The International Conference on Energy Meteorology 2013

 
ICEM_header

Bonjour!

I have the privilege of writing my most recent blog post from Toulouse, France at the International Conference Energy Meteorology – ICEM 2013!  And boy has it been an exciting and interesting event!

You would be amazed what is going on in the Energy Meteorology field these days – the science is truly impressive.  With industry and university level delegates from around the world, it is providing me with an excellent overview of the state of the art in solar and wind forecasting and resource assessments.

In particular, the solar forecasting field is fast moving.  Here is a snapshot!

In Europe, it is already in full-swing. 

In France, EDF (Electricité De France) and Meteo-France are already producing operational solar energy forecasts, using numerical weather prediction (NWP), statistical/machine-learning forecasts and cloud imaging.   Research presentations in this field included:

  • Laurent Dubus (EDF) – “Requirements of the energy sector on meteorological information”.  Take away from this talk:
    •  a 1°C changes total demand by 2.3GW!
    • There are 250k solar energy installations in France
    • You can see their generation and load in real-time online
  • Michaël Zamo (Meteo-France) - A Benchmark of Statistical Regression Methods for Short-Term Forecasting of Photovoltaic Electricity Production

    •  A great benchmarking exercise looking at daily solar energy generation on a day ahead forecast horizon
    • Random Forests did the best, here’s a graphic:
meteofrance_RMSE
  • Bruno Charbonnier  (EDF) – “A Cloud Cover Estimation System and It’s Comparison to the One Provided by Total Sky Imager”
    • They built their own sky-imager for $2k that does better detecting clouds than the $15k Yankee TSI 440



In Germany, Jan Kuehnert of University of Oldenberg gave a talk about cloud motion vectors, and satellite based forecasting for the massive number of distributed PV sites in Germany.  Turns out, they have some of the same challenges we do in Australia – knowing on the installed capacity of PV systems and the approximate location.  We are looking forward to collaborating on solving this issue.

Jan’s talk – “Satellite Based Short Term Prediction of Photovoltaic Power for the Application at the Energy Market”

Solar Energy Research Institute of Singapore

Other notable talks in the field were those from Andre Nobre and Wilfred Walsh of the Solar Energy Research Institute of Singapore.  It’s always fun to hear their talks – seeing what a progressive, well-funded research organization can do with solar forecasting!  They’re ramping up (oh yeah, that’s a bad pun) to create a full-on solar energy forecasting system for Singapore.  Their talks:

  • Andre Nobre – “Spatially-Resolved Real-Time Irradiance Measurements for Advanced Solar Resource Forecasting for Photovoltaic Applications”

 

  • Wilfred Walsh – “A Solar Forecasting System for Singapore”

Follow their work at their webpage and while you’re at it, their YouTube video.

NCAR's New Solar Forecasting Project

But the show was really stolen by the new effort underway at NCAR, as presented by Sue Ellen Haupt.  Suffice it to say, if they succeed in all their goals, they’ll put the rest of us out of a job!  Entitled:  "A Public-Private-Academic Partnership to Advance Solar Forecasting"

Take aways:     

  •  WRF-Solar is happening
  • A full spectrum of forecasts from short (minutes) through to days ahead will be created
  • Forecasts will be focused on large solar plants around the U.S. (e.g. De Soto (50MW), SMUD (150MW), Xcel’s new one (90MW)
  • It’s all going to be Open-source!

In Australia

Down Under, we heard from:

  • Alberto Troccoli (CSIRO) - Predicting a Daily Variability Index of Solar Radiation Using CCAM

 

  • Peter Coppin (CSIRO) – The Australian Solar Energy Forecasting System (Phase 1)

 

  • Armin Dehghan (UNSW) - Solar Radiation Forecasting in Eastern Australia
    • Applying the TAPM air pollution model to solar forecasting

 

  • John Boland (University of South Australia) – Statistical Analysis in Energy Meteorology
    • Presented hourly forecasts via his new CARDs model (get the paper)

 

And of course, myself!  But I’m going to save that for a blog post of its own, which I hope to write tomorrow after my talk:

Nicholas Engerer - "Short-Term Machine Learning Based Power Output Forecasts for Collectives of Rooftop Photovoltaics"

 

Until then, Au Revoir!

 

/*