The Smart4RES project aims to enhance renewable energy source (RES) forecasting through innovative research and collaboration across six countries, involving twelve partners over a budget of €4 million. It addresses significant gaps in current forecasting technologies, focusing on weather prediction tailored for RES applications, improving accuracy, and integrating new forecasting tools. The project seeks to optimize the value of forecasting products and explore business models that incentivize data sharing while maintaining privacy and confidentiality.

1. Smart4RES - Data science
for renewable energy
prediction
Georges Kariniotakis, MINES ParisTech, ARMINES
Pierre Pinson, Technical University of Denmark (DTU)
05 June 2020
ISGAN Academy webinar #22
Recorded webinars available at: https://www.iea-isgan.org/our-work/annex-8/

2. ISGAN in a Nutshell
Created under the auspices of:
the Implementing
Agreement for a
Co-operative
Programme on Smart
Grids
5/6/2020 ISGAN IN A NUTSHELL 2
Strategic platform to support high-level government
knowledge transfer and action for the accelerated
development and deployment of smarter, cleaner
electricity grids around the world
International Smart Grid Action Network is
the only global government-to-
government forum on smart grids.
an initiative of the
Clean Energy
Ministerial (CEM)
Annexes
Annex 1
Global
Smart Grid
Inventory Annex 2
Smart Grid
Case
Studies
Annex 3
Benefit-
Cost
Analyses
and
Toolkits
Annex 4
Synthesis
of Insights
for
Decision
Makers
Annex 5
Smart Grid
Internation
al
Research
Facility
Network
Annex 6
Power
T&D
Systems
Annex 7
Smart Grids
Transitions
Annex 8:
ISGAN
Academy
on Smart
Grids

3. ISGAN’s worldwide presence
1/8/2018 ISGAN, PRESENTATION TITLE 3
5/6/2020 ISGAN IN A NUTSHELL

4. Value proposition
4
ISGAN
Conference
presentations
Policy
briefs Technology
briefs
Technical
papers
Discussion
papers
Webinars
Casebooks
Workshops
Broad international
expert network
Knowledge sharing,
technical assistance,
project coordination
Global, regional &
national policy support
Strategic partnerships
IEA, CEM, GSGF,
Mission Innovation, etc.
Visit our website:
www.iea-isgan.org
5/6/2020 ISGAN IN A NUTSHELL

5. Agenda
1. Introduction to the Smart4RES project Georges Kariniotakis (MINES ParisTech, ARMINES)
2. Challenges addressed Georges Kariniotakis (MINES ParisTech, ARMINES)
3. Research for innovative weather and RES
production forecasting products
Pierre Pinson (Technical University of Denmark, DTU)
05/06/2020 SMART4RES - DATA SCIENCE FOR RENEWABLE ENERGY PREDICTION 5

6. Agenda
1. Introduction to the Smart4RES project Georges Kariniotakis (MINES ParisTech, ARMINES)
2. Challenges addressed Georges Kariniotakis (MINES ParisTech, ARMINES)
3. Research for innovative weather and RES
production forecasting products
Pierre Pinson (DTU)
05/06/2020 SMART4RES - DATA SCIENCE FOR RENEWABLE ENERGY PREDICTION 6

7. Introduction
• Renewable Energy Sources (RES)
forecasting is a “mature” technology with
operational tools and services used by
different actors
• However, there are several gaps and
bottlenecks in the model & value chain
stimulating significant research worldwide
• Our objective today:
Present the challenges and innovative
research towards the next generation
tools for RES forecasting and related
decision making
SMART4RES - DATA SCIENCE FOR RENEWABLE ENERGY PREDICTION 705/06/2020

8. What is Smart4RES?
• 6 countries, 12 partners
• Budget: 4 M€
• Duration: 11/2019 - 04/2023
• End-users / Industry / Research /
Universities / Meteorologists
• TRLs: 1-5
SMART4RES - DATA SCIENCE FOR RENEWABLE ENERGY PREDICTION 805/06/2020
A new collaborative research project aiming to give a new boost to the
RES forecasting technology through some disruptive ideas.

