Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
by
HUIYING ZHENG
SHUHUI LI, COMMITTEE CHAIR
TIM A. HASKEW
JABER ABU QAHOUQ
DAWEN LI
MIN SUN
A DISSERTATION
TUSCALOOSA, ALABAMA
2013
ABSTRACT
Solar photovoltaic (PV) energy is becoming an increasingly important part of the worlds
renewable energy. In order for effective energy extraction from a solar PV system, this research
investigates solar PV energy generation and conversion from devices to grid integration.
First of all, this dissertation focuses on IV and PV characteristics of PV modules and
arrays, especially under uneven shading conditions, and considers both the physics and electrical
characteristics of a solar PV system in the model development. The dissertation examines how
different bypass diode arrangements could affect maximum power extraction characteristics of a
solar PV module or array. Secondly, in order to develop competent technology for efficient
energy extraction from a solar PV system, this research investigates typical maximum power
point tracking (MPPT) control strategies used in solar PV industry, and proposes an adaptive and
close-loop MPPT strategy for fast and reliable extraction of solar PV power. The research
focuses especially on how conventional and proposed MPPT methods behave under highly
variable weather conditions in a digital control environment. A computational experiment system
is developed by using MatLab SimPowerSystems and Opal-RT (real-time) simulation
technology for fast and accurate investigations of the maximum power extraction under high
frequency switching conditions of power converters. A hardware experiment system is built to
compare and validate the conventional and the proposed MPPT methods in a more practical
condition. Advantages, disadvantages and properties of different MPPT techniques are studied,
ii
evaluated, and compared. Thirdly, in order to develop efficient and reliