Building The Brazilian Smart Grid with Photovoltaic Systems: Project “51 Rooftops” Conception, design, implementation and expansion. Getúlio Antero de Deus Júnior 1 ; Ricardo Henrique Fonseca Alves 2 ; Flávio Henrique Teles Vieira 3 ; Marcelo Stehling de Castro 4 ; Rodrigo Pinto Lemos 5 ; Sérgio Granato de Araújo 6 Department of Electrical, Mechanical and Computer Engineering, Federal University of Goiás (UFG) Av. Universitária, n.th 1488, Setor Leste Universitário, Goiânia, Goiás, Brazil 1 [email protected]; 2 [email protected]; 3 [email protected]; 4 [email protected]; 5 [email protected]; 6 [email protected]Rafael Nielson 7 Celg Distribution S.A. (Celg-D) Rua 2, n.th 505, Jardim Goiás, Goiânia, Goiás, Brazil, 7 [email protected]Abstract—The main goal of the “51 Rooftops in Nova Veneza- GO” Research and Development (R&D) Project is to use the infrastructure of a pre-Smart Grid network located in Nova Veneza-GO in a systemic way taking into account the concepts of a Smart Grid to monitor the power balance. This monitoring will occur in an integrated manner with 51 bidirectional smart meters to collect measures every 15 minutes. The rest of the system to be monitored consists of two transformers of the city of Nova Veneza-GO substation with remote monitoring; five power quality meters Customer Consumer Units (CCU); 51 photovoltaic systems connected to the power grid of the utility; and 51 interactive inverters connected for real-time monitoring of the energy generated. This paper proposes a methodology for the selection of 51 consumers in Nova Veneza-GO connected to two transformers in the pre-Smart Grid network. The methodology consists of ten stages ranging from the grouping of consumers with the same power consumption profile using a neural network, that is, a Non Parametric Self-Organizing Map (PSOM), until the complete and optimal allocation of financial resources through of an Integer Linear Programming. We obtained 12 different groups (clusters) of consumers of the two transformers with the same power consumption profile using the network PSOM algorithm. This grouping (clustering) was considered in the dimensioning and design of Photovoltaic Systems Connected to the Grid (Grid-Tie Systems) using three different computational tools, among them, an approach based on the PVSyst software. In addition, a study of Economic Engineering was carried out to expand the R&D pilot project aiming at the implementation of Grid Tie Systems for all the consumers of Nova Veneza-GO, requiring a Capital Expenditure (Capex) in the order of $ 5,591,673.83 and an annual Operational Expenditure (Opex) of $ 636,718.24. A payback of 12 years was obtained considering a Minimum Acceptable Rate of Return (MARR) of 5% and an allowance of 50% of capital investment. The results show that the business will be financially attractive when considering tax incentives in Brazil for the development of the national industry in the photovoltaic area. On the other hand, the environmental impact results for the city of Nova Veneza-GO show a reduction in CO 2 emissions in the order of 184.3 tons per year and a saving of 503.4 billion liters of water in water reservoirs of hydroelectric plants in Brazil. This R&D pilot project research is unique and the first in this area of the power distribution company Celg Distribution S.A. (Celg-D). Keywords—Economic Engineering; Smart Grid; Solar Energy; Photovoltaic Generation; Non Parametric Self-Organizing Map (PSOM). I. INTRODUCTION The Project No. 253 of CELG Distribution (Celg-D) entitled “Application of Intelligent Network (Smart Grid) on the Supervision of Electricity Supply Medium and Low Voltage Using Different Communication Technologies" Research and Development (R&D) was carried out between 2012 and 2013 with the R&D program funds approved by the National Electric Energy Agency (Aneel) [1]. The project covers different areas of applications related to Smart Grids as Advanced Metering Infrastructure (AMI), Distribution Automation (DA) and Integration of Celg-D systems. In addition, it was developed a consumer portal, opinion polls with consumers and a methodology for positioning of concentrators in a mesh network. Thus, the R&D project No. 253 allowed the Energy utility to have a direct contact with technologies and systems for testing and technical evaluation in these areas [2]. Fig. 1 shows a simplified scheme for the installation of meters, with the need of hiring a GPRS data circuit (General Packet Radio Service) to allow sending data collected from the meters for The Supervision Center, located in Goiânia-GO, based on the premises of Celg-D. Thus, meter data are forwarded to a concentrator device with General Packet Radio Service (GPRS data connection) that takes readings directly to the energy utility [2]. International Scientific Journal Journal of Environmental Science http://environment.scientific-journal.com/
14
Embed
Building The Brazilian Smart Grid with Photovoltaic ...environment.scientific-journal.com/articles/5/17.pdfBuilding The Brazilian Smart Grid with Photovoltaic Systems: Project “51
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Building The Brazilian Smart Grid with
Photovoltaic Systems: Project “51 Rooftops” Conception, design, implementation and expansion.
Getúlio Antero de Deus Júnior1; Ricardo Henrique
Fonseca Alves2; Flávio Henrique Teles Vieira
3;
Marcelo Stehling de Castro4; Rodrigo Pinto Lemos
5;
Sérgio Granato de Araújo6
Department of Electrical, Mechanical and Computer
Engineering, Federal University of Goiás (UFG)
Av. Universitária, n.th 1488, Setor Leste Universitário,
Abstract—The main goal of the “51 Rooftops in Nova Veneza-
GO” Research and Development (R&D) Project is to use the
infrastructure of a pre-Smart Grid network located in Nova
Veneza-GO in a systemic way taking into account the concepts of
a Smart Grid to monitor the power balance. This monitoring will
occur in an integrated manner with 51 bidirectional smart meters
to collect measures every 15 minutes. The rest of the system to be
monitored consists of two transformers of the city of Nova
Veneza-GO substation with remote monitoring; five power
quality meters Customer Consumer Units (CCU); 51
photovoltaic systems connected to the power grid of the utility;
and 51 interactive inverters connected for real-time monitoring
of the energy generated. This paper proposes a methodology for
the selection of 51 consumers in Nova Veneza-GO connected to
two transformers in the pre-Smart Grid network. The
methodology consists of ten stages ranging from the grouping of
consumers with the same power consumption profile using a
neural network, that is, a Non Parametric Self-Organizing Map
(PSOM), until the complete and optimal allocation of financial
resources through of an Integer Linear Programming. We
obtained 12 different groups (clusters) of consumers of the two
transformers with the same power consumption profile using the
network PSOM algorithm. This grouping (clustering) was
considered in the dimensioning and design of Photovoltaic
Systems Connected to the Grid (Grid-Tie Systems) using three
different computational tools, among them, an approach based
on the PVSyst software. In addition, a study of Economic
Engineering was carried out to expand the R&D pilot project
aiming at the implementation of Grid Tie Systems for all the
consumers of Nova Veneza-GO, requiring a Capital Expenditure
(Capex) in the order of $ 5,591,673.83 and an annual Operational
Expenditure (Opex) of $ 636,718.24. A payback of 12 years was
obtained considering a Minimum Acceptable Rate of Return
(MARR) of 5% and an allowance of 50% of capital investment.
