Sensitivity of portuguese electricity market prices to solar PV penetration : an analysis of 2016 prices

Sensitivity of Portuguese Electricity Market Prices to Solar PV Penetration An analysis of 2016 prices Master Thesis for M.Sc. in Energy Engineering, part of EIT Innoenergy M.Sc. in Energy for Smart Cities. Author: Jo˜ao Cordeiro de Sousa Coordinator: Jordi de la Hoz Casas Supported by: Galp Energia Date: September 12 2017 Abstract The reduction in price of solar PV technology led, in the recent years, multiple investors to apply for installing new solar PV power plants in Portugal which would operate without subsidies or feed-in-tariffs. In 2016 it was reported the approval of construction of such power plants and given the low variable cost of this technology it is expected that their penetration would reduce the electricity market prices. Hence, before doing the economic assessment of potential new solar PV power plants included in this regime it is important to quantify the impact that these units themselves would have on the market prices, and consequently on their revenues. This thesis aims to calculate this potential maximum impact of a given scenario of 480 MWp with every solar PV producer in it willing to produce at 0 e/MWh. Acknowledgements After concluding this thesis I have completely acknowledged the benefits that it will bring both to my professional and personal life, as it was a fruitful lesson of self discipline, overcoming adversities and self realization. Thus, I want to thank to Jordi de la Hoz Casas, my supervisor from UPC, for all the support and technical advice through out the whole process, to Luis Alexandre Silva, from Galp Energia, for supervising the project and let me integrate his team for the development of this work and to Jo˜ao Marques, from Galp Energia, for bringing this project together. In addition, I want to thank my family for the unconditional support they gave through out my whole education in every decision I made and, of course, for giving me the possibility of having this enriching and amazing experience of studying abroad. Lastly, I want to thank my girlfriend and my closest friends, the ones who grew with me in Lisbon and the ones with who I shared this international experience, not only for being present but for making these last years amazing, full of life changing experiences and new realities. 1 Contents 1 Introduction 7 2 The Portuguese Electricity Market 9 2.1 Evolution ..................................... 9 2.1.1 Building the Market . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 2.1.2 TheEnergyMix ............................. 13 2.1.3 Evolution of Electricity Prices . . . . . . . . . . . . . . . . . . . . . . 22 2.2 CurrentSituation................................. 27 2.2.1 Operations in the Market . . . . . . . . . . . . . . . . . . . . . . . . 27 2.2.2 TheEnergyMix ............................. 31 2.3 TheElectricityPrice............................... 32 3 GCPVS in Portugal: Regulatory Framework 35 3.1 Before1988:Monopoly ............................. 35 3.2 1988-1994: Opening the electricity sector . . . . . . . . . . . . . . . . . . . 35 3.3 1995-1998: The new SEN . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36 3.4 1999-2005: Empowering renewable energy . . . . . . . . . . . . . . . . . . . 37 3.5 2006-2011: Reorganization of the SEN. Liberalization of the electricity sector 42 3.6 2012-Present: Promoting competition . . . . . . . . . . . . . . . . . . . . . 44 4 Methodology and Calculations 46 4.1 TheScenario ................................... 46 4.2 Calculations.................................... 48 4.3 Simulations .................................... 52 5 Results Analysis 54 5.1 ByMonth..................................... 54 5.1.1 January.................................. 54 5.1.2 February ................................. 55 5.1.3 March................................... 56 5.1.4 April.................................... 57 5.1.5 May .................................... 59 5.1.6 June.................................... 59 5.1.7 July .................................... 61 5.1.8 August .................................. 62 5.1.9 September................................. 64 5.1.10 October.................................. 64 5.1.11 November................................. 66 5.1.12 December................................. 67 5.2 Averages...................................... 68 5.3 EconomicImpact................................. 71 2 6 Conclusion 72 A Matlab Code 78 3 List of Figures 2.1 Quantitative evolution of the electricity generation market between the 30s and90s ...................................... 10 2.2 Evolution of transmission grid. . . . . . . . . . . . . . . . . . . . . . . . . . 12 2.3 Evolution of International Lines . . . . . . . . . . . . . . . . . . . . . . . . . 13 2.4 International transmission capacity . . . . . . . . . . . . . . . . . . . . . . . 14 2.5 Average transmission capacity and market splitting shares from 2008 to 2016.[57,58].................................... 14 2.6 Evolution of thermal and hydro power plants from 1930 to 1975. . . . . . . 15 2.7 ThermalPowerplants .............................. 16 2.8 HydroPower ................................... 17 2.9 WindPower.................................... 18 2.10WindPower.................................... 19 2.11Biomass...................................... 20 2.12SupplyDistribution ............................... 22 2.13 Domestic Electricity Tariff System for the city of Lisbon 1929-1975. . . . . 23 2.14 Domestic Electricity Tariffs for the city of Lisbon from 1950 to 1975 . . . . 24 2.15 Electricity price decomposition . . . . . . . . . . . . . . . . . . . . . . . . . 24 2.16TariffsEvolution ................................. 25 2.17CIEGEvolution ................................. 26 2.18StructureofOMIP................................ 28 2.19 Technical Restriction Impact . . . . . . . . . . . . . . . . . . . . . . . . . . 29 2.20 Intraday market sessions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 2.21 Operations of daily and intraday markets . . . . . . . . . . . . . . . . . . . 31 2.22 Cumulative capacity in December 2016 and July 2017. . . . . . . . . . . . . 32 2.23 Annual generation mix on July of 2015, 2016 and 2017. . . . . . . . . . . . 32 2.24Referencetariffs.................................. 33 3.1 Organization of SEN according to Law Decree 182/95 . . . . . . . . . . . . 37 3.2 Remuneration System inputs and outputs of Law-Decree 168/99. . . . . . . 40 3.3 Remuneration System inputs and outputs after Law-Decree 337-C/2001. . . 41 3.4 Organization of SEN according to Law Decree 29/2006 . . . . . . . . . . . . 43 4.1 GCPVS that were accounted for the analysis and their equivalent single unit 47 4.2 Interface of PVGIS platform .......................... 47 4.3 Example of PVGIS estimation of daily radiation. . . . . . . . . . . . . . . . 48 4.4 Example of PVGIS estimation of yearly production. . . . . . . . . . . . . . 48 4.5 Example of data analysis exercise for January. . . . . . . . . . . . . . . . . . 49 4.6 Omieresults.................................... 50 4.7 Visual representation of the pretended analysis to be applied to each hour. 50 4.8 Demonstration of supply and demand curves construction . . . . . . . . . . 51 4.9 Matlab code used to calculate intersections . . . . . . . . . . . . . . . . . . 52 4.10 Matlab plotting the demand and supply curves . . . . . . . . . . . . . . . . 52 4 4.11 Results obtained for the 12th January, 2016. . . . . . . . . . . . . . . . . . . 52 4.12 Daily Production Distribution of Solar PV on given location . . . . . . . . . 53 5.1 Simulation results for January 12 2016. . . . . . . . . . . . . . . . . . . . . 55 5.2 Simulation results for February 15 2016. . . . . . . . . . . . . . . . . . . . . 56 5.3 Generation mix on January 12 2016 . . . . . . . . . . . . . . . . . . . . . . 56 5.4 Generation mix on February 15 2016 . . . . . . . . . . . . . . . . . . . . . . 56 5.5 Simulation results for March 15 2016. . . . . . . . . . . . . . . . . . . . . . 57 5.6 Simulation results for April 11 2016. . . . . . . . . . . . . . . . . . . . . . . 58 5.7 Generation mix on March 15 2016. . . . . . . . . . . . . . . . . . . . . . . . 58 5.8 Generation mix on April 11 2016. . . . . . . . . . . . . . . . . . . . . . . . . 58 5.9 Simulation results for May 16 2016. . . . . . . . . . . . . . . . . . . . . . . . 59 5.10 Simulation results for June 23 2016. . . . . . . . . . . . . . . . . . . . . . . 60 5.11 Generation mix on May 16 2016. . . . . . . . . . . . . . . . . . . . . . . . . 61 5.12 Generation mix on June 23 2016. . . . . . . . . . . . . . . . . . . . . . . . . 61 5.13 Simulation results for July 14 2016. . . . . . . . . . . . . . . . . . . . . . . . 62 5.14 Simulation results for August 15 2016. . . . . . . . . . . . . . . . . . . . . . 63 5.15 Generation mix on July 16 2016. . . . . . . . . . . . . . . . . . . . . . . . . 63 5.16 Generation mix on August 15 2016. . . . . . . . . . . . . . . . . . . . . . . . 63 5.17 Simulation results for September 15 2016. . . . . . . . . . . . . . . . . . . . 64 5.18 Simulation results for October 14 2016. . . . . . . . . . . . . . . . . . . . . 65 5.19 Generation mix on September 15 2016. . . . . . . . . . . . . . . . . . . . . . 66 5.20 Generation mix on October 14 2016. . . . . . . . . . . . . . . . . . . . . . . 66 5.21 Simulation results for November 15 2016. . . . . . . . . . . . . . . . . . . . 67 5.22 Simulation results for December 15 2016. . . . . . . . . . . . . . . . . . . . 68 5.23 Generation mix on November 15 2016. . . . . . . . . . . . . . . . . . . . . . 68 5.24 Generation mix on December 14 2016. . . . . . . . . . . . . . . . . . . . . . 68 5.25Averagesbymonth................................ 69 5.26 Average daily impact ordered by increasing average price . . . . . . . . . . 69 5.27Averagesbyhour................................. 70 5.28 Average hourly impact ordered by increasing average price . . . . . . . . . . 71 5 List of Tables 2.1 Average price on the Portuguese and Spanish daily markets in the last 12 months ...................................... 34 3.1 Remuneration according to Law Decree 168/99 . . . . . . . . . . . . . . . . 40 3.2 Remuneration according to Law Decree 339-C/2001 and Law Decree 33- A/2005 ...................................... 42 4.1 List of GCPVS that were accounted for the analysis, respective location andrateproduction[4].............................. 46 5.1 Estimated production of given scenario on a January reference day. . . . . . 54 5.2 Estimated production of given scenario on a February reference day. . . . . 55 5.3 Estimated production of given scenario on a March reference day. . . . . . . 57 5.4 Estimated production of given scenario on an April reference day. . . . . . . 58 5.5 Estimated production of given scenario on a May reference day. . . . . . . . 59 5.6 Estimated production of given scenario on a June reference day. . . . . . . . 60 5.7 Estimated production of given scenario on a July reference day. . . . . . . . 61 5.8 Estimated production of given scenario on an August reference day. . . . . 62 5.9 Estimated production of given scenario on a September reference day. . . . 64 5.10 Estimated production of given scenario on an October reference day. . . . . 65 5.11 Estimated production of given scenario on a November reference day. . . . . 66 5.12 Estimated production of given scenario on a December reference day. . . . . 67 5.13 Estimated annual revenue of the given scenario with sample prices before technical restrictions and reduced prices due to solar penetration . . . . . . 71 6.1 Significant figures of calculated results. . . . . . . . . . . . . . . . . . . . . . 74 6 Chapter 1 Introduction Unlike to what happens with other forms of energy, the impossibility of storing electricity in large scale developed a very complex and dynamic electricity system. Being incapable of producing now the consumption of the future requires the system to maintain a constant equilibrium between current supply and demand. With the Portuguese and Spanish electricity markets merged with a competitive regime means that at a each moment in the Iberian peninsula the market price and transacted energy result from the demand and supply equilibrium. More specifically, the daily market timeline is divided in periods of one hour and for each of these, on the day before and based on predictions of what will be the consumption and production conditions, producers and suppliers bid a certain volume at a specific price which result in a demand and supply curves. Thus, these curves intersections dictate a price and transacted energy for each hour of every day energy in this pool, where producers and suppliers negotiate the electricity before this is sold to consumers. Therefore, the electricity market more than any other consumable good market experiences constant price variations, both in the long and the short run due to three main factor. Firstly, as the environmental conditions and technical restrictions vary at every hour the available energy mix on the supply side varies with it. Since each technology has its own corresponding cost, the supply curve can assume completely different shapes in two consecutive hours. Secondly, this same principle can also be applied to a much larger timescale. With the Portuguese energy mix integrating technologies such as thermal power plants and dams for hydro power, which production cost depend on factors such as global oil price and annual rainfall, respectively, the shape of the supply curve will also experience variations correlated to these parameters which occur much slower than the first, typically associated to an whole year or month. Thirdly, the volume of demand will also affect the price fixed at each hour, increasing the prices at periods of high demand such as early evening on a daily basis and months of extreme temperatures on an annual basis, and decreasing them on periods of low demand. In conclusion, it is clear that being the electricity market a competitive playground its prices will constantly experience variations but will always get closer to the marginal costs of each production technology as competition increases. Furthermore, on top of the variable cost of electricity negotiated in the pool market, there are additional tariffs charged to consumers that represent other costs of the system such as transmission, distribution and system’s general costs. While the first two are quite stable and represent cost of technology maintenance and infrastructure investments, which hardly could be avoided, the