Report Linear Model Uncertainty – this is a self-contained implementation of the uncertainty model proposed by Clerk, Stuart and Anderson of RES. It combines the correlated and uncorrelated uncertainties contained in the met mast objects as part of the met data uncertainty described below, but it is independent of any other settings described in the section. It is possible that this method may be modified and incorporated as an option to specify wind-flow modeling uncertainty in the UL method in a future update.
Report Energy Sensitivity - an alternative to calculating the uncertainty in Openwind is to simply calculate the sensitivity of the energy capture routine to a percentage change in either the free or waked wind speed.
Report UL Model Uncertainty Estimates – this model allows the user to combine loss uncertainty, wake model uncertainty, wind-flow modelling uncertainty and met data uncertainty together with inter-annual variability to produce a table of net energy estimates that vary with both period and probability of exceedance.
The implementation of uncertainty in Openwind takes three forms: loss uncertainty including the uncertainty in the wake model as well as taking account of inter-annual variability and exceedance values; met data uncertainty; and the wind-flow modelling uncertainty.
UL Uncertainty Model cycles through the various evaluation periods (annual, 1st year, evaluation period 1, …, evaluation period N) and probability of exceedance values (P75, P84, …., P99) and for each turbine calculates the following:
•The number of standard deviations, Ns, to use for this probability of exceedance level
•For each calculation step, the difference in energy production, Lv, due to reducing the free velocity by the uncertainty in the wind speed multiplied by Ns
•For each calculation step, the difference in energy production, Lpc, due to the uncertainty in the power curve multiplied by the gross power for this calculation step, multiplied by Ns
•The wake loss factor, Fw, which is the wake loss uncertainty in % multiplied by Ns
•The performance loss factor, Ff, which is the performance loss uncertainty in % multiplied by Ns
•The availability loss factor, Fa, which is the availability loss uncertainty in % multiplied by Ns
•The electrical loss factor, Fe, which is the electrical loss uncertainty in % multiplied by Ns
•The environmental loss factor, Fg, which is the environmental loss uncertainty in % multiplied by Ns
•The curtailment loss factor, Fc, which is the curtailment loss uncertainty in % multiplied by Ns
•The wind speed frequency factor, Fws, which is the wind speed frequency uncertainty from the met mast which is used for this turbine, multiplied by Ns
Lv and Lpc are both summed over all calculation steps and divided by the gross energy, Eg, for each turbine.
If any of the loss uncertainties are expressed as a percentage of the loss (rather than as a percentage of net energy), the corresponding factors are converted to an uncertainty in net energy by multiplying by that loss as a fraction of the net energy as calculated before application of that loss.
All the loss factors are then combined as the square root of the sum of the squares for each turbine. The individual turbine uncertainty. The uncertainty reported for each turbine is that calculated for the annual case and one standard deviation.
The total uncertainties per site are combined by weighting the individual turbine uncertainties by the P50 net energy of that turbine divided by the total P50 net energy for that site layer. All uncertainties are treated as or converted to fractions of the net energy.
In the reported uncertainty table, the total wind resource uncertainty is the square root of the sum of the squares of the:
•Field verification uncertainty
•Measurement uncertainty
•Uncertainty due the MCP process in estimating the long-term average
•Uncertainty due to the variability in the wind over the evaluation period
•Wind shear uncertainty
•Wind-flow modelling uncertainty
The wind speed frequency distribution loss uncertainty, is calculated using the number of measurement years, Ym, at the mast and the number of years in the evaluation period, Yev, as shown below:
The total plant losses uncertainty is the square root of the sum of the squares of the uncertainty in the net energy due to:
•Wake losses
•Availability losses
•Performance losses
•Environmental losses
•Electrical losses
•Curtailment losses
The power curve uncertainty is given by the term Lpc defined above and summed over all calculation steps.
The total energy production uncertainty can be seen as being equal to the square root of the sum of the squares of:
•Total wind resource uncertainty (calculated taking account of power curve sensitivity)
•Wind speed frequency distribution uncertainty
•Total plant loss uncertainty
•Power curve uncertainty
For any given evaluation period:
Met Data uncertainty is input per met mast (see section on MetMastLayers) and is intended to include the uncertainty in the wind frequency table arising from the following:
•Site documentation and verification uncertainty – a somewhat subjective measure of the quality and reliability of the data herein
•Wind speed measurement uncertainty - the uncertainty in the original measurements due to calibration tolerances, met mast orientation, installation and instrument degradation
•Long-term average speed uncertainty (correlated) - the uncertainty in the long-term wind speed due to the MCP process and which is common to all masts in the project.
