Distributed Software Infrastructure for Evaluating the Integration of
Photovoltaic Systems in Urban Districts
Lorenzo Bottaccioli
1
, Edoardo Patti
1
, Michelangelo Grosso
2
, Gaetano Rascon
`
a
2
, Angelo Marotta
3
,
Salvatore Rinaudo
3
, Andrea Acquaviva
1
and Enrico Macii
1
1
DAUIN, Politecnico di Torino, Torino, Italy
2
ST-Polito s.c.a.r.l., Torino, Italy
3
STMicroelectronics s.r.l., Catania, Italy
Keywords:
Photovoltaic, GIS, Distributed Software Infrastructure, Urban Planning, Spatio-temporal Analysis, Renewable
Energy Planning.
Abstract:
Nowadays, the adoption of renewable energy sources distributed across the city is crucial for planning and de-
veloping the future Smart City. An accurate simulation and modelling of energy sources, such as Photovoltaic
Panels (PV), is necessary to evaluate both economical and environmental benefits. With the growth of renew-
able sources in the city simulations of energy production became crucial for the DSO for evaluating retrofits
or for network balancing events. In this paper, we present a software infrastructure for simulating the solar
radiation and estimating the energy production of a district. The infrastructure simulates the PV production
and evaluates the integration of such systems considering real electricity consumption data. In its core, the
proposed solution models the behaviours of PV systems taking into account the digital surface of rooftops and
sub-hourly meteorological data (e.g. solar radiation and temperature) to compute real-sky conditions. Then,
such information is used to feed a model of the hardware components of PV systems to gain more accurate
estimations of energy production in the district in real-sky conditions.
1 INTRODUCTION
Nowadays, we are moving forward to more smart and
sustainable cities that aim at reducing greenhouse gas
emissions. A Smart City approach fosters a smart
energy use also taking advantage from an increasing
renewable energy sources deployment. In this con-
text, Information and Communication Technologies
(ICTs) play a crucial role in both planning and moni-
toring of distributed energy sources. The crucial roles
of ICTs and the emerging Internet-of-Things (IoT)
are highlighted by the spread diffusion of heteroge-
neous and pervasive sensors in our houses, district and
cities. IoT devices and sensors allow to collect large
amounts of energy related data capable of describing
the consumption behaviours of the citizens. Electric-
ity consumption data can be used in simulation pro-
cesses for evaluating: i) energy management actions;
This work was partially supported by the EU projects
DIMMER and FLEXMETER, and by the Italian project
”Edifici a Zero Consumo Energetico in Distretti Urbani In-
telligenti”.
ii) management of electricity distribution networks;
iii) integration of renewable sources in the city.
In this work, we present a methodology for de-
veloping a distributed software infrastructure to fos-
ter the usage of solar energy. Our solution exploits
a Microservices approach for integrating heteroge-
neous sensors, services and simulation tools for evalu-
ating the integration of Photovoltaic (PV) energy sys-
tems in urban districts. If a large amount of fluctu-
ating energy sources are planned to be installed, an
integration analysis has to be performed considering
network constraints. Such analysis can be achieved
correlating simulated PV systems energy production
with electricity consumption data coming from IoT
devices. In its core, our solution includes hardware
models of PV system components to give more accu-
rate estimations of energy production. The proposed
methodology takes advantages of weather informa-
tion to simulate sub-hourly real-sky solar radiation of
rooftops and to analyse the hardware component per-
formance of the PV system.
The rest of the paper is organized as follows. In
Bottaccioli, L., Patti, E., Grosso, M., Rasconà, G., Marotta, A., Rinaudo, S., Acquaviva, A. and Macii, E.
Distributed Software Infrastructure for Evaluating the Integration of Photovoltaic Systems in Urban Districts.
In Proceedings of the 5th International Conference on Smar t Cities and Green ICT Systems (SMARTGREENS 2016), pages 357-362
ISBN: 978-989-758-184-7
Copyright
c
2016 by SCITEPRESS Science and Technology Publications, Lda. All rights reserved
357
Section 2 the actual state of art of methodologies and
services for PV energy simulation are presented. Sec-
tion 3 analyse the motivation that prompt us to embark
on this research. Section 4 presents the specifications
and the methodology to develop our solution. Finally,
Section 5 provides the concluding remarks.
