Self-sufficient renewable energy supply in urban areas: Application to the city of Seville

Depósito de Investigación de la Universidad de Sevilla https://idus.us.es/ This is an Accepted Manuscript of an article published by Elsevier in Sustainable Cities and Society, Vol. 46, on April 2019, , available at: https://doi.org/10.1016/j.scs.2019.101450 © 2019 Elsevier. En idUS Licencia Creative Commons CC BY-NC-ND Accepted Manuscript Title: Self-sufficient renewable energy supply in urban areas: application to the city of Seville Authors: Angel Arcos-Vargas, Antonio Gomez-Exposito, Francisco Gutierrez-Garcia PII: S2210-6707(18)32083-3 DOI: https://doi.org/10.1016/j.scs.2019.101450 Article Number: 101450 Reference: SCS 101450 To appear in: Received date: 12 October 2018 Revised date: 4 January 2019 Accepted date: 28 January 2019 Please cite this article as: Arcos-Vargas A, Gomez-Exposito A, Gutierrez-Garcia F, Self-sufficient renewable energy supply in urban areas: application to the city of Seville, Sustainable Cities and Society (2019), https://doi.org/10.1016/j.scs.2019.101450 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. 1 Self-sufficient renewable energy supply in urban areas: application to the city of Seville Angel Arcos-Vargas1, Antonio Gomez-Exposito and Francisco Gutierrez-Garcia University of Seville, Spain 1 Corresponding author [email protected] Highligths:  A method to evaluate if a city can be self-sufficient energetically is proposed.  Only public, available and no cost information is used.  The energy of transport and heat has been considered electrified  The case of Seville shows a self-sufficient renewable system presents lower costs.  Seasonal storage obtains non-economic solutions. Abstract: This paper proposes a methodology that, using only publicly available information, assesses the capacity of urban agglomerations to be self-sufficient and energy sustainable. The methodology evaluates the potential of the urban surface to accommodate thermal and PV solar facilities, as well as the energy storage requirements on a year-round basis. When applied to the city of Seville, considering not only the current electricity demand but also the one arising from the electrification of both urban transport and the entire thermal energy currently supplied by gas, the conclusion is that it is possible to have a self-supplied energy system based almost exclusively on PV and thermal panels. Given the high cost of seasonal storage, the resulting system is economically unviable, so it makes sense to keep the transmission grid as a back-up system to feed power into the city at certain times of the year (up to 8% of annual consumption). In the case study, the renewable energy comes mainly from PV facilities roof-mounted (72%), or groundmounted in the surrounding urban lots (25%), plus a modest contribution from biogas (wastewater). A discussion is included regarding the role of regulated utilities, that should be reconsidered in the light of these upcoming scenarios. Keywords: Self-Supply; Regulation; Cities; Emissions; Renewables; Storage 1. Introduction ACCEPTED MANUSCRIPT 2 The sustainability of cities will be one of the great challenges of the 21st century. Of the 10 billion people expected to live on earth in 2050, almost 7 billion will live in cities 1 . This will pose numerous resource availability and sustainability problems, both locally and globally, that will have to be solved efficiently. One of the issues that most concerns society is air pollution in large cities, where today we already find examples that may be conditioning their growth, due to its impact on mortality rates (Chen et al., 2017). The two most important sources of emissions in the cities are associated with vehicle traffic and electricity production. In this context, all links of the electric system chain (generation, transmission, distribution and consumption) will play a more crucial role in assuring that urban citizens enjoy a safe, resilient, costcompetitive, clean and sustainable energy delivery. In recent years, the technologies associated with generation, storage, measurement and control of electricity have improved exponentially, thanks to a plethora of developments including: emission-free distributed generation El-Khattam and Salama, 2012, at similar or even lower prices than conventional technologies providing electricity through the wires; the growth of the concept of zero energy buildings Cao Xiadong et al, 2016; the evolution of electrochemical storage devices, in terms of both performance (energy density) and cost, driven mainly by the development of the electric vehicle (EV) Hannan et al, 2017; the