Elia Group SA/NV · EURONEXT:ELI
Elia Group SA/NV — Technical Presentation (July 2019)
53 slides from Elia Group SA/NV's technical presentation, July 2019
Slide 1: Market Response 2019
Market Response 2019 Task Force ISR presentation Brussels, July 2019 1
Slide 2: Table of content
Table of content 1 Goal, scope and planning of the MR update – Reminder 2 Results of the aggregated curves analysis – 2019 update 3 Conclusion 2
Slide 3: Market Response volume determination is essential to size the volumes of Strategic Reserves
REMINDER Market Response volume determination is essential to size the volumes of Strategic Reserves Market Response corresponds to the response of electricity consumers in periods of tension and high prices in the electricity grid Available resources Needs Variables Demand Reserves • Generation • H
Slide 4: 2017 project scope: Market Response encompasses all DR/DSR not subject to a contract with the TSO
REMINDER 2017 project scope: Market Response encompasses all DR/DSR not subject to a contract with the TSO DR type Description Associated product This type of DSR is directly activated by the Elia products With TSO TSO in periods of scarcity Perimeter of the Market Response Contract based This categ
Slide 5: 2017 project scope: under a certain threshold, market response is already taken into account
REMINDER 2017 project scope: under a certain threshold, market response is already taken into account in the load forecast by Elia – the methodology aims at assessing Market Response above 150€/MWh, above normal conditions 1. Market response under 150€/MWh 2. Ancillary services volumes Price (€/MWh)
Slide 6: A list of 8 possible methodologies was established
REMINDER After a benchmark and a group reflection during the 4 workshops with stakeholders, a list of 8 possible methodologies was established Case by case analysis of the volume variations of the total load of Elia. Market A Load curve of Elia response would be seen as a load reduction in periods o
Slide 7: All methodologies were confronted to criteria established with the stakeholders
REMINDER In 2017, all methodologies were then confronted to criteria established with the stakeholders, and key conclusions were drawn for each methodology The methodologies were confronted to criteria established during a brainstorm session during WS2: 1. Perimeter of the study : Each methodology w
Slide 8: A first proposal was discussed with the stakeholders to come to a final methodology
After the criteria confrontation, a first proposal was discussed with the stakeholders to come to a final methodology, validated by the stakeholders during the fourth workshop Methodology proposal Final methodology Methodology selection and discussion proposal 1. Perimeter of the study Each methodol
Slide 9: A robust methodology was established based on the aggregated curves
REMINDER In 2017, a robust methodology was established based on the aggregated curves, and complemented with a qualitative Q&A to define the details of the activation Residential Tertiary Industrial A Aggregated curves analysis: quantitative approach To provide a robust estimation for the future yea
Slide 10: In the aggregated curves of EPEX DAM Belgium, Market Response volumes appear as demand decrease or offer increase
REMINDER In the aggregated curves of EPEX DAM Belgium, Market Response volumes appear as a demand decrease or as an offer increase Market Response volumes valued in the DA market Demand decrease Offer increase • This part can be analyzed directly in the aggregated • Instead of a demand decrease, sup
Slide 11: Table of contents
Table of contents 1 Goal, scope and planning of the MR update – Reminder 2 Results of the aggregated curves analysis – 2019 update 3 Conclusion 11
Slide 12: The update is based on the exact same methodology as in 2017 and 2018
The update of the Market Response Study is based on the exact same methodology as the one performed in 2017 and 2018 The process followed four key steps to come to a pertinent volume of Market Response: The Market Response volumes were first extracted from the aggregated curves of EPEX DAM Extractio
Slide 13: Agenda
Agenda 2 Results of the aggregated curves analysis – 2019 update A Extraction of the Market Response volumes B Refinement of the dataset C Statistical analysis D Implementation 13
Slide 14: EPEX DAM Belgium provides hourly aggregated curves of purchase and sale orders
A VOLUME EXTRACTION – REMINDER EPEX DAM Belgium provides hourly aggregated curves of the purchase and sale orders EPEX DAM Belgium aggregated curve Price Illustration (€/MWh) • The curves determine the clearing price: at 3000 the intersection of the demand and supply curve. 2500 • From the curves, w