2 STATE OF THE ART
Geographic Information Systems (GIS) tools have
been applied for solar energy applications in urban
context as reported by (Freitas et al., 2015). One
of the major limitation of actual GIS tools consists
on neglecting time domain, as reported by (Camargo
et al., 2015). Both (Camargo et al., 2015; Freitas
et al., 2015) highlight the fact that an accurate solar ra-
diation estimation is needed and a detailed modelling
of the components of the PV system is required.
GIS tools have been exploited also for web-
based applications for fostering photovoltaic poten-
tial and system integration platforms (Suri et al.,
2008; Mapdwell Solar System, ; de Sousa et al.,
2012; De Amicis et al., 2012). Such platforms do
not provide time-dependent simulation about energy
production and PV system performances. Time de-
pendent simulation of PV production, with an accu-
rate evaluation of components performance, is crucial
for: ii) planning deployment activities; ii) business
plan evaluation; iii) monitoring of existing plants and
smart energy use. The main limitation of such ser-
vices are summarized in the following and reported in
Figure 1: i) they provide yearly data in real-sky con-
ditions; ii) they do not provide hourly or sub-hourly
data in real-sky conditions; iii) they do not take into
account PV system hardware components to model
their performance and behaviour. In addition, such
services do not correlate electricity consumption data
with PV production simulation, which is relevant to
evaluate PV systems integration.
Figure 1: Confront with available services.
With respect to state of the art solutions our
methodology provides real-time energy production
data of existing and of feasible PV systems in the dis-
trict. In the simulation process, our solution takes into
account also the operation efficiency of each com-
ponent of the PV systems. Taking advantage of the
correlation of PV energy production simulation with
electricity consumption data our solution is able to
evaluate the energetic impact of the integration of PV
systems in the district.
3 MOTIVATION AND EXPECTED
OUTCOMES
This research aims at developing a software infras-
tructure that exploits GIS tools, weather and electric-
ity consumption data for evaluating PV systems inte-
gration in urban context. Our infrastructure aims at
providing sub-hourly information of real-sky incident
radiation on rooftops. It is devoted to both planning
and monitoring phases of PV systems, spanning all
scales starting from single building up to block, dis-
trict and city. Such infrastructure can be exploited in
order to produce forecast simulation of PV systems
energy production for smart energy use. This solu-
tion is intended to satisfy the needs of different end-
users such as: i) Single citizen can evaluate the eco-
nomic and environmental savings achievable with the
installation of a PV system; ii) Energy aggregators
and Energy Communities can use the simulations to
schedule consumption of their clients for maximiz-
ing self-consumption and minimizing energy bills. In
particular Energy Communities can exploit such in-
frastructure to perform feasibility studies as proposed
in our previous research (Bottaccioli et al., 2015);
iii) PV system engineers can simulate the behaviour
of converters with the application of realistic condi-
tions. This simulation helps in dimensioning, vali-
dating and optimizing each system before and after
installation; iv) Distribution system operators (DSO)
can take advantage of the proposed solution for net-
work balancing and for planning retrofits and/or ex-
tensions of the existing distribution grid; v) Energy
and City planners can exploit the infrastructure for
evaluating the impacts of large PV systems installa-
tions or for monitoring the performance of existing
ones.
4 PV SYSTEM INTEGRATION
New methodologies and procedures for a high de-
tailed simulations, in both spatial and temporal do-
mains, are recently emerging in the research filed of
PV potential estimation (Camargo et al., 2015; Jaku-
biec and Reinhart, 2013; Luka et al., 2014). These
SMARTGREENS 2016 - 5th International Conference on Smart Cities and Green ICT Systems
358
Figure 2: Distributed Software Infrastructure.
works have opened the way to spatio-temporal anal-
ysis in the assessment of PV potential. In particu-
lar, (Camargo et al., 2015) highlights the necessity
to integrate simulated PV production with electricity
consumption data for a correct PV integration to avoid
network congestions. The limitations of the state of
art solution in PV energy simulation are: i) they run as
Desktop applications and simulated results can not be
provided easily to external users and/or other software
components; ii) they do not integrate weather data;
iii) they do not take advantage of hardware compo-
nents models for evaluating the efficiency of a PV sys-
tem; iv) they do not correlate PV simulated data with
real electricity consumption data. Hence, the main
objective of our research are: i) the development of a
distributed software infrastructure for PV energy sim-
ulation; ii) the integration of the meteorological data
coming from the third-party services; iii) the adop-
tion of hardware components models of PV systems
in order to evaluate the overall efficiency; iv) the cor-
relation between PV production and electricity con-
sumption data in order to evaluate the integration of
PV systems in the district considering network con-
straints.