9. The RES forecasting model & value
chain
SMART4RES - DATA SCIENCE FOR RENEWABLE ENERGY PREDICTION 905/06/2020
Power, weather variables measurements,
satellite images, sky cameras, radars, lidars….
Forecasts of RES production for the next
minutes up to the next days

10. The RES forecasting model & value
chain
…."mature technology", but forecasting
accuracy remains low
05/06/2020 SMART4RES - DATA SCIENCE FOR RENEWABLE ENERGY PREDICTION 10
• Financial losses in electricity markets
• Increased need for costly remedies (reserves, storage, demand
response…)
• Limited capacity of RES plants to deliver reliable ancillary services (AS)
• Lower RES acceptability by operators
• RES curtailment
• Higher maintenance costs for RES plants
• …
Impacts

11. The RES forecasting model & value
chain
…."mature technology", but forecasting
accuracy remains low and new needs are emerging
05/06/2020 SMART4RES - DATA SCIENCE FOR RENEWABLE ENERGY PREDICTION 11
• Forecasts for aggregated RES plants
• Forecasts for net load at different points of the grid
• Dedicated forecasts for ancillary service provision
• …
new
needs

12. The Smart4RES vision
05/06/2020 SMART4RES - DATA SCIENCE FOR RENEWABLE ENERGY PREDICTION 12
Achieve outstanding improvement in
RES predictability through a holistic
approach, that covers the whole model
and value chain related to RES
forecasting

13. The Smart4RES objectives
05/06/2020 SMART4RES - DATA SCIENCE FOR RENEWABLE ENERGY PREDICTION 13
2 RES-dedicated weather forecasting with 10-15% improvement using
various sources of data and very high resolution approaches.
3 New generation of RES production forecasting tools enabling 15%
improvement in performance.
4 Streamline the process of getting optimal value through new forecasting
products, data market places, and novel business models
5 New data-driven optimisation and decision aid tools for power system
management and market participation
6 Validation of new models in living labs and assessment of forecasting
value vs remedies.
1 Requirements for forecasting solutions to enable 100% RES penetration

14. Agenda
1. Introduction to the Smart4RES project Georges Kariniotakis (MINES ParisTech, ARMINES)
2. Challenges addressed Georges Kariniotakis (MINES ParisTech, ARMINES)
3. Research for innovative weather and RES
production forecasting products
Pierre Pinson (DTU)
05/06/2020 SMART4RES - DATA SCIENCE FOR RENEWABLE ENERGY PREDICTION 14

15. Gaps and bottlenecks
05/06/2020 SMART4RES - DATA SCIENCE FOR RENEWABLE ENERGY PREDICTION 15
Challenges in the models and the connections:
Need for Numerical Weather Prediction (NWP) products
adapted to RES use-cases.
4
5

16. Gaps and bottlenecks (NWPs)
05/06/2020 SMART4RES - DATA SCIENCE FOR RENEWABLE ENERGY PREDICTION 16
• Improved RES-oriented modelling of NWP variables. I.e.:
• Refined cloud radiative impact for solar irradiance forecast. Dynamic consideration of aerosols.

17. Gaps and bottlenecks (NWPs)
05/06/2020 SMART4RES - DATA SCIENCE FOR RENEWABLE ENERGY PREDICTION 17
• Improved RES-oriented modelling of NWP variables. I.e.:
• Refined cloud radiative impact for solar irradiance forecast. Dynamic consideration of aerosols.
AROME operational
domain WHIFFLE-LES application
• Need for seamless NWPs in applications
• Need for higher spatial/temporal resolution & updates frequency
AROME-EPS
NWP variable
(Ensembles)
HorizonsDay + 2 Day + 4
Association
ARPEGE-EPS IFS-EPS

18. Gaps and bottlenecks (NWPs)
05/06/2020 SMART4RES - DATA SCIENCE FOR RENEWABLE ENERGY PREDICTION 18
• Improved RES-oriented modelling of NWP variables. I.e.:
• Refined cloud radiative impact for solar irradiance forecast. Dynamic consideration of aerosols.
▪ Camera only x 13 (5 planned)
▪ Camera + meteo (RSI) x 7 (5 planned)
▪ Camera + ref. meteo x 2
▪ Camera in PV plant x 1 (Solarpark
Ammerland)
▪ Ceilometer x 6
• Better modelling of weather conditions through remote sensing (sky imaging)
• Need for seamless NWPs in applications.
• Need for higher spatial/temporal resolution & updates frequency

19. 5
Gaps and bottlenecks
05/06/2020 SMART4RES - DATA SCIENCE FOR RENEWABLE ENERGY PREDICTION 19
Need for NWP products adapted to RES use-cases.
Limitations of RES prediction models to exploit large amounts
of heterogenous data
4
Challenges in the models and the connections:

