speed profile available in SNwind and SNLWIND-3D.
This profile uses the power-law wind profile for the wind speeds at heights on the rotor disk and
the logarithmic profile for heights not on the rotor disk. For example, if URef is specified at a
RefHt below the rotor disk, the logarithmic profile is used to calculate the HubHt mean wind
speed. Then the power-law profile would be used with the HubHt wind speed to calculate winds
across the rotor disk. This profile could cause a discontinuity in the wind profile at the bottom of
the rotor disk (this discontinuity would be noticed with tower points and with grids where
GridWidth < GridHeight).

Low-Level Jet Wind Profile

The low-level jet wind profile is derived from LLLJP 10-minute SODAR measurements and is
available with only the GP_LLJ spectral model. This profile type is unique because it generates
both wind-speed and wind-direction profiles.

The low-level jet wind-speed profile is defined using Chebyshev polynomials,

10
( z)
u= ∑c
n =0
n ⋅ Tn ( z ) (70)

where z is the height above ground, u ( z ) is the mean wind speed at height z, Tn(z) is the nth
order Chebyshev polynomial, and cn is a Chebyshev coefficient. The Chebyshev coefficients are
u
derived from LLLJP data and are a linear combination of the jet wind speed, ZJetMax , and input
parameters RICH_NO and UStar:

cn = C1,n u ZJetMax + C2,n RICH _ NO + C3,n UStar + C4,n (71)

56
The coefficients, Ci ,n , i = 1, 2, 3, 4, are determined by the input parameter ZJetMax.

The low-level jet wind-direction profile, like the wind-speed profile, is a Chebyshev polynomial
with coefficients derived from the same parameters in the LLLJP data. The wind-direction
profile is a relative horizontal direction and is always zero at the hub height. The HFlowAng
rotation is added to the relative direction provided from this profile.

Figure 24 plots example jet wind-speed and wind-direction profiles for three different jet heights.
The profiles have been generated with RICH_NO = 0.05, and an 80-m (hub-height) wind speed
of 12 m/s. The UStar parameter is 0.411 m/s, which is the default for these GP_LLJ conditions.

Figure 24. Example jet wind profiles with a 12 m/s wind speed at 80 m and RICH_NO = 0.05

API Wind Profile

The API wind profile is defined by the equation

( )
u ( z ) =URef 1 + 0.0573 1 + 0.15 URef ln z ( RefHt ) (72)

where z is the height above ground, u ( z ) is the mean wind speed at height z, URef is the one-
hour mean wind speed, and RefHt is 10 meters.

User-Defined Velocity Profiles

When WindProfileType is USR or TS, TurbSim linearly interpolates the input velocity and
direction profiles. These profiles are input directly for the USR model; for the TS model, the
profiles are calculated from the mean values of the input time series.

The profiles are extrapolated by using a nearest-neighbor approach: the profiles are constant at
heights above or below the heights where the input profiles are defined.

57
Coherent Structures
For analysis purposes, coherent structures have been defined in terms of CTKE (see Eq. (7) for
the CTKE definition). A coherent structure is an event where the 3-s mean CTKE meets a
specified threshold value, determined by the mean background levels of a particular site. The
event lasts from the time the threshold is first met until the 3-s mean CTKE falls below the
threshold value. For the LLLJP data, the threshold chosen was 2 m2/s2, and for the LIST and
wind-farm data, the threshold chosen was 5 m2/s2. Figure 25 gives an example of CTKE
measured in the NWTC LIST experiment and shows the detected coherent structures.

The background flow that is produced in TurbSim (i.e., the wind speed data contained in the FF
and HH output files) does contain coherent structures, using the definition above. These wind
files, however, do not always generate as many coherent structures as observed in the
atmosphere. To obtain more events with realistic spatial-temporal characteristics, sections (in
time) of numerical simulations of a Kelvin-Helmholtz billow are added randomly to the
background flow when the input parameter WrACT is “true.” TurbSim generates a coherent
turbulence time-step file (“.cts”) with the information describing how to scale the billow and
where the events should be added. These events then are superimposed on the background flow
in AeroDyn v13.

An example of the superimposed structures is shown in Figure 26. The black line in the plot
shows the 3-s mean CTKE of the background flow at one point on the grid; the green line shows
the 3-s mean CTKE of the background with the addition of events in the “.cts” file at the same
grid point. It should be noted that the “.cts” files can decrease the CTKE of the background as
well as increase it.

Figure 25. Coherent structures in a 10-minute period from the NWTC LIST dataset: the time
series shows the 3-s mean CTKE (solid black line) crossing the dashed threshold line, which
indicates the location of the coherent structures (indicated in blue)

58
 Figure 26. Example time series from the NWTCUP model: the black line indicates the
background wind file; the green shows the addition of events in a coherent time-step file
(“.cts”)

Adding and Scaling the Coherent Structures

The Kelvin-Helmholtz billow has been broken up into several different pieces, which are a fixed
non-dimensional size with non-dimensional velocities. Before adding these pieces to the
background flow, they must be scaled in space (through the DistScl input parameter) and in time
to determine the dimensional velocities. TurbSim randomly chooses the start times of the billow
pieces from an exponential distribution; the choice of which piece of the billow is inserted at
those places is determined from a uniform random distribution.

The coherent structure scaling for the site-specific spectral models has been determined from
analysis of sonic anemometer measurements at each of the respective sites, which are described
in the Spectral Models section of this guide. The SMOOTH model uses the same scaling as the
GP_LLJ model. Coherent structures are not added to the IEC spectral models.

The three non-input parameters for scaling the non-dimensional pieces of the billow and adding
them to the background time series are discussed below. A flow chart with these parameters is
included in Appendix C.

Interarrival Times
The interarrival time is the time from the start of one event to the start of the next event. These
times are exponentially distributed random variables with rate parameters determined from the
analyzed datasets. For the GP_LLJ and SMOOTH models, the random distribution is influenced
by the height and wind speed, u center , at the center of the billow. For the NWTCUP and the wind-
farm models, the random distribution is influenced by u center and RICH_NO.

Expected Length of Coherent Structures

The length of coherent structures is the total amount of time that contains coherent structures in a
given record. The expected lengths for each of the non-IEC spectral models are selected from a
random distribution whose probability density function matches the data from their respective
datasets.

59
TurbSim concatenates extra pieces of the billow to pieces that already have been added to the
coherent structure file until the total length of the events is at least the expected length of the
coherent structures from the datasets.

Peak Coherent Turbulent Kinetic Energy

The velocities for the coherent events are scaled to achieve a specific peak value in CTKE in the
set of events added to the background. The peak CTKE is a function of several different
parameters, depending on the spectral model. These parameters include height, z; mean wind
speed of the background flow at the center of the billow, u center ; shear across the billow
(difference in mean wind speed between the top and bottom of the billow), ∆u ; standard
deviation of vertical wind speed at the center of the billow, σ w ; and input parameters
center

RICH_NO and UStar. Some models also include a random component. Table 13 shows which
particular parameters are used for each of the non-IEC spectral models.

Table 14. Coherent Structure Peak CTKE Scaling

Random
TurbModel Predictors of Peak CTKE
Component?

GP_LLJ/SMOOTH RICH_NO, ∆u , UStar, z Yes

NWTCUP RICH_NO, ∆u , σ w , u center Yes

center

WF_UPW RICH_NO , ∆u , σ w , u center No

center

WF_07D RICH_NO, ∆u , σ w , UStar Yes

center

WF_14D RICH_NO, ∆u , σ w , u center No

center

Using Coherent Turbulence Time-Step Files with AeroDyn v13

To use the coherent time-step files that TurbSim generates (files with the “.cts” extension), a
coherent turbulence parameter input file must be created for AeroDyn v13’s InflowWind
module. This file must have a “.ctp” extension, and the name of this “.ctp” file must be entered
on the WindFile parameter line in the AeroDyn input file (using v12.57 through v13.*).