Figure 2 shows the three layers of our distributed
software infrastructure. The bottom layer is the data
source integration layer that is in charge of collecting
the required information from the following heteroge-
neous data sources: i) Digital Surface Model(DSM),
which is a raster image that represents terrain el-
evation considering the presence of manufactures;
ii) Linke Turbidity coefficients that express the at-
tenuation of solar radiation related to air pollution;
iii) Weather data of third party services for collecting
solar radiation, temperature and wind speed; iv) Elec-
tricity consumption data that are used for evaluating
the integration of PV systems.
The middle layer is the core of our methodol-
ogy. It consists of simulation and integration ser-
vices that are summarized in the following: i) GIS-DB
stores clear-sky and suitable surface maps; ii) Suitable
area identifies suitable surface for PV modules on
rooftops; iii) Clear-Sky simulation; iv) Real-Sky sim-
ulation; v) Solar decomposition provides diffuse and
direct components of solar radiation (as described in
Section 4.2); vi) PV energy estimation module consid-
ers hardware components models for evaluating the
performance behaviours of the PV system; vii) PV
integration module takes into account network con-
straints for evaluating the integration of PV systems
into the grid. Furthermore, it implements algorithm
for load shifting and demand side management. The
PV integration module performs also economic eval-
uation of feasible PV energy systems.
The upper layer is devoted to user applications,
such as Web-Map interface and Dashboards. Both of
them can provide information about performed simu-
lation across the city with different level of details.
4.1 Software Infrastructure for PV
Simulation
In this section, we present the methodology we are
exploiting to develop our distributed simulation in-
frastructure for PV system planning and monitoring
in a Smart City context. The simulation of energy
production is performed taking into account the ef-
ficiency behaviours of each hardware component of
the PV system. The GIS-Server module, showed in
Figure 3, is the core of our solution. We selected
the open-source software GRASS-GIS that provides,
through r.sun, functionalities for computing clear-sky
solar radiation. This tool has been proved to provide
an accurate simulation of solar radiation in urban con-
texts (Ronzino et al., ; Freitas et al., 2015).
The inputs required by the clear-sky simulation
process are: i) the Digital Surface Models (DSM) and
ii) monthly Linke turbidity coefficients. In addition,
the GIS-Server stores in its database (GIS-DB) pre-
and post-processed data: i) DSM; ii) generated clear-
sky hourly radiation maps and iii) suitable rooftops
surface maps.
The proposed distributed infrastructure takes ad-
vantage of a Microservices approach in order to in-
tegrate different software and models in a form of
interoperable services. It exploits the Web Process-
ing Services (WPS), Web Feature Service (WFS) and
Web mapping Services (WMS) that are the standards
defined by the Open Geospatial Consortium (OGC).
OGC specifies a service interface for publishing and
performing geospatial process over the web. WPS are
used for file upload and for executing the simulation
process. They provide the rules for standardizing in-
puts and outputs of the process. WFS are used for
querying and retrieving features about the elements of
Distributed Software Infrastructure for Evaluating the Integration of Photovoltaic Systems in Urban Districts
359
Figure 3: PV Simulation Infrastructure Time Diagram.
a polygon-map. WMS are used for the visualization
of the produced map in the Web-Map interface.
Figure 3 shows the interactions between the soft-
ware actors of our distributed infrastructure for sim-
ulating PV systems energy production and hardware
components behaviours. The computation starts when
the User uploads the DSM through a WPS to GIS-
Server.The upload of the file automatically executes
the clear-sky simulation module by specifying the in-
puts directly in request. Based on the DSM, given a
date and time this process produces the related set of
clear-sky radiation maps. The maps are saved in the
GIS-DB and the User is notified. Then the User can
execute the task for identifying the suitable areas for
PV systems installation. This is computed by the Suit-
able Surface module that takes the DSM, slopes and
orientation as inputs. The resulting maps are stored
again in the GIS-DB. Alternatively the user can up-
load in the GIS-DB his own suitable surface maps.
