Impact on Electricity Consumption and Market Pricing of Energy and Ancillary Services During Pandemic of COVID-19 in Italy

Article Impact on Electricity Consumption and Market Pricing of Energy and Ancillary Services during Pandemic of COVID-19 in Italy

Emilio Ghiani * , Marco Galici, Mario Mureddu and Fabrizio Pilo

Department of Electrical and Electronic Engineering, University of Cagliari, 09123 Cagliari, Italy; [email protected] (M.G.); [email protected] (M.M.); [email protected] (F.P.) * Correspondence: [email protected]

Received: 24 April 2020; Accepted: 19 June 2020; Published: 1 July 2020

Abstract: At the moment of writing, in Italy, there is an ongoing pandemic of coronavirus disease 2019 (COVID-19), caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). Its outbreak is leading to severe global socioeconomic disruptions impacting on all economic sectors from tourism, industry and the tertiary sector, up to the operational and opening of public offices, the closure of schools and the organization of families. Measures adopted by the Italian government to deal with the COVID-19 emergency have had direct effects both on people’s daily lives and on the activity of most industrial and commercial production companies. These changes have been unequivocally reflected also on the Italian electricity system, which has shown unprecedented behavior in terms of both energy consumption and volume—and subsequently, in the observed share of renewable and conventional production technologies. The goal of this study is to show the impact on the power industry of all the restrictions and lockdown of the activities in Italy and to discuss the effects of COVID-19 outbreak on the bulk power system and the entire electricity sector. In particular, the consequences on load profiles, electricity consumption and market prices in Italy, including the environmental aspects, are examined.

Keywords: pandemic COVID-19; bulk power system; electricity consumption; load proﬁles; day-ahead market

1. Introduction The coronavirus SARS-CoV-2 emerged in Wuhan, a city of 11 million people in China’s Hubei province in late 2019. Cases of the COVID-19 disease grew by several thousand per day in China in late January and February 2020 [1] since then, the outbreak evolved in a pandemic centered in Italy, with major outbreaks in South Korea, Iran and the US. In particular, the northern Italian regions of Lombardia, Veneto and Emilia-Romagna have been most affected by the outbreak, with thousands of infected patients and deaths. In an attempt to slow down the infection, the Italian government started a nationwide lockdown on 8th March 2020, with huge economic repercussions on the country. During the lockdown, the government imposed countrywide precautionary measures. This involved not only the forced quarantine for the population, closing of schools and churches, shops, bar, restaurants, retail shops, but also shut down of national and international flights, closure of non-essential ports and airports and non-essential firms and industries. This caused a reduction in energy consumption, leading to a decrease in the overall demand in the energy market. The COVID-19 measures in Italy quickly started to cause a noticeable impact on energy consumption, which reflected in daily profiles and market prices. Notably, the forced closure of industries and tertiary activities highly flattened the working hour consumption curves in comparison

Energies 2020, 13, 3357; doi:10.3390/en13133357 www.mdpi.com/journal/energies Energies 2020, 13, 3357 2 of 19 with the daily pre-lockdown ones. Although many people started home working through smart working, the increasing of residential consumption did not compensate for the general reduction of demand. As a result, the daily consumption curve is now less pronounced than a typical spring day. In the morning, electricity consumption ramps up slower, since there is no morning commute. The evening peak is still observable, as people prepare meals, turn lights on and spend time watching TV and playing games—but it is much lower than would normally be expected. The COVID-19 outbreak on a global scale is deeply influencing the whole energy industry, causing an overall reduction of oil and natural gas demand which, coupled with strategic tensions among the leading producing nations, led to an unprecedented reduction of their selling prices. BRENT oil price plummeted about 59% from the last peak reached on 17 of February (40 days before the writing of this study) to a minimum value of USD 24 per barrel, an all-time minimum since 2002. Similarly, natural gas prices fell to USD 1.63 per million metric British thermal units (mmBtu), showing a relative decrease of 38% from the November 2019 peak, reaching its minimum since November 1995. The combination of both local and global effects caused a strong reduction in electricity demand prices, which is going to influence the entire electricity supply chain in the short and middle term [2]. In addition, the IEA reports forecasted a global yearly total demand decrease of 6%, especially hitting the production from non-renewable energy sources [3]. Europe, moreover, showed a consumption decrease of roughly 10% during the final weeks of March and the first week of April [4]. This decrease has been observed to be particularly severe in Italy, which reportedly reached a decrease in consumption of about 23% [4] as described also in the IAEE Energy Forum [5] which analyses worldwide the impacts of COVID-19 on the energy industries in term of demand drops, supply-side shocks, facilities shutdowns and new patterns of electricity demand. Due to the aforementioned reasons, the effects associated with the COVID-2019 outbreak are difficult to assess with the traditional evaluation of the performance of critical infrastructures, based on classical risk assessment methods, which consider credible and predictable hazards and threats [6]. The ubiquitous effects on the power system should be indeed studied from a resilience perspective. Resilience definitions and frameworks have been proposed in different disciplines and applied in a variety of case studies worldwide [7]. In most resiliency studies, energy systems are subject to major disruptions affecting economic activities, operation of infrastructure and the society as a whole [8]. Even though, in this case, the resiliency issues arise from a crisis that has occurred in civil society and impacted the operation of the Italian power system and its related energy market prices. Thus, the data relative to the COVID-19 outbreak period will be of vital importance for testing the reliability of the currently employed planning, scenario-making and resilience testing methodologies. In particular, accurate analysis of the current situation will allow a better estimation of scenarios with higher penetration of renewable energy sources (RES) and their effects on the future power industry. In this context, the goal of this study is to provide detailed and comprehensive information related to the alterations on load profiles, electricity demand and both wholesale and ancillary services market prices in Italy due to the restrictions and lockdown of the activities, providing powerful insights about the resilience of the entire electricity sector for a prolonged period of low consumption/production ratio and posing issues which can serve as inspiration for future research.

2. Generation and Consumption in Italy before the COVID-19 Outbreak

2.1. Generation The Italian electrical system is the energy backbone of the third-largest economy of the European Union. In 2018, the Italian electricity system had a peak consumption of around 60 GW and a generation capacity of 115.20 GW [9]. As depicted in Figure1, the generation capacity in Italy consists mainly of thermal generation (coal, gas, oil) and renewable energy sources (RES); nuclear production is forbidden by law since 1987 and about 15% of the electricity is imported from abroad. Energies 2020, 13, 3357 3 of 19

RES consolidated their leading role in the Italian energy system, conﬁrming themselves as a Energiesdecisive 2020, 1 factor3, x FOR for PEER the REVIEW sustainable development of the country in the energy transition. According3 of 22 to Italian national energy strategy (NES) [10], the integrated national energy and climate plan [11] and Italianthe nationalnational electricityenergy strategy transmission (NES) [10], grid the development integrated plannational [12], energy RES capacity and climate is expected plan [11] to grow and up theto national 93 GW electricity in 2030, with transmission almost 40 grid additional development GW with plan respect [12], RES to 2017. capacity The is main expected contribution to grow will up be to 93from GW solar in 2030, photovoltaic with almost (PV), 40 with additional almost GW+30 with GW, followedrespect to by 2017. wind, The with main almost contribution+9 GW. will be from solar photovoltaic (PV), with almost +30 GW, followed by wind, with almost +9 GW. 140

120

100

Capacity [GW] Capacity 40

0 Thermal Hydro Photovoltaic Wind Geothermal Installed capacity

FigureFigure 1. Italian 1. Italian power power capacity. capacity. The actual gross electricity production from RES in summarized in Table1[ 13]. Thermal generation The actual gross electricity production from RES in summarized in Table 1 [13]. Thermal is fueled by natural gas sources (72.6%); whereas coal and oil account for 20.9% and 6.5%, respectively. generation is fueled by natural gas sources (72.6%); whereas coal and oil account for 20.9% and 6.5%, respectively. Table 1. Gross electricity production (TWh) from renewable energy sources (RES) in Italy compared to total load. Table 1. Gross electricity production (TWh) from renewable energy sources (RES) in Italy compared to total load. Source 2015 2016 2017 2018 2019 SourceHydro 2015 46,436 2016 43,771 2017 37,545 2018 49,918 2019 47,063 Wind 14,705 17,522 17,565 17,556 19,992 Hydro 46,436 43,771 37,545 49,918 47,063 Solar 22,438 21,629 23,864 22,121 24,138 GeothermalWind 14,705 5814 17,522 5865 17,565 5784 17,556 5707 19,992 5688 SolarBiomasses 22,438 17,180 21,629 17,212 23,864 17,010 22,121 16,896 24,138 16,881 GeothermalTot RES 5814 106,573 5865 105,999 5784 101,768 5707 112,198 5688 113,762 BiomassesThermal 17,180 162,55 17,212 168,906 17,010 177,497 16,896 160,822 16,881 163,558 Tot LoadRES 106,573 316,95 105,999 314,3 101,768 320,578 112,198 321,500 113,762 319,552 Thermal% RES 162,55 33.6 168,906 33.7 177,497 31.7 160,822 34.9 163,558 35.6 Load 316,95 314,3 320,578 321,500 319,552 2.2. Load Proﬁle of the% ItalianRES Power33 System.6 33.7 31.7 34.9 35.6