introduction of smart meters Alahakoon and Yu., 2016, allowing almost instantaneous knowledge of energy flows at all levels of the network (HV, MV and LV); and the so-called voltage source converter, radically simplifying the task of interfacing conventional AC systems to ubiquitous DC systems and renewable generators. These advances make it possible to design a new energy model, either for residential users 1 UN World Population Prospects, https://esa.un.org/unpd/wpp/DataQuery/  ACCEPTED MANUSCRIPT 3 Arcos-Vargas et al, 2018 or whole cities Chourabi et al, 2012, intended to allow their sustainable and efficient growth in future scenarios characterized by a further electrification of the energy demand, and part of other energy consumptions, such as heat and cooling. A first vision of this new energy model for urban areas can be found in Gomez-Exposito et al, 2018, although it only gives a very general overview without providing an analytic or quantitative assessment. Several studies proposing emission-free energy models have been published, such as that of Jacobson (Jacobson et al., 2018), the one published by the Economist for Energy group (2018) or Linares and Declercq (2018). The first is an emission-free, low-cost energy model for 139 countries, while the second is an energy scenario for Spain, proposing an electricity mix where gas accounts for more than 50% of generation by 2050. Both have been developed at the country level, not taking into account the local specificities of cities. Jacobson et al. (2018b) have carried out a study to define an energy roadmap for 53 towns and cities in North America. Although it reaches specific quantitative results, it does not provide a practical tool exploiting the actual topography and layout of each city. This paper presents a methodology that, based on publicly available data, performs a systematic assessment of the year-round energy self-supply capability of urban areas, based on emissions-free production (mainly PV on roofs and city outskirt areas plus a very modest contribution from biogas wastewater treatment plants) along with energy storage (Li-ion batteries). Then, the city of Seville is used as a case study to illustrate in detail the proposed evaluation scheme. Firstly, the urban surface energy potential is calculated and, subsequently, the expected electrical, transportation 2 and thermal demands are estimated. Based on those values, the 2 It is considered that 100% of urban transport is electrified, including buses, cars, vans and bikes. ACCEPTED MANUSCRIPT 4 hourly balances throughout an average year, as well as the associated storage needs, are then obtained. The methodology is applied to three possible supply scenarios, differing mainly in the amount and management of the installed storage. The conclusions show that, at least for cities located at low enough latitudes with moderate industrialization levels, there are several alternatives to design a 100% renewable energy system embedded within the urban perimeter, the required investment being significantly influenced by the amount of storage assumed. The overall cost may become comparable to that of the traditional centralized system when the storage capacity is dimensioned just for daily needs, which calls for some electricity (hopefully also renewable) to be procured through the transmission grid to fully balance the city demand. The rest of the paper is organized as follows: after this introduction, Section 2 offers an overview of the methodology proposed. This methodology can be applied to any city, without the need for specific information; i.e., it is applied using the customary statistical and cartographic data of any city. Section 3 presents the application of the proposed method to the particular case of the city of Seville (the fourth largest in Spain) showing the results obtained for each of the three electricity supply alternatives considered herein. Finally, Section 5 presents the main conclusions of the work along with recommendations for interested municipalities. 2. Methodology For the sake of clarity, the reader is referred to Figure 1, showing a schematic overview of the proposed methodology. ACCEPTED MANUSCRIPT 5 Figure 1. Methodology flowchart. 2.1 Energy potential The objective of this section is to evaluate the renewable energy that can be harvested by covering all suitable rooftops in the city with solar systems (both thermal and PV), eventually supported by extra PV panels located at surrounding urban lots and the biogas produced by the wastewater treatment plants serving the municipality. 2.1.1. Solar Potential The methodology to analyse the solar potential comprises three stages: 1. The available surface is estimated. 