Morten

Hi VareKey, The REWC files derived from reanalysis data should be used with care. My colleagues working with mesoscale models warn about problems in complex terrain, very high roughness, and areas with tropical cyclones. So it is a good idea to make comparisons with statistics based on high-quality data from similar projects in the region you are operating in. Unfortunately, the REWC file format has changed from WEng2 to WEng3. I have been assigned the task of updating our reanalysis-based REWC files to the new format but unfortunately, I have been working with other projects. We are implementing an alternative method based on the peak-over –threshold approach for the WEng 3.1 upgrade version,which should be available very soon. This will allow you to prepare EWC statistics from time series of short duration, but the statistical uncertainty will be very high – there is no way to change that. Best regards, Morten

Hi VareKey, Unfortunately, we do not have an EWC estimator for WEng3. The extreme-wind model in WEng3 depends on observation of extreme events rather than fitted distributions. The best way to generate this input (*.OEWC files) is with the new WACA2. You can also input WEng2-style (*.OEWC) files generated by WACA1, but if you still have the data it is better to recalculate the *.OEWC files with WACA2, as then you will model effects of terrain-induced flow direction changes more correctly. Unfortunately, the old (*.EWC) format is not supported in WEng3 and this is why you are having problems using the output from the EWC estimator or REWC files generated from this. The EWC estimator predicted extreme-wind Gumbel distributions and stored the parameters in the OEWC file instead of the observed winds required by WEng3, so to make it work we would have to make a simulation pseudo-random “observations”. It could be done, but the results would be random. I remember that we discussed the problem during the development of WEng3 and decided against it. Maybe you need this tool because you only have a short time series available. In that case it might be a possibility to use data from a remote site and then translate that to regional statistics, see http://www.wasp.dk/Products/WEng/ExtremeWinds.aspx We are also developing a peak-over-threshold method for WEng3 and it should be available quite soon. Cheers, Morten

Hi Timm, You got it all right, except the last point. The results in the *.wtr file are for the wind speed you select in the turbulence intensity rose script, and it may differ from 15 m/s depending on your choice. The results of this script are calculated by the Mann turbulence model and not the formula you quote. This is used in of the methods for the mean TI inside WAT. It is actually not the preferred method if you have access to data from a reference mast. Best regards, Morten

Hi Jan, WAT is using rather crude estimates for this. One method is to use a modified version of the IEC NTM model. Another one is to use measured statistics. Please read the WAT help file section called "modeling with WAT| ambient turbulence models". Regards, Morten

Hi Giuseppe, WAT is using the WEng turbulence estimates to calculate the the special turbulence measure needed for an IEC site assessment. Read more on this in my post today at http://www.wasptechnical.dk/forum/viewtopic.php?pid=1455#p1455 Best regards, Morten

Hi Timm, The report you have , Ris-R-900(EN), is a good introduction to the WEng flow model, LINCOM. In addition, I recommend Ris-R-1107(EN) for details on how WEng models surface roughness over water and Ris-R-1179(EN), chapter 2, for details on how the flow equations are solved. As you say, WEng uses the LINCOM flow results as input to the Mann turbulence model, based on rapid distortion theory. My favourite references to the turbulence model are [1] J Mann (2000) The spectral velocity tensor in moderately complex terrain, J. Wind Eng. Ind. Aerodyn. 94 (2006) 581–602 (explaining the model) and [2] J Mann (1998) Wind field simulation, Prob. Eng. Mech.13, 269-282 (explaining how simulation how the model is tuned to classic text-book spectra like the one of Kaimal and how to simulate pseudo-random time series). The WEng turbulence predictions are valid for ideal steady-state turbulence, i.e. without trends in mean speed and direction, for neutral atmospheric stability only and without wake effects from buildings and neighboring turbines. The main input is the terrain elevation and surface roughness maps. The modeled turbulent fluctuations over land are proportional to the mean wind speed and thus the modeled turbulence intensity is independent on wind speed. Over water surfaces there is a slight effect of the wind-speed dependent surface-roughness. The calculations in WAT adapt the WEng turbulence results to the IEC 61400-1 standard. It adds turbulence from neighboring turbines and calculates the so-called effective turbulence intensity, which is the single turbulence level giving the same fatigue damage as turbulent winds from all directions. This averaging depends on mean wind statistics, and that is why you need the *.lib wind atlas file. The turbulence level used by the IEC is actually not the mean turbulence or WEng, but the 90% level of the random distribution of turbulence intensity during random 10-min periods. WAT has various (crude) ways to correct the WEng turbulence predictions. You can either use 1) the same correction as in the IEC standard, or 2) observed turbulence statistics at a reference mast, or 3) supply mean and spread of the turbulence intensity modeled by your own model. See the WAT help file sections "modeling with WAT| ambient turbulence models" and "IEC 61400 standards| IEC 61400-1, Annex D - windfarm turbulence" for further details. Best regards, Morten Nielsen