2.2. LoadFigure Profile2 ofshows the Italian the di Powerfferent System contributions of consumption for each sector in Italy during the year 2018. The main consumption comes from the residential load (39%). The secondary (industrial) and tertiaryFigure (services)2 shows the sectors different have 37%contributions and 22% of totalconsumption load, respectively, for each sector whereas in Italy the primary during the sector year(agriculture, 2018. The main raw materials)consumption account comes for from 2% ofthe total residential demand load [9]. (39%). The secondary (industrial) and tertiaryThe load(services) profile sectors is the shapehave 37% of the and electrical 22% of loadtotal overload, time. respect Theively, load whereas profile forthe a primary given area sectorvaries (agriculture, considerably raw according materials) to account the categories for 2% of of users, total demand including [9]. commercial, residential or industrial users. Peak electricity demands also vary according to seasonal and holiday needs. A typical Italian

Energies 2020, 13, x FOR PEER REVIEW 4 of 22

Energies 2020, 13, 3357 4 of 19 140

120 load proﬁle in 2019, for winter (January) and summer (July) weekdays and weekends, is shown in Figure3. Electricity100 consumption normally follows a regular daily shape; quick rising as people wake up and go to school or work, smoothening throughout the day and then peaking at dinner time when people get back home.80 In Italy, the ﬂattening of the load curve during the daytime, as shown in Figure3b, is exacerbated in summer by the production from PV distributed generation in distribution 60 Energiesnetworks 2020, [ 1413,, x15 FOR]. PEER REVIEW 4 of 22 40 140

Energy Consumption [TWh] 20 120 0 100 Primary Sector Secondary Tertiary Sector Residential Sector Sector 80 Figure 2. Subdivision among sectors of the energy consumed in Italy. 60 The load profile is the shape of the electrical load over time. The load profile for a given area varies considerably40 according to the categories of users, including commercial, residential or industrial users. Peak electricity demands also vary according to seasonal and holiday needs. A

typical ItalianEnergy Consumption [TWh] load20 profile in 2019, for winter (January) and summer (July) weekdays and weekends, is shown in Figure 3. Electricity consumption normally follows a regular daily shape; quick rising as people wake up 0and go to school or work, smoothening throughout the day and then peaking at dinner time when peoplePrimary get back Sector home. SecondaryIn Italy, the flatteningTertiary of Sectorthe load curveResidential during the daytime, as shown in Figure 3b, is exacerbated in summerSector by the production from PV distributedSector generation in distribution networks [14,15]. FigureFigure 2. 2. SubdivisionSubdivision among among sectors sectors of of the the energy energy consumed consumed in in Italy Italy..

60 60 The load profile is the shape of the electrical load over time. The load profile for a given area 50 varies50 considerably according to the categories of users, including commercial, residential or industrial40 users. Peak electricity demands also vary40 according to seasonal and holiday needs. A [GW] typical[GW] 30 Italian load profile in 2019, for winter (January)30 and summer (July) weekdays and weekends, Load Load is shown20 in Figure 3. Electricity consumption normally20 follows a regular daily shape; quick rising as people wake up and go to school or work, smoothening throughout the day and then peaking at 10 10 dinner time wheWorkingn people Day Winter get backHolyday home. Winter In Italy, the flattening Workingof the Day load Summer curve duringHolyday Summerthe daytime, 0 0 as shown1 2 in3 Figure4 5 6 7 3b,8 9 10is11 exacerbated12 13 14 15 16 17 18 in19 summer20 21 22 23 24 by the1 production2 3 4 5 6 7 8from9 10 11PV12 13distributed14 15 16 17 18 19 generation20 21 22 23 24 in distribution networks [14Time,15 [h] ]. Time [h] a) b) 60 60 FigureFigure 3.3. LoadLoad proﬁleprofile ( a)) Weekday/Weekend Weekday/Weekend (January/Winter) (January/Winter) (b ()b (July/Summer)) (July/Summer).. 50 50

40 OperatingOperating pre pre lockdown lockdown conditions conditions were were characterized characteri40 zed by aby peak a peak load load of about of about 60 GW 60 inGW summer in [GW] (58.81[GW] 30summer GW on(58.81 25 JulyGW 2019)on 25 and July a 2019) minimum and a load minimum of 2030 GW load in of winter 20 GW (18.27 in winter GW on (18.27 25 December GW on 2019).25

In winter, on the contrary, the peak load is 55 GW andLoad the summer minimum load is 21 GW [9,14]. Load December 2019). In winter, on the contrary, the peak load is 55 GW and the summer minimum load 20 20 is 21 GW [9,14]. 3.10 Impact on Electricity Consumption and Load Proﬁles10 during COVID-19 Working Day Winter Holyday Winter Working Day Summer Holyday Summer 03. ImpactThe lockdown on Electricity sequence Consumption consists of and three Load main Profiles dates:0 during COVID-19 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 The lockdown sequenceTime [h] consists of three main dates: Time [h] 6th March 2020: closing of all schools and universities; • • 12th6th MarchMarch 2020:2020: closing Alla) territory of all schools of Italy and becomes universities a “red; zone”. Non-essentialb) factories, companies • • and12th publicMarch buildings 2020: All areterritory closed of (partial Italy becomes shutdown); a “red zone”. Non-essential factories, companies Figure 3. Load profile (a) Weekday/Weekend (January/Winter) (b) (July/Summer). 21stand March public 2020: buildings airports are closed and seaports (partial closedshutdown); to passenger traﬃc (shutdown). • OperatingThe forced pre closure lockdown of public conditions buildings, were industries characteri and tertiaryzed by activities a peak duringload of COVID-19 about 60 outbreakGW in summerrestrictions (58.81 highly GW reduced on 25 theJuly consumption 2019) and a curves minimum during load the of daily 20 workingGW in winter hours and(18.27 in theGW evening. on 25 December 2019). In winter, on the contrary, the peak load is 55 GW and the summer minimum load is 21 GW [9,14].

3. Impact on Electricity Consumption and Load Profiles during COVID-19 The lockdown sequence consists of three main dates: • 6th March 2020: closing of all schools and universities; • 12th March 2020: All territory of Italy becomes a “red zone”. Non-essential factories, companies and public buildings are closed (partial shutdown);

Energies 2020, 13, x FOR PEER REVIEW 5 of 22

• 21st March 2020: airports and seaports closed to passenger traffic (shutdown).

EnergiesThe2020 ,forced13, 3357 closure of public buildings, industries and tertiary activities during COVID-195 of 19 outbreak restrictions highly reduced the consumption curves during the daily working hours and in the evening. In Figure 4, the daily load profile of the days 6th and 21st (for 2020, 23rd, the first Inavailable Figure4 ,Monday) the daily loadMarch profile is compared of the days for 6th the and last 21st three (for years, 2020, 23rd,although the firstneglecting available weather Monday) Marchdifferences is compared and further for the related last three economic years, issues, although a consistent neglecting reduction weather of di thefferences consumption and further has been related economicregistered. issues, As a result a consistent of the lockdown, reduction the of thedaily consumption consumption hascurve been changed registered. their typical As a result pattern. of the lockdown,In the morning, the daily electricity consumption consumption curve changedramps up their more typical slowly pattern. and the In evening the morning, peaks remain electricity consumptionsustained as rampspeople up prepare more slowlymeals and and spend the evening time with peaks their remain families sustained watching as peopleTV, using prepare electric meals andappliances spend timeand withelectronic their devices; families these watching peaks TV, are using anyway electric lower appliances than one andwould electronic normally devices; be theseexpected. peaks are anyway lower than one would normally be expected.