The results show that the business will be financially attractive
when considering tax incentives in Brazil for the development of
the national industry in the photovoltaic area. On the other hand,
the environmental impact results for the city of Nova Veneza-GO
show a reduction in CO2 emissions in the order of 184.3 tons per
year and a saving of 503.4 billion liters of water in water
reservoirs of hydroelectric plants in Brazil. This R&D pilot
project research is unique and the first in this area of the power
distribution company Celg Distribution S.A. (Celg-D).
Keywords—Economic Engineering; Smart Grid; Solar Energy;
Photovoltaic Generation; Non Parametric Self-Organizing Map
(PSOM).
I. INTRODUCTION
The Project No. 253 of CELG Distribution (Celg-D) entitled “Application of Intelligent Network (Smart Grid) on the Supervision of Electricity Supply Medium and Low Voltage Using Different Communication Technologies" Research and Development (R&D) was carried out between 2012 and 2013 with the R&D program funds approved by the National Electric Energy Agency (Aneel) [1]. The project covers different areas of applications related to Smart Grids as Advanced Metering Infrastructure (AMI), Distribution Automation (DA) and Integration of Celg-D systems. In addition, it was developed a consumer portal, opinion polls with consumers and a methodology for positioning of concentrators in a mesh network. Thus, the R&D project No. 253 allowed the Energy utility to have a direct contact with technologies and systems for testing and technical evaluation in these areas [2].
Fig. 1 shows a simplified scheme for the installation of meters, with the need of hiring a GPRS data circuit (General Packet Radio Service) to allow sending data collected from the meters for The Supervision Center, located in Goiânia-GO, based on the premises of Celg-D. Thus, meter data are forwarded to a concentrator device with General Packet Radio Service (GPRS data connection) that takes readings directly to the energy utility [2].
International Scientific Journal Journal of Environmental Science
Fig. 1. Network scheme for smart metering in Nova Veneza-GO [2].
In the development of the R&D Project No. 253 the integration and the automation of the System Of Service Orders and the measurement system was carried out by proposing that the service request is not created only by the Commercial Billing System (CBILL), but directly in the Technical System of Operation Management (SGT-OPER). The process was based on information collected from the meters of Customer Consumer Units (CCU) and consolidated by the Measurement Center. The advantage of this proposal is to eliminate the user's need to notify the Call Center, concentrating the interface between the Measurement Center and the SGT-OPER, thereby increasing the efficiency of the system [2].
In the R&D Project No. 253, it was also proposed an approach for positioning concentrators in a ZigBee mesh network of smart meters in order to minimize the average delay of messages sent to the GPRS concentrators, resulting in a better network performance. The K-means clustering algorithm is used to distribute the meters into subnetworks. Queuing Theory is used to estimate the average network delay and Binary Linear Programming (BLP) to determine the location of the concentrators. In addition, computer simulations were carried out to identify network performance from the position determined by the proposed methodology [2]. Other relevant work of the R&D project No. 253 were documented and published in [3] [4] [5] [6] [7].
The R&D project No. 253 provided excellent results and know-how, leading us to propose the R&D project No. 364 “51 Rooftops in Nova Veneza-GO” [8].
In order to verify the feasibility of Grid-Connected Photovoltaic Systems in the city of Nova Veneza, state of Goiás (GO), it was conducted a historical survey of the electricity consumption for consumers connected to the TA and TB transformers shown in Fig. 1.
It is important to note the existence of a smart metering system that monitors 123 consumers of Class B. There are 62 consumers being linked to the extension of low voltage transformer called TA, with nominal power of 112.5 kVA, and 61 consumers linked to the extension of low voltage transformer called TB with the same rating.
The Smart Metering system has the function of record and send data on the energy consumption of Customer Consumer Units (CCU) for the Measurement Center of Celg-D utility with the possibility to acknowledge power failure in certain three-phase consumer units. The system also checks the status of the difference between the power delivered in the secondary of the transformer and the power actually consumed by the CCU, i.e. non-technical losses.
Considering the historical survey of electricity consumption for consumers connected to the TA and TB transformers, it was proposed a methodology to choose 51 roofs in Nova Veneza-GO. The methodology consists of ten stages ranging from the grouping of consumers with the same power consumption profile using a neural network, that is, a Non Parametric Self-Organizing Map (PSOM) [10], until the complete and optimal allocation of financial resources by an Integer Linear Programming.
In the following sections will be presented the methodology for allocating the financial resources of the R&D Project No. 364, as well as a grouping (clustering) proposal for the curves of consumption profiles of the consumer units of Nova Veneza-GO using Artificial Neural Networks. It will be also presented the traditional method used to project the PV systems that will be connected to the grid of Nova Veneza-GO and a socio-economic feasibility study for implementation of solar energy using Grid-Connected Photovoltaic Systems with and without financial incentive to all consumers of the Class B in Nova Veneza-GO.
International Scientific Journal Journal of Environmental Science
http://environment.scientific-journal.com/
II. “51 ROOFTOPS IN NOVA VENEZA-GO” RESEARCH AND
DEVELOPMENT (R&D) PROJECT
The R&D Project No. 364 aims to use the infrastructure of a pre Smart Grid located in the urban area of Nova Veneza-GO where a measuring solution via telemetry allows an automation of a remote reading, cutting and reconnection for a set of clients from the Class B connected to two transformers in Nova Veneza-GO [9]. The consumer behavior analysis was essential for dimensioning the Photovoltaic Systems, the grouping of consumers and the choice of 51 roofs to be benefited with Grid-Connected Photovoltaic Systems.
The structure implemented in the city of Nova Veneza-GO will be extremely important for monitoring the network after the insertion of the 51 solar roofs that will be connected to the electrical system of Celg-D. The monitoring of the energy balance of production and consumption of electricity of the 51 solar roofs will occur seamlessly with:
51 bidirectional smart meters to collect measures every 15 minutes;
Two smart transformers of the city of Nova Veneza-GO substation with remote monitoring;
five power quality meters Customer Consumer Units (CCU);
51 photovoltaic systems connected to the power grid of the utility;
51 interactive inverters connected for real-time monitoring of the energy generated.