2. The optimal configuration or combination of solar energy collectors is determined. ACCEPTED MANUSCRIPT 6 3. The hourly production is calculated considering the performance of the solar panels, surface orientations and irradiation conditions. A similar approach is developed by Ordoñez et al., 2010, who estimate the PV energy capacity of the residential rooftops in Andalusia by calculating the monthly values of each sub-region. The useful roof surface is determined by random stratified sampling applied to the buildings of all cities. Surface Analysis A cartographic analysis is carried out using coloured city maps, leading to a classification of the overall urban surface in three types:  Residential/commercial/public (schools, sport centres, etc.) buildings.  Industrial buildings.  Empty or unused lots (i.e., urban areas still free of buildings or infrastructures). In turn, the first two types are subdivided into two categories: flat roof and pitched roof. Monuments and protected institutional buildings (palaces, historical, etc.) are excluded. The total surface is calculated using the horizontal projection of the building areas. Adobe Photoshop CC software application quantifies the number of pixels within a selection of an image, thereby providing the percentage of selected pixels over the total and, hence, the fraction of selected surface from the total area represented on the map. Brito et al., 2012, and Santos et al., 2014, propose a methodology to assess the total area of the rooftop using Light Detection and Ranging (LiDAR) data. In the same way, Jacques et al., 2014 propose another methodology using low-resolution LiDAR data to estimate the available surface on the rooftop considering common roof profiles and types over a certain region. The assessment proposed in this work does not resort to LiDAR information, as this dataset is not available for every city and is expensive to obtain. ACCEPTED MANUSCRIPT 7 Both the available surface suitable for solar system installation and the pitched roof orientation are estimated by comparing satellite images with the mapping representation using a representative sample for each class of buildings. When identifying the suitable flat roof building areas, the space occupied by objects such as elevator shafts, stairwells, chimneys, partitions between adjacent buildings, etc., as well as the surface unavailable by shadow effects (produced by those objects and nearby buildings) are estimated. The final value is obtained by taking the arithmetic mean among all the samples. Kartesis et al., 2013, estimate the available surface on the rooftops by making a 3D digital mock-up using Google SketchUpTM software application and simulating the shadow effects and permanent obstacles for only one type of building. When the orientations of the pitched roof buildings are taken into account, five categories of roofs arise, as shown in Table 1. Roof type (possible orientations) Flat roof No slope Pitched roof (oriented to:) South North East West Table 1. Roofs classification. Optimal layout The main goal of this subsection is to establish the optimum configuration of PV solar panels that provides the best performance in terms of energy yield for a specific location. A simulation model is developed to which the optimization process is applied for a given ACCEPTED MANUSCRIPT 14 Yearly publications of the main institutions ruling the cities, regions or the country offer a large amount of data about total electric consumption. Additionally, the hourly distribution of the electric demand can be estimated using the typical consumption curves provided by the national system operator or, if available, from the aggregated or average (anonymous) data provided by smart meters. 2.2.2 Electric vehicle energy consumption An estimation is performed of the additional electric energy demand arising from the upcoming electrification of transportation. The calculation considers the number and type of vehicles currently registered in the city (cars, buses, vans and motorcycles), their specific consumption (Wh/km) and the average driven distances. The hourly charging distribution of the vehicles has been obtained from Postma, 2014 which analyses the charging process of the EVs in the Netherlands and provides the average of total power charge transactions. 