Hedging the Risk of Renewable Energy Sources in Electricity
Transcription
Hedging the Risk of Renewable Energy Sources in Electricity
Outline RES penetration Case study Modeling Results Conclusions Hedging the Risk of Renewable Energy Sources in Electricity Production Giorgia Oggioni1 Cristian Pelizzari2 Mercati energetici e metodi quantitativi: un ponte tra Università e Aziende Padova October 8th, 2015 1 University of Brescia, Department of Economics and Management, 25122 Brescia, Italy. E-mail: giorgia.oggioni@unibs.it. 2 University of Brescia, Department of Economics and Management, 25122 Brescia, Italy. E-mail: cristian.pelizzari@unibs.it. 1/ 38 Outline RES penetration Case study Modeling Results Conclusions Outline 1 Effects of renewable energy sources penetration 2 Wind strategies 3 Modeling wind penetration in a risk neutral world Reference equilibrium model: no wind energy production Modeling wind energy production 4 Results 5 Conclusions 2/ 38 Outline RES penetration Case study Modeling Results Conclusions The 20-20-20 European targets Europe 2020, the 2020 climate and energy package, sets demanding climate and energy targets to be met by 2020, known as the “20-20-20” targets: 20% reduction of GHG emissions by 2020 compared to 1990 through the EU Emissions Trading System (Directives 2003/87/EC and 2009/29/EC); 20% share of renewable energy sources (RES) based energy in final energy consumption by 2020 (Directive 2009/28/EC); 20% reduction in EU primary energy consumption by 2020, compared with projected levels, to be achieved by improving energy efficiency. In addition, Europe 2030, 2030 framework for climate and energy policies, and Europe 2050, Roadmap for moving to a low-carbon economy in 2050, set more ambitious objectives, to the aim of a full decarbonization of the energy sector. 3/ 38 Outline RES penetration Case study Modeling Results Conclusions Effects of RES penetration BUT RES penetration implies: 1 Intermittence of energy production; 2 Reduction of electricity prices that implies a significant revenue drop and thus: reduction of incentives to invest in conventional power plants; mothballing and/or dismantling of existing power plants, with the result that the security of supply becomes riskier and riskier. RES penetration has some side effects that need to be quantified in relation to the relevant market design! 4/ 38 0 Outline RES penetration Case study BULGARIA 690.5 Modeling ITALY 8,662.9 PORTUGAL* 4,914.4 Results Conclusions FYROM** 37 SPAIN 22,986.5 TURKEY 3,762.5 GREECE 1,979.8 Wind installed capacity in Europe MALTA 0 Installed End 2013 Installed 2013 2014 EU Capacity (MW) Austria 308.4 1,683.8 411.2 Belgium 275.6 1,665.5 293.5 Bulgaria 7.1 681,1 9.4 Croatia 81.2 260.8 85.7 Cyprus 146.7 Czech Republic 8 268.1 14 Denmark* 694.5 4,807 67 Estonia 10.5 279.9 22.8 Finland 163.3 449 184 France 630 8,243 1,042 Germany 3,238,4 34,250.2 5,279,2 Greece 116.2 1,865,9 113.9 Hungary 329.2 Ireland 343.6 2,049.3 222.4 Italy 437.7 8,557.9 107.5 Latvia 2.2 61.8 Lithuania 16.2 278.8 0.5 Luxembourg 58.3 Malta Netherlands 295 2,671 141 Poland 893.5 3,389.5 444.3 Portugal* 200 4,730.4 184 Romania 694.6 2,599.6 354 Slovakia 3.1 Slovenia 2.3 2.3 0.9 Spain 175.1 22,959.1 27.5 Sweden 689 4,381.6 1,050.2 UK 2,075 10,710.9 1,736.4 12,440.3 Total EU-28 128,751.4 11,357.3 117,383.6 11,791.4 End 2014 2,095 1,959 690.5 346.5 146.7 281.5 4,845 302.7 627 9,285 39,165 