Extrapolation of Extreme Response for Wind Turbines based on Field Measurements Authors: Henrik Stensgaard Toft, Aalborg University, Denmark John Dalsgaard.

Slides:



Advertisements
Similar presentations
Prediction of design wind speeds
Advertisements

ALBERTA WIND POWER VARIABILITY STUDY Represented by Tommi Pensas.
1 John Dalsgaard Sørensen 1,2, Jens Nørkær Sørensen 2 & Jørgen Lemming 2 1) Aalborg University, Denmark 2) DTU Wind Energy, Denmark Introduction Risk assessment.
Challenge the future Delft University of Technology Blade Load Estimations by a Load Database for an Implementation in SCADA Systems Master Thesis.
Hydrologic Statistics
N.D.GagunashviliUniversity of Akureyri, Iceland Pearson´s χ 2 Test Modifications for Comparison of Unweighted and Weighted Histograms and Two Weighted.
Spring INTRODUCTION There exists a lot of methods used for identifying high risk locations or sites that experience more crashes than one would.
EPIDEMIOLOGY AND BIOSTATISTICS DEPT Esimating Population Value with Hypothesis Testing.
Lecture ERS 482/682 (Fall 2002) Flood (and drought) prediction ERS 482/682 Small Watershed Hydrology.
Quantitative Methods for Flood Risk Management P.H.A.J.M. van Gelder $ $ Faculty of Civil Engineering and Geosciences, Delft University of Technology THE.
Comparison of Eddy Covariance Results By Wendy Couch, Rob Aves and Larissa Reames.
CHAPTER 6 Statistical Analysis of Experimental Data
Distribution of Defects in Wind Turbine Blades & Reliability Assessment of Blades Containing Defects Authors: Henrik Stensgaard Toft, Aalborg University.
Hybrid System Performance Evaluation Henrik Bindner, Tom Cronin, Per Lundsager, Oliver Gehrke Risø National Laboratory, Roskilde, Denmark.
Flood Frequency Analysis
Aspects of Relevance in Offshore Wind Farm Reliability Assessment Nicola Barberis Negra 2 nd PhD Seminar on Wind Energy in Europe Risø.
The impacts of hourly variations of large scale wind power production in the Nordic countries on the system regulation needs Hannele Holttinen.
Wind loading and structural response Lecture 18 Dr. J.D. Holmes
Presentation of Wind Data  The wind energy that is available at a specific site is usually presented on an annual basis.  There are several methods by.
Detected Inhomogeneities In Wind Direction And Speed Data From Ireland Predrag Petrović Republic Hydrometeorological Service of Serbia Mary Curley Met.
Monday, 18 May 2015 Stefan Goossens
Smart Rotor Control of Wind Turbines Using Trailing Edge Flaps Matthew A. Lackner and Gijs van Kuik January 6, 2009 Technical University of Delft University.
Southern Taiwan University Department of Electrical engineering
February 3, 2010 Extreme offshore wave statistics in the North Sea.
Dong-Yun Kim, Chao Han, and Evan Brooks Virginia Tech Department of Statistics.
1 Saxony-Anhalt State Environmental Protection Agency Wolfgang Garche Bukarest EU-Twinning Project RO 2006 IB EN 09 Wolfgang Garche Saxony-Anhalt.
Chanyoung Park Raphael T. Haftka Paper Helicopter Project.
Wolf-Gerrit Früh Christina Skittides With support from SgurrEnergy Preliminary assessment of wind climate fluctuations and use of Dynamical Systems Theory.
Stats 95.
Going to Extremes: A parametric study on Peak-Over-Threshold and other methods Wiebke Langreder Jørgen Højstrup Suzlon Energy A/S.
Problems related to the use of the existing noise measurement standards when predicting noise from wind turbines and wind farms. Erik Sloth Vestas Niels.
Statistical Post-Processing of General Time Series Data - With Wind Turbine Applications LeRoy Fitzwater, Lance Manuel, Steven Winterstein.
ECE 7800: Renewable Energy Systems
EWEC2007 Milano, 8 May 2007 Extrapolation of extreme loads acc. to IEC Ed.3 in comparison with the physics of real turbine response Dirk Steudel,