2.2.3 Thermal energy consumption The thermal energy requirements are estimated by assuming that all the natural gas and Liquid Petroleum Gas (LPG) delivered to the city are used to produce thermal energy, which is to be provided theoretically by solar thermal panels in the future scenarios considered in this work. Moreover, the demand associated to high-temperature industrial process is also considered in the thermal energy balance to be matched with solar thermal panels. This assumption is adopted for the sake of simplicity, as the high-temperature thermal energy required by certain industrial processes cannot be efficiently or readily replaced by electricity with todays’s technology. A detailed case-by-case analysis would be necessary to more accurately assess which industrial thermal consumptions could be actually electrified, which is out of the scope of this paper. ACCEPTED MANUSCRIPT 15 2.3 Energy balance analysis The energy balance analysis is performed for several scenarios differing from each other in the degree of sustainability (self-supply), storage needs and costs. The scenarios considered are:  The first scenario assesses the possibility of achieving a fully (self-sufficient) renewable energy system by exclusively resorting to PV panels for electricity production, located whenever possible on rooftop surfaces. In this case, a huge storage system is assumed so that the whole city energy consumption-production balance is secured at any time of the year, by shifting the summer energy surplus to the winter.  A less ambitious model has been also assessed which restricts the amount of energy storage to the daily needs (i.e., a storage system providing nightly energy consumption from the overproduction obtained over the sunny preceding hours).  Finally, a third scenario is considered in which some extra land, in addition to rooftops, is used as required to meet the daily demand for the less favourable day of the year. As in the previous case, a minimal storage system is assumed to satisfy the hourly demand on a daily basis. Similar approaches have been analysed by Sisternes et al., 2016, and O’Shaughnessy et al., 2018. The first one assesses the storage system value in the decarbonization of the electricity sector, while the second carries out a storage system analysis considering batteries as the best option to satisfy the hourly consumption for the whole day. The storage system proposed is based on the fulfilment of the energy balance for most of the year. ACCEPTED MANUSCRIPT 16 2.3.1 Energy balances For the energy balances, two different cases are considered, one assessing only the electric production-consumption balance, the other adding the thermal equilibrium to the equation. Electric energy balance only A comparison between the city self-contained renewable generation capacity and the electric energy demand is performed for each hour of the year, assuming all of the available roof surfaces are devoted to PV panels. This way, an overview is obtained of the energy production model, its capacity and, particularly, the periods with deficit and overproduction. Electrical and thermal energy balance The thermal solar panels are installed on flat and south oriented surfaces in order to assure the best performance. The surface required to provide the current city thermal energy demand is calculated and, then, the electric energy balance is carried out as mentioned before with the rest of the surface. 2.3.2 Energy storage system With the aim of satisfying the energy demand at any hour of the year, a storage system must be considered. Sevket and Tepe, 2017, and Gómez-Exposito et al., 2017, assess the performance of different energy storage systems, depending on their technical characteristics and application. The storage system considered in this work is based on liion batteries, featuring about 5% of losses per charge-discharge cycle. The sizing of the storage capacity is estimated for each scenario through a MATLAB algorithm which uses the data arising in the hourly electric energy balance. The best and worst demand matching periods of the year are established regarding the overproduction ACCEPTED MANUSCRIPT 17 or the deficit of the renewable generation. Those periods are identified by summation (integration) of the hourly net balance for every hour of the year, considering the previous ones. In addition to the hourly energy production and consumption, the algorithm duly considers in the integration process the current status (charging or discharging) of the battery, providing in this way daily and monthly energy balances. A graphical representation of those values (see Figure 12 below) shows if there is an overproduction (positive slope) or deficit (negative slope). Moreover, losses for every daily chargedischarge cycle are taken into account. The maximum amount of energy that must be seasonally stored is defined as the difference between the maximum and minimum values, being the time lapse between the two extreme points the number of days when the city is supplied in part with stored energy. For the two scenarios that neglect the possibility of seasonal deficit/surplus compensation, only the energy overproduced and stored during the central hours of the day is to be delivered over the next night. This way, in order to properly size the energy storage system, it is necessary to identify the worst-case day, which is the day with largest consumption during the non-production hours among all days with positive total balance. 