1 Impact of High Variable Reewable Geeratio o Future Market Prices ad Geerator Reveue P. Vithayasrichareo, Member, IEEE, J. Riesz, Member, IEEE ad I. MacGill, Member, IEEE Abstract This study assesses the potetial impact of high reewable geeratio o the spot electricity prices, geerator reveue ad profits i a -oly electricity market. I particular, it presets modellig outcomes for the Australia Natioal Electricity Market (NEM) with a rage of possible reewable peetratios i 2030. It is assumed that the curret reliability stadard is maitaied ad participats deploy short ru margial cost biddig. The study foud that icreasig the share of wid ad PV geeratio would likely result i lower average spot prices ad subsequetly reveue ad profit of geerators. The reveue impact o large-scale PV was foud to be very severe ad could lead to isufficiet reveue to cover the costs, particularly at higher reewable peetratios. Chages i market mechaisms, such as icreasig the Market Price, may be required to esure reveue sufficiecy ad log-term resource adequacy i a -oly market with high reewables. Idex Terms Reveue sufficiecy, oly market, reewable, Australia Natioal Electricity Market (NEM) I. INTRODUCTION ENEWABLE geeratio sources, particularly wid ad Rsolar photovoltaics (PV), are fast becomig major geeratio sources i a umber of electricity idustries. This is due to fallig solar ad wid techology costs ad growig cocers over climate chage ad security. Due to the variable availability ad somewhat upredictable ature of wid ad PV geeratio, there are cocers over the potetial impacts of such reewable sources o the electricity idustry. For restructured electricity idustries with competitive market arragemets, the high capital yet low operatig costs (short ru margial cost or SRMC) of these techologies poses some iterestig additioal challeges. I particular, growig peetratios of low SRMC reewable geeratio i -oly wholesale markets are likely to reduce spot electricity prices ad hece market returs to all geerators. The risk of isufficiet reveue to recover both fixed ad variable operatio costs is oe of the major cocers for geerators. Cocers over reveue sufficiecy are also shared by may policy makers ad market regulators give that this might lead to log-term resource adequacy challeges by promotig early retiremet ad deferred etry to the market, which ca reduce the reliability of the electricity supply [1, 2]. This work has bee supported i part by Australia Reewable Eergy Agecy (ARENA) ad the CSIRO Future Grid Project. The authors are with the Cetre for Eergy ad Evirometal Markets ad School of Electrical Egieerig ad Telecommuicatios, UNSW Australia, Sydey, Australia (email: peerapat@usw.edu.au). The Australia Natioal Electricity Market (NEM) is a moderately sized market (aroud 35GW of peak demad ad 200TWh per year) with growig wid ad solar deploymet ad sigificat reewable resource potetial, It features a relatively trasparet -oly market with relatively few costraits imposed o geeratio offers, ad therefore provides a iterestig case study for aalysis of high reewable scearios, ad their reveue implicatios. Previous studies have explored the techical feasibility ad ecoomics of high reewable scearios i the NEM, icludig scearios of 100% reewable [3, 4]. However, these studies have ot directly quatified the reveue implicatios of these high reewable systems. Some observers have raised questios about the feasibility of the NEM s -oly market desig i high reewable scearios, icludig claims that a system composed of a majority of low SRMC geeratio may ot deliver appropriate commercial icetives for assured resource adequacy [5]. This study aims to examie the possible impact of high reewable peetratios o spot electricity prices, geerator reveues ad profits i a future Australia NEM i 2030, with a view to assessig the potetial viability of the preset -oly market ad its mechaisms to esure resource adequacy ad hece log-term reliability. The paper provides quatitative aalysis usig a log-term geeratio portfolio plaig ad ivestmet modellig tool first developed i [6]. A umber of high reewable peetratios are cosidered icludig ucertaities associated with these. The modellig assumes that the curret NEM reliability stadard is maitaied ad participats deploy SRMC biddig. II. METHODOLOGY This study uses a probabilistic geeratio portfolio modellig tool which exteds the commoly applied load duratio curve (LDC) based optimal geeratio mixes by usig Mote Carlo simulatio to icorporate key ucertaities ito the assessmet [6]. These ucertaities iclude future gas costs, carbo policies ad electricity demads. The tool determies a probability distributio of aual reveue, operatig costs ad profits/losses of each geeratio techology for differet possible geeratio portfolios. The expected aual reveue, operatig cost ad profit of each geeratio techology for a particular portfolio represet the average of all the simulated reveue, costs ad profits from every Mote Carlo ru. Geerators obtai reveue through a spot market based upo the spot electricity price (or market clearig price ) i each period.
