Presentation on theme: "Www.eprg.group.cam.ac.uk Auctions and Future Energy Systems: Micro auctions for distributed generation with flexible zones Thomas Greve and Michael G."— Presentation transcript:
Auctions and Future Energy Systems: Micro auctions for distributed generation with flexible zones Thomas Greve and Michael G. Pollitt and 2 nd Interdisciplinary Workshop on Smart Grid Design and Implementation University of Florida 29 March 2014
2 This project is part of: – EPSRC Autonomic Power Project. – Started in 2011 and ends in – 8 universities and 9 industrial partners. This project focuses on the electricity network of 2050: – the energy networks of the future will be far more difficult to manage and design than those of today, for technical, social and commercial reasons. – need for ’self-*’ network. The Autonomic Power Project Source: APS (2011)
3 Four key areas of research 1.Self* network operation and control. 2.Resilience and risk management. 3.Active participation of consumers. 4.Autonomic economics. - WP1: Economic thinking zones for future complex power systems. - WP2: Micro auctions for distributed generation with flexible zones.
The challenge In short: (1) increasing energy prices (possibly!) we need a tool to hold down prices. (2) changing distribution of demand we need a tool to secure that cheapest supplier follows demand. (3) new energy sources and an increasing in small-scale renewables we need a tool to respond to an unpredictable market. (4) operators to consider multiple markets we need a tool to respond to the possible exercise of market power and collusion. Assumption: Unlimited computer power.
5 In short: (1) increasing energy prices (possibly!) we need a tool to hold down prices and increase price legitimacy – auction design. (2) changing distribution of demand we need a tool to secure that cheapest supplier follows demand – flexible zones. (3) new energy sources and an increasing in small-scale renewables we need a tool to respond to an unpredictable market – proxy agent (auction design) and Quality of service. (4) operators to consider multiple markets we need a tool to respond to the possible exercise of market power and collusion – proxy agent (auction design), auction design and flexible zones. Our contribution
6 The Beauty of Auctions Fairness: select the outcome that achieves a certain fairness criterion in utility. Budget-balanced: implement outcomes that have balanced transfers across players. Efficiency (truth-telling): select the outcome that maximizes total utility. Revenue maximization/Cost minimisation (typically referred to as optimality): select the outcome that maximizes revenue to a seller/minimizes costs to a buyer. Pareto optimality (alternative): occurs when no one can be made better off without making someone worse off. With truth-telling: the winner will always be the person with the highest valuation of the good. Therefore, the item belongs to the person/group who values it most, social welfare is optimised.
7 Second-price sealed-bid auction All sellers submit bids. Seller submitting the lowest bid wins and gets paid the second lowest bid. Bidding strategy is easy –Bidding one’s true valuation is a dominant strategy. Intuition: –The amount a bidder pays is not dependent on her bid. $700 $400 $500 $300 WINNER! Receives $400
8 Uniform price auction Also known as a the "clearing price auction“. A multi-unit auction. A number of identical units are sold for the same price. In theory, the uniform-price auction provides an incentive for bidders to bid insincerely (unless each bidder has demand for only a single unit). An inefficient equilibrium. Example revenue auction Consider a situation in which there are six units (K =6) to be sold to three bidders and the submitted bid vectors are –b1 = (50, 47, 40, 32, 15, 5), –b2 = (42, 28, 20, 12, 7, 3), –b3 = (45, 35, 24, 14, 9, 6), –b = (50, 47, 45, 42, 40, 35), next is 32. The market-clearing price is 32. Bidder 1 wins three units and pays a total of 3*32 = 96. Bidder 2 wins one unit and pays a total of 1*32 = 32. Bidder 3 wins two units and pays a total of 2*32 = 64.
9 Proxy agent The proxy agent will be used: –as an aggregator of bids for a number of different small-scale renewables. –to determine the seller’s Willingness To Accept (WTA) and therefore, compute the bidders optimal strategy. The optimal strategy for a seller is to tell the proxy agent its WTA. A proxy agent is an agent or software which submits bids on a bidder behalf. The bid is decreased by a minimum increment (in theory conducted with negligibly small bid increments) until the second-lowest submitted proxy bid is exceeded by another bid. The bidder who submitted the lowest proxy bid wins the object and receives the price that is equal to the second-lowest (proxy) bid.
10 At the moment the auction is not efficient. At the moment there is no variation in bids to help and handle special groups (like small scale suppliers). Active use of proxy agents - the optimal bid. Why auctions!