Finally, the User invokes the PV Simulation mod-
ule for estimating the energy production of a PV sys-
tem exploiting also meteorological conditions. PV
Simulation module needs the already stored infor-
mation about suitable surface and clear-sky radiation
maps. If new parameters for identifying suitable sur-
face are given, such maps are re-calculated by Suit-
able Surface. Meteorological conditions are needed
as well and they are provided by the nearest weather
station though third-party web-services (e.g. (Weather
Underground, )). In particular, solar radiation data
are useful for estimating real-sky conditions by defin-
ing clear-sky indexes for both direct and diffuse ra-
diation. Wind speed and air temperature are needed
for evaluating detailed performance of each PV sys-
tem hardware component (e.g. PV modules, Inverter
and Maximum Power Point Tracker). Optionally, if
direct and diffuse solar radiation components are not
provided by third party services, the solar radiation
decomposition modules is used by the simulation pro-
cess. More information about weather data integra-
tion are reported in Section 4.2. The PV Simulation
modules query the PV array data sheet in order to col-
lect information on the PV array characteristic needed
by the hardware components models.
In a nutshell, the described software infrastruc-
ture is able to simulate sub-hourly real-sky solar ra-
diation of rooftops in a given city district area. Then
it provides an estimation of PV systems energy pro-
duction, also analysing its hardware components per-
formances. Section 4.3 presents the model for energy
conversion efficiency of PV hardware components,
that is the core of the PV Simulation module.
4.2 Weather Data Integration
In the last years the availability of weather station data
present in our cities is strongly increased. This is due
to the lower price of weather stations and the appear-
ance of open web-services for data management and
publication, such as (Weather Underground, ). In or-
der to simulate real-sky solar radiation on a pitched
surface, information on direct and diffuse radiation
is needed. The majority of present weather stations
do not provide direct and diffuse radiation measure-
ment, because accurate and expensive sensors would
be required. In order to overcome such limitation, in
our solution we have developed a module for extract-
ing direct and diffuse radiation from global horizon-
tal radiation. The module exploits solar radiation de-
composition techniques present in the literature such
as (Boland et al., 2013; Orgill and Hollands, 1977;
Erbs et al., 1982). Those experimental techniques use
SMARTGREENS 2016 - 5th International Conference on Smart Cities and Green ICT Systems
360
information only on global radiation to predict the di-
rect and diffuse components. The user can specify
which technique is the most suitable for the area of
interest. This because decomposition models do not
perform with the same accuracy all over the world
due to their experimental nature. Indeed, depending
from latitude, longitude and environmental condition
the accuracy of the decomposition model can change.
This module gives to our infrastructure a big flex-
ibility of employment due to the possibility of inte-
grating common and available weather station data.
4.3 Energy Conversion Efficiency
Modelling
Photovoltaic cells produce direct current (DC) en-
ergy at only a fraction of a volt. The utility wiring
or grid and appliances within the home typically use
alternating current (AC) power with voltages greater
than 100V. To convert power from DC to AC, an in-
verter must be integrated into the PV system. Typ-
ically, another DC-DC converter is added to step up
the low voltage DC produced by photovoltaic cells to
the substantially higher DC voltage necessary at the
input of the inverter. ”Smart” converters embed Max-
imum Power Point Tracker (MPPT) logic that adapt
the power transfer to the changing working conditions
in real time.
The traditional grid-tied architecture of photo-
voltaic systems concentrates all the electronics in the
central inverter. This is the centralized approach. To
gain in terms of global system energy production, re-
liability, safety, communication and monitoring, the
trend today is to move towards the distributed ap-
proach where the electronics is partially or fully dis-
tributed close to each panel (microinverter). In this
way, the power transfer related to non-uniform shad-
ing conditions can be maximized using local MPPT.
