Conference programme

Back to the programme printer.gif Print



Wednesday, 18 November 2015
17:00 - 18:30 Annual energy production - let's get it right!
Resource assessment  
Onshore      Offshore    


Room: Montparnasse

To improve investor confidence in our energy yield assessments it’s vital that we can demonstrate improvements in the accuracy of our predictions. Only through doing this will be able to access cheaper sources of capital which will in turn reduce the cost of energy.

Learning objectives

  • Identification of the errors in the model chain.
  • How using multiple models can improve prediction accuracy.
  • How to use vertical extrapolation models in combination with long term correlation techniques.
  • Using the due diligence process to add project value.
Lead Session Chair:
Mike Anderson, RES Ltd.
Olivier Coupiac Maia Eolis, France
Co-authors:
Alban Mercier (1) F Olivier Coupiac (1) Jourdier Bénédicte (2) Girard Nicolas (1) Bossy Mireille (3) Rousseau Antoine (3) Kraria Sélim (3) Drobinsky Philippe (3) Omrani Hiba (3)
(1) Maia Eolis, LILLE, France (2) IPSL/LMD, PARIS, France (3) Inria, Paris, France (4) ADEME, Paris, France

Share this presentation on:

Printer friendly version: printer.gif Print

Presenter's biography

Biographies are supplied directly by presenters at EWEA 2015 and are published here unedited

Olivier Coupiac is currently a wind expert at Maia Eolis.
After a Ph-D in solid state physics, he dedicated himself to wind energy : first designing and building small wind turbines in Central America, then as a project manager in wind development in Germany. He is now in charge of site and energy yield assessment, measurement campaigns and research and developpement at Maia Eolis since 2012.

Abstract

Vertical extrapolation methods - a comparative study

Introduction

A permanent challenge for industrial wind assessment is to select the simplest and cheapest way of carrying out on site measurement, associated with the most reliable calculation method to estimate the long term wind specifications. Performing long measurement campaigns at great heights is expensive and can cause difficulties. This work shows the limits and the strengths of lidar technology, to improve knowledge about vertical extrapolation and long term correlation.

Approach

To simulate wind speed for a 80m hub height, this analysis focus on the use of a 40m meteo mast during 1 year, combined to a near Lidar during 3 months (in the same year).
Our previous work proves that long term correlations and vertical extrapolations are major sources of uncertainty. Therefore we tested various methods, first focalising on vertical extrapolation only, and then combining vertical and long term extrapolation.
Since our mast measurements are limited to 40m, we analysed various techniques to use 3 month Lidar data to estimate gradient in a relevant way.
Then we compared long term correlations between Lidar, meteo mast, and long term reference (more than 15 years).


Main body of abstract

To simulate one year wind at 80m high, six vertical extrapolation methods have been confronted, computing mast and Lidar wind speed gradient:
(1) alpha series: gradient is calculated each 10min between 10m and 40m (mast), then used to calculate 80m wind speed from each 40m value (mast).
(2) gradients are averaged for classes of wind direction and speed, between 10m and 40m (mast), then used to calculate 80m wind speed from each 40m value (mast).
(3) gradients are averaged for classes of wind direction, speed and thermal stability, between 10m and 40m (mast), then used to calculate 80m wind speed from each 40m value (mast).
(4) gradients are averaged for classes of wind direction and speed, between 40m and 80m (lidar), then used to calculate 80m wind speed from each 40m value (mast).
(5) gradients are averaged for classes of wind direction, speed and thermal stability, between 40m and 80m (lidar), then used to calculate 80m wind speed from each 40m value (mast).
(6) correlation of series of gradient between 10m and 40m (mast) and between 40m and 80m (lidar) during 3 months, to simulate a gradient 40-80 of 1 year.

We performed these calculations on 80m masts to check our results. Method (5) turned out to be the most reliable, considering power density error. Gradient between 10 and 40m should not be used to evaluate 80m wind speed.

Eventually, 4 techniques have been tested to estimate a long term wind speed at 80m high, using matricial EMD correlation model:
(1) correlation between 80m (lidar, 3 months) and long term reference.
(2) correlation between 80m (lidar, 3 months) and 40m (mast, 3 months), then correlated with long term reference.
(3) calculation of 80m wind speed from 40m using extrapolation method (5), then correlated with long term reference.
(4) correlation of 40m wind speed (mast, 1 year) and 80m with long term reference, then WAsP extrapolation.
For these calculations, 6 measurement campaigns of at least two years have been used, in order to check our results. Method (3) turned out to be the most reliable considering power density error, but method (4) is very close.


Conclusion

Measnet recommends to carry out measurements at least at ⅔ of hub height, to limit vertical extrapolation error. However, using a lidar during 3 months can significantly increase the reliability of these measures (on average 2% for power density). Some new work on thermal stability are now tested, and should improve the results furthermore. Indeed, more meteorological data are available each day, and can be relevant to estimate local gradient.


Learning objectives
- quantify uncertainty of each step of wind assessment calculation
- improve long term correlation and vertical extrapolation techniques.
- valorize 40m masts and Lidar to reduce campaign costs.