1,979.8 329,2 2,271.7 8,662.9 61.8 279.3 58.3 2,805 3,833.8 4,914.4 2,953.6 3.1 3.2 22,986.5 5,424.8 European Union: 128,751.4 MW Candidate Countries: 3,799.5 MW EFTA: 882.6 MW Total Europe: 133,968.2 MW Installed 2013 Candidate Countries (MW) FYROM Serbia Turkey 646.3 Total 646.3 EFTA (MW) Iceland 1.8 Liechtenstein Norway 110 Switzerland 13.3 Total Other (MW) Belarus Faroe Islands Russia Ukraine Total Total Europe End 2013 CYPRUS 146.7 Installed 2014 End 2014 2,958.5 2,958.5 37 804 841 37 3,762.5 3,799.5 1.8 771.3 60.3 1.2 48 - 3 819.3 60.3 125.1 833.4 49.2 882.6 4.5 95.3 99.8 12,228.5 3.4 6.6 15.4 371.2 396.7 11.7 126.3 138.0 12,819.6 3.4 18.3 15.4 497.5 534.7 133,968.2 121,572.2 * Provisional data ** Former Yugoslav Republic of Macedonia Note: due to previous year adjustments, 423.5 MW of project decommissioning, repowering and rounding of figures, the total 2014 end-of-year cumulative capacity is not exactly equivalent to the sum of the 2013 end-of-year total plus the 2014 additions. EWEA (2015). Wind in Power - 2014 European Statistics. Available at http://www.ewea. THE EUROPEAN WIND ENERGY ASSOCIATION org/fileadmin/files/library/publications/statistics/EWEA-Annual-Statistics-2014.pdf. 4 5/ 38 Outline RES penetration Case study Modeling Results Conclusions Wind strategies 6/ 38 Outline RES penetration Case study Modeling Results Conclusions Wind policies and assumptions Wind penetration levels 1 2 No wind penetration (reference case without wind production); Wind penetration (priority dispatch). Load and wind electricity production uncertainty 1 Load and wind-power scenarios. Wind derivatives (in a risk neutral world) 1 2 3 Call option (hedge of “too strong” wind); Put option (hedge of “too weak” wind); Monte Carlo pricing based on wind speed scenarios. Link between wind speed scenarios and load/wind-power scenarios 1 2 3 closeness of simulated wind-power duration curves to observed wind-power duration curve; probability distribution of observed distances; probabilistic assignment of wind speed scenarios to wind-power scenarios. 7/ 38 Outline RES penetration Case study Modeling Results Conclusions Reference equilibrium model: no wind energy production 8/ 38 Outline RES penetration Case study Modeling Results Conclusions Notation Sets m ∈ M : Set of plant types. M = res ∪ conv, where res and conv respectively indicate wind and conventional power plants; b ∈ B: Set of demand blocks. Parameters Gm : Capacity of plant type m (MW); db : Power consumed in block b (MWh); cm : Variable costs of plant type m (e/MWh); em : Emission factor associated to plant type m (ton/MWh); pCO2 : Allowance price (e/ton); pc: Price cap (e/MWh); Hb : Duration in hours of each block b. Variables gb,m : Power generated in block b by plant type m (MWh); gsb : Power sold in block b (MWh); nb : Shortage in block b (MWh); pb : Electricity price in block b (e/MWh). 9/ 38 Outline RES penetration Case study Modeling Results Conclusions Reference equilibrium model Generator’s profit maximization problem " X X Max Hb · pb · gsb − cm · gb,m − pCO2 · b m∈conv # X em · gb,m m∈conv subject to: Gm − gb,m ≥ 0 (ϕb,m ) X gb,m = gsb ∀b ∀m ∈ conv (ηb ) ∀b m∈conv gb,m ≥ 0 ∀b ∀m ∈ conv gsb ≥ 0 ∀b Clearing of the energy market Min X Hb · pc · nb b subject to: gsb + nb − db = 0 nb ≥ 0 (pb ) ∀b ∀b 10/ 38 Outline RES penetration Case study Modeling Results Conclusions Complementarity formulation of the reference