Solar radiation and temperature have a huge influ-
ence on the characteristics and performance of each
photovoltaic module, so modelling is mandatory and
very useful to quantify how these environmental fac-
tors influence the performance of the system. The
availability of models of the converter chain compo-
nents is very important in system sizing, cost analy-
sis, and monitoring. For a power electronics engineer
working with renewable energies, it is imperative to
have an accurate model as it can support testing and
development of optimal solutions. At the same time,
the models need to be fed with realistic input condi-
tions taking into account solar radiation, panel loca-
tion and inclination and weather data.
The authors in (Marotta et al., 2011) describe a
behavioural steady-state averaged model for the so-
lar boost converter SPV1020 by STMicroelectron-
ics, equipped with logic running a Perturb&Observe
MPPT algorithm. The same methodology is being
adapted here for modelling and evaluating the effi-
ciency of the various converters used in the district
depending on their actual components and on the spe-
cific environmental conditions. It must be noted that
the system engineers are more concerned with long-
term behaviour than in the transients: this is the rea-
son for which a steady-state behavioural model is
employed. The rate of change of ambient temper-
ature and insulation usually varies over minutes or
hours, so slow-changing stimuli are handled. In ad-
dition, the simulation with a switching model would
be extremely slow due to the prohibitively small time
step needed. The behavioural steady-state averaged
model is implemented to handle series and parallel
connected PV panels.
4.4 Electricity Consumption Data
Integration
With the increase of Smartness in our cities, a large
amount of IoT sensors and actuators have appeared.
During this process energy monitoring and manage-
ment are receiving great attention and many sensors
are recording energy-related data and fluxes. Thanks
to smart-meters, more accurate information about
user energy profiles is available. Furthermore, smart-
plugs (Ganu et al., 2012) or Non intrusive load mon-
itoring algorithm (Zoha et al., 2012) provide detailed
information about load consumption of each appli-
ance. Such detailed information can be used by En-
ergy aggregators or Energy managers to schedule the
consumption of each user with respect to best prices
or renewable production. At the same time consump-
tion data can be use to evaluate the integration of
renewable sources considering self-consumption and
network constrains.
Thanks to a web-services based approach, our in-
frastructure allows the interaction among our solu-
tion and other distributed software architectures or
services devoted to the integration of real-time data
coming from the Smart-Grid (Patti et al., 2015; Patti
et al., ). Hence, our solution is able to easily in-
tegrate data provided by smart-meters and/or smart-
plugs. The integration of electricity consumption data
gives us the possibility to evaluate with more detail
the level of self-consumption of each user. Such de-
tailed analysis is crucial to evaluate economics in-
dex such as Return on Investment, Rate of Return or
Pay Back Time. Additionally, information about con-
sumption and production profiles can be used from
Energy aggregators or Energy Communities to sched-
Distributed Software Infrastructure for Evaluating the Integration of Photovoltaic Systems in Urban Districts
361
ule the consumption profiles of their users for maxi-
mizing the self-consumption and the economics sav-
ings. DSOs can use consumption profile information
for network balancing, active network management or
for ancillary services. Finally, electricity consump-
tion data are used to evaluate the integration of PV
system in the selected area considering also the distri-
bution network constraints.
In many areas, smart-meters are still not deployed
and data regarding users consumption load profiles
are not available. To give flexibility to our methodol-
ogy, we have integrated a load profile simulation mod-
ule that is able to estimate load profile with a good
accuracy as reported in our previous work (Bottacci-
oli et al., 2015). The load profile simulation module
is able to simulate the consumption profiles for differ-
ent users. For residential users, the module requires
information regarding the size of the houses in square
meters and the number of inhabitants for each house-
hold. For industrial and commercial customers, nor-
malized standard load profiles are used. Those stan-
dard profiles are rescaled with respect to total yearly
or monthly electricity consumption.
5 CONCLUSION
In this paper, we presented a methodology for the de-
velopment of a distributed software infrastructure for
simulating PV system behaviours and evaluating their
integration in a Smart City context. Combining re-
alistic radiation modelling framework and electricity
consumption data, our infrastructure can offer to users
detailed information of PV energy production in real-
sky conditions. With such detailed results, different
users can take the optimal decisions in defining the
structure and the architecture of a solar plant in an ur-
ban context, spanning all scales starting from single
building up to block, district and city.