The Brazil stands out with a unique solar energy potential. Thus, in addition to the pre Smart Grid structure in Nova Veneza-GO the solar energy potential in this city was of great importance for the realization of the R&D No. 364, with an annual average irradiance in the region of around 5.6 kWh / m
2
[11].
The R&D project No. 364 is considered unique, being the first scalable pilot project of the CELG Distribution (Celg-D). Thus, the experience in the project will allow the local utility to test future problems, such as changing the voltage levels; energy generation affected by shading areas; automatic shutdown of inverters without apparent cause; inappropriate consumer behavior in the use of photovoltaic technology; power balance of generated and consumed energy; among other problems.
III. METHODOLOGY OF FINANCIAL RESOURCES ALLOCATION
AND THE DESIGN OF GRID-CONNECTED PHOTOVOLTAIC
SYSTEMS
The choice of the 51 roofs for the R&D project No. 364 is based on the proposed methodology that uses data collection of electricity consumption of 123 potential consumers connected to the TA and TB transformers. The final goal of the proposed methodology is the allocation of financial resources for the development of projects, purchase and installation of photovoltaic sets that will be installed in the consumer units and will be subsequently approved by Celg-D.
A. Methodology Allocation of Financial Resources
The methodology used for the selection of the 51 roofs between the consumer units of Nova Veneza-GO is shown in the flowchart of Fig. 3. As it can be seen, the flowchart are provided by ten stages ranging from the grouping of consumers with the same consumption profile using a PSOM (Non Parametric Self-Organizing Map) neural network [10] until the completion of allocation of financial resources by using a Integer Linear programming.
Fig. 2. Methodology proposed for allocatingthe financial resources of the
R&D project No. 364 (“51 Rooftops in Nova Veneza-GO”).
International Scientific Journal Journal of Environmental Science
http://environment.scientific-journal.com/
B. Proposal for Grouping of Consumer Profile Using a
PSOM Network
The PSOM network (Non Parametric Self-Organizing Map) is a non-parametric architecture network that presents some changes in the training algorithm for self-organizing maps (SOM), originally proposed by Kohonen [13].
The PSOM network has a training algorithm based on a
pruning procedure, whose ultimate goal is the reduction of the
dimension of the topological map generated. Fig. 4 shows a
comparison of typical architectures for SOM (two-
dimensional structure) and PSOM (dimensional structure),
with the neighboring node j, NEj (Nc) published in [10].
Nc=0
Nc=1
Nc=2
Nc=3
NEj (Nc=4)
(a)
Nc=0
Nc=1Nc=2
NEj (Nc=3){ [ | ( ) | ] }
(b)
Fig. 3. Typical architectures with neighboring node j, NEj (Nc) [10], (a) two-
dimensional structure (SOM) and (b) one-dimensional structure (PSOM).
The PSOM network allows us to group the consumers from TA and TB transformers of Nova Veneza-GO in 12 different groups. The initial settings for the PSOM network training are listed in Table I. The algorithm begins the training with 200 clusters and end with the target of 12 clusters, i.e., there is a reduction in the size of the predicted topological map. The initial learning rate and the number of epochs used were sufficient for the convergence of the training algorithm.
Figures 4 and 5 show the outputs map of the PSOM network for the consumers connected to transformers TA and TB. As can be seen in both figures, it was obtained 12 consumer clusters with the same consumption profile for both analyzed transformers.
Fig. 4. Outputs map of the neural network PSOM for the group of consumers
connected to the transformer TA (12 Groups).
Fig. 5. Outputs map of the neural network PSOM for the group of consumers
connected to the transformer TB (12 Groups).
Thus, Table II shows two groups of consumers with the same “consumer profile” connected to the transformer TA, obtained by using the classifier PSOM network as standards. The Table III shows the grouping of consumers connected to the transformer TB. The obtained clustering is very interesting because in some cases, the results found by the PSOM network do not take into account only the average power consumption, but the consumption pattern of the series over a year, as shown in Figures 6 and 7.
International Scientific Journal Journal of Environmental Science
http://environment.scientific-journal.com/
Fig. 6. Grouping of the PSOM networkfor the group 8 of the TA transformer.
Fig. 7. Grouping of the PSOM networkfor the group 7 of the TB transformer.
International Scientific Journal Journal of Environmental Science
http://environment.scientific-journal.com/
TABLE I. INITIAL NETWORK SETTINGS PSOM.
Initial number of clusters 200
Minimum number of clusters at the end of simulation 12
Initial learning rate 0.05
Maximum number of epochs 1,000
TABLE II. GROUPING OF TA TRANSFORMER CONSUMERS OBTAINED
FROM THE USE OF TRAINED PSOM NETWORK (12 GROUPS).
CCU Annual average
(KWh) Group Phase-Connected
CCU1 3.82 1 C
CCU2 5.64 1 A
CCU3 15.79 1 B
CCU4 22.05 1 A
CCU5 21.26 2 A
CCU6 26.79 2 A
CCU7 31.45 2 ABC
CCU8 37.85 2 A
CCU9 44.15 2 B
CCU10 46.82 2 B
CCU11 56.22 3 A
CCU12 60.44 3 A
CCU13 69.15 3 C
CCU14 69.45 3 B
CCU15 70.53 3 B
CCU16 78.34 3 C
CCU17 82.25 3 B
CCU18 82.68 3 B
CCU19 89.33 4 A
CCU20 102.96 4 C
CCU21 113.34 4 A
CCU22 113.74 4 ABC
CCU23 114.50 4 B
CCU24 124.53 5 B
CCU25 132.20 5 A
CCU26 143.74 5 A
CCU27 145.66 6 B
CCU28 156.16 6 B
CCU29 156.82 6 A
CCU30 160.07 6 A
CCU31 166.64 6 A
CCU Annual average
(KWh) Group Phase-Connected
CCU32 174.79 7 ABC
CCU33 187.31 7 ABC
CCU34 195.29 7 C
CCU35 200.95 7 B
CCU36 203.62 7 A
CCU37 207.42 7 ABC
CCU38 211.82 7 A
CCU39 230.67 8 A
CCU40 231.36 8 A
CCU41 231.83 8 B
CCU42 256.16 8 ABC
CCU43 211.20 9 ABC
CCU44 214.04 9 A
CCU45 238.28 9 ABC
CCU46 240.07 9 C
CCU47 242.02 9 A
CCU48 265.07 10 B
CCU49 265.37 10 A
CCU50 265.88 10 A
CCU51 307.66 10 C
CCU52 328.21 11 ABC
CCU53 336.04 11 ABC
CCU54 349.92 11 C
CCU55 355.10 11 A
CCU56 362.73 11 A
CCU57 364.52 11 A
CCU58 374.06 11 ABC
CCU59 611.47 12 A
CCU60 793.05 12 B
CCU61 1,259.48 12 ABC
CCU62 1,991.16 12 ABC
TABLE III. GROUPING OF TB TRANSFORMER CONSUMERS OBTAINED
FROM THE USE OF TRAINED PSOM NETWORK (12 GROUPS).
