Content uploaded by Simon Camal
Author content
All content in this area was uploaded by Simon Camal on May 20, 2020
Content may be subject to copyright.
This project has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No 864337
Towards the next generation forecasting tools of
renewable energy production
George Kariniotakis, Simon Camal | ARMINES, MINES ParisTech, Center PERSEE
in behalf of the Smart4RES team (see abstract for list of names).
EGU 2020, 4-8 May 2020, Online
Introduction
2
Context
▪Renewable Energy Sources (RES)
forecasting is a mature technology with
operational tools and services used by
different actors
▪However, there are gaps and bottlenecks in
the value/model chain…
▪Significant research worldwide
3
Number of publications with “solar” or “ photovoltaic” and “ forecast
” in Scopus Database
Year
Power, weather variables measurements, satellite
images, sky cameras, radars, lidars….
The RES forecasting model & value chain
4
Forecasts of RES production for the next minutes
up to the next days
A "mature technology", but...
5
Power output of a wind farm on a day
Forecast
Measurements
…Forecasting accuracy remains low
A "mature technology", but...
6
0
2000
4000
6000
8000
10000
12000
00:00
01:15
02:30
03:45
05:00
06:15
07:30
08:45
10:00
11:15
12:30
13:45
15:00
16:15
17:30
18:45
20:00
21:15
22:30
23:45
time
Power [MW]
measurement
prediction
Example of aggregated production in Germany:
Path of low-pressure system was different than predicted,
Max. error: 5500 MW…
…Forecasting accuracy remains low
A "mature technology", but...
7
•Imbalance costs for traders
•Increased need for costly remedies (reserves,
storage, demand response…)
•Limited capacity to support the grid with reliably
ancillary services (AS)
•Lower RES acceptability by operators
•RES curtailment
•Higher maintenance costs for RES plants
•…..
Impacts
…Forecasting accuracy remains low
A "mature technology", but...
8
… New forecasting needs emerge for applications like
provision of ancillary services by VPPs
If service can not be fulfilled,TSO may charge high penalties
9
Gaps and bottlenecks
The consortium
10
6 countries, 12 partners
End-users / Industry / Research / Universities / Meteorologists
▪Budget: 4 Mio€
▪Duration: 3.5 years, 11/2019 - 04/2023
11
The project
Our Vision: Achieve outstanding improvement in RES predictability through
a holistic approach, that covers the whole model and value chain related to RES
forecasting (from weather prediction up to end-use applications).
12
Objectives
1/ Requirements for forecasting technologies to enable near 100% RES penetration.
Overarching objective: Propose and test a number of forecasting and decision-aid tools to
cover gaps and/or prepare the energy system of 2030 and beyond
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
5/ New data-driven optimisation and
decision-aid tools for RES and storage.
4/ Streamline the process of getting optimal value through new forecasting products, datamarket
places, and novel business models.
6/ Validation of new models in living labs and assesment of forecasting values vs remedies.
Weather RES prediction Applications
13
Concept & methodology
Focus on NWP improvements
•Very/Ultra high resolution NWPs
•Regional solar irradiance nowcasting
14
Focus –NWP improvements
▪Refined cloud radiative impact for solar irradiance
forecast
▪Advanced tools to take advantage of ensemble
simulations for wind and solar forecasts
▪Seamless forecast from very short to medium ranges
15
Focus –NWP improvements
▪Very high resolution NWPs
16
Focus –NWP improvements
▪Advanced analysis of camera images for cloud characterization
▪Eye2Sky network of 35 sky cameras (Northern Germany)
18
Focus on NWP improvements
•Very high resolution NWPs
•Regional solar irradiance nowcasting
19
19
Focus on methods to improve RES
production forecasts
How can we improve accuracy without increasing complexity of the overall model chain?