equilibrium model 0 ≤ cm + em · pCO2 + ϕb,m − ηb ⊥gb,m ≥ 0 0 ≤ −pb + ηb ⊥ gsb ≥ 0 ∀b ∀m ∈ conv ∀b 0 ≤ Gm − gpb,m ⊥ ϕb,m ≥ 0 ∀ b ∀m ∈ conv X gb,m − gsb = 0 (ηb f ree) ∀ b m∈conv gsb + nb − db = 0 (pb 0 ≤ pc − pb ⊥ nb ≥ 0 f ree) ∀b ∀b 11/ 38 Outline RES penetration Case study Modeling Results Conclusions Modeling wind energy production 12/ 38 Outline RES penetration Case study Modeling Results Conclusions Notation Additional Sets s ∈ S: Set of scenarios considered in each block b. Additional Parameters θs,b : Wind power capacity factor in scenario s and block b (%); ds,b : Power consumed in scenario s and block b (MWh); τs,b : Probability of scenario s associated to block b; α: Wind derivative (call/put option) price (e/MWh); βs,b : Wind derivative (call/put option) payoff in scenario s and block b (e/MWh). Variables gs,b,m : Power generated in scenario s and block b by existing plant of type m (MWh); gss,b : Power sold in scenario s and block b (MWh); ns,b : Shortage in scenario s and block b (MWh); ps,b : Electricity price in scenario s and block b (e/MWh). 13/ 38 Outline RES penetration Case study Modeling Results Conclusions Generator’s profit maximization problem Max X τs,b · Hb · pb · gss,b − s,b − X τs,b · Hb · τs,b · Hb · pCO2 · X X em · gs,b,m m∈conv s,b + cm · gs,b,m m s,b X X τs,b · Hb · (βs,b − α) · X gs,b,m m∈res s,b subject to: Gm − gs,b,m ≥ 0 (ϕs,b,m ) ∀s, b ∀m ∈ conv Gm · θs,b − gs,b,m ≥ 0 (ϕs,b,m ) ∀s, b ∀m ∈ res X gs,b,m = gss,b (ηs,b ) ∀s, b m gs,b,m ≥ 0 gss,b ≥ 0 ∀s, b, m ∀s, b 14/ 38 Outline RES penetration Case study Modeling Results Conclusions Clearing of the energy market Min X τs,b · Hb · pc · ns,b s,b subject to: gss,b + ns,b − ds,b = 0 ns,b ≥ 0 (ps,b ) ∀s, b ∀s, b 15/ 38 Outline RES penetration Case study Modeling Results Conclusions Complementarity formulation of the wind equilibrium model 0 ≤ cm + em · pCO2 + ϕs,b,m − ηs,b ⊥gs,b,m ≥ 0 ∀ s, b ∀m ∈ conv 0 ≤ cm + ϕs,b,m − ηs,b + α − βs,b ⊥gs,b,m ≥ 0 ∀ s, b ∀m ∈ res 0 ≤ −ps,b + ηs,b ⊥ gss,b ≥ 0 0 ≤ Gm − gs,b,m ⊥ ϕs,b,m ≥ 0 ∀ s, b ∀ s, b ∀m ∈ conv 0 ≤ Gm · θs,b − gs,b,m ⊥ ϕs,b,m ≥ 0 ∀ s, b ∀m ∈ res X gs,b,m − gss,b = 0 (ηs,b f ree) ∀ s, b m gss,b + ns,b − ds,b = 0 (ps,b 0 ≤ pc − ps,b ⊥ ns,b ≥ 0 f ree) ∀s, b ∀ s, b 16/ 38 Outline RES penetration Case study Modeling Results Conclusions Wind derivatives 17/ 38 Outline RES penetration Case study Modeling Results Conclusions Wind derivatives: call and put option payoffs t: hour of the reference year, t ∈ {1, ..., 8760} Wt : wind speed at hour t, measured in m/s Wind call option designed to hedge “too strong” wind: it pays for hours with wind speed higher than a “strike price” strike price x × w̄t,Y : for each hour t, it is a factor x of the average wind speed w̄t,Y of the previous Y years χ: conversion coefficient from m/s to e/MWh expiration: end of the reference year Asian style option: the option sums the excess wind speeds of each hour of the reference year payoff: 8760 X payof fc = χ max(0, Wt − x × w̄t,Y ) t=1 Wind put option hedging of “too weak” wind payoff: payof fp = 8760 X χ max(0, x × w̄t,Y − Wt ) t=1 18/ 38 Outline RES penetration Case study Modeling Results Conclusions Wind derivatives: call and put option prices Pricing of wind options Call option fair price: αc = e −r EQ (payof fc ) assumptions: r (continuously compounded interest rate on an annual basis), Q ≡ P (risk neutral world) Monte Carlo method: generate N wind speed