CCU Annual average
(KWh) Group Phase-Connected
CCU63 0.00 1 ABC
CCU64 5.30 1 C
CCU65 16.47 1 A
International Scientific Journal Journal of Environmental Science
http://environment.scientific-journal.com/
CCU Annual average
(KWh) Group Phase-Connected
CCU66 19.96 1 C
CCU67 24.55 1 C
CCU68 26.38 1 ABC
CCU69 30.04 1 B
CCU70 43.55 1 A
CCU71 73.26 2 C
CCU72 78.60 2 C
CCU73 78.73 2 B
CCU74 82.85 2 A
CCU75 93.46 3 B
CCU76 98.21 3 A
CCU77 100.75 3 ABC
CCU78 103.84 3 B
CCU79 112.11 3 A
CCU80 130.32 4 B
CCU81 132.83 4 B
CCU82 135.72 4 C
CCU83 137.67 4 A
CCU84 138.30 4 A
CCU85 145.74 4 ABC
CCU86 139.84 5 C
CCU87 151.45 5 C
CCU88 155.48 5 A
CCU89 165.61 5 ABC
CCU90 170.21 5 C
CCU91 170.71 5 A
CCU92 173.26 5 A
CCU93 191.43 6 ABC
CCU94 206.20 6 ABC
CCU95 211.70 6 C
CCU96 212.93 6 C
CCU97 215.44 6 B
CCU98 220.16 6 ABC
CCU99 235.43 6 A
CCU100 241.53 6 A
CCU101 243.74 6 A
CCU102 252.90 6 A
CCU103 308.74 7 ABC
CCU104 322.35 7 B
CCU Annual average
(KWh) Group Phase-Connected
CCU105 327.99 7 B
CCU106 348.94 7 B
CCU107 330.28 8 A
CCU108 335.36 8 ABC
CCU109 341.95 8 ABC
CCU110 353.14 8 A
CCU111 394.32 9 ABC
CCU112 428.89 9 AB
CCU113 431.63 9 ABC
CCU114 476.42 10 ABC
CCU115 484.20 10 B
CCU116 546.43 10 B
CCU117 710.16 11 ABC
CCU118 833.04 11 ABC
CCU119 862.50 11 C
CCU120 959.97 11 ABC
CCU121 943.70 12 C
CCU122 1,098.47 12 ABC
CCU123 1,445.85 12 ABC
C. Design of Grid-Connected Photovoltaic Power Systems
The first step to accomplish the design of Grid-Connected Photovoltaic Systems is to determine the amount of energy to produce, taking the following selection criteria [12]:
Monthly average consumption of electricity;
The space available for installation of the PV modules;
Economic Criteria.
The design and dimensioning of the Grid-Tie Systems can be done through various computational tools such as MatLab, PVSyst, among other softwares. However, in this work, it was used a “Traditional Design”.
In order to design the PV systems, it is important to consider your average monthly electricity consumption. However, in this dimensioning it should be considered that consumers have to pay for the cost of energy availability of Celg-D. Therefore, Aneel regulates the following minimum costs depending on the type of connection of the Class B consumers in the Celg-D network: (1) single-phase (30 kWh); (2) Two-Phase (50 kWh); e (3) three-phase (100 kWh).
In order to dimension using the proposed traditional design, it is necessary to have the technical data of the PV modules that will be used to determine the amount of energy produced by the panel.
International Scientific Journal Journal of Environmental Science
http://environment.scientific-journal.com/
The number of PV modules to be implemented can be inferred from [12] as follows:
a project efficiency:
Np = (Esystem*1000) /(30*Es*n*Pmodule), (1)
where:
Np is the number of modules of the photovoltaic installation;
Esystem is the energy produced by the system [kWh] in the considered time interval, subtracted the availability rate that is 30 kWh for single-phase, 50 kWh for Two-Phase and 100 kWh for three phase;
Pmodule is the peak power [W] of the module;
Es is the daily insolation [kWh/m2/dia]; and
n is the efficiency of the photovoltaic project (default inference) equal to 83% (loss in generating and transmitting power).
Considering a module Pmodule = 250 Wp (module C-SI M 60 NA42117, Bosch manufacturer), a daily insolation of 5.6 kWh/m
2/dia in the city of Nova Veneza-GO and a 83%
efficiency of the photovoltaic, the design of the number of photovoltaic modules for some consumers connected to the TA transformer can be calculated using equation (1). Thus, Tables IV and V show the design for some consumers connected to TA and TB transformers, respectively.
It is possible to calculate the number of modules required for each CCU applying equation (1) for all consumers of Nova Veneza-GO. Thus, Table VI shows the number of Grid-Tie systems required to attain all demands of customers of Nova Veneza-GO connected to TA and TB transformers.
TABLE IV. GRID-TIE SYSTEM DESIGN FOR SOME CONSUMERS CONNECTED
TO THE TRANSFORMER TA.
Consumer Profile System Capacity
(Power)
Generation
Photovoltaics
1st Group (Single Phase
Consumer) (CCU4) * *
1st Group (Consumer Three-
Phase) * *
2nd Group (Single Phase
Consumer) (CCU10) 0.1 kWp 0.15 MWh
2nd Group (Three Phase
Consumer) (CCU7) * *
3rd Group (Single Phase
Consumer) (CCU18) 0.3 kWp 0.44 MWh
3rd Group (Three Phase
Consumer) ** **
4th group (Single Phase
Consumer) (CCU23) 0.6 kWp 0.88 MWh
4th group (Three Phase
Consumer) (CCU22) * *
5th Group (Single Phase
Consumer) (CCU26) 0.8 kWp 1.18 MWh
5th Group (Three Phase
Consumer) ** **
6th Group (Single Phase
Consumer) (CCU31) 0.9 kWp 1.33 MWh
6th Group (Three Phase
Consumer) ** **
Consumer Profile System Capacity
(Power)
Generation
Photovoltaics
7th Group (Single Phase
Consumer) (CCU38) 1.3 kWp 1,92 MWh
7th Group (Three Phase
Consumer) (CCU37) 0.7 kWp 1.03 MWh
8th Group (Single Phase
Consumer) (CCU41) 1.4 kWp 2.06 MWh
8th Group (Three Phase
Consumer) (CCU42) 1.1 kWp 1.62 MWh
9th Group (Single Phase
Consumer) (CCU47) 1.5 kWp 2.21 MWh
9th Group (Three Phase
Consumer) (CCU45) 0.9 kWp 1.33 MWh
10th Group (Single Phase
Consumer) (CCU51) 2.0 kWp 2.94 MWh
10th Group (Three Phase
Consumer) ** **
11th Group (Single Phase
Consumer) (CCU57) 2.4 kWp 3.54 MWh
11th Group (Three Phase
Consumer) (CCU53) 1.9 kWp 2.80 MWh
12th Group (Single Phase
Consumer) (CCU60) 5.5 kWp 8.11 MWh
12th Group (Three Phase
Consumer) (CCU62) 13.6 kWp 20.04 MWh
Ps.: * The energy consumption is equal to or less than the minimum consumption (cost of availability). ** There are not consumers in this category.