2 Geerators are dispatched based o their SRMC with the objective of miimizig the total system operatig cost of meetig demad i a year subject to demad balacig costraits. SRMC is the sum of the fuel, variable operatios ad maiteace (O&M) ad greehouse emissios costs of each uit. The modellig assumes that geerators bid ito the market at their SRMCs ad the spot price is the cost to supply the last MW of electricity to meet demad. PV ad wid geeratio is icorporated ito the modellig through the use of a residual (et) load duratio curve (RLDC) approach to capture the chroology of PV ad wid resource variability ad its match to NEM electricity demad, based upo historical correlatios observed i 2010. As the lowest SRMC geeratio, PV ad wid geeratio is dispatched first i the merit order. With this approach, hourly simulated PV ad wid geeratio is subtracted from hourly demad over the year to obtai residual demad, which is the rearraged to obtai a RLDC. It is this curve which has to be met by covetioal techologies i the portfolio. A 15% miimum sychroous geeratio requiremet is applied i all dispatch periods to provide adequate system iertia, fault feed-i levels ad system stability [7]. This represets the miimum amout to which aggregate covetioal geerators ca be tured dow. This costrait is importat for high reewable scearios sice some of the most promisig kids of reewable geeratio (otably wid ad PV) are o-sychroous ad therefore do ot geerally provide iertia ad fault feed-i curret to the system [1]. For the purposes of this study, coal, gas ad hydro plats are assumed to provide sychroous geeratio (some types of reewable geeratio such as solar thermal, geothermal ad biomass are also sychroous, although these techology types have ot bee modelled i this study). I additio to the market reveue, covetioal geerators also receive a supplemetary paymet i periods i which they are dispatched out of merit order to satisfy the 15% sychroous requiremet costrait. This supplemetary paymet is referred to i this study as a costraied o paymet, ad is determied based upo the SRMC of the most expesive covetioal geerator that is dispatched to meet the sychroous requiremet. For each Mote Carlo ru, aual reveue, of each geeratio techology is calculated accordig to Eq. (1) (3). Spot market reveue Costraied o paymet REVSpot P MC (1) T t1 T t1 CP Pcov SRMC max Aual reveue REVTotal = REVSpot + CP (3) where P is the geeratio output of techology (MW), MC t is the market clearig price ($/MWh), Pcov is the output of covetioal geerator (MW) that is dispatched out of merit order to meet the sychroous requiremet, SRMCmax t is the SRMC of the most expesive geerator ($/MWh) that is dispatched to meet the sychroous requiremet i period t. Aual operatig cost ad profit of the geerator are determied based upo Eq. (4) (5) respectively. t t (2) OPEX P SRMC (4) T t1 OPProfit = REVTotal OPEX FOM (5) where SRMC is the SRMC ($/MWh) of techology i period t ad FOM is the aual fixed operatig ad maiteace (O&M) costs ($) of techology. III. THE AUSTRALIAN NATIONAL ELECTRICITY MARKET (NEM) CASE STUDY Six differet reewable peetratio scearios for the NEM i 2030 were cosidered: 15%, 30%, 40%, 60%, 75% ad 85% (by cotributio). Eight techologies were icluded: coal, combied cycle gas turbie (CCGT), ope cycle gas turbie (OCGT), co-geeratio, distillate, utilityscale PV (sigle axis trackig), wid (o shore) ad hydro. The reewable peetratio scearios ad the percetage of each reewable techology are summarised i TABLE I. Note that the proportio of PV ad wid for each reewable peetratio were selected based o assumptios o future ivestmet scearios, as explaied i [8]. Reewable peetratio scearios TABLE I DIFFERENT RENEWABLE PENETRATION SCENARIOS Achieved total reewable peetratio % PV % Wid % Hydro % Fossil 15% 14 4 4 6 86 30% 27 7 14 6 73 40% 40 10 24 6 60 60% 60 20 34 6 40 75% 73 30 37 6 27 85% 83 39 35 9 17 The maximum spot price is set at $13,500/MWh, which is the curret Market Price Cap (MPC) for the NEM [9]. This price is triggered i periods whe demad exceeds available geeratio capacity. The istalled capacity was determied so that each geeratio portfolio