Available online at www.sciencedirect.

com

ScienceDirect
Energy Procedia 57 (2014) 168 – 177

2013 ISES Solar World Congress

Computational Tool for Sizing and Assessment of Grid-

Connected Photovoltaic Systems
Gilberto Figueiredoa, Wilson Macêdob, João Tavares Pinhob
a
Instituto de Energia e Ambiente (IEE/USP), Av. Prof. Luciano Gualberto, 1289, São Paulo, Brazil
b
Grupo de Estudos e Desenvolvimento de Alternativas Energéticas (GEDAE/UFPA), Rua Augusto Corrêa, 1, Belém, Brazil.

Abstract

This work presents the utilization of mathematical models which represent the energy processing on Grid-Connected
Photovoltaic Systems. Such models are part present on literature and part proposed by the authors. All models are
implemented in MATLAB GUIDE code which allows the analysis, helps on the design and permits the operational
behavior and energy contribution simulation of GCPV with different sizes. The work also presents comparison
between the data generated by the program and some measured data from installed Grid-Connected Photovoltaic
Systems.

© 2013The
© 2014 The Authors.
Authors. Published
Published by Ltd.
by Elsevier Elsevier
This is Ltd.
an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/3.0/).
Selection and/or peer-review under responsibility of ISES
Selection and/or peer-review under responsibility of ISES.
Keywords: Gri-Connected Photovoltaic Systems, Computational Tool, Energy Forecasting

1. Introduction

The interest in simulating the behavior of Grid-Connected Photovoltaic Systems (GCPV) accurately is
related to the fact that solar energy is particularly different from conventional forms of Distributed
Generation (DG), such as diesel. That said, using models which are able to show the performance of
GCPV in distinct climatic conditions as close as possible to reality becomes a necessity.
The mathematical models used in the software presented on this paper are further shown. The
modeling includes the Photovoltaic (PV) generation and DC to AC conversion stage which comprises the
representative models of Maximum Power Point Tracking (MPPT), DC to AC conversion efficiency and
the power limitation due to power and/or temperature. Electrical losses are also estimated.

1876-6102 © 2014 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/3.0/).
Selection and/or peer-review under responsibility of ISES.
doi:10.1016/j.egypro.2014.10.021
 Gilberto Figueiredo et al. / Energy Procedia 57 (2014) 168 – 177 169

Nomenclature

ܲ௠௣ Power at the maximum power point (MPP) ‫ܫ‬ Generator current
଴
ܲ௉௏ Generator peak power ‫ܫ‬௅ Photo generated current
‫ܪ‬௣௟௔௡௘ Irradiance level at generator’s plane ܸ௢௖ Open circuit voltage
ߛ௠௣ MPP Temperature coefficient ܸ Generator voltage
ܶ௖ Cell temperature ܴ௦ Series resistance
a IxV curve shape factor ߟெ௉௉் Maximum power point tracking efficiency
‫݌‬௖௖ Normalized PV power M0, M1 MPPT model power factors
ܸ௠௔௫ Maximum input voltage ܸ௜௡௜௧௜௔௟ Start-up MPPT voltage
௜௡௩ ଴
ܲ௖௖ି௠௔௫ Maximum inverter input power ܲ௜௡௩ Inverter nominal power
‫݌‬௢௨௧ Normalized inverter output power ‫݌‬௉௏ Normalized PV power
ܸ௠௣ Voltage at MPP k0V, k1V, e k2V Losses linear coefficients
௜௡௩
s0V, s1V e s2V Losses angular coefficients ܲ௠௔௫ Maximum inverter output power
ߟூ௡௩ Inverter efficiency ܶ௜௡௩ଶ ǡ ܶ௜௡௩ଵ Inverter temperatures at t1 and t2
‫ܨ‬௖௔௣ Inverter thermal capacity factor ‫ܨ‬஽ Inverter dissipation factor
ܶ௔௠௕ Ambient temperature ȟ‫ݐ‬ Time interval
ߟ௠௔௫ Maximum inverter efficiency ܶ௠௔௫ Maximum inverter operational temperature
‫݌݁ݐݏ‬ Temperature limitation process counter ܲ௜௡௜௧̴௟௜௠ Power level just before the limitation
‫ܮ‬஽஼ ǡ ‫ܮ‬஺஼ Inverter electrical losses (input and output) ISF Inverter’s sizing factor

