- Email: [email protected]
The importance of facades for the solar PV potential of a Mediterranean city using LiDAR data
Renewable Energy 111 (2017) 85e94
Contents lists available at ScienceDirect
Renewable Energy journal homepage: www.elsevier.com/locate/renene
The importance of facades for the solar PV potential of a Mediterranean city using LiDAR data ~es, C. Catita, P. Redweik M.C. Brito, S. Freitas*, S. Guimara Instituto Dom Luiz, Faculdade de Ci^ encias, Universidade de Lisboa, 1749-016, Lisboa, Portugal
a r t i c l e i n f o
a b s t r a c t
Article history: Received 7 October 2016 Received in revised form 13 March 2017 Accepted 28 March 2017 Available online 1 April 2017
The aim of this work is to discuss the potential of facades and other vertical features for the photovoltaic potential of the cityscape. The photovoltaic potential in two representative case studies in the city of Lisbon, Portugal, is computed using a digital surface model determined from LiDAR (Light Detection And Ranging) measurements and local typical meteorological year time series. Results are compared with estimated local electricity demand derived from the population distribution. The annual analysis shows that roof and facade PV potential exceeds the local non-baseload demand and can contribute to 50e75% of the total electricity demand. Hourly breakdown shows peak PV power can only achieve winter mid-day electricity demand if the solar potential of facades is also taken into account. Its added value for off-peak PV supply is less significant in winter since non-south facades are not particularly exposed. In summer, however, facades can satisfy non-baseload morning and afternoon demand. A conservative economic analysis shows payback times below 10 years can only be achieved with PV on roofs while a 50/50 mix would lead to payback times of 15 years. © 2017 Elsevier Ltd. All rights reserved.
Keywords: Solar potential GIS Facades Demand
1. Introduction In an era of growing urbanization when most of energy demand is concentrated in cities, energy sustainability requires that a significant fraction of this energy demand can be fulfilled with local, clean and abundant sources of energy. As it has been highlighted by Hernandez [1], solar power is an unavoidable piece of the fabric of sustainable cities; solar power is plentiful in most regions of the globe, it is a renewable source of energy and CO2 emission free. However, solar power has relatively low energy densities, thus requiring considerably larger areas to produce relevant amount of electricity. As modern cities are characterized by high density populations, living in high rise buildings, the available roof area becomes in short supply for solar power to fulfill the local energy demand. As such, building facades offer an attractive and complementary option. Although vertical solar panels receive less solar radiation than roofs and horizontal surfaces, in particular in the summer months, and are more affected by the compactness of the urban layout [2], facades feature high areas; in a building with 4 floors, the area of
* Corresponding author. E-mail address: [email protected] (S. Freitas). http://dx.doi.org/10.1016/j.renene.2017.03.085 0960-1481/© 2017 Elsevier Ltd. All rights reserved.
the facades is about 4 times the area of the roof1. If the whole available area of such building was used for solar panels and shadings from neighboring buildings neglected, the total annual electricity production would triple that of the roof. This ratio will obviously increase further for taller buildings. However, it is important to point out that in this example, the cost of the generated solar electricity (in V/kWh) would also be 4 times higher. Hence, the deployment of PV on less than optimum inclination/ orientation has to be weighted by economic constraints. Nevertheless, the recent trend of fast decreasing costs of PV, which is expected to proceed in future years [3], opens a window of opportunity for this type of applications. Additionally, vertical PV facades will produce relatively more power in winter and less in summer, and more in the early and late hours of the day, when the sun is lower in the sky. Since a building will typically have four, or at least two, exposed facades with opposite orientations, the different solar facades of a building will produce at maximum power at different times of the day. As has been recently highlighted by Hummon et al. [4], this leads to a widening of the peak of power production throughout the day and
1 Assuming a 10 10 m2 area building with 4 storeys, each 3 m high; hence, the roof area is about 100 m2 while the facade area is 4 facades 4 storeys 3 m/ storey 10 m length ¼ 480 m2 facade area.
86
M.C. Brito et al. / Renewable Energy 111 (2017) 85e94
hence a closer match to the load diagram, which can then result in significant savings in electricity storage and/or fossil fuel based backup power. Also in Ref. [5] the importance of adjusting tilt and azimuth of PV installations, with higher tilts being the preferable option for south facing scenarios, is highlighted as a means to spread the electricity generation over a larger period of time. It is also worth mentioning that the lower level of solar irradiation on facades has prompted consideration of different concepts and practical solutions for increasing energy harvesting on vertical surfaces. These include reshaping of the facades themselves [6] or the development of concentration photovoltaic systems for application in buildings [7]. Furthermore, the use of facades for PV generation, and in particular Building Integrated Photovoltaics (BIPV) [8] – defined as photovoltaic cells integrated into the building envelope as part of the building structure – offers interesting opportunities as it can replace conventional building materials whilst creating a harmonious architecture by blending into the design of the building [8,9]. Since it replaces other construction materials, its net cost (per unit area) is already cost-effective in a number of applications [10]. As an added benefit, the air flow behind the solar modules reduces their temperature improving the system efficiency and longevity [11]. Another approach is the use of semitransparent BIPV which allows daylight to enter the interior spaces [12]. The purpose of this work is to analyze the cityscape PV potential, including both roofs and facades, and to understand when facades might act as added advantage to the rooftops. The methodology is described in section 2. Using Light Detection And Ranging (LiDAR) data to describe the topography of the city, including both the terrain and buildings, and reference meteorological data, the PV potential is determined for two case study areas in the city of Lisbon, Portugal, presented in section 3. The two different areas are characterized by differing building morphologies. Then, hourly PV generation potential of roofs and facades is compared with estimated local electricity demand. Results are presented in sections 4, 5 and 6, which discusse the overall PV potential of facades and the added value of better adjustment to the load diagram. In the end, conclusions are drawn regarding the contribution of PV facades to the solar potential of a city. 2. Methodology The estimation of the cityscape PV potential is based on a 3D PV potential tool which uses LiDAR data and reference meteorological data to determine the expected insolation of all points on the ground, roof and facades of buildings in the study areas. The discussion entails the comparison of the PV generation potential with the local electricity demand, which is determined from the population distribution and the per capita average load. 2.1. 3D photovoltaic potential The estimation of solar potential in the urban environment typically requires combining radiation models, accounting for the apparent movement of the sun in the sky, as well as the beam and diffuse fractions of the global solar irradiation, and Geographical Information Systems (GIS) to describe the opaque obstacles met by beam irradiation (i.e. shadow casting) and sky view factors (SVF) that determine the amount of diffuse irradiation. For an overview of solar potential models in the urban environment see Freitas et al. [13]. In this study, the solar power potential of roofs and facades is determined using the SOL algorithm, a software tool described in Redweik et al. [14] and Catita et al. [15]. The algorithm starts from a geo-referenced LiDAR data cloud, re-sampled for a 1 1m2 raster. As vertical surfaces are not directly detected by the aerial LiDAR
sensor, which only measures the surface height, they are decomposed into hyperpoints, i.e. each XY point is used to produce a set of 3D points sharing the same XY coordinates (cartographic coordinates) but with different Z (elevation) located in space along a vertical column on a facade. The local typical meteorological year (TMY) data set associated to the SOLTERM database is used. This includes hourly mean values for horizontal direct and diffuse irradiation calculated over 30 years of local observations. The hourly global irradiance, G, on any point of the tilted surfaces is determined by its two main components: direct and diffuse irradiance (eq. (1), adapted from Ref. [16]).