will, o average, meet the preset NEM reliability stadard of 0.002% aual userved (USE). A. Hourly Demad ad Geeratio A hourly electricity demad profile for 2029-2030 was obtaied from aalysis by the Australia Eergy Market Operator (AEMO) o a 100% reewables system uder a moderate ecoomic growth sceario. Hourly wid ad solar output profiles for 2030 were simulated from hourly traces of 1-MW o-shore wid ad solar PV (sigle axis trackig) geeratio i differet locatios across the NEM provided by AEMO [7]. For hydro geeratio a aual hydro dispatch limit of 13 TWh was applied, based upo the logterm average hydro geeratio estimated by AEMO [7]. Geeratio output of each thermal techology (coal, CCGT, OCGT, coge ad distillate) i each period was determied usig merit order dispatch based upo their SRMCs i 2030. Techical ad cost parameters of geeratig plats were based upo a previous study preseted i [8].
3 B. Modellig Ucertaities Key ucertai parameters cosidered i the modellig are gas prices, carbo prices ad electricity demad as they have experieced a higher degree of ucertaity tha other variables [10, 11]. Logormal distributios were applied to model future fuel ad carbo prices to reflect the asymmetric dowside risk associated with high price outcomes. Demad ucertaity was modelled assumig a ormal distributio of residual peak demad for each reewable peetratio sceario. Both logormal ad ormal distributios ca be characterized by their mea (expected value) ad stadard deviatio (SD). The mea ad SD of fuel prices ad carbo prices were determied based upo Australia Govermet estimates for 2030 [12, 13]. Correlated samples of coal, gas ad carbo prices were simulated from their margial logormal distributios 10,000 times usig Multivariate Mote Carlo simulatio techiques described i [6]. The mea ad SD of peak demad were estimated based o the Probability of Exceedace (POE) demad projectios i 2029-2030 provided i [7], ad were explaied i detail i [8]. Residual peak demad were also simulated 10,000 times. I order to achieve the 0.002% USE reliability stadard o average, there were istaces where the simulated residual peak demads exceeded the istalled fossil-fuel geeratio capacity. IV. MODELLING RESULTS AND ANALYSIS With Mote Carlo simulatio techiques, the modellig calculated overall geeratio costs, emissios, reveue ad operatig profits for each techology withi each possible geeratio portfolio for 10,000 simulated future fuel prices, carbo prices ad electricity demads. The cost of USE is valued at the MPC ($13,500/MWh), ad is icluded i the overall geeratio cost. Fig. 1 illustrates the efficiet frotiers cosistig of optimal geeratio portfolios i terms of expected geeratio cost ad cost risk (SD of cost) for differet geeratig portfolios, ragig from 15% to 85% reewable geeratio. Each dot is a plot of a portfolio s expected costs (agaist the vertical axis) ad the cost risk (agaist the horizotal axis), calculated over 10,000 simulatios. 1 As illustrated i Fig. 1, the lowest cost geeratio portfolio features 60% reewable, with a expected cost of $92/MWh. The costs rise as reewable icreases to 75% ad 85%, ad are also higher for the reewable peetratio levels below 60%. For the purpose of this discussio, oly the reveue ad profits of each techology i the least cost portfolio for differet reewable peetratios are quoted. For example, the least cost portfolio for the 15% reewable portfolio is the oe that cosists of 41% coal, 21% CCGT, 7% OCGT, ad this lowest cost portfolio is used as the basis for aalysis. The average spot price duratio curve for the least cost portfolio i each reewable peetratio for the highest 2% price periods is show i Fig. 2. The figure also shows the correspodig PV ad wid geeratio outputs i those periods. The results suggest that the magitude of price spikes 1 For each reewable peetratio, the amout of distillate, cogeeratio, hydro, PV ad wid capacity was fixed for every possible thermal portfolio. icreases with higher reewable peetratios but the high price periods (e.g. greater tha $500/MWh) are less frequet. For example, the average highest spot price i the 85% reewables sceario is aroud $8,500/MWh compared to $1,500/MWh i the 15% reewables sceario. However the umber of periods where the spot prices are greater tha $500/MWh is less tha 0.4% of the time (35 hours per year) i the 85% reewables sceario compared to 2% of the time (75 hours per year) i the 15% reewables sceario. 