last includes, among other, the cost associated to policies implemented in the electricity system, which have proved that can be quite damaging for consumers and possibly could be avoided. A typical example of this is the cost that renewable generation subsidies implied to the Portuguese system. To promote renewable 7 production the government created a remuneration regime for producers of this type fixing a price for all the energy these generated clearly above the pool market prices. These subsidized tariffs were called feed-in-tariffs and the difference between them and market prices has been accounted as a system’s general cost and charged to consumers in the final price. Thus, it is clear that the electricity price is sensitive to both technical, economic and legislative measures in the long and short run. At the same time, with the consumption of electricity increasing every year its price volatility has also been gaining an ever increasing attention and submitted to deeper analysis and studies. This thesis was proposed after being reported that the Portuguese electricity market regulator has approved, in 2016, the construction of several solar PV power plants in Portugal which will not benefit from any subsidized tariff and therefore will compete in the liberalized market with every other technology. Given the fact that the penetration of these producers will most likely decrease the market prices, it is important that they calculate their impact in the market before estimating potential revenues considering current market prices. In fact, there will be a saturation point for the market where the entrance of an additional MW of solar PV, or of another technology with equally low variable cost, will push the market prices below the limit that sets the edge of a profitable investment. Hence, and considering the sensitivity of electricity prices, it was established as final objective of this thesis to attempt to quantify the maximum impact that such solar PV penetration would have on the Portuguese electricity market prices. Consequently, it will be assumed the penetration of 480 MWp of cumulative capacity, which is the scenario of approved production units in Portugal described on the most recent reports. Nevertheless, the model used for such calculation could be applied to any other solar PV capacity dimension. The study will firstly go through an analysis of the evolution of the market and its current situation, in chapter 2. Then, in chapter 3, it will be done a review of the legal framework regarding solar PV technology in order to to get an idea of the political strategies applied so far and correlate them with the evolution of the market. Finally, the model used to quantify the impact established as final objective will be described in chapter 4 and its results presented in chapter 5. 8 Figure 2.6: Evolution of thermal and hydro power plants from 1930 to 1975. [6]. and during the 90s after the construction of the Sines thermal power plant of 1200 MW, still the biggest generation facility in Portugal. Fuel-oil power plants were introduced in the 70s supplying the majority of the consumption during the 70s and 80s, however since the beginning of the XXI century that the technology had been losing its production share and in 2012 the last fuel oil power plant installed in Portugal was shutdown. The third type, natural gas power units were only introduced to the sector in 1997 as a cogeneration technology, meaning that produces heat and electricity in the same combustion cycle, but rapidly became the main thermal power plant technology in the energy mix by surpassing coal production in the mid 2000s. Hydro Power Hydro power is together with thermal the oldest technology used to generate electricity in Portugal. Although it has the unpredictability issue related to its dependence on rain, this technology has two main advantages over the fossil fuel alternative, firstly it is more environmental friendly and the its source does not have to be imported. The first hydro power plants were concluded in 1894 and 1985 in the north of Portugal and by the end the of the century the public lighting of Braga was already being fed by one of those plants with three 125 hp5turbines. Until 1930 several hydro power plants almost all between 100 kW and 2 MW, excluding the Lindoso power plant that in 1922 was activated producing electricity with 7,5 MW turbines, were constructed to supply local consumption in remote villages or industries. Between the 1930 and 1950 the hydro power industry was represented by a dissemination of small power plants, with nearly one hundred of hydro power units and five hundred thermal units operating but producing small amounts of energy and at excessive prices. In fact, in 1940 all these small units combined generated less than the three biggest existing power plants in Portugal, which were the Lindoso power plant that had been upgraded in 15 MW since the original project and the two thermal units supplying Lisbon. In the 51 Horsepower = 0.7457 kW 15 (a) (b) Figure 2.7: (a)Thermal gross production since 1930. Adapted from [6–10, 56].(b)Thermal installed capacity. [47] 50s, resulting from the Law 2002 of 1944 promoting hydro generation, the so called golden era for hydro power began with several heavy hydro units construction being completed. During this decade almost 1000 MW of hydro power were installed, with the 139 MW hydro power unit of Castelo de Bode replacing the thermal facilities and becoming the main electricity supplier of Lisbon and with another two 180 MW and 174 MW hydro power units constructed in the north being the most notorious. Considering that between 1950 and 1960 the total hydro power production increased from 941,8 GWh to 3263,5 GWh and its contribution increased from 46% to 95% it is clear the impact that this industry gained in ten years. However, due to the increase of consumption and impossibility to keep growing the hydro power sector infinitely, on the following decade two heavy thermal units were constructed slowing down the industry of hydro power and its impact on total production. The decades of the 70s and 80s were a continuation of the 60s, with the hydro power sector growing, yet much slower than the consumption and thermal production. From the 90s until today the increase of hydro power installed relied in the construction of less but bigger power plants, as it is possible to see in figure 2.8b the clear steps of 1992, when a 630 MW hydro power facility was installed in Alto Lindoso, 2003, when another of 240 MW was constructed in Alqueva, and 2011 when other two combining 436 MW started its operations in the Douro river [6, 45, 47, 56]. 16 (a) (b) Figure 2.8: (a) Total hydro gross production since 1930. Data extracted from: [6, 45, 56].(b) Hydro Power Installed [47] Wind Power Just as it happened with hydro power in the 50s, wind power was introduced to the Portuguese energy sector due to two obvious reasons: its beneficial environmental impact and to decrease the energy dependency on imported sources. Furthermore, wind power could also improve the supply reliability and help on the control of technical parameters of the grid such as frequency control[3]. In that regard, in the 90s the electric system was reorganized with the legislation promoting competition and the entrance of renewables in the generation mix, as it will be explained better in chapter 3, and wind power plants projects started to appear. In 1996 the first wind park was installed in Portugal’s main land6and since then that has been the most present renewable technology in the Portuguese energy sector[61]. By the end of 2001 the installed capacity in the country was already 134 MW and three years later, in 2004, it had quadrupled. In 2005 an important legislative measure influenced the growth of renewables, the Law Decree 33-A/2005 increased the feed-in-tariffs, and attracted many investors. Consequently, and only because the development of the wind power technology allowed it, from 2005 to 2010 Portugal witnessed an exponential growth of wind power penetration in its electric system. As [20] describes for the similar Spanish situation, this growth was the result of a combination of legislative, technical and economic conditions. At the same time the government was 6In 1992 the first wind park was installed Madeira island. 17 promoting renewable generation, these technologies were becoming more efficient and a global economic crisis was erupting. In the late 2000s private investors saw an opportunity to invest in this already reliable wind power technology for which they would be paid a fix amount per unit of energy generated independently from the market conditions making it a very safe investment in a period of economic instability. Consequently, by the end 2010 the the installed capacity in Portugal’s was already 4500 MW, nearly 6,5 times the 700 MW installed in the beginning of 2005, equivalent to an annual growth rate of 36%. Since 2010 until today the expansion has been much slower, with an average growth of 3%, mainly due to the overcost that the feed-in-tariffs implied to the national electric system and which resulted in a promotion retreat in terms of subsidized tariffs and conceded operation licenses. (a) (b) Figure 2.9: (a) Total wind power generation. Data extracted from [45].(b) Cumulative wind power installed capacity. Data extracted from [45]. Solar PV Similarly to wind power, Solar PV technology entered the energy mix quite recently when the combination of the three parameters legislative, technical and economical allowed it. The Law Decree that in 2005 increased the remuneration of wind power did the same for solar PV (see table 3.2), in fact photovoltaic was the technology that most benefited from this incentive setting its remuneration above all the others. Such privilege was mainly 18 because it was one of the most expensive renewable technologies and because there was great potential for the sector to be well developed in Portugal being this one of the most sunniest countries of Europe [60]. Hence, the first solar PV power plants were installed in 2006 and in 2008 the biggest solar PV farm in Europe, with 46 MW, was constructed in Portugal. The expansion since then has been consistent, with an average annual growth of 66% [19], yet not even getting closer to wind power in terms of cumulative capacity. In solar PV it is important to differentiate the type of production between micro, mini production and normal power plants production. Micro and mini production are special regimes designed to allow consumers to generate power up to 5 kW and 250 kW of power, respectively, at their properties and then sell it to the grid. These regimes were presented through Law Decrees 363/2007 and 34/2011 and since their integration that the total installed solar PV capacity and production has been heavily influenced by them, being today responsible for more than 30% of the total production as figure 2.10a represents. (a) (b) Figure 2.10: (a) Total solar power generation distributed between normal production, miniproduction and microproduction. Data extracted from [19].(b) Cumulative solar PV power installed capacity. Data extracted from [19]. Biomass Biomass technology transforms natural residues, like forestry residues, urban solid residues or industrial organic waste into heat and electricity. In Portugal this technology has been 19 implemented either as conventional thermal power plants or as cogeneration power plants, producing heat and electricity with the same combustion (as natural gas combined cycle power plants mentioned before). The introduction of biomass into the electricity sector was a bit different than the other renewable technologies mentioned before, arising not only from political interest on decreasing the CO2 emissions of the country, on diversifying the energy mix or creating new industries to enrich the economy, but also due to the paper and pulp industry in which the country is very strong. This industry transforms wood and forestry residues into paper or pulp leaving some residues at the end of the process, which can then be used as fuel on biomass power plants. Obviously, these companies saw biomass power generation as a potential secondary business since they already had the primary source required for it. In addition, since biomass power plants also consume forestry waste and are typically located on rural areas the implementation of these would bring two beneficial side effects for the country, that were the creation of jobs in these rural under populated villages and helped cleaning the forests decreasing the risk of natural fires [23]. When grouping the different technologies, biogas and urban solid residues (USR) power plants are typically considered in the same category than biomass technology since they also consume organic matter or waste to generate electricity. (a) (b) Figure 2.11: (a) Total biomass power generation including cogeneration, no cogeneration, biogas and urban solid residues. Data extracted from [18, 19].(b) Cumulative biomass installed capacity. Data extracted from [18, 19]. 20 In the 1990s, the first biomass power plants were installed in Portugal and they used cogeneration technology. Plants with no cogeneration, just as biogas and USR, only appeared in 1999 in smaller projects and by the end of that year there were installed 441 MW in total of biomass, biogas and USR power plants. In the late 2000s, the feed-in-tariffs established for this renewable technology attracted investment into the activity which forced the expansion represented in figure 2.11b. The mix In conclusion, until the XXI century the available sources of electricity in Portugal were either hydro or fossil fuels with both of them dominating the generation on different periods, hydro in the decades of 50s and 60s and fossil thermal power plants on the 70s, 80s and 90s. Since the beginning of the XXI century that renewable production started to have its impact on the mix being wind the most present technology reaching 24% of the national consumption in 2013. However, the annual energy mix is highly dependent on two factors: meteorological conditions and fossil fuels price. In fact 64% of the installed capacity rely on rain, sun or wind to produce electricity and 33% on imported fuels. Thus, the technologies that suffer higher annual variations are hydro power and fossil fuels as it can be seen for example, between 2015 and 2016, where the hydro installed capacity was increased by 13% while its production increased 82%. Furthermore, besides the imports of fuels consumed in thermal power plants, the Portuguese electricity sector has typically been dependent on the imports of electricity from Spain to supply the national consumption. The integration of renewables decreased the relative dependency, not by decreasing substantially the imports, but by allowing the consumption and total national production to increase without increasing the first. Hence, due to the combination of a varied renewable technologies mix and rainy half year, in 2016, for the first time since 1999 the electricity sector had a positive balance of electricity exports. 