2. Mathematical Models

2.1. Photovoltaic Generation

The PV generator converts the energy contained in photons of sunlight into DC electricity. The correct
sizing of the PV generator in means of power capacity and configuration to be installed is essential for the
safe and efficient operation of the system.
The computational tool developed in this paper allows the evaluation of the PV generator operation in
two ways: first using a model that considers the power loss coefficient by temperature. In this part the
software returns to the user graphical information about the relationship between global efficiency and
annual yield of the system related to the ratio between the inverter installed power and PV peak power.
This factor is called Inverter Sizing Factor – ISF (As the interface was made in Portuguese, such variable
will appear in the figures as FDI).
The temperature of the module is almost never maintained at the value set by the Standard Test
Conditions (STC) during of the PV generator operation so it is important that modeling considers this
170 Gilberto Figueiredo et al. / Energy Procedia 57 (2014) 168 – 177

particularity. The term "temperature coefficient" is used to quantify the behavior of electrical parameters
of the module, i.e. it allows analyzing the rate of change with respect to temperature, current, voltage and
power. The Evans model [1] to estimate the maximum power point for a given climatic condition of
irradiance and temperature is shown in equation (1).

଴ ு೛೗ೌ೙೐
ܲ௠௣ ൌ ܲ௉௏ ൈ ൈ ൣͳ െ ߛ௠௣ ൫ܶ௖ െ ܶ௖ǡ௥௘௙ ൯൧ (1)
ுೝ೐೑

Variables with the subscript ref represent the value in STC.

The PV generator is commonly composed by PV modules associations. In the second part of the
simulation the user may choose considering the information obtained by the indicated ISF and must size
the generator by entering the number of PV modules in series and the number of rows in parallel in order
to achieve the desired installed power. The results will help to analyze the performance in terms of
voltage, current and power of the GCPV in different conditions.
To perform this simulation it is used the four parameters model for the PV generator which allows the
calculation of its IxV behavior. This is crucial since the software aims to use models that make possible
analyze continuously the performance of GCPV even considering the voltage variations. Thus, it is
possible to estimate the power to be managed by the inverter considering the dynamic variability of the
PV generator throughout the day. The model is shown in equation (2).

௏೚೎ǡೝ೐೑ ሺ௏ାூήோೞ ሻ
‫ ܫ‬ൌ ‫ܫ‬௅ǡ௥௘௙ െ ‫ܫ‬௅ǡ௥௘௙ ή ‡š’ ቀെ ቁ ή ݁‫ ݌ݔ‬൬ ൰ (2)
௔ ௔ೝ೐೑

2.2. Maximum Power Point Tracking

There is a single point on the IxV curve where the PV generator maximum power is extracted.
Therefore, the inverter must have an algorithm whose tracks the Maximum Power Point (MPP).
Rampinelli [3] designed formulation that estimates the static efficiency of the MPPT depending on the
inverter load under clear sky conditions (irradiance and temperature constants for 1 minute) and assuming
that the MPPT meets MPP at the same range. From the MPPT efficiency curve (measured) as a function
of the load inverter, numerical regression shows that the behavior of the static MPPT is given by equation
(3), where M0 and M1 are power coefficients.
௣೎೎
ߟெ௉௉் ൌ (3)
௣೎೎ ାሺெబ ାெభ ௣೎೎ ሻ

Parameters M0 and M1 can be estimated using equations (4) and (5), obtained by multiple linear
regression.

଴
௜௡௩
‫ܯ‬଴ ൌ ሺെͳǡͺ ൈ ͳͲିହ ሻܸ௠௔௫ ൅ ሺʹǡʹͻ ൈ ͳͲିହ ሻܸ௜௡௜௧௜௔௟ ൅ ͲǡͲͲͳʹܲ௖௖ି௠௔௫ െ ͲǡͲͲʹܲ௜௡௩ ൅ ͲǡͲͳͳ (4)

݅݊‫ݒ‬ Ͳ
‫ܯ‬ଵ ൌ ሺെͷ ൈ ͳͲି଺ ሻܸ௠௔௫ െ ሺ͸ǡͷͷ ൈ ͳͲିହ ሻܸ௜௡௜௧௜௔௟ െ ͲǡͲʹͶܲܿܿെ݉ܽ‫ ݔ‬൅ ͲǡͲʹͷܲ݅݊‫ ݒ‬൅ ͲǡͲͳͶ (5)

2.3. Conversion efficiency as a function of operational voltage

The relationship between inverter conversion efficiency and output power can be expressed in terms of
consumption and losses dependents on loading. However, as shown in [3], the conversion efficiency also
depends on the PV operating voltage. Equation (6) is an adaptation of the classical model and considers
 Gilberto Figueiredo et al. / Energy Procedia 57 (2014) 168 – 177 171

the characteristic parameters of the losses as linear functions of the operating voltage. It is solved as a 2º
equation with pout as variable.