FigureFigure 4. 4. ItalianItalian Load profile profile comparison comparison 6th 6th March March (a ()a and) and 20–21 20–21 March March (b) ( b2018,) 2018, 2019, 2019, 2020. 2020. The evening peak is reached with a much greater gradient than in the past, putting under stress The evening peak is reached with a much greater gradient than in the past, putting under stress the regulation capabilities and flexibility of thermoelectric power plants that are working close to the the regulation capabilities and flexibility of thermoelectric power plants that are working close to the minimum, especially if the regulating capacity of hydroelectric plants (impoundment facility and minimum, especially if the regulating capacity of hydroelectric plants (impoundment facility and pumpingpumping storage) storage) shouldshould not be sufficient. sufficient. However, However, the the electric electric system system must must be beable able to manage to manage at at anyany time time such such changeschanges inin loadload in a range of of values values that, that, in in 2018, 2018, fluctuated fluctuated between between a minimum a minimum of of 20.220.2 GW GW and and a a maximummaximum of 57.8 GW. GW. To To assess assess that that issue issue the the residual residual load, load, defined defined as the as thedifference difference betweenbetween current current load load and and production production fromfrom renewablerenewable sources, has assumed remarkable remarkable importance importance in thein lastthe last years, years, due due to to the the increase increase of of the the renewable renewable resources, in in particular particular with with the the high high penetration penetration ofof PV PV plants. plants. TheThe residualresidual load corresponds to to the the actual actual load load that that must must be becovered covered by bydispatchable dispatchable generationgeneration plants.plants. Besides,Besides, the shape of of the the residual residual load load has has evolved evolved in inthe the last last years, years, even even more more didifferentiatingfferentiating from from the the pattern pattern of theof wholethe whole electrical elec profile,trical profile, and this and phenomenon this phenomenon will be increasinglywill be evidentincreasingly in the evident forecasted in the future forecasted scenarios future due scenarios to the expecteddue to the increase expected in increase PV generation. in PV generation. In the days ofIn this the observation, days of this theobservation, residual loadthe residual curve is load characterized curve is characterized by the typical by duckthe typical curve, duck with curve, relevant variationswith relevant during variations daytime during and in thedaytime steepness and ofin thethe eveningsteepness ramp of the due evening to the simultaneous ramp due to increase the insimultaneous demand and increase reduction in demand in renewable and reduction energy in production, renewable energy which determinesproduction, which the need determines for a rapid increasethe need in for production a rapid increase from programmable in production from sources. programmable The RES growth, sources. occurred The RES in growth, the Italian occurred territory, hasin the shown Italian to territory, be often has not shown consistent to be withoftenthe not pointsconsistent of majorwith the consumption. points of major On consumption. the other hand, On the other hand, especially for PV, the installation of the plants led to increasing network especially for PV, the installation of the plants led to increasing network congestion situations, due to congestion situations, due to the asymmetry in their installation location (most of the production is the asymmetry in their installation location (most of the production is located in South Italy for the located in South Italy for the solar energy availability). solar energy availability). In Figure 5 the variation of the peak in the demand is considered for March 2020 and the first In Figure5 the variation of the peak in the demand is considered for March 2020 and the first week of April. The profiles show how increasing lockdown measures caused a sustained decrease in weeknational of April. energy The demand. profiles In showparticular how, from increasing the last lockdown weeks of March, measures a decrease caused in a load sustained was observed, decrease in nationalfollowing energy the consecutive demand. In closure particular, of production from the facilities. last weeks In ofFigure March, 6, the a decrease peak variation in load iswas compared observed, followingfor the last the three consecutive years. To closurebetter clarify of production the phenom facilities.ena, in Figure In Figure 7 is 6reported, the peak a comparison variation is between compared for2018, the last2019 three and years.2020 Toof betterthe peak clarify variation the phenomena, profiles. Figure in Figure 7 7shows is reported an evident a comparison reduction between in 2018,consumption, 2019 and 2020which of reached the peak at variation least 25% profiles. less consumption Figure7 shows than the an evidentaverage reductionin the previous in consumption, year. which reached at least 25% less consumption than the average in the previous year. The analysis was brought to a higher level of detail, to evaluate the impacts of the health emergency, on all the areas of Italy. In particular, the profiles on the northern part of Italy was evaluated. The northern part of Italy, composed by Valle d’Aosta, Piemonte, Liguria, Emilia-Romagna, Lombardia, Veneto, Friuli Venezia Giulia and Trentino Alto Adige is the main productive zone of the nation, accounting for about half of the national gross domestic product. Figure8 shows the load profiles Energies 2020, 13, 3357 6 of 19 in the North of Italy. As shown in Figure8, their trends reflect those of the whole nation, describing a lowering of the load by almost half compared to normal conditions. The northern part of Italy is characterized by the highest penetration of industrial facilities and the impact on consumption of the lockdown was quite remarkable. Energies 2020, 13, x FOR PEER REVIEW 6 of 22

Pre - lockdown First Days lockdown Partial lockdown Full lockdown Full lockdown th st 1st – 7th March 8th – 14th March 15th – 21st March 22nd – 29th March 30th March – 5th April 47,09 GW 45,30 GW 47,09 GW 45,30 GW 40,74 GW 37,24 GW 37,60 GW @ 4th March @ 11th March th th st @ 4 March @ 11 March @ 18th March @ 25th March @ 1st April

-4,3% -15,1% -22,4%

FigureFigure 5. 5.Five-week-load Five-week-load profile proﬁle,, 1st 1st March March–5th–5th April April 2020 2020..

FigureFigure 6. Five-week-load6. Five-week-loadproﬁle profile comparison between between 2018, 2018, 2019 2019 and and 2020. 2020.

Figure 7. Five-week-peak comparison between 2018, 2019 and 2020.

Energies 2020, 13, x FOR PEER REVIEW 7 of 22

The analysis was brought to a higher level of detail, to evaluate the impacts of the health emergency, on all the areas of Italy. In particular, the profiles on the northern part of Italy was evaluated. The northern part of Italy, composed by Valle d’Aosta, Piemonte, Liguria, Emilia- Romagna, Lombardia, Veneto, Friuli Venezia Giulia and Trentino Alto Adige is the main productive zone of the nation, accounting for about half of the national gross domestic product. Figure 8 shows the load profiles in the North of Italy. As shown in figure 8, their trends reflect those of the whole nation, describing a lowering of the load by almost half compared to normal conditions. The northern

Energiespart of2020 Italy, 13 ,is 3357 characterized by the highest penetration of industrial facilities and the impact on7 of 19 consumption of the lockdown was quite remarkable.

Figure 8. 8. NorthNorth Italy Italy load load profile profile comparison comparison (a)( 6tha) 6th March March and and (b) (21b) March 21 March 2018, 2018, 2019 2019 and 2020. and 2020. 4. Impact on Electricity Market during COVID-19 4. Impact on Electricity Market during COVID-19 4.1. Italian Electricity Market 4.1. Italian Electricity Market In Italy, the electricity market (Italian Power Exchange—IPEX) was set up as a result of Legislative In Italy, the electricity market (Italian Power Exchange—IPEX) was set up as a result of Decree no. 79 dated 16 March 1999 (“Bersani Decree”) as part of the implementation of the EU directive Legislative Decree no. 79 dated 16 March 1999 (“Bersani Decree”) as part of the implementation of onthe the EU creation directive of on an internalthe creation energy of marketan internal (Directive energy 96 market/92/EC (Directive repealed by96/92/EC Directive repealed 2003/54 by/EC). DirectiveThe Italian2003/54/EC). electricity market is divided into: - TheDay-ahead Italian electricity market (MGP); market is divided into: -- Day-aheadIntra-day market market (MI);(MGP); -- Intra-dayDispatching market services (MI); market (MSD). - Dispatching services market (MSD). In the MGP and MI—also referred to as energy markets—producers, wholesalers and end In the MGP and MI—also referred to as energy markets—producers, wholesalers and end customers, as well as Acquirente Unico (AU), responsible for acquiring energy for captive customers customers, as well as Acquirente Unico (AU), responsible for acquiring energy for captive customers and Gestore dei Servizi Energetici (GSE), responsible for selling energy for renewable producers, and Gestore dei Servizi Energetici (GSE), responsible for selling energy for renewable producers, buy buy and sell wholesale quantities of electricity for the next day. These markets, which are managed by and sell wholesale quantities of electricity for the next day. These markets, which are managed by GestoreGestore dei Mercati Mercati Energetici Energetici (GME), (GME), define define system system marginal marginal prices prices at which at which the energy the energy is traded. is traded. InIn thethe MSD,MSD, the the transmission transmission system system operator operator (TSO), (TSO), Terna, Terna, procures procures the the resources resources it needs it needs to manage to andmanage control and the control system the (solving system (solving intranational intranationa congestions,l congestions, creating creating energy energy reserves, reserves, real-time real-time balancing). Givenbalancing). that transportGiven that capacity transport limits capacity exist, limits the Italian exist, the power Italian system power is subdividedsystem is subdivided into market into zones andmarket are zones cleared and through are cleared the localthrough market the local algorithm. market algorithm. The number The of number constrained of constrained zones or zones “point of limitedor “point production”, of limited production”, are zones whose are zones maximum whose maximum generation generation exportable exportable to the rest to of the the rest grid of isthe lower thangrid is the lower maximum than the possible maximum generation possible generation owing to owing insuffi tocient insufficient Transmission Transmission capacity capacity [14]. Currently, [14]. theCurrently, zones are the aszones follows are as (Figure follows9): (Figure 9): • Geographical zone: zone: representing representing a aportion portion of ofthe the national national grid. grid. Geographical Geographical zones zones are northern are northern • ItalyItaly (NORD),(NORD), central-northerncentral-northern ItalyItaly (CNOR), (CNOR), central-southern central-southern Italy Italy (CSUD), (CSUD), southern southern Italy Italy (SUD), Sicilia(SUD),(SICI), Sicilia (SICI), Sardegna Sardegna (SARD); (SARD); • National virtual zone: constrained zone (“point or pole of limited production”). It includes National virtual zone: constrained zone (“point or pole of limited production”). It includes • Rossano (ROSN); Rossano (ROSN);

Foreign virtual zone: point of interconnection with neighboring countries. It includes France • (FRAN), Switzerland (SVIZ), Austria (AUST), Slovenia (SLOV), Slovenia coupling representing the interconnection dedicated to the market coupling between Italy and Slovenia (BSP); Corsica (CORS), Corsica AC (COAC), Greece (GREC), France coupling (XFRA), Austria coupling (XAUS), Malta (MALT), Switzerland coupling (XSVI) and Montenegro (MONT). Market zone: aggregation of geographical and/or virtual zones such that the flows between the • same zones are lower than the transmission limits notified by the TSO. This aggregation is defined on an hourly basis as a result of the resolution of the day-ahead market and intra-day market. In the same hour, different market zones may have different zonal prices. Energies 2020, 13, x FOR PEER REVIEW 8 of 22

• Foreign virtual zone: point of interconnection with neighboring countries. It includes France (FRAN), Switzerland (SVIZ), Austria (AUST), Slovenia (SLOV), Slovenia coupling representing the interconnection dedicated to the market coupling between Italy and Slovenia (BSP); Corsica (CORS), Corsica AC (COAC), Greece (GREC), France coupling (XFRA), Austria coupling (XAUS), Malta (MALT), Switzerland coupling (XSVI) and Montenegro (MONT). • Market zone: aggregation of geographical and/or virtual zones such that the flows between the same zones are lower than the transmission limits notified by the TSO. This aggregation is defined on an hourly basis as a result of the resolution of the day-ahead market and intra-day Energies 2020, 13, 3357 8 of 19 market. In the same hour, different market zones may have different zonal prices.