G ¼ Gbh
cos q SC þ Gdh Fd SVF cos qz
(1)
where Gbh and Gdh are respectively the hourly direct and the diffuse horizontal irradiance components, SC denotes a Shadow Caster b is the transposition factor for isotropic attribute (0 or 1), Fd ¼ 1þcos 2 diffuse radiation, SVF denotes the Sky View Factor (i.e. the fraction of sky visible from a point), qz is the sun's zenith angle and q is the angle of incidence of the sun rays on the tilted plane calculated according to the following equation [16]:
cos q ¼ cos d sin b sin g sin u þ ðcos f cos b þ sin f sin b cos gÞcos d cos u þ ðsin f cos b cos f sin b cos gÞsin d
(2)
where d is the declination angle, f is the latitude of the location, b is the surface tilt, u is the hour angle and g is the surface azimuth. The model does not consider reflected light. At any given time, a shadow algorithm takes each point of the Digital Surface Model (DSM), including trees, as a shadow caster along the line opposite to the direction of the sun. As long as this line is not interrupted, i.e. whenever a DSM cell along that profile features a Z value lower than the shadow line height at that XY position, the pixel is in shadow and receives the Shadow Caster attribute 0, creating then a binary map. For facades the process is further refined by taking into account shadow height and behavior of the neighbourhood. The result is that at any given time, points belonging to the same facade hyperpoint may have different Shadow Caster attributes, i.e. such a hyperpoint may be totally shaded, partially shaded or totally unshaded. The diffuse irradiation on a given point is constrained by its SVF, a measure of the amount of sky seen from the location. For the same facade, ground floors normally have a lower SVF than top floors because of an increase in view obstructions to the sky. Therefore, even when a point is in shadow, diffuse irradiance is still available according to its SVF. This is taken into consideration in the SOL algorithm whereby the SVF is calculated for each hyperpoint element (as well as for the roofs and ground). The SOL algorithm delivers hourly irradiation data on every point of roof and facade for every day of a year in form of maps and tables, which can be used to obtain solar irradiation estimates for different periods and timescales. The PV potential mapping is determined by multiplying the solar irradiation by an efficiency of 15% for a typical module. Because the effect of module temperature on the efficiency of silicon based modules is not negligible, this is also taken into consideration [17]:
hmod ¼ href 1 Dh Ta þ
TNOCT 20 G 25 800
(3)
where hmod is the module operating efficiency, Dh is the
M.C. Brito et al. / Renewable Energy 111 (2017) 85e94
temperature coefficient for efficiency (0,45%/ C), Ta is the ambient temperature in C, NOCT is the Nominal Operating Cell Temperature (47 C). For simplicity, partial shading is not considered as this correction is not expected to be very relevant since the spatial resolution (1 point per square meter) is of the order of magnitude of the area of the considered PV module. If a finer grid was used, the impact of the partial shading effect on the solar yield would become important but its impact can be minimized by a number of techniques, as reviewed by Bidram et al. [18]. This methodology includes some conceptual simplifications and limitations which are worth discussing. On the one hand, the inclination and orientation of the solar PV panels are considered to be those of the building surfaces where they are installed. Thus, PV modules on facades are vertical with the orientation of the facade while PV modules on the roof have the inclination of the roof. This option will underestimate the solar potential. Indeed, one could envision PV modules with an optimum tilt mounted on shading sunscreens (for the case of facades) or on flat roofs. Although it depends on the local architecture, Verso et al. [19] suggests that the option for optimum tilt on flat roofs could represent a 12% increase in the solar potential of an urban area. For the facades, the vertical assumption (instead of tilted sunscreens) could represent an underestimation of the solar potential of c.a. 20%. On the other hand, and for the particular case of facades, this methodology does not take into consideration the existence of windows because LiDAR input data does not provide any information on features of the vertical surfaces. Although windows may be used as solar active area, e.g. using semi-transparent PV modules, the average conversion efficiencies would be about ½ of opaque standard PV modules. With a rough estimate of 30% of glazed area and 7.5% conversion efficiency means that the disregard for the windows will overestimate the facades potential in about 15%. The result is that this more or less balances the underestimation discussed above. Nevertheless, and in order to accommodate these uncertainties, we have conservatively considered an 80% performance ratio for both roof and facades. It ought to be underlined that the approach developed in this methodology, suitable for the large scale analysis of an urban region, requires simplifications regarding local conditions that may vary significantly among particular applications, such as wind speed, reflected radiation, partial shading, air temperature, etc. As demonstrated by Yoo [20], these may strongly impact the solar energy yield of a particular building or façade. As such, the approach used here is not appropriate for individual buildings. 2.2. Electricity demand The electricity demand is determined by estimating the population distribution and multiplying it by the average electricity demand. This is a typical top-down approach, where there is no concern for the stochastic behavior of electricity demand at the individual level. Since the electricity consumption is distributed and it is not constrained to the building itself, the importance resides on the per capita aggregated electricity demand, which encompasses electricity consumption beyond the domestic individual end-users. As discussed in Swan and Ugursal [21], a bottom-up alternative based on the extrapolation of energy consumption of representative individual houses would refine the understanding of the details associated with the energy consumption. However, this approach would require data (e.g. statistical data on household occupancy, appliances availability and use, etc [22]) which was not available for the present study. To determine the population distribution, the census population at block group level [23] is disaggregated by building using the ‘residential volume’ as a proxy variable. The procedure starts by
87
using the LXI [24] to characterize the use of each building in the study area as ‘residential’ or ‘non-residential’. Furthermore, residential homes may be distinguished according to the type of building, e.g. houses or apartments. An additional restrictive criterion is implemented: only buildings with at least one floor above ground (with a mean height above 2.5 m) are considered suitable for habitation, and may therefore receive resident population. The resident number per building was determined by the method proposed by Ural et al. [25]. The set of buildings receives a proportional share of the resident population based on the ratio of their volume and the total resident volume in the block group, which is then weighted by the height of the building. The electricity demand of each building was then calculated by multiplying the estimated number of inhabitants by the per capita electricity demand. For that purpose, we considered the hourly load diagrams for a full year [26] normalized by the number of inhabitants in the country. The flowchart in Fig. 1 illustrates the process described above. In the discussion below (c.f. section 3.3), the PV generation potential is also compared to the non-baseload load diagram. As the baseload is defined as the minimum level of demand on an electrical grid over 24 h the non-baseload hourly profile was determined by subtracting from the hourly demand load the minimum load of the previous 24 h. This minimum is expected to occur during the night period, when electricity demand tends to be lower. 3. Methodology implementation through case study The methodology is implemented in two representative albeit dissimilar neighborhoods in the city center of Lisbon, Portugal. Area 1 was built in the 1950e1960s and features high rise buildings with about 10 floors (Fig. 2). On the other hand, Area 2 is located in a parish originally developed in the early 20th century, with narrower roads and high trees. Typical buildings have three to five floors and inner quadrangle. The total number of residential buildings in Area 1 and Area 2 are 244 and 294, respectively. For both cases, the modelled area is 500 500 m2. €ppen climate classification scheme, climate According to the Ko in Lisbon is Subtropical-Mediterranean climate, with short and very mild winters and warm summers. A summary of the most important features of both Areas is presented in Table 1. Using the methodology described in section 2.2, the building's resident number and electricity demand were estimated for the two case study areas. Results are shown in Fig. 3. One can observe that in Area 1, the population density and hence the electricity demand is higher in the central high rising buildings. Area 2 is more compact and uniform. For Area 1, the resident number was estimated at 3313 and therefore the overall electricity demand was estimated at 63 GWh/year. As for Area 2, the estimated resident number is slightly higher at 3742 inhabitants resulting then in an also slightly higher estimated electricity demand of 72 GWh/year. 4. Annual energy production For the PV potential in roofs and facades of the two areas of interest, the methodology described in section 2.1 was implemented. Solar radiation and PV power was determined for every hour of the day, for all ground and rooftops raster points and for all vertical surface hyperpoints. Fig. 4 maps the annual solar irradiation of the two areas of interest. Its most conspicuous feature is the fact that roofs and ground are clearly superior to vertical facades, which have significantly lower levels of irradiation, both in Area 1, with high rise and generally unobstructed buildings and in Area 2, with a compact
88
M.C. Brito et al. / Renewable Energy 111 (2017) 85e94
Fig. 1. Flowchart of the methodology followed for electricity demand estimation.
Fig. 2. Street view and aerial view of case study areas: Area 1 (top) and Area 2 (bottom).
arrangement of 3-story buildings. A more detailed observation of Fig. 4 reveals that south-facing
facades feature higher annual yields than east- or west-facing facades (the latter are not seen in this particular perspective view).
M.C. Brito et al. / Renewable Energy 111 (2017) 85e94
89
Table 1 Summary of buildings characteristics for Areas A and B.
Area A Area B
Residential footprint area [%]
Average building height [m]
Maximum building height [m]
Rooftop area [m2]
Façade area [m2]
13 19
18 20
67 51
47133 32850
135902 98372
Fig. 3. Annual per building electricity demand for Area 1 (left) and Area 2 (right).
Fig. 4. Annual solar irradiation for Area 1 (left) and Area 2 (right).
This effect is easily observed on the central neighborhoods in Area 2. On the other hand, the variation of PV potential within the same facade is of interest; this is clearly seen on the high rise buildings in Area 1, where a certain degree of mutual shading in the lower floors is present. The distribution of the solar potential and the range of values achieved can be better understood from Fig. 5, which presents the histogram of the annual solar potential for Area 1 and Area 2, both for roofs (dashed lines) and facades (solid and colored lines). These results confirm that roofs feature much higher levels of irradiation, from 1000 to 1800 kWh/m2/year, depending on their particular orientation and inclination. Facades, on the other hand, typically receive solar radiation only in the 100e1000 kWh/m2/year range, depending on their orientation (higher annual irradiation on south facing facades and the lower on north facing facades) and shading from neighboring buildings and trees. In both cases, east and west facing facades feature a wide range of irradiation levels, from 100 to
800 kWh/m2/year, and the sum of their potential might surpass those facing south. It is worth mentioning that the inner quadrangles and indented topography of the buildings in Area 2 lead to a higher facade area which leads to higher total facade irradiation but higher mutual shading, and therefore feature lower average irradiation density. Fig. 6 shows the accumulated monthly PV potential on roofs (in dark brown) and facades (different shades of light brown, according to 4 different classes: irradiation above 900 kWh/m2/year, between 700 and 900, between 500 and 700, and below 500 kWh/m2/year). Also shown is the estimated local electricity demand; the solid line describes non-baseload demand while the dashed line represents the full load demand for each area. The discussion below focus on the non-baseload demand. This implicitly assumes that other (nonurban) renewable energy sources, such as biomass, large hydro or wind (if associated to storage) would satisfy the baseload demand. Results show that the PV potential in Area 2 is more favorable
90
M.C. Brito et al. / Renewable Energy 111 (2017) 85e94
Fig. 5. Annual solar potential histogram for Area 1 (left) and Area 2 (right). Dashed lines refer to roofs and solid lines to facades with different colors according to South, East, West and North orientation, i.e. points with azimuth inside the intervals [45 , 135 [, [135 , 225 [, [225 , 315 [ e [315 , 45 [ where 0 denotes North. (For interpretation of the references to colour in this figure caption, the reader is referred to the web version of this paper.)
