[PPT] - Active Intratemporal Prosumer Bidding Relaxing the Price Taking PowerPoint Presentation

SLIDE 1

Active Intratemporal Prosumer Bidding

Relaxing the Price Taking Assumption Presenter Christian Spindler (University of Vienna) Co-authors Oliver Woll (ZEW Mannheim) Dominik Schober (ZEW Mannheim) IAEE, Vienna 2017

1 / 22

SLIDE 2

Active Prosumer

New type of agent: Active Prosumer

Intratemporal, day-ahead market at each bidding period adapt strategy depending on market supply and demand forecasts Stage 1: prosumer decides how much of his own generation to use for his own demand Stage 2: remaining generation and demand is added to market supply and demand, final market equilibrium price and quantity are determined by the market operator

2 / 22

SLIDE 3

Recent developments in the Electricity Market

Decentralization and Decreasing Investment Costs for small scale generation units Ownership structure pivots towards small scale owners Decreasing/Limited Feed-in-Tariffs (BMJ 2017; OeMAG 2017)

Figure 1: Average Retail Price vs. FIT, Germany (Kairies et al. 2016)

3 / 22

SLIDE 4

Other Forms of Market Participation

Demand Side Bidding: consumers offer demand shifts or demand reductions, no production involved Intertemporal Arbitrage: production and demand shifts to use/generate different prices in different time periods; storage optional Virtual Bidding (Jha and Wolak 2015) : e.g. by financial institutions at EPEX, arbitrage between markets (day-ahead/real-time),no ownership of demand or generation units, no risk of operation Aggregator (Ottesen et al. 2016) legal entity that aggregates flexible generation and demand that has not been contracted forward and trades net demand/supply on the markets Ñ focus mainly on production side, e.g. oekostrom AG

4 / 22

SLIDE 5

Research Questions

Given this new type of agent in the electricity market, viz. a prosumer who controls both his

wn demand and his own generation:

What is this agent’s profit-maximizing strategy when we relax the price taking assumption? What are the implications for competition in the day-ahead market and its market design?

5 / 22

SLIDE 6

Model Setup I

Assumptions: competitive fringe / complete information prosumer’s generation cost and willingness to pay lower or equal to respective industry maximum copper plate capacity withholding possible no price taking

6 / 22

SLIDE 7

Model Setup II

Prosumer Decision Variables:

qPS

d

demand purchased on the market qPS

s

production offered on market

Linear Demand and Supply: pD

♣Q,α♣qPS

d qq ✏ aD ✁ ηD♣1 αq✁1 ☎ Q

(1) pS

♣Q,β♣qPS

s

qq ✏ aS ηS♣1 βq✁1 ☎ Q

(2)

where ηD and ηS = slope coefficients α and β = prosumer’s relative market sizes Q ✏ rqPS, q✁PSs, aD/aS = intercepts

7 / 22

SLIDE 8

Model Setup III

9 Possible Market Outcomes

Figure 2: Window of Attainable Market Clearing Prices and Quantities

8 / 22

SLIDE 9

Model – Constrained Optimization I

Objective Function: πPS

♣qPS

s

,qPS

d q ✏ p✝♣qPS

s q ✁ ♣p✝ τq♣qPS d q ✁ cPS♣qPS s

CPS

dem ✁ qPS d q

s.t. CPS

gen ➙ qPS s

CPS

dem ✁ qPS d

(3) CPS

dem ➙ CPS dem ✁ qPS d

(4) CPS

gen ➙ CPS dem ✁ qPS d

(5) Qind

s

qPS

s

✁ Qind

d

✁ qPS

d

➙ 0 (6) nonnegativity for decision variables and Lagrange multipliers (7)

9 / 22

SLIDE 10

Model – Constrained Optimization II

Figure 3: Active Prosumer Bidding; Optimal Generation

10 / 22

SLIDE 11

Model – Constrained Optimization III

Active prosumption weakly dominates passive prosumption.