A Process Control Screen for Multiple Stream Processes An Operator Friendly Approach Richard E. Clark Process & Product Analysis.
Chalmers University of Technology Department of Mathematical Sciences Göteborg, Sweden Effect of whipping on ship fatigue- Gaussian VS non-Gaussian modelling.
Uncertainty on Fatigue Damage Accumulation for Composite Materials Authors: Henrik Stensgaard Toft, Aalborg University, Denmark John Dalsgaard Sørensen,
Reliability Analysis of Wind Turbines
1 Statistical Analysis – Descriptive Statistics Dr. Jerrell T. Stracener, SAE Fellow Leadership in Engineering EMIS 7370/5370 STAT 5340 : PROBABILITY AND.
Frankfurt (Germany), 6-9 June 2011 M. Khederzadeh Power & Water University of Technology (PWUT) Tehran, IRAN. M.Khederzadeh – IRAN – RIF S3 – Paper 0066.
Far Shore Wind Climate Modelling Background Far shore wind conditions are expected to be favorable for wind energy power production due to increased mean.
Extreme value statistics Problems of extrapolating to values we have no data about Question: Question: Can this be done at all? unusually large or small.
1 National Wind Technology Center Wind Turbine Design According to IEC (Onshore) and -3 (Offshore) Standards Overview for NWTC November 8, 2005.
What can we learn about dynamic triggering in the the lab? Lockner and Beeler, 1999.
Machine Design Under Uncertainty. Outline Uncertainty in mechanical components Why consider uncertainty Basics of uncertainty Uncertainty analysis for.
Horizontal Axis Wind Turbine Systems: Optimization Using Genetic Algorithms J. Y. Grandidier, Valorem, 180 Rue du Marechal Leclerc, F B ´ Begles,
Extreme Value Prediction in Sloshing Response Analysis
Probability distributions
Sundermeyer MAR 550 Spring Laboratory in Oceanography: Data and Methods MAR550, Spring 2013 Miles A. Sundermeyer Computing Basic Statistics.
Effective static loading distributions Wind loading and structural response Lecture 13 Dr. J.D. Holmes.
Statistics.  Percentiles ◦ Divides a data set into 100 equal parts  A score of 1700 on the SAT puts students in the 72 nd Percentile. ◦ 72% score 1700.
Hydrological Forecasting. Introduction: How to use knowledge to predict from existing data, what will happen in future?. This is a fundamental problem.
Chapter 6: Descriptive Statistics. Learning Objectives Describe statistical measures used in descriptive statistics Compute measures of central tendency.
Tailoring the ESS Reliability and Availability needs to satisfy the users Enric Bargalló WAO October 27, 2014.
Probability plots.
EGS-AGU-EUG Joint Assembly Nice, France, 7th April 2003
Alternative Turbulence Correction Methods
Structural Reliability Aspects in Design of Wind Turbines
Enhancement of Wind Stress and Hurricane Waves Simulation
Power curve loss adjustments at AWS Truepower: a 2016 update
INVESTIGATION OF IDLING INSTABILITIES IN WIND TURBINE SIMULATIONS
Extreme Value Prediction in Sloshing Response Analysis
Flood Frequency Analysis
WindEurope Summit th September, Hamburg
Retuning the turbulent gust component in the COSMO model
Hydrologic Statistics
Summary descriptive statistics: means and standard deviations:
Chapter 1: Exploring Data
HYDROLOGY Lecture 12 Probability
Eric TROMEUR, Sophie PUYGRENIER, Stéphane SANQUER
Presentation transcript:

Extrapolation of Extreme Response for Wind Turbines based on Field Measurements Authors: Henrik Stensgaard Toft, Aalborg University, Denmark John Dalsgaard Sørensen, Aalborg University / Risø-DTU, Denmark

Contents Introduction Load Extrapolation based on Extreme Response in Storms Numerical Example Conclusion Future Work

Introduction The extreme load can occur in two situations: Standstill position – The wind turbine is parked and behave like a “normal” civil engineering structure (wind speed > 25m/s). Operational condition – The wind turbine is producing power and behave like a machine (wind speed 3-25m/s).