A new version of the WAsP Map Editor version 11 is available from http://www.wasp.dk/ It has the following new features: - GIS-type coordinate meta-files for scaling/calibration of background maps are now supported. - BLN-map files to be immported - Shape files to be impported - Faster, more efficient check of LFR-errors

WAsP version 11.0 release C is now available from http://www.wasp.dk/ It fixes some problems regarding: - Wind farm in workspace-root obstructs associations between turbine/wind farms and vector map - With an open Wind atlas, inserting a met station causes an error - Expired temporary licence notification - Error uploading file - Cannot export workspace to WAsP 10 format - Vector map with no elevation - WAsP 11 needs to look for and see the WAsP Map Editor 11 - Copy- to clipboard problems in Reference site and Met station - Gaussian hill generator does not close the contours perfectly - Windows user names with Ä character cause licence registration failure - Change the mail in case of the message "calculation server access information is invalid..."

Hi Bepi, No not yet, but we have plans of using WACA for generating observed turbulence statistics as input for the WAT program. Best regards, Morten

Hi Bepi, I am not familiar with the WindPro turbulence model, but in WAsP Engineering the standard deviations of the wind speed is a model result and not an input. The input is a reference mean wind speed and direction plus the terrain elevation and surface roughness map. An important limitation of the WEng turbulence model is that it cannot model effects of variable stability or random effects of measurements with 10-min sample duration. This introduces some differences at low wind speeds. A good description of the WEng turbulence model is found in “Mann (2000) The spectral velocity tensor in moderately complex terrain, J. Wind Eng. Ind. Aerodyn. 94 (2006) 581–602” Best regards, Morten Nielsen

Hi Jan, There is no official release date for WAT 3.1, but I work on this as often as teaching and work on commercial projects allows me. I don’t like issuing too many beta versions either, because in my experience this is often confusing and complicates technical support. At the present stage I would also have to hide some half-finished or undocumented new features before compiling a new beta release. So for the moment you will have to live with the work-around method for the problem you mention, which is to export results for each site one by one. Sorry for the inconvenience. Best regards, Morten

Dear Jan, WAT has three methods for modelling representative ambient TI, defined as the 90% percentile of a typical scattered distribution. The simplest of these methods, and the first one published, is to give the IEC model the offset which makes it converge to the WEng estimate at very high wind speed. A quick-and-dirty fix is to change the input from WEng. This is feasible if you select the WEng output in Excel format and manually correct the data in the column called TI. NB: I am not really recommending this, so please read on for better solutions. The real problem is that the IEC model is designed to be conservative, and it is not quite suitable as a the most likely estimate for the ambient TI. It is probably better to use one of the other WAT methods. Here, you can 1) provide observed TI statistics at one or several reference masts and let WAT translate these conditions to your turbine sites by as described in the WAT help file, or 2) you can prescribe your own modelled TI statistics for each of the turbine sites. Getting observed or modelled TI stats into WAT 3.0 can be a hassle, so please contact WEng support if you have problems with this. Regards, Morten