FigureFigure 9. 9.Italian Italian load load subdivision subdivision zones zones for for the the Italian Italian Power Power Exchange Exchange (IPEX). (IPEX). The day-ahead market algorithm runs the first time at the national level. If the capacity limits The day-ahead market algorithm runs the first time at the national level. If the capacity limits among zones are not reached, the outcome is an equal price for all the zones, otherwise, the market is among zones are not reached, the outcome is an equal price for all the zones, otherwise, the market split into one or more zones. All supply offers and demand bids are valued at the marginal clearing is split into one or more zones. All supply offers and demand bids are valued at the marginal clearing price of the zone to which they belong. This price is determined, for each hour, by the intersection of price of the zone to which they belong. This price is determined, for each hour, by the intersection of the zonal demand and supply curves and is different from one zone to another zone when transmission the zonal demand and supply curves and is different from one zone to another zone when capacity limits are saturated. The accepted demand bids pertaining to consuming units belonging to transmission capacity limits are saturated. The accepted demand bids pertaining to consuming units Italian geographical zones are valued at the “Prezzo Unico Nazionale” (PUN—national single price); belonging to Italian geographical zones are valued at the “Prezzo Unico Nazionale” (PUN—national this price is the weighted mean of the prices of the geographical zones, weighted for the quantities single price); this price is the weighted mean of the prices of the geographical zones, weighted for the purchased in these zones [14]. quantities purchased in these zones [14]. Commonly, Italy shows significant differences among the market prices of different zones. This is Commonly, Italy shows significant differences among the market prices of different zones. This due to bottlenecks in the north-south transmission lines that are unable to manage the regional is due to bottlenecks in the north-south transmission lines that are unable to manage the regional imbalance of generation and demand between the two zones. Currently, plans have been made to imbalance of generation and demand between the two zones. Currently, plans have been made to strengthen transmission lines and to achieve further transmission capacity to accommodate higher strengthen transmission lines and to achieve further transmission capacity to accommodate higher renewables according to the national electricity transmission grid development plan [12] and ENTSO-E renewables according to the national electricity transmission grid development plan [12] and TSO cooperation regarding the network development and long-term planning defined in Ten-Year ENTSO-E TSO cooperation regarding the network development and long-term planning defined in Network Development Plan (TYNDP) [16]. For instance, the construction of three main high voltage Ten-Year Network Development Plan (TYNDP) [16]. For instance, the construction of three main high direct current (HVDC) lines is being considered, which will connect south and north of Italy through voltage direct current (HVDC) lines is being considered, which will connect south and north of Italy the mainEnergies Islands 2020 of, 13 Sardinia, x FOR PEER and REVIEW Sicily (Figure 10)[17]. 9 of 22 through the main Islands of Sardinia and Sicily (Figure 10) [17].

HV DC SA.CO.I.3

HV DC SA.PE.I

HV DC Tyrrhenianlink

Figure 10.FigureThe 10. option The option of building of building a new a new HVDC HVDC Tyrrhenian Tyrrhenian link link to to Sicily Sicily/Sardinia/Sardinia—the—the year 2030 2030..

Besides, the growth of renewable power has forced down wholesale electricity prices given that the clearing power price is determined by the “merit order”—the sequence in which power producers contribute power to the market, with the cheapest offer made by the producer. Power from renewable installations such as wind and PV is offered at null cost (price takers). That means they lower the entrance price and push more expensive conventional producers down in the merit order list (Figure 11).

Pr i ce (€/ M Wh) Peaking Peaking gener ati on

Turbogas Oil gener ati on Inter m ed i ate Base gener ati on

Gas Coal Renew abl e gener ati on Import Wind, Solar, H ydro

Quantity(MWh)

Figure 11. Merit order sequence for bidding in the IPEX market.

The growth of renewable sources has led to a significant decrease in the hours of use of thermal plants with the highest production costs and a significant reduction in the PUN. As can be seen in Figure 12, high quantities of RES lead to low prices of energy in the market, specially recorded in the summer months.

Energies 2020, 13, x FOR PEER REVIEW 9 of 22

Energies 2020 13 , , 3357 9 of 19 Figure 10. The option of building a new HVDC Tyrrhenian link to Sicily/Sardinia—the year 2030. Besides, the growth of renewable power has forced down wholesale electricity prices given thatBesides, the clearing the growth power of price renewable is determined power has by theforced “merit down order”—the wholesale electricity sequence prices in which given power that producersthe clearing contribute power price power is determined to the market, by the with “merit the cheapest order”—the oﬀer sequence made by in the which producer. power Power producers from contributerenewable power installations to the market, such as with wind the and cheapest PV is o offerﬀered made at null by the cost producer. (price takers). Power That from means renewable they installationslower the entrance such as price wind and and push PV more is offered expensive at nu conventionalll cost (price producers takers). That down means in the they merit lower order the list (Figureentrance 11 price). and push more expensive conventional producers down in the merit order list (Figure 11).

Price (€/ MWh) Peaking Peaking generation

Turbogas Oil generation Intermediate Base generation Gas Coal Renewable generation Im port Wind, Solar, Hydro

Quantity(MWh)

Figure 11. Merit order sequence for bi biddingdding in the IPEX market.

The growth of renewable sources has led to a significant signiﬁcant decrease in the hours of use of thermal plants with the highest production costs and a sign signiﬁcantificant reduction in the PUN. As can be seen in Figure 12,12, high quantities of RES lead to low prices of energy in the market, specially recorded in the summerEnergies 2020 months., 13, x FOR PEER REVIEW 10 of 22

Price Price (€/ MWh) (€/ MWh) High High High High Demand Demand Low Low Demand Demand Low Renewable

4 3 2 1 T.gas T.gas Renewable Im port Coal Gas Oil Renewable Im port Coal Gas Oil Quantity(MWh) Quantity(MWh) a) b)

Figure 12. (a) dashed line 2: price with low demand/low RES, dashed line 4 price with high Figure 12. (a) dashed line 2: price with low demand/low RES, dashed line 4 price with high demanddemand/low/low RES, ( (bb)) dashed dashed line line 1: 1: price price with with low low demand/h demandigh/high RES RES and and dashed dashed line line4: price 4: price withwith highhigh demanddemand/high/high RES. RES. 4.2. Statistical Comparison of Pre- and Post-Lockdown Day-ahead Market Prices 4.2. Statistical Comparison of Pre- and Post-Lockdown Day-Ahead Market Prices To understand the effects of the COVID-19 lockdown, a statistical analysis was performed, To understand the effects of the COVID-19 lockdown, a statistical analysis was performed, aimingaiming toto comparecompare the different different pre pre and and post post lockdown lockdown MGP MGP prices prices distributions. distributions. FiguresFigures 13 and 1414 showshow the the boxplots boxplots of of the the rela relativetive deviation deviation from from the themean mean of pre-lockdown of pre-lockdown WD P marketmarket pricespricespre, ,in in order order to to perform perform a diff-in-diff a diff-in-di analysis.ff analysis. ℎ : the observed hourly PUN prices () were aggregated in sets of observed prices per each week of the reference year, for days between Monday and Friday (weekdays), represented by the sets { }. Then, sets of their relative differences (in percentage) from the pre-lockdown mean price { } were computed using (1), where the pre-lockdown mean price was calculated as in (2).

− = × 100 (1)

=⋃∈{ } (2) The same aggregation was performed for weekend days (Saturday and Sunday), obtaining the sets { }. These sets normalized the observed market prices concerning the average prices observed in the month before. In this way, it was possible to compare the normalized observables in pre- and post-lockdown times, and to add a comparison with the behavior of observed prices in 2019. The results of the performed analysis are shown as weekly boxplots, for weekdays and weekend days, respectively and both the year 2019 and 2020, for the sake of comparison. The figures show how the normalized prices started to drop after the lockdown (from week 12). In particular, a strong decrease was observed in the median and quartiles, for both weekdays and weekend days. The median weekday prices dropped about 40% during week 13, whereas the median drop of weekend prices (approx. 50%) was observed during the weekend of week 13. Besides, the boxplot relative to week 15 shows that 25% of the observed weekend days prices were below 70% of the pre-lockdown average. All the highlighted observations were not present in the same period of 2019.