Figure 4: Profit comparison with changing size of market demand Qind

d

black: active prosumer blue: self supply, excess production (Case B) red: self supply (Case A)

range: no self supply, no excess production (Case C)

purple: pure consumer (Case D)

11 / 22

SLIDE 12

Empirical Parameterization

Data: EEX hourly supply and demand bids, equilibrium prices and quantities, 2014-2016 Estimation of Demand and Supply Functions 6 typical days with 24h (144 typical hours) estimation based on Bigerna and Bollino (2014) ln♣demhq ✏ αhln♣phq ➳

i

βi,hdi ✝ ln♣phq (8) ln♣suphq ✏ αhln♣phq ➳

i

βi,hdi ✝ ln♣phq (9) where di represents working day/weekend, summer/winter/transition seasons

12 / 22

SLIDE 13

Results I

30 Prosumer Scenarios Capacity: 200/400/600/800/1000/1200 MW Ñ avg. demand 550 MW Production Costs: 10/20/30/40/50 EUR/MWh Ñ avg. market clearing price 29 EUR/MWh deriving optimal market production, market demand and self supply for each scenario

13 / 22

SLIDE 14

Results II

Figure 5: Market Prices and Corresponding Volumes for 144 Typical Hours

14 / 22

SLIDE 15

Results III

(a) (b) Figure 6: Prosumer Decision Variables for 144 Typical Hours, Capacity 800 MW, (a) Prod.Costs 20 EUR/MWh, (b) Prod.Costs 40 EUR/MWh

15 / 22

SLIDE 16

Results IV

Figure 7: Average Impact on Market Clearing Price

16 / 22

SLIDE 17

Results V

(a) (b)

Figure 8: Equilibrium outcomes according to (a) production capacity and (b) production costs Cases: A: Lone Wolf, B: Pure Monopolist, C: Merchant, D: Pure Monopsonist, AB: Strat Monopolist, AD: Strategic Monopsonist, BC: Partial Demander, CD: Partial Supplier, ABCD: Partial Demander and Supplier

17 / 22

SLIDE 18

Results VI

(a) (b)

Figure 9: Average profits of (a) passive prosumption and (b) active prosumption per typical hour

18 / 22

SLIDE 19

Results VII

Figure 10: Profit Advantage of Active Prosumption

19 / 22

SLIDE 20

Summary

Active Prosumer as a new type of agent (“Un-Unbundling”?) increasing importance for the electricity market active prosumption allows for a profit advantage Ñ Model potentially applicable to other markets: Airbnb Uber Amazon/Google Servers

20 / 22

SLIDE 21

Active Intratemporal Prosumer Bidding

Relaxing the Price Taking Assumption Presenter Christian Spindler (University of Vienna) Co-authors Oliver Woll (ZEW Mannheim) Dominik Schober (ZEW Mannheim) IAEE, Vienna 2017

21 / 22

SLIDE 22

References I

Bigerna, S. and C. A. Bollino (2014). “Electricity Demand in Wholesale Italian Market”. In: The Energy Journal 35.3,

pp. 25–46.

BMJ (2017). Gesetz f¨ ur den Ausbau ernerubarer Energien, EEG 2017. Berlin, DE. url: https://www.gesetze-im-internet.de/eeg_2014/BJNR106610014.html (visited on 05/29/2017). Jha, A. and F. A. Wolak (2015). Testing for Market Efficiency with Transaction Costs: An Application to Financial Trading in Wholesale Electricity Markets. url: https://web.stanford.edu/group/fwolak/cgi-bin/?q=node/3 (visited on 05/15/2017). Kairies, K.-P., D. Haberschusz, J. v. Ouwerkerk, J. Strebel, O. Wessels, D. Magnor, J. Badeda, and D. U. Sauer (2016). Wissenschaftliches Mess- und Evaluierungsprogramm Solarstromspeicher. RWTH Aachen, Aachen, Germany. url: http://www.speichermonitoring.de/fileadmin/user_upload/%20Speichermonitoring_Jahresbericht_2016_ Kairies_web.pdf (visited on 05/15/2017). OeMAG (2017). F¨

rderung. Photovoltaik. Vienna, AT. url: http://www.oem-ag.at/de/foerderung/photovoltaik/

(visited on 05/29/2017). Ottesen, S., A. Tomasgard, and S.-E. Fleten (2016). “Prosumer bidding and scheduling in electricity markets”. In: Energy 94, pp. 828–843.