2 Fig. 1. Efficiet frotiers cotaiig optimal geeratio portfolios for differet reewable peetratios i 2030. The capacity of fossil-fuel techologies i each portfolio is show i GW (i brackets) ad percetage share. The coloured boxes show the share of each techology by capacity istalled. Fig. 2. Average market price duratio curve for the top 2% of the price periods ad the correspodig PV ad wid geeratio. Sice the model does ot icorporate strategic biddig behaviour, high spot prices i the model are drive by periods where userved is occurrig. This meas that the price duratio curves illustrate that there are fewer periods of supply ad demad imbalace as the reewable peetratio icreases. However, the magitude of USE occurrig i each of those periods is higher (USE is cocetrated ito fewer periods as 2 Note that the figure shows the average spot price across 10,000 simulated fuel prices, carbo price ad electricity demad for each period. Without modellig the ucertaities, the highest spot price show o the graph would be $13,500/MWh.
4 the reewable percetage icreases, keepig i mid that the total USE for each geeratio portfolio is the same). Fig. 3 shows the expected aual geerator reveue ad operatig profit of each techology i the least cost portfolio for each reewable peetratio. The aual average spot prices are also show i Fig. 3(a). The impact of the carbo price o the reveue, operatig costs ad hece profits of the fossil fuel plats are apparet. Although the reveues of PV ad wid plats are relatively low, their operatig profits of PV ad wid plats are sigificatly higher tha those of coal ad CCGT, particularly at low to moderate reewable peetratios (i.e. from 15% to 60% reewable peetratio). This is due the low operatig cost of reewable geeratio ad the impact of carbo price o the high operatig costs of fossil fuel plats. The operatig profit of each techology geerally reduces as the amout of reewables icreases due to lower aual average spot prices iflueced by the low SRMCs of wid ad PV. However, fossil fuel geerators are able to make operatig profit eve at high reewable peetratio. This is likely iflueced by the 15% miimum sychroous geeratio requiremet applied i the modellig, which eforced thermal geeratig plats (most likely coal) to supply at least 15% of demad i every period. Hece they are able to ear reveue to most periods. This is particularly crucial durig scarcity or ear scarcity periods whe the spot prices are extremely high. (a) (b) Fig. 3. (a) Expected aual reveue of each techology ad aual average spot prices (b) Expected aual operatig profit of each techology for each reewable peetratio. The capacity (MW) of each techology is also show. O the other had, the profits of PV ad wid reduce far more sigificatly tha for thermal geeratio techologies. This is particularly the case for PV as show by its egligible operatig profit at a 85% reewable peetratio eve without takig ito accout aual capital repaymets. Sice PV, ad to a lesser extet wid, do ot ofte geerate durig high price periods (as show i Fig. 2), they were uable to beefit from the high spot prices. This result may due to the very large proportio of PV icluded i the 75% ad 85% reewable portfolios, which may be higher that ecoomically optimal, resultig i high costs ad almost egligible profits. These issues warrat further ivestigatio The modellig results suggest that, at high reewable peetratio levels ad give the curret market arragemets, PV ad wid plats might ot ear sufficiet reveue to cover their costs. I cotrast, coal ad CCGT plat appear to maitai operatig profitability followig a iitial declie. OCGT plat appear to maitai operatig profitability regardless of the reewable peetratio level, suggestig that peakig plat may be relatively immue to the reducig average wholesale price, ad able to flexibly adjust as required to access high priced periods. Oe of the optios for