See Appendix H in this document for an example of the “.ctp” input file. Do not add or delete
lines from the file because AeroDyn v13 assumes parameters are on specific lines. The
parameters in the file are discussed below.

CTSpath: Name of path to coherent turbulence binary data files [-]

The CTSpath parameter is the name of the path that contains the binary data files for the coherent
structures, which you must get from the coherent structure archive on the TurbSim Web site (in
folder x90_i16). This directory must contain files called “Scales.les” and “Scales.dns,” which
contain scaling parameters for the two event types, and are used to read and convert normalized

60
16-bit integer binary data to real numbers. There should also be three folders in this directory,
named “u,” “v,” and “w” respectively, containing data for the three wind components. Each of
these three directories contains files named something like “u_i16_xxxxx.les.”

CTTSfile: Name of TurbSim CTS file [-]

The parameter CTTSfile is the name of the coherent time-step file generated by TurbSim. It has a
“.cts” extension. This file name must be specified relative to the directory from which AeroDyn
v13will be run.

CTbackgr: Name of TurbSim background FF file [-]

The parameter CTbackgr is the name of the background turbulence file. This should be the FF
wind file with the “.wnd” or “.bts” extension that was generated at the same time the “.cts” file
was created. This file name also must be specified relative to the directory from which AeroDyn
will be run. AeroDyn v13 automatically looks for the “.sum” summary file that goes with a
binary “.wnd” file.

CT_DF_Y: Decimation factor in the Y direction [-]

The CT_DF_Y parameter is used for decimating the binary coherent turbulence data in the
horizontal, Y, direction. Enter the horizontal decimation factor: A value of 1 uses every point in
the Y direction, 2 uses every other point, etc. It is recommended that you always use the entire
grid (i.e., CT_DF_Y = 1).

CT_DF_Z: Decimation factor in the Z direction [-]

The CT_DF_Z parameter is used for decimating the binary coherent turbulence data in the
vertical, Z, direction. Enter the vertical decimation factor: A value of 1 uses every point in the Z
direction, 2 uses every other point, etc. It is recommended that you always use the entire grid
(i.e., CT_DF_Z = 1).

Suggestions for Generating Coherent Turbulent Structures

Effort has been made in TurbSim to randomize the occurrence and scaling of coherent event
structures that occur in natural, nocturnal boundary layer flows. Simulations that generate
coherent turbulence time-step files have up to 10 degrees of stochastic freedom—in addition to
the random phases associated with each frequency at each grid point and wind component—and
are designed to give some feel of the expected variability in the atmosphere. Because of the
degree of variability, using more than 30 different random seeds 3 for a specific set of boundary
conditions is recommended.

To test the effects of a coherent structure (KH billow), we recommend using the “KHTEST”
option in the IECturbc input parameter with the NWTCUP spectral model. This test function
superimposes one intense coherent event in the middle of the output time series, reducing the
number of stochastic degrees of freedom to no more than two (plus the random phases). The
gradient Richardson number (RICH_NO) and wind shear (PLExp) of the background flow are
overwritten, and TurbSim uses fixed values to scale the LES-type event. This test function is

3
As a general rule of thumb, the number 30 is the dividing line between large and small sample statistics.

61
designed to generate intense turbulence, and does not necessarily reflect the variability for given
boundary conditions.

The choice of the gradient Richardson number and hub wind speed largely control the impact of
coherent structures on turbine response. It is recommended that at least one series of runs be
made at rated wind speed and a Richardson number between 0.02 and 0.05. Further discussion
on the impact of coherent turbulent structures on wind turbines is found in [40].

Warnings
• AeroDyn v12.57 or a later version is required to read TurbSim files correctly.
• If you compile AeroDyn v13, you must use the compiler option “/assume:byterecl” to read
the TurbSim coherent structures binary files correctly. If you use ADAMS2AD [41], be sure
to use v12.17 or later so that this compiler option is set.
• Hub-height time series from HH wind files and FF wind files (without UsableTime = “ALL”)
do not have events happening at the same time because InflowWind shifts the FF files (see
Figure 17).
• Because of the way the FFT routine works, extra time may be added to the analysis time to
get the FFT to run efficiently. Due to this plus the fact that the output time could be shorter
than the analysis time, the mean wind speed for the portion of the run actually used could be
different from what was specified in the input file.
• The statistics calculated in the summary file are based on the complete time series generated
(the analysis time plus any extra time added for the FFT calculations). Because the output
time can be less than the analysis time, these statistics might differ from what can be
calculated from the output files.
• Be cautious when using mean flow angle inputs with full-field grids for InflowWind.
InflowWind marches FF grids through the turbine along its Propogation Direction at the
mean hub-height wind speed, regardless of the flow angles. This can give strange results if
the mean flow angles are not small.

Limitations
• The GP_LLJ spectral model is estimated to be applicable up to a height of 230 m.
• The SMOOTH spectral model and the coherent turbulence time-step files are both currently
estimated to be applicable up to a height of 120 m.
• The NWTCUP spectral model is estimated to be applicable up to heights of 120 m.
• The wind farm spectral models (WF_UPW, WF_07D, and WF_14D) are valid only up to
heights of about 50 m.

Possible Future Enhancements

• Define the grid by specifying the domain bounds instead of assuming the rotor is centered in
the upper portion of the grid.
• Add the Mann model.

62
• Add other site-specific models if data becomes available.

Caveats
NREL makes no guarantees about the usability or accuracy of TurbSim, which is essentially a
beta code. NREL does not have the resources to provide full support for this program.

Feedback
If you have questions about TurbSim, please use our forums. We will respond to your needs if
time and resources permit, but please do not expect an immediate response. You can apply for an
account on the forum here: https://wind.nrel.gov/forum/wind/viewforum.php?f=17

63
References
1. Platt, A.; Jonkman, B.J.; Jonkman, J.J. InflowWind User’s Guide. Draft Version, October
2016. http://nwtc.nrel.gov/InflowWind/. Accessed June 1, 2016.

2. Jonkman, J. M.; Buhl Jr., M. L. FAST User’s Guide. NREL/EL-500-29798. Golden, CO:
National Renewable Energy Laboratory, August 2005.
http://wind.nrel.gov/designcodes/simulators/fast/. Accessed August 25, 2009.

3. MSC Software. 2005. MSC.ADAMS®. MSC.Software Corporation, Santa Ana, CA.

http://www.mscsoftware.com/. Accessed August 25, 2009.

4. Laino, D. J.; Hansen, A.C. User’s Guide to the Wind Turbine Dynamics Aerodynamics
Computer Software AeroDyn. Salt Lake City, UT: Windward Engineering, LC, December 2002.
http://nwtc.nrel.gov/AeroDyn/. Accessed August 25, 2009.

5. Veers, P.S. Three-Dimensional Wind Simulation. SAND88-0152. Albuquerque, NM: Sandia

National Laboratories, March 1988.

6. Kelley, N.D. Full Vector (3-D) Inflow Simulation in Natural and Wind Farm Environments
Using an Expanded Version of the SNLWIND (VEERS) Turbulence Code. NREL/TP-442-5225.
Golden, CO: National Renewable Energy Laboratory, November 1992.

7. Bossanyi, E.A. GH Bladed Version 3.6 User Manual. Document 282/BR/010 Issue 12. Garrad
Hassan and Partners Limited, 2003.

8. Buhl, M.L. Jr. SNwind User’s Guide. NREL/EL-500-30121. Golden, CO: National Renewable
Energy Laboratory, October 2001.

9. Swarztrauber, P.N. FFTPACK Version 4.1. Boulder, CO: National Center for Atmospheric
Research, November 1988. http://www.scd.ucar.edu/softlib/FFTPACK.html. Accessed August
25, 2009.

10. Kelley, N.D.; Jonkman, B.J. Overview of the TurbSim Stochastic Inflow Turbulence
Simulator Version 1.21. NREL/TP-500-41137. National Renewable Energy Laboratory, Golden,
CO, March 2007.

11. Buhl, Jr. M.L. Installing NWTC CAE Tools on PCs Running Windows®. Golden, CO:
National Renewable Energy Laboratory, August 2013.