Fig. 6. Monthly PV potential (roofs: dark brown column; facades: lighter brown columns according to 4 different classes: above 900 kWh/m2/year, between 700 and 900, between 500 and 700, and below 500 kWh/m2/year) and electricity demand (blue solid line: non-baseload monthly electricity demand; blue dashed line: monthly total electricity demand) for Area 1 (left) and Area 2 (right). (For interpretation of the references to colour in this figure caption, the reader is referred to the web version of this paper.)
than in Area 1, with higher PV potential for all months of the year. In both areas, and in annual terms, the roof PV potential exceeds the local non-baseload demand and can contribute to 26% (Area 1) and 36% (Area 2) of the total electricity demand. Furthermore, if the potential of PV facades is added to that of the roofs, the total PV potential increases 54% for Area 1 and 73% for Area 2. Closer inspection of Fig. 6 shows that for the summer months the total roof PV potential exceeds the non-baseload demand in both areas. Also, if both facades and roofs are taken into account then the buildings' PV potential is of the order of magnitude of the total load demand; for Area 2 this threshold is even reached for all of the summer months. These results highlight the relevant role that building integrated photovoltaics can provide to the electricity system in urban environments. On the other hand, for the winter months, solar irradiation decreases significantly while the load demand increases slightly, due to shorter and colder days which require more artificial lighting and some heating. Thus, it is not surprising that the total roof PV potential is not able to satisfy the non-baseload electricity demand
Payback ½years ¼
during 5 months in Area 1 and 4 months in Area 2. However, this non-baseload demand would be satisfied if the total PV potential of facades was deployed (Table 2). As discussed in the following section, it is not reasonable to expect that all facades could be covered with PV modules at current or foreseen PV costs in the near or medium term. 5. Payback time analysis The classification of facades according to yearly solar irradiation (the different shades of brown in Fig. 6) may be interpreted as different classes of financial payback of investment (Table 3); hence, darker brown (above 900 kWh/m2/year) are cost-effective PV investments even in today's market conditions, where one may assume an average cost for a rooftop PV system of 300 V/m2 [10], and a grid selling price of 0.1555V/kWh [27]. Lighter browns (below 500 kWh/m2/year) could only be interesting as an investment when PV costs dropped significantly. The Payback time estimation is detailed in eq. (4).
300 V m2 System class kWh m2 year href PR 0:1555½V=kWh
(4)
M.C. Brito et al. / Renewable Energy 111 (2017) 85e94
91
Table 2 Annual energy demand and solar electricity production for the different system classes represented in Fig. 6.
Total load demand [GWh/year]
including baseload not including baseload Roofs Façade
Energy production [GWh/year]
1800 kWh/m2/year 500 kWh/m2/year 500e700 kWh/m2/year 700e900 kWh/m2/year 900 kWh/m2/year
Roofs and façades
Table 3 Financial payback time of investment for an average rooftop system and the threshold of the different facade classes. System class Rooftop Facade
Payback time (years) 1800 kWh/m2/year 900 kWh/m2/year 700 kWh/m2/year 500 kWh/m2/year
8.9 17.9 23.0 32.2
where href is 0.15 the typical efficiency of a 1 m2 module, PR is a performance ratio of 0.80 and System class can be found in Table 3. It is interesting to note that the combined payback of roofs and high performing facades (above 700 kWh/m2/year) would be 19.5 and 18.7 years for areas 1 and 2, respectively. Furthermore, Table 4 shows the optimum mix of PV systems on roofs and facades for different levels of combined payback times. The results of Table 4 were obtained by ordering the points on roofs and facades according to their payback time (lower paybacks first). The global payback is the weighted average of the paybacks, where the weights are the fraction of the area with that particular payback. Results provided by the model show that for both areas payback times below 10 years can only be achieved with PV on roofs whilst a 50/50 mix would lead to payback times of 15 years. These long payback times confirm that, with current costs, PV on facades can only be attractive in very particular conditions. However, expected significant reductions in PV costs in the medium term (halving costs by 2030 [3]) will trigger the mass deployment solar facades in the urban environment. In this context, the results presented in Fig. 6 show that, for Area 2, non-baseload electricity demand could be satisfied by cost effective PV investments on roofs and facades for 10 months of the year. For Area 1, where most facades have modest yearly performance (mostly east- and west-facing facades generating less than 500 kWh/m2/year) the contribution of cost-effective PV deployment on facades would only be marginal, reducing the deficit from 5 to 4 months. It ought to be noticed that the estimation of these payback times critically depend on the average efficiency of the modules. If 20% PV efficiency was to be assumed, as it is expected by 2025, payback times for rooftops would decrease to just over 6 years and best locations in facades would reach about 13 years paybacks. This would translate to 38% and 37% of PV on facades for a combined
Table 4 Mix of roof and facade PV systems for different combined payback time periods. Combined payback time
10 12 15
Area 1
Area 2
Roofs
Facades
Roofs
Facades
100% 78% 50%
e 22% 50%
100% 78% 48%
e 22% 52%
Area A
Area B
17.9 4.2 4.6 1.7 1.9 1.1 0.4 9.7
15.9 3.7 6.4 2.4 2.1 1.4 0.7 13
payback time below 10 years, for Areas 1 and 2, respectively. Fig. 6 also shows how the non-baseload electricity demand may be satisfied by solar electricity, in particular in the summer months. These results also highlight the need for alternative (not solar) fuels for the baseload production since the non-baseload only accounts for about ¼ of the total load demand. Coincidentally, a very similar plot would be that of the monthly residential load. In fact, for the district of Lisbon, 28% of the electricity demand is estimated to be associated to residential demand (data for 2012, c.f. [28]). Hence, these results show that the PV generation would satisfy the residential electricity demand of the city (on a monthly or annual basis, not hourly, since there is residential demand during night periods) and therefore one may conclude that PV on roofs and facades can be a critical tool to achieve Net Zero Energy Buildings in the urban environment.
6. Hourly photovoltaic supply For the analysis of the impact of non-optimum inclination and orientation of PV facades and, thus, their off-peak production, it is informative to observe Fig. 7, which shows for the two areas under study the solar radiation at midday in a winter day and 09:00 LST on a summer day. At midday in the winter day, with relatively low overall solar irradiation, south facing facades have higher irradiation than roofs (or ground) while east and west-facing (not seen in the images due to the perspective) facades receive little solar power at this time. On the other hand, for the early hours of a summer day, the results show that the solar potential is much higher than the winter peak potential, even at this early time of the day. It is also clear that roofs (and ground) have much higher solar irradiation density than facades, regardless of their orientations. Fig. 8 shows the hourly balance between local electricity demand and the photovoltaic potential in roofs and facades, for typical days of summer and winter in the two areas under analysis. For the winter day, the results confirm that due to the low levels of solar irradiation and the relatively shorter days, the overall PV supply is significantly lower than the local daily electricity demand for both areas. Also, peak PV power can only achieve mid-day electricity demand if the solar potential of facades is also taken into account. One can observe that the added value of off-peak PV supply is not significant since non-south facades are not particularly exposed during winter days. In fact, the peak production of the facades is reached at the same time of the peak production of roofs, which is not coincident with the morning or the evening demand peak, typically associated with illumination. As for the summer day, the total roof-only potential peak power exceeds demand at midday. It is clear that the facade-only potential peak power exceeds morning demand in both areas; this effect is much more pronounced in Area 2. Peak production of east- and west-facing facades at 9am and 4pm (morning and afternoon) respectively is evidently higher that
Fig. 7. Solar radiation for Area 1 (left) and Area 2 (right) at 12:00 LST on December 21st (top) and 09:00 LST on June 21st (bottom).
Fig. 8. Hourly electricity demand (dark blue line) and photovoltaic potential of roofs (black dashed line), all facades (black solid line), south facades (orange), east facades (yellow), west facades (green), north facades (light blue) and roofs and facades (red), for Area 1 (left) and Area 2 (right) for a winter day (top) and a summer day (bottom). (For interpretation of the references to colour in this figure caption, the reader is referred to the web version of this paper.)