11 At the moment there are no flexible zones! Can secure demand equal to supply within an internally unconstrained area: –Place a zone in the right place, ensure local supply=demand. Can secure that lowest cost suppliers deliver. Therefore, help the auction to secure welfare improvement or optimality. Why flexible zones!
12 Existing (distribution) networks are not ready to cope with the uncertainty and complexity in future networks and far away from being ready for the complexity which might exist by An efficient auction. Flexible zones. We need a method to help and handle small scale suppliers. We need an auction and zoning design to secure truth-telling (”efficiency”) and an efficient allocation (”optimality”) in all zones. Aim: lowest possible consumer prices and welfare improvement Autonomic Power Project
13 To operate with flexible zones -improve initial zones and use flexible zones to find lowest cost combination of suppliers (i.e. generators or DR aggregators). -Locational pricing with competition can provide larger benefits. To design an auction that can take advantages of the flexible zones and handle or use small scale suppliers. Rules for suppliers: 1.A supplier has to be signed up for auction if it wants to participate. 2.A proxy agent will aggregate all small small-scale renewables and submit bids. 3.If a small-scale supplier is not connected to a proxy agent, it may sign-up for the auction. If a supplier is not signed up, the supplier cannot get access to a distribution line. 4.Quality of service/subscriptions - to respond to the unpredictable nature of renewable generation: Gold subscription = electricity delivered when demanded Silver subscription = electricity delivered within 15 min Bronze subscription = electricity delivered within 30 min Tools - Flexible zones and auctions Assumption: Active consumers
14 Auction design 1 – English auction with proxy bidding (ebay auction)
15 It is March 26, auction starts for March 26, time interval am 8 suppliers or 8 proxy agents have signed up for auction: Zone 1 Gas 1 Wind 1 SSS 1a SSS 1b ND 1 ND 2 Zone 2 Wind 2 Gas 2 SSS 2b Wind 2a Demand 20 Zone 1 Demand 20 Zone 2 10 MWh Gold Initial zoning SSS: small scale suppliers ND: negative demand 10 MWh Silver 10 MWh Gold 10 MWh Silver
16 Consider the bidding process for Gold in Zone 1. Two identical items offered for sale – a licence to deliver 1 MWh Gold and a licence to deliver 1 MWh Gold, in 10 simultaneous auctions. Assume that each bidder only desires one item (in each auction). Gas 1 instructs her proxy to bid down to £163 per MWh for the item, and wants three MWh. Wind 1 instructs her proxy to bid down to £120 per MWh, and wants three MWh. SSS 1a instructs her proxy to bid down to £123 per MWh, and wants that it wants five MWh. SSS 1b instructs her proxy to bid down to £122, and wants three MWh. The reserve price is set to £165 per MWh. Initial zoning – proxy agent bids
17 The rules of the auction Every bidder has to submit bids (/values) to a proxy agent. Each bidder can make one bid in each auction. Multi-unit action. The auctioneer looks at all the bids and divides the auctions (the 10 auctions) after highest valuations. The two lowest proxy bidders win one item each at a price that is an increment below the highest losing bid.
18 Initial zoning, zone 1 – submitted bids Table 1 Submitted bids in zone 1, auction 1 Winners: Wind 1 and SSS 1b (win one MWh each). Price: £ per MWh.