12. The MathWorks. 2008. MATLAB®. Natick, MA. http://www.mathworks.com/. Accessed

August 25, 2009.

13. Anderson, E.; Bai, Z.; Bischof, C.; Blackford, S.; Demmel, J.; Dongarra, J.; Du Croz, J.;
Greenbaum, A.; Hammarling, S.; McKenney, A.; Sorensen, D. LAPACK User’s Guide, 3rd
edition. Philadelphia, PA: Society for Industrial and Applied Mathematics, August 1999.
http://www.netlib.org/lapack. Accessed August 25, 2009.

64
14. Lawson, C.L.; Hanson, R.J.; Kincaid, D.; Krogh, F.T. “Basic Linear Algebra Subprograms
for FORTRAN usage.” ACM Trans. Math. Soft., Vol. 5, September 1979; pp. 308-323.

15. Dongarra, J.J.; Du Croz, J.; Hammarling, S.; and Hanson, R.J. “An extended set of
FORTRAN Basic Linear Algebra Subprograms.” ACM Trans. Math. Soft., Vol. 14, March 1988;
pp. 1-17.

16. Meissner, Loren P. (August 1995) “ranlux.f90.”

http://flash.uchicago.edu/~tomek/htmls/refs/ranlux.f90.html. Accessed October 19, 2005. Copy
of document on file with the author.

17. Buhl, M.L., Jr. “NWTC Subroutine Library” NWTC Design Codes,
http://wind.nrel.gov/designcodes/miscellaneous/nwtc_subs. Last modified 18-Jan-2008;
Accessed August 25, 2009.

18. L'ecuyer, P. “Efficient and Portable Combined Random Number Generators.”

Communications of the ACM, Vol. 31, June 1988; pp. 742-751.

19. Lüscher, M. “A portable high-quality random number generator for lattice field theory
simulations.” Computer Physics Communications, Vol. 79, February 1994; pp. 100-110.

20. James, F. “RANLUX: A Fortran implementation of the high-quality pseudorandom number

generator of Lüscher.” Computer Physics Communications, Vol. 79, February 1994; pp. 111-
114.

21. IEC 61400-1 “Wind turbines-Part 1: Design requirements.” 3rd edition. Geneva, Switzerland:
International Electrotechnical Commission, August 2005.

22. IEC 61400-2 “Wind turbines-Part 2: Design requirements for small wind turbines.” 2nd
edition. Geneva, Switzerland: International Electrotechnical Commission, March 2006.

23. IEC 61400-3 “Wind turbines-Part 3: Design requirements for offshore wind turbines.” Draft
1st edition. Geneva, Switzerland: International Electrotechnical Commission, January 2006.

24. IEC 61400-1 “Wind turbine generator systems-Part 1: Safety requirements.” 2nd edition.
Geneva, Switzerland: International Electrotechnical Commission, 1999.

25. Panofsky, H.A.; Dutton, J.A. Atmospheric Turbulence: Models and Methods for Engineering
Applications. New York: Wiley-Interscience, 1984. 397 pp.

26. Dutton, J.A.; Panofsky, H.A.; Larko, D.; Shirer, H.N.; Stone, G.; Vilardo, M. Statistics of
wind fluctuations over complex terrain. Report No. DOE/ET/20560-1. University Park, PA:
Pennsylvania State University, Department of Meteorology, October 1979.

27. ESDU. Report No. 85020. Characteristics of atmospheric turbulence near the ground, Part II:
Single point data for strong winds (neutral atmosphere). London: Engineering Sciences Data
Unit, April 1993.

65
29. Højstrup, J. “Velocity Spectra in the Unstable Planetary Boundary Layer.” Journal of the
Atmospheric Sciences, Vol. 39, October 1982; pp. 2239-2248.

30. Olesen, H.R.; Larsen, S.E.; Højstrup, J. “Modeling Velocity Spectra in the Lower Part of the
Planetary Boundary Layer.” Boundary-Layer Meteorology, Vol. 29, July 1984; pp. 285-312.

31. Kelley, N.; Hand, M.; Larwood, S.; and McKenna, E. The NREL Large-Scale Turbine Inflow
and Response Experiment – Preliminary Results. NREL/CP-500-30917. Golden, CO: National
Renewable Energy Laboratory, January 2002.

32. Kelley, N.D.; Shirazi, M.; Jager, D.; Wilde, S.; Adams, J.; Buhl, M.; Sullivan, P.; Patton, E.
Lamar Low-Level Jet Project Interim Report. NREL/TP-500-34593. Golden, CO: National
Renewable Energy Laboratory,. January 2004.

33. Kelley, N.D.; Wright, A.D. A Comparison of Predicted and Ovserved Turbulence Wind
Fields Present in Natural and Internal Wind Park Environments. NREL/TP-257-4508. Golden,
CO: National Renewable Energy Laboratory,. October 1991.

34. Tangler, J.; Smith, B.; Jager, D.; Olsen, T. SERI Thin-Airfoil Blade Atmospheric
Performance Test: Final Results NREL/TP-257-4076, Golden, CO: National Renewable Energy
Laboratory, September 1990.

35. Thomson, J.; Polagye, B.; Durgesh, V.; Richmond, M.; “Measurements of Turbulence at
Two Tidal Power Sites in Puget Sound, WA (USA)”, Journal of Oceanic Engineering, July
2012. doi: 10.1109/JOE.2012.2191656

36. Det Norske Veritas, “Recommended Practice DNV-RP-C205: Environmental Conditions and
Environmental Loads,” October 2010. https://rules.dnvgl.com/docs/pdf/DNV/codes/docs/2010-
10/RP-C205.pdf. Accessed June 3, 2016.

37. Thresher, R.W.; Holley, W.E.; Smith, C.E.; Jafarey, N.; and Lin, S.-R. Modeling the
Response of Wind Turbines to Atmospheric Turbulence. Report No. RL0/2227-81/2. Corvallis,
OR: Oregon State University, Department of Mechanical Engineering, August 1981.

38. Solari, G. “Turbulence Modeling for Gust Loading,” ASCE Journal of Structural
Engineering, Vol 113 (7), July 1987; pp. 1550-1569.

39. Davenport, A.G. “The Spectrum of Horizontal Gustiness Near the Ground in High Winds,”.
Quarterly Journal of the Royal Meteorological Society, Vol. 87, 1961; pp. 194-211.

40. Kelley, N.D.; Jonkman, B.J.; Scott, G.N.; Bialasiewicz, J.T.; Redmond, L.S. The Impact of
Coherent Turbulence on Wind Turbine Aeroelastic Response and Its Simulation. NREL/CP-500-
38074. Golden, CO: National Renewable Energy Laboratory, August 2005.

41. Laino, D.J.; Hansen, A.C. User’s Guide to the Computer Software Routines AeroDyn
Interface for ADAMS®. Salt Lake City, UT: Windward Engineering, LC, September 2001.
http://wind.nrel.gov/designcodes/simulators/adams2ad/. Accessed August 25, 2009.