M.C. Brito et al. / Renewable Energy 111 (2017) 85e94
the peak production of south-facing facades around solar noon. At this time of the year, the south-facing facades do not contribute significantly to the production diagram, since at midday the solar height favors solar roof and not vertical surfaces. If the roof and facade solar potential was fully exploited, the total electricity demand during daylight hours would be several times satisfied by local photovoltaic production, doubled for Area 1 and more than tripled in Area 2. More interestingly, the fact that the peak production changes with orientation leads to a complementarity between roofs and facades, especially for east- and westfacing facades. Hence, even if regulatory or technical restrictions to the use of photovoltaic systems on roofs and/or facades significantly limited the technical potential of building surfaces, we can conclude that during daylight hours in the summer months, PV systems could easily satisfy the total local load. As the available area on roofs and facades is fixed, an average efficiency of about 20%, expected to be reached by 2025, would represent a þ30% increase on the solar energy yield for all surfaces, at all times of the day and year, since the efficiency is a multiplicative factor in the present model. This would significant impact on the presented results. On the one hand, it would help solar roofs to address the electricity demand on their own; PV on roofs and the best locations on facades (>900 kWh/m2/year) would clearly satisfy non-baseload consumption throughout the year. On the other hand, higher efficiency PV modules on the best spots on facades would also increase their contribution to the adjustment of electricity generation to the load diagram. We can thus conclude that the development of cost-competitive higher efficiency PV modules would have an enabling effect on large scale deployment of urban photovoltaics and its impact on the electricity landscape. 7. Conclusions The purpose of this study was to analyze the relevance of facades and other vertical features in the urban environment for solar power generation. Hourly solar irradiation on every unit area of roofs and vertical walls of buildings in two representative areas in the city of Lisbon are calculated using LiDAR data to describe the topography of the landscape and local typical meteorological year data set. The model includes mutual shading between buildings and sky view factor restrictions to diffuse radiation. The solar irradiation results are compared to local electricity demand, estimated from the population distribution. Results show that, as expected, facades receive lower levels of solar irradiation than roofs, irrespective of their inclinations. On annual terms, south facing PV facades feature higher yields than facades with other orientations. During the summer months the total roof potential exceeds the non-baseload demand. At that time of the year, if the facades' potential is also taken into account, the total PV potential reaches the total local electricity demand. During the winter months, with higher electricity demand and lower irradiation available, both roofs and facades are required to reach the non-baseload demand. Overall, it is shown that for the case study areas, the non-baseload electricity demand could be satisfied by cost effective PV investments on roofs and facades at today's market conditions for up to 10 months of the year. Analysis of hourly PV generation of roofs and facades has shown that, in summer days, facades are able to spread the peak PV production throughout the day, in particular in the early and later hours of the day when the demand can only be satisfied by PV facades. During winter, facades have the potential to double the solar potential, due to the more favorable inclination, but mostly at midday, and therefore their contribution to modifying the generation profile is less relevant than in summer. The impact of PV facades on monthly or annual time scales
93
result from the fact that there is a significant area available on vertical surfaces in contemporary cities; particularly, for Area 1 the ratio between roof and facade areas is 0.33 and 0.35 for Area 2. Moreover, the fact that two neighborhoods with the same area and comparable resident populations feature relevant differences in the solar potential of roofs and facades, with Area 2 presenting a significant higher potential in spite of an average lower irradiation density due to the larger facade area, highlights the role of architecture and urban planning for the design of modern cities that can take the full potential of the solar resource [29] to meet its local electricity needs. Acknowledgements Part of this work was supported by MIT Portugal Program for Sustainable Energy Systems and FCT grant SFRH/BD/52363/2013 as well as projects MITP-TB/CS/0026/2013 (SUSCITY), PTDC/EMS-ENE/ 4525/2014 (PVCITY), UID/GEO/50019/2013 - Instituto Dom Luiz, and Calouste Gulbenkian Foundation award. References [1] R.R. Hernandez, M.K. Hoffacker, C.B. Field, Efficient use of land to meet sustainable energy needs, Nat. Clim. Change (2015), http://dx.doi.org/10.1038/ NCLIMATE2556. [2] Andreas Molin, Simon Schneider, Patrik Rohdin, Bahram Moshfegh, Assessing a regional building applied PV potential e spatial and dynamic analysis of supply and load matching, Renew. Energy 91 (June 2016) 261e274, http:// dx.doi.org/10.1016/j.renene.2016.01.084. ISSN 0960-1481. [3] IEA, Technology Roadmap: Solar Photovoltaic Energy - 2014 edition, Retrieved from https://www.iea.org/publications/freepublications/publication/ technology-roadmap-solar-photovoltaic-energye-2014-edition.html. [4] M. Hummon, P. Denholm, R. Margolis, Impact of photovoltaic orientation on its relative economic value in wholesale energy markets, Prog. Photovolt. Res. Appl. 21 (2013) 1531e1540, http://dx.doi.org/10.1002/pip.2198. [5] Nahid Mohajeri, Govinda Upadhyay, Agust Gudmundsson, Dan Assouline, ro ^me K€ Je ampf, Jean-Louis Scartezzini, Effects of urban compactness on solar energy potential, Renew. Energy 93 (August 2016) 469e482, http:// dx.doi.org/10.1016/j.renene.2016.02.053. ISSN 0960-1481. [6] C. Hachem, A. Athienitis, P. Fazio, Energy performance enhancement in multistory residential buildings, J. Appl. Energy 116 (C) (2014) 9e19. [7] Daniel Chemisana, Building integrated concentrating photovoltaics: a review, Renew. Sustain. Energy Rev. 15 (1) (January 2011) 603e611, http://dx.doi.org/ 10.1016/j.rser.2010.07.017. ISSN 1364-0321. [8] Bjørn Petter Jelle, Christer Breivik, Hilde Drolsum Røkenes, Building integrated photovoltaic products: a state-of-the-art review and future research opportunities, Sol. Energy Mater. Sol. Cells 100 (May 2012) 69e96, http://dx.doi.org/ 10.1016/j.solmat.2011.12.016. ISSN 0927-0248. [9] G.K. Singh, Solar power generation by PV (photovoltaic) technology: a review, Energy 53 (1) (May 2013) 1e13, http://dx.doi.org/10.1016/j.energy.2013.02.057. ISSN 0360-5442. [10] G. Verberne, P. Bonomo, F. Frontini, M.N. van den Donker, A. Chatzipanagi, K. Sinapis, W. Folkerts, BIPV products for facades and roofs: a market analysis, in: 29th European Photovoltaic Solar Energy Conference and Exhibition, Amsterdam, 2014. [11] A. Henemann, BIPV: built-in solar energy, Renew. Energy Focus 9 (14) (2008) 16e19, http://dx.doi.org/10.1016/S1471-0846(08)70179-3. , A comprehensive review [12] G. Quesada, D. Rousse, Y. Dutil, M. Badache, S. Halle of solar facades. Transparent and translucent solar facades, Renew. Sustain. Energy Rev. 16 (5) (2012) 2643e2651, http://dx.doi.org/10.1016/ j.rser.2012.02.059. [13] S. Freitas, C. Catita, P. Redweik, M.C. Brito, Modelling solar potential in the urban environment: state-of-the-art review, Renew. Sustain. Energy Rev. 41 (2015), http://dx.doi.org/10.1016/j.rser.2014.08.060. January, 915e931. [14] P. Redweik, C. Catita, M.C. Brito, Solar energy potential on roofs and facades in an urban landscape, Sol. Energy 97 (2013), http://dx.doi.org/10.1016/j.solener.2013.08.036. November, 332e341, ISSN 0038-092X. [15] C. Catita, P. Redweik, J. Pereira, M.C. Brito, Extending solar potential analysis in buildings to vertical facades, Comput. Geosci. (2014), http://dx.doi.org/ 10.1016/j.cageo.2014.01.002. Elsevier, (66) May, 1e12. [16] M. Iqbal, An Introduction to Solar Radiation, Academic Press, New York, 1983, pp. 169e213. Chap. 7. [17] B. Marion, A method for modeling the currentevoltage curve of a PV module for outdoor conditions, Prog. Photovolt. Res. Appl. 10 (2002) 205e214. [18] A. Bidram, A. Davoudi, R.S. Balog, Control and Circuit Techniques to mitigate partial shading effects in photovoltaic arrays, IEEE J. Photovolt. (2012) 532e546, 10.1109/JPHOTOV.2012.2202879. [19] A. Verso, A. Martin, J. Amador, J. Dominguez, GIS-based method for evaluating
94
[20]
[21]
[22]
[23] [24]
M.C. Brito et al. / Renewable Energy 111 (2017) 85e94 photovoltaic potential in urban environments: the particular case of Miraflores de la Sierra, Sol. Energy 117 (2015) 236e245. July 2015. Seung-Ho Yoo, Simulation for an optimal application of BIPV through parameter variation, Sol. Energy 85 (7) (July 2011) 1291e1301, http:// dx.doi.org/10.1016/j.solener.2011.03.004. ISSN 0038-092X. L.G. Swan, V.I. Ugursal, Modeling of end-use energy consumption in the residential sector: a review of modeling techniques, Renew. Sustain. Energy Rev. 13 (8) (2009) 1819e1835, http://dx.doi.org/10.1016/j.rser.2008.09.033. Jacopo Torriti, A review of time use models of residential electricity demand, Renew. Sustain. Energy Rev. 37 (September 2014) 265e272, http://dx.doi.org/ 10.1016/j.rser.2014.05.034. ISSN 1364-0321. INE (Instituto Nacional de Estatística), Censos 2011 Resultados Definitivos Portugal, 2011. Retrieved from, http://www.ine.pt/. LXI, Online SIG Platform of the Lisbon Council, 2012. Retrieved from, http://lxi.
cm-lisboa.pt/lxi/. [25] S. Ural, E. Hussain, J. Shan, Building population mapping with aerial imagery and GIS data, Int. J. Appl. Earth Obs. Geoinf. 13 (6) (2011) 841e852. [26] REN, Electricity - Real Time Information, 2011. Retrieved from, http://www. centrodeinformacao.ren.pt/EN/Pages/CIHomePage.aspx. ^ncia no mercado liberalizado de energia ele trica e ga s [27] ERSE, Preços de refere natural em Portugal continental, 2016. Retrieved from, http://www.erse.pt/pt/ Simuladores/Documents/Pre%C3%A7osRef_BTN.pdf. ^neo), Consumo de energia [28] PORDATA (Base de Dados Portugal Contempora trica, 2013. Retrieved from, http://www.pordata.pt/. ele [29] J. Kanters, M. Wall, A planning process map for solar buildings in urban environments, Renew. Sustain. Energy Rev. 57 (2016) 173e185, http:// dx.doi.org/10.1016/j.rser.2015.12.073. May 2016.