19 Zone 1 - result Table 2: Winning bidders and allowed price, zone 1, Gold Result Auction 1: Wind 1 and SSS 1b win one MWh each. Auction 2 and 3: the same result. Auction 4 and 5: Gas 1 and SSS 1a win two MWh each, price of £165 per MWh. Auction 6-10: No bidders. Remember
20 Zone 2 - result Table 3: Winning bidders and allowed price, zone 2, Gold Result Auction 1: Wind 2 and SSS 2b win one MWh each. Auction 2-7: the same result. Auction 8-10: Gas 2 and SSS 2a win three MWh each, price of £160 per MWh. Note Wind 1 and SSS 2b can deliver extra 4 MWh. Notice
21 New zoning? The APS will self-optimise and self-configure if it can see a lower cost network. Assume that all bidders can deliver to consumers in both zones, and physical constraints allow this. Suppose that it is possible to combine Wind 1, Wind 2, SSS 1b and SSS 2b in a new zone. Old zone 1 Gas 1 Wind 1 SSS 1a SSS 1b ND 1 ND 2 Old zone 2 Wind 2 Gas 2 SSS 2b Wind 2a New zone
22 New zone Table 4: Winning bidders and allowed price, new zone, Gold Table 5 Result with and without flexible zones, new zone, Gold
23 Disadvantages Difficult to explain the second-price rule, but with proxy bidding it makes sense. Dynamic auction, rounds, but with unlimited computer power this disadvantage is eliminated. Not a package auction. Have heterogeneous – Gold, Silver and Bronze. Is there a better auction? Advantages The Autonomic Power Project idea is in the auction design. The auction design means that the auctioneer does not have to choose the prices before learning bidders’ preferences. The auction allows bidders to submit bids on more variables – Gold, Silver, Bronze, Reserve, Big consumers etc. The auction is efficient. We can help, handle and control small-scale renewables. Active use of proxy agents promote competition and makes the second-price rule interesting. Gives the consumers more options – consumers participation (washing machine). Day-Ahead auction and Real-Time Market can work together. Advantages/Disadvantages of the design
27 Initial zoning, zone 1, the lots Table 6: Number of lots offered for sale and reserve prices per lot Table 7: Penalty for bids on lot Gold1-3 and lot Silver1-3 Lots to auctioned – two categories 1.Available if the network self-optimise and self-configure. 2.Only deliver to initial zone = penalty.
28 Principal stage – primary round A clock auction. Promote price discovery. Gives a sense of what the prices are and possible package to win. Bid increments together with activity rules yield higher efficiency. Table 8: The clock auction
29 Principal stage – supplementary round A single round opportunity - best offer for all available packages. A sealed-bid round. Subject to caps. Can fully express their preference for packages of lots/express its true value for these packages. Table 9: The supplementary round
30 Winner determination and base prices Vickrey-Clarke-Groves (VCG) mechanism, a truth-telling mechanism. Table 10: All bids made in the auction Table 11: Lowest value combination
31 Assignment stage Will determine the specific assignment of lots to each winning bidder, subject to the number of lots won in the principal stage. Is there because of the restriction on line capacity lots Gold 1-3 and Silver 1-3 may be more desirable. A sealed-bid round. Maximize total value across all winning bidders. Table 11: All bids made in the auction Table 12: All bids made in the auction
32 Final assignment and allowed price for desired lots Table 13: Winning bidders and allowed price for desired lots SSS 1a and SSS 1b want the same package. However, SSS 1a has submitted the highest price reduction. SSS 1a wins Gold 1-3 and Silver 1-3. Wind 1 and SSS 1b pay zero. SSS 1a’s base price is £882. First, according the VCG mechanism SSS 1a’s price has to be reduced by £1, SSS 1b’s bid. Second, SSS 1a has to pay a penalty of 1% for both Gold and Silver lots. SSS 1a pays a price of (£882-£1)*0.02 =£18. Remember
33 Zone 2 Table 15: Winning bidders and allowed price for desired lots, zone 2 Table 14: All bids made in the auction Notice
34 Zone 2 and new zone Table 17: Winning bidders and allowed price for desired lots, new zone Table 16: Winning bidders and allowed price for desired lots, Zone 1 and 2
35 Disadvantages Dynamic auction, rounds, but with unlimited computer power this disadvantage is eliminated. Greater efficiency. Advantages The Autonomic Power Project idea is in the auction design. The auction design means that the auctioneer does not have to choose the prices before learning bidders’ preferences. The auction allows bidders to submit bids on more variables – Gold, Silver, Bronze, Reserve, Big consumers etc. A package auction. We can help, handle and control small-scale renewables. Active use of proxy agents promote competition and makes the second-price rule interesting. Gives the consumers more options – consumers participation (washing machine). Day-Ahead auction and Real-Time Market can work together. Advantages/Disadvantages of the design
36 Design 1 – English auction with proxy bidding - (1) Efficient, (2) Not a package auction Design 2 – Package Clock Auction – (1) Greater efficiency, (2) A package auction SO – (1)If efficient – placing the MWh in hands other than those who value them the most. (2)the flexible zones secure – “the right supplier” is matched with “the right consumer” -lower prices -Pareto improvement or maybe optimality (2) Proxy agent: -Can handle and help small scale suppliers and promote competition -Secure the optimal WTA Conclusion
Next Steps Working on different auction designs. Need to test our flexible zoning for feasibility. Need to test our auction designs for feasibility. Mechanism for market power. A software to check for collusion – filter.