66
 Appendix A: Sample TurbSim Input Files

---------TurbSim v2.00.* Input File------------------------

Example input file for TurbSim.
---------Runtime Options-----------------------------------
False Echo - Echo input data to <RootName>.ech (flag)
2318573 RandSeed1 - First random seed (-2147483648 to 2147483647)
RANLUX RandSeed2 - Second random seed for intrinsic pRNG, or other pRNG: "RanLux" or "RNSNLW"
False WrBHHTP - Output HH turbulence parameters in GenPro-binary form? (Generates RootName.bin)
False WrFHHTP - Output HH turbulence parameters in formatted form? (Generates RootName.dat)
False WrADHH - Output hub-height time-series data in AeroDyn form? (Generates RootName.hh)
False WrADFF - Output FF time-series data in TurbSim/AeroDyn form? (Generates Rootname.bts)
True WrBLFF - Output FF time-series data in BLADED/AeroDyn form? (Generates RootName.wnd)
False WrADTWR - Output tower time-series data? (Generates RootName.twr)
False WrFMTFF - Output FF time-series data in formatted (readable) form? (RootName.u, .v, .w)
True WrACT - Output coherent turbulence time steps in AeroDyn form? (Generates RootName.cts)
True Clockwise - Clockwise rotation looking downwind? (Used only for FF binary files w/ BLADED)
0 ScaleIEC - Scale IEC turbulence models to exact target std deviation? [0=none;1=hub;2=all]

--------Turbine/Model Specifications-----------------------
13 NumGrid_Z - Vertical grid-point matrix dimension
13 NumGrid_Y - Horizontal grid-point matrix dimension
0.05 TimeStep - Time step [s]
600 AnalysisTime- Length of analysis time series [s] (program will add time if necessary)
"ALL" UsableTime - Usable length of output time series [s] (GridWidth/MeanHHWS s added if not "ALL")
84.30 HubHt - Hub height [m] (should be > 0.5*GridHeight)
80.00 GridHeight - Grid height [m]
80.00 GridWidth - Grid width [m] (should be >= 2*(RotorRadius+ShaftLength))
0 VFlowAng - Vertical mean flow (uptilt) angle [degrees]
0 HFlowAng - Horizontal mean flow (skew) angle [degrees]

--------Meteorological Boundary Conditions-------------------

"SMOOTH" TurbModel - Turbulence model (see Table 4 for valid codes)
"unused" UserFile - Name secondary input file for user-defined spectra or time series inputs
"1-ED2" IECstandard - Number of the IEC standard (61400-x, x=1,2,3) with optional 61400-1 ed. number
"A" IECturbc - IEC turbulence characteristic ("A", "B", "C" or TI in %) or KHTEST
"NTM" IEC_WindType- IEC turbulence type ("NTM", "xETM", "xEWM1", or "xEWM50" for x=class 1, 2, or 3)
default ETMc - IEC Extreme turbulence model "c" parameter [m/s] (or "default")
"PL" ProfileType - Wind profile type (see Table 6 for valid codes)
"unused" ProfileFile - Name of the file that contains user-defined input profiles
84.30 RefHt - Height of the reference wind speed [m]
18.2 URef - Mean wind speed at the reference height [m/s] [must be 1-hr mean for API model]
450 ZJetMax - Height of the low-level jet [m] (70-490 m or "default", only for "JET" profile)
default PLExp - Power law exponent (or "default")
default Z0 - Surface roughness length [m] (or "default")

--------Non-IEC Meteorological Boundary Conditions------------

default Latitude - Site latitude [degrees] (or "default")
0.05 RICH_NO - Gradient Richardson number [-]
default UStar - Friction or shear velocity [m/s] (or "default")
default ZI - Mixing layer depth [m] (or "default")
default PC_UW - Hub mean u'w' Reynolds stress [m^2/s^2] (or "default" or "none")
default PC_UV - Hub mean u'v' Reynolds stress [m^2/s^2] (or "default" or "none")
default PC_VW - Hub mean v'w' Reynolds stress [m^2/s^2] (or "default" or "none")

--------Spatial Coherence Parameters----------------------------

default SCMod1 - u-component coherence model ("GENERAL","IEC","API","NONE", or "default")
default SCMod2 - v-component coherence model ("GENERAL","IEC","NONE", or "default")
default SCMod3 - w-component coherence model ("GENERAL","IEC","NONE", or "default")
default InCDec1 - u-component coherence parameters [-, m^-1] ("a b" in quotes or "default")
default InCDec2 - v-component coherence parameters [-, m^-1] ("a b" in quotes or "default")
default InCDec3 - w-component coherence parameters [-, m^-1] ("a b" in quotes or "default")
default CohExp - Coherence exponent for general model [-] (or "default")

--------Coherent Turbulence Scaling Parameters-------------------

"M:\coh_events\eventdata" CTEventPath - Name of the path where event data files are located
"Random" CTEventFile - Type of event files ("LES", "DNS", or "RANDOM")

67
True Randomize - Randomize the disturbance scale and locations? (true/false)
1.0 DistScl - Disturbance scale (ratio of wave height to rotor disk).
0.5 CTLy - Fractional location of tower center from right to L of dataset looking downwind
0.5 CTLz - Fractional location of hub height from the bottom of the dataset
30.0 CTStartTime - Minimum start time for coherent structures in RootName.cts [s]

Figure A-1. Sample Primary TurbSim input file (Note: figure is continued from previous page.)

--------------TurbSim v2.00.* User Time Series Input File-----------------------

Time series input 2 sonic anemometers from 01134_16_40_00_013.mat. Using rotated series.
--------------------------------------------------------------------------------
3 nComp - Number of velocity components in the file
2 nPoints - Number of time series points contained in this file (-)
2 RefPtID - Index of the reference point (1-nPoints)
Pointyi Pointzi ! nPoints listed in order of increasing height
(m) (m)
0.00000 30.00000
0.00000 76.00000
--------Time Series-------------------------------------------------------------
Elapsed Time Point01u Point01v Point01w Point02u Point02v Point02w
(s) (m/s) (m/s) (m/s) (m/s) (m/s) (m/s)
0.0000 10.0239 -6.5673 0.1700 10.7104 -4.3265 -0.2657
0.0500 9.8543 -6.6871 0.2014 10.5539 -4.5656 -0.1635
0.1000 9.6866 -6.9837 -0.0274 10.6105 -4.1738 -0.1907
0.1500 9.7324 -7.0552 -0.0051 10.6691 -4.4155 -0.0675
0.2000 10.6893 -6.8507 -0.9577 10.3897 -4.9771 0.2487
0.2500 9.9231 -7.3007 0.7656 10.4993 -4.6568 0.1041
0.3000 10.6087 -7.4602 1.1109 10.6404 -4.6216 0.4016
0.3500 10.7004 -6.5530 1.5361 10.6060 -5.0307 0.2697
0.4000 10.6239 -6.5870 0.9715 10.2804 -5.5762 0.2131
0.4500 10.3173 -6.9557 0.7657 9.7826 -5.9725 0.4581
0.5000 10.1416 -7.2209 0.7567 10.0303 -4.9716 0.6309
0.5500 10.5047 -6.7512 0.6150 9.2657 -4.9317 0.3516
0.6000 10.7474 -6.2916 1.0679 9.8545 -4.6793 0.9724
0.6500 10.0867 -7.4206 0.5036 9.7205 -4.9432 1.1458
0.7000 9.8459 -7.4542 -1.2710 8.9698 -4.7850 1.0775
0.7500 9.6427 -7.2455 -1.4315 9.3917 -4.6785 1.1891
0.8000 9.5695 -7.7153 -0.7343 9.5739 -4.4328 1.1584
0.8500 10.2921 -6.9918 -0.9979 9.6578 -4.3620 1.2127
0.9000 9.8191 -6.6210 0.3998 9.8743 -4.2941 1.0936
0.9500 10.0563 -6.9999 0.2417 10.3157 -4.2559 1.1260
1.0000 9.2220 -5.9308 0.8000 9.9854 -4.1755 1.2094
1.0500 10.0784 -5.5374 2.1954 9.5217 -4.6836 0.8753
1.1000 9.5813 -5.8415 2.4204 10.2011 -4.7455 0.9099
1.1500 10.1393 -5.7391 1.2873 9.5294 -5.2682 0.6955
1.2000 10.3018 -6.1910 0.7048 9.3079 -5.5758 0.5641
1.2500 10.4492 -6.5951 1.0127 9.5492 -6.0838 0.6965
1.3000 9.7664 -7.2437 0.7676 9.8434 -6.0361 1.7628
1.3500 8.8919 -7.6760 -0.0979 10.1855 -5.7703 2.1307
1.4000 8.5238 -7.3008 -0.3770 10.8332 -4.6349 1.7131
1.4500 8.8623 -7.0775 -0.9606 11.0740 -3.6287 1.5952
1.5000 8.9728 -7.6597 -1.1552 10.7549 -4.2620 1.7992
1.5500 8.8930 -7.7153 -1.7600 10.7559 -5.3923 1.5490
< Lines omitted >
599.7500 21.6185 1.3266 0.5301 20.7629 1.8099 0.4765
599.8000 20.6428 2.2662 0.6105 20.6605 2.0787 0.8918
599.8500 20.0781 2.4219 1.1325 20.0819 2.0141 1.2528
599.9000 19.9940 1.8457 1.7090 20.2872 2.2371 1.4736
599.9500 20.6705 2.1299 2.4844 20.4711 2.0164 1.8634