21 (a) (b) Figure 2.12: (a) Supply distribution from 2001 to 2006. (b) Supply distribution from 2007 to 2016. Adapted from: [54]. 2.1.3 Evolution of Electricity Prices Since the integration of electricity in the activities of the Portuguese society and because the electricity sector was considered a monopolist market until the recent pro liberalization changes mentioned above, an important tool has been used to regulate the industry and protect consumers which was the concession contracts. These were contracts through which the regulating agency could control the tariffs practiced by the private companies. Such supervision derived from the special treatment that electricity gained in the moment it started having an impact in the life of the Portuguese people. Thus, the price of kWh paid by the consumers was used during the XX century as a tool to lead the economy depending on whatever the strategy was, either to promote consumption and investment expanding the electricity sector, to incite energy efficiency and decrease dependency on a foreign supplier, or like it happened in the mid XX century to encourage families to invest in household appliances and boost another industry [2]. In order to analyze the evolution of the electricity price data referent to the domestic tariffs applied to the city of Lisbon will be used for period of 1930-1975 as the total national electricity consumption by then was concentrated on the capital and the policies 22 implemented in Lisbon represented the political strategy applied for the rest of the country. In 1928 a concession contract was signed between the producer CRGE and the city council of Lisbon where it was established a single tariff of 1,896 PTE7for the electricity consumed in Lisbon. This single tariff meant that the whichever the quantity consumed there would be not differentiated price per unit, and despite the fact that it had an updating parameter on its calculation formula it ended up by remaining constant until 1945. Yet, between 1936 and 1942 a new system was introduced to the single tariff system, which were the decreasing tariffs. These implemented the concept of economies of scale in the domestic electricity consumption by establishing 3 levels of consumption which would vary with the seasons of the year and the total rooms on the consumer house. Basically, the first level was designed for domestic lighting while the second and third level would only be achieved with electric appliances. However, as between these dates most of the domestic consumers did not own any electrical appliances the vast majority never reached the second and third levels. Being the first level price identical the previous single tariff system this measure was practically symbolic [2]. Figure 2.13: Domestic Electricity Tariff System for the city of Lisbon 1929-1975. Adapted from [2] In 1942 the second world war had brought changes in all aspects and one of the them was scarcity and increase of coal prices, which was the main source of electricity at the time. Consequently, the government while trying to avoid such expensive imports decided to increase consumption efficiency by abolishing the decreasing tariffs system and bringing back the single tariff combined with a fine system of 10 PTE per kWh consumed above a certain limit. In 1945 the single tariff was increased to 2,50 PTE, which represented an increase of 32% on a single year compensating the 16 years of tariffs stagnation but in 1947, two years after the end of the war, a new period characterized by promoting electricity consumption began. Thus, in 1947 the fine system was abolished, in 1948 the decreasing tariffs were implemented again and in 1951 all the tariffs were reduced a new tariff regime was designed specifically for economically fragile consumers which had an annual consumption below 100 kWh. As result between 1938 and 1953 the average electricity price paid by consumers was reduced by 29% in real prices [2], considering inflation, which was compensated with an proportional increase of consumption. After that, during more than 20 years the electricity tariffs were kept constant, which from the real prices perspective means that were constantly reduced as figure 2.14 represents, defining clear period of consumption promotion between 1951 and 1975. Still, it is important to remember that in this post war period the industrial and agricultural activities were considered crucial activities for the national economic development and had specific tariffs at even lower prices than domestic tariffs. The end of this period was set in 1976 with an increase of electricity prices and with the appearance, in 1977, of a new tariff system where for the first time ever contracts differentiated the energy consumed and the power at which was consumed. 7The Portuguese currency before euro. 1P T E = 0,005Eur 23 Figure 2.14: Domestic Electricity Tariffs for the city of Lisbon from 1950 to 1975 - at real prices of 1950. Adapted from [2] Figure 2.15: Electricity price decomposition since the beginning of the liberalization of the electricity sector in the decade of 1990 In the 90s five different consumption power levels were already established. The very high voltage or MAT, the high voltage or AT, the medium voltage or MT, the normal low voltage or BTN and the special low voltage or BTE, which was differentiated from the previous because it was mainly designed for local businesses instead of residential consumers. The logic behind different voltage levels is based on the principle of the higher the voltage the lower the price since consumers uses less distribution grids8. However, when looking at tariffs it is important to be aware of how these are built. Figure 2.15 represents how the final electricity price is obtained in the Portuguese electricity sector since the liberalization of the sector. In addition to the cost of the producer, TSO, DSO and supplier, consumers still have to pay the System General Usage tariff, UGS. This includes costs that are required for the system to perform well but that are not exclusively related to any of the four activities of the supply chain. Thus, the UGS tariff is composed by three different categories of costs, the Systems Management Costs, General Economic Interest Cost or CIEG, and the Power Warranty Cost. However, the CIEG is the only requiring special attention since it has been having a substantial impact on the final price paid by the consumers. In figure 2.16a it is represented the evolution in real prices, referent to 2016, of the different regulated tariffs applied between 1990 and 2010 by the last resort supplier, the public supplier which will be described in chapter 3. After 2011 the tariffs MAT, AT, MT and BTE stopped being supplied by this last resort supplier, consequently between 2011 and 2016 figure 2.16a represents the average price practiced by competitive suppliers for these voltage levels. 8There are high, medium and low voltage distribution grids that operate sequentially 24 Figure 2.21: Operations of the daily and intraday markets [40] activities are the only that are still regulated on the electricity market since the retail is also done in a competitive market, except for situations like consumers from remote locations or requiring social support who use the last resort supplier. As in terms of stakeholders figures operating in the production, transmission, distribution and retail of the Portuguese electricity market, not including the special regime, there are [11]: •4 producers on ordinary regime; •1 transmission system operator; •13 distribution system operators - of which 12 are regional distributors controlled by municipalities; •13 regulated last resort suppliers - of which 12 are regional suppliers controlled by municipalities; •18 domestic suppliers; •21 industrial suppliers; 2.2.2 The Energy Mix Until the current moment of the year 2017, the energy mix installed in Portugal is very similar to the one by the end of 2016 as there were not any heavy generation units installed since then. Hence, the current installed production capacity relies on coal, natural gas and biomass thermal power plants, hydro power plants, wind farms and photovoltaic technology, having increased by 1,31% in total since 2016, mainly due to an increase of hydro power capacity. In figure 2.22 it is possible to compare the installed capacity between December of 2016 and July 2017. Unlike the installed capacity, which typically is quite constant since its variations depend on the construction or dismantling if physical infrastructure, the energy mix used to supply the consumption is much more volatile. Parameters like meteorological conditions and fossil fuels global price affect heavily the generation mix used in Portugal since 64% of the installed capacity rely on rain, sun or wind to produce electricity and 33% on imported fossil fuels. Therefore, what technologies are available to be dispatched and at which cost are two parameters strongly dictated by external factors and that each day experience clear variations. The current situation of Portugal represents this concept very clearly as the first 7 months of 2017, until July, were characterized by being very dry reducing the production 31 Figure 2.22: Cumulative capacity in December 2016 and July 2017. Data from [54, 55] Figure 2.23: Annual generation mix on July of 2015, 2016 and 2017. Data from [54, 55] of hydro power substantially when compared to the first months of 2016. In fact, the amount of rain that fell between December 2016 and March 2017 was 69% lower than the average [21] resulting in a hydro power production on the first half of 2017 60% shorter than the one on the first half of 2016. Due to such cut on the hydro production and given the impossibility to increase the renewable production, fossil fueled thermal power plants were used more frequently during these first 7 months increasing its production by 52% compared to the same period of 2016. Hence, during the first half of 2017 the total production combined reached 32487 GWh, of which 16,4% derived from hydro power plants, 30,2% from conventional natural gas, 16,9% from coal power plants, 22,3% from wind power, 5,1% from biomass, 1,5% from solar PV, and 8,5% from natural gas combined cycle power plants included in the special regime. Furthermore, just like it happened in the previous year, until July 2017 the Portuguese electricity market had a positive exports balance delivering 7% of its production in Spain. As mentioned before, the dispatched set depends heavily on the technologies availability and cost and while some benefit from a lower variable cost others benefit from a higher control. Thus, each day the production set is organized by a merit order which sorts by ascending variable cost the available technologies. The ones with low variable costs such as solar and wind power, come in first place and technologies with high variable costs, such as thermal power plants come in last. Hydro power plants, exceptionally, define their variable cost according to the market and environmental conditions as they use a free of cost source but are able to store it, which does not happen with any other technology. Hence, these production units have its variable cost equal to the opportunity costs of using the available water to generate electricity at specific moment instead of saving it for another. 2.3 The Electricity Price In this section, it will done an analysis to the electricity price that consumers are currently paying. For that, it is important to remember that there are two markets where consumers are currently obtaining the electricity from, the regulated market through the last resort supplier which practices regulated tariffs and the liberalized market supplied through 32 competitive suppliers at market prices. Also, remember that there are several voltage levels, the MAT, AT, MT, BTN and BTE, each with a specific tariff and which is composed as described in figure 2.15. Furthermore, as it will be better described in chapter 4, note that in 2012 it was established that normal consumers at any voltage level could no longer choose the regulated tariffs of the last resort supplier over market prices, preserving the first exclusively for consumers requiring extra social support, hence, with this tariff gaining the name of social tariff. However, as before 2012 there was such choice, in order to smooth this transition of thousands of consumers there were created the transitory tariffs, which are applied from last resort supplier to consumers who still did not switch to the liberalized market. Consequently, there are still consumers on the AT, MT, BTN and BTE voltage levels being supplied by the last resort supplier through the transitory tariffs, while MTA is the only voltage level for which regulated tariffs are already extinct. As expected, these tariffs applied by the last resort supplier have been higher than the average competitive market tariffs in order to discourage consumers to remain on this regime. However, its composition is very similar to the first. In conclusion, in 2017 there is the liberalized market supplying in all voltage levels and a regulated market supplying in all voltages levels excluding MAT, yet on a provisional regime. In order to promote a fair competitive market, in the liberalized market at the beginning of each year, suppliers send their reference tariffs to the market regulator ERSE, which is an estimation of the prices that they will charge during that year. The market regulator then complies all the information and publishes a document describing these tariffs in order to provide consumers an idea of the electricity retail market prices. Figure 2.24 represents the reference tariffs of the liberalized market for the year of 2017. (a) (b) Figure 2.24: Reference tariffs for 2017: (a) Absolute values. (b) Distribution in %.