Contents lists available at ScienceDirect
Renewable Energy journal homepage: www.elsevier.com/locate/renene
The importance of facades for the solar PV potential of a Mediterranean city using LiDAR data ~es, C. Catita, P. Redweik M.C. Brito, S. Freitas*, S. Guimara Instituto Dom Luiz, Faculdade de Ci^ encias, Universidade de Lisboa, 1749-016, Lisboa, Portugal
a r t i c l e i n f o
a b s t r a c t
Article history: Received 7 October 2016 Received in revised form 13 March 2017 Accepted 28 March 2017 Available online 1 April 2017
The aim of this work is to discuss the potential of facades and other vertical features for the photovoltaic potential of the cityscape. The photovoltaic potential in two representative case studies in the city of Lisbon, Portugal, is computed using a digital surface model determined from LiDAR (Light Detection And Ranging) measurements and local typical meteorological year time series. Results are compared with estimated local electricity demand derived from the population distribution. The annual analysis shows that roof and facade PV potential exceeds the local non-baseload demand and can contribute to 50e75% of the total electricity demand. Hourly breakdown shows peak PV power can only achieve winter mid-day electricity demand if the solar potential of facades is also taken into account. Its added value for off-peak PV supply is less significant in winter since non-south facades are not particularly exposed. In summer, however, facades can satisfy non-baseload morning and afternoon demand. A conservative economic analysis shows payback times below 10 years can only be achieved with PV on roofs while a 50/50 mix would lead to payback times of 15 years. © 2017 Elsevier Ltd. All rights reserved.
Keywords: Solar potential GIS Facades Demand
1. Introduction In an era of growing urbanization when most of energy demand is concentrated in cities, energy sustainability requires that a significant fraction of this energy demand can be fulfilled with local, clean and abundant sources of energy. As it has been highlighted by Hernandez [1], solar power is an unavoidable piece of the fabric of sustainable cities; solar power is plentiful in most regions of the globe, it is a renewable source of energy and CO2 emission free. However, solar power has relatively low energy densities, thus requiring considerably larger areas to produce relevant amount of electricity. As modern cities are characterized by high density populations, living in high rise buildings, the available roof area becomes in short supply for solar power to fulfill the local energy demand. As such, building facades offer an attractive and complementary option. Although vertical solar panels receive less solar radiation than roofs and horizontal surfaces, in particular in the summer months, and are more affected by the compactness of the urban layout [2], facades feature high areas; in a building with 4 floors, the area of
* Corresponding author. E-mail address: [email protected] (S. Freitas). http://dx.doi.org/10.1016/j.renene.2017.03.085 0960-1481/© 2017 Elsevier Ltd. All rights reserved.
the facades is about 4 times the area of the roof1. If the whole available area of such building was used for solar panels and shadings from neighboring buildings neglected, the total annual electricity production would triple that of the roof. This ratio will obviously increase further for taller buildings. However, it is important to point out that in this example, the cost of the generated solar electricity (in V/kWh) would also be 4 times higher. Hence, the deployment of PV on less than optimum inclination/ orientation has to be weighted by economic constraints. Nevertheless, the recent trend of fast decreasing costs of PV, which is expected to proceed in future years [3], opens a window of opportunity for this type of applications. Additionally, vertical PV facades will produce relatively more power in winter and less in summer, and more in the early and late hours of the day, when the sun is lower in the sky. Since a building will typically have four, or at least two, exposed facades with opposite orientations, the different solar facades of a building will produce at maximum power at different times of the day. As has been recently highlighted by Hummon et al. [4], this leads to a widening of the peak of power production throughout the day and
1 Assuming a 10 10 m2 area building with 4 storeys, each 3 m high; hence, the roof area is about 100 m2 while the facade area is 4 facades 4 storeys 3 m/ storey 10 m length ¼ 480 m2 facade area.
86
M.C. Brito et al. / Renewable Energy 111 (2017) 85e94
hence a closer match to the load diagram, which can then result in significant savings in electricity storage and/or fossil fuel based backup power. Also in Ref. [5] the importance of adjusting tilt and azimuth of PV installations, with higher tilts being the preferable option for south facing scenarios, is highlighted as a means to spread the electricity generation over a larger period of time. It is also worth mentioning that the lower level of solar irradiation on facades has prompted consideration of different concepts and practical solutions for increasing energy harvesting on vertical surfaces. These include reshaping of the facades themselves [6] or the development of concentration photovoltaic systems for application in buildings [7]. Furthermore, the use of facades for PV generation, and in particular Building Integrated Photovoltaics (BIPV) [8] – defined as photovoltaic cells integrated into the building envelope as part of the building structure – offers interesting opportunities as it can replace conventional building materials whilst creating a harmonious architecture by blending into the design of the building [8,9]. Since it replaces other construction materials, its net cost (per unit area) is already cost-effective in a number of applications [10]. As an added benefit, the air flow behind the solar modules reduces their temperature improving the system efficiency and longevity [11]. Another approach is the use of semitransparent BIPV which allows daylight to enter the interior spaces [12]. The purpose of this work is to analyze the cityscape PV potential, including both roofs and facades, and to understand when facades might act as added advantage to the rooftops. The methodology is described in section 2. Using Light Detection And Ranging (LiDAR) data to describe the topography of the city, including both the terrain and buildings, and reference meteorological data, the PV potential is determined for two case study areas in the city of Lisbon, Portugal, presented in section 3. The two different areas are characterized by differing building morphologies. Then, hourly PV generation potential of roofs and facades is compared with estimated local electricity demand. Results are presented in sections 4, 5 and 6, which discusse the overall PV potential of facades and the added value of better adjustment to the load diagram. In the end, conclusions are drawn regarding the contribution of PV facades to the solar potential of a city. 2. Methodology The estimation of the cityscape PV potential is based on a 3D PV potential tool which uses LiDAR data and reference meteorological data to determine the expected insolation of all points on the ground, roof and facades of buildings in the study areas. The discussion entails the comparison of the PV generation potential with the local electricity demand, which is determined from the population distribution and the per capita average load. 2.1. 3D photovoltaic potential The estimation of solar potential in the urban environment typically requires combining radiation models, accounting for the apparent movement of the sun in the sky, as well as the beam and diffuse fractions of the global solar irradiation, and Geographical Information Systems (GIS) to describe the opaque obstacles met by beam irradiation (i.e. shadow casting) and sky view factors (SVF) that determine the amount of diffuse irradiation. For an overview of solar potential models in the urban environment see Freitas et al. [13]. In this study, the solar power potential of roofs and facades is determined using the SOL algorithm, a software tool described in Redweik et al. [14] and Catita et al. [15]. The algorithm starts from a geo-referenced LiDAR data cloud, re-sampled for a 1 1m2 raster. As vertical surfaces are not directly detected by the aerial LiDAR
sensor, which only measures the surface height, they are decomposed into hyperpoints, i.e. each XY point is used to produce a set of 3D points sharing the same XY coordinates (cartographic coordinates) but with different Z (elevation) located in space along a vertical column on a facade. The local typical meteorological year (TMY) data set associated to the SOLTERM database is used. This includes hourly mean values for horizontal direct and diffuse irradiation calculated over 30 years of local observations. The hourly global irradiance, G, on any point of the tilted surfaces is determined by its two main components: direct and diffuse irradiance (eq. (1), adapted from Ref. [16]).