scenarios evaluate the option payoff for each wind speed scenario average the N option payoffs take the previous average as an approximation of EQ (payof fc ) PN i=1 payof fc,i , EQ (payof fc ) ≈ N where payof fc,i is the call payoff (in e/MWh) associated to wind speed scenario i 19/ 38 Outline RES penetration Case study Modeling Results Conclusions Wind derivatives: wind speed scenarios Input data: Hourly wind speeds of Germany in 2014 Data from NCAR are u-components and v-components of wind collected every 6 hours at 10 metre heights on a grid of 48 intersection points between parallels and meridians, transformed into wind speeds and then made hourly N = 100 Scenarios are based on Weibull distributions fitted on the 6-hour 48-point grid data, then made hourly Dependence of data is indirectly taken into account because scenarios are composed of wind speeds with the same time order of the input data 20/ 38 Outline RES penetration Case study Modeling Results Conclusions Wind derivatives: other settings Strike price: w̄t,Y : hourly average of German wind speeds in the last 12 years (2002 − 2013) call option: x is set equal to 1.01, 1.03, 1.05, 1.10, 1.15, and 1.20 put option: x is set equal to 0.99, 0.97, 0.95, 0.90, 0.85, and 0.80 r = 0.05% Conversion coefficient χ based on the slope coefficient of a linear regression model fitted to the 2014 data of wind speeds and wind electricity productions in Germany 100 wind speed scenarios (and corresponding option payoffs) assigned probabilistically to the 36 load/wind-power scenarios based on their closeness 21/ 38 Outline RES penetration Case study Modeling Results Conclusions Link between wind speed scenarios and load/wind-power scenarios 22/ 38 Outline RES penetration Case study Modeling Results Conclusions Load/Wind-power scenarios 1.00 0.80 0.90 0.70 0.80 0.60 0.70 0.50 Wind factor Demand factor 0.60 0.50 0.40 0.40 0.30 0.30 0.20 0.20 0.10 0.10 0.00 0.00 1 501 1001 1501 2001 2501 3001 3501 4001 4501 Hour 5001 5501 6001 6501 7001 7501 8001 8501 1 501 1001 1501 2001 2501 3001 3501 4001 4501 Hour 5001 5501 6001 6501 7001 7501 8001 8501 Figure: Load and wind-power capacity factor duration curves of Germany in 2014 - 3 levels for each of the 4 blocks of the load and wind-power duration curves. Scenarios based on the load and wind-power capacity factor duration curves of Germany in 2014. A total of 36 scenarios, 3 levels of wind-power capacity factors times 3 levels of load capacity factors times 4 load blocks. Baringo, L., A.J., Conejo (2013). Correlated wind-power production and electric load scenarios for investment decisions, Applied Energy, 101, 475–482. 23/ 38 Outline RES penetration Case study Modeling Results Conclusions Link between wind speed scenarios and load/wind-power scenarios (1) IDEA assess the similarity of each wind speed scenario generated for the wind derivatives to the wind-power scenarios of the equilibrium model assign wind speed scenarios to wind-power scenarios in a probabilistic way, in particular option payoffs payof fc,i and payof fp,i , with i = 1, ..., 100 do this for each of the 4 load blocks calculate the option payoffs, βs,b,c and βs,b,p for the call and the put respectively, for each wind-power scenario and load block as the average of the option payoffs assignments 24/ 38 Outline RES penetration Case study Modeling Results Conclusions Link between wind speed scenarios and load/wind-power scenarios (2) 35000 30000 Wind electricity production (MWh) 25000 20000 15000 10000 5000 0 0 1 2 3 4 5 Wind speed (m/s) 6 7 8 9 Figure: Wind electricity production against wind speed in Germany in 2014. Lydia, M., Kumar, S.S., Selvakumar, A.I., and G.E.P., Kumar (2015). Wind resource estimation using wind speed and power curve models, Renewable Energy, 83, 425–434. 