Energies 2020, 13, 3357 10 of 19

WD Ppre have been calculated as f ollows : the observed hourly PUN prices P(t) were aggregated in sets of observed prices per each week of the reference year, for days between Monday and Friday n WDo (weekdays), represented by the sets Pw . Then, sets of their relative diﬀerences (in percentage) from n WDo the pre-lockdown mean price RDPw were computed using (1), where the pre-lockdown mean price WD Ppre was calculated as in (2). WD WD Pw Ppre WD − RDPw = 100 (1) WD × Ppre

WD n WDo Ppre = mean Uw pre lockout Pw (2) ∈ − The same aggregation was performed for weekend days (Saturday and Sunday), obtaining the sets n WEo RDPw . These sets normalized the observed market prices concerning the average prices observed in the month before. In this way, it was possible to compare the normalized observables in pre- and post-lockdown times, and to add a comparison with the behavior of observed prices in 2019. The results of the performed analysis are shown as weekly boxplots, for weekdays and weekend days, respectively and both the year 2019 and 2020, for the sake of comparison. The ﬁgures show how the normalized prices started to drop after the lockdown (from week 12). In particular, a strong decrease was observed in the median and quartiles, for both weekdays and weekend days. The median weekday prices dropped about 40% during week 13, whereas the median drop of weekend prices (approx. 50%) was observed during the weekend of week 13. Besides, the boxplot relative to week 15 shows that 25% of the observed weekend days prices were below 70% of the pre-lockdown average.

All the highlightedEnergiesEnergies 20202020,, observations1313,, xx FORFOR PEERPEER REVIEWREVIEW were not present in the same period of 2019. 1111 ofof 2222

Figure 13. Weekly boxplots of the normalized WDprices (weekday), for 2019 and 2020. Figure 13.FigureWeekly 13. Weekly boxplots boxplots of the of normalizedthe normalizedRDP w pricesprices (weekday), (weekday), for 2019 for and 2019 2020. and 2020.

Figure 14. Weekly boxplots of the normalized WEprices (weekend days), for 2019 and 2020. Figure 14. Weekly boxplots of the normalized RDPw prices (weekend days), for 2019 and 2020. 4.3. Italian Day-ahead Market Prices during the COVID-19 Outbreak During the COVID-19 outbreak, thethe ItalianItalian PUNPUN priceprice steeplysteeply dedecreasedcreased inin bothboth peakpeak andand off-off- peak hours (Figure 15). In particular, the daily peak registered on the 23rd of march displayed an average PUN reduction by 37% and a decrease of 24% of the peak price with respect of the second week of the same month, updating the negative historicalstorical recordrecord toto 26.326.3 €/MWh€/MWh [18].[18]. InIn FiguresFigures 1616 and 17 the average PUN in comparison with previous years is shown. In particular, in Figure 17, throughoutthroughout thethe MarchMarch period,period, thethe averageaverage PUNPUN reachereachess peakspeaks aboutabout 37%37% lessless thanthan inin previousprevious years.years.

Energies 2020, 13, 3357 11 of 19

4.3. Italian Day-ahead Market Prices during the COVID-19 Outbreak During the COVID-19 outbreak, the Italian PUN price steeply decreased in both peak and off-peak hours (Figure 15). In particular, the daily peak registered on the 23rd of march displayed an average PUN reduction by 37% and a decrease of 24% of the peak price with respect of the second week of the same month, updating the negative historical record to 26.3 €/MWh [18]. In Figures 16 and 17 the average PUN in comparison with previous years is shown. In particular, in Figure 17, throughout the March period, the average PUN reaches peaks about 37% less than in previous years. This behavior is due to low consumptions and high RES production, as depicted for example in Figure 18, for the 23rd March 2020 when the hourly production from RES in the middle hours overcomes the 40% of the total consumption. Figure 19 shows that the minimum values of the prices have not been modified compared to the maximum values, which instead have larger fluctuations. Energies 2020,, 13,, xx FORFOR PEERPEER REVIEWREVIEW 12 of 22

FigureFigure 15.15. NationalNational single single price price (PUN) (PUN) on onIPEX. IPEX. Dates Dates 1st March–5th 1st March–5th April April2020. 2020.

1stst –7thth March 8thth –14thth March 15thth –21stst March 22ndnd –29thth March 30thth March – 5thth April

130 2018 2019 2020 120

110

100

70 €/MWh €/MWh 60

FigureFigure 16. 16. PUNPUN comparison comparison between between 2018, 2018, 2019 2019 and 2020. and 2020.

Energies 2020, 13, x FOR PEER REVIEW 13 of 22

Energies 2020, 13, 3357 12 of 19

Figure 17. Average PUN comparison between 2018, 2019 and 2020.

This behavior is due to low consumptions and high RES production, as depicted for example in Figure 18, for the 23rd March 2020 when the hourly production from RES in the middle hours overcomes the 40% of the total consumption. Figure 19 shows that the minimum values of the prices have not been modified compared to the maximum values, which instead have larger fluctuations. Figure 17. Average PUN comparison between 2018, 2019 and 2020.

40 60%

35 50%

40% 25 % 20 30% RES

15 20% Production\Consumption [GW] Production\Consumption 10

10% 5 Total Load GW Renewable1 Generation RES %

0 0% 123456789101112131415161718192021222324

Energies 2020, 13, x FOR PEER REVIEW 14 of 22 FigureFigure 18. 18.Low Low demand demand and and highhigh RES for for the the day day of of23rd 23rd March March 2020. 2020.

Figure 19. Average, min and max (error bars) price of energy in the IPEX. Dates 1st–28th March 2020. Figure 19. Average, min and max (error bars) price of energy in the IPEX. Dates 1st–28th March 2020.

This situation hits the electricity-producing industry in two ways. First of all, the daily needed generation was cut about 40%, cutting out the high marginal costs power plants shown in Figure 12. Second, the low PUN prices highly decreased the profit for all the generation sector. This second issue can be overcome with the market dynamics since the competition will likely raise the bids placed by the thermal generation facilities in the ancillary services market instead of the day-ahead market. The first one, on the other hand, has a difficult solution and will likely cause the high marginal cost generation facilities to shut down their production, possibly leading to decreased flexibility of the power system in the short and middle term. For this reason, it will be interesting to evaluate the effects of this phenomenon, which will provide useful insights for enhancing power systems resilience towards exogenous factors. Besides, the price difference among the different market zones highly decreased during the lockdown weeks, as shown in Figures 20 and 21. This is due to a strong reduction in contingencies with the unloading of the power system consequently to demand reduction. It is easy to notice how their standard deviations are not zero or anyway very small (< 1 €) only in sporadic events during the average- and post-lockdown times, thus reducing unfair outcomes in national wholesale markets. The existence of such high price spreads is due to the high demand from the NORTH area and the strong production from the SOUTH area. This led to the violations of some constraints on the power lines and forced the separation of the Italian market into different market areas. There is a strong likelihood that such behavior is determined by the combination of two main factors: low demand and high generation in particular from renewable sources during the middle hour. During the lockdown, however, the reduction of north consumption reflected on the zonal exchanges, strongly reducing contingencies and price splitting among zones, leading to the strong spread reduction observed in the last weeks of Figure 21. Besides, in Figure 22 it is shown a representative snapshot of the situation of the 23rd of March, one of the days in which the commonly observed price spread between zonal prices almost disappeared. In particular, sales prices ranged between 10.70 €/MWh, at noon, to 29.89 €/MWh, at 19:00. Such prices turn out to be different since, as described in Section 4, the hourly clearing price is calculated considering the matching between expected consumption and generation. During day time the large quantity of renewable generation, mostly PV, contributes to the reduction of the market price up to 10.70 €/MWh; in the evening the clearing prices in the market go up to 29.89 €/MWh as a consequence of the selection of the offers of the thermoelectric plant that are selected as marginal units. However, a very interesting aspect is the fact that price has remained unchanged among zones, this denotes that all technical constraints between the zones are satisfied leading therefore to a single

Energies 2020, 13, 3357 13 of 19

This situation hits the electricity-producing industry in two ways. First of all, the daily needed generation was cut about 40%, cutting out the high marginal costs power plants shown in Figure 12. Second, the low PUN prices highly decreased the profit for all the generation sector. This second issue can be overcome with the market dynamics since the competition will likely raise the bids placed by the thermal generation facilities in the ancillary services market instead of the day-ahead market. The first one, on the other hand, has a difficult solution and will likely cause the high marginal cost generation facilities to shut down their production, possibly leading to decreased flexibility of the power system in the short and middle term. For this reason, it will be interesting to evaluate the effects of this phenomenon, which will provide useful insights for enhancing power systems resilience towards exogenous factors. Besides, the price difference among the different market zones highly decreased during the lockdown weeks, as shown in Figures 20 and 21. This is due to a strong reduction in contingencies with the unloading of the power system consequently to demand reduction. It is easy to notice how their standard deviations are not zero or anyway very small (<1 €) only in sporadic events during the average- and post-lockdown times, thus reducing unfair outcomes in national wholesale markets. The existence of such high price spreads is due to the high demand from the NORTH area and the strongEnergies production2020, 13, x FOR from PEER the REVIEW SOUTH area. This led to the violations of some constraints on the power15 of 22 lines and forced the separation of the Italian market into different market areas. There is a strong likelihoodprice for the that whole such behaviorof Italy. It is is determined possible to byobserv the combinatione that the evening of two prices main factors:are identical low demand for all Italy. and highThis generationmeans that in the particular technical from grid renewableconstraints sources are in duringfor all areas the middle and no hour. price During splitting the has lockdown, to occur however,for removing the reductioncontingencies. of north A similar consumption shape is reflectedobservable on in the the zonal early exchanges, hours of the strongly day. In reducingthis time, contingenciesthere are slight and differences price splitting in amongthe price zones, between leading the to NORTH the strong and spread the reductionSOUTH zone, observed but inthese the lastdifferences weeks ofare Figure anyway 21. very small if compared to the standard, pre-lockdown ones shown in Figure 21.