[13] On the current year, it is possible to conclude that prices paid by consumers range between 0,0773e, for MAT, and 0,1932e, for BTN, having increased 0,4% on average when 33 compared to 2016. In terms of activity tariffs, the highest variations on the average price were the 20% increase on the transmission grid usage tariff and the 5,3% reduction in the energy tariff. However, as in the previous years, the system general usage tariff, UGS, keeps having an heavy influence in the price paid by consumers, specially for BTN consumers who pay more for this cost than for any other activity. Remember that the UGS tariff includes the CIEG, which in turn include the overcost of PRE and the CMEC, the two most expensive political measures for consumers implemented recently in the Portuguese electricity sector. The first represents the overcost generated by the feed-in-tariffs paid to producers in the special regime, which is generated each time there is a difference between market prices and these subsidized tariffs. However, since 2012 that this this annual cost has been distributed by the following five years in order to stabilize the financial situation but which ended up accumulating debt for the energy system. This means that the in addition to PRE overcost of 2017 that consumers are paying this year shares of the PRE overcosts since 2013 also being charged on the final price. The second, was implemented in 2007 as a strategy to replace the contracts CAE that were active and which acted against the competitive market principles of MIBEL as they had established fixed prices for some conventional power plants. However, the CMEC, which were supposed to turn these competitive, were merely a bureaucratic change as they put these power plants now selling their energy at the market price but ensure them that their NPV remains equal to when they had the CAE contracts by charging the difference between this value and the results obtained in the market on the CIEG tariff. Thus, as it can be seen in figure 2.24, this year the average consumer is paying an electricity tariff of 0,14 e/kWh of which only 40% correspond to energy costs, 26% to transmission and distributions costs, being the low voltage distribution far more expensive than the others, and 34% to system global costs, UGS. Of these system global costs the PRE overcost and the CMEC alone have a combined impact of 34% for BTN consumers, 12% for BTE, 15% for MT, 18% for AT and 20,8% for MAT consumers. Going back on the construction of the electricity prices charged to consumers, before the integration of regulated tariffs and supplier costs, analyzing the pool market from where these get the electricity, it becomes clear that fixing an energy cost on the final tariff is only possible due to the derivatives market. For example, considering table 2.1, the average prices on the Portuguese daily market from July 2016 until June 2017 ranged between 40 e/MWh and 71 e/MWh while the energy tariff for 2017 remained almost constant during the whole time. Furthermore, in addition to the monthly variations of the market price there are also the daily and even hourly variations which result from demand highs and low, and the environmental conditions. Thus, in order to be protected from sudden price spikes, suppliers have at least part of their transactions negotiated with futures contracts. PT daily market (e/MWh) ES daily market (e/MWh) July 2016 40,36 40,53 August 2016 41,14 41,16 September 2016 43,61 43,59 October 2016 52,78 52,83 November 2016 56,25 56,13 December 2016 60,27 60,49 January 2017 71,52 71,49 February 2017 51,39 51,74 March 2017 43,95 43,19 April 2017 44,18 43,69 May 2017 47,12 47,11 June 2017 50,22 50,22 Total 50,24 50,19 Table 2.1: Average price on the Portuguese and Spanish daily markets in the last 12 months 34 Chapter 3 GCPVS in Portugal: Regulatory Framework In this chapter it will be done a review to the evolution of the regulatory framework that has been applied to the the energy system in general but highlighting the policies that most influenced the solar photovoltaic technology and its applications in the Portuguese market. Hence, the review will be structured by periods that marked the industry in a particular way, which might have been already mentioned in the previous sections but not from the regulatory perspective. 3.1 Before 1988: Monopoly The very first legislation structuring the electricity sector was established in 1944 through Law 2002 which reserved the concession of the activities of production, transmission and distribution of electricity for Portuguese citizens or for companies with most of its capital Portuguese. Three decades later, following the political revolution against the dictatorial right wing party, the Law Decree 205-G/75 nationalized most of the companies working on the electricity sector. Still on the same year, the Law Decree 502/76 created the public electricity company called EDP which had the exclusive concession of both production, transmission and distribution activities. During the following decade, any access to electricity sector was prohibited for private entities and the state was managing a monopoly. 3.2 1988-1994: Opening the electricity sector In 1988, the Law Decree 189/88 took a major step to start inverting the situation by allowing the entrance of small private producers into the national electric system, or SEN. Furthermore, it was the first time the small producer of renewable energy was included in the Portuguese legislation. Such document defined the operating conditions for producers which did not exceed the apparent power 10 MWA and that used renewable sources or national fuels as the primary resource. Still on 1988, the Law Decree 449/88 finally opened the access of private entities to the activities of production, transmission and distribution with no restrictions. Between 1991 and 1995 the importance of having a competitive electricity sector led to its vertical disintegration which implied that the companies operating in it had to be specifically focused on only one of the activities of production, transmission and distribution. At the same time, the sector was divided in two electric systems, one of public administration and another side market designed for private entities who wanted to operate on the production or commercialization on . 35 3.3 1995-1998: The new SEN In 1995, The Law Decree 182/95 redefined the organization of the electricity sector and it kept a two electric systems structure that existed before, yet making some changes. It defined a public system called SEP, and a non public system called Independent Electric System or SEI. Hence, from now on it is important to be aware of the following definitions: •Low voltage (BT): below 1 kV; •Medium voltage (MT): between 1 kV and 45 kV; •High voltage (AT): between 45 kV and 110 kV; •Very high voltage (MAT): above 110 kV; •Bounding contract: long term contract by which a producer is committed to deliver all the electricity it generates to the SEP and a distributor is committed to distribute all the electricity it receives from it; •Bounded license: license by which the its holder is committed to feed the SEP or be fed by it and follow the rules of this system; •Not bounded license: license by which the its holder is not committed to feed the SEP dedicating its activities for self interest or for a third party interest under a unregulated contract; On the one hand, according to the Law Decree 182/95 the public electric system SEP included four entities which were: one responsible for the planning of the system, holders of production bounded licenses, the managers of the TSO and the holders of distribution bounded licenses. For example, in a case where the entity responsible for the planning concluded that it was necessary the integration of an additional producer in the SEP production set, it would inform the management team of the TSO about the situation. The latter would then select among several alternatives the best producer and establish a bounding contract with a minimum duration of 15 years. The remuneration of such contract would be a mix of a fixed and a variable share. Hence, after the electricity was injected to the TSO it was sold to holders of distribution bounded licenses until it reached consumers of the SEP. On the other hand, the Independent Electric System, SEI, defined in the Law Decree 182/95 was composed by the non bounded electric system (SENV), cogeneration units, hydro power plants up to 10 MVA and renewable energy power plants. The SENV was composed by private entities operating in the production and distribution sections with own infrastructure at medium and high voltage levels and that had specific customers which were obliged to consume at least 100 GWh/year. These producers, distributors and consumers were holders of not bounded licenses and would arrange their electricity trading between each other. Still in 1995 the Law Decree 313/95 complemented the operation conditions of renewable technologies and established that after approval of operations from the responsible governmental agency, DGE1, the renewable power units would see their production remunerated as: 1. For units with connection power below than 10 MVA the monthly income was the result of the following equation: Monthlyrevenue = 0.8∗P P ∗p0 1Direc¸c˜ao Geral de Energia, today with the name Direc¸c˜ao Geral de Energia e Geologia, DGEG 36 Figure 3.1: Organization of SEN according to Law Decree 182/95 Where: PP is the monthly price of average usage tariff; p0is the minimum between P1 and P2 where: P1 = Ep Tp P2 = Ep+Ec Tp+Tc Being: Epthe monthly energy supplied at peak hours (kWh); Ecthe monthly energy supplied at full hours 2(kWh); Tpthe monthly duration of peak hours (h); Tcthe monthly duration of full hours (h); 2. For units with connection power above 10 MVA of apparent power, on those months where these units registered productions above the 10 MVA, the first 10 MVA were remunerated as specified in number 1 being the rest remunerated according to the avoided costs criteria during a period of 15 years. The values for the avoided costs were published each year by the regulating agency; 3.4 1999-2005: Empowering renewable energy In 1999 two main forces drove the implementation in the Law Decree 168/99: the policies to liberalize the electricity market that had been approved in the previous years and 2full hours is a translation from the Portuguese ”horas cheias” and are used in the tri-hour tariff system. Correspond to periods that are neither peak hours nor base hours. 37 the increase of environmental awareness combined with the objective of reducing carbon emissions. Hence, this document changed completely the legislation responsible for the renewable energy generation by updating its remuneration system, reorganizing the regulations for renewable energy producers and changing the way new grid connections were attributed. The new remuneration system published in this document was much more complex than the previous and it was based on the sum of a fixed, a variable and an environmental parameters with additional adjustments. The equation below represents the formula use to calculate the remuneration per unit of energy supplied, VRD: V RDm=KMHOm∗[PFm+PVm+PAm]∗IPCm−1 IPCref ∗1 (1 −LEV )(3.1) Where: •V RDmis the applicable feed-in-tariff in month, m; •KMHO is an optional 3coefficient that modulates PFmand PVmaccording to the hour of energy production; •PFmis a fixed amount paid to the renewable producers at month m; •PVmis a variable amount paid to the renewable producers at month m; •PAmis the amount paid to the producers at the month mcorresponding to the environmental benefits of the renewable power generation. •IPCm−1is the consumer price index, without housing, of the month before; •IPCref is the reference consumer price index, without housing, corresponding to December 1998; •LEV represents the losses in the transmission and distribution grids avoided by local renewable generation. For power plants with installed capacity, POTdec ≥5MW ⇒ LEV = 0.015. For power plants with P OTdec <5MW ⇒LEV = 0.035; If KMHO was not chosen it would be assumed the value of 1, otherwise it would be calculated as follows: KMHO =KMHOpc ∗ECRpc,m +KMHOv∗ECRv,m ECRm Where: •KMHOpc takes the value of 1.25 and corresponds to the peak and full hours; •ECRpc,m energy generated at peak and full hours by the renewable power plant; •KMHOvtakes the value of 0.65 and corresponds to the base hours 4; •ECRv,m energy generated at base hours by the renewable power plant; •ECRmenergy generated by the renewable power plant during month m; 3The producer must choose if wants to operate either with or without the coefficient at the time of license approval 4In the winter from 22h:00 to 08:00h and in the summer from 23:00h to 09:00h. The rest of the hours correspond to peak and full hours 38 The fixed share PFmin equation 3.1 takes into account the investment avoided by in new power plants thanks to the existence of such renewable energy injection, the power guarantee it provides to the grid and its average power injection. PFm=PFref ∗COEFpot,m ∗POTmed,m Where: •PFref = 5.45 e/kWh 5and it corresponds the the investment avoided in other electricity generation projects; •COEFpot,m is a coefficient related to the power guarantee that the grid is able to ensure thanks to the renewable producer; •POTmed,m =MIN P OTdec;ECRm 24∗30 corresponds to the average power, in kW, injected to the grid during month m.; The COEFpot,m was calculated as: COEFpot,m =NHPref,m NHOref,m =ECRm 0.8∗24 ∗30 ∗POTdec Where: •NHPref,m is the number of hours during mthe facility generated electricity at the declared power; •NHPref,m is the reference number of hours during mused for the calculation and equals 0.8∗24 ∗30; •POTdec is the declared installed capacity of the renewable energy generation facility; The variable share PVmin equation 3.1 corresponds the amount of operating and maintenance costs avoided in the construction of new electricity generation projects thanks to existence of the renewable power plant and is calculated as follows: PVm=PVref ∗ECRm Where the PVref takes the value 0.025 e/kWh 6and is the reference unit value for the operations and maintenance costs avoided thanks to each renewable power plant. Finally, the P Amin equation 3.1 corresponds to the environmental benefits of the renewable energy generation and is calculated as follows: PAm=ECEref ∗CCRref ∗ECRm Where: •ECEref is the reference unit for the value of carbon emissions avoided by the renewable electricity generation and it takes the value 7.5∗10−5e/g 7; •CCRref is the per unit amount of CO2 emitted by the reference power plant and it has the value of 370 g/kWh; 5conversion from 1090 PTE, the Portuguese currency before e 6converted from 5 PTE/kWh; 7converted from 0.015 PTE/g; 