25/ 38 Outline RES penetration Case study Modeling Results Conclusions Link between wind speed scenarios and load/wind-power scenarios (3) Sigmoid regression model fitted to the 2014 data of wind speeds and wind electricity productions in Germany Each wind speed scenario transformed into a wind electricity production scenario Sigmoid function value altered by a normally distributed number accounting for variability of production at different speeds Wind-power capacity factors calculated for each wind electricity production scenario Assessment of closeness of capacity factors of wind electricity production scenarios to capacity factors of wind-power scenarios (capacity factors adjusted by their standard deviation) Probability distribution estimation with higher probability to lower absolute differences Probabilistic assignment of wind electricity production scenarios to wind-power scenarios The last 3 steps repeated for each load block 26/ 38 Outline RES penetration Case study Modeling Results Conclusions Model additional assumptions Input data of the equilibrium model Electricity market: Germany Reference year: 2014 EU-ETS No EU-ETS: CO2 price 0 e/ton EU-ETS: CO2 price 40 and 50 e/ton Available technologies RES based plants: wind Conventional plants: nuclear, lignite, coal, CCGT, oil Conventional plant dismantling/mothballing Dismantling of 30% of the available nuclear capacity Mothballing of 30% of the available CCGT capacity 27/ 38 Outline RES penetration Case study Modeling Results Conclusions Results 28/ 38 Outline RES penetration Case study Modeling Results Conclusions No wind penetration Production per block and technology in MWh (reference capacity) 80,000 70,000 60,000 50,000 Electricity produc.on (MWh) Oil Gas 40,000 Coal Lignite Nuclear 30,000 Wind 20,000 10,000 0 No EU-‐ETS EU-‐ETS 40 EU-‐ETS 50 No EU-‐ETS EU-‐ETS 40 EU-‐ETS 50 No EU-‐ETS EU-‐ETS 40 EU-‐ETS 50 No EU-‐ETS EU-‐ETS 40 EU-‐ETS 50 Block 1 Block 2 Block 3 Block 4 29/ 38 Outline RES penetration Case study Modeling Results Conclusions No wind penetration Profits (Ke) No EU-ETS Revenues Costs Reference capacity 70% nuclear 70% CCGT 70%-70% 21,525,027 169,904,105 180,108,829 831,852,306 8,928,503 9,578,034 8,978,623 10,000,624 - - - - 12,596,524 160,326,071 171,130,206 821,851,683 Emissions Profits EU-ETS CO2 40 e/ton Revenues Costs Emissions Profits 33,501,692 180,059,147 193,309,454 837,123,275 8,957,224 9,594,775 9,007,344 10,017,365 14,519,622 15,054,114 14,450,594 14,982,118 10,024,845 155,410,259 169,851,516 812,123,792 EU-ETS CO2 50 e/ton Revenues 37,172,267 183,274,317 196,989,411 Costs 11,480,140 10,407,480 11,748,578 Emissions 15,393,016 16,458,072 16,528,039 Profits 10,299,110 155,067,667 170,053,892 838,820,818 11,219,308 17,408,326 810,193,183 30/ 38 Outline RES penetration Case study Modeling Results Conclusions No Wind vs. Wind penetration Prices e/MWh, no EU-ETS Wind No wind 1600.00 1400.00 1200.00 1000.00 800.00 600.00 400.00 200.00 0.00 Initial capacity 70% nuclear 70% CCGT 70%-70% 31/ 38 Outline RES penetration Case study