FigureFigure 20.20.Weekly Weekly boxplots boxplots of theof standardthe standard deviations deviatio betweenns between zonal prices zonal in theprices observed in the weekdays. observed weekdays. Besides, in Figure 22 it is shown a representative snapshot of the situation of the 23rd of March, one of the days in which the commonly observed price spread between zonal prices almost disappeared. In particular, sales prices ranged between 10.70 €/MWh, at noon, to 29.89 €/MWh, at 19:00. Such prices turn out to be diﬀerent since, as described in Section4, the hourly clearing price is calculated considering the matching between expected consumption and generation. During day time the large quantity of renewable generation, mostly PV, contributes to the reduction of the market price up to 10.70 €/MWh; in the evening the clearing prices in the market go up to 29.89 €/MWh as a consequence of the selection of the oﬀers of the thermoelectric plant that are selected as marginal units. However, a very interesting

Figure 21. Weekly boxplots of the standard deviations between zonal prices in the observed weekend days’ timeframe.

Energies 2020, 13, x FOR PEER REVIEW 15 of 22

price for the whole of Italy. It is possible to observe that the evening prices are identical for all Italy. This means that the technical grid constraints are in for all areas and no price splitting has to occur for removing contingencies. A similar shape is observable in the early hours of the day. In this time, there are slight differences in the price between the NORTH and the SOUTH zone, but these differences are anyway very small if compared to the standard, pre-lockdown ones shown in Figure 21.

Energies 2020, 13, 3357 14 of 19 aspect is the fact that price has remained unchanged among zones, this denotes that all technical constraints between the zones are satisfied leading therefore to a single price for the whole of Italy. It is possible to observe that the evening prices are identical for all Italy. This means that the technical grid constraints are in for all areas and no price splitting has to occur for removing contingencies. A similar shape is observable in the early hours of the day. In this time, there are slight differences in the price betweenFigure the 20. NORTH Weekly and boxplots the SOUTH of the zone, standard but thesedeviatio diffnserences between are zonal anyway prices very in small the observed if compared to theweekdays. standard, pre-lockdown ones shown in Figure 21.

FigureFigure 21.21. WeeklyWeekly boxplotsboxplots ofof thethe standardstandard deviationsdeviations betweenbetween zonalzonal pricesprices inin thethe observedobserved weekendweekend Energiesdays’days’ 2020 timeframe.timeframe., 13, x FOR PEER REVIEW 16 of 22

a) b)

FigureFigure 22. 22.Zonal Zonal pricesprices observedobserved in Italy at at noon noon (a (a) )and and at at 19:00 19:00 (b ()b of) of 23rd 23rd of ofMarch March 2020. 2020. 4.4. Italian Ancillary Services Costs during the COVID-19 Outbreak 4.4. Italian Ancillary Services Costs during the COVID-19 Outbreak Ancillary services are crucial for system operation security and reliability. In Italy, these services Ancillary services are crucial for system operation security and reliability. In Italy, these services areare provided provided by by the the dispatching dispatching servicesservices market (MSD). (MSD). MSD MSD supplies supplies the the resources resources necessary necessary for for thethe system system management management and and controlcontrol (resolution(resolution of of intrazonal intrazonal congestion, congestion, creation creation of ofenergy energy reserve, reserve, real-timereal-time balancing). balancing). ItIt isis splitsplit intointo twotwo markets: the the MSD MSD “ex-ante” “ex-ante” for for forward forward contracts contracts and and the the balancingbalancing market market (MB) (MB) for for the the intradayintraday trading of power power reserves. reserves. Power Power reserve reserve trading trading takes takes place place in three ways: frequency-controlled reserve (FCR) is the most lucrative, but also the most technically demanding since the required power reserves must be set up within seconds. The automatic frequency restoration reserve (aFRR), in Italy still known as “secondary reserve energy”, is slightly slower and must be set up 15 min after the TSO’s activation command. The third option is the replacement reserve (RR), known as “other services”, used for the management of possible congestions; the preset power must be completely available 15 min after the activation command from the TSO. MSD ex-ante follows the results of MGP by preparing the power reserves for the next day. Part of the reserves accepted by the TSO is delivered on the MB. The MSD prices are highly dependent on the demand/offer ratio and it is easy to observe high price peaks in the presence of high RES fluctuations. Moreover, the offer of flexibility is highly related to the availability of the ramping capability of thermal power plants. Assuming that few thermal generators are available for offering ancillary services, the demand/offer ratio is going to rise. Thus, if few thermal generating units are kept online during the daytime, (the same period in which much RES fluctuations are observed), it will be easier to observe market price spikes [19]. In Figure 23, the weekly total cost sustained by the TSO for ancillary services is illustrated. As expected, the MSD costs highly raised from the COVID-2019 lockdown.

Energies 2020, 13, 3357 15 of 19 in three ways: frequency-controlled reserve (FCR) is the most lucrative, but also the most technically demanding since the required power reserves must be set up within seconds. The automatic frequency restoration reserve (aFRR), in Italy still known as “secondary reserve energy”, is slightly slower and must be set up 15 min after the TSO’s activation command. The third option is the replacement reserve (RR), known as “other services”, used for the management of possible congestions; the preset power must be completely available 15 min after the activation command from the TSO. MSD ex-ante follows the results of MGP by preparing the power reserves for the next day. Part of the reserves accepted by the TSO is delivered on the MB. The MSD prices are highly dependent on the demand/offer ratio and it is easy to observe high price peaks in the presence of high RES fluctuations. Moreover, the offer of flexibility is highly related to the availability of the ramping capability of thermal power plants. Assuming that few thermal generators are available for offering ancillary services, the demand/offer ratio is going to rise. Thus, if few thermal generating units are kept online during the daytime, (the same period in which much RES fluctuations are observed), it will be easier to observe market price spikes [19]. In Figure 23, the weekly total cost sustained by the TSO for ancillary services is illustrated. Energies 2020, 13, x FOR PEER REVIEW 17 of 22 As expected, the MSD costs highly raised from the COVID-2019 lockdown.

30 [M€]

0 Week 4 Week 5 Week 6 Week 7 Week 8 Week 9 Week 10 Week 11 Week 12 Week 13 FigureFigure 23. 23. WeeklyWeekly system system costs costs paid paidby the by Italian the tran Italiansmission transmission system operator system (TSO) operator for ancillary (TSO) for services.ancillary services.

InIn particular, particular, a considerable incrementincrement ofof approx.approx. 100%100% ofof thethecosts costs is is registered registered between between week week 9 9and and week week 13 13 (full (full lockdown lockdown started started on on week week 11, 11, partial partial lockdown lockdown was was in in place place from from week week 9). 9). Figure Figure 24 24shows shows how how the the nationwide nationwide costs costs are are subdivided subdivided per per market market zone. zone. FigureFigure 2424 showsshows a a detailed representation representation of of ho howw both both the volume and the prices distribution increasedincreased duringduring the the lockdown lockdown period, period, leading leading to the to overallthe overall cost increase cost increase showed showed in Figure in 25 Figure. Anyway, 25. Anyway,it is important it is important to notice alsoto notice how historically-stablealso how historically-stable zones as north zones and as centrenorth,north and centrenorth, which often which show oftenlow MSDshow costs, low MSD show costs, a great show relative a great increase relative in increase costs during in costs the during lockdown the lockdown phase, as phase, a signal as ofa signalstressful of situationstressful spreadsituation on spread the entire on powerthe entire system. power This system. result showsThis result how theshows eﬀects how of the lockdowneffects of theworked lockdown as a stress worked test foras a the stress electricity test for system, the elec whichtricity can system, be of greatwhich interest can be for ofunderstanding great interest for the understandingrelation between the RES relation/Load between ratio and RES/Load system ﬂexibility. ratio and system flexibility. The southern and south-central zones are the ones that experienced the largest cost increases.

This eﬀect is due to aNORD large PV-productionCNORD CSUD in the southernSUD SARD part ofSICI Italy, andROSN a general lack of hydro pumping30 facilities in the same areas. Investigating deeper these results, the MSD results of 2019 and 2020 were statistically compared in Figure 26. In this ﬁgure, the weekly boxplots of hourly MSD costs are shown for weeks 5–13 of 2019 and 2020. The plot shows a strong increase in all the statistical cost 25 indicators during weeks 12 and 13. In particular, the median costs almost doubled concerning the

20 [M€] 15

0 Week 4 Week 5 Week 6 Week 7 Week 8 Week 9 Week 10 Week 11 Week 12 Week 13 Figure 24. Weekly system costs for ancillary services, per each market zone.