G ¼ Gbh
cos q SC þ Gdh Fd SVF cos qz
(1)
where Gbh and Gdh are respectively the hourly direct and the diffuse horizontal irradiance components, SC denotes a Shadow Caster b is the transposition factor for isotropic attribute (0 or 1), Fd ¼ 1þcos 2 diffuse radiation, SVF denotes the Sky View Factor (i.e. the fraction of sky visible from a point), qz is the sun's zenith angle and q is the angle of incidence of the sun rays on the tilted plane calculated according to the following equation [16]:
cos q ¼ cos d sin b sin g sin u þ ðcos f cos b þ sin f sin b cos gÞcos d cos u þ ðsin f cos b cos f sin b cos gÞsin d
(2)
where d is the declination angle, f is the latitude of the location, b is the surface tilt, u is the hour angle and g is the surface azimuth. The model does not consider reflected light. At any given time, a shadow algorithm takes each point of the Digital Surface Model (DSM), including trees, as a shadow caster along the line opposite to the direction of the sun. As long as this line is not interrupted, i.e. whenever a DSM cell along that profile features a Z value lower than the shadow line height at that XY position, the pixel is in shadow and receives the Shadow Caster attribute 0, creating then a binary map. For facades the process is further refined by taking into account shadow height and behavior of the neighbourhood. The result is that at any given time, points belonging to the same facade hyperpoint may have different Shadow Caster attributes, i.e. such a hyperpoint may be totally shaded, partially shaded or totally unshaded. The diffuse irradiation on a given point is constrained by its SVF, a measure of the amount of sky seen from the location. For the same facade, ground floors normally have a lower SVF than top floors because of an increase in view obstructions to the sky. Therefore, even when a point is in shadow, diffuse irradiance is still available according to its SVF. This is taken into consideration in the SOL algorithm whereby the SVF is calculated for each hyperpoint element (as well as for the roofs and ground). The SOL algorithm delivers hourly irradiation data on every point of roof and facade for every day of a year in form of maps and tables, which can be used to obtain solar irradiation estimates for different periods and timescales. The PV potential mapping is determined by multiplying the solar irradiation by an efficiency of 15% for a typical module. Because the effect of module temperature on the efficiency of silicon based modules is not negligible, this is also taken into consideration [17]:
hmod ¼ href 1 Dh Ta þ
TNOCT 20 G 25 800
(3)
where hmod is the module operating efficiency, Dh is the
M.C. Brito et al. / Renewable Energy 111 (2017) 85e94
temperature coefficient for efficiency (0,45%/ C), Ta is the ambient temperature in C, NOCT is the Nominal Operating Cell Temperature (47 C). For simplicity, partial shading is not considered as this correction is not expected to be very relevant since the spatial resolution (1 point per square meter) is of the order of magnitude of the area of the considered PV module. If a finer grid was used, the impact of the partial shading effect on the solar yield would become important but its impact can be minimized by a number of techniques, as reviewed by Bidram et al. [18]. This methodology includes some conceptual simplifications and limitations which are worth discussing. On the one hand, the inclination and orientation of the solar PV panels are considered to be those of the building surfaces where they are installed. Thus, PV modules on facades are vertical with the orientation of the facade while PV modules on the roof have the inclination of the roof. This option will underestimate the solar potential. Indeed, one could envision PV modules with an optimum tilt mounted on shading sunscreens (for the case of facades) or on flat roofs. Although it depends on the local architecture, Verso et al. [19] suggests that the option for optimum tilt on flat roofs could represent a 12% increase in the solar potential of an urban area. For the facades, the vertical assumption (instead of tilted sunscreens) could represent an underestimation of the solar potential of c.a. 20%. On the other hand, and for the particular case of facades, this methodology does not take into consideration the existence of windows because LiDAR input data does not provide any information on features of the vertical surfaces. Although windows may be used as solar active area, e.g. using semi-transparent PV modules, the average conversion efficiencies would be about ½ of opaque standard PV modules. With a rough estimate of 30% of glazed area and 7.5% conversion efficiency means that the disregard for the windows will overestimate the facades potential in about 15%. The result is that this more or less balances the underestimation discussed above. Nevertheless, and in order to accommodate these uncertainties, we have conservatively considered an 80% performance ratio for both roof and facades. It ought to be underlined that the approach developed in this methodology, suitable for the large scale analysis of an urban region, requires simplifications regarding local conditions that may vary significantly among particular applications, such as wind speed, reflected radiation, partial shading, air temperature, etc. As demonstrated by Yoo [20], these may strongly impact the solar energy yield of a particular building or façade. As such, the approach used here is not appropriate for individual buildings. 2.2. Electricity demand The electricity demand is determined by estimating the population distribution and multiplying it by the average electricity demand. This is a typical top-down approach, where there is no concern for the stochastic behavior of electricity demand at the individual level. Since the electricity consumption is distributed and it is not constrained to the building itself, the importance resides on the per capita aggregated electricity demand, which encompasses electricity consumption beyond the domestic individual end-users. As discussed in Swan and Ugursal [21], a bottom-up alternative based on the extrapolation of energy consumption of representative individual houses would refine the understanding of the details associated with the energy consumption. However, this approach would require data (e.g. statistical data on household occupancy, appliances availability and use, etc [22]) which was not available for the present study. To determine the population distribution, the census population at block group level [23] is disaggregated by building using the ‘residential volume’ as a proxy variable. The procedure starts by
87
using the LXI [24] to characterize the use of each building in the study area as ‘residential’ or ‘non-residential’. Furthermore, residential homes may be distinguished according to the type of building, e.g. houses or apartments. An additional restrictive criterion is implemented: only buildings with at least one floor above ground (with a mean height above 2.5 m) are considered suitable for habitation, and may therefore receive resident population. The resident number per building was determined by the method proposed by Ural et al. [25]. The set of buildings receives a proportional share of the resident population based on the ratio of their volume and the total resident volume in the block group, which is then weighted by the height of the building. The electricity demand of each building was then calculated by multiplying the estimated number of inhabitants by the per capita electricity demand. For that purpose, we considered the hourly load diagrams for a full year [26] normalized by the number of inhabitants in the country. The flowchart in Fig. 1 illustrates the process described above. In the discussion below (c.f. section 3.3), the PV generation potential is also compared to the non-baseload load diagram. As the baseload is defined as the minimum level of demand on an electrical grid over 24 h the non-baseload hourly profile was determined by subtracting from the hourly demand load the minimum load of the previous 24 h. This minimum is expected to occur during the night period, when electricity demand tends to be lower. 3. Methodology implementation through case study The methodology is implemented in two representative albeit dissimilar neighborhoods in the city center of Lisbon, Portugal. Area 1 was built in the 1950e1960s and features high rise buildings with about 10 floors (Fig. 2). On the other hand, Area 2 is located in a parish originally developed in the early 20th century, with narrower roads and high trees. Typical buildings have three to five floors and inner quadrangle. The total number of residential buildings in Area 1 and Area 2 are 244 and 294, respectively. For both cases, the modelled area is 500 500 m2. €ppen climate classification scheme, climate According to the Ko in Lisbon is Subtropical-Mediterranean climate, with short and very mild winters and warm summers. A summary of the most important features of both Areas is presented in Table 1. Using the methodology described in section 2.2, the building's resident number and electricity demand were estimated for the two case study areas. Results are shown in Fig. 3. One can observe that in Area 1, the population density and hence the electricity demand is higher in the central high rising buildings. Area 2 is more compact and uniform. For Area 1, the resident number was estimated at 3313 and therefore the overall electricity demand was estimated at 63 GWh/year. As for Area 2, the estimated resident number is slightly higher at 3742 inhabitants resulting then in an also slightly higher estimated electricity demand of 72 GWh/year. 4. Annual energy production For the PV potential in roofs and facades of the two areas of interest, the methodology described in section 2.1 was implemented. Solar radiation and PV power was determined for every hour of the day, for all ground and rooftops raster points and for all vertical surface hyperpoints. Fig. 4 maps the annual solar irradiation of the two areas of interest. Its most conspicuous feature is the fact that roofs and ground are clearly superior to vertical facades, which have significantly lower levels of irradiation, both in Area 1, with high rise and generally unobstructed buildings and in Area 2, with a compact
88
M.C. Brito et al. / Renewable Energy 111 (2017) 85e94
Fig. 1. Flowchart of the methodology followed for electricity demand estimation.
Fig. 2. Street view and aerial view of case study areas: Area 1 (top) and Area 2 (bottom).
arrangement of 3-story buildings. A more detailed observation of Fig. 4 reveals that south-facing
facades feature higher annual yields than east- or west-facing facades (the latter are not seen in this particular perspective view).
M.C. Brito et al. / Renewable Energy 111 (2017) 85e94
89
Table 1 Summary of buildings characteristics for Areas A and B.
Area A Area B
Residential footprint area [%]
Average building height [m]
Maximum building height [m]
Rooftop area [m2]
Façade area [m2]
13 19
18 20
67 51
47133 32850
135902 98372
Fig. 3. Annual per building electricity demand for Area 1 (left) and Area 2 (right).
Fig. 4. Annual solar irradiation for Area 1 (left) and Area 2 (right).
This effect is easily observed on the central neighborhoods in Area 2. On the other hand, the variation of PV potential within the same facade is of interest; this is clearly seen on the high rise buildings in Area 1, where a certain degree of mutual shading in the lower floors is present. The distribution of the solar potential and the range of values achieved can be better understood from Fig. 5, which presents the histogram of the annual solar potential for Area 1 and Area 2, both for roofs (dashed lines) and facades (solid and colored lines). These results confirm that roofs feature much higher levels of irradiation, from 1000 to 1800 kWh/m2/year, depending on their particular orientation and inclination. Facades, on the other hand, typically receive solar radiation only in the 100e1000 kWh/m2/year range, depending on their orientation (higher annual irradiation on south facing facades and the lower on north facing facades) and shading from neighboring buildings and trees. In both cases, east and west facing facades feature a wide range of irradiation levels, from 100 to
800 kWh/m2/year, and the sum of their potential might surpass those facing south. It is worth mentioning that the inner quadrangles and indented topography of the buildings in Area 2 lead to a higher facade area which leads to higher total facade irradiation but higher mutual shading, and therefore feature lower average irradiation density. Fig. 6 shows the accumulated monthly PV potential on roofs (in dark brown) and facades (different shades of light brown, according to 4 different classes: irradiation above 900 kWh/m2/year, between 700 and 900, between 500 and 700, and below 500 kWh/m2/year). Also shown is the estimated local electricity demand; the solid line describes non-baseload demand while the dashed line represents the full load demand for each area. The discussion below focus on the non-baseload demand. This implicitly assumes that other (nonurban) renewable energy sources, such as biomass, large hydro or wind (if associated to storage) would satisfy the baseload demand. Results show that the PV potential in Area 2 is more favorable
90
M.C. Brito et al. / Renewable Energy 111 (2017) 85e94
Fig. 5. Annual solar potential histogram for Area 1 (left) and Area 2 (right). Dashed lines refer to roofs and solid lines to facades with different colors according to South, East, West and North orientation, i.e. points with azimuth inside the intervals [45 , 135 [, [135 , 225 [, [225 , 315 [ e [315 , 45 [ where 0 denotes North. (For interpretation of the references to colour in this figure caption, the reader is referred to the web version of this paper.)