Hi Francesco, I recomend the following references: - J Mann (1998) Wind field simulation, Prob. Eng. Mech. vol. 13, p. 269-282 - J Mann (2000) The spectral velocity tensor in moderately complex terrain, J. Wind Eng. Ind. Aerodyn. vol. 88, p. 581–602 - J C Kaimal and J J Finnigan (1994) Atmospheric Boundary Layer Flows, Oxford University Press. Mann (1998) explains the WEng turbulence model for flat terrain and Mann (2000) explains extensions for terrain effects. I suggest that you start by reading Mann (2000). Kaimal & Finnigan (1994) is my favorite text book on micrometeorology. It explains the physics, but unfortunately you will not find a simple formula for terrain effects on standard deviations of velocity perturbations. With best regards, Morten

Hi Francesco, I know this kind of scaling is used by load engineers to match the turbulence intensity of simulated series with prescribed values. The reason why statistics of the WEng simulated time series deviate from their target values is their limited duration. There are both systematic and random effects on the statistics, just as when you sample atmospheric turbulence with an anemometer. If you use individual scaling for the three velocity components, you will create divergence in the simulated turbulence field. You will have to ask a load engineer, but I guess that is only a minor problem. Length scales and correlations should not be affected by scaling the variances. I am more worried about your suggestion for the ratios of the standard deviations as they are valid for flat terrain only. It would be more correct to use the variances listed in the WEng ‘site view’ which include terrain effects. They are calculated by the smooth target spectra, not random simulations. Regards, Morten

Hi Petar, The sites are sorted multiple times - by name, wind directions, distance from reference site, et cetera. Unfortunately, these repeated sortings confused the report you mention. The solution was to introduce a separate site list for this specific report. It is working in my current developer version, so the problem should disapear in the comming version WAT 3.1. The release will be announced on wasp.dk. A temporary work-around solution is to export these tables for individual sites. Cheers, Morten

Hi Petar, 'p(u)' is our modelled site-specific wind speed distribution based on the Wind Atlas *.lib file you provide when setting up the WAT project. 'p(u)_IEC' is a turbine-specific wind speed distribution determined by the IEC turbine class. The combination of this distribution and some standard models for turbulence and wind shear was used by the manufaturer to verify turbine safety with respect to fatigue damages. For the site assessment we may assume safe turbine deployment when on-site turbulence intensity and wind-speed probability density are lower in the range of turbine operation that assumed for the turbine-classification load cases. Read more about this in the WAT help file or, better, in the IEC 61400-1 standard. Cheers, Morten

Hi Francesco, First I like to repeat that it is perfectly natural when spectra of simulated or measured times series derivate from those of the target process. This is a consequence of finite sampling time and time step. There are alternative methods where you get exactly the spectra you prescribe. They operate with complex input variables of fixed amplitude and random phase, instead of the complex Gaussian variables used in the Mann model. The stretching correction you describe sounds a little like that. It would correct spectra and probably only slightly corrupt cross-spectra and probability distribution. I don’t know exactly how the high frequency compensation is implemented. My guess is that Jacob Mann first analyze the smoothing effect theoretically, then apply a compensating correction to the target spectra, and finally simulate the wind field by the usual method. With best regards, Morten

Hi Francesco, As you mention, the high-frequency variations of the simulated wind field are normally reduced by the dicretization filter. The purpose of the ‘high-frequency compensation’ is to avoid this effect. The help file section ‘WAsP Engineering modeling| turbulence| IEC turbulence simulator’ shows some spectra simulated with and without this option. High-frequency compensation is relevant if you prefer realistic variations at each node in the computational mesh. On the other hand, you should not use it if you prefer fluctuations averaged over each mesh cell. High-frequency compensation will not normalize the spectra. In fact, most of the variance is associated with the energy at lower frequencies. The WEng simulations are like samples of a stochastic process. There will be random variations, but if you increase the duration of the sample or average multiple simulations you will approach the target spectral model. Normalization is often applied on wind fields fed to aeroelastic models. This is probably because the load simulations are done on relatively short time series. The best reference to the simulation method is "Mann J. (1998), Wind field simulation, in Probabilistic Engineering Mechanics, Vol. 13 (4), page 269-282". In here, you can also find references to the spectral models by Kaimal, ESDU and Simiu & Scanlan. Cheers, Morten