Energies 2020, 13, x FOR PEER REVIEW 17 of 22

30 [M€]

0 Week 4 Week 5 Week 6 Week 7 Week 8 Week 9 Week 10 Week 11 Week 12 Week 13 Figure 23. Weekly system costs paid by the Italian transmission system operator (TSO) for ancillary services.

In particular, a considerable increment of approx. 100% of the costs is registered between week 9 and week 13 (full lockdown started on week 11, partial lockdown was in place from week 9). Figure 24 shows how the nationwide costs are subdivided per market zone. Figure 24 shows a detailed representation of how both the volume and the prices distribution

Energiesincreased2020, 13during, 3357 the lockdown period, leading to the overall cost increase showed in Figure 1625. of 19 Anyway, it is important to notice also how historically-stable zones as north and centrenorth, which often show low MSD costs, show a great relative increase in costs during the lockdown phase, as a previoussignal of reference stressful weekssituation and spread the previous on the entire year. power The same system. observation This result can shows be made how on the the effects statistical of maximumthe lockdown of the worked distribution, as a stress showing test strongfor the changes electricity in thesystem, observed which prices. can be Besides, of great it interest is worthwhile for tounderstanding notice the existence the relation of the between very high RES/Load outliers ratio of week and 12.system flexibility.

NORD CNORD CSUD SUD SARD SICI ROSN 30

20 [M€] 15

0 Week 4 Week 5 Week 6 Week 7 Week 8 Week 9 Week 10 Week 11 Week 12 Week 13 Energies 2020, 13, xFigure FORFigure PEER 24. 24. REVIEWWeekly Weekly systemsystem costs for ancillary ancillary services, services, per per each each market market zone. zone. 18 of 22

160.000 250 Volume (MWh) Average Price (€) 140.000 200 120.000

100.000 150

80.000

100 60.000 Volume [MWh]

40.000 Price of MSD bids[€/MWh] 50 20.000

0 0 Week 4 Week 5 Week 6 Week 7 Week 8 Week 9 Week 10 Week 11 Week 12 Week 13 FigureFigure 25. 25.Weekly Weekly system costs costs paid paid by by the the Italian Italian TSO TSO in the in southern the southern Italy (CSUD) Italy (CSUD) area. area.

TheThe dispatching southern and services south-central costs increase zones are substantially the ones that due experienced to the unexpected the largest cost decrease increases. in demand thatThis strengthened effect is due to the a large challenges PV-production of balancing in the so theuthern Italian part power of Italy, system. and a general Normally, lack of to hydro follow the pumping facilities in the same areas. Investigating deeper these results, the MSD results of 2019 and ramps (evening and morning), rapid balancing actions are required and created by dispatchable 2020 were statistically compared in Figure 26. In this figure, the weekly boxplots of hourly MSD costs units with high modulation capacities and fast response times. The best resources for this service are are shown for weeks 5–13 of 2019 and 2020. The plot shows a strong increase in all the statistical cost theindicators hydroelectric during production weeks 12 and and 13 pumping. In particular, units, the which median can costs enter almost service doubled and concerning vary the production the in aprevious very short reference time. weeks Thermoelectric and the previous plants year. can The also same be observation used, but can they be mustmade beon the kept statistical at a technical minimummaximum in of the the hours distribution, when their showing production strong ischan notges needed in the (although observed theyprices. have Besides, unsurmountable it is physicalworthwhile limits to of notice minimum the existence uptime, of the downtime, very high rampoutliers up of/ downweek 12. capabilities). Besides, as a result of the non-programmableThe dispatching services renewable costs increase sources, substantia the rampslly due (especially to the unexpected the evening decrease one) in become demand steeper. Forthat this strengthened reason, it is, the therefore, challenges necessary of balancing an increased the Italian reserve power capacitysystem. Normally, to cover theto follow steeper the loading rampsramps accentuated (evening and by morning), the loss ofrapid photovoltaics balancing actions in the are evening required hours and created and/or by the dispatchable loss of the units wind in the with high modulation capacities and fast response times. The best resources for this service are the morning hours, in particular in South Italy, leading to an increase in dispatching costs (for services that hydroelectric production and pumping units, which can enter service and vary the production in a can be oﬀered almost exclusively by thermoelectric units). very short time. Thermoelectric plants can also be used, but they must be kept at a technical minimum in the hours when their production is not needed (although they have unsurmountable physical limits of minimum uptime, downtime, ramp up/down capabilities). Besides, as a result of the non- programmable renewable sources, the ramps (especially the evening one) become steeper. For this reason, it is, therefore, necessary an increased reserve capacity to cover the steeper loading ramps accentuated by the loss of photovoltaics in the evening hours and/or the loss of the wind in the morning hours, in particular in South Italy, leading to an increase in dispatching costs (for services that can be offered almost exclusively by thermoelectric units).

Energies 2020, 13, 3357 17 of 19 Energies 2020, 13, x FOR PEER REVIEW 19 of 22

FigureFigure 26. Statistical 26. Statistical comparison comparison between between 2019 and and 2020 2020 Dispatching Dispatching services services market market (MSD) costs. (MSD) costs. ShownShown quantities quantities are relativeare relative to to hourly hourly MSD MSD costs and and are are aggregated aggregated weekly. weekly.

4.5. Power System CO2 Emissions during COVID-2019 Outbreak 4.5. Power System CO2 Emissions during COVID-2019 Outbreak In this last part of the study, the environmental impact of the COVID-19 is evaluated by Inconsidering this last part the of changes the study, in CO2 the emissions environmental for the impactItalian power of the system. COVID-19 To quantitatively is evaluated evaluate by considering the changesthe reduction in CO2 in emissions CO2 emissions, for the the Italian share of power energy system. produced To by quantitatively CO2 emitting generation evaluate thefacilities reduction in CO2 emissions,was calculated, the shareby using of energythe approach produced described by COin [20].2 emitting This value generation was then facilitiesmultiplied was by the calculated, by usingnational the approach Emission describedFactor (EF) infor [ 20electricity]. This valueproduction was thenof thermal multiplied origin in by 2017, the nationalwhich was Emission 433.2 g Factor (EF) forCO2eq/kWh. electricity This production value is far of thermalbelow the origin European in 2017,average, which equal was to 575 433.2 g CO2eq/kWh g CO2eq/ kWh.[20]. Figure This value is 27 shows the total estimated emissions of the whole electricity system during the observed weeks. As far below the European average, equal to 575 g CO2eq/kWh [20]. Figure 27 shows the total estimated a suggestion for future research, a more precise EF can be estimated for this period to better estimate emissionsthis value of the since whole the shares electricity of energy system supplied during by different the observedthermal technologies weeks. As will a likely suggestion change. At for future research,the atime more of writing precise this EF study, can however, be estimated these data for are this unfortunately period to better not available. estimate this value since the shares of energyFor instance, supplied considering by diff theerent open thermal data in [15], technologies the total production will likely decreased change. by At approximately the time of writing this study,15% between however, the these last week data of are March unfortunately 2019 and the not last available. of March 2020, whereas the RES production Forwas instance, approximately considering unchanged the open(−3%) data and inthe [15 thermoelectric], the total production generation decreasedwas strongly by reduced approximately (−26%). This reduced thermal generation caused a change in total CO2 emissions which reached a 15% betweendecrease theof 40% last during week the of last March week 2019 of March. and theSure lastly this of phenomenon March 2020, is whereaspositive for the theRES national production was approximately unchanged ( 3%) and the thermoelectric generation was strongly reduced ( 26%). development plan, as well as the− for the European one. Such a reduction in emissions through a short − This reducedtime is certainly thermal interesting generation because caused it could a change lead to infuture total developments CO2 emissions in thewhich field of reachedCO2 reduction. a decrease of 40% duringIt would the certainly last week be interesting of March. as Surely a future this stud phenomenony to analyze whether is positive this decreasing for the national trend is going development plan, asto well continue, as the perhaps for the with European some fluctuations one. Such or a whethe reductionr it will in return emissions to thethrough historical aemission short time values is certainly known. interesting because it could lead to future developments in the field of CO2 reduction. It would certainly be interesting as a future study to analyze whether this decreasing trend is going to continue, perhaps with someEnergies fluctuations 2020, 13, x FOR or PEER whether REVIEWit will return to the historical emission values known.20 of 22

20 2018 2019 2020 18 16 14 12 10

Mt CO2 8 6 4 2 0 Week 4 Week 5 Week 6 Week 7 Week 8 Week 9 Week 10 Week 11 Week 12 Week 13

Figure 27. Weekly CO2 emissions (in Mt) before and during COVID-19 lockdown in Italy. Figure 27. Weekly CO2 emissions (in Mt) before and during COVID-19 lockdown in Italy.