3. Case study In this section the above methodology is applied to the city of Seville, located at the South of Spain (coordinates: 37°23′00″N 5°59′00″W). The main results of the assessment are shown in the sequel. 3.1 Energy potential. 3.1.1 Solar potential Surface Analysis. ACCEPTED MANUSCRIPT 18 The total surface of the municipality of Seville (urban area and beyond) is 141 km2, of which 75 km2 are properly speaking urban areas (streets and buildings). Figure 2 is a map showing all the surface area in the municipality of Seville (obtained from the Spatial Data Infrastructure of the Town Hall of Seville through the online digital map geoSEVILLA 5 ), where the area enclosed within the green line contains the urban fabric. All the buildings are represented on this map in different colours, depending on their function. The residential/commercial/public buildings are defined in light brown; the industrial zones are defined by two purple colours, the darker one representing buildings; the institutional buildings and monuments are shown in dark brown. Moreover, the rural (i.e. not covered by buildings) areas are represented in beige colour. As most industrial buildings have pitched roof, it is assumed for simplicity that all of them correspond to this type of roof. Table 3 collects the total surface of each type for all the buildings in the municipality. Although difficult to assess visually, roughly 25% of the urban municipality surface is occupied by building rooftops. Also, the total available surface of the non built-up areas is about 49 km2. 5 http://sig.urbanismosevilla.org/visorgis/ ACCEPTED MANUSCRIPT 19 Figure 2. Map of the city of Seville. Type of buildings m2 Flat roof residential buildings 10,397,722 Pitched roof residential buildings 1,331,809 Pitched roof industrial buildings 3,336,071 Total 15,065,602 Table 3. Total building surface. To calculate the availability factor and roof orientations, several samples corresponding to each type have been analysed in detail and then extrapolated: 8 samples for flat roof ACCEPTED MANUSCRIPT 20 residential buildings, which represent a majority in the city, 7 samples for pitched roof residential buildings and 6 samples for pitched roof industrial buildings. The mean availability factor for flat roof residential buildings is 68%, according to the sample. Regarding pitched roof buildings, both residential and industrial, Table 4 shows the surface corresponding to each orientation, including flat surface, and non-available areas. Surface (m2) South North East West Flat surface Non-available Total Residential buildings 306,316 292,998 239,725 213,089 159,817 119,862 1,331,809 Industrial buildings 1,434,513 433,690 467,050 467,050 467,050 66,721 3,336,076 Table 4. Available surface of pitched roofs. Optimal layout. On the one hand, to calculate the optimal layout it is necessary to obtain the solar irradiation received by the optimally oriented/inclined panels on a flat surface. In Table 5 the values obtained from PVGIS are shown. Fixed system: inclination=33°, orientation=0° Month Ed Em Hd Hm Jan 3.29 102 4.23 131 Feb 4 112 5.2 145 Mar 4.58 142 6.14 190 Apr 4.62 139 6.3 189 May 4.86 151 6.76 210 Jun 5.08 152 7.19 216 Jul 5.24 162 7.51 233 Aug 5.14 159 7.4 229 Sep 4.62 139 6.48 195 Oct 4.19 130 5.72 177 Nov 3.55 106 4.66 140 ACCEPTED MANUSCRIPT 21 Dec 3.12 96.7 4.02 125 kWh kWh kWh/m2 kWh/m2 Yearly average 4.36 133 5.97 182 Total for year 1590 2180 Table 5. Solar irradiation and energy production (optimal orientation and inclination). Ed and Em are the mean daily and monthly values of the energy production in kWh/kWp; Hd and Hm are the mean daily and monthly values of global irradiation received by the panels in kWh/m2. On the other hand, the sun elevation angles for the 21st day of every month are presented in Table 6. Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Sun elevation (o) 32 42 53 64 72 76 73 64 53 42 33 29 Table 6. Sun elevation angle. The solar thermal and PV panels have been chosen from the current market product brochures:  Photovoltaic panel: SPR-E19-320, SunPower E-Series Residential Solar Panels, nominal peak power 320W, efficiency 19.9% and size 1558x1046x46 mm.  