REFERENCES
Boland, J., Huang, J., and Ridley, B. (2013). Decomposing
global solar radiation into its direct and diffuse com-
ponents. Renew. Sustainable Energy Rev., 28:749–
756.
Bottaccioli, L., Patti, E., Acquaviva, A., Macii, E., Jarre,
M., and Noussan, M. (2015). A tool-chain to foster
a new business model for photovoltaic systems inte-
gration exploiting an energy community approach. In
Proc. of IEEE ETFA2015. IEEE.
Camargo, L. R., Zink, R., Dorner, W., and Stoeglehner, G.
(2015). Spatio-temporal modeling of roof-top photo-
voltaic panels for improved technical potential assess-
ment and electricity peak load offsetting at the munic-
ipal scale. Comput. Environ. Urban Syst., 52:58–69.
De Amicis, R., Conti, G., Patti, D., Ford, M., and Elisei,
P. (2012). I-Scope-Interoperable Smart City Services
through an Open Platform for Urban Ecosystems. na.
de Sousa, L., Eykamp, C., Leopold, U., Baume, O., and
Braun, C. (2012). iguess-a web based system integrat-
ing urban energy planning and assessment modelling
for multi-scale spatial decision making. In Porc. of
iEMSs 2012.
Erbs, D., Klein, S., and Duffie, J. (1982). Estimation
of the diffuse radiation fraction for hourly, daily
and monthly-average global radiation. Solar Energy,
28(4):293–302.
Freitas, S., Catita, C., Redweik, P., and Brito, M. (2015).
Modelling solar potential in the urban environment:
State-of-the-art review. Renew. Sustainable Energy
Rev., 41:915–931.
Ganu, T., Seetharam, D. P., Arya, V., Kunnath, R., Hazra, J.,
Husain, S. A., De Silva, L. C., and Kalyanaraman, S.
(2012). nplug: a smart plug for alleviating peak loads.
In Proc. of e-Energy 2012, page 30. ACM.
Jakubiec, J. A. and Reinhart, C. F. (2013). A method for pre-
dicting city-wide electricity gains from photovoltaic
panels based on lidar and gis data combined with
hourly daysim simulations. Solar Energy, 93:127–
143.
Luka, N., Seme, S., laus, D., tumberger, G., and alik, B.
(2014). Buildings roofs photovoltaic potential assess-
ment based on lidar (light detection and ranging) data.
Energy, 66:598–609.
Mapdwell Solar System. http://www.mapdwell.com.
Marotta, A., Ciccazzo, A., and Rinaudo, S. (2011). Mod-
eling of a smart photovoltaic panel integrated self-
powered and high efficiency DC-DC Boost converter.
PCIM Europe 2011.
Orgill, J. and Hollands, K. (1977). Correlation equation for
hourly diffuse radiation on a horizontal surface. Solar
Energy, 19(4):357–359.
Patti, E., Pons, E., Martellacci, D., Castagnetti, F. B., Ac-
quaviva, A., and Macii, E. (2015). multiflex: Flex-
ible multi-utility, multi-service smart metering archi-
tecture for energy vectors with active prosumers. In
Proc. of SMARTGREENS 2015, pages 288–293.
Patti, E., Syrri, A. L. A., Jahn, M., Mancarella, P., Acqua-
viva, A., and Macii, E. Distributed software infras-
tructure for general purpose services in smart grid.
Ronzino, A., Osello, A., Patti, E., Bottaccioli, L., Danna,
C., Lingua, A. M., Acquaviva, A., Macii, E., Grosso,
M., Messina, G., and Rascon, G. The energy effi-
ciency management at urban scale by means of inte-
grated modelling. In Proc.of SEB-15. Elsevier.
Suri, M., Huld, T., Dunlop, E., and Cebecauer, T. (2008).
Geographic aspects of photovoltaics in europe: con-
tribution of the pvgis website. J-STARS, IEEE Journal
of, 1(1):34–41.
Weather Underground. http://www.wunderground.com/.
Zoha, A., Gluhak, A., Imran, M. A., and Rajasegarar, S.
(2012). Non-intrusive load monitoring approaches
for disaggregated energy sensing: A survey. Sensors,
12(12):16838–16866.
SMARTGREENS 2016 - 5th International Conference on Smart Cities and Green ICT Systems
362