Figure A-2. Sample input file for User-Defined Time Series

68
-------- User-Defined Spectra (Used only with USRINP spectral model) ------------------------------------
- The Kaimal spectra IEC 61400-1 Ed. 3 for Vhub=12 m/s; Zhub=90 m; Class="B"; -
---------------------------------------------------------------------------------------------------------
20000 NumUSRf - Number of Frequencies [determines how many lines to read from this file]
1.0 SpecScale1 - scaling factor for the input u-component spectrum
1.0 SpecScale2 - scaling factor for the input v-component spectrum
1.0 SpecScale3 - scaling factor for the input w-component spectrum
.........................................................................................................
Frequency u-component PSD v-component PSD w-component PSD
(Hz) (m^2/s) (m^2/s) (m^2/s)
---------------------------------------------------------------------------------------------------------
0.001 364.644672 92.196417 9.432145
0.002 290.820811 84.504829 9.221093
0.003 238.306635 77.790863 9.017498
0.004 199.474346 71.891456 8.821001
0.005 169.860744 66.676649 8.631266
0.006 146.703651 62.041773 8.447977
0.007 128.214357 57.901706 8.270838
0.008 113.190378 54.186616 8.099568
0.009 100.797523 50.838749 7.933903
0.010 90.441383 47.809980 7.773593
0.011 81.688510 45.059936 7.618404
0.012 74.216397 42.554527 7.468113
0.013 67.780802 40.264798 7.322511
0.014 62.193824 38.166019 7.181398
0.015 57.308865 36.236959 7.044586
0.016 53.010107 34.459303 6.911897
0.017 49.205001 32.817181 6.783162
0.018 45.818825 31.296779 6.658222
0.019 42.790679 29.886027 6.536923
0.020 40.070501 28.574339 6.419121
0.021 37.616811 27.352396 6.304680
< Lines omitted >
19.994 0.000616 0.000819 0.000814
19.995 0.000616 0.000819 0.000814
19.996 0.000616 0.000819 0.000814
19.997 0.000616 0.000819 0.000814
19.998 0.000615 0.000819 0.000814
19.999 0.000615 0.000818 0.000814
20.000 0.000615 0.000818 0.000814

Figure A-3. Sample input file for User-Defined Spectra

---------TurbSim v2.00.* Profile Input File------------------------

Example file using completely made up profiles
-------- User-Defined Profiles (Used only with USR wind profile or USRVKM spectral model) -------------
5 NumUSRz - Number of Heights
1.092 StdScale1 - u-component scaling factor for the input standard deviation
1.0 StdScale2 - v-component scaling factor for the input standard deviation
0.534 StdScale3 - w-component scaling factor for the input standard deviation
-----------------------------------------------------------------------------------
Height Wind Speed Wind Direction Standard Deviation Length Scale
(m) (m/s) (deg, cntr-clockwise ) (m/s) (m)
-----------------------------------------------------------------------------------
15.0 3 00 .100 3
25.0 4 00 .200 4
35.0 5 00 .300 6
45.0 6 00 .100 9
55.0 7 00 .500 13

Figure A-4. Sample input file for User-Defined Profiles

69
Appendix B: TurbSim Quick-Start Guidelines for IEC
Turbulence
To generate IEC-type turbulence, many of the parameters in the TurbSim input file can be
ignored. Figure B-1 shows a TurbSim input file set up to generate IEC 61400-1 3rd ed., category
“B” turbulence for the NTM using the Kaimal model. It creates a FF Bladed-style “.wnd” file
containing 600 seconds of (periodic) usable data, using a time step of 0.05 s.

All of the unused parameters have been crossed out in Figure B-1. The parameters in black
typically do not need to be changed. The input parameters that typically might have to be
changed are mentioned below, along with suggestions for typical values. The Input Files section
of this guide describes the parameters in more detail.

The parameters in blue italics in Figure B-1 should be changed based on the particular turbine
for which the wind field is being generated:

ScaleIEC: Change this parameter to the type of scaling desired. If you are unsure, use 0.

NumGrid_Z: The number of vertical grid points should be set so there is sufficient vertical grid
resolution. A typical value is an odd integer that is close to the GridHeight divided by the mean
chord of the turbine’s blades.

NumGrid_Y: The number of lateral grid points should be set so there is sufficient lateral grid
resolution. A typical value is an odd integer that is close to the GridWidth divided by the mean
chord of the turbine’s blades.

HubHt: This is the hub height in meters of the turbine for which the turbulence is being
generated.

GridHeight: The grid height (in meters) typically is 10% larger than the turbine rotor diameter. It
must be larger for turbines that have significant displacements.

GridWidth: The grid width (in meters) typically is the same as GridHeight.

IECturbc: The turbulence category should be “A,” “B,” or “C,” depending on the desired
61400-1 category. Category “A” is the most turbulent.

RefHt: The reference height is the height (in meters) where the input wind speed is defined. It is
typically the same as HubHt.

The parameters in bold red in Figure B-1 typically are changed for each case when running
design load cases:

RandSeed1: The random seed, which initializes the pseudo-random number generator, should be
a different number for each simulation. For each case, several different seeds should be used,
keeping all other input parameters constant.

70
 IEC_WindType: This is the wind condition for the (turbulent) IEC load cases. It often is NTM.
For other conditions, see Table 5 of this guide.

URef: This is the reference wind speed (in meters per second) at the RefHt. It typically ranges
from cut-in to cut-out in 2 m/s increments.

---------TurbSim v2.00.* Input File------------------------

Example input file for TurbSim.
---------Runtime Options-----------------------------------
False Echo - Echo input data to <RootName>.ech (flag)
1234567 RandSeed1 - First random seed (-2147483648 to 2147483647)
RANLUX RandSeed2 - Second random seed for intrinsic pRNG, or other pRNG: "RanLux" or "RNSNLW"
False WrBHHTP - Output HH turbulence parameters in GenPro-binary form? (Generates RootName.bin)
False WrFHHTP - Output HH turbulence parameters in formatted form? (Generates RootName.dat)
False WrADHH - Output hub-height time-series data in AeroDyn form? (Generates RootName.hh)
False WrADFF - Output FF time-series data in TurbSim/AeroDyn form? (Generates Rootname.bts)
True WrBLFF - Output FF time-series data in BLADED/AeroDyn form? (Generates RootName.wnd)
False WrADTWR - Output tower time-series data? (Generates RootName.twr)
False WrFMTFF - Output FF time-series data in formatted (readable) form? (RootName.u, .v, .w)
False WrACT - Output coherent turbulence time steps in AeroDyn form? (Generates RootName.cts)
True Clockwise - Clockwise rotation looking downwind? (Used only for FF binary files w/ BLADED)
0 ScaleIEC - Scale IEC turbulence models to exact target std deviation? [0=none;1=hub;2=all]

--------Turbine/Model Specifications-----------------------
13 NumGrid_Z - Vertical grid-point matrix dimension
13 NumGrid_Y - Horizontal grid-point matrix dimension
0.05 TimeStep - Time step [s]
600 AnalysisTime- Length of analysis time series [s] (program will add time if necessary)
"ALL" UsableTime - Usable length of output time series [s] (GridWidth/MeanHHWS s added if not "ALL")
84.30 HubHt - Hub height [m] (should be > 0.5*GridHeight)
80.00 GridHeight - Grid height [m]
80.00 GridWidth - Grid width [m] (should be >= 2*(RotorRadius+ShaftLength))
0 VFlowAng - Vertical mean flow (uptilt) angle [degrees]
0 HFlowAng - Horizontal mean flow (skew) angle [degrees]