Introduction The extreme load during operation is dependent on: The mean wind speed. The turbulence intensity. The type and settings of the control system. Flap Bending Moment Pitch controlled wind turbine Tower Mudline Moment Pitch controlled wind turbine

Introduction IEC : Extreme load determined from 10min. simulations of the response during operation over the range of significant wind speeds. The method is based on the following assumptions: Local extremes in the response are independent. Individual 10min. time series are independent. In the present paper are these assumptions investigated.

Introduction In the present paper is a new method for estimation of the extreme load presented. By comparing the extreme load determine by this method and the method in IEC can the assumptions about independence be studied. The new method uses a long measured time series which rarely is available – Therefore is the new method only used for validation of the existing method.

Load Extrapolation based on Extreme Response in Storms The new method for estimating the extreme load is based on: Existing methods for estimating e.g. the extreme wave height. Assumption about independent storms.

Load Extrapolation based on Extreme Response in Storms A storm wind speed U storm should be defined so extreme loads occur for mean wind speeds above U storm.

Load Extrapolation based on Extreme Response in Storms U storm is defined from the nominal wind speed U nom (typically m/s) and the standard deviation for the turbulence  1. Independence of storms: Storms are combined if the mean wind speed between them not are below a percentile of U storm. Storms are separated a minimum number of hours.

Load Extrapolation based on Extreme Response in Storms 1.The measured time series is divided into independent storms. 2.For each storm is the extreme response extracted. 3.To the largest extreme responses is a distribution function fitted. 4.The characteristic load is calculated for the probability:

Numerical Example Stall controlled onshore wind turbine. Dataset for response measured over 62 days in winter and spring. (52 days of complete measurements) Dataset for mean wind speed measured over 4 years. Site: Less severe than class III in IEC Turbulence intensity I ref = 0.12.

Numerical Example Nominal wind speed U nom Storm wind speed U storm Storms per year  Length of storm T s 14m/s9.5m/s2028.0h 15m/s10.2m/s1657.4h 16m/s11.0m/s1356.7h Investigation of storm definition. Storms are combined if the mean wind speed between them not are below 80% of U storm. Storms are separated at least 2 hours. Higher nominal wind speed leads to less and shorter storms.

Numerical Example Load extrapolation according to IEC Extremes extracted by the Peak Over Threshold method. Characteristic loads calculated with/without statistical uncertainty (Hessian matrix). Normalized characteristic loads. ThresholdCharacteristic load without stat. unc.with stat. unc

Numerical Example DistributionCharacteristic load without stat. unc.with stat. unc. Weibull Gumbel Load extrapolation based on extreme response in storms. 25 largest extremes from independent storms are used. Measured time series are short – 52 days of complete measurements Choice of distribution function has a significant influence on the results.

Conclusion New method for calculation of the extreme response is proposed. Definition of a storm is proposed. Higher characteristic load using the new method which could indicate that the 10min. time series or local extremes are dependent. Based on a stall controlled wind turbine and a short measured time series.

Future Work Refinement of the storm definition based on: Longer measured time series. More severe sites. Estimation of the “correct” long-term distribution for the extreme response: Weibull Gumbel Etc. Comparison of the new method and IEC for a long measured time series.

Extrapolation of Extreme Response for Wind Turbines based on Field Measurements Authors: Henrik Stensgaard Toft, Aalborg University, Denmark John Dalsgaard Sørensen, Aalborg University / Risø-DTU, Denmark