20. Gaps and bottlenecks (RES models)
05/06/2020 SMART4RES - DATA SCIENCE FOR RENEWABLE ENERGY PREDICTION 20
Data for horizon interval 1
(e.g. 5 min – 1 h)
Data for horizon interval 2
(e.g. 1 h – 3 h)
Data for horizon interval 3
(e.g. 3 h – 6 h)
Probabilistic forecasting tool @
horizon interval 1
Probabilistic forecasting tool @
horizon interval 2
Probabilistic forecasting tool @
horizon interval 3
Multi-time scale
Power Forecast
pdf pdf pdf
pdf pdf pdf pdf pdf
pdf
! !
• State of the art consists in separate models for different time frames (e.g. 5 min to 1 h, 1h to 6h, 6h to 48h
ahead…), each exploiting different data sources as input.
• Need for seamless and generic forecasting approaches, able to consider simultaneously heterogenous
data. => Possible a convergence of forecasting solutions?
Horizons

21. 5
Gaps and bottlenecks
05/06/2020 SMART4RES - DATA SCIENCE FOR RENEWABLE ENERGY PREDICTION 21
Need for NWP products adapted to RES use-cases.
Limitations of RES prediction models to exploit large amounts
of heterogenous data
Open loop between forecasts generation and their use in apps
4
4
Challenges in the models and the connections:

22. Gaps and bottlenecks (“open loop ”)
05/06/2020 SMART4RES - DATA SCIENCE FOR RENEWABLE ENERGY PREDICTION 22
• RES forecasting models today are centered on accuracy.
• An alternative could be to “tune” them considering, not only accuracy, but also the “value” they bring
when used in a specific application (i.e. €s in trading).
Forecasting
Trading
Forecasting
Trading

23. Gaps and bottlenecks (“open loop ”)
05/06/2020 SMART4RES - DATA SCIENCE FOR RENEWABLE ENERGY PREDICTION 23
• RES forecasting models are tuned today upon their accuracy.
• But why not use AI to simplify the whole model chain?
• An alternative could be to tune them considering, not only accuracy, but also the “value” the bring
when used in a specific application (i.e. revenue € in trading).
Forecasting
Trading
Forecasting + trading

24. 5
Gaps and bottlenecks
05/06/2020 SMART4RES - DATA SCIENCE FOR RENEWABLE ENERGY PREDICTION 25
Need for NWP products adapted to RES use-cases.
Limitations of RES prediction models to exploit large amounts
of heterogenous data
Open loop between forecasts generation and their use in apps.
4
4
Decaying commercial value of forecast products.
Challenges in the models and the connections:

25. 5
Gaps and bottlenecks
05/06/2020 SMART4RES - DATA SCIENCE FOR RENEWABLE ENERGY PREDICTION 26
4
Lack of price incentives to share data
Lack of meaningful open data (privacy issues)
Challenges in the models and the connections:

26. Gaps and bottlenecks (value from data)
05/06/2020 SMART4RES - DATA SCIENCE FOR RENEWABLE ENERGY PREDICTION 27
• In an era of energy digitalization, the focus is on more data.
But what real value can be extracted from data?
• Many works have shown the benefits of integrating spatially
distributed information (neighbor PV/wind farms as sensors).
• With Smart4RES we will exploit data science techniques, like
federated learning, to develop a framework for collaborative
forecasting through data sharing that respects privacy and
confidentiality constraints
• And a data market concept to foster data sharing

27. 5
Gaps and bottlenecks
05/06/2020 SMART4RES - DATA SCIENCE FOR RENEWABLE ENERGY PREDICTION 28
Lack of price incentives to share data
4
Lack of standardisation
Lack of meaningful open data (privacy issues)
Need for decision aid models adequate for high RES
integration scenarios
Need for business cases that demonstrate the value of
integrating uncertainty forecasts to industry
5
Challenges in the models and the connections:

28. 05/06/2020 SMART4RES - DATA SCIENCE FOR RENEWABLE ENERGY PREDICTION 29
Gaps and bottlenecks (the apps…)

29. Agenda
1. Introduction to the Smart4RES project Georges Kariniotakis (MINES ParisTech, ARMINES)
2. Challenges addressed Georges Kariniotakis (MINES ParisTech, ARMINES)
3. Research for innovative weather and RES
production forecasting products
Pierre Pinson (DTU)
05/06/2020 SMART4RES - DATA SCIENCE FOR RENEWABLE ENERGY PREDICTION 30

30. SMART4RES - DATA SCIENCE FOR RENEWABLE ENERGY PREDICTION 31
05/06/2020
What is a forecast product?
Data streams
Models
Forecasting approaches
Forecasting expertise
Analytics and visualization
Forecastproduct
( problem-oriented approach)