TABLE V. GRID-TIE SYSTEM DESIGN FOR SOME CONSUMERS CONNECTED
TO THE TRANSFORMER TB.
Consumer Profile System Capacity
(Power)
Generation
Photovoltaics
1st Group (Single Phase
Consumer) (CCU70)
* *
1st Group (Three Phase
Consumer) (CCU68)
* *
2nd Group (Single Phase
Consumer) (CCU74)
0.3 kWp 0.44 MWh
2nd Group (Three Phase
Consumer)
** **
3rd Group (Single Phase
Consumer) (CCU79)
0.5 kWp 0.74 MWh
3rd Group (Three Phase
Consumer) (CCU77)
* *
4th Group (Single Phase
Consumer) (CCU84)
0.7 kWp 1.03 MWh
4th Group (Three Phase
Consumer) (CCU85)
0.3 kWp 0.44 MWh
5th Group (Single Phase
Consumer) (CCU92)
1.0 kWp 1.47 MWh
5th Group (Three Phase
Consumer) (CCU89)
0.4 kWp 0.589 MWh
6th Group (Single Phase
Consumer) (CCU102)
1.6 kWp 2.36 MWh
6th Group (Three Phase
Consumer) (CCU98)
0.8 kWp 1.18 MWh
7th Group (Single Phase
Consumer) (CCU106)
2.3 kWp 3.39 MWh
7th Group (Three Phase
Consumer) (CCU103)
1.5 kWp 2.21 MWh
8th Group (Single Phase
Consumer) (CCU110)
2.3 kWp 3.39 MWh
8th Group (Three Phase
Consumer) (CCU109)
1.7 kWp 2.51 MWh
9th Group (Consumidor Two
Phase) (CCU112)
2.7 kWp 3.98 MWh
International Scientific Journal Journal of Environmental Science
http://environment.scientific-journal.com/
Consumer Profile System Capacity
(Power)
Generation
Photovoltaics
9th Group (Three Phase
Consumer) (CCU113)
2.3 kWp 3.39 kWh
10th Group (Single Phase
Consumer) (CCU116)
3.7 kWp 5.45 MWh
10th Group (Three Phase
Consumer) (CCU114)
2.7 kWp 3.98 MWh
11th Group (Single Phase
Consumer) (CCU119)
6.0 kWp 8.84 MWh
11th Group (Three Phase
Consumer) (CCU120)
6.2 kWp 9.14 MWh
12th Group (Single Phase
Consumer) (CCU121)
6.5 kWp 9.58 MWh
12th Group (Three Phase
Consumer) (CCU123)
9.7 kWp 14.29 MWh
Ps.: * The energy consumption is equal to or less than the minimum consumption (cost of availability). ** There are not consumers in this category.
D. Application of the Methodology for allocation of financial
resources
Aiming at optimization of resources to be used in the R&D project No. 364, we focused on an allocation of resource optimization by using Integer Linear Programming. The study was based on assumptions of available values for the dimensioning of photovoltaic projects in a Celg-D project, considering that all consumers connected to the two (2) transformers have roofs eligible for the implementation of their photovoltaic system.
The Linear Programming was carried out using the LINDO Software (trial version 6.1). The LINDO (Linear Interactive Discrete and Optimizer) Software is a convenient and powerful tool for solving linear problems, integer and quadratic programming problems.
The number of modules found to consumers of the two (2) transformers were gathered in Table VI.
TABLE VI. NUMBER OF CONSUMERS IN NOVA VENEZA-GO CONNECTED
TO THE TWO (2) TRANSFORMERS IN FUNCTION OF THE PROJECTED MODULES.
Number of Modules Number of Consumers (Potential)
1 13
2 18
3 11
4 7
5 9
6 13
7 2
8 4
9 7
10 2
12 1
14 1
Number of Modules Number of Consumers (Potential)
16 1
17 1
20 1
22 1
24 2
26 1
28 1
33 1
38 1
54 1
The costs of the kits of Grid-Tie Systems were calculated
by applying the Table VII where the values in US dollars for the installed Wp and approved project in Celg-D, were based on an initial market research.
TABLE VII. VALUES OF THE WP INSTALLED WITH THE APPROVED
PROJECT IN CELG-D.
Power (Wp) Values ($)
P < 500 4.84
500 ≤ P ≤ 2.000 3.09
2.000 < P ≤ 3.000 2.55
3.000 < P ≤ 5.000 2.35
5.000 < P ≤ 8.000 2.08
P 8.250 1.95
The results of financial resource allocations are presented in
Table VIII, considering that the project possessed a capital of $ 201,569.55 to be optimized with Integer Linear Programming through the cost function f, based on a maximizing power production (higher Wp).
TABLE VIII. FINANCIAL RESOURCE ALLOCATION OF THE R&D PROJECT
“51 ROOFTOPS IN NOVA VENEZA-GO” USING INTEGER LINEAR PROGRAMMING
IN LINDO SOFTWARE (TRIAL VERSION 6.1), WITH THE COST FUNCTION F1.
International Scientific Journal Journal of Environmental Science
http://environment.scientific-journal.com/
Number of Modules $ 201,569.55
4 1 3,090.73 1,000 4
9 1 5,744.73 2,250 9
22 1 11,455.87 5,500 22
24 2 24,994.62 12,000 48
26 1 13,538.76 6,500 26
28 1 14,580.20 7,000 28
33 1 16,075.04 8,250 33
38 1 18,510.80 9,500 38
54 1 26,304.83 13,500 54
Total 51 201,014.70 85,250 341
Ps.: (1) Number of consumers to be benefited in the two transformers, as prior design. (2) Cost in US dollars. (3) Total number of PV panels of 250 Wp.