Focus
20
▪Improve very short-term PV and wind forecasting (cameras, lidar.. / machine learning)
Very short-term RES forecasting
Focus
21
Weights on sources at runtime 12.00
Multi-source input PV forecasting
▪Towards models fed with multi-source data inputs
oIncrease quantity/quality of inputs
oMethodologies to tackle challenges from multiple
sources of input (dimensionality, heterogeneous
correlations between data sources and production…)
▪Towards better seamless forecasting
oLooking for generic models: assess validity for various
renewable energy sources, cover multiple time frames
oBest candidates to reproduce temporal correlations /
exploit temporal information? (e.g. recurrent models,
adversarial networks…)
Towards generic RES forecasting
▪The traditional approach to forecast for a single location:
▪Many works have shown the benefits from integrating
spatially distributed information (neighbor wind farms or PV
plants as sensors)
▪In Smart4RES we develop a framework for data sharing
preserving privacy and confidentiality constraints
▪And a data market concept to valorise data sharing
Focus
22
Collaborative RES forecasting
23
Focus on a selection of innovative
Use Cases
How to maximise the value of RES forecasts in Electrical Grids, Electricity Markets
and Storage Operation?
Markets
:
24
How to optimize the
operation and trading performance in electricity markets
of a
combined vRES and Storage system?
Focus
Challenge
Application: bidding strategy for RES + storage (1/2)
Today
, the use of storage systems only to compensate RES imbalances on the
energy market is not sufficient to pay back the investment in storage
.
•Energy (day ahead, intra-day)
•Ancillary services,
•Local flexibility...
Focus
25
Data
-driven bidding strategy for any vRES + storage on multiple electricity markets
Forecast RES production (DA + ID)
Forecast energy market
quantities (DA, ID, RT)
Forecast ancillary service market
quantities (DA, ID, RT)
Trading strategy including
storage constraints
Data on RES, storage, markets
Bids on energy + AS markets
Data-driven bidding tool
•Emulation of environment
•Neural nets learning value & policy
•Exploration / Exploitation
KPI: +20-25%
revenue
vs benchmarks
Approach
Application: bidding strategy RES + storage (2/2)
DA: Day ahead
ID: Intra-day
RT: Real time
AS: Ancillary Services.
Focus
▪
▪Extend the state-of-the-art approach for RES trading to include the human behaviour
of the traders.
26
“Automated
trading” Human
decision-maker
Market insights,
portfolio information,
etc.
Offer
Application: Include the human in trading
Focus
27
How can a DSO optimize the predictive management of local flexibilities
?
Today
, DSO do not fully consider uncertainties from Distributed Sources (DER)
o
peration and their local impact on booking and activation of flexibilities.
Challenge
DSO DER operator
Local Flexibility Market
2. Booking of
flexibilities
1.
Deterministic simulations
of constraints induced by
DER production/load
3.
Activation
of flexibilities
Substation
Influence
zone of DER
pdf
DER production/load
?
Application: Predictive management of grids (1/2)
Focus
28
Approach
Uncertainty
-aware booking of flexibilities + data-driven local activation controller
Evaluated on distribution grids in Europe
DSO DER operatorLocal Flexibility Market
2.Uncertainty-aware
booking of flexibilities
1.
Power flow
simulations of
constraints on ensembles
3. Data-
driven activation
of flexibilities
Optimized real-time
activation controller
Data from DER & grid
Ensemble of DER trajectories
KPI: +50% increased DER hosting capacity
pdf
V,P,Q
Application: Predictive management of grids (2/2)
THANK YOU !
29
This project has received funding from the European Union’s
Horizon 2020 research and innovation programme under grant
agreement No 864337
INCOMING ACTIVITIES
▪Webinar, June 5th 14:00 CEST, G.Kariniotakis and P.Pinson
(hosted by ISGAN)
oThe Role of RES forecasting in future power system
applications: innovative use cases (ARMINES)
oResearch for innovative weather and RES production
forecasting products (DTU)
▪Questionnaires to stakeholders on their experience and
expectations regarding RES forecasting and applications
www.smart4res.eu
Contact: info -at- smart4RES.eu