31. Motivations for new forecast products
05/06/2020 SMART4RES - DATA SCIENCE FOR RENEWABLE ENERGY PREDICTION 32
Regulatory and
market framework
• New markets e.g.
p2p or flexibility
markets
• New market
products e.g.
resolution,
ancillary services
• Requirements
from the
regulatory side
Technology and its
development
• Larger wind
turbines
• Hybrid power
plants
• Prosumer setups
• Storage
• New paradigms in
power system
operation
End-users
expressed needs
”Please tell us!”
A push from the
R&D side
• Forecasting R&D
is very active
• New concepts
have continuously
emerged
• Profit of latest
advances in
analytics and
decision-support

32. Motivations for new forecast products
05/06/2020 SMART4RES - DATA SCIENCE FOR RENEWABLE ENERGY PREDICTION 33
Regulatory and
market framework
• New markets e.g.
p2p or flexibility
markets
• New market
products e.g.
resolution,
ancillary services
• Requirements for
the regulatory
Technology and its
development
• Larger wind
turbines
• Hybrid power
plants
• Prosumer setups
• Storage
• New paradigms in
power system
operation
End-users
expressed needs
”Please tell us!”
A push from the
R&D side
• Forecasting R&D
is very active
• New concepts
have continuously
emerged
• Profit of latest
advances in
analytics and
decision-support
Exemple of high-resolution case
• Temporal resolution
ancillary services
• Prosumer markets
e.g. p2p
• Etc.

33. Motivations for new forecast products
05/06/2020 SMART4RES - DATA SCIENCE FOR RENEWABLE ENERGY PREDICTION 34
Regulatory and
market framework
• New markets e.g.
p2p or flexibility
markets
• New market
products e.g.
resolution,
ancillary services
• Requirements for
the regulatory
Technology and its
development
• Larger wind
turbines
• Hybrid power
plants
• Prosumer setups
• Storage
• New paradigms in
power system
operation
End-users
expressed needs
”Please tell us!”
A push from the
R&D side
• Forecasting R&D
is very active
• New concepts
have continuously
emerged
• Profit of latest
advances in
analytics and
decision-support
Exemple of high-resolution case
• Temporal resolution
ancillary services
• Prosumer markets
e.g. p2p
• Etc.
• Offshore wind farm
control
• Storage operation
• Etc.

34. Motivations for new forecast products
05/06/2020 SMART4RES - DATA SCIENCE FOR RENEWABLE ENERGY PREDICTION 35
Regulatory and
market framework
• New markets e.g.
p2p or flexibility
markets
• New market
products e.g.
resolution,
ancillary services
• Requirements for
the regulatory
Technology and its
development
• Larger wind
turbines
• Hybrid power
plants
• Prosumer setups
• Storage
• New paradigms in
power system
operation
End-users
expressed needs
”Please tell us!”
A push from the
R&D side
• Forecasting R&D
is very active
• New concepts
have continuously
emerged
• Profit of latest
advances in
analytics and
decision-support
• Many examples e.g.
Copenhagen airport
Exemple of high-resolution case
• Temporal resolution
ancillary services
• Prosumer markets
e.g. p2p
• Etc.
• Offshore wind farm
control
• Storage operation
• Etc.

35. Motivations for new forecast products
05/06/2020 SMART4RES - DATA SCIENCE FOR RENEWABLE ENERGY PREDICTION 36
Regulatory and
market framework
• New markets e.g.
p2p or flexibility
markets
• New market
products e.g.
resolution,
ancillary services
• Requirements for
the regulatory
Technology and its
development
• Larger wind
turbines
• Hybrid power
plants
• Prosumer setups
• Storage
• New paradigms in
power system
operation
End-users
expressed needs
”Please tell us!”
A push from the
R&D side
• Forecasting R&D
is very active
• New concepts
have continuously
emerged
• Profit of latest
advances in
analytics and
decision-support
• Many examples e.g.
Copenhagen airport
• Early push towards
high resolution in the
late 90s
• Today different
(very) high resolution
approaches
• Especially relevant
for offshore wind
Exemple of high-resolution case
• Temporal resolution
ancillary services
• Prosumer markets
e.g. p2p
• Etc.
• Offshore wind farm
control
• Storage operation
• Etc.

