Conference programme

Back to the programme printer.gif Print



Wednesday, 18 November 2015
11:30 - 13:00 Atmospheric flow over terrain
Resource assessment  
Onshore     


Room: Montparnasse

Complex terrain poses a challenge for onshore wind farms in terms of uncertainty on the energy production and difficult to predict turbulence, shear and gusts. To solve these challenges researcher employ numerical modeling, physical modeling and conduct field experiments.

Session scope:

onshore wind energy only

Learning objectives

The delegates with be able to explain:

  • The importance of field experiments for model validation
  • The strength and limitation of physical flow modeling
  • The possibilities and drawbacks of numerical modelling
  • The possibilities with a completely new kind of wind modeling facility
Lead Session Chair:
Jakob Mann, Professor, DTU Wind Energy / President, EAWE, Denmark
Øyvind Byrkjedal Kjeller Vindteknikk, Norway
Co-authors:
Øyvind Byrkjedal (1) F Emilie Iversen (1) Ove Undheim (1)
(1) Kjeller Vindteknikk, Kjeller, Norway

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

Byrkjedal has been working in Kjeller Vindteknikk for the past 8 years, and has currently the R&D manager of Kjeller Vindteknikk. Byrkjedal has a background as a meteorologist and holds a PhD in meteorology from the University of Bergen, Norway.

He has been working in the field of meteorological modeling during the past 8 years and has lead the development of the Norwegian wind- and icing atlases and has also developed wind and icing atlases for Sweden and Finland.

Abstract

High resolution meso-scale calculations at two moderately complex sites

Introduction

At wind power sites with moderate to high terrain complexity the industry standard micro-scale models are often not able predict the wind conditions to sufficiently low degree of uncertainty. When comparing to site measurements cross prediction errors are often large resulting in large uncertainties in the final annual energy production (AEP) estimates for the planned wind farm.

Approach

Two wind farm sites in Norway have been studied by the use of high resolution meso-scale simulations with the Weather Research and Forecast model (WRF). For both sites the model has been set up with a spatial resolution of 333 m x 333 m horizontally. The resolution was a compromise to have high enough resolution to represent the main features of the topography at the site and at the same time include the meso-scale and stability effects that the typical micro-scale models do not capture. The simulations were carried out for a period of one year which allows for sufficient statistics to compare the modeled data to the measurements carried out at the sites. The meso-scale simulations were compared to results from the micro-scale models WAsP and WindSim.

Main body of abstract

The first site is the Kjøllefjord Wind farm in the north of Norway. Data at this site were available from two masts which were located approximately 4 km apart from each other in addition to SCADA data from the 17 Siemens 2.3 MW turbines in the wind farm. The wind industry standard micro scale models have shown to be unable to simulate the observed differences in wind climate between the two masts resulting in a large deviation in AEP for the Kjøllefjord wind farm depending on which mast is used as basis for the energy calculation. The industry standard models resulted in cross prediction errors for the two mast locations in the range 10-18%.
The results from Kjøllefjord showed that the meso-scale model was able to accurately calculate the wind conditions at the two masts and we were able to realistically predict the wind conditions within the wind farm. The cross prediction error for the meso-scale model was 3 % at the mast locations. Calculations of the wind energy in the wind farm were greatly improved by using the meso-scale model instead of micro-scale models to estimate the wind conditions within the wind farm. The validation was carried out using concurrent data from the wind farm.
The second site was located in the south western part of Norway. Data from 5 tall meteorological masts in the area were available for validating the simulations. The masts were all located within 7 km from each other. The cross prediction errors between some of the masts were moderate to high using the industry standard micro-scale models with an average error 5.2 %. The cross prediction errors were significantly smaller using the high resolution meso-scale model resulting in an average error of 2.4 %.


Conclusion

The high resolution meso-scale model simulations were found to outperform the classical micro-scale models for the two sites with moderately complex terrain. Two effects have been identified as important in describing the observed differences between the mast locations in this study: 1. Large scale topographical features in the vicinity of the wind farm sites create meso scale influences that are not captured by the micro-scale models. 2. The meso scale model is able to capture differences that arise from flow over a hill under stable atmospheric conditions whereas the micro-scale models to not capture this effect.


Learning objectives
High resolution meso-scale models can reduce the uncertainty in AEP estimates for planned wind farms located in moderatly complex terrain.