Slide 15: Market Response volumes were extracted from the aggregated curves
A VOLUME EXTRACTION The Market Response volumes were extracted from the aggregated curves. The 2019 update added 8472 hours to the dataset Market Response study 2017 Update 2018 Update 2019 2014 2015 2016 2017 2018 2019 Timeframe of 01/01/2014 01/05/2017 17/04/2018 05/04/2019 the dataset 00:00 23:00
Slide 16: Agenda
Agenda 2 Results of the aggregated curves analysis – 2019 update A Extraction of the Market Response volumes B Refinement of the dataset C Statistical analysis D Implementation 16
Slide 17: The aggregated dataset refinement follows the same approach as in previous years
B DATA REFINEMENT The aggregated dataset refinement follows the same approach as in the previous years Market Response study 2017 Update 2018 Update 2019 Raw dataset Refined dataset Refined dataset Refined dataset Update 2018 Update 2019 General strike days Considered as Considered as Considered as
Slide 18: Agenda
Agenda 2 Results of the aggregated curves analysis – 2019 update A Extraction of the Market Response volumes B Refinement of the dataset C Statistical analysis D Implementation 18
Slide 19: Numerous analyses were conducted without any strong correlations
C STATISTICAL ANALYSIS Numerous analyses were conducted to explain the volume patterns, yet without any strong correlations Various analyses were conducted : § Simple correlations and multivariate regressions: – Day-ahead prices – Temperatures – Normal temperatures – Daily maximum price – Load – Gas
Slide 20: Various correlations were computed without satisfying results (1/2)
STATISTICAL ANALYSIS – Example Various correlations were computed (temperature, price, normal temperature) without any satisfying results: R2 remains very low (1/2) Principle Results – 2017 study Results – 2018 study ² Results – 2019 study A regression was • 150€/MWh: R2 = 0.03 • 150€/MWh: R2 = 0.06
Slide 21: Various correlations were computed without satisfying results (2/2)
STATISTICAL ANALYSIS – Example Various correlations were computed (temperature, price, normal temperature) without any satisfying results: R2 remains very low (2/2) Principle Results – 2017 study Results – 2018 study Results – 2019 study The regression was • 150€/MWh : R2 = 0.09 • 150€/MWh : R2 = 0.
Slide 22: The impact of various parameters was assessed on the new dataset
C STATISTICAL ANALYSIS The impact of various parameters was assessed on the new dataset to verify the coherence with the analyses conducted in the last years Analyses conducted: Restriction of the Focus on the load impact of parameters dataset • 3 analyses were conducted to Among the previous parame
Slide 23: Only restrictions of the additional 2019 dataset to the winter period decrease volatility
C STATISTICAL ANALYSIS – Restriction to pertinent periods Contrary to the 2017 study, only the restrictions of the additional 2019 dataset to the winter period decrease the volatility, while further restrictions to weekdays and peak hours increase it Restriction 1: Restriction 2: Restriction 3: Glob
Slide 24: Only restrictions of the complete dataset up to 2019 to the winter period decrease volatility
C STATISTICAL ANALYSIS – Restriction to pertinent periods Contrary to the 2017 study, only the restrictions of the complete dataset up to 2019 to the winter period decrease the volatility, while further restrictions to weekdays and peak hours increase it Restriction 1: Restriction 2: Restriction 3:
Slide 25: The restrictions also decrease the average MR values
C STATISTICAL ANALYSIS – Study of some specific days of additional data In addition to amplifying the volatility of the dataset, the restrictions also decrease the average MR values EVOLUTION OF THE DISTRIBUTION WITH THE REFINEMENT Refined dataset distribution Restriction 1 : season N° of hours N° o
Slide 26: Categorization of the additional data is less coherent
C STATISTICAL ANALYSIS – Restriction to pertinent periods Contrary to the previous years, the categorization of the additional data is less coherent with the 2017 dataset: it leads to a standard deviation increase, and to a slightly lower market response volume Summary of the two categories (refined
Slide 27: Agenda
Agenda 2 Results of the aggregated curves analysis – 2019 update A Extraction of the Market Response volumes B Refinement of the dataset C Statistical analysis D Implementation 27
Slide 28: Total Market Response has grown +1% on average p.a.