39 The period of application for this remuneration system defined in equation 3.1 was established at 144 months. After that producers would me remunerated as: V RDm=KMHOm∗IP Cm−1 IPCref ∗[PFm+PVm] + PAm∗1 (1 −LEV )(3.2) Solving the mathematical expressions defined above assuming certain parameters, the price at which a producer of electricity using solar PV technology would sell each kWh is presented in table 3.1. Figure 3.2 illustrates the inputs needed to calculate such value: LD 168/99 (e/kWh) KMHO 1 PF 0.0095 PV 0.0250 PA 0.0278 IPCm/IPCref 1.0 LEVP >5MW 0.015 LEVP≤5MW 0.035 V RDm,P >5MW 0.0632 V RDm,P ≤5MW 0.0645 Table 3.1: Remuneration according to Law Decree 168/99 Figure 3.2: Remuneration System inputs and outputs of Law-Decree 168/99. In 2001, although political strategy was focused on promoting the installation of new renewable power plants the grid capacity availability was not being able to keep up with the pace of new projects. Thus, many proposals were rejected due to grid limitations which resulted in the creation of Law-Decree 312/2001 that reformed the reception capacity management system of the public grid. It established that: 1. Every investment done by the transmission operator would be supported by the transmission grid usage tariff (URT) charged to consumers; 2. In order to promote the special regime, if the distribution grid had to invest in a new connection for this type of producers and such connection was smaller than 50 MVA, its cost would be charged to consumers through the distribution grid usage tariff and not to the investors; 40 Figure 4.1: GCPVS that were accounted for the analysis and their equivalent single unit The third column on the table represents the average daily production per kWp of capacity installed at each project location. Such values were obtained using the CM-SAF database of the Photovoltaic Geographic Information System,PVGIS, which is an online simulator platform of solar PV production supported by the European Commission. Hence, according to its description the data retrieved from this software is based on calculations from satellite images during a period of 12 years, between 1998 and 2011, which provides estimations with an approximated margin of error of 5% [5]. On the last row and the third column of the table it is represented the weighted average daily rate of production of this solar generation mix, which can be used to evaluate the system as a whole by replacing these several units, each with its parameters, by a single production unit with the cumulative capacity. All it is needed is to find a location sufficiently close to these units to ensure the same daylight hours throughout the year and with an average production rate equal to 4,32 kWh/kWp. Note that there are three projects that belong to the same promoter in Alc´acer do Sal, Castelo de Vide and Santar´em and that it is known that they will accumulate 56 MW of peak power but nothing is known about their distribution. It was assumed each of them will have one third of the total capacity assigned by the promoter. Finally, the chosen location that represents the whole system has the coordinates of 37,964 Latitude, -7,718 Longitude and all the following simulations allocated a total installed capacity of 480 MWp to this location. Figure 4.2: Interface of PVGIS platform [4]. 47 Figure 4.3: PVGIS estimation of daily radiation on January for given location [4] Figure 4.4: PVGIS estimation of yearly production for given location and capacity [4] Furthermore, it was assumed all the PV panels are of the fixed plane type, are oriented towards south at the optimal slope angle for this location, 34 degrees, and that the system has 14% of losses, which is the standard value of the software. When setting these parameters, the installed capacity and the location on PVGIS it is possible to obtain the daily radiation for every month and the yearly estimated production. The first provides the average solar irradiance W/m2at each 15 minutes for every month. The second provides the expected average daily production for the given scenario for every month taking into account the 14% losses of the system, 12% losses due to temperature and low irradiance 1and 2,6% loss due to angular reflectance effects. 4.2 Calculations Given the daily radiation shown in figure 4.3 as example it is possible to know how will the total daily generation be distributed along the day. Since the frequency of solar intensity data provided by PVGIS is at every 15 minutes it was assumed that during this period such parameter remains constant implying that, in this model, each hour of sunlight will have a maximum of four different solar radiation intensities. Hence, every hour was divided in four equal intervals 2and the solar radiation values obtained in the simulation were attributed to the respective interval. Considering that the solar irradiance Gis given in W/m2, the energy generated in 15 minutes, in W h/m2, with such solar irradiance is G/4. Adding the energy generated in each of these four intervals results in the total energy generated on average at such hour of such month per square meter. Finally, having calculated this for every hour of solar production it is possible to obtain the hourly distribution of solar production for a generic day of each month. The second step is to apply such distribution to the expected average daily production of our system retrieved from the other PVGIS simulation, represented in figure 4.4. In summary, it has been calculated the average hourly production of the given scenario for a generic day of each month based on the data provided by 12 years of satellite images. Figure 4.5 represents the process for the first hours of January as an example. Note that the third column of first table is equal to the one obtained in the simulations of figure 4.3. Following, to analyze the impact that the additional production of solar energy would have on the market price it was required to access relevant figures about the market during the period study. Thus, it was considered most appropriate to evaluate the impact 1using local ambient temperature 2xx:00h - xx:15h, xx:15h - xx:30h, xx:30h - xx:45h and xx:45h - yy:00h 48 Figure 4.5: Example of data analysis exercise for January. that such solar penetration would have had on the year of 2016 as it has the most recent annual market results. Furthermore, remember it was said that this study aims to evaluate the worst scenario a solar producer would face in the given scenario, which requires the assumption of certain conditions even acknowledging that some are not expected to happen so frequently on the current market. Thus, the following assumptions have to be considered: 1. The first assumption of the model is that there is no market splitting between Portugal and Spain, which already happens on most than 90% of the time and on what the MIBEL is working to improve. A congested interconnection line either rises or drops the price on the Portuguese region depending if it is exporting or importing. As it is not possible to predict in which direction the power will flow nor if the lines will constraint in the future because it depends on the generation mix and demand of the day, such ambiguity would bring inconsistent results. Consequently only days of 2016 without market splitting were studied. 2. The second assumption might be the most critical for the model as it has a substantial impact on the results. It was assumed that there is no technical restrictions on the daily market. As mentioned above, these are adjustments made, if necessary, by the grid operations manager depending on a daily grid security analysis and which are not supposed impact more than 5% of the energy transacted. However it has been seen in the recent times that this value is frequently exceeded which typically increases the price with the dispatch of expensive technologies such as thermal and nuclear. Hence, the solar producer if was still dispatched after such adjustments would only benefit from the situation by selling at a higher price. Since the aim is to evaluate the worst scenario for the producer it is reasonable to not include such technical restriction. Furthermore, as for the first assumption, being such adjustments unpredictable, their inclusion in the model would also bring inconsistencies to the results unless another model to predict such restrictions was used. 3. The third assumption consists on that the solar producers responsible for the 480 MW of GCPVS will always bid at 0 e/MWh to guarantee they are always dispatched if market price P≥0e/MWh. As result the study included the analysis of 12 days, one for each month, that were chosen based on three criteria: be weekdays, be the closest to the 15th day of the month and not have market splitting on the hours of solar production. Following, the analysis was based on affecting the intersection of buy and sell orders of each productive hour with 49 (a) (b) Figure 4.6: OMIE Daily market hourly price on:(a) January 15, 2016. (b) January 12, 2016. [38] Figure 4.7: Visual representation of the pretended analysis to be applied to each hour. Adapted from [38]. the respective estimated solar production of the given scenario. Hence, all the remaining information needed to complete the model could be found on the OMIE public website where it is provided a database with all the results of the spot market. Figure 4.6 represents the interface of the OMIE platform for the hourly price results on two distinct days. Note that the horizontal axis of both figures has the daily hours given in the format hour 1, 2, etc, which correspond to first hour of the day (00:00h - 01:00h), second hour (01:00h - 02:00h) and so on. This is the format that will be used on the rest of the model. Also, in the same figure it is possible to see a price difference between the Portuguese and Spanish markets - market splitting - from hour 7 to hour 10 of January 15th, consequently other days around this date were verified until that the January 12th was the closest day with a single price for the whole solar productive period and consequently was included in the analysis. The next step consisted on accessing the aggregated supply and demand curves for each hour of the day and shift the supply curve with the respective amounts of estimated solar production. Figure 4.7 illustrates an example of the intended objective through a visual representation, however it is merely descriptive as its values do not represent any real values calculated on the model. Note that curve affected is the sale offers curve and not the matched sale offers curve as the second includes the technical restrictions, here ignored. Thus, all the excel sheets containing the market bids for each hour of solar production 50 Figure 4.8: Demonstration of supply and demand curves construction for the first productive hour of January. [38] were downloaded from the database and for each one the data was grouped by categories of sale offer and purchase offer. This data was then processed to get the real aggregated demand and supply curves in the tabular form, which enabled the computation of the hypothetical supply curve with additional PV generation where the production of the given scenario was also included, bidding at 0 e/MWh. Again, figure 4.8 exemplifies such process for the first productive hour of January 12th, where the upper table represents the one containing market orders provided by OMIE with column F specifying if the order is a buy with a ”C”, or a sell with a ”V”, and the column I specifying if it is matched after technical restrictions with ”C”, or not with ”O”. The lower table represents the data already filtered and processed with the first row of last two columns corresponding to the solar production of the given scenario at such hour, in this hour 32,31 MWh. After repeating this process for each productive hour of all 12 days finally it was calculated the intersections between the three curves, demand - supply and demand - new supply containing additional solar PV, for which was used the software Matlab. Hence, for each hour it was created a script with the following structure: 1. Import data from the excel file. 2. Interpolate linearly each set of data - demand curve, supply curve and new supply curve - to each 0,1 MWh. 3. Find the coordinates where demand curve intersects supply curve. 4. Find the coordinates where demand curve intersect new supply curve. 5. Plot graphic. While the first step refuse any further explanation, the second step is critical for the model as without it it would impossible to find the intersections. Since the tables with bids imported from OMIE are composed by discrete buying and selling orders these had to be converted into linear functions in order to be possible to find an intersection between them. Hence, it was required to interpolate the data to the 0,1 MWh as this it is the smallest unit provided on data of OMIE database. For that, the Matlab function interp1 was used as it can be seen in figure 4.9. To find the intersection coordinates the function find was asked to return the first demand coordinates at which the price difference between the demand and supply curves, and repeated for the new supply curve, was below eps3, which is the smallest unit a computer is able to process. 3eps = 2−52 51 Figure 4.9: Matlab code used to calculate intersections Figure 4.10: Matlab plotting the demand and supply curves Figure 4.11: Results obtained for the 12th January, 2016. Having calculated all the intersections, two per hour of solar production, all it was remaining was to quantify the difference between intersections of each hour and present them clearly. Therefore, all the intersections were ”called” on a new file, which got all their values in a single run, and a final script was coded to present the absolute impact, in e/MWh, and the relative impact, in %, that the price of each analyzed hour would have been submitted. This final script was divided by sections where at each section the representative day of each month is studied. Figure 4.11 illustrates, as demonstration, the results obtained with the code on Appendix for January 12th 2016. 4.3 Simulations As mentioned above, this study was based on a sample of 12 days, one from each month of 2016, testing the impact that 480 MWp of solar PV capacity would have had on each hour of these days. As each month has its own daylight schedules, which determines the solar 52 Figure 4.12: Daily Production Distribution of Solar PV on given location production periods, the total amount of simulations were not evenly distributed between all days. Based on the data provided by the PVGIS software the daily solar production on the given location occurs between hours 6 and 20 on the largest days, in June, and hours 8 and 17 on the shortest days, in January and December. More specifically, figure 4.12 represents the daily production distribution on the location of the hypothetical base scenario for each month. Consequently, a total of 150 hours belonging to the following days have been studied according to the methodology described on the previous section: 1. January 12, 2016; 2. February 15, 2016; 3. March 15, 2016; 4. April 11, 2016; 5. May 16, 2016; 6. June 23, 2016; 7. July 14, 2016; 8. August 15, 2016; 9. September 15, 2016; 10. October 14, 2016; 11. November 15, 2016; 12. December 15, 2016; 53 Chapter 5 Results Analysis 5.1 By Month 5.1.1 January The electricity production on the given scenario was estimated to reach 1470 MWh on a reference day of January distributed as in table 5.1. Hence the simulations applied to the January 12th resulted on a total of 10 analyzed hours, from hour 8 to hour 17 as represented in figure 5.1, where the maximum impact on the price, before technical restrictions, was felt when this was at its highest value, between 08:00h and 09:00h. On the opposite, the lowest impact on the price took place between 16:00h and 17:00h where both supply curves intersected the same horizontal branch of the demand curve resulting in a 0 e/MWh impact on the price. In conclusion, the injection of 1470 MWh to the daily production decreased on average 0,8836 e/MWh the price on the hours that solar PV was producing, which is equivalent to 5,99%. January distribution MWh 07:00 - 08:00 2,2% 32,34 08:00 - 09:00 7,8% 114,95 09:00 - 10:00 11,4% 167,27 10:00 - 11:00 13,8% 202,86 11:00 - 12:00 15,0% 220,98 12:00 - 13:00 15,0% 220,98 13:00 - 14:00 13,8% 202,86 14:00 - 15:00 11,4% 167,27 15:00 - 16:00 7,8% 114,95 16:00 - 17:00 1,7% 25,55 TOTAL 100% 1470 Table 5.1: Estimated production of given scenario on a January reference day. However, it is important to expose the generation mix of January 12 2016 as it influences heavily the calculated results. Hence, according to OMIE monthly reports, the technologies producing during this day were mainly of the renewable type in both countries which allowed them to share a relatively low price without congesting the lines. While in Spain wind power dominated most of the production with hydro and nuclear coming right after, in Portugal it was hydro who dominated the production. See figure 5.3. 54 Figure 5.1: Simulation results for January 12 2016. 5.1.2 February For February, it was estimated that the daily electricity production of the 480 MWp of PV technology on our scenario would to reach the 1920 MWh between the hour 8 and 18 as shown in table 5.2. Hence the simulations done to the February 15th curves resulted on the impacts shown on figure 5.2 which, as it is possible to concluded, were almost zero except for hour 13 and 14. However, it is also possible to see that all the original prices, before technical restrictions, assumed a very low value which could only be practiced by renewable technologies. Hence, the this means that the demand curve intersected the supply curve when this was still on its first steps - mostly horizontal - and consequently the new supply curve which is only shifted to the right could not have great impact. In conclusion, the additional solar PV would have reduced the prices, on average during its productive hours, by 0,0778 e/MWh, or 4,4%. This value would have been much less if hours 13 and 14 were not included in the calculations as these have price impacts substantially higher than the rest. February distribution MWh 07:00 - 08:00 3,9% 75,72 08:00 - 09:00 7,8 % 149,96 09:00 - 10:00 10,9% 209,72 10:00 - 11:00 13,1% 251,12 11:00 - 12:00 14,2% 272,15 12:00 - 13:00 14,2% 272,15 13:00 - 14:00 13,1% 251,12 14:00 - 15:00 10,9% 209,72 15:00 - 16:00 7,8 % 149,96 16:00 - 17:00 3,9 % 75,72 17:00 - 18:00 0,1 % 2,67 TOTAL 100% 1920,00 Table 5.2: Estimated production of given scenario on a February reference day. Such low prices can be explained by looking at figure 5.4 where it can be seen that in this day there were almost no thermal power plants supplying the demand in the Iberian peninsula. The combination of wind and hydro power volume was considerably bigger than the most of the days of February. Therefore, on those days when more expensive technology produced more, probably the supply curve on the analyzed hours would have 55 Figure 5.2: Simulation results for February 15 2016. a higher slope resulting in a bigger impact of additional solar PV on the price. Figure 5.3: Generation mix on January 12 2016. Adapted from [28]. Figure 5.4: Generation mix on February 15 2016.Adapted from [29]. 5.1.3 March For March it was estimated that the 480 MWp of solar power would generate 2200 MWh on an average day distributed between hour 7 and 18. On March 15, the simulations showed that the impact on the price was considerably stable in most of the productive hours not exceeding the 0,32 e/MWh except on hour 15 which impact reached the 1,18 e/MWh. On average, the solar production on this day reduced the price by 0,2861 e/MWh, or 1,53%, on its productive hours. Note that the market prices, before technical restrictions, on this day were considerably higher than the days presented above due to the inclusion of a higher share of coal on the energy mix, as it is presented in figure 5.7. 56 Figure 5.14: Simulation results for August 15 2016. Figure 5.15: Generation mix on July 16 2016. Adapted from [30]. Figure 5.16: Generation mix on August 15 2016. Adapted from [26]. 63 5.1.9 September In September it was estimated that the daily solar production was distributed between hour 6 and 19 and reached a total generation of 2270 MWh. Hence, the 15th of September was analyzed which resulted in the outcomes of figure 5.17. It is possible to see that with the original prices before technical restriction being quite stable in all productive hours, ranging between 28 e/MWh and 35 e/MWh, the absolute impact was always below 0,3 e/MWh, excluding hour 11 when it reached 1,45 e/MWh. On average, the additional solar PV decreased the price by 0,22 e/MWh, or 0,72%. The energy mix on this day is represented on figure 5.19 and as it is possible to conclude, on this day coal dominated the Portuguese energy mix and nuclear the Spanish, while wind power came second on both markets. September distribution MWh 06:00 - 07:00 1,5% 34,12 07:00 - 08:00 4,6% 104,94 08:00 - 09:00 7,8% 177,06 09:00 - 10:00 10,5% 237,29 10:00 - 11:00 12,3% 279,68 11:00 - 12:00 13,3% 301,22 12:00 - 13:00 13,3% 301,22 13:00 - 14:00 12,3% 279,68 14:00 - 15:00 10,5% 237,29 15:00 - 16:00 7,8% 177,06 16:00 - 17:00 4,6% 104,94 17:00 - 18:00 1,5% 34,12 18:00 - 19:00 0,1% 1,38 TOTAL 100% 2270,00 Table 5.9: Estimated production of given scenario on a September reference day. Figure 5.17: Simulation results for September 15 2016. 5.1.10 October The daily solar production in October was estimated to reach 2020 MWh distributed between hours 7 and 18. So, the day submitted to the simulations was October 14 and as figure 5.18 demonstrates, the solar penetration on this day focused its impact between 64 hours 13 and 15 decreasing the market price in this period by nearly 0,9 e/MWh on average, almost four times the highest impact excluded from this period, at hour 16. Consequently, the average price reduction on the productive hours, before technical restrictions, was calculated to reach 0,2906 e/MWh, or 0,74%. In terms of electricity supply the energy mix of this day in the Portuguese region was mostly focused on thermal power plants with both coal and natural gas leading the total production, while in the Spanish region it was nuclear who dominated the production with coal and wind coming in second and third place, respectively. October distribution MWh 06:00 - 07:00 0,2% 3,61 07:00 - 08:00 4,3% 86,39 08:00 - 09:00 7,9% 158,59 09:00 - 10:00 10,8% 218,10 10:00 - 11:00 12,8% 259,18 11:00 - 12:00 13,9% 280,16 12:00 - 13:00 13,9% 280,16 13:00 - 14:00 12,8% 259,18 14:00 - 15:00 10,8% 218,10 15:00 - 16:00 7,9% 158,59 16:00 - 17:00 4,3% 86,39 17:00 - 18:00 0,6% 11,55 TOTAL 100% 2020,00 Table 5.10: Estimated production of given scenario on an October reference day. Figure 5.18: Simulation results for October 14 2016. 65 Figure 5.19: Generation mix on September 15 2016. Adapted from [36]. Figure 5.20: Generation mix on October 14 2016. Adapted from [35]. 5.1.11 November For the month of November it was estimated a total daily solar production of 1660 MWh, distributed between hours 8 and 17 as shown in table 5.11. Hence, the day analyzed in the simulations was the November 15th and as results on figure 5.21 represent, the impact cause by the solar penetration on the market price before technical restrictions on this day never exceeded the 0,28 e/MWh. Furthermore, it is also possible to conclude that the original prices were considerably higher on this day than on any of the days seen until now, never falling below 43 e/MWh on the solar productive hours. The average price impact on these hours was calculated to be 0,0828 e/MWh, or 0,19%. To conclude, the energy mix supplying the analyzed day is represented on figure 5.23 and shows that, while wind power was increased compared to the previous simulation and supplied most of the consumption on both countries, hydro power and other renewable technologies production was decreased which resulted on a bigger share of fossil fuels production and a price increase. November distribution MWh 07:00 - 08:00 2,9% 48,61 08:00 - 09:00 8,8% 145,62 09:00 - 10:00 12,3% 203,89 10:00 - 11:00 14,6% 241,74 11:00 - 12:00 15,5% 257,95 12:00 - 13:00 15,2% 252,62 13:00 - 14:00 13,6% 225,43 14:00 - 15:00 10,7% 177,25 15:00 - 16:00 5,7% 94,12 16:00 - 17:00 0,8% 12,76 TOTAL 100% 1660,00 Table 5.11: Estimated production of given scenario on a November reference day. 66 Figure 5.21: Simulation results for November 15 2016. 5.1.12 December In December, it was estimated that the given scenario would provide a daily solar production of 1380 MWh, making it the less productive month. This daily production would then be distributed between hours 8 and 17, as shown in table 5.12, which also makes it the month with shortest daylight periods, together with January. So, the simulations were applied to December 15th, which was the day within the sample with highest prices, following the trend seen in the previous months. Hence, the results of figure 5.22 indicated that every hour of the solar productive period of the day was affected by this PV penetration, which in this study had only happened in March 15. However, only between hours 9 and 13 this impact was above the 0,1 e/MWh. On this day the price before technical restrictions was reduced on average, during the productive hours, by 0,2802 e/MWh, or 0,5%. Such high prices mentioned above are explained by the energy mix of the day, with thermal and coal technologies leading the production on both countries. December distribution MWh 07:00 - 08:00 1,4% 19,71 08:00 - 09:00 7,7% 106,56 09:00 - 10:00 11,6% 159,42 10:00 - 11:00 14,2% 195,57 11:00 - 12:00 15,5% 213,87 12:00 - 13:00 15,5% 213,87 13:00 - 14:00 14,2% 195,57 14:00 - 15:00 11,6% 159,42 15:00 - 16:00 7,7% 106,56 16:00 - 17:00 0,7% 9,43 TOTAL 100% 1380,00 Table 5.12: Estimated production of given scenario on a December reference day. 67 Figure 5.22: Simulation results for December 15 2016. Figure 5.23: Generation mix on November 15 2016. Adapted from [34]. Figure 5.24: Generation mix on December 14 2016. Adapted from [27]. 5.2 Averages In order to have a broad perspective of the results of the simulations it is important to present the average values of these. Thus, in figures 5.25 and 5.27 it is presented, for the solar productive period, the average price, the average absolute impact and the average relative impact of each day and hour. However, remember that the sample size used to calculate each of these values was not constant as the daylight periods are also variable. For example, the average values of hour 20 only include the results of June and July as these are the only months with solar production at this hour. The findings of figure 5.25 indicate that, as expected, the three days with higher daily production, which are the ones belonging to June, July and August, are among the days with higher average absolute impact. However, January 12th was also among these days despite the fact that it had the second smallest daily production rate, which might seem quite contradictory. Looking at the figure it can be seen that the three days of June, July and August had an average original price before technical restrictions quite similar, ranging from 19,91 e/MWh to 22,89 e/MWh and that the closest practiced prices to 68 this range belonged to the days of January and March, 10,37e/MWh and 28,19e/MWh, respectively, which are exactly the other two days in the top five days with highest absolute impact. However, note that January 12 absolute impact was three times higher than March’s day impact. Therefore, based on the sample used, it can be concluded that the market prices suffered a substantially higher absolute impact when the average price, during solar productive hours and before technical restrictions, was between between the 10,37 e/MWh and 22,89 e/MWh. Actually, for the four days having its average price within this range this was reduced by an average of 0,8375 e/MWh while the calculated average reduction when all days were included was less than half of it, by 0,41 e/MWh. Furthermore, as represented in figure 5.26, with the exception of October and December days, as daily average price approaches to the interval 10,37 - 22,89 e/MWh the absolute impact also increases, supporting the theory that the average price most sensitive to the solar penetration of the given scenario relies within this interval. However, with January 12th average price being so distinct from the prices of June, July and August days, the other days with highest impacts, it deserves a particular attention. Looking at this day results on figure 5.1 it is clear that such absolute impact derives mainly from hour 9 alone with its price being decreased by 5,2 e/MWh, more than four times the second highest impact of the day. In addition, it is possible to see that the price at hour 9 was 18,2 e/MWh, which is also substantially higher than any other price of the day and much closer to the average price of July, June and August, indicating that the most sensitive price relies closer to 20 e/MWh than to 10 e/MWh. Figure 5.25: Averages by month Figure 5.26: Average daily impact ordered by increasing average price Nevertheless, while that for consumers having its consumption volume depending on 69 the market price (elastic demand) the absolute impact is the most interesting parameter to be analyzed since it will influence the energy that these will be able to buy, for consumers with a fixed consumption rate (rigid demand) it is more important to analyze the relative impact. For this consumers it has more meaning knowing that their electricity expenses could be reduced by 5% or 10% after such solar PV penetration. Hence, looking at figure 5.25 it is possible to conclude that the relative impact, even though it is not strictly inversely proportional to the average market price, it clearly assumes different magnitudes for daily average prices before technical restrictions above and below the 22 - 28 e/MWh interval. While that on the days from January to August excluding March, which correspond to the lowest price days, the average price was reduce by 3,4% to 8,52%, on the rest this reduction was never above 1,01%. An alternative analysis that can be done to complement the one above is to calculate the average impact at each hour of solar production on the given sample. This approach provides the perspective of the average solar impact through out the day. Therefore, as figure 5.27 represents, it is clear that the strong solar production impact relies between hour 9 and 15. However, while the maximum production occurs at hours 12 and 13, with these combining between 25% and 31% of total daily productions depending on the month, it seems that the maximum impact happens at hour 9 which is never responsible for more than 8,8% of total production, see figure 4.12. Thus, this might be explained by the average price of hour 9 or by the average energy mix supplying at hour 9 on the days of the sample. When looking at the remaining hours of the period with highest absolute impact, from hour 9 to 15, it is possible to conclude that, with the exceptions of hours 11 and 12, as the other prices approach the price of hour 9 their absolute impact also increases. Thus, it seems that an average hourly price of 24,44 e/MWh, practiced at hour 9, is closer to the price that experiences a higher impact and which could be the one of hour 7 or 8, as these are the only above such price, but since solar production on these hours is substantially lower their impact does not represent it. Finally, based on the sample used, the hourly average price decrease reached a maximum of 0,78 e/MWh at hour 9 and a minimum of 0,01 e/MWh at hour 19, setting an average of 0,37 e/MWh. On the contrary to what happens with the average daily prices where the values experience great variations between each other, the average hourly price is much more constant ranging between 14,66 e/MWh and 25,32 e/MWh. Furthermore, the relative impact and absolute impact on the average hourly price, except for hour 18 were always proportional as figure 5.28 represents. Figure 5.27: Averages by hour To conclude, having evaluate the average impact at each hour and day of the given 70 Figure 5.28: Average hourly impact ordered by increasing average price sample, it seems clear that the higher average absolute impact takes place when the price before technical restrictions is around 24 e/MWh. In fact, for both analysis presented above it was concluded that the maximum impact was neither at the most productive day nor hour but when the average price was closer to this value. However, it is important to recall that it must be taken into account the energy mix that was present at each day and hour as it is such mix that defines the supply curve and consequently influences the sensitivity of price. 5.3 Economic Impact Having presented the results obtained from the simulations it is time to translate these into what they might mean for a solar producer included on those 480 MWp of the analyzed scenario. Thus, assuming that each analyzed day of 2016 represents its whole month in terms of solar daily production, average market price before technical restrictions and average impact of additional solar PV in such price is possible to calculate the economic impact that the the given scenario would have caused on this hypothetical 2016. However, before presenting the results it is important to remember that as the prices of this scenario do not include technical restrictions, because they try to simulate the worst scenario from the perspective of the solar producer and these might increase the price, the revenues presented on table 5.13 will be below of what investors would ever expect if these include them on their calculations. Revenue with sample prices Revenue with additional PV Diff January 467 180,54 e431 812,00 e35 368,53 e February 99 648,21 e93 352,44 e6 295,77 e March 62 572,13 e61 848,42 e723,70 e April 241 874,01 e227 521,55 e14 352,47 e May 296 625,79 e277 836,33 e18 789,46 e June 1 530 354,03 e1 466 423,99 e63 930,04 e July 1 557 941,35 e1 454 317,10 e103 624,25 e August 1 695 036,32 e1 593 769,20 e101 267,12 e September 2 059 207,44 e2 038 016,33 e21 191,11 e October 2 455 799,22 e2 431 082,64 e24 716,58 e November 2 162 144,94 e2 156 720,53 e5 424,41 e December 2 341 834,53 e2 326 483,69 e15 350,84 e TOTAL 14 970 218,51 e14 559 184,21 e411 034,30 e Table 5.13: Estimated annual revenue of the given scenario with sample prices before technical restrictions and reduced prices due to solar penetration As seen in table 5.13 the results of the economic analysis, extrapolating each hourly price before technical restrictions from the sample to the rest of the respective month, indicate that the additional 480 MWp of solar PV would have had a revenue with prices of 2016 411 043 ehigher than with reduced prices. Such economic impact corresponds to a decrease of 2,75% on the annual income and an average monthly revenue 34 252eshorter. 71 Chapter 6 Conclusion Being the electricity market a playground for a competitive regime, meaning that its prices will get closer to the marginal costs of each production technology as competition increases, one cannot assume that the penetration of generation units in energy mix will not influence the price. Therefore, before investing in a new production unit to penetrate the market and estimating its potential revenue with current prices, it is important that the investor does the exercise of calculating the impact that itself will have on the market prices before doing such economic evaluation. This because there will be a point where the penetration of additional low variable cost technology, as solar PV, in a market which is already saturated will push the market prices below the minimum limit to keep its production profitable. In Portugal, with the construction approval of new solar PV power plants which will compete in the market with other producers this analysis of calculating the saturation point has become an even more important. With that in mind, this thesis aimed to quantify the impact that the additional solar PV technology penetrating the Portuguese market and not benefiting from feed-in-tariffs might have on its electricity prices. Based on the recent reports, until the moment 480 MWp of this type of production were approved for construction which served as base scenario for the simulations. Hence, one day from each month of 2016 was chosen based on the following criteria: be the closest to the 15th day of the month, not have market splitting on solar productive hours and be weekdays. Following, these days were analyzed by considering their supply and demand curves without technical restrictions of each solar productive hour. These supply curves were then adjusted to include the respective solar production of the base scenario assuming these producers were bidding at 0 e/MWh to ensure they were always dispatched. As expected and demonstrated in the results of the simulations, such penetration would have caused a reduction in market prices depending on several parameters such as the original market price before technical restrictions, the energy mix supplying the demand at each hour, which is also related to the first, and the volume of solar production coming from the base scenario at such hour. The results of each day and hour were considerably different from each other as the the parameters mentioned above defining them were also very different. In fact, the second half of 2016 was much more dry than the first half, implying higher market prices on the days from September to December than on the rest, never falling below 30 e/MWh, due to a more frequent dispatch of fossil fuels thermal power plants. On the contrary, the analyzed days of February, April and May were supplied almost exclusively by renewable sources which set their average prices before technical restrictions far below the other days, never exceeding the 4 e/MWh. However, when looking at the daily average impacts on figure 5.25 it is possible to see that its maximum would not have occurred at any of these two periods, but when the averages prices during solar productive periods and before technical restrictions were set at lows 20 e/MWh. More specifically, this conclusion resulted from 72 Step 2: Compiling all the files 1%% Compilation of r e s u l t s , 3 min to run%% 2c l e a r a l l ; 3 4%January 5jan h8 ; jan h9 ; jan h10 ; jan h11 ; jan h12 ; jan h13 ; jan h14 ; jan h15 ; jan h16 ; jan h17 ; 6 7%February 8feb h8 ; feb h9 ; feb h10 ; feb h11 ; feb h12 ; feb h13 ; feb h14 ; feb h15 ; feb h16 ; feb h17 ; feb h18 ; 9 10 %March 11 mar h7 ; mar h8 ; mar h9 ; mar h10 ; mar h11 ; mar h12 ; mar h13 ; mar h14 ; mar h15 ; mar h16 ; mar h17 ; mar h18 ; 12 13 %April 14 apr h6 ; apr h7 ; apr h8 ; apr h9 ; apr h10 ; apr h11 ; apr h12 ; apr h13 ; apr h14 ; apr h15 ; apr h16 ; apr h17 ; apr h18 ; 15 apr h19 ; 16 17 %May 18 may h6 ; may h7 ; may h8 ; may h9 ; may h10 ; may h11 ; may h12 ; may h13 ; may h14 ; may h15 ; may h16 ; may h17 ; may h18 ; 19 may h19 ; 20 21 %June 22 jun h6 ; jun h7 ; jun h8 ; jun h9 ; jun h10 ; jun h11 ; jun h12 ; jun h13 ; jun h14 ; jun h15 ; jun h16 ; jun h17 ; jun h18 ; 23 jun h19 ; jun h20 24 25 %July 26 j u l h 6 ; j u l h 7 ; j u l h 8 ; j u l h 9 ; j ul h 1 0 ; j u l h 1 1 ; j u l h 1 2 ; j u l h 1 3 ; jul h14 ; jul h15 ; jul h16 ; jul h17 ; jul h18 ; 27 jul h19 ; jul h20 ; 28 29 %August 30 aug h6 ; aug h7 ; aug h8 ; aug h9 ; aug h10 ; aug h11 ; aug h12 ; aug h13 ; aug h14 ; aug h15 ; aug h16 ; aug h17 ; aug h18 ; 31 aug h19 ; 32 33 %September 34 sep h7 ; sep h8 ; sep h9 ; sep h10 ; sep h11 ; sep h12 ; sep h13 ; sep h14 ; sep h15 ; sep h16 ; sep h17 ; sep h18 ; sep h19 ; 35 36 %October 37 oct h7 ; oct h8 ; oct h9 ; oct h10 ; oct h11 ; oct h12 ; oct h13 ; oct h14 ; oct h15 ; oct h16 ; oct h17 ; oct h18 ; 38 39 %November 79 40 nov h8 ; nov h9 ; nov h10 ; nov h11 ; nov h12 ; nov h13 ; nov h14 ; nov h15 ; nov h16 ; nov h17 ; 41 42 %December 43 dec h8 ; dec h9 ; dec h10 ; dec h11 ; dec h12 ; dec h13 ; dec h14 ; dec h15 ; dec h16 ; dec h17 ; Step 3: Working the Results - January as example 1%% January 2% begin c o l o r l i s t 3myblue=’ [ 0 0.5 1] ’ ; 4mytextblue=’ [ 0 0.1 1 ] ’ ; 5myorange=’ [1 .5 0] ’ ; 6mytextorange=’ [1 .2 0 ] ’ ; 7mygreen=’ [0 0.7 0 . 3 ] ’ ; 8mytextgreen=’ [ 0 0.4 0 . 3 ] ’ ; 9% end c o lo r l i s t 10 j a n p r i c e a r r a y =[ p r i c e j a n h 8 ; p r i c e j a n h 9 ; p r i c e j a n h 1 0 ; pr ic e j an h1 1 ; p ri ce ja n h 12 ; p r ic e j an h1 3 ; pr ic e j an h1 4 ; pr ic e j an h1 5 ; 11 pr ic e j an h1 6 ; pr ic e j an h1 7 ] ; 12 13 jan price array PV =[ price jan h8 PV ; price jan h9 PV ; price jan h10 PV ; price jan h11 PV ; price jan h12 PV ; price jan h13 PV ; price jan h14 PV ; 14 price jan h15 PV ; price jan h16 PV ; price jan h17 PV ]; 15 16 jan combined =[ j a n p r i c e a r r a y ( : ) , jan p ri ce a rray PV ( : ) ] ; 17 18 jan impact abs =( j an p r ic e a rr ay −jan price array PV); 19 jan impa ct perc=( jan impact abs . / j a n p r i c e a r r a y ) ∗100; 20 21 jan hours =8:17; 22 23 %Display Results 24 f i g u r e 25 subplot ( 2 , 2 , [ 1 , 2 ] ) ; %f i r s t bar chart 26 hb1=bar( jan hours , jan combined , 0 . 8 , ’grouped ’) ; 27 hb1 (1) . FaceColor = myblue ; 28 hb1 (2) . FaceColor = ’ red ’ ; 29 xlabel(’ Hour of the Day (h) ’ ) 30 ylabel(’ E l e c t r i c i t y Price ( Eur/MWh) ’ ) 31 32 title(’JANUARY 12 th 2016 ’ ) 33 text( jan hours , jan combined ( : , 1 ) , num2str ( jan combined ( : , 1 ) , ’%0.2 f ’ ) , ’ HorizontalAlignment ’ ,’ r ig h t ’ ,’VerticalAlignment ’ ,’bottom ’,’ co l or ’ , mytextblue) ; 34 text( jan hours , jan combined ( : , 2 ) , num2str ( jan combined ( : , 2 ) , ’%0.2 f ’ ) , ’ HorizontalAlignment ’ ,’ l e f t ’ ,’VerticalAlignment ’ ,’bottom ’ ,’ co l or ’ ,’ red ’ ) ; 35 legend1jan = legend (hb1 ( : ) , ’ Real s i t u a t i o n ’ ,’Situation with 80 ad di ti o na l 480 MW of Solar PV’ ) ; 36 ax = gca ;% current axes 37 ax . YLim = [ 0 2 2 ] ; 38 39 subplot (2 ,2 ,3) ; %second bar chart 40 hb2=bar( jan hours , jan impact abs , ’FaceColor ’ , myorange ) ; 41 xlabel(’ Hour of the Day (h) ’ ) 42 ylabel(’ Absolute Impact ( Eur/MWh) ’ )% l e f t y−axis 43 text( jan hours , jan impact abs , num2str (jan impact abs , ’ %0.2 f ’ ) , ’ HorizontalAlignment ’ ,’ center ’ ,’VerticalAlignment ’ ,’bottom ’ ,’ co l or ’ ,mytextorange) ; 44 legend2jan=legend (hb2 , ’ Absolute impact ’ ) ; 45 46 subplot (2 ,2 ,4) ; %Third bar chart 47 hb3=bar( jan hours , jan impact perc , ’ FaceColor ’ , mygreen ) ; 48 xlabel(’ Hour of the Day (h) ’ ) 49 ylabel(’ Relative Impact (%) ’ ) ; 50 text( jan hours , jan impact perc , num2str( jan impact perc , ’ %0.2 f ’ ) , ’ HorizontalAlignment ’ ,’center ’ ,’VerticalAlignment ’ ,’bottom ’ ,’ co l or ’ ,mytextgreen) ; 51 legend3jan=legend (hb3 , ’ Relative impact ’ ) ; 52 ax = gca ;% current axes 53 ax . YLim = [ 0 3 5 ] ; 54 55 ja n p ri c e av g=mean( j a n p r i c e a r r a y ) ; 56 jan pricePV avg=mean( jan price array PV ) ; 57 58 jan impact perc avg=mean( jan impact perc ) ; 59 jan impact abs avg=mean( jan impact abs ) ; 60 61 %% Repeat f o r every month 62 %...... 