•Long-term average speed uncertainty (uncorrelated) – the uncertainty in the long-term wind speed due to the MCP process and which is peculiar to this met mast (perhaps due to the length of record at this mast and/or the quality of the correlation with the primary site mast)
•Shear uncertainty – the uncertainty in the long-term wind speed introduced by shearing from the top measurement height to this mast height. If no shearing was done because this height is within the range of the original met mast instrument heights then this can be set to zero.
A met mast object in Openwind can contain data at multiple heights, combining different processes and so the above fields are specified for every height in the met mast object.
In addition to the above, the number of measurement years are required. This should be the number of years of valid measurements before any data reconstruction and does not need to be an integer number of years.
Wind-Flow Modelling Uncertainty is the uncertainty introduced by the modelling process as we attempt to extrapolate the wind regime from the mast positions into unmonitored areas of the site. Of these, the simplest is to use the first option to set the wind-flow modelling uncertainty to a fixed value for all turbine positions. The second and third options allow the wind-flow modelling uncertainty to vary with the turbine location, relative to the met mast location. These options are described below.
UL has carried out an inter-mast analysis across several sites using Sitewind. The resulting modelling uncertainty formula is available for use in Openwind. However, this formula only applies to wind maps or WRGs created with the Sitewind method.
Modelling uncertainty may be linked to many different metrics depending on the wind-flow method used. Openwind provides the capability to define a modelling uncertainty function based on the following inputs:
•Distance from the mast
•Directional difference between the met mast location and each location in the WRG
•Topographic difference between the met mast location and each location in the WRG
•User input uncertainty rasters (2D or 3D)
•Ruggedness of the terrain (RIX) by conversion to an uncertainty raster
The modelling uncertainty does not take into account uncertainties in the wake models used—only in the variation of the wind resource across the site.
The topographic difference (TD) is the frequency-weighted sum of the squares of the deviations from unity of the speed-ups calculated relative to the mast location. Which is to say, the more the mean wind speeds at a location vary from the mast location, the greater TD becomes.
For the purposes of this graphic:
•TD is the topographic difference
•n is the number of directions
•P is the probability at the point of the wind coming from direction i
•Ui,point is the mean wind speed at the point from direction i
•Ui,mast is the mean wind speed at the mast from direction i
The directional difference (DD) is the sum of the differences in probability for all directions.
For the purposes of this graphic:
•DD is the directional difference
•n is the number of directions
•P is the probability at the point of the wind coming from direction i
The uncertainty capability in Openwind does not adjust the energy losses. These must be adjusted by the user.
If you are using data from Sitewind, you can use the UL uncertainty function. Otherwise, the uncertainty function can be defined using the modelling uncertainty dialog shown in figure 152
One can make a variety of uncertainty functions with components relating to distance, directional difference, topographic difference and uncertainty input in the form of raster layers or 3D raster layers (see WRB format). The terms can be manipulated individually before being combined into the uncertainty function.
The uncertainty function can be used to blend the influence of several met masts by right-clicking the WRG and choosing "Adjust to Masts", then choosing "Modelling Plus Measurement Uncertainty (see white paper)”. The resulting blended WRG will tend to have a lower uncertainty1 than that predicted by the uncertainty equation used to generate it and for this reason a new uncertainty raster is generated at the same time. This raster 3D gives the uncertainty that should be used in conjunction with the PXX energy capture. In order to make use of this combined uncertainty estimate, one must check the option "Use uncertainty rasters from blended WRGs when available or calc from single WRG" in the uncertainty settings.
The easiest way to create and use uncertainty rasters generated from an uncertainty weighted adjust to mast is to setup the uncertainty similarly to that shown above then use uncertainty weighted adjust to mast with option to automatically create validity layers and WRG copies.
At the bottom of the uncertainty dialog is an option to include the uncorrelated met mast uncertainty when adjusting to masts. It could be argued that this is the more correct way of running an uncertainty-weighted adjust to masts. However, this method tends to result in the WRG not matching the met mast wind speeds and frequency distribution at the met mast location. This is because it is effectively allowing met masts with high uncertainty to be overruled by other masts or at least it represents a consensus view of the wind resource at each mast location. Elsewhere in Openwind we tend to assume that the WRG matches the met mast data at the met mast location and so this option is not checked by default. It should be noted that whether this option is check or not, the resulting uncertainty raster remains the same and that this therefore has a limited impact on the uncertainty estimates.