D Extrapolation of the results – Historical volumes evolution Over the last winters, total Market Response has grown with +1% on average p.a. on the last 5 years Evolution of the volumes of Market Response - Winter Months1) MW +1%/y.2) 1,185 1,075 1,049 995 1,026 4474) 342 414 504 399 738 653 612 65
Slide 29: Several extrapolations can be defended
D Extrapolation of the results Several extrapolations can be defended 1% global volume growth 3% global volume growth 5% global volume growth § The +1% global market § The +3% global market § The +5% extrapolation response volume is a more response volume corresponds to the cautious approach based o
Slide 30: Extrapolation growth varies from +1%/y. to +5%/y.
D Extrapolation of the results Based on the historical growth trends, the total market growth extrapolation growth varies from +1%/y. to +5%/y. A historical extrapolation factor can be suggested : +1% based on 4 years (14/15 to 18/19). Two additional scenarios taking +3%, and +5% total growth per ye
Slide 31: Table of contents
Table of contents 1 Goal, scope and planning of the MR update – Reminder 2 Results of the aggregated curves analysis – 2019 update 3 Conclusion 31
Slide 32: Conclusion: quantitative part updated with recent data
Conclusion According to the methodology designed in 2017, the quantitative part (aggregated curve analysis) was updated with recent data 2017 Market response study 2019 Study update A Aggregated curves analysis Raw dataset Refinement: The year 2014 and national strikes days were excluded Verificatio
Slide 33: The update leads to a 699 MW Market Response volume and 3 extrapolation scenarios
Conclusion The update of the study leads to a 699 MW Market Response volume and 3 extrapolation scenarios ranging from 1% to 5% total volume growth Output of the 2019 study 3 extrapolation scenarios +1% total market growth 699 MW +3% total market growth Market Response Volume +5% total market growth
Slide 34: Table of contents
Table of contents 1 Goal, scope and planning of the 2019 update – Reminder 2 Results of the aggregated curves analysis – 2019 update 3 Conclusion X Back-up 34
Slide 35: Agenda
Agenda X Back-up A Extraction of the Market Response volumes B Refinement of the dataset C Statistical analysis D Implementation 35
Slide 36: There are multiple ways of looking at the data
There are multiple ways of looking at the data, depending on the focus of the analysis Cumulated dataset Additional data Winter months1) Winter 14/15 Refined dataset 2015-2017 Winter 15/16 2018 update (w/o 2014 and 2015-2018 Winter 16/17 strike days) 2019 update 2015-2019 Winter 17/18 Winter 18/19 D
Slide 37: Additional volumes of the 2019 update are coherent with the 2017 and 2018 datasets
C STATISTICAL ANALYSIS The additional volumes of the 2019 update do not present a specific behavior of the customers and are coherent with the dataset of the Market Response study 2017 and 2018 Distribution for the offer side only, cumulated dataset Raw dataset – 2017 Dataset w.o. y. 2014 – Dataset
Slide 38: The distribution of the updated dataset presents similar characteristics
C STATISTICAL ANALYSIS The distribution of the updated dataset (cumulated values since 2015) presents similar characteristics as the volume distribution of the 2017 study N° of hours 900 Volume distribution – Refined results 800 Standard deviation 700 update 2019 500€/MWh: 215 MW 600 150€/MWh: 195 M
Slide 39: Agenda
Agenda X Back-up A Extraction of the Market Response volumes B Refinement of the dataset C Statistical analysis D Implementation 39
Slide 40: Market response in winter 2018/2019 is not higher than previous years
C STATISTICAL ANALYSIS Despite a higher yearly MR average, the market response in the 2018/2019 winter season is not higher than the previous years, except for a peak in the 2nd and 6th weeks of 2019 Offer + demand (MW) 1 200 1 050 900 750 600 450 300 150 Week 44 45 46 47 48 49 50 51 52 1 2 3 4 5 6
Slide 41: Weekly averages are below those of last year
C STATISTICAL ANALYSIS The weekly averages for both the high and the low bounds are below those of last year (non-restricted hours), yet higher than winter periods 2015/2016 and 2016/2017 Offer + demand (MW) 1 200 1 050 900 750 600 450 300 150 Week Winter Winter Winter Winter 15/16 16/17 17/18 18/19