When evaluating the results presented in Table VIII it is important to note that the investment of $ 201,569.55 has solutions that attain the 51 planned roofs. In order to do an investment of $ 201,569. , it is achieved a maximization of the generated energy (85,250 Wp), which can be very interesting to “stress” the electrical grid and to evaluate the performance of electrical parameters monitored in tele monitored transformers in the city of Nova Veneza- GO.
IV. SOCIO-ECONOMIC FEASIBILITY STUDIES FOR THE
IMPLEMENTATION OF PHOTOVOLTAIC SYSTEMS THROUGHOUT
THE REGION OF NOVA VENEZA-GO
In order to develop good solutions to meet the demands of business, we need to demonstrate the financial and technical viability. These demonstrations are accomplished when the Business Analyst build scenarios to be presented in a clear and objective manner to stakeholders that the project solution has financial and technical feasibility.
In simple terms, it must be shown that the solution will add value to the business.
The Modern Management uses the Internal Rate of Return (IRR) and Net Present Value (NPV) as tools to demonstrate the financial viability of a business [14].
In order to expand the studies of the R&D Project “51 Rooftops in Nova Veneza-GO”, it was carried out a survey of the energy consumption of all consumers of the Class B in Nova Veneza-GO in order to predict the impacts of deployment of photovoltaic systems connected to the grid for any consumer from class B of the region. The PV design of all these consumers was conducted as shown in section III.
A. Scenario 1: No Financial Incentive
The first scenario refers to a socio-economic feasibility study concerning the installation of micro solar power generation using Grid-Tie Systems, in the absence of financial
incentives, for all Customer Consumer Units (CCU) of Class B in Nova Veneza-GO, excluding public lighting and industries (Class A).
Table IX summarizes the minimum prices of electricity in US dollars for Class B CCU in the state of Goiás, according to kWh consumption range, as retrieved from a query to Celg-D.
TABLE IX. PRICE OF ELECTRICITY CHARGED BY CELG-D FOR CCU.
Minimum prices of electricity (values without taxes)
Class B
Single-
Phase
(30 kWh)
Two-
Phase
(50 kWh)
Three-Phase
(100 kWh)
Residential $ 3.7621 $ 6.2701 $ 12.5403
Other Classes (Industrial -
Commercial - Services -
Government - Public Service -
Self-consumption).
$ 3.7621 $ 6.2701 $ 12.5403
The minimum prices of electricity shown in Table IX were calculated based on fixed-rate plans for which Celg-D charges $ 0.1254/kWh. Note that prices are the same for residential and other consumers inside Class B. Furthermore, the minimum prices of electricity including taxes (ICMS, PIS and Cofins) totalize $ 0.1890/kWh.
The values considered for the PV system are approximate numbers and were calculated to supply 100% of the electrical demand discounting a minimum consumption of electrical energy that corresponds to the cost of availability.
In this work, the PV design did not take into account local neighborhood conditions informed to install the system and that can lead to an electrical production revision due to shading of the modules, such as trees or nearby buildings.
It can be noted that the energy bill will never be zero, as residential consumers in Class B, including farms, must pay monthly, at least the cost of availability. For each consumer profile (single-phase, two-phase and three-phase), the energy bill has a different cost of availability.
For each consumer profile (single-phase, two-phase or three-phase), has a cost of different availability. For example, the cost of availability for a Three Phase Consumer is 100 kWh. Even if the consumer has not used the network power, the concessionaire has complied with the obligation to provide the necessary infrastructure to bring power to the consumer, which is why there is such a minimal cost.
The initial investment (Capital Expenditure- Capex) with the installation of the PV system was obtained for each CCU profile, considering the kits based on table VII. In addition to the purchase of equipment, the costs to provide the installation of the kits were estimated. Thus, Table X shows the CAPEX (year 0) for each residential CCU profile without any government incentives. In the dimensioning, consumers with system capacity (power) of less than 250 Wp, it was used modules of 50 Wp and consumers with greater capability than this value, modules of 250 Wp were used.
International Scientific Journal Journal of Environmental Science
http://environment.scientific-journal.com/
TABLE X. CAPEX FOR CCU PROFILE FROM CLASS B (NOVA VENEZA-GO).
CAPEX (Initial investment in the installation of microgeneration
Table XI presents the annual maintenance value with the PV power system and the annual savings with the system installation. As can be seen, the value of Operational Expenditure (OPEX) is positive for the CCU profile analyzed.
TABLE XI. OPEX FOR CCU PROFILE FROM CLASS B (NOVA VENEZA-GO).
OPEX (Maintenance and annual savings with the installation of
microgeneration Photovoltaic System)
Annual maintenance of systems with modules of 50Wp - $ 7,874.65
Annual maintenance of systems with modules of 250Wp - $ 152,547.84
Annual savings with the installation of the PV system
(Nova Veneza-GO): Systems of 50Wp $ 7,070.29
Annual savings with the installation of the PV system
(Nova Veneza-GO): Systems of 250Wp $ 790,070.44
TOTAL OPEX (Nova Veneza-GO: all the consumers
with modules of 50 Wp and 250Wp) $ 636,718.24
From the considerations made so far, you can apply and simulate the Net Present Value (NPV) for the CCU proposal (Scenario 1). Fig. 8 shows in graph form the NPV considering a Minimum Acceptable Rate of Return (MARR) equal to 9%, 5% and 1%. As can be seen in the results, the solution proposed for a MARR = 9% is prohibitive. For TMA = 5% (see Fig. 3), the payback begins to be viable where the investment return time is approximately 43 years (mean time longer than the lifetime of a PV system). When analyzing the case where an MARR = 1% the investment return time is less than 19 years, which is less than the lifetime of a PV system, however in this case we have nothing attractive at the investment (MARR = 1%).
B. Scenario 2: With financial incentive of 50%
The second scenario refers to the study of socio-economic feasibility for the implementation of micro solar power generation using Grid-Tie Systems with a Government financial incentive of 50% for all Consumer Units in Nova Veneza GO.
From a financial point of view, this strategy is important for both parts. On the one hand, the country increases your energy production and on the other hand, the consumers of the city of Nova Veneza-GO have a financial return in a shorter time, allowing the implementation of the PV systems. Therefore, it is an extremely interesting model for the consumer.
The Table XII presents the CAPEX (year 0) for each residential CCU profile with a government incentive of 50%.
TABLE XII. CAPEX FOR CCU PROFILE FROM CLASS B (NOVA VENEZA-GO).
CAPEX (Initial investment in the installation of microgeneration
Table XIII presents the annual maintenance value with the PV power system and the annual savings with the system installation. As can be seen, the value of OPEX is positive for the CCU profile analyzed.