--------Meteorological Boundary Conditions-------------------

"IECKAI" TurbModel - Turbulence model (see Table 4 for valid codes)
"unused" UserFile - Name secondary input file for user-defined spectra or time series inputs
"1-ED3" IECstandard - Number of the IEC standard (61400-x, x=1,2,3) with optional 61400-1 ed. number
"B" IECturbc - IEC turbulence characteristic ("A", "B", "C" or TI in %) or KHTEST
"NTM" IEC_WindType- IEC turbulence type ("NTM", "xETM", "xEWM1", or "xEWM50" for x=class 1, 2, or 3)
default ETMc - IEC Extreme turbulence model "c" parameter [m/s] (or "default")
"PL" ProfileType - Wind profile type (see Table 6 for valid codes)
"unused" ProfileFile - Name of the file that contains user-defined input profiles
84.30 RefHt - Height of the reference wind speed [m]
18.2 URef - Mean wind speed at the reference height [m/s] [must be 1-hr mean for API model]
450 ZJetMax - Height of the low-level jet [m] (70-490 m or "default", only for "JET" profile)
default PLExp - Power law exponent (or "default")
default Z0 - Surface roughness length [m] (or "default")

--------Non-IEC Meteorological Boundary Conditions------------

default Latitude - Site latitude [degrees] (or "default")
0.05 RICH_NO - Gradient Richardson number [-]
default UStar - Friction or shear velocity [m/s] (or "default")
default ZI - Mixing layer depth [m] (or "default")
default PC_UW - Hub mean u'w' Reynolds stress [m^2/s^2] (or "default" or "none")
default PC_UV - Hub mean u'v' Reynolds stress [m^2/s^2] (or "default" or "none")
default PC_VW - Hub mean v'w' Reynolds stress [m^2/s^2] (or "default" or "none")

--------Spatial Coherence Parameters----------------------------

default SCMod1 - u-component coherence model ("GENERAL","IEC","API","NONE", or "default")
default SCMod2 - v-component coherence model ("GENERAL","IEC","NONE", or "default")
default SCMod3 - w-component coherence model ("GENERAL","IEC","NONE", or "default")
default InCDec1 - u-component coherence parameters [-, m^-1] ("a b" in quotes or "default")
default InCDec2 - v-component coherence parameters [-, m^-1] ("a b" in quotes or "default")

71
default InCDec3 - w-component coherence parameters [-, m^-1] ("a b" in quotes or "default")
default CohExp - Coherence exponent for general model [-] (or "default")

--------Coherent Turbulence Scaling Parameters-------------------

"M:\coh_events\eventdata" CTEventPath - Name of the path where event data files are located
"Random" CTEventFile - Type of event files ("LES", "DNS", or "RANDOM")
True Randomize - Randomize the disturbance scale and locations? (true/false)
1.0 DistScl - Disturbance scale (ratio of wave height to rotor disk).
0.5 CTLy - Fractional location of tower center from right to L of dataset looking downwind
0.5 CTLz - Fractional location of hub height from the bottom of the dataset
30.0 CTStartTime - Minimum start time for coherent structures in RootName.cts [s]

Figure B-1. Sample TurbSim input file for IEC turbulence: parameters shown in blue should be
changed based on the turbine configuration; parameters shown in red should be changed for
each load case and simulation. (Note: figure is continued from previous page.)

72
Appendix C: Flow Charts

TurbSim Input Frequency Time

File
Velocity spectra
IFFT
Zero-mean time series
Runtime Spatial coherence
Options
Phases Optional scaling (for
IEC statistics or cross-
component correlation)
Turbine/Model
Specifications
Background time series

Random Variates
Meteorological
Boundary
Conditions Coherent structures

Mean Wind Profiles

Non-IEC Wind speed
Meteorological
Boundary
Wind direction
Conditions

Coherent Space
Turbulence Output Files
Scaling (y,z) position for each
Parameters simulated point (grid,
hub, and tower)

Assumed turbine /
rotor size

Figure C-1. Overview of the TurbSim simulation method; blue lines indicate processes influenced
by input-file parameters; black lines indicate internal variables and processes

73
TurbSim Input Frequency Time
File Velocity spectra
Zero- Optional scaling Back-
IFFT mean (for IEC statistics or ground Coherent
Spatial time cross-component time structures
Runtime series correlation) series
coherence
Options
Phases Mean Wind Profiles
RandSeed1 Space
position of simulation Wind Speed Direction
points (grid, hub, tower)
RandSeed2 Random Variates
Assumed turbine size Output Files
WrBHHTP Hub-height parameters, binary (*.bin) Hub

WrFHHTP Hub-height parameters, text (*.dat) Hub

WrADHH AeroDyn hub-height wind, text (*.hh) Hub

WrADFF TurbSim full-field binary (*.bts) Grid, Tower

WrBLFF BLADED full-field binary (*.wnd) Grid

Tower
WrADTWR Tower binary (*.twr)

WrFMTFF Full-field, text (*.u, *.v, *.w) Grid, Hub

WrACT Coherent turbulence, text (*.cts)

Clockwise Summary file, text, always

generated (*.sum)
ScaleIEC

Figure C-2. Parameters in the Runtime Options section of the input file

74
 TurbSim Input Frequency Time
File
Velocity spectra Zero-mean time
IFFT
Turbine/ series
Model Specs Spatial coherence

Phases Optional scaling (for

IEC statistics or
NumGrid_Z
cross-component
correlation)

NumGrid_Y
Background time
Random Variates series
TimeStep

Mean Wind Coherent structures

AnalysisTime
Profiles
Wind speed
UsableTime

Wind direction
HubHt

Output Files
GridHeight Space
(y,z) position for
each simulated
GridWidth point (grid, hub,
and tower)

VFlowAng Assumed turbine /

rotor size

HFlowAng

Figure C-3. Parameters in the Turbine/Model Specifications section of the TurbSim input file

75
 TurbSim Input Frequency Time
File
Velocity spectra Zero-mean time
Meteorological IFFT
series
Boundary Spatial
Conditions coherence
Optional scaling (for
Phases IEC statistics or
TurbModel
cross-component
correlation)
IECstandard
Random Background time
series
IECturbc Variates

IEC_Wind
Coherent structures
Type
Mean Wind
ETMc
Profiles
Wind speed
WindProfile
Type Wind direction
IEC, PL only
IEC, LOG only

RefHt
Output Files
Space
URef (y,z) position for
each simulated
point (grid, hub,
ZJetMax and tower)

Assumed
PLExp turbine / rotor
size

Z0

Figure C-4. Parameters in the Meteorological Boundary Conditions section of the TurbSim input
file (for IECKAI and IECVKM models only)

76
 TurbSim Input Frequency Time
File
Velocity spectra Zero-mean time
Meteorological IFFT
series
Boundary Spatial
Conditions coherence
Optional scaling (for
Phases IEC statistics or
TurbModel
cross-component
correlation)
IECstandard
Random Background time
IECturbc series
Variates
(KHTEST)

IEC_Wind
Coherent structures
Type KHTEST only

ETMc
Mean Wind
Profiles
WindProfile
Type Wind speed
IEC, PL only
IEC, LOG only

RefHt Wind direction

Output Files
JET only

URef Space
(y,z) position for
ZJetMax each simulated
point (grid, hub,
and tower)
PLExp
Assumed
turbine / rotor
size
Z0

Figure C-5. Parameters in the Meteorological Boundary Conditions section of the TurbSim input
file (for models other than IECKAI and IECVKM)

77
TurbSim Input Frequency Time
File
Velocity spectra Zero-mean time
Non-IEC IFFT
series
Meteorological Spatial
Boundary coherence
Conditions Optional scaling (for
Phases IEC statistics or
Latitude cross-component
correlation)

RICH_NO
Random Background time
series
Variates
UStar
WF_07D only
Coherent structures
ZI
JET only