Fig. 6. Monthly PV potential (roofs: dark brown column; facades: lighter brown columns according to 4 different classes: above 900 kWh/m2/year, between 700 and 900, between 500 and 700, and below 500 kWh/m2/year) and electricity demand (blue solid line: non-baseload monthly electricity demand; blue dashed line: monthly total electricity demand) for Area 1 (left) and Area 2 (right). (For interpretation of the references to colour in this figure caption, the reader is referred to the web version of this paper.)
than in Area 1, with higher PV potential for all months of the year. In both areas, and in annual terms, the roof PV potential exceeds the local non-baseload demand and can contribute to 26% (Area 1) and 36% (Area 2) of the total electricity demand. Furthermore, if the potential of PV facades is added to that of the roofs, the total PV potential increases 54% for Area 1 and 73% for Area 2. Closer inspection of Fig. 6 shows that for the summer months the total roof PV potential exceeds the non-baseload demand in both areas. Also, if both facades and roofs are taken into account then the buildings' PV potential is of the order of magnitude of the total load demand; for Area 2 this threshold is even reached for all of the summer months. These results highlight the relevant role that building integrated photovoltaics can provide to the electricity system in urban environments. On the other hand, for the winter months, solar irradiation decreases significantly while the load demand increases slightly, due to shorter and colder days which require more artificial lighting and some heating. Thus, it is not surprising that the total roof PV potential is not able to satisfy the non-baseload electricity demand
Payback ½years ¼
during 5 months in Area 1 and 4 months in Area 2. However, this non-baseload demand would be satisfied if the total PV potential of facades was deployed (Table 2). As discussed in the following section, it is not reasonable to expect that all facades could be covered with PV modules at current or foreseen PV costs in the near or medium term. 5. Payback time analysis The classification of facades according to yearly solar irradiation (the different shades of brown in Fig. 6) may be interpreted as different classes of financial payback of investment (Table 3); hence, darker brown (above 900 kWh/m2/year) are cost-effective PV investments even in today's market conditions, where one may assume an average cost for a rooftop PV system of 300 V/m2 [10], and a grid selling price of 0.1555V/kWh [27]. Lighter browns (below 500 kWh/m2/year) could only be interesting as an investment when PV costs dropped significantly. The Payback time estimation is detailed in eq. (4).
300 V m2 System class kWh m2 year href PR 0:1555½V=kWh
(4)
M.C. Brito et al. / Renewable Energy 111 (2017) 85e94
91
Table 2 Annual energy demand and solar electricity production for the different system classes represented in Fig. 6.
Total load demand [GWh/year]
including baseload not including baseload Roofs Façade
Energy production [GWh/year]
1800 kWh/m2/year 500 kWh/m2/year 500e700 kWh/m2/year 700e900 kWh/m2/year 900 kWh/m2/year
Roofs and façades
Table 3 Financial payback time of investment for an average rooftop system and the threshold of the different facade classes. System class Rooftop Facade
Payback time (years) 1800 kWh/m2/year 900 kWh/m2/year 700 kWh/m2/year 500 kWh/m2/year
8.9 17.9 23.0 32.2
where href is 0.15 the typical efficiency of a 1 m2 module, PR is a performance ratio of 0.80 and System class can be found in Table 3. It is interesting to note that the combined payback of roofs and high performing facades (above 700 kWh/m2/year) would be 19.5 and 18.7 years for areas 1 and 2, respectively. Furthermore, Table 4 shows the optimum mix of PV systems on roofs and facades for different levels of combined payback times. The results of Table 4 were obtained by ordering the points on roofs and facades according to their payback time (lower paybacks first). The global payback is the weighted average of the paybacks, where the weights are the fraction of the area with that particular payback. Results provided by the model show that for both areas payback times below 10 years can only be achieved with PV on roofs whilst a 50/50 mix would lead to payback times of 15 years. These long payback times confirm that, with current costs, PV on facades can only be attractive in very particular conditions. However, expected significant reductions in PV costs in the medium term (halving costs by 2030 [3]) will trigger the mass deployment solar facades in the urban environment. In this context, the results presented in Fig. 6 show that, for Area 2, non-baseload electricity demand could be satisfied by cost effective PV investments on roofs and facades for 10 months of the year. For Area 1, where most facades have modest yearly performance (mostly east- and west-facing facades generating less than 500 kWh/m2/year) the contribution of cost-effective PV deployment on facades would only be marginal, reducing the deficit from 5 to 4 months. It ought to be noticed that the estimation of these payback times critically depend on the average efficiency of the modules. If 20% PV efficiency was to be assumed, as it is expected by 2025, payback times for rooftops would decrease to just over 6 years and best locations in facades would reach about 13 years paybacks. This would translate to 38% and 37% of PV on facades for a combined
Table 4 Mix of roof and facade PV systems for different combined payback time periods. Combined payback time
10 12 15
Area 1
Area 2
Roofs
Facades
Roofs
Facades
100% 78% 50%
e 22% 50%
100% 78% 48%
e 22% 52%
Area A
Area B
17.9 4.2 4.6 1.7 1.9 1.1 0.4 9.7
15.9 3.7 6.4 2.4 2.1 1.4 0.7 13
payback time below 10 years, for Areas 1 and 2, respectively. Fig. 6 also shows how the non-baseload electricity demand may be satisfied by solar electricity, in particular in the summer months. These results also highlight the need for alternative (not solar) fuels for the baseload production since the non-baseload only accounts for about ¼ of the total load demand. Coincidentally, a very similar plot would be that of the monthly residential load. In fact, for the district of Lisbon, 28% of the electricity demand is estimated to be associated to residential demand (data for 2012, c.f. [28]). Hence, these results show that the PV generation would satisfy the residential electricity demand of the city (on a monthly or annual basis, not hourly, since there is residential demand during night periods) and therefore one may conclude that PV on roofs and facades can be a critical tool to achieve Net Zero Energy Buildings in the urban environment.
6. Hourly photovoltaic supply For the analysis of the impact of non-optimum inclination and orientation of PV facades and, thus, their off-peak production, it is informative to observe Fig. 7, which shows for the two areas under study the solar radiation at midday in a winter day and 09:00 LST on a summer day. At midday in the winter day, with relatively low overall solar irradiation, south facing facades have higher irradiation than roofs (or ground) while east and west-facing (not seen in the images due to the perspective) facades receive little solar power at this time. On the other hand, for the early hours of a summer day, the results show that the solar potential is much higher than the winter peak potential, even at this early time of the day. It is also clear that roofs (and ground) have much higher solar irradiation density than facades, regardless of their orientations. Fig. 8 shows the hourly balance between local electricity demand and the photovoltaic potential in roofs and facades, for typical days of summer and winter in the two areas under analysis. For the winter day, the results confirm that due to the low levels of solar irradiation and the relatively shorter days, the overall PV supply is significantly lower than the local daily electricity demand for both areas. Also, peak PV power can only achieve mid-day electricity demand if the solar potential of facades is also taken into account. One can observe that the added value of off-peak PV supply is not significant since non-south facades are not particularly exposed during winter days. In fact, the peak production of the facades is reached at the same time of the peak production of roofs, which is not coincident with the morning or the evening demand peak, typically associated with illumination. As for the summer day, the total roof-only potential peak power exceeds demand at midday. It is clear that the facade-only potential peak power exceeds morning demand in both areas; this effect is much more pronounced in Area 2. Peak production of east- and west-facing facades at 9am and 4pm (morning and afternoon) respectively is evidently higher that
Fig. 7. Solar radiation for Area 1 (left) and Area 2 (right) at 12:00 LST on December 21st (top) and 09:00 LST on June 21st (bottom).
Fig. 8. Hourly electricity demand (dark blue line) and photovoltaic potential of roofs (black dashed line), all facades (black solid line), south facades (orange), east facades (yellow), west facades (green), north facades (light blue) and roofs and facades (red), for Area 1 (left) and Area 2 (right) for a winter day (top) and a summer day (bottom). (For interpretation of the references to colour in this figure caption, the reader is referred to the web version of this paper.)