Solar thermal panel: T20MS, Termicol Gama Excel Familia T-MS, Zero-loss efficiency 81.2%, 1st and 2nd order heat loss coefficient a1=3.812 W/m2k and a2=0.021 W/m2k2 respectively and size 2130x970x83 mm. The application of the optimization algorithm calculates the total energy production by the PV panels installed in vertical arrangement (length>width) on a model surface of 100m2 as a function of the separation distance between panels, as Figure 3 shows. ACCEPTED MANUSCRIPT 22 Figure 3.Electric energy production depending on the separation distance between panels. As can be noted from the above illustration, the electric energy production increases with the separation between panels, until the distance is so large that a row of panels must be removed, which explains the saw-teeth shape of the curve. Accordingly, the optimal theoretical separation distance in Seville is 0.432m, which is rounded in practice to 0.4m. Also, the annual utilisation factor of the panel due to the shadow effects, as a function of the spacing distance, is shown in Figure 4, calculated considering the sun elevation and the shadow projection over the panels behind, for every day during a whole year. ACCEPTED MANUSCRIPT 23 Figure 4: Utilisation factor due to shadow projection over the panels. The number of panels that can be installed in the 100m2 model surface is 54, yielding a mean surface of 1.85m2 per panel that, considering the technical performances of the chosen PV solar panels, see above, amounts to 176 W/m2. Therefore, the total energy production, considering shadow effects, is 24.794 MWh/year, which is roughly 10% less than the theoretical yield (27.475 MWh/year ignoring shadow). For pitched roofs, the same spacing has been assumed to facilitate maintenance operation, considering also the limited available surface to install panels in this case owing to the typical trapezoidal shape of these roofs. An additional simulation was carried out, with panels arranged horizontally (width>length), providing worse results. ACCEPTED MANUSCRIPT 30 3.2 Demand Estimation. 3.2.1 Current electric energy consumption. The current electric energy consumption of the city of Seville is 2,540,292 MWh which has been obtained from Anuario Estadístico de la Provincia de Sevilla, 2018. The total energy is prorated among the 8760 hours of the year in accordance to the profile consumption curves provided by the Spanish Transmission System Operator, REE, as shown in Figure 10. Figure 10. Current daily electric consumption. 3.2.2 Electric vehicle energy consumption. The annual EV energy consumption is expected to be 890,013 MWh, as shown in Table 7. The number of each vehicle type has been obtained from Anuario Estadístico de la Provincia de Sevilla, 2018; the annual driven distance has been compared among different - 1,000.00 2,000.00 3,000.00 4,000.00 5,000.00 6,000.00 7,000.00 8,000.00 9,000.00 Electric consumption (MWh) Day Daily electric consumption ACCEPTED MANUSCRIPT 31 sources for several countries (National Statistical Institute8, Federal Highway Administration 9 , Department for Transport 10 , Dirección General de Tráfico 11 and Real Academia de Ingeniería de España, 2018); the specific consumptions are calculated by the mean of the specific consumption of several EVs in the current market for each type. Vehicle type Amount Driven distance (km) Specific consumption (Wh/km) Electric consumption (MWh) Cars 320,571 12,500 163 653,163 Motorbikes 56,804 7,700 68 29,743 Vans 16,041 19,500 235 73,508 Buses 1,090 55,000 1,300 77,935 Motorcycle s 41,660 11,000 32 14,664 Total 436,166 105,700 1,798 849,013 Table 7. Electric vehicle energy consumption. As done with the current electricity demand, the total annual energy consumed by EVs is prorated among the 365 days of a year in accordance to the daily consumption curves provided by Postma, 2014, as shown in Figure 11. 9 https://www.fhwa.dot.gov/ 10 https://www.gov.uk/government/organisations/department-for-transport 11 http://www.dgt.es/ ACCEPTED MANUSCRIPT 32 Figure 11. EV daily consumption 3.2.2 Thermal energy consumption The thermal energy consumption has been obtained from the Plan de Acción por el Clima y la Energía Sostenibles de Sevilla, 2018. In the city, the total consumption of natural gas and liquid petroleum gas (LPG) are 22.2 ktep and 21.9 ktep, respectively, which are equivalent to a yearly total of 512,883 MWh (roughly 20% of the electrical demand) and a monthly mean of 42,740 MWh. This demand considers all the industrial thermal energy processes, as mentioned in the description of the methodology. 