PC_UW Mean Wind

JET, LOG Profiles
only
Wind speed
PC_UV

Wind direction
PC_VW Output Files

Space
IncDec1
(y,z) position for
each simulated
IncDec2 point (grid, hub,
and tower)

Assumed
IncDec3
turbine / rotor
size
CohExp

Figure C-6. Parameters in the Non-IEC Meteorological Boundary Conditions section of the
TurbSim input file

78
 TurbSim Input File Default Input Values
Turbine Specs
HubHt
Meteorological Boundary
Conditions
TurbModel

IECstandard

IECturbc Meteorological
Boundary
IEC_WindType Conditions

ETMc ETMc = 2 m/s

WindProfileType WindProfileType Random

Distributions
RefHt
PDFs from
URef URef (JET only) site-specific
data analysis
ZJetMax ZJetMax

PLExp PLExp

Z0 Z0
Background
Non-IEC Conditions Non-IEC Conditions
Information
Latitude Latitude = 45° Height
RICH_NO Rotor-disk
shear
UStar UStar
uhub
ZI ZI

PC_UW PC_UW
Site-specific
models only

PC_UV PC_UV

PC_VW PC_VW

IncDec1 IncDec1

IncDec2 IncDec2

IncDec3 IncDec3

CohExp CohExp = 0

Figure C-7. Default input values for the for the Meteorological Boundary Conditions and Non-IEC
Meteorological Boundary Conditions sections of the TurbSim input file

79
 75% centered and covering rotor disk
12.5% each on upper or lower half of rotor

TurbSim Input File Coherent Structures Background data

Coherent
Turbulence Shear across
Generate CTS file?
Scaling billow
Parameters

CTEventPath w-component
Type of structure standard
(DNS/LES) deviation at the
CTEventFile hub
(not GP_LLJ)
Randomize Space scale
and location relative to
not GP_LLJ Average u at the
DistScl
rotor disk or WF_07D billow center

CTLy
Interarrival time Height at the
billow center
CTLz (GP_LLJ only)

CTStartTime Start time of each

coherent structure Random
Runtime Options Distributions
WrACT
Which piece(s) of KH
Uniform
billow to add
Non-IEC
Meteorological
Boundary
Expected length of
Conditions
coherent structures Exponential
RICH_NO (time) in record
UStar
(WF_07D, PDFs from site-
GP_LLJ Velocity scale
not WF_UPW specific data
only) (from peak CTKE) or WF_14D analysis

Not GP_LLJ

Figure C-8. Parameters for coherent structures and the Coherent Turbulence Scaling Parameters
section of the TurbSim input file; the SMOOTH model uses the GP_LLJ scaling

80
Appendix D: Full-Field TurbSim Binary File Format
Table D-1. Full-Field TurbSim Binary File Header Format

Type (Bytes) Parameter Description

Identifies the file as a TurbSim binary file. ID should

Integer (2) ID
have the value 7 (not periodic) or 8 (periodic).
Integer (4) NumGrid_Z The number of grid points in the vertical direction.
Integer (4) NumGrid_Y The number of grid points in the horizontal direction.
Integer (4) ntower The number of tower points below the grid.

Integer (4) nt The number of time steps.

The distance in meters between two adjacent points
Real (4) dz
in the vertical direction.
The distance in meters between two adjacent points
Real (4) dy
in the horizontal direction.
Real (4) TimeStep The time in seconds between consecutive grids.

Real (4) uhub The mean wind speed in m/s at hub height.
Real (4) HubHt The height in meters of the hub.
Real (4) Zbottom The height in meters of the bottom of the grid.

for i = 1, 2, 3
th 4
The slope used to scale the i velocity component
Real (4) Vslope(i)
from 4-byte reals into 2-byte integers.
th
The intercept used to scale the i velocity
Real (4) Vintercept(i) 4
component from 4-byte reals into 2-byte integers.
end i

The number of characters in the ASCII string that

gives the TurbSim version number, date, and time
Integer (4) ncharacters
the file was generated. This number is no larger
than 200.
for i = 1, 2, … ncharacters
th
The ASCII integer representation of the i character
of the string that gives the TurbSim version number,
Integer (1) Characteri
date, and time the file was generated.
ACHAR(Characteri) gives the character.
end i

4
The three wind components are defined as U = 1, V = 2, and W = 3.

81
 Table D-2. FF TurbSim Binary File Grid Format

Type (Bytes) Parameter Description

for it = 1, 2, … nt
for iz = 1, 2, … NumGrid_Z
for iy = 1, 2, … NumGrid_Y
for i = 1, 2, 3
th 5
The normalized i velocity component of the
Integer (2) Vgrid_norm(i,iy,iz,it) wind speed at time step, it, and grid location
(y(iy), z(iz)).
end i
end iy
end iz

for iz = 1, 2, … ntower
for i = 1, 2, 3
th
The normalized i -component5 of the wind
Integer (2) Vtower_norm(i,iz,it)
speed at time step, it, and tower height, ztower(iz).
end i
end iz
end it

To convert the normalized wind in the FF TurbSim binary file to velocities in units of meters per
second, use the following equations:

Vgrid _ norm ( i, iy, iz , it ) − Vintercept ( i )

Vgrid ( i, iy, iz , it ) = (D-1)
Vslope ( i )

Vtower _ norm ( i, iz , it ) − Vintercept ( i )

Vtower ( i, iz , it ) = (D-2)
Vslope ( i )
.
The corresponding lateral locations, Y, and vertical locations, Z, of the grid and/or tower points
are given in units of meters by

Ygrid ( iy ) =
−0.5 ( NumGrid _ Y − 1) dy + ( iy − 1) dy
(D-3)
Z grid ( iz=
) Zbottom + ( iz − 1) dz
and

Ytower = 0
. (D-4)
Z tower ( iz=
) Zbottom − ( iz − 1) dz

5
The three wind components are defined as U = 1, V = 2, and W = 3.

82
Appendix E: Full-Field Bladed-Style Binary File Format
Table E-1. Full-Field Bladed-Style Binary File Header Format

Type (Bytes) Parameter Description

Identifies the file as a Bladed-style binary file. ID should have

Integer (2) ID
the value -99.
Integer (2) ID2 ID2 should have the value 4 to include the next 7 parameters.
Integer (4) nc The number of wind components. nc should be 3.

Real (4) Latitude This value is not used in AeroDyn.

Real (4) Z0 The surface roughness. This value is not used in AeroDyn.
Real (4) Ztmp The height at the center of the grid, in meters.

Real (4) 100 * TI(u) The turbulence intensity of the u component, in percent.
Real (4) 100 * TI(v) The turbulence intensity of the v component, in percent.
Real (4) 100 * TI(w) The turbulence intensity of the w component, in percent.

Real (4) dz The grid spacing in the vertical direction, in meters.

Real (4) dy The grid spacing in the lateral direction, in meters.
Real (4) uhub * TimeStep The longitudinal grid resolution, in meters.

Integer (4) nt / 2 Half the number of points in the longitudinal direction.

Real (4) uhub The mean wind speed (in meters per second) at hub height.
Real (4) Unused The value 0. This parameter is not used in AeroDyn.

Real (4) Unused The value 0. This parameter is not used in AeroDyn.
Real (4) Unused The value 0. This parameter is not used in AeroDyn.
Integer (4) Unused The value 0. This parameter is not used in AeroDyn.

Integer (4) RandSeed1 This value is not used in AeroDyn.

Integer (4) NumGrid_Z The number of grid points vertically.
Integer (4) NumGrid_Y The number of grid points laterally.

Integer (4) Unused The value 0. This parameter is not used in AeroDyn.
Integer (4) Unused The value 0. This parameter is not used in AeroDyn.
Integer (4) Unused The value 0. This parameter is not used in AeroDyn.

Integer (4) Unused The value 0. This parameter is not used in AeroDyn.
Integer (4) Unused The value 0. This parameter is not used in AeroDyn.
Integer (4) Unused The value 0. This parameter is not used in AeroDyn.