22 / 22

SLIDE 23

Appendix I

BACKUP

23 / 22

SLIDE 24

Appendix II

Nomenclature

qPS

d

demand purchased on the market pS,pD inverse supply and demand functions qPS

s

production offered on market aS, aD intercepts of supply and demand functions cPS production costs of prosumer’s production unit CPS

gen

production capacity of prosumer cind maximum costs of production units available CPS

dem

total (hourly) prosumer demand ηD slope of demand function Qind

d

market demand of competitive fringe ηS slope of supply function Qind

s

production capacity of competitive fringe

24 / 22

SLIDE 25

Appendix III

Linear Demand and Supply: pD

♣Q,qPS

wnq ✏ aD ✁

aD Qind

d

qPS

d

☎ Q pS

♣Q,qPS

m q ✏ aS ♣cind ✁ aSq

Qind

s

qPS

s

☎ Q

25 / 22

SLIDE 26

Appendix IV

0.10
0.09
0.08
0.07
0.06
0.05
0.04
0.03
0.02
0.01

0.00 h1 h2 h3 h4 h5 h6 h7 h8 h9 h10 h11 h12 h13 h14 h15 h16 h17 h18 h19 h20 h21 h22 h23 h24

SO Work

(a)

0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.10 h1 h2 h3 h4 h5 h6 h7 h8 h9 h10 h11 h12 h13 h14 h15 h16 h17 h18 h19 h20 h21 h22 h23 h24

SO Work

(b)

Figure 11: Estimation Results for Price Elasticities: (a) Demand Elasticities, (b) Supply Elasticities

26 / 22

SLIDE 27

Appendix V

Direct and Indirect Effects of Decision Variables: dπ dqPS

d

✏ ❇π ❇qPS

d

❇π ❇p✝ ❇p✝ ❇α ❇α ❇qPS

d

(10) dπ dqPS

s

✏ ❇π ❇qPS

s

❇π ❇p✝ ❇p✝ ❇β ❇β ❇qPS

s

(11)

27 / 22

SLIDE 28

Appendix VI

Case Distinction:

q✝,PS

d

✏ ✩ ✬ ✬ ✬ ✬ ✬ ✬ ✬ ✫ ✬ ✬ ✬ ✬ ✬ ✬ ✬ ✪ max if dπ dqPS

d

→ 0 ñ C PS

dem

if p✝ τ ❇π ❇p✝ ❇p✝ ❇α ❇α ❇qPS

d

➔ cPS Y if dπ dqPS

d

✏ 0 ñ Y if p✝ τ ❇π ❇p✝ ❇p✝ ❇α ❇α ❇qPS

d

✏ cPS min if dπ dqPS

d

➔ 0 ñ max♣0, C PS

dem ✁ C PS genq

if p✝ τ ❇π ❇p✝ ❇p✝ ❇α ❇α ❇qPS

d

→ cPS (12) q✝,PS

s

✏ ✩ ✬ ✬ ✬ ✬ ✬ ✬ ✬ ✫ ✬ ✬ ✬ ✬ ✬ ✬ ✬ ✪ max if dπ dqPS

s

→ 0 ñ C PS

gen ✁ C PS dem q✝,PS d

if p✝ ❇π ❇p✝ ❇p✝ ❇β ❇β ❇qPS

s

→ cPS X if dπ dqPS

s

✏ 0 ñ X if p✝ ❇π ❇p✝ ❇p✝ ❇β ❇β ❇qPS

s

✏ cPS min if dπ dqPS

s

➔ 0 ñ if p✝ ❇π ❇p✝ ❇p✝ ❇β ❇β ❇qPS

s

➔ cPS (13)

28 / 22

SLIDE 29

Appendix VII

Figure 12: Profit advantage passive vs. active prosumption (x-axis: cPS; y-axis: π (in kEUR)) black: active prosumer blue: self supply, excess production (Case B) red: self supply (Case A)

range: no self supply, no excess production (Case C)

purple: pure consumer (Case D)

29 / 22