TABLE XIII. OPEX FOR CCU PROFILE FROM CLASS B (NOVA VENEZA-GO).
OPEX (Maintenance and annual savings with the installation of
microgeneration Photovoltaic System)
Annual maintenance of systems with modules of 50Wp - $ 7,874.65
Annual maintenance of systems with modules of 250Wp - $ 152,547.84
Annual savings with the installation of the PV system
(Nova Veneza-GO): Systems of 50Wp $ 7,070.29
Annual savings with the installation of the PV system
(Nova Veneza-GO): Systems of 250Wp $ 790,070.44
TOTAL OPEX (Nova Veneza-GO: all the consumers
with modules of 50 Wp and 250Wp) $ 636,718.24
Thus, simulations were performed to obtain the Net Present Value (NPV) for the consumers of Class B in Nova Veneza-GO proposal (Scenario 2). Fig. 4 shows in graph form the NPV considering a Minimum Acceptable Rate of Return (MARR) equal to 9%, 5% and 1%. As can be seen in the results, the solution proposed for a MARR = 9% the payback begins in about 17 years, with now is a favorable scenario for an investor by encouraging 50% of government. It is observed that for a 5% and 1% MARR the return occurs in a much shorter time with a very low attractiveness.
C. Socioeconomic and Environmental Impacts
Consider that the “water footprint”, proposed by [15] presented in Table XIV. Also taking in consideration the ten largest Brazilian reservoirs volume of water storage capacity presented in Table XV [15] [16]. Based on data published by the Itaipu Hydroelectric Power Plant in 2013, it is possible to calculate the volume of water consumed for four hydroelectric plants presented in Table XVI.
The Carbon Offset is the reduction in emissions of carbon dioxide or greenhouse gases by purchasing carbon credits or external project financing to the activity of a company, industry or country [17].
With the installation of PV systems for all consumers of the Class B in Nova Veneza-GO, it will be possible to obtain a reduction of 184.3 tons of carbon per year, considering an emission factor of 29 g CO2 / kWh [18] [19].
International Scientific Journal Journal of Environmental Science
http://environment.scientific-journal.com/
Fig. 8. NPV for a MARR = 9%, MARR = 5% and MARR = 1%, without financial incentive (Scenario 1).
Fig. 9. NPV for a MARR = 9%, MARR = 5% and MARR = 1%, with financial incentive (Scenario 2).
International Scientific Journal Journal of Environmental Science
http://environment.scientific-journal.com/
TABLE XIV. “WATER FOOTPRINT” FOR VARIOUS ENERGY SOURCES [15].
Source
“Water Footprint ''
Average (m3/GJ)
“Water Footprint ''
Average (m3/kWh)
Wind power 0 0
Nuclear 0.1 0.00036
Natural gas 0.1 0.00036
Wood 0.2 0.00072
Solar 0.3 0.00108
Oil 1.1 0.00396
Hydropower 22.0 0.07920
Biomass 72.0 0.25920
TABLE XV. THE TEN LARGEST BRAZILIAN RESERVOIRS IN VOLUME OF
WATER STORAGE CAPACITY [15].
Power plant Reservoir
volume (106 m
3)
Flooded
area (km2)
Power
(MW)
Power/Area
(MW/km2)
Serra da
Mesa 55,200 1,784 1,275 0.714686099
Tucuruí 45,500 2,850 8,340 2.926315789
Sobradinho 34,100 4,210 1,050 0.249406176
Itaipu 29,000 1,350 14,000 10.37037037
Furnas 22,950 1,440 1,216 0.844444444
Ilha Solteira 21,166 1,195 3,444 2.882008368
Três Marias 21,000 1,040 396 0.380769231
Porto
Primavera 18,500 2,250 1,800 0.800000000
Balbina 17,500 2,360 250 0.105932203
Itumbiara 17,030 778 2,082 2.676092545
TABLE XVI. ANNUAL WATER CONSUMPTION FOR FOUR HYDROELECTRIC
PLANTS.
Record
production
(mi de
MWh)
Year's
record
Average of
the best
four years
Average of
the best four
years (mi de
GJ)
Water
consumption
(billon m3)
98.631 2013 94.27 339.372 7,466,184
98.112 2012 84.21 303.156 6,669,432
53.413 2008 51.10 183.960 4,047,120
41.434 2009 39.52 142.272 3,129,984
Ps. : 1Itaipu (Brazil),
2Three Gorges (Chine),
3Guri
(Venezuela) and 4Tucuruí (Brazil).
Table XVII shows the reductions of CO2 and water that would be avoided with the implementation of photovoltaic systems connected to the grid for consumers of Class B in Nova Veneza-GO.
TABLE XVII. ENVIRONMENTAL IMPACTS WITH THE INSTALLATION OF
PHOTOVOLTAIC SYSTEMS FOR THE CLASS B OF THE ENTIRE CITY OF NOVA
VENEZA-GO.
Class B
in Nova Veneza-
GO
Annual water savings in Nova Veneza-GO (100%) (mi
m3) 503.4 billon liters
Avoided CO2 emissions (toneladas/ano) 184.3 ton
Figures 10 and 11 shows a Fig. 10. Graphic art to raise awareness of the importance of adherence to the installation of PV systems of all Customer Consumer Units in Nova Veneza-GO.
Fig. 10. Graphic art to raise awareness of the importance of adherence to the
installation of PV systems of all Customer Consumer Units in Nova
Veneza-GO.
Fig. 11. Graphic art to raise awareness of the importance of adherence to the
installation of PV systems of all Customer Consumer Units in Nova
Veneza-GO.
International Scientific Journal Journal of Environmental Science
http://environment.scientific-journal.com/
V. CONCLUSION
In this work, it was presented a method for resource allocation as well as a way of grouping the consumption of consumer profile in Nova Veneza-GO connected to the smart transformers measured using neural networks, more specifically, by using PSOM network.
It is observed that the methodology was fundamental for the selection of the 51 roofs to the Research and Development Project “51 Rooftops in Nova Veneza-GO” once through the clustering and allocation of resources was possible to propose a selection focused on the optimization of financial resources and energy production.
The results show that the installation of PV systems for all the consumers of Class B in Nova Veneza-GO has not only socio-economic impact (pollution reduction in carbon credits) as documented in this paper, but also has an important aspect from an environmental point of view in the economical water (water resources).
It can be concluded with the results that the presence of incentives by the government to the implementation of photovoltaic systems for consumers in the state of Goiás would provide a substantial increase in a matter of feasibility for the implementation of this renewable source in the energy matrix of the state. In this way, there would be an increase in energy supply and a decrease of not clean sources.