icreasig geerator reveue is to icrease the MPC from the curret $13,500/MWh, sice a higher MPC will lead to more reveue eared durig high demad periods ad hece higher profits for geerators [14]. This will be examied i future work. V. CONCLUSIONS This paper assesses the impact of variable reewable geeratio o spot market prices ad geerator reveues i a -oly electricity market. The Australia Natioal Electricity Market (NEM) with differet reewable peetratios i 2030 uder ucertai gas prices, carbo pricig policy ad electricity demad was used as a case study. Modellig results idicate that the aual average spot price geerally reduces as the amout of reewable geeratio icreases due to the low operatig costs of wid ad PV geeratio. Although there were fewer periods of demad ad supply imbalace as the reewable peetratio icreases, the magitude of the imbalace ad hece average price spikes were greater. Geerally, the reductio i the average spot price results i reduced reveues ad profitability of geerators ad potetially leads to isufficiet reveue to meet costs, particularly for large scale wid ad PV geerators. The reveue impacts o PV ad wid geeratio are very severe at the high reewable peetratios cosidered. Therefore, chages i market mechaisms such as icreasig market price cap may be required to esure reveue sufficiecy ad log-term resource adequacy i a oly market with high reewables. Further work is warrated to explore these issues. There are some limitatios i this study. The fidigs are highly depedet o modellig ad iput assumptios. For the reewable peetratio greater tha 60%, the proportio of PV ad wid geeratio chose i the modellig may ot be the most ecoomically optimal, resultig i higher idustry costs
5 ad almost egligible operatig profits for PV. Furthermore, there may be mechaisms other tha imposig a miimum sychroous geeratio costrait, which is a costly optio, to provide system iertia ad frequecy respose at times of high o-sychroous reewable peetratios. These limitatios represet areas for future work. VI. REFERENCES [1] E. Ela, M. Milliga, A. Bloom, A. Botterud, T. A., ad T. Levi, "Evolutio of Wholesale Electricity Market Desig with Icreasig Levels of Reewable Geeratio," Natioal Reewable Eergy Laboratory, 2014. [2] IEA, "The Power of Trasformatio: Wid, Su ad the Ecoomics of Flexible Power Systems," Iteratioal Eergy Agecy, 2014. [3] AEMO, "100 per cet Reewables Study - Modellig Outcomes," Australia Eergy Market Operator, 2013. [4] B. Ellisto, I. MacGill, ad M. Diesedorf, "Least cost 100% reewable electricity scearios i the Australia Natioal Electricity Market," Eergy Policy, vol. 59, 0, pp. 270-282, 8// 2013. [5] J. Riesz ad M. Milliga, "Desigig electricity markets for a high peetratio of variable reewables," WIREs Eergy Eviromet doi: 10.1002/wee.137, 2014. [6] P. Vithayasrichareo ad I. F. MacGill, "A Mote Carlo based decisiosupport tool for assessig geeratio portfolios i future carbo costraied electricity idustries," Eergy Policy, vol. 41, pp. 374-392, 2012. [7] AEMO, "100 Per cet Reewables Study - Modellig Assumptios ad Iput," Australia Eergy Market Operator, 2012. [8] P. Vithayasrichareo, J. Riesz, ad I. F. MacGill, "Usig reewables to hedge agaist future electricity idustry ucertaities A Australia case study," Eergy Policy, vol. 76, pp. 43-56, 2015. [9] AER, "State of the Eergy Market 2014," Australia Eergy Regulator, 2014. [10] S. Ji, S. Rya, J.-P. Watso, ad D. Woodruff, "Modelig ad solvig a large-scale geeratio expasio plaig problem uder ucertaity," Eergy Systems, vol. 2, 3-4, pp. 209-242, 2011/11/01 2011. [11] F. A. Roques, D. M. Newbery, ad W. J. Nuttall, "Fuel mix diversificatio icetives i liberalized electricity markets: A Mea- Variace Portfolio theory approach," Eergy Ecoomics, vol. 30, 4, pp. 1831-1849, 2008. [12] BREE, "Australia Eergy Techology Assessmet 2012," Bureau of Resources ad Eergy Ecoomics, Australia Govermet, 2012. [13] Australia Treasury, "Strog growth, low pollutio: Modellig a carbo price," Australia Govermet, Caberra, 2011. [14] J. Riesz ad I. MacGill, "100% Reewables i Australia: Will a Capacity Market be Required?," i 3rd Solar Itegratio Workshop Lodo, UK, 2013.