3.3 Balance analysis 3.3.1 Energy balances Electric only energy balance As mentioned before, the selection of the surface to install PV panels is guided by their relative performance in accordance to the surface orientation (Figure 7). In line with this assumption, the flat surfaces and those oriented to south are first chosen, and then those oriented to east and west (north faced rooftops surfaces are not used), yielding a total of - 20 40 60 80 100 120 140 160 180 200 01:00 02:00 03:00 04:00 05:00 06:00 07:00 08:00 09:00 10:00 11:00 12:00 13:00 14:00 15:00 16:00 17:00 18:00 19:00 20:00 21:00 22:00 23:00 23:59 MWh Hour EV daily consumption ACCEPTED MANUSCRIPT 33 2,613,603 MWh. For the city of Seville, this is more than sufficient to meet the current annual demand (2,540,292 MWh), with a residual overproduction (73,111 MWh) that can be eventually used to compensate for the losses associated with the charge/discharge cycles of the storage system. When the future EV demand is also considered (849,013 MWh), an additional surface of 3.5 km2 of rural area (blue area in Figure 2) is necessary to satisfy the total demand of the city. Excluding the northfacing sloped roofs, due to its low efficiency, the additional PV panels must be accommodated on extra land surface. If the electricity produced by the wastewater treatment stations is considered, then the PV panels oriented to the west would not be actually needed for the current demand. In this case, the yearly electricity production would reduce to 2,541,982 MWh, still sufficient to satisfy the current demand of Seville if storage losses are ignored. Figure 12 graphically summarizes the available surface and the PV generation potential for each orientation, as well as the potential of wastewater treatment plants, the total demand of Seville and the balance achieved in both cases. Note that, should Seville be located at higher latitudes, e.g. in Germany, or should the electricity demand increase with respect to current values, which can be expected in a more electrified and efficient future, then the urban rooftop surface would not suffice to achieve a 100% sustainable ACCEPTED MANUSCRIPT 34 city. Figure 12. Electric energy balance. Electrical and thermal energy balance According to the assumptions made in the methodology, the thermal solar panels installed on the flat rooftop surfaces need 707,686 m2 to satisfy the thermal demand throughout the year. Note that this panel surface is determined by the consumption of December, the least favourable month for solar energy production. ACCEPTED MANUSCRIPT 35 The utilisation of the remaining rooftop surfaces for PV panels, along with the electricity provided by the wastewater treatment plants, yield enough electricity to satisfy the current electric consumption of Seville, with a residual overproduction of 56,951 MWh. For the EV demand, the same additional peripherical area will be necessary to cover its consumption. It is worth noting that, as the thermal panels are sized according to the demand of December, there is an overproduction of thermal energy during the remaining months, which in summer could be used for instance for cooling. 3.3.2 Energy storage system The base case scenario is the one in which sufficient storage is deployed to achieve a perfect hourly balance between the PV production (on flat rooftop surfaces and those oriented to south, east and west) and the electricity demand throughout the year (the electricity produced by wastewater treatment plants is ignored in this analysis). The results obtained by the application of the MATLAB algorithm described in the previous section are shown in Figure 13. ACCEPTED MANUSCRIPT 36 Figure 13. Hourly electric energy balance (Storage sizing). The pseudo-sinusoidal curve represents the algebraic sum (integration) of the energy unbalances for every hour of the year, starting on the 1st of August at 00:00. Considering the charge/discharge losses, the size of the storage system is 293,438 MWh (8.7% of the annual demand), leading to an overproduction of 2,738 MWh at the end of the integration period. Additionally, the graphic shows the periods in which the storage system is charged or discharged. The charging period (daily PV production higher than demand) lies between the 25th of February at 9:00 and the 17th of October at 18:00, the remaining time being the net discharge period. The second scenario considered in this work, as defined above, assumes the same PV system as in the base case, but the storage system is dimensioned just for the daily needs. For this purpose, the day with largest negative balance, i.e. largest negative area under the net load curve, is analysed, which