Back to the programme printer.gif Print




Delegates are invited to meet and discuss with the poster presenters in this topic directly after the session 'How does the wind blow behind wind turbines and in wind farms?' taking place on Tuesday, 11 March 2014 at 16:30-18:00. The meet-the-authors will take place in the poster area.

Matthieu Boquet LEOSPHERE, France
Co-authors:
Paul MAZOYER (1) F P Matthieu Boquet (1) Mehdi Machta (1)
(1) LEOSPHERE, Orsay, France

Printer friendly version: printer.gif Print

Abstract

wind simulator : a coupled atmosphere lidar simulator for improvment of remote sensing.

Introduction

Simulation is key factor of success for improving products. From aeronautics to oil company, the first step of development is simulation. Our aim is, therefore, to develop a tool which permits to simulate the atmosphere and the measurement of the Lidar. It permits to perform R&D study case, to analyses system sensibility , to create proofs of concepts but also to develop new design for Lidar. Results of simulation can drive to new concept of Lidar with better performance. Nevertheless, simulation should always be followed by a real life test and validation is essential.



Approach

Lidar measurement contains several different physical phenomenon which we decided to modelise separately. First step is modelise of atmosphere which should be able to represent different phenomenon which occurs in the boundary layer (wind shear, turbulence, wakes vortex in wind farm, gust, low level jet,...).Then, we modelise the Lidar measurement which contains geometric and physical aspect, different parameters from Laser one to Detectors ones are available, backscatter and extinction parameters are also available; We allow the user to set whatever geometric configuration on the Lidar (number of beams, scanning speed,…) and we modelise a fine description of the optical process of Lidar. Finally, we modelise a very fexible algorithm of wind reconstruction which can be changed by the user easily. Different performance boards are available in the software and allow the user to get a complete bench of information.

Several simulation modelise has been developed from now on. Many of them uses Large Eddy Scale simulation (LES) which is impossible to run on a classis computer (important need of CPU time) and very few offers to modify all parameters of the Lidar measurement process especially turbulence ones ( [1], [2]). Challenge was to build a simulator that modelise all of the Lidar process so that coupling effect between issues can be modelise and a better understanding of Lidar performance is available. We describe here the design of wind simulator and each parts specification then we give an example of study carried out. Then a validation method of the simulator using real data is explained.


Main body of abstract

Wind simulator design :

Wind Field :
The aim is to represent all phenomenon which interact with Lidar measurement as wind shears, turbulence, rain or density of aerosol. We did focus on wind phenomenon which can be classified in two parts: unsteady and mean wind field which we can also identify to inner and outter scale looking at a 10 minutes simulation. Two algorithm are widely used for wind field simulation: large eddy scale simulation (LES) and Mann (or Sandia) modelise. LES resolves the energy containing scale of turbulent motion and modelise the turbulent motion of the inner scale. This technique permits to represent accurately the evolution of wind speed with a large scope of phenomenon (like wake, shears,…) and offers a good model of eddies dissipation in the inner scale. Nevertheless, LES algorithm are quite complicated to set and heavy to run. Using LES limits the number of case available and can’t be used on a classic computer. In figure 1, we can see a LES simulation of wake in a wind farm.

Sandia simulation algorithm is a statistical model designed for load calculation on blades. It provides a 2D temporal wind field modelising theoretical homogeneous and statistically steady atmosphere with spectra on which settings are, for basic model, mean wind speed, turbulence intensity and length scale of coherence. Spectra are layer dependent and can be chosen from Von Karman spectra to empirical model form measurement, its do not modelise horizontal shears for mean wind field or other coherent phenomenon (wake). Some improvement of the model permit to modelise few coherent phenomenon such as Low Level Jet or Kelvin-Helmholtz billow. Besides, it is possible to set different kind of mean wind profiles with this model ( Log, exp,...).
Considering that turbulence is frozen in the mean wind, we can set equivalence between spatial axis X and temporal axis t and therefore transform 2d temporal wind field in 3d temporal wind field. The software used to implement the model is TurbSim. A wind simulation with a coherent event (Low Level Jet) with a 10 m/s mean wind speed at 300 meters is represented (figure 3 shows slice of the wind field). Spectra model for this simulation is based on measurement (See [4])
For simple to moderate complex topography, those generic turbulence description give a good rate of representativeness error (as will be shown in validation model part). Wind phenomenon don’t change in a significant way for a moderate wind (from 4m/s to 16 m/s). Besides, several spectra models are available which describes a large bench of case from convective atmosphere (negative Richardson number) to really stable atmosphere. Limits can be founded in the fact that spectra model are stationary. Mann 's model which also can be used purposes a RDT (rapid distortion theory) which describes well the evolution of eddies in shear wind. We decided to used stationary model from Sandia method which describes well wind phenomenon for short period (up to 10 minutes). Real paramount of this method is its easiness and the fact that it allows the user to simulate many different cases quickly.