83
 Table E-2. Format of Grid Velocities in Full-Field Bladed-Style Binary File Format

Type (Bytes) Parameter Description

for it = 1, 2, … nt
for iz = 1, 2, … NumGrid_Z
for iy = 1, 2, … NumGrid_Y
The normalized U component of the wind speed
Integer (2) U grid _ norm ( iy, iz , it )
at time step, it, and grid location (y(iy), z(iz)).
The normalized V component of the wind speed
Integer (2) Vgrid _ norm ( iy, iz , it )
at time step, it, and grid location (y(iy), z(iz)).
The normalized W component of the wind speed
Integer (2) Wgrid _ norm ( iy, iz , it )
at time step, it, and grid location (y(iy), z(iz)).
end iy
end iz
end it

To convert the normalized wind in the FF Bladed-style binary file to velocities in units of meters
per second, use the following equations:

 TI ( u ) 
=U grid ( iy, iz , it ) uhub  U grid _ norm ( iy, iz , it ) + 1 , (E-1)
 1000 

 TI ( v ) 
Vgrid ( iy, iz , it ) = uhub  Vgrid _ norm ( iy, iz , it )  , (E-2)
 1000 
and

 TI ( w ) 
Wgrid ( iy, iz , it ) = uhub  Wgrid _ norm ( iy, iz , it )  . (E-3)
 1000 
Here TI represents the turbulence intensity as a decimal, not a percentage.

The corresponding vertical locations, Z, of the grid points are given in units of meters by

Z grid ( iz ) =
−0.5 ( NumGrid _ Z − 1) dz + HubHt − HeightOffset + ( iz − 1) dz (E-4)

using values of HubHt and HeightOffset from the summary file. The lateral locations, Y, of the
grid points depend on the input value Clockwise (read from the summary file) and are given by

−0.5 ( NumGrid _ Y − 1) dy + ( iy − 1) dy, Clockwise =

 FALSE
Ygrid ( iy ) =  . (E-5)
 0.5 ( NumGrid _ Y − 1) dy − ( iy − 1) dy, Clockwise =
 TRUE

84
Appendix F: Tower Data Binary File Format
TurbSim tower files have a “.twr” extension. Each file contains a header of 4-byte real and
integer values, followed by 2-byte integer time series of the three wind components at each point
on the tower grid. The wind components at the tower points are normalized and stored in 2-byte
binary integers, exactly the same way that Bladed-style full-field wind files are written. The
tower files have the same vertical resolution as the full-field grid, with points going downward
from the bottom of the full grid in a single line at the tower centerline.

Table F-1: Format of Header in TurbSim Binary Tower-Data File

Type
(Bytes) Parameter Description

Real (4) dz Vertical grid resolution, in meters.

Real (4) uhub * TimeStep Longitudinal grid resolution, in meters.
Real (4) Zmax The height of the highest tower point, in meters.

Real (4) nt The number of points in the longitudinal direction.

Real (4) nz The number of vertical tower points.
Real (4) uhub The mean wind speed, in meters per second.

Real (4) 100 * TI(u) The turbulence intensity of the u component, in percent.
Real (4) 100 * TI(v) The turbulence intensity of the v component, in percent.
Real (4) 100 * TI(w) The turbulence intensity of the w component, in percent.

Table F-2: Format of Grid Velocities in TurbSim Binary Tower-Data File

For each increasing time step (nt points),and starting at the top of the grid, going downward (nz
points) the data are stored as:
Type (Bytes) Parameter Description
for it = 1, 2, … nt
for iz = 1, 2, … nz
Normalized U component of the wind speed at
Integer (2) U tower _ norm ( iz , it ) time step, it, and height z(iz).
Normalized V component of the wind speed at
Integer (2) Vtower _ norm ( iz , it ) time step, it, and height z(iz).
Normalized W component of the wind speed
Integer (2) Wtower _ norm ( iz , it ) at time step, it, and height z(iz).
end iz
end it

85
To convert the normalized wind in the tower data binary file to velocities in units of meters per
second, use the following equations:

 TI ( u ) 
=U tower ( iz , it ) uhub  U tower _ norm ( iz , it ) + 1 , (F-1)
 1000 

 TI ( v ) 
Vtower ( iz , it ) = uhub  Vtower _ norm ( iz , it )  , (F-2)
 1000 
and

 TI ( w ) 
Wtower ( iz , it ) = uhub  Wtower _ norm ( iz , it )  . (F-3)
 1000 
Here TI represents the turbulence intensity as a decimal, not a percentage.

The corresponding lateral locations, Y, and vertical locations, Z, of the tower points are given in
units of meters using values of Zmax from the file header:

Ytower = 0
(F-4)
Z tower ( iz ) = Z max − ( iz − 1) dz
.
.

86
Appendix G: Velocity Spectra Comparison Plots

Default UStar (m/s)

SMOOTH 0.668
NWTCUP 0.779
GP_LLJ 0.550
WF_UPW 0.728
WF_07D 1.395
WF_14D 1.221

Figure G-1. Neutral velocity spectra for the 8 spectral models available in TurbSim, using a 15 m/s
rd
wind speed at 80 m; IECKAI and IECVKM use NTM category “B” and 61400-1 3 ed. scaling; the
non-IEC models use RICH_NO = 0 and UStar = “default”

87
Figure G-2. Neutral velocity spectra for the 8 spectral models available in TurbSim, using a 15 m/s
rd
wind speed at 80 m; IECKAI and IECVKM use NTM category “B” and 61400-1 3 ed. scaling; the
non-IEC models use RICH_NO = 0 and UStar = 1.1 m/s

88
 Default UStar (m/s)

SMOOTH 0.644
NWTCUP 0.716
GP_LLJ 0.495

WF_UPW 0.709
WF_07D 1.344
WF_14D 1.184

Figure G-3. Stable velocity spectra using a 15 m/s wind speed at 80 m; the non-IEC models use
RICH_NO = 0.05 and UStar = “default”; The IEC models, which are neutral (RICH_NO = 0), were
rd
added for reference; they use NTM category “B” and 61400-1 3 ed. scaling

89
Figure G-4. Stable velocity spectra using a 15 m/s wind speed at 80 m; the non-IEC models use
RICH_NO = 0.05 and UStar = 1.1 m/s; the IEC models, which are neutral (RICH_NO = 0), were
rd
added for reference; they use NTM category “B” and 61400-1 3 ed. scaling

90
 Default UStar (m/s)

SMOOTH 0.656
NWTCUP 0.741
GP_LLJ 0.537
WF_UPW 0.917

WF_07D 1.485
WF_14D 1.304

Figure G-5. Unstable velocity spectra using a 15 m/s wind speed at 80 m; the non-IEC models use
RICH_NO = -0.05 and UStar = “default”; the IEC models, which are neutral (RICH_NO = 0), were
rd
added for reference; they use NTM category “B” and 61400-1 3 ed. scaling

91
Figure G-6. Unstable velocity spectra using a 15 m/s wind speed at 80 m; the non-IEC models use
RICH_NO = -0.05 and UStar = 1.1 m/s; the IEC models, which are neutral (RICH_NO = 0), were
rd
added for reference; they use NTM category “B” and 61400-1 3 ed. scaling

92
Appendix H: Sample AeroDyn v13 Coherent
Turbulence Parameter Input File

Example Coherent Turbulence Parameter input file (TurbSim_AD.ctp). Valid with AeroDyn 12.57.
# Parameters that can vary from one turbine simulation to the next:
"H:\x90_i16" | CTSpath - Path to coherent turbulence data files
"TurbSim.cts" | CTTSfile - File containing time steps of the coherent turbulence event files
"TurbSim.wnd" | CTbackgr - Name of file containing background wind data (quoted string)
1 | CT_DF_Y - Decimation factor for wind data in the y direction
1 | CT_DF_Z - Decimation factor for wind data in the z direction

==================================================
NOTE: Do not add or remove any lines in this file!
==================================================
For decimation factors, 1 = use every point, 2 = use every other point, etc.

93