ଶ
‫݌‬௢௨௧ ൌ ‫݌‬௉௏ െ ሾሺ݇଴௏ േ ‫ݏ‬଴௏ Ǥ ܸ௠௣ ሻ ൅ ሺ݇ଵ௏ േ ‫ݏ‬ଵ௏ Ǥ ܸ௠௣ ሻ ή ‫݌‬௢௨௧ ൅ ሺ݇ଶ௏ േ ‫ݏ‬ଶ௏ Ǥ ܸ௠௣ ሻ ή ‫݌‬௢௨௧ (6)

Limitation losses shall be estimated at GCPV design phase. Equations (7), (8) and (9) show the
algorithm implemented in the software to represent the inverter power limitation by power.

௜௡௩ ௜௡௩
݂݅ܲ௢௨௧ ൒ ܲ௠௔௫ ֜ ܲ௢௨௧ ൌ ܲ௠௔௫  (7)

଴
݂݅ܲ௉௏ ൑ ሺ݇଴௏ േ ‫ݏ‬଴௏ Ǥ ܸ௠௣ ሻ ή ܲூ௡௩ ֜ ܲ௢௨௧ ൌ Ͳ (8)

௜௡௩
݂݅ሺ݇଴௏ േ ‫ݏ‬଴௏ Ǥ ܸ௠௣ ሻ ൑ ܲ௢௨௧ ൑ ܲ௠௔௫ ֜ ܲ௢௨௧ ൌ ܲ௢௨௧ (9)

The limitation actually occurs at the DC side and the software also considers this characteristic by
means of the maximum DC accepted power.

2.4. Temperature Estimation

The gradual increase in temperature of the inverter, which occurs in normal operating situations, tends
to be accentuated when power limitation occurs. Once reached the temperature limit, inverters develop a
second power limitation stage, now due to the temperature. This happens in practice by increasing the
operating voltage of the PV generator and the power decreases gradually until the device reaches safe
temperature. The model proposed in [3] and shown in equation (10) considers that all the power not
transformed into useful energy is dissipated as heat by convection and radiation from the inverter to the
environment.

ଵିఎ಺೙ೡ ிವ
ܶ௜௡௩ଶ ൌ ܶ௜௡௩ଵ ൅ ൤൬ ൰ Ǥ ܲ௖௖ Ǥ ȟ‫ݐ‬൨ െ ൤൬ ൰ ሺܶூ௡௩ଵ െ ܶ௔௠௕ ሻǤ ȟ‫ݐ‬൨ (10)
ி೎ೌ೛ ி೎ೌ೛

It is possible to calculate the thermal parameters with equations (11) and (12). Both were designed
from multiple linear regressions using measured values presented in [3]. It considers that FCAP and FD are
dependent on the inverter maximum and nominal power, besides its maximum efficiency and maximum
permissible operating temperature (variables usually provided in the manufacturer datasheet).

଴
௜௡௩
‫ܨ‬௖௔௣ ൌ ሺെʹǡͷ ൈ ͳͲଷ ሻܲ௖௖ି௠௔௫ ൅ ሺʹ͸ǡͺ ൈ ͳͲଷ ሻܲ௜௡௩ െ ሺͷ ൈ ͳͲଷ ሻߟ௠௔௫ ൅ ͷͺͷܶ௠௔௫ ൅ ሺͶͷǡʹ ൈ ͳͲଷ ሻ (11)

௜௡௩ ଴
‫ܨ‬஽ ൌ ሺͳǡͺʹሻܲ௖௖ି௠௔௫ െ ͳǡ͸ͺܲ௜௡௩ ൅ ͲǡͶͷߟ௠௔௫ െ ͲǡͲͳͷܶ௠௔௫ െ ͵ͺǡͺ (12)

2.5. Power behavior at temperature limitation conditions

PV generator oversizing on GCPV is a practice that can be used by designers to maximize the
electricity production and minimize losses, for example. However, the smaller the ISF is, the greater the
incidence of power limitation situations. As the computational tool proposed in this paper seeks to
represent the performance of GCPV as close as possible to reality, it is necessary to represent the process
of temperature limitation and the account of energy losses related to this fact. Figueiredo [4] proposes a
172 Gilberto Figueiredo et al. / Energy Procedia 57 (2014) 168 – 177

logarithmic model, seen in equation (13), to represent the operating power behavior in limitation
situations.