Modeling Results Conclusions Wind penetration Production per block and technology in MWh (reference capacity) 80,000 70,000 Electricity produc.on (MWh) 60,000 50,000 Oil Gas 40,000 Coal Lignite Nuclear 30,000 Wind 20,000 10,000 0 No EU-‐ETS EU-‐ETS 40 EU-‐ETS 50 No EU-‐ETS EU-‐ETS 40 EU-‐ETS 50 No EU-‐ETS EU-‐ETS 40 EU-‐ETS 50 No EU-‐ETS EU-‐ETS 40 EU-‐ETS 50 Block 1 Block 2 Block 3 Block 4 32/ 38 Outline RES penetration Case study Modeling Reference capacity 70% nuclear 70% CCGT 70%-70% 21,135,150 38,868,742 105,009,753 327,265,260 Wind revenues 1,640,170 2,184,352 4,932,441 20,060,014 Costs 7,080,919 7,661,617 7,170,246 7,825,957 - - - - 15,694,402 33,391,476 102,771,947 339,499,317 Results Conclusions Wind penetration Profits (Ke) No EU-ETS Revenues Emissions Profits EU-ETS CO2 40 e/ton Revenues 32,857,787 49,625,557 117,120,313 335,587,664 Wind revenues 3,535,022 3,892,320 6,873,776 21,629,047 Costs 7,211,097 7,753,520 7,300,424 7,917,860 12,694,553 13,403,830 12,684,668 13,370,863 32,360,527 104,008,997 335,927,989 Emissions Profits 16,487,159 EU-ETS CO2 50 e/ton Revenues 36,272,524 52,866,220 120,522,051 338,189,507 4,030,516 4,368,912 7,396,694 22,083,426 Costs 10,282,913 10,615,683 9,263,817 9,715,466 Emissions 12,530,962 13,639,757 13,715,935 14,750,785 Wind revenues Profits 17,489,166 32,979,692 104,938,993 335,806,682 33/ 38 Outline RES penetration Case study Modeling Results Conclusions Introducing call options Wind electricity production (MWh) Call payoff net of call price (e) 100,000,000 30,000 29,000 50,000,000 28,000 Call 1% Call 3% Call 5% Call 10% Call 15% Call 20% 27,000 No EU-ETS No EU-ETS -50,000,000 EU-ETS EU-ETS 26,000 -100,000,000 25,000 24,000 -150,000,000 23,000 No Call Call 1% Call 3% Call 5% Call 10% Call 15% Call 20% -200,000,000 34/ 38 Outline RES penetration Case study Modeling Results Conclusions Introducing put options Wind electricity production (MWh) Put payoff net of put price (e) 300,000,000 35,000 30,000 250,000,000 25,000 200,000,000 20,000 No EU-ETS No EU-ETS 150,000,000 EU-ETS EU-ETS 15,000 100,000,000 10,000 50,000,000 5,000 No Put Put 1% Put 3% Put 5% Put 10% Put 15% Put 20% Put 1% Put 3% Put 5% Put 10% Put 15% Put 20% 35/ 38 Outline RES penetration Case study Modeling Results Conclusions Both call and put options Call and put payoffs net of corresponding call and put prices (e) Wind electricity production (MWh) 29,200 180,000,000 29,000 160,000,000 140,000,000 28,800 120,000,000 No EU-ETS 28,600 EU-ETS 100,000,000 No EU-ETS EU-ETS 80,000,000 28,400 60,000,000 40,000,000 28,200 20,000,000 28,000 No Call-Put Call-Put 1% Call-Put 3% Call-Put 5% Call-Put 10% Call-Put 15% Call-Put 20% Call-Put 1% Call-Put 3% Call-Put 5% Call-Put 10% Call-Put 15% Call-Put 20% 36/ 38 Outline RES penetration Case study Modeling Results Conclusions Conclusions and further steps Options can be beneficial, with exceptions (that could be the reason of a small (and OTC) market of wind derivatives) Design of scenarios to favor an integration between a financial approach and an (economic) equilibrium approach Consider a risk averse world by introducing a Value-at-Risk-based objective function on the side of the electricity producer in the equilibrium model and a market price of risk for the underlying asset (wind speed) of the wind derivatives 37/ 38 Outline RES penetration Case study Modeling Results Conclusions Thank you for your attention! 38/ 38