REFERENCES
[1] S. G. de Araújo et Al., “Aplicação da Rede Inteligente (Smart Grid) na Supervisão do Fornecimento de Energia Elétrica em Média e Baixa Tensão Utilizando Diferentes Tecnologias de Comunicação” (in portuguese), Projeto de Pesquisa & Desenvolvimento No. 253 Celg-D, PD-6072-0253/2010 (Aneel), 17p.
[2] T. L. Vasques et Al., “Building The Brazilian Smart Grid: Implementation of Smart Grid Technologies in Goiás”, In: Energy for Sustainability 2015, Sustainable Cities: Designing for People and the Planet, 14-15 May, 2015, Coimbra, 6p.
[3] R. V. R. de Carvalho, F. H. T. Vieira, S. G. de Araújo and C. R. Lima, “A Protection Coordination Scheme for Smart Grid Based Distribution Systems Using Wavelet Based Fault Location and Communication Support”, In: 2013 IEE PES Conference on Innovative Smart Grid Technologies Latin America (ISGT LA), Print ISBN: 978-1-4673-5272-7, INSPEC Accession Number: 13653764, DOI: 10.1109/ISGT-LA.2013.6554458, 15-17 April, 2013, São Paulo-SP: IEEE, 8p.
[4] R. V. R. de Carvalho, R. Nielson, C. R. Lima and I. Kopcak, “Operational Conditions Analyses of Distributed Generation (DG) Connected on Power Distribution Grid Based on IEEE Std. 1547”, In: 2013 IEE PES Conference on Innovative Smart Grid Technologies Latin America (ISGT LA), 2013 IEEE PES Conference On, Print ISBN: 978-1-4673-5272-7 2012, INSPEC Accession Number: 13653786, DOI: 10.1109/ISGT-LA.2013.6554481, 15-17 April, 2013, São Paulo-SP: IEEE, 8p.
[5] R. V. R. de Carvalho, F. H. T. Vieira, S. G. de Araújo and C. R. Lima, “A Wavelet Based Fault Location Method with Wireless
Communication Support Applied to Protection Coordination of Smart Grid Based Distribution Systems”, In: International Workshop on Telecommunications (IWT 2013), 6-9 May, 2013, Santa Rita do Sapucaí-MG: Instituto Nacional de Telecomunicações (Inatel), 8p.
[6] G. Souza, F. H. T. Vieira, C. Lima, G. A. de Deus Júnior, M. S. de Castro and S. G. de Araújo, “Optimal Positioning of GPRS Concentrators in Smart Grids Considering Routing in Mesh Networks”, In: International Workshop on Telecommunications (IWT 2013), 6-9 May, 2013, Santa Rita do Sapucaí-MG: Instituto Nacional de Telecomunicações (Inatel), 8p.
[7] G. Souza, F. H. T. Vieira, C. Lima, G. A. de Deus Júnior, M. S. de Castro and S. G. de Araújo, “Optimal Positioning of GPRS Concentrators for Minimizing Node Hopes in Smart Grids Considering Routing in Mesh Networks, In: 2013 IEEE PES Conference on Innovative Smart Grid Technologies Latin America (ISGT LA), Print ISBN: 978-1-4673-5272-7, INSPEC Accession Number: 13653727, DOI: 10.1109/ISGT-LA.2013.6554419, 15-17 April, 2013, São Paulo: IEEE, 7p.
[8] M. S. de Castro et Al., “Projeto 51 Telhados em Nova Veneza-GO: Sistema Solar Fotovoltaico de Microgeração Distribuída Conectada à Rede de Baixa Tensão da Celg-D em 51 Unidades Consumidoras (UC) dentro do Conceito de Redes Elétricas Inteligentes” (in portuguese), Projeto de Pesquisa & Desenvolvimento No. 364 Celg-D, P&D-FCP-Projeto 364/2014 (Aneel), 11p.
[9] G. B. C. Souza, Otimização do posicionamento de concentradores GPRS em redes elétricas inteligentes utilizando programação linear e teoria de filas (in portuguese). Dissertaçãoo de Mestrado, UFG, 2014.
[10] G. A. de Deus Júnior, L. N. de Castro and J. Portugheis, “Mapa Auto-Organizável Não-Paramétrica (PSOM) Aplicado À Decisão De Lógica Majoritária”, SBAI, São Paulo, 1999, pp. 100-155.
[11] E. B. Pereira, F. R. Martins, S. L. de Abreu and R. Rütte, Atlas Brasileiro de Energia Solar (in portuguese). São José dos Campos: Instituto Nacional de Pesquisas Espaciais, 2006.
[12] M. G. Villalva and J. R. Gazoli, Energia solar fotovoltaica: conceitos e aplicações (in portuguese). São Paulo, Érica, 2012.
[13] S. S. Haykin, Neural networks and learning machines. Upper Saddle River: Pearson Education, 3rd ed., 2009.
[14] D. G. Newnan, T. Eschenbach and J. P. Lavelle, Engineering economic analysis. Oxford: Oxford University Press, 2013.
[15] G. N. Pinto de Moura. A Relação entre água e energia: gestão energética nos sistemas de abastecimento de água das companhias de saneamento básico do Brasil (in portuguese). Dissertação de Mestrado em Ciências em Planeamento Energético. Rio de Janeiro: Universidade Federal do Rio de Janeiro, 2010.
[16] E. von Sperling, Morfologia de lagos e represas (in portuguese). Belo Horizonte: Editora da UFMG, 1999.
[17] United Nations Framework Convention on Climate Change (UNFCCC), La Convention-cadre des Nations Unies sur les Changements Climatiques (CCNUCC), “Consolidated baseline methodology for grid-connected electricity generation from renewable sources”, CDM (Executive Board ACM0002), revision to the approved consolidated baseline methodology ACM0002, version 19 May 2006.
[18] A. G. Bumpus and D. M. Liverman, “Accumulation by decarbonization and the governance of carbon offsets”, In: Economic Geograph, Online ISSN: 1944-8287, DOI: 10.1111/j.1944-8287.2008.tb00401.x, vol. 84, issue 2, Boston: Clark University, 2008, pp. 127-155.
[19] D. L. Medeiros, S. T. Oliva and A. Kiperstok, “Methodological Inconsistências Metodológicas em Estimativas de Emissões de Gases do Efeito Estufa na Matriz Elétrica Brasileira” (in portuguese), In: 4th International Workshop Advances in Cleaner Production (IWACP), 2013, 22-24 May, 2013, São Paulo-SP: Unip, 7p.
International Scientific Journal Journal of Environmental Science