ܲ௠௣ ൌ െͷͳǡ͵͹ ൈ Žሺ‫݌݁ݐݏ‬ሻ ൅ ܲ௜௡௜௧̴௟௜௠  (13)

It is worthy mention that the empirical model proposed in this paper was based on experimental data
obtained from two GCPV which had many power/temperature limitation situations occurring power over
a year. Such systems had the same inverter model and it is desirable to compare the results with data from
different devices. The verification of the reliability of the model through validation with experimentally
measured data is at the end of this work.

2.6. Losses

At last, the representation of GCPV must predict the amount of power lost which is not related to
losses due to the limitation of power and DC/AC convert These losses are relate to cables, fuses, diodes,
and protections both on DC and AC sides.
According to Rodrigues [5] the losses mentioned above are equivalent, at maximum and at standard
conditions, to 2% of maximum power, at both DC and AC sides. This is in relation to PV generator peak
power and maximum power of the inverter at AC side. Assuming that this behavior takes place in a linear
shape, the equations (14) and (15) are valid.

௉೘೛
‫ܮ‬஽஼ ൌ ͲǡͲʹ ൈ బ (14)
௉ುೇ

௉೚ೠ೟
‫ܮ‬஺஼ ൌ ͲǡͲʹ ൈ ೔೙ೡ (15)
௉೘ೌೣ

Modules dispersion losses were considered as 2%.

3. Interface

The models presented in the previous section were implemented in an algorithm developed in GUIDE
(Graphical User Interface) environment of MATLAB in order to facilitate the use of the models, in GCPV
performance analysis and energy estimation.
The program recognizes files with .txt or .dat extensions and formatted as shown in Table 1 as climate
data input.

Table 1 – Example of climate data input.

Month Day Hour Irradiance (W/m²) Temperature (ºC)

1 1 5:45 0 22
1 1 5:55 0 22.5
... ... ... ... ...
1 2 10:00 248 27.5
... ... ... ... ...
12 31 15:20 584.6 31

Fig. 1 shows the first simulation screen. The user can choose the equipment already registered
(modules and inverters), or sign up for a new one provided that user declares the variables necessary for
 Gilberto Figueiredo et al. / Energy Procedia 57 (2014) 168 – 177 173

the models shown so far. In the first part the user can make successive simulations and is able to compare
the results of different equipment in terms of yearly yield and conversion efficiency as a function of ISF.

Fig. 1. First part of the simulation

The program suggests to the user the ISF that sets the highest yield. DC losses (due to limitation
phenomena) and total losses (DC losses plus conversion losses) both related to the total energy supplied
by the PV generation, as a function of ISF, are also shown, as seen in Figure 2. With this data, the user
can define what will be the configuration of the PV array more interesting for him. It is worth considering
the MPPT voltage operating range of the inverter when deciding the number of modules in series.

Fig. 2. Total and DC losses as a function of ISF

174 Gilberto Figueiredo et al. / Energy Procedia 57 (2014) 168 – 177

The user can access the second part of the interface after doing or not the first part. The devices must
be selected and the configuration of the system must be indicated. If the climate data file have not been
already chosen at the first part it will be asked before the second part of the simulation runs. Fig. 3 shows
the results of the second part of the simulation, which performs better return to the user in terms of
performance indicators and energy contribution. The left part shows bar graphs regarding energy (in daily
and monthly means) and monthly performance indicators (Final Yield and Performance Ratio). The right
part returns the instantaneous trace of the AC GCPV power, where can be clearly seen the limitation
phenomena being represented. The period of visualization of this part shall be set by the user who still
gets information about energy, Final Yield and Capacity Factor for the period.

Fig. 3. Results concerning the second part of the simulation

The user can also assess the determined system by means of the voltage (which influences in the
conversion efficiency and well performance of the GCPV), irradiance and power occurrences, as shown
in Fig. 4.

4. Energetic Comparison

The results presented by the program were validated using measured data from five GCPV installed at
the Institute of Energy and Environment, University of São Paulo (IEE / USP). Table 2 shows the module
type, PV generator configuration and the ISF for each GCPV analyzed. At the time of measurement the
systems operated with the same inverter model: SB 1100.
 Gilberto Figueiredo et al. / Energy Procedia 57 (2014) 168 – 177 175

Fig. 4. Occurrences diagrams

Table 2 – Basic characteristics of the GCPV.