corresponds to June 28. That day, the total negative balance (i.e., surplus PV energy to be stored) is 5,006 MWh, so the size of the storage system, considering losses, should be 5,269 MWh. This represents less than 2% of the ACCEPTED MANUSCRIPT 37 storage system capacity arising in the base case scenario, fully dimensioned for the worstcase seasonal unbalance (100% renewable self-supply). When the whole year is considered, such daily storage system allows 3,133,985 MWh of PV energy to be self-consumed, representing about 92% of the total consumption of Seville. This somewhat counterintuitive result should be remarked: with just 2% of the full-size storage system, a 92% sustainable system is achieved, with major implications in terms of investments (see below). In this scenario, 8% of the energy must be supplied from the grid during less sunny seasons, but barely the same amount will be in excess during summer, which could be applied to alternative uses of electricity or sold in wholesale markets. The last scenario (dayly storage 100% ECO) considered in this work aims to achieve a 100% renewable production system by increasing the surface devoted to PV panels beyond that strictly available on building rooftops, i.e., by resorting to non-urban area (red area in Figure 2), while at the same time keeping a reasonably small storage system capable of balancing the least favourable day. For the city of Seville, the critical day is December 14, leading to an increase of PV peak power from 2,477 MWp to 3,919 MWp (158% that of the base scenario) and a storage system of 6,715 MWh (2.3% that of the base scenario). Compared to the base scenario, the ambitioned sustainability is achieved by trading more PV panels than those strictly required to achieve the annual balance for much less storage devices, provided the required land for the extra panels is somewhere available, as in the case of Seville. Note that this intermediate scenario can be very attractive for a city like Seville, located at relatively low latitudes, but not necessarily elsewhere. ACCEPTED MANUSCRIPT 38 4. Economic analysis. The assessment is completed with an economic analysis of the impact of the three scenarios from the customers perspective, keeping in mind the current retail prices of electricity. First of all, the following hypotheses are established:  The thermal energy balance is not considered in the analysis. Indeed, as a large enough fraction of the available roof surface is reserved for thermal panels, the thermal subproblem is decoupled from the major electricity procurement problem.  As the wastewater treatment plant generation represents just 2.5% of the total demand, only the PV panels are considered as electric energy production systems.  The PV panels are installed on all flat surfaces and those oriented to south, east and west, as considered in the base case scenario. Rooftops facing north have not been considered, due to their reduced efficiency.  The depreciation periods are 20 years for the solar panels and 10 years for the liion batteries.  No public incentive or aid is considered in the analysis Full storage Daily storage Daily storage (100% eco) Photovoltaic system Installed Power (MWp) 2,477 2,477 3,919 Annual production (GWh) 3,480 3,480 5,375 Roof surface (km2) 10.8 10.8 10.8 No roof surface (km2) 3.5 3.5 11.2 ACCEPTED MANUSCRIPT 39 Table 8. Results of the economic analysis. In accordance with Moskowitz, 2018, Sussams and Leaton, 2017 and Henbest, 2018, the market prices for the installed PV power and the li-ion batteries are about 1.00 €/kWp and 100 €/kWh, respectively. As can be seen in Table 8, the full storage system allows to balance all the demand throughout the year, but the resulting mean cost per kWh is extremely high when 12 http://ec.europa.eu/eurostat Total PV investment (M€) 2,477 2,477 3,919 Depreciation (M€/year) 124 124 196 Storage system Capacity (MWh) 293,438 5,269 6,715 Total Storage investment (M€) 29,344 527 672 Depreciation (M€/year) 2,934 53 67 Total investment (M€/year) 3,058 177 263 Self-supplied energy (GWh) 3,389 (100%) 3,134 (92%) 3,389 (100%) Energy fed into the grid (MWh) 0 255,3 1.901,7 Energy taken from the grid (MWh) 0 255,3 0 Energy cost (c€/kWh) 90.2 5.6 7.8 Mean retail electricity price in Spain in 2016 (c€/kWh)12 Household 22.8 Industry 10.3 Average 14.5 ACCEPTED MANUSCRIPT 46 45) SunCalc application web https://www.suncalc.org 46) Unite Kingdom Government, Department for transport, 2017 https://www.gov.uk/government/organisations/department-for-transport 47) United Nations. Population Division – World Population Prospects. 2017. https://esa.un.org/unpd/wpp/DataQuery/ ACCEPTED MANUSCRIPT