This is a fairly usual problem and can have several explanations. There are some guidelines at http://www.wasp.dk/Support/FAQ/WengMaps.aspx

Hi Madajavi, Perhaps you are refering to the WAsP homepage, which has a short explanation of this concept. I think something has been lost while formating the page for our new design. Please find a more complete version in the WAsP help file section 'modeling with WAsP| power production estimation| wind speed distributions'.

First of all, please note that the representative TI defined by the IEC standard is the 90% percentile of all observations including natural variation induced by variable atmospheric stability and unsteady wind speed. Thus representative TI is higher than mean TI. To model this effect, your CFD model needs to both model the mean TI and the standard deviation of the TI. WAT will then model the 90% percentile as Mean(TI) + 1.28 x StDev(TI) at each wind speed and direction. WAT makes a distinction between observed and TI statistics, modelled by e.g. a CFD model. Modelled TI statistics is added for each turbine site from the WAT main menu. Observed TI statistics is added to a TI reference mast and the deviation of the WEng TI predictions at the TI reference mast is used to correct the WEng estimates at the turbine sites. Normally you don't have observations at all wind speeds, and this is why WAT applies an extrapolation for high wind speeds. You could make this extrapolation yourself, and import the corrected statistics.

If you are a new user you probably have downloaded all the latest versions of our software. However, you might need a piece of software enabling the WEng2 script to talk invoke WAsP8/9/10. This was previously distributed with the WAT installation package, but I think it was excluded from the latest WAT 3.0 package. Try to first install WAT 2.5 from http://www.wasp.dk/Download/Software/Weng2_Installation.aspx and then install WAT 3.0 from http://www.wasp.dk/Download/Software/Weng3_Installation.aspx Sorry for the inconvenience.

WAT will apply the background turbulence intensity predicted by WAsP Engineering without corrections, if this option is deselected. This will result in a TI which is lower than the 'representative TI' defined in the IEC 61400-1 standard. Read more in the WAT help file sections 'IEC 61400 standards| IEC 61400-1 normal turbulence model' and 'Modelling wint WAT| Ambient Turbulence models'.

Hi Pedro, The way to do this is to 1) Insert a reference site for your windfarm and then define its position and measurement height 2) Recalculate the windfarm 3) Right-click the windfarm object and select 'show| WF power curve' 4) Click somewhere in the WF power curve, press ctrl+C to copy to clipboard, and paste data to an Excel worksheet or similar If you want to modify the resolution of the WF power curve, you can right-click the WAsP project object, select 'Edit project configuration| Turbine wake modelling', adjust the parameters, and recalculate the project. However, please bear in mind that this might slow-down or reduce the accuracy of future wake calculations. So you might want to reset these parameters after obtaining the WF power curve. When using the WF power curve, you should remember that it is based on the free wind at the reference site without wake effects from operating turbines. Read more in the help file under 'WAsP user guide| the hierarchy members| the wind farm window' Cheers, Morten

Hi Matt, The GUI version of the stand-alone program is available from http://www.wasp.dk/Products/WEng/IECTurbulenceSimulator.aspx (not *.htm) The command-line version is the oldest one, but we still offer it as an alternative because some people prefer to run multiple simulations from scripts. The reason why you cannot make "simEx1.inp" work is that you don't have a LINCOM output file (*.dat) feeding terrain effects to the simulation model. For this purpose I suggest to call the turbulence simulations from within WAsP Engineering. Detailed questions on the command-line version are best adressed directly to Jakob Mann, but please bear in mind that this model is now freeware and no longer covered by the WEng license. With best regards, Morten