System Module Series Parallel ISF

N1 MSX-77 13 1 0,99
N3 MSX-83 12 2 0,52
N4 A-75 12 2 0,56
N5 A-75 11 2 0,61
N6 MSX-70 10 2 0,65

Fig. 5 shows specific comparison of the PV power in occasions of power and temperature limitation
on system N3. The measured power and the calculated values of the PV generator can be seen. The trace
labeled as “managed” represents the amount of available PV power in the inverter input and the trace
labeled as “to be inverted” is the power which was actually inverted after the MPPT and limitation
processes. As can be clearly seen, the calculated trace was pretty close to the measured value, including in
the most critical moments involving the limitation due power and temperature.
Table 3 shows the energetic comparison between the data calculated by the program and the measured
data of the 5 system listed in Table 2. The errors, calculated in percentage, are also shown in Table 3 and
did not exceed 6%, which can be considered as a good approximation in energy year basis.
Nevertheless is important to notice that such errors can be related to the mathematical models
implemented since some are empirical and needs better adjustment. Another thing is the measurement
apparatus which can also incur in errors. Still then such results are pleasant since in most cases the error
did not exceed 2 % for the bigger oversized systems. That fact can be related to the proposed power
limitation due temperature model which, as can be seen in Fig. 4., returned results calculated by the
software close to the actual power developed by the PV generator.
Future works shall involve improvements in the models presented at this work. Attention will be
given at the conversion efficiency model depending on the operational voltage since its approximation
may not include all inverters models which have different inversion topologies that can cause distinct
176 Gilberto Figueiredo et al. / Energy Procedia 57 (2014) 168 – 177

performance characteristics. The model regarding the power behavior at temperature limitation conditions
will also be considered due to the reasons listed above.

DC Power Comparison
Managed

To be inverted

Measured
PV
Po
we
r
(W
)

Fig. 4. Instantaneous comparison

Table 3 – Energetic comparison.

SYSTEM N1
Measured Calculated
1023,6 kWh 1084 kWh
Error (%) 5,9%
SYSTEM N2
Measured Calculated
1967,9 kWh 1982,6 kWh
Error (%) 0,69%
SYSTEM N4
Measured Calculated
1977,7 kWh 1950,4 kWh
Error (%) -1,39%
SYSTEM N5
Measured Calculated
1794,0 kWh 1811,6 kWh
Error (%) 1%
SYSTEM N6
Measured Calculated
1665 kWh 1654,1 kWh
Error (%) -0,65 %

The interface shown in this paper facilitates the handling and understanding of modeling
implemented. This is important in the design and analysis of GCPV as the software is a tool with multiple
 Gilberto Figueiredo et al. / Energy Procedia 57 (2014) 168 – 177 177

graphical results. Some of the program input data are easily found in the equipment datasheets and others
can be extracted from experimental measurements or in the literature.
The tool approached the simulation results of the values found in experimental measurements. Thus,
the better the information provided more reliable is the outcome estimation. The validation done showed
that the data generated by the program are closer to reality mostly in oversized systems.

References

[1] Evans, D. L. Simplified Method for Prediction Photovoltaic Array Output. Solar Energy, vol. 27, n. 6, pp. 555-560, 1981.
[2] De Soto, W. Improvement and Validation of a Model for Photovoltaic Array Performance. Master’s Thesis, Solar Energy
Laboratory, Winsconsin-Madison University, 2004.
[3] Rampinelli, G. A. Estudo de Características Elétricas e Térmicas para Inversores para Sistemas Fotovoltaicos Conectados à
Rede. PhD Thesis, Programa de Pós-Graduação em Engenharia Mecânica, UFRGS, 2010.
[4] Figueiredo, Gilberto. Ferramenta computacional para dimensionamento e avaliação de desempenho de sistemas fotovoltaicos
conectados à rede elétrica. Master’s Thesis, Programa de Pós-Graduação em Engenharia Elétrica, 2012.
[5] RODRIGUES, C. N. Centrais Solares Fotovoltaicas Ligadas à Rede Eléctrica. Prova de Acesso à Categoria de Investigador
Auxiliar, Instituto Nacional de Engenharia, Tecnologia e Inovação, Portugal, 2008.