Lidar optical simulation :
Optical simulation come from Simulid (see [4]). All physical process of optical acquisition are modelised in Simulid and several parameter are accessible from Laser ones to detectors ones.Others inputs are radial velocity on beams from wind box and backscattering and extinction coefficient (called atmospheric parameters).

Geometric configuration :
In order to simulate a large scope of Lidar, simulator can handle any geometric configuration of Lidar. From profiler ones to scanning but also moving Lidar (buoy Lidar). Figure 5 shows profiler and scanning configuration in the wind field

Reconstruction algorithm and performance display :
The simulator has been made very adaptable therefor it is easy to try a new algorithm for reconstruction. Currently, two reconstruction algorithm are available, its are specific to profiler : the scalar averaging which reconstruct mean horizontal speed using the mean radial speed and the vector averaging which reconstruct mean horizontal speed using the instantaneous horizontal speed.
User can compare performance of the Lidar with different performance boards showing mean wind speed over the simulation duration and statistical properties (turbulence intensity,…). Figure 9 shows an example of performance board :



Conclusion

Wind simulator is efficient to provide basic study. For instance we tryed to figure out the impact of reconstruction algorithm on mean wind speed measurement. We simulated a wind field with turbulence spectra of Kaimal (defined in [6]) for 20% turbulence intensity with 10 m/s mean wind speed at 300 meters. Lidar setting were 5 beams with a 75 degrees elevation (Lidar described in figure 5), an acquisition frequency of 1hz and 25 gates were defined from 25 meters to 300 meters height. Simulation lasts 10 minutes and mean wind speed is calculated along this duration.

As expected, vector average gives good results with a correlation of 0,9915 and overestimation of less than 1%. Scalar average is less efficient with a 3% overestimation and a correlation of 0,9909. 3 % overestimation maybe too much and simulation with more point maybe need to be done in order to evaluate in a more accurate way the overestimation.

Way of validating the simulator is to compare real life measurement with simulator simulation which would close to the real wind field. Many Lidar campaign have been carried out on simple topography which are well represented by wind field simulation chosen.

In its current configuration, the simulator is efficient to provide basics studies with turbulent atmosphere especially for profiler Lidar. Several wind field are available and a large scope of parameters can be used for modelising atmosphere or for describe Lidar. The physical model from the wind field box is precise enough and gives reasonable error of representativeness. Simulator can be used efficiently on classic computer and gives good results.

Nevertheless, those model are efficient for small grid with moderate time and space step. LES simulation are more efficient for longer scale and to simulate long range Lidar. To simulate a whole scope of atmosphere phenomenon on a large grid, LES simulation is necessary.

For several topics of Lidar technologies, simulation appears to be paramount for development even if it might never replace campaign. For instance, in the case of buoy Lidar, for which, each campaign represent a important time and financial investment, a simulation could help to gain benefits.


Learning objectives
Attenders will learn about wind field simulation especially what is available as regards to CPU performance. Lidar measurement simulation with all physical process modelising will be described. Then several study case showing abality of simulation to answer question will be presented.


References
[1] Banakh, V.A.; Werner, C.; 2004; Computer simulation of coherent Doppler lidar measurement of wind velocity and retrieval of turbulence wind statistics; Optical Enginnering 44(7), 071205-1 071205-19 (2005)
[2] Frehlich, R.;1998; Simulation of coherent Doppler lidar performance for space based platforms; Journal of Applied Meteorology 39; 245-262 (1999)
[3] Matthew J. ChurchField; Sang Lee; Patrick J. Moriarty; Luis A. Martinez; Stefano Leonardi; Ganesh Vijayakumar and James G Brasseur; A LES Simulation of Wind Plant Aerodynamics. National Renewable Energy; 09/01/2012
[4] Kelley,N. ;Hand, M; Larwood, S; and McKenna, E. The NREL Large Scale Turbine Inflow and Response Experiment – Preliminary Results. NREL/CP-500-3097. Golden, CO : National Renewable Energy Laboratory,. January 2004.)
[5] SIMULID® developed in partnership between LEOSPHERE and l’ONERA. « SIMULID, Pulsed Focused Lidar simulation software”, Mehdi MACHTA et Jean-Pierre CARIOU (LEOSPHERE), Matthieu VALLA (ONERA), ILRC2012)
[6] IEC 61400-1 “Wind turbines-Part 1 : Design requirements.” 3rd edition. Geneva, Switzerland. International Electrical Commission, August 2005