63 %% Def ining new v ec tors needed f o r Conclusion 64 65 66 %d e fi n in g p r i c e s per hour in v ec tor s 67 pr ic e h 6 a rr a y =[ p ric e ap r h6 ; price may h6 ; p ri c e ju n h 6 ; p r i c e j u l h 6 ; price aug h6 ] ; 68 pr ice h7 ar ray =[ price mar h7 ; p ric e a pr h 7 ; price may h7 ; pr ice ju n h7 ; p r i c e j u l h 7 ; price aug h7 ; p ri ce s ep h 7 ; 69 price oct h7 ]; 70 p r i c e h 8 a r r a y =[ p r i c e j a n h 8 ; p r i c e f e b h 8 ; price ma r h8 ; pr ice ap r h 8 ; price may h8 ; pr i ce j un h 8 ; p r i c e j u l h 8 ; 71 price a ug h8 ; p r i ce s e p h 8 ; p r i c e o c t h 8 ; price nov h8 ; pri ce d ec h 8 ] ; 72 p r i c e h 9 a r r a y =[ p r i c e j a n h 9 ; p r i c e f e b h 9 ; price ma r h9 ; pr ice ap r h 9 ; price may h9 ; pr i ce j un h 9 ; p r i c e j u l h 9 ; 73 price a ug h9 ; p r i ce s e p h 9 ; p r i c e o c t h 9 ; price nov h9 ; pri ce d ec h 9 ] ; 74 p r ic e h 1 0 a r r ay =[ pr i c e j a n h 1 0 ; p r i c e f e b h 1 0 ; price mar h10 ; pri ce a pr h 10 ; price may h10 ; pri ce jun h 10 ; p r i c e j u l h 1 0 ; 81 75 price aug h10 ; pric e se p h 1 0 ; p ri c e o c t h 10 ; price nov h10 ; p rice de c h10 ] ; 76 p r ic e h 1 1 a r r ay =[ pr i c e j a n h 1 1 ; p r i c e f e b h 1 1 ; price mar h11 ; pri ce a pr h 11 ; price may h11 ; pri ce jun h 11 ; p r i c e j u l h 1 1 ; 77 price aug h11 ; pric e se p h 1 1 ; p ri c e o c t h 11 ; price nov h11 ; p rice de c h11 ] ; 78 p r ic e h 1 2 a r r ay =[ pr i c e j a n h 1 2 ; p r i c e f e b h 1 2 ; price mar h12 ; pri ce a pr h 12 ; price may h12 ; pri ce jun h 12 ; p r i c e j u l h 1 2 ; 79 price aug h12 ; pric e se p h 1 2 ; p ri c e o c t h 12 ; price nov h12 ; p rice de c h12 ] ; 80 p r ic e h 1 3 a r r ay =[ pr i c e j a n h 1 3 ; p r i c e f e b h 1 3 ; price mar h13 ; pri ce a pr h 13 ; price may h13 ; pri ce jun h 13 ; p r i c e j u l h 1 3 ; 81 price aug h13 ; pric e se p h 1 3 ; p ri c e o c t h 13 ; price nov h13 ; p rice de c h13 ] ; 82 p r ic e h 1 4 a r r ay =[ pr i c e j a n h 1 4 ; p r i c e f e b h 1 4 ; price mar h14 ; pri ce a pr h 14 ; price may h14 ; pri ce jun h 14 ; p r i c e j u l h 1 4 ; 83 price aug h14 ; pric e se p h 1 4 ; p ri c e o c t h 14 ; price nov h14 ; p rice de c h14 ] ; 84 p r ic e h 1 5 a r r ay =[ pr i c e j a n h 1 5 ; p r i c e f e b h 1 5 ; price mar h15 ; pri ce a pr h 15 ; price may h15 ; pri ce jun h 15 ; p r i c e j u l h 1 5 ; 85 price aug h15 ; pric e se p h 1 5 ; p ri c e o c t h 15 ; price nov h15 ; p rice de c h15 ] ; 86 p r ic e h 1 6 a r r ay =[ pr i c e j a n h 1 6 ; p r i c e f e b h 1 6 ; price mar h16 ; pri ce a pr h 16 ; price may h16 ; pri ce jun h 16 ; p r i c e j u l h 1 6 ; 87 price aug h16 ; pric e se p h 1 6 ; p ri c e o c t h 16 ; price nov h16 ; p rice de c h16 ] ; 88 89 p r ic e h 1 7 a r r ay =[ pr i c e j a n h 1 7 ; p r i c e f e b h 1 7 ; price mar h17 ; pri ce a pr h 17 ; price may h17 ; pri ce jun h 17 ; p r i c e j u l h 1 7 ; 90 price aug h17 ; pric e se p h 1 7 ; p ri c e o c t h 17 ; price nov h17 ; p rice de c h17 ] ; 91 92 p r ic e h 1 8 a r r ay =[ p r i c e f e b h 1 8 ; price mar h1 8 ; p ric e a pr h 18 ; price may h18 ; p ric e j un h1 8 ; p r i c e j u l h 1 8 ; 93 price aug h18 ; pric e se p h 1 8 ; p ri c e o c t h 18 ] ; 94 95 pr ice h19 ar ray =[ pr ice apr h1 9 ; price may h19 ; pri ce ju n h19 ; p r i c e j u l h 1 9 ; 96 price aug h19 ; p ri ce s ep h 19 ] ; 97 98 pr ice h 20 a rra y =[ pri ce j un h2 0 ; p r i c e j u l h 2 0 ] ; 99 100 %−−−−− For new p r i c e s 101 price h6 PV array =[ price apr h6 PV ; price may h6 PV ; price jun h6 PV ; pric e jul h 6 PV ; price aug h6 PV ] ; 102 price h7 PV array =[ price mar h7 PV ; price apr h7 PV ; price may h7 PV ; price jun h7 PV ; price jul h7 PV ; 103 price aug h7 PV ; price sep h7 PV ; p r i c e o c t h 7 ] ; 104 price h8 PV array =[ price jan h8 PV ; price feb h8 PV ; price mar h8 PV ; price apr h8 PV ; price may h8 PV ; 82 105 price jun h8 PV ; price jul h8 PV ; 106 price aug h8 PV ; price sep h8 PV ; price oct h8 PV ; price nov h8 PV ; price dec h8 PV ] ; 107 price h9 PV array =[ price jan h9 PV ; price feb h9 PV ; price mar h9 PV ; price apr h9 PV ; price may h9 PV ; price jun h9 PV ; price jul h9 PV ; 108 price aug h9 PV ; price sep h9 PV ; price oct h9 PV ; price nov h9 PV ; price dec h9 PV ] ; 109 price h10 PV array =[ price jan h10 PV ; price feb h10 PV ; price mar h10 PV ; price apr h10 PV ; price may h10 PV ; price jun h10 PV ; price jul h10 PV ; 110 price aug h10 PV ; price sep h10 PV ; price oct h10 PV ; price nov h10 PV ; price dec h10 PV ]; 111 price h11 PV array =[ price jan h11 PV ; price feb h11 PV ; price mar h11 PV ; price apr h11 PV ; price may h11 PV ; price jun h11 PV ; price jul h11 PV ; 112 price aug h11 PV ; price sep h11 PV ; price oct h11 PV ; price nov h11 PV ; price dec h11 PV ]; 113 price h12 PV array =[ price jan h12 PV ; price feb h12 PV ; price mar h12 PV ; price apr h12 PV ; price may h12 PV ; price jun h12 PV ; price jul h12 PV ; 114 price aug h12 PV ; price sep h12 PV ; price oct h12 PV ; price nov h12 PV ; price dec h12 PV ]; 115 price h13 PV array =[ price jan h13 PV ; price feb h13 PV ; price mar h13 PV ; price apr h13 PV ; price may h13 PV ; price jun h13 PV ; price jul h13 PV ; 116 price aug h13 PV ; price sep h13 PV ; price oct h13 PV ; price nov h13 PV ; price dec h13 PV ]; 117 price h14 PV array =[ price jan h14 PV ; price feb h14 PV ; price mar h14 PV ; price apr h14 PV ; price may h14 PV ; price jun h14 PV ; price jul h14 PV ; 118 price aug h14 PV ; price sep h14 PV ; price oct h14 PV ; price nov h14 PV ; price dec h14 PV ]; 119 price h15 PV array =[ price jan h15 PV ; price feb h15 PV ; price mar h15 PV ; price apr h15 PV ; price may h15 PV ; price jun h15 PV ; price jul h15 PV ; 120 price aug h15 PV ; price sep h15 PV ; price oct h15 PV ; price nov h15 PV ; price dec h15 PV ]; 121 price h16 PV array =[ price jan h16 PV ; price feb h16 PV ; price mar h16 PV ; price apr h16 PV ; price may h16 PV ; price jun h16 PV ; price jul h16 PV ; 122 price aug h16 PV ; price sep h16 PV ; price oct h16 PV ; price nov h16 PV ; price dec h16 PV ]; 123 price h17 PV array =[ price jan h17 PV ; price feb h17 PV ; 83 price mar h17 PV ; price apr h17 PV ; price may h17 PV ; price jun h17 PV ; price jul h17 PV ; 124 price aug h17 PV ; price sep h17 PV ; price oct h17 PV ; price nov h17 PV ; price dec h17 PV ]; 125 price h18 PV array =[ price feb h18 PV ; price mar h18 PV ; price apr h18 PV ; price may h18 PV ; price jun h18 PV ; price jul h18 PV ; 126 price aug h18 PV ; price sep h18 PV ; price oct h18 PV ] ; 127 price h19 PV array =[ price apr h19 PV ; price may h19 PV ; price jun h19 PV ; price jul h19 PV ; 128 price aug h19 PV ; price sep h19 PV ] ; 129 price h20 PV array =[ price jun h20 PV ; price jul h20 PV ] ; 130 131 132 %% Average pr ic e per hour 133 price h6 avg=mean( p ri ce h6 ar r ay ) ; price h6 PV avg=mean( price h6 PV array ) ; 134 price h7 avg=mean( p ri ce h7 ar r ay ) ; price h7 PV avg=mean( price h7 PV array ) ; 135 price h8 avg=mean( p ri ce h8 ar r ay ) ; price h8 PV avg=mean( price h8 PV array ) ; 136 price h9 avg=mean( p ri ce h9 ar r ay ) ; price h9 PV avg=mean( price h9 PV array ) ; 137 price h10 avg=mean( pr ice h10 ar r ay ) ; price h10 PV avg=mean( price h10 PV array ) ; 138 price h11 avg=mean( pr ic e h1 1 ar ray ) ; price h11 PV avg=mean( price h11 PV array ) ; 139 price h12 avg=mean( pr ic e h1 2 ar ray ) ; price h12 PV avg=mean( price h12 PV array ) ; 140 price h13 avg=mean( pr ic e h1 3 ar ray ) ; price h13 PV avg=mean( price h13 PV array ) ; 141 price h14 avg=mean( pr ic e h1 4 ar ray ) ; price h14 PV avg=mean( price h14 PV array ) ; 142 price h15 avg=mean( pr ic e h1 5 ar ray ) ; price h15 PV avg=mean( price h15 PV array ) ; 143 price h16 avg=mean( pr ic e h1 6 ar ray ) ; price h16 PV avg=mean( price h16 PV array ) ; 144 price h17 avg=mean( pr ic e h1 7 ar ray ) ; price h17 PV avg=mean( price h17 PV array ) ; 145 price h18 avg=mean( pr ic e h1 8 ar ray ) ; price h18 PV avg=mean( price h18 PV array ) ; 146 price h19 avg=mean( pr ic e h1 9 ar ray ) ; price h19 PV avg=mean( price h19 PV array ) ; 147 price h20 avg=mean( pr ic e h2 0 ar ray ) ; price h20 PV avg=mean( price h20 PV array ) ; 148 149 %put i t in the vector form 150 price avg byhour =[ pri ce h6 avg ; price h7 a vg ; price h8 avg ; price h9 avg ; price h10 avg ; price h11 avg ; 84 151 price h12 avg ; price h13 avg ; price h14 avg ; price h15 avg ; price h16 avg ; price h17 avg ; price h18 avg ; 152 price h19 avg ; price h20 avg ] ; 153 154 pricePV avg byhour =[ price h6 PV avg ; price h7 PV avg ; price h8 PV avg ; price h9 PV avg ; price h10 PV avg ; price h11 PV avg ; 155 price h12 PV avg ; price h13 PV avg ; price h14 PV avg ; price h15 PV avg ; price h16 PV avg ; price h17 PV avg ; price h18 PV avg ; 156 price h19 PV avg ; price h20 PV avg ] ; 157 158 159 byhour combined=[ price avg byhour ( : ) , pricePV avg byhour ( : ) ] ; 160 161 hours =6:20; 162 % Calculate Impacts 163 impact byhour abs=(price avg byhour −pricePV avg byhour) ; 164 impact byhour perc=(impact byhour abs ./ price avg byhour ) ∗100; 165 166 %Display the r e s u l t s 167 f i g u r e 168 169 mynewgreen = [1 49 8 ] ./ 255; 170 subplot ( 2 , 2 , [ 1 , 2 ] ) ; %f i r s t bar chart 171 hb1=bar( hours , byhour combined , 0 . 8 , ’grouped ’) ; 172 hb1 (1) . FaceColor = ’ [ 0 . 1 0.5 0 . 2 ] ’ ; 173 hb1 (2) . FaceColor = mytextorange ; 174 xlabel(’ Hour of the Day (h) ’ ) 175 ylabel(’ E l e c t r i c i t y Price ( Eur/MWh) ’ ) 176 177 title(’Average −by Hour ’ ) 178 text( hours , byhour combined ( : , 1 ) , num2str ( byhour combined ( : , 1 ) , ’ %0.2 f ’ ) , ’ HorizontalAlignment ’ ,’ r ig h t ’ ,’VerticalAlignment ’ ,’ bottom ’ ,’ c o l or ’ , mynewgreen) ; 179 text( hours , byhour combined ( : , 2 ) , num2str ( byhour combined ( : , 2 ) , ’ %0.2 f ’ ) , ’ HorizontalAlignment ’ ,’ l e f t ’ ,’VerticalAlignment ’ ,’ bottom ’ ,’ c o l or ’ , mytextorange) ; 180 legend1dec = legend (hb1 ( : ) , ’ Average pr ic e before simulations ’ ,’ Average p ri ce a f t e r s imulations ’ ) ; 181 mynewblue = [11 36 89] . / 255; 182 subplot (2 ,2 ,3) ; %second bar chart 183 hb2=bar( hours , impact byhour abs , ’FaceColor ’ , mynewblue ) ; 184 xlabel(’ Hour of the Day (h) ’ ) 185 ylabel(’ Absolute Impact ( Eur/MWh) ’ )% l e f t y−axis 186 text( hours , impact byhour abs , num2str ( impact byhour abs , ’ %0.2 f ’ ) , ’ HorizontalAlignment ’ ,’center ’ ,’VerticalAlignment ’ ,’bottom ’ ,’ co l or ’ , mynewblue ) ; 187 legend2dec=legend ( hb2 , ’ Absolute impact ’ ) ; 188 ax = gca ;% current axes 189 ax . YLim = [ 0 0 . 9 ] ; 85 190 191 mynewred = [163 51 40] ./ 255; 192 subplot (2 ,2 ,4) ; %Third bar chart 193 hb3=bar( hours , impact byhour perc , ’ FaceColor ’ ,mynewred) ; 194 xlabel(’ Hour of the Day (h) ’ ) 195 ylabel(’ Relative Impact (%) ’ ) ; 196 text( hours , impact byhour perc , num2str(impact byhour perc , ’ %0.2 f ’ ) , ’ HorizontalAlignment ’ ,’center ’ ,’VerticalAlignment ’ ,’ bottom ’ ,’ co l or ’ ,mynewred) ; 197 legend3dec=legend ( hb3 , ’ Relative impact ’ ) ; 198 199 %% Average pr ic e per month/day 200 price avg bymonth =[ j a n p r i c e a v g ; f e b p r i c e a v g ; mar price av g ; apr price avg ; 201 may price avg ; j un pr ice av g ; j u l p r i c e a v g ; aug price avg ; 202 s e p p r i ce a v g ; o c t p r i c e a v g ; nov p rice avg ; dec pri ce av g ] ; 203 204 pricePV avg bymonth=[ jan pricePV avg ; feb pricePV avg ; mar pricePV avg ; apr pricePV avg ; 205 may pricePV avg ; jun pricePV avg ; jul pricePV avg ; aug pricePV avg ; 206 sep pricePV avg ; oct pricePV avg ; nov pricePV avg ; dec pricePV avg ] ; 207 208 bymonth combined=[ price avg bymonth ( : ) , pricePV avg bymonth ( : ) ] ; 209 months = [ 1 : 1 2 ] ; 210 %Calculate Impacts 211 impact bymonth abs=(price avg bymonth−pricePV avg bymonth ) ; 212 impact bymonth perc=(impact bymonth abs ./ price avg bymonth ) ∗100; 213 214 %Display Results 215 f i g u r e 216 217 mynewgreen = [1 49 8 ] ./ 255; 218 subplot ( 2 , 2 , [ 1 , 2 ] ) ; %f i r s t bar chart 219 hb1=bar( months , bymonth combined , 0 . 8 , ’grouped ’) ; 220 hb1 (1) . FaceColor = ’ [ 0 . 1 0.5 0 . 2 ] ’ ; 221 hb1 (2) . FaceColor = mytextorange ; 222 xlabel(’Month ’ ) 223 ylabel(’ E l e c t r i c i t y Price ( Eur/MWh) ’ ) 224 225 title(’Average −by Month ’ ) 226 text( months , bymonth combined ( : , 1 ) , num2str (bymonth combined (: ,1) , ’ %0.2 f ’ ) , ’ HorizontalAlignment ’ ,’ r i g h t ’ ,’VerticalAlignment ’ ,’ bottom ’ ,’ c o l or ’ , mynewgreen) ; 227 text( months , bymonth combined ( : , 2 ) , num2str (bymonth combined (: ,2) , ’ %0.2 f ’ ) , ’ HorizontalAlignment ’ ,’ l e f t ’ ,’VerticalAlignment ’ ,’ bottom ’ ,’ c o l or ’ , mytextorange) ; 228 legend1dec = legend (hb1 ( : ) , ’ Average pr ic e before simulations ’ ,’ 86 Average p ri ce a f t e r s imulations ’ ) ; 229 mynewblue = [11 36 89] . / 255; 230 set(gca ,’xtick ’ ,1:12 ,... 231 ’ x t i c k l a b e l ’ ,{’ Jan 12 th ’ ,’ Feb 15 th ’ ,’Mar 15 th ’ ,’ Apr 11 th ’ ,’May 16 th ’ ,’ Jun 23 rd ’ ,’ Jul 14 th ’ ,’Aug 15 th ’ ,’ Sep 15 th ’ ,’ Oct 14 th ’ ,’Nov 15 th ’ ,’ Dec 15 th ’ }) ; 232 ax = gca ;% current axes 233 ax . YLim = [ 0 6 5 ] ; 234 subplot (2 ,2 ,3) ; %second bar chart 235 hb2=bar( months , impact bymonth abs , ’ FaceColor ’ , mynewblue ) ; 236 xlabel(’Month ’ ) 237 ylabel(’ Absolute Impact ( Eur/MWh) ’ )% l e f t y−axis 238 text( months , impact bymonth abs , num2str ( impact bymonth abs , ’ %0.2 f ’) , ’ HorizontalAlignment ’ ,’center ’ ,’VerticalAlignment ’ ,’bottom ’,’ co l or ’ , mynewblue ) ; 239 legend2dec=legend ( hb2 , ’ Absolute impact ’ ) ; 240 set(gca ,’xtick ’ ,1:12 ,... 241 ’ x t i c k l a b e l ’ ,{’ Jan ’ ,’ Feb ’ ,’Mar ’ ,’Apr ’ ,’May ’ ,’ Jun ’ ,’ Jul ’ ,’Aug ’ ,’ Sep ’ ,’ Oct ’ ,’Nov ’ ,’ Dec ’ }) ; 242 ax = gca ;% current axes 243 ax . YLim = [ 0 1 . 0 5 ] ; 244 245 mynewred = [163 51 40] ./ 255; 246 subplot (2 ,2 ,4) ; %Third bar chart 247 hb3=bar( months , impact bymonth perc , ’ FaceColor ’ ,mynewred) ; 248 xlabel(’Month ’ ) 249 ylabel(’ Relative Impact (%) ’ ) ; 250 text( months , impact bymonth perc , num2str (impact bymonth perc , ’ %0.2 f ’ ) , ’ HorizontalAlignment ’ ,’center ’ ,’VerticalAlignment ’ ,’ bottom ’ ,’ c o l or ’ ,mynewred) ; 251 legend3dec=legend ( hb3 , ’ Relative impact ’ ) ; 252 ax = gca ;% current axes 253 ax . YLim = [ 0 1 0 ] ; 254 set(gca ,’xtick ’ ,1:12 ,... 255 ’ x t i c k l a b e l ’ ,{’ Jan ’ ,’ Feb ’ ,’Mar ’ ,’Apr ’ ,’May ’ ,’ Jun ’ ,’ Jul ’ ,’Aug ’ ,’ Sep ’ ,’ Oct ’ ,’Nov ’ ,’ Dec ’ }) ; 256 %% All impacts mean 257 all impacts sum=sum( jan impact abs )+sum( feb impact abs )+sum( mar impact abs )+sum(apr impact abs)+sum( may impact abs )+sum( jun impact abs)+sum( j u l i mpact ab s )+sum( aug impact abs )+sum( sep impact abs )+sum( oct impact abs )+sum(nov impact abs)+sum( dec impact abs) ; 258 al l i mp act s c ou nt=numel ( jan impact abs )+numel ( feb impact abs )+ numel ( mar impact abs )+numel ( apr impact abs )+numel ( may impact abs )+numel ( jun impact abs )+numel ( j ul i mpact ab s ) . . . 259 +numel ( aug impact abs )+numel ( sep impact abs )+numel ( oct impact abs )+numel ( nov impact abs )+numel ( dec impact abs ) ; 260 all impacts mean=all impacts sum / a ll i mp act s co un t ; 87