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
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
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
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
Research Questions Given this new type of agent in the electricity market, viz. a prosumer who controls both his own 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
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
Model Setup II Prosumer Decision Variables: q PS demand purchased on the market d q PS production offered on market s Linear Demand and Supply: d qq ✏ a D ✁ η D ♣ 1 � α q ✁ 1 ☎ Q p D (1) ♣ Q ,α ♣ q PS qq ✏ a S � η S ♣ 1 � β q ✁ 1 ☎ Q p S (2) ♣ Q ,β ♣ q PS s where η D and η S = slope coefficients α and β = prosumer’s relative market sizes Q ✏ r q PS , q ✁ PS s , a D / a S = intercepts 7 / 22
Model Setup III 9 Possible Market Outcomes Figure 2: Window of Attainable Market Clearing Prices and Quantities 8 / 22
Model – Constrained Optimization I s q ✁ ♣ p ✝ � τ q♣ q PS Objective Function: π PS d q ✏ p ✝ ♣ q PS d q ✁ c PS ♣ q PS � C PS dem ✁ q PS d q , q PS s ♣ q PS s s.t. C PS gen ➙ q PS � C PS dem ✁ q PS (3) s d C PS dem ➙ C PS dem ✁ q PS (4) d C PS gen ➙ C PS dem ✁ q PS (5) d Q ind � q PS ✁ Q ind ✁ q PS ➙ 0 (6) s s d d nonnegativity for decision variables and Lagrange multipliers (7) 9 / 22
Model – Constrained Optimization II Figure 3: Active Prosumer Bidding; Optimal Generation 10 / 22
Model – Constrained Optimization III Active prosumption weakly dominates passive prosumption. Figure 4: Profit comparison with changing size of market demand Q ind d blue: self supply, excess production (Case B) black: active prosumer red: self supply (Case A) orange: no self supply, no excess production (Case C) purple: pure consumer (Case D) 11 / 22
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 ♣ dem h q ✏ α h ln ♣ p h q � β i , h d i ✝ ln ♣ p h q (8) i ➳ ln ♣ sup h q ✏ α h ln ♣ p h q � β i , h d i ✝ ln ♣ p h q (9) i where d i represents working day/weekend, summer/winter/transition seasons 12 / 22
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
Results II Figure 5: Market Prices and Corresponding Volumes for 144 Typical Hours 14 / 22
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
Results IV Figure 7: Average Impact on Market Clearing Price 16 / 22
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
Results VI (a) (b) Figure 9: Average profits of (a) passive prosumption and (b) active prosumption per typical hour 18 / 22
Results VII Figure 10: Profit Advantage of Active Prosumption 19 / 22
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
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
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¨ orderung. 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
Appendix I BACKUP 23 / 22
Appendix II Nomenclature q PS demand purchased on the market p S , p D inverse supply and demand functions d q PS production offered on market a S , a D intercepts of supply and demand functions s c PS C PS production costs of prosumer’s production unit production capacity of prosumer gen c ind C PS maximum costs of production units available total (hourly) prosumer demand dem Q ind η D slope of demand function market demand of competitive fringe d Q ind η S slope of supply function production capacity of competitive fringe s 24 / 22
Appendix III Linear Demand and Supply: a D p D own q ✏ a D ✁ ☎ Q ♣ Q , q PS Q ind � q PS d d m q ✏ a S � ♣ c ind ✁ a S q p S ☎ Q ♣ Q , q PS Q ind � q PS s s 25 / 22
Appendix IV SO Work SO Work 0.00 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 0.09 -0.01 0.08 -0.02 0.07 -0.03 0.06 -0.04 0.05 -0.05 0.04 -0.06 0.03 -0.07 0.02 -0.08 0.01 -0.09 0.00 -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 (a) (b) Figure 11: Estimation Results for Price Elasticities: (a) Demand Elasticities, (b) Supply Elasticities 26 / 22
Appendix V Direct and Indirect Effects of Decision Variables: ❇ p ✝ d π ❇ π � ❇ π ❇ α ✏ (10) dq PS ❇ q PS ❇ p ✝ ❇ q PS ❇ α d d d ❇ p ✝ ❇ π � ❇ π ❇ β d π ✏ (11) ❇ p ✝ dq PS ❇ q PS ❇ β ❇ q PS s s s 27 / 22
Recommend
More recommend
