Conference programme

Back to the programme printer.gif Print



Friday, 20 November 2015
12:00 - 13:30 Breakthrough session
Breakthrough sessions  
Onshore      Offshore    


Room: Belleville

A thorough review of abstracts cannot be done overnight – hence the main call for abstracts having closed in May. This is fine in most cases but, as delegates have told us, it is too early to be able to propose the very latest findings. Hence, we hold a call for ‘breakthrough abstracts’ presenting work that is genuinely ground-breaking and has in that form never been made public before. We introduced this at EWEA OFFSHORE 2015 and given the positive outcome, we decided to repeat this at EWEA 2015.

This call for ‘breakthrough abstracts’ took place from 1-14 September 2015. We would like to thank all the submitters forward to a highly interesting breakthrough session at EWEA 2015!

This breakthrough session covers topics from the resource to power performance, from turbines to wind farms and combines measurements and models. In a mixture of scientific and technical presentations attendees will learn about the experience and increasing acceptance of floating lidar technologies, covered by international IEA Wind experts. Turbulence remains of course a challenge for lidar measurements as well as for simplified but manageable models. A talk on turbulence intensities in large offshore wind farms will shine some light on the performance of models if compared with real measurements. The combination of turbulent fluctuations and large rotor blades leads to non-trivial blade deflections and modern control methods try to cope with this. However in order to be successful, the current deflection needs to be known and session attendees will see a new measurement concept that promises do this and which has just been tested on a large turbine this August. Finally combining this knowledge of turbulent flows with aero-elastic turbine models will be used to predict power performances in non-standard conditions.
 

Learning objectives

After attending this session, delegates will be able to:

  • estimate the current possibilities of using floating lidars in offshore wind farms.
  • quantify the performance of commonly used turbulence intensity models in real life.
  • explain how blade deflection of large blades in turbulent flows can be tackled.
  • evaluate the potential of using large wind farm data sets and modern computer power.
Lead Session Chair:
Stephan Barth, ForWind - Center for Wind Energy Research, Germany

Co-chair(s):
Tim Robinson, EWEA - The European Wind Energy Association asbl/vzw, Belgium
Taylor Geer DNV GL, United States
Co-authors:
Taylor Geer (1) F Carl Ostridge (1) Vineet Parkhe (1)
(1) DNV GL, Portland, United States

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

Taylor joined the wind industry in 2004 as an energy analyst for Garrad Hassan. Since then he has held numerous positions focused on assessing the energy production of renewable projects. Taylor helped lead the harmonization of the GL GH and DNV Kema Energy Assessment methodology. In his current role as Service Line Leader – Project Development, he ensures DNV GL can provide innovative solutions and opportunities for our customers to maximize their development activities. Taylor is a US delegate on the IEC 61400-15: Assessment of Wind Resource, Energy Yield, and Site Suitability committee.

Abstract

Understanding turbine performance with the dynamic trio: Super Measurements, Super Computers, and Super Models

Introduction

Understanding turbine performance in all conditions is a hot topic in the wind industry. A number of standard approaches to the challenge have been proposed in the industry, including “inner-outer” range calculations and performance matrices based on various parameters. In this presentation, DNV GL will present a suite of emerging approaches to better understand the issue and ensure we are on the right path to finding a solution, including: applying a machine learning algorithm to a large dataset of power performance test data and using an aero-elastic turbine model to predict performance in non-standard conditions.

Approach

DNV GL will apply a machine learning algorithm to over 75 power performance test data sets to identify the key variables impacting the variation in turbine performance. Additionally, DNV GL will use an aero-elastic turbine model (Bladed) to model the impact of complex flow conditions. The results of these two analyses will be used to develop a new approach to estimating turbine performance in preconstruction assessments.

Main body of abstract

The goal of this presentation will be to present a suite of emerging approaches to better understand the issue and ensure we are on the right path to finding a solution. DNV GL will anchor these emerging approaches with measured validations and a discussion around quantifying the uncertainty.
DNV GL has used a variety of approaches to better understand turbine performance including: comparing actual turbine performance to predicted performance in over 350 power performance tests, simulating power output with complex input conditions using an aero-elastic turbine model (Bladed), and applying machine learning algorithms to test data to derive realistic power curves for various inflow conditions. This presentation will build on previous work in a coordinated way to advance the industry’s understanding of what’s driving turbine production by:
• Identifying the key inflow conditions affecting turbine performance,
• identifying the extent to which performance can be predicted using aero-elastic turbine models,
• and identifying the conditions in which modeled and measured performance diverge
By applying a multivariable regression tree to a large database of power performance tests, the key drivers in turbine performance will be identified. Next, an aero-elastic turbine model will be used to perform a parameter sweep of these key drivers to identify the impact on turbine performance. The results of the parameter sweeps will be compared to measured turbine performance to identify any gaps in the modeled output.
The results of this exercise will be used to inform guidelines for pre-construction wind resource measurements and adjust energy assessments for site specific conditions.


Conclusion

Demonstrating that the energy output of a wind farm can be accurately modeled is essential for building and maintaining investor confidence in the wind industry. One of the biggest questions currently being asked is whether a turbine will perform as expected. DNV GL is in a unique position to answer that question through the combination of a large number of data sets and advanced computational power. This presentation will help to advance the industries knowledge on this topic and ensure that future work is focused on the right path


Learning objectives
1. Identify which inflow conditions primarily affect turbine performance
2. If aero-elastic models capture the effect of these conditions on turbine performance compared to measured performance
3. How to increase accuracy and transform the model into a super model.