Slide 42: Main analysis of the 2017 study reconducted on the 2019 dataset
BACKUP – STATISTICAL ANALYSES The main analysis of the 2017 study was reconducted on the updated 2019 dataset, to assess the impact of Load, Temperature and Day-Ahead prices The goal of this analysis is to assess the impact of various parameters (price, temperature, load on Goal of the the Elia Grid
Slide 43: Restricting the dataset to important load does not increase precision
BACKUP – STATISTICAL ANALYSES Restricting the dataset to situations of important load on the Elia Grid does not increase the precision of the dataset: there is no significant impact on the average, but the standard deviation increases Average and standard deviation evolution Evolution of the size of
Slide 44: Restricting the dataset to high DA prices does not improve precision
BACKUP – STATISTICAL ANALYSES Restricting the dataset to situations of high DA prices does not improve the precision of the dataset: there is no impact on the standard deviation despite an increase of the average Impact of an increasing Day-ahead prices threshold Evolution of the size of the dataset
Slide 45: Restricting to certain temperature conditions gives no satisfying results
BACKUP – STATISTICAL ANALYSES Restricting the dataset to certain temperature conditions doesn’t provide any satisfying results: the volumes remain stable while the standard deviation increases Impact of a temperature threshold: restriction to all hours with a temperature lower than x °C Standard dev
Slide 46: Winter 2018-2019 shows a decrease of market response volumes but higher peaks
BACKUP – STATISTICAL ANALYSES During winter 2018-2019, we observe a global decrease of the market response volumes compared to the other winters, but higher peaks EVOLUTION OF THE DAILY AVERAGED MARKET RESPONSE VOLUME [01/01/2015-04/05/2019; Peak hours] ~ similar to ~ similar to non ~ similar to non
Slide 47: In December, a week of tension reveals important market response volumes
C STATISTICAL ANALYSIS – Study of some specific days of additional data In December, a week of tension on the electricity grid reveals important market response volumes, up to ~1000 MW Avrg. W18/19 A. Hourly day-ahead prices from December 3rd, 2018 to December 7th, 2018 [€/MWh – Peak hours] value 10
Slide 48: In January, a week of tension reveals important market response volumes
C STATISTICAL ANALYSIS – Study of some specific days of additional data In January, a week of tension on the electricity grid reveals important market response volumes, from ~1000MW up to ~2000 MW Avrg. W18/19 A. Hourly day-ahead prices from January 7th, 2019 to January 11th, 2019 [€/MWh – Peak hour
Slide 49: In February, a week of tension reveals important market response volumes
C STATISTICAL ANALYSIS – Study of some specific days of additional data In February, a week of tension on the electricity grid reveals important market response volumes, up to ~1100 MW Avrg. W18/19 A. Hourly day-ahead prices from February 4th, 2019 to February 8th, 2019 [€/MWh – Peak hours] value 10
Slide 50: Agenda
Agenda X Back-up A Extraction of the Market Response volumes B Refinement of the dataset C Statistical analysis D Implementation 50
Slide 51: Ancillary services projections for the 2019 update
Ancillary services projections provided for the 2019 update of the Market Response study R1 DR R3+DR Sum - DR FRR R3+ R1 Total Avg % DR R2 Total R2 DR Avg % DR need Total Min-max Min-max 2017 11831a 681a 171b 25,0% 1441a 0 10301a 5151b 50,0% 9-34 435-573 532 2018 11901a 811a 101b 12,3% 1391a 0 10801
Slide 52: Ancillary services projections for the 2018 update
Ancillary services projections provided for the 2018 update of the Market Response study R1 DR R3+DR Sum - DR FRR R3+ R1 Total Avg % DR R2 Total R2 DR Avg % DR need Total Min-max Min-max 2016 12031a 731a 141b 19,2% 1401a 0 10201a 4471b 43,8% 0-30 413-480 461 2017 11831a 681a 171b 25,0% 1441a 0 10301
Slide 53: Ancillary services projections for the 2017 study
Ancillary services projections provided for the 2017 Market Response study R1 DR R3+DR Sum - DR FRR R3+ R1 Total Avg % DR R2 Total R2 DR Avg % DR need Total Min-max Min-max 2015 12401a 831a 231b 27.7% 1401a 0 9111a 3211b 35,2% 19-26 344 2016 12031a 731a 141b 19,2% 1401a 0 10201a 4241b 41,6% 0-30 390