Day one, 8 September

Opening speeches

SPEAKERS

SESSION DESCRIPTION

The electricity system will double in size compared to today and wind will be half of that by 2050. So resource and operational assessment are more important than ever. Where can we still improve operational and resource assessment to ensure the cost-effective management of a climate neutral energy system? Digitalisation is a big driver to improve the performance of wind assets and optimise system integration. How do artificial intelligence and machine learning contribute to improving resource and operational assessment? All our models are historical. But Climate Change is testing our modelling. While we help address Climate Change we also need to be mindful of its impacts on our resources and operations. How can we deal with this? Join us for a conversation with Salvatore Bernabei, CEO of ENEL green power to explore these questions and get ready for 3 days of in-depth presentations and discussions.

Giles Dickson

CEO, WindEurope

Welcoming address

Salvatore Bernabei

CEO, Enel Green Power

Opening remarks

Artificial Intelligence: from design to operation

SESSION DESCRIPTION

Artificial Intelligence (AI) provides powerful solutions that support all steps of the value chain of wind energy. In this session participants will learn about contributions of machine learning in the ressource assessment phase for obtaining more accurate annual energy production estimates, but also in the operational phase, for detecting underperformance of wind turbines.

The session will address the scalability of AI tools for predictive maintenance of wind turbines, where massive data are considered. Finally, the participants will hear about AI-based state-of-the-art forecasting methods for wind generation and extreme wind situations and also for optimising the use of forecasts in applications like trading to electricity markets.

SESSION CHAIR

Georges Kariniotakis

Professor, Head of Renewables & SmartGrids Group, MINES ParisTech

SPEAKERS

Jordan Perr-Sauer

Staff Researcher – Data Science, National Renewable Energy Laboratory (NREL)

Lowering long-term energy estimate uncertainty through better wind plant power curves

Henrique Diogenes

Performance & Reliability Coordinator, Casa dos Ventos

A machine learning approach for underperformance detection in wind turbines

Gianmarco Pizza

CEO, Nispera

A scalable solution for predictive maintenance of operating wind farms based on deep learning

Miguel Ángel Prosper Fernández

Meteorologist, Siemens Gamesa Renewable Energy

Detection system for extreme wind events based on WRF high-resolution simulations and Deep Learning image recognition

Simon Camal

Senior Researcher, Project Manager, MINES ParisTech – Centre PERSEE

A holistic approach to improve the model & value chain of renewable energy forecasting – the smart4RES project

Improving simulations to reduce uncertainty

SESSION DESCRIPTION

The maturity of the wind industry has led it to become a much more competitive market than in the past, requiring greater precision in studies by reducing its uncertainties. This session will show innovative techniques aimed at reducing uncertainty in different aspects relevant to wind farms, ranging from the optimal use of measurement systems, to the development and calibration of new models of wakes, plant losses or the performance of wind turbines.

SESSION CHAIR

Jose Vidal

Technical Director, Energy Advisory Services, UL

SPEAKERS

Lee Cameron

Data Science Manager, RES

Designing scanning LiDAR campaigns to Reduce Yield Assessment Uncertainty

Morten Lybech Thøgersen

Senior Wind Energy Specialist, EMD International A/S

Shorter measuring campaigns using geospatial predictors and machine learning

Neil Atkinson

Principal Specialist, K2 Management

A multiple wind farm validation of turbine performance predictions from a 3D turbine performance matrix

Camille Dubreuil-Boisclair

Principal Researcher, Equinor ASA

Ensemble methods for wake parameter calibration

Marion Le Doeuff

Analytics Manager, Natural Power

Method to improve accuracy of production modelling and loss interaction in the time domain