5. Conclusions The economic effects of crises arising from COVID-19 have manifested themselves in the disruption of supply chains and the consequent interruption of productive and commercial activities. These changes have caused in the electricity sector tangible changes in power consumption profiles due to the shutdown of public offices, schools and various business facilities, with consequences on the entire operation of the electrical system. These effects on the Italian electricity system can be regarded as an unexpected country size experiment and resiliency test performed on the electric power system. The study shows that the pandemic caused a reduction of consumption up to 37% compared to the same time as the previous year. The observed reduction in electricity consumption has had immediate effects on the day-ahead electricity market. The reduction in consumption excluded part of the costly thermoelectric generation, which is automatically taken out of the market due to the production cost compared with renewable generation. As a consequence, wholesale energy prices decreased about 30% in the last weeks of march and in the first week of April, and in some cases reached a value of 0 €/MWh. This circumstance has had immediate consequences also on the reduction of CO2 emissions. Starting from these observations, this study investigates the observed price changes, to understand if they can only be related to exogenous factors or if the unexpected crisis triggered already existent systemic weaknesses. The observations highlighted a strong increase in the ancillary market costs, which showed an average increase of about 70% for the weeks before the lockdown, and almost doubled during the last week of March. These extra costs were determined by the need of TSO to achieve system adequacy and security recurring to the flexibility of conventional plants for providing ancillary services such as real-time load balancing, frequency regulation and reserve margins formation, which are obtained in a competitive context and cannot be guaranteed from renewable sources. The ancillary cost growth was common in all the Italian market zones, but it was observed to be significant in the Italian Southern zone, where the majority of the RES generation is installed. Finally, the impact of COVID-19 pandemic on the generation mix is discussed. In particular, it was observed that the reduction in overall consumption did not affect the RES generation, but rather the amount of energy supplied by conventional sources. The share of energy supplied by RES steeply increased, reaching a daily RES penetration higher than 40%, when the average seasonal value was about 23%. The contents of this study allow improving the knowledge about techno-economic effects on the power industry and society due to unpredictable events such a pandemic and can stimulate deeper investigation about the resilience of power systems with high RES penetration. Altogether, the

Energies 2020, 13, 3357 18 of 19

5. Conclusions The economic effects of crises arising from COVID-19 have manifested themselves in the disruption of supply chains and the consequent interruption of productive and commercial activities. These changes have caused in the electricity sector tangible changes in power consumption profiles due to the shutdown of public offices, schools and various business facilities, with consequences on the entire operation of the electrical system. These effects on the Italian electricity system can be regarded as an unexpected country size experiment and resiliency test performed on the electric power system. The study shows that the pandemic caused a reduction of consumption up to 37% compared to the same time as the previous year. The observed reduction in electricity consumption has had immediate effects on the day-ahead electricity market. The reduction in consumption excluded part of the costly thermoelectric generation, which is automatically taken out of the market due to the production cost compared with renewable generation. As a consequence, wholesale energy prices decreased about 30% in the last weeks of march and in the first week of April, and in some cases reached a value of 0 €/MWh. This circumstance has had immediate consequences also on the reduction of CO2 emissions. Starting from these observations, this study investigates the observed price changes, to understand if they can only be related to exogenous factors or if the unexpected crisis triggered already existent systemic weaknesses. The observations highlighted a strong increase in the ancillary market costs, which showed an average increase of about 70% for the weeks before the lockdown, and almost doubled during the last week of March. These extra costs were determined by the need of TSO to achieve system adequacy and security recurring to the flexibility of conventional plants for providing ancillary services such as real-time load balancing, frequency regulation and reserve margins formation, which are obtained in a competitive context and cannot be guaranteed from renewable sources. The ancillary cost growth was common in all the Italian market zones, but it was observed to be significant in the Italian Southern zone, where the majority of the RES generation is installed. Finally, the impact of COVID-19 pandemic on the generation mix is discussed. In particular, it was observed that the reduction in overall consumption did not affect the RES generation, but rather the amount of energy supplied by conventional sources. The share of energy supplied by RES steeply increased, reaching a daily RES penetration higher than 40%, when the average seasonal value was about 23%. The contents of this study allow improving the knowledge about techno-economic effects on the power industry and society due to unpredictable events such a pandemic and can stimulate deeper investigation about the resilience of power systems with high RES penetration. Altogether, the availability of testing data will grant the possibility to achieve a much better tuning of scenario-based analysis, allowing more accurate planning phases and processes. One key message from the paper is that it is possible to run the power system with high shares of generation from RES but the cost for ancillary services could be very high without a high level of flexibility on the demand side and the installation of significant energy storage capacity.

Author Contributions: All the authors contributed equally to the conceptualization, data curation, writing of the original draft preparation, review and editing. All authors have read and agreed to the published version of the manuscript. Funding: This research received no external funding. Conﬂicts of Interest: The authors declare no conﬂict of interest.

References

1. Callaway, E.; Cyranoski, D.; Mallapaty, S.; Stoye, E.; Tollefson, J. Coronavirus by the numbers. Nature 2020, 579, 26. 2. Tuohy, A.; Kelly, A.; Deaver, B.; Lannoye, E.; Brooks, D. COVID-19 Bulk System Impacts. Demand Impacts and Operational and Control Center Practices. EPRI Transmission Operations and Planning. EPRI Product ID: 3002018602. 27 March 2020. Available online: http://mydocs.epri.com/docs/public/covid19/3002018602R2.pdf (accessed on 28 March 2020). Energies 2020, 13, 3357 19 of 19

3. IEA. Global Energy Review 2020. 2020. Available online: https://www.iea.org/reports/global-energy-review -2020/global-energy-and-co2-emissions-in-2020 (accessed on 10 June 2020). 4. Cicala, S. Early Economic Impacts of COVID-19 in Europe: A View from the Grid; University of Chicago: Chicago, IL, USA, 2020. Available online: https://home.uchicago.edu/~{}scicala/papers/real_time_EU/real_time_EU. pdf (accessed on 10 June 2020). 5. IAEE Energy Forum/Covid-19 Issue 2020. Available online: https://www.iaee.org/documents/2020EnergyFor umSI.pdf (accessed on 10 June 2020). 6. Linkov, I.; Bridges, T.; Creutzig, F.; Decker, J.; Fox-Lent, C.; Kröger, W.; Lambert, J.H.; Levermann, A.; Montreuil, B.; Nathwani, J.; et al. Changing the resilience paradigm. Nat. Clim. Chang. 2014, 4, 407–409. [CrossRef] 7. Hosseini, S.; Barker, K.; Ramirez-Marquez, J.E. A review of definitions and measures of system resilience. Reliab. Eng. Syst. Saf. 2016, 145, 47–61. [CrossRef] 8. Gasser, P.; Lustenberger, P.; Cinelli, M.; Kim, W.; Spada, M.; Burgherr, P.; Hirschberg, S.; Stojadinovic, B.; Sun, T.Y.A review on resilience assessment of energy systems. Sustain. Resil. Infrastruct. 2019, 1–27.[CrossRef] 9. Terna. Transmission System Operator. Statistical Data. Available online: www.terna.it (accessed on 28 March 2020). 10. Italy’s National Energy Strategy-SEN 2017. Available online: https://www.mise.gov.it/images/stories/docum enti/BROCHURE_ENG_SEN.PDF (accessed on 28 March 2020). 11. Integrated National Energy and Climate Plan–31/12/2018. Available online: https://ec.europa.eu/energy/site s/ener/files/documents/it_final_necp_main_en.pdf (accessed on 28 March 2020). 12. Terna. Transmission System Operator. The National Electricity Transmission Grid Development Plan. 2019. Available online: https://www.terna.it/en-gb/sistemaelettrico/pianodisviluppodellarete.aspx (accessed on 28 March 2020). 13. GSE. Gestore dei Servizi Energetici Operator. Available online: https://www.gse.it/dati-e-scenari/statistiche (accessed on 28 March 2020). 14. Available online: http://www.mercatoelettrico.org/en/tools/glossario.aspx#Constrained%20Zone (accessed on 8 April 2020). 15. Terna. Transmission System Operator. Transparency Platform. Available online: https://www.terna.it/it/sist ema-elettrico/transparency-report (accessed on 28 March 2020). 16. Carlini, E.M.; Schroeder, R.; Birkebæk, J.M.; Massaro, F. EU transition in power sector: How RES affects the design and operations of transmission power systems. Electr. Power Syst. Res. 2019, 169, 74–91. [CrossRef] 17. Corona, P.; Ghiani, E.; Contreras, J. Analysis of Sardinia-Italy Energy Flows with Future Transmission Investments for Increasing the Integration of RES. In Proceedings of the 2019 1st International Conference on Energy Transition in the Mediterranean Area (SyNERGY MED), Cagliari, Italy, 28–30 May 2019; pp. 1–6. 18. GME. Energy Market Operator. Statistical Data. Available online: www.mercatoelettrico.org (accessed on 8 April 2020). 19. Mureddu, M.; Meyer-Ortmanns, H. Extreme prices in electricity balancing markets from an approach of statistical physics. Phys. A Stat. Mech. Appl. 2018, 490, 1324–1334. [CrossRef] 20. Istituto Superiore per la Protezione e la Ricerca Ambientale (ISPRA). Fattori di Emissione Atmosferica di Gas a Effetto Serra nel Settore Elettrico Nazionale e nei Principali Paesi Europei. Atmospheric Greenhouse Gas Emission Factors in the National Electricity Sector and in the Main European Countries; ISPRA Rapporti 303/2019; ISPRA: Rome, Italy, 2019.

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).