36. From high-resolution information and
data…
05/06/2020 SMART4RES - DATA SCIENCE FOR RENEWABLE ENERGY PREDICTION 37
• A company like Whiffle (among others) can provide very-high resolution weather
forecasts e.g. here for Horns Rev (spatial: 50-100m – temporal: app. 30s)
• How to use the produced information optimally?
C. Gilbert et al. (2020) Statistical post-processing of turbulence-resolving weather forecasts for offshore wind power forecasting. Wind Energy 23(4), pp. 884-897

37. … to meaningful forecast products
through post-processing
• Post-processing is usually considered for improving forecast quality (e.g.
model output statistics, machine learning. etc.)
• However, it can also be used to go from ”raw forecast data” to a tailored
forecast product
05/06/2020 SMART4RES - DATA SCIENCE FOR RENEWABLE ENERGY PREDICTION 38
Classical hourly single-valued forecasts
(or 5-15 mns)
Variability forecasts
Forecasts of potential minimum
and maximum production values
within periods

38. The probabilistic side
• Probabilistic forecasts are increasingly used in operations and electricity markets
05/06/2020 SMART4RES - DATA SCIENCE FOR RENEWABLE ENERGY PREDICTION 39
• Such concepts and
visualizations were
pushed forward in the
early 2000’s… now
getting quite common

39. New probabilistic forecasting products
• For many novel approaches to
decision-making under
uncertainty, those probabilistic
forecasts are not sufficient…
05/06/2020 SMART4RES - DATA SCIENCE FOR RENEWABLE ENERGY PREDICTION 40
• This is why scenarios and now
polyhedra and ellipsoids
(uncertainty regions) were
proposed

40. Data and forecasts are products
themselves!
• What if various agents and operators sell each other forecasts, data or
contribution to collaborative analytics?
05/06/2020 SMART4RES - DATA SCIENCE FOR RENEWABLE ENERGY PREDICTION 41
• Assume for instance a
central agent (a wind farm)
that wants to improve its
forecasts and contract
neighboring ones to help…
• How to design mechanisms for that purpose, while also respecting potential
privacy and business competition concerns?
Radar images
payment
payment
payment
Power data
Weather forecasts

41. New forecast products for grid
management
05/06/2020 SMART4RES - DATA SCIENCE FOR RENEWABLE ENERGY PREDICTION 42
• RES forecasts at plant and aggregated level +
Grid topology
• = New product: Grid-aware flexibility forecast
of distributed resources at a HV/MV interface
• How can data from distributed RES help
power system operators?
M. Kalantar-Neyestanaki, F. Sossan, M. Bozorg and R. Cherkaoui, "Characterizing the Reserve Provision Capability
Area of Active Distribution Networks: A Linear Robust Optimization Method," in IEEE Transactions on Smart Grid,
vol. 11, no. 3, pp. 2464-2475, May 2020
RES cluster 1
pdf
Aggregated RES
pdf
Aggregated Load
pdf
RES cluster 2
pdf
Load 1
pdfLoad 2
pdf
Coherent hierarchical
RES forecast
Active Power
Reactive
power
Forecast of flexibility
curve of the system

42. Take away messages
• RES-oriented research for improving weather forecasting
• Seamless approaches to permit convergence of the technology
• Data science approaches for alternative forecasting and decision-making
paradigms.
• Data sharing and data markets to extract the value out of data!
SMART4RES - DATA SCIENCE FOR RENEWABLE ENERGY PREDICTION 4305/06/2020

43. Be involved
• Want to take part of the Smart4RES
project?
Complete our questionnaire on your
use of forecasts and forecast-based
decision-aid tools
 Be involved in our Reference Group
SMART4RES - DATA SCIENCE FOR RENEWABLE ENERGY PREDICTION 4405/06/2020

44. Stay tuned
SMART4RES - DATA SCIENCE FOR RENEWABLE ENERGY PREDICTION 45
Smart4RES Webinar Series (under preparation):
Episode #2 - October 2020 RES forecasting and data marketplace: a collaborative framework
Episode #3 - December 2020 Optimising participation of RES generation in electricity markets: new
opportunities and the role of forecasting
 Newsletter: subscribe at www.smart4res.eu
Info @ smart4res.eu
05/06/2020

45. Georges Kariniotakis, MINES ParisTech
georges.kariniotakis@mines-paristech.fr
Pierre Pinson, Technical University of Denmark (DTU)
ppin@elektro.dtu.dk

46. This project has received funding from the European Union’s
Horizon 2020 research and innovation programme under grant
agreement No 864337
APPENDIX
COPYRIGHT (except initial slides on ISGAN presentation):
Smart4RES project: “Next Generation Modelling and Forecasting of Variable
Renewable Generation for Large-scale Integration in Energy Systems and
Markets”
DISCLAIMER
The European Commission or the European Innovation and Networks Executive
Agency (INEA) are not responsible for any use that may be made of the
information in this presentation.
PROJECT COORDINATOR & CONTACTS
Georges Kariniotakis, ARMINES/MINES ParisTech, Centre PERSEE,
Sophia-Antipolis, France.
info @ smart4res.eu
www.smart4res.eu