Wednesday, 28 September 2016
09:00 - 10:30 Cold climate issues in resource assessment
Resource assessment  
          

Sites subject to cold climate not only have temperatures outside the normal limits of standard wind turbines but also atmospheric icing conditions which are frequent and may account for a significant loss in annual production. This session addresses the most recent advancements in the field of atmospheric icing effects on wind resource yield assessment. Assessing, measuring and estimating icing losses in the resource assessment phase of a project is of crucial importance for the successful business case of a wind farm in cold climates. Icing of the rotor blades can significantly reduce the energy yield of a wind farm up to 10% or more of the annual production and it also influences wind measurements by reducing availability.

You attended this session?

Please give us feedback

 

Learning objectives

• Learn state-of-the-art measurement techniques and data analysis approaches for cold climate sites and therefore decrease the uncertainty in yield assessments;
• Discover the latest findings from over 20 sites and 100 met mast years of data in cold climates from Scandinavia and Germany;
• Learn to execute more reliable pre-construction energy yield assessments in cold climate sites.

This session will be chaired by:

Martin Strack Deutsche WindGuard Consulting GmbH, Germany
Co-authors:
Martin Strack (1) ^F Dr. Rene Pforte (2)
(1) Deutsche WindGuard Consulting GmbH, Varel, Germany (2) Svevind AB, Umea, Sweden

Big data approach of wind resource and operational data analysis in cold climate

Introduction

Within this presentation it is shown how big data analysis approaches facilitated the thorough analysis of wind measurement and wind turbine operational data for a large wind farm project in cold climate.

Approach

The enormous amount of data from a comprehensive wind measurement campaign (>100 sensors) and turbine operational data in cold climate was analysed by a new integral data evaluation concept, which bases on time- and wind direction dependent scatter- and correlation-matrices calculated between all related wind data. Based on this, a procedure for a seasonally dependent data gap filling and data assessment scheme was developed.

Main body of abstract

The “Markbygden 1101” wind farm, located in northern Sweden, is the largest onshore wind farm yet planned in Europe, including the erection of 1101 wind turbines (up to 4 GW) in several phases, from which the pilot phases (about 50 turbines) have already been installed.

For assessment of the wind resource at the intended hub height of 130-140 m, a comprehensive wind measurement campaign was started in 2003, extended over the time and is still running. Altogether 13 well equipped masts have been operated, several between 120 and 152 m high, with about 80 wind speed, 40 wind direction and numerous further atmospheric measurements, accompanied by mesoscale wind data from different grid points as well as operational turbine data. Due to the enormous data amount and the challenging climatic conditions, the data quality control and evaluation exceeded the feasible limits of conventional data evaluation procedures. This concerns the detection of icing events in measurement and operational data, data filtering, correlation and filling process, but also the alignment of measurement data to the same period, in order to allow accurate evaluations and analyses of atmospheric conditions.

A new approach has been developed to handle this with “big data” analysis methods. The evaluation of scatter- and correlation-matrices between all involved wind data series, including an analysis of seasonal dependency of these, provided the basis for developing a systematic quality assessment and data filtering process. Likewise, these analyses gave impressing insights of atmospheric conditions and phenomena relevant at the site. It became clear, that conventional procedures for MCP extrapolations and data gap filling will introduce a bias. A strategy was developed to tackle these tasks with minimized uncertainty and avoiding such bias. This procedure utilized an integral multi-correlation approach based on all available data, ranked sector-wise by correlation coefficient and goodness of self-consistency tests, and applied on basis of a moving-window analysis to acknowledge seasonal variability.

The results of this process were analysed and compared to conventional approaches, in order to assess the effectiveness of the computationally demanding procedure. As the data were further evaluated not only for energy yield analysis for upcoming project phases, but also for operational turbine data analysis, the uncertainty assessment was handled on time series basis, in order to allow a reasonable consideration of error propagation and better assessment of significance of observed deviations and incidents.


Conclusion

When evaluating wind measurement or operational turbine data, a thorough quality control, and possibly filtering and filling, is indispensable for accurate analysis. Demanding conditions like icing impacts impose special requirements for these tasks. Based on the example of a large project in cold climate conditions with an outstanding measurement campaign, the challenges related to these tasks have been elaborated, and a procedure based on big data analysis approaches has been developed, improving the state of the art.


Learning objectives
The tasks of quality assessment, filtering and filling of measurement and operational data can be effectively improved by an extensive, integral "big" data evaluation. The presented ideas may be used by delegates as starting point for improvement of such procedures in similar conditions, like cold climate, but also for other purposes, where wind data from large height range is relevant, like evaluation of remote sensing data. Last, but not least, is it exciting to see real meteorological effects visualized by correlation matrices from many wind measurements and their seasonal variation.