M.C. Brito et al. / Renewable Energy 111 (2017) 85e94
the peak production of south-facing facades around solar noon. At this time of the year, the south-facing facades do not contribute significantly to the production diagram, since at midday the solar height favors solar roof and not vertical surfaces. If the roof and facade solar potential was fully exploited, the total electricity demand during daylight hours would be several times satisfied by local photovoltaic production, doubled for Area 1 and more than tripled in Area 2. More interestingly, the fact that the peak production changes with orientation leads to a complementarity between roofs and facades, especially for east- and westfacing facades. Hence, even if regulatory or technical restrictions to the use of photovoltaic systems on roofs and/or facades significantly limited the technical potential of building surfaces, we can conclude that during daylight hours in the summer months, PV systems could easily satisfy the total local load. As the available area on roofs and facades is fixed, an average efficiency of about 20%, expected to be reached by 2025, would represent a þ30% increase on the solar energy yield for all surfaces, at all times of the day and year, since the efficiency is a multiplicative factor in the present model. This would significant impact on the presented results. On the one hand, it would help solar roofs to address the electricity demand on their own; PV on roofs and the best locations on facades (>900 kWh/m2/year) would clearly satisfy non-baseload consumption throughout the year. On the other hand, higher efficiency PV modules on the best spots on facades would also increase their contribution to the adjustment of electricity generation to the load diagram. We can thus conclude that the development of cost-competitive higher efficiency PV modules would have an enabling effect on large scale deployment of urban photovoltaics and its impact on the electricity landscape. 7. Conclusions The purpose of this study was to analyze the relevance of facades and other vertical features in the urban environment for solar power generation. Hourly solar irradiation on every unit area of roofs and vertical walls of buildings in two representative areas in the city of Lisbon are calculated using LiDAR data to describe the topography of the landscape and local typical meteorological year data set. The model includes mutual shading between buildings and sky view factor restrictions to diffuse radiation. The solar irradiation results are compared to local electricity demand, estimated from the population distribution. Results show that, as expected, facades receive lower levels of solar irradiation than roofs, irrespective of their inclinations. On annual terms, south facing PV facades feature higher yields than facades with other orientations. During the summer months the total roof potential exceeds the non-baseload demand. At that time of the year, if the facades' potential is also taken into account, the total PV potential reaches the total local electricity demand. During the winter months, with higher electricity demand and lower irradiation available, both roofs and facades are required to reach the non-baseload demand. Overall, it is shown that for the case study areas, the non-baseload electricity demand could be satisfied by cost effective PV investments on roofs and facades at today's market conditions for up to 10 months of the year. Analysis of hourly PV generation of roofs and facades has shown that, in summer days, facades are able to spread the peak PV production throughout the day, in particular in the early and later hours of the day when the demand can only be satisfied by PV facades. During winter, facades have the potential to double the solar potential, due to the more favorable inclination, but mostly at midday, and therefore their contribution to modifying the generation profile is less relevant than in summer. The impact of PV facades on monthly or annual time scales
93
result from the fact that there is a significant area available on vertical surfaces in contemporary cities; particularly, for Area 1 the ratio between roof and facade areas is 0.33 and 0.35 for Area 2. Moreover, the fact that two neighborhoods with the same area and comparable resident populations feature relevant differences in the solar potential of roofs and facades, with Area 2 presenting a significant higher potential in spite of an average lower irradiation density due to the larger facade area, highlights the role of architecture and urban planning for the design of modern cities that can take the full potential of the solar resource [29] to meet its local electricity needs. Acknowledgements Part of this work was supported by MIT Portugal Program for Sustainable Energy Systems and FCT grant SFRH/BD/52363/2013 as well as projects MITP-TB/CS/0026/2013 (SUSCITY), PTDC/EMS-ENE/ 4525/2014 (PVCITY), UID/GEO/50019/2013 - Instituto Dom Luiz, and Calouste Gulbenkian Foundation award. References [1] R.R. Hernandez, M.K. Hoffacker, C.B. Field, Efficient use of land to meet sustainable energy needs, Nat. Clim. Change (2015), http://dx.doi.org/10.1038/ NCLIMATE2556. [2] Andreas Molin, Simon Schneider, Patrik Rohdin, Bahram Moshfegh, Assessing a regional building applied PV potential e spatial and dynamic analysis of supply and load matching, Renew. Energy 91 (June 2016) 261e274, http:// dx.doi.org/10.1016/j.renene.2016.01.084. ISSN 0960-1481. [3] IEA, Technology Roadmap: Solar Photovoltaic Energy - 2014 edition, Retrieved from https://www.iea.org/publications/freepublications/publication/ technology-roadmap-solar-photovoltaic-energye-2014-edition.html. [4] M. Hummon, P. Denholm, R. Margolis, Impact of photovoltaic orientation on its relative economic value in wholesale energy markets, Prog. Photovolt. Res. Appl. 21 (2013) 1531e1540, http://dx.doi.org/10.1002/pip.2198. [5] Nahid Mohajeri, Govinda Upadhyay, Agust Gudmundsson, Dan Assouline, ro ^me K€ Je ampf, Jean-Louis Scartezzini, Effects of urban compactness on solar energy potential, Renew. Energy 93 (August 2016) 469e482, http:// dx.doi.org/10.1016/j.renene.2016.02.053. ISSN 0960-1481. [6] C. Hachem, A. Athienitis, P. Fazio, Energy performance enhancement in multistory residential buildings, J. Appl. Energy 116 (C) (2014) 9e19. [7] Daniel Chemisana, Building integrated concentrating photovoltaics: a review, Renew. Sustain. Energy Rev. 15 (1) (January 2011) 603e611, http://dx.doi.org/ 10.1016/j.rser.2010.07.017. ISSN 1364-0321. [8] Bjørn Petter Jelle, Christer Breivik, Hilde Drolsum Røkenes, Building integrated photovoltaic products: a state-of-the-art review and future research opportunities, Sol. Energy Mater. Sol. Cells 100 (May 2012) 69e96, http://dx.doi.org/ 10.1016/j.solmat.2011.12.016. ISSN 0927-0248. [9] G.K. Singh, Solar power generation by PV (photovoltaic) technology: a review, Energy 53 (1) (May 2013) 1e13, http://dx.doi.org/10.1016/j.energy.2013.02.057. ISSN 0360-5442. [10] G. Verberne, P. Bonomo, F. Frontini, M.N. van den Donker, A. Chatzipanagi, K. Sinapis, W. Folkerts, BIPV products for facades and roofs: a market analysis, in: 29th European Photovoltaic Solar Energy Conference and Exhibition, Amsterdam, 2014. [11] A. Henemann, BIPV: built-in solar energy, Renew. Energy Focus 9 (14) (2008) 16e19, http://dx.doi.org/10.1016/S1471-0846(08)70179-3. , A comprehensive review [12] G. Quesada, D. Rousse, Y. Dutil, M. Badache, S. Halle of solar facades. Transparent and translucent solar facades, Renew. Sustain. Energy Rev. 16 (5) (2012) 2643e2651, http://dx.doi.org/10.1016/ j.rser.2012.02.059. [13] S. Freitas, C. Catita, P. Redweik, M.C. Brito, Modelling solar potential in the urban environment: state-of-the-art review, Renew. Sustain. Energy Rev. 41 (2015), http://dx.doi.org/10.1016/j.rser.2014.08.060. January, 915e931. [14] P. Redweik, C. Catita, M.C. Brito, Solar energy potential on roofs and facades in an urban landscape, Sol. Energy 97 (2013), http://dx.doi.org/10.1016/j.solener.2013.08.036. November, 332e341, ISSN 0038-092X. [15] C. Catita, P. Redweik, J. Pereira, M.C. Brito, Extending solar potential analysis in buildings to vertical facades, Comput. Geosci. (2014), http://dx.doi.org/ 10.1016/j.cageo.2014.01.002. Elsevier, (66) May, 1e12. [16] M. Iqbal, An Introduction to Solar Radiation, Academic Press, New York, 1983, pp. 169e213. Chap. 7. [17] B. Marion, A method for modeling the currentevoltage curve of a PV module for outdoor conditions, Prog. Photovolt. Res. Appl. 10 (2002) 205e214. [18] A. Bidram, A. Davoudi, R.S. Balog, Control and Circuit Techniques to mitigate partial shading effects in photovoltaic arrays, IEEE J. Photovolt. (2012) 532e546, 10.1109/JPHOTOV.2012.2202879. [19] A. Verso, A. Martin, J. Amador, J. Dominguez, GIS-based method for evaluating
94
[20]
[21]
[22]
[23] [24]
M.C. Brito et al. / Renewable Energy 111 (2017) 85e94 photovoltaic potential in urban environments: the particular case of Miraflores de la Sierra, Sol. Energy 117 (2015) 236e245. July 2015. Seung-Ho Yoo, Simulation for an optimal application of BIPV through parameter variation, Sol. Energy 85 (7) (July 2011) 1291e1301, http:// dx.doi.org/10.1016/j.solener.2011.03.004. ISSN 0038-092X. L.G. Swan, V.I. Ugursal, Modeling of end-use energy consumption in the residential sector: a review of modeling techniques, Renew. Sustain. Energy Rev. 13 (8) (2009) 1819e1835, http://dx.doi.org/10.1016/j.rser.2008.09.033. Jacopo Torriti, A review of time use models of residential electricity demand, Renew. Sustain. Energy Rev. 37 (September 2014) 265e272, http://dx.doi.org/ 10.1016/j.rser.2014.05.034. ISSN 1364-0321. INE (Instituto Nacional de Estatística), Censos 2011 Resultados Definitivos Portugal, 2011. Retrieved from, http://www.ine.pt/. LXI, Online SIG Platform of the Lisbon Council, 2012. Retrieved from, http://lxi.
cm-lisboa.pt/lxi/. [25] S. Ural, E. Hussain, J. Shan, Building population mapping with aerial imagery and GIS data, Int. J. Appl. Earth Obs. Geoinf. 13 (6) (2011) 841e852. [26] REN, Electricity - Real Time Information, 2011. Retrieved from, http://www. centrodeinformacao.ren.pt/EN/Pages/CIHomePage.aspx. ^ncia no mercado liberalizado de energia ele trica e ga s [27] ERSE, Preços de refere natural em Portugal continental, 2016. Retrieved from, http://www.erse.pt/pt/ Simuladores/Documents/Pre%C3%A7osRef_BTN.pdf. ^neo), Consumo de energia [28] PORDATA (Base de Dados Portugal Contempora trica, 2013. Retrieved from, http://www.pordata.pt/. ele [29] J. Kanters, M. Wall, A planning process map for solar buildings in urban environments, Renew. Sustain. Energy Rev. 57 (2016) 173e185, http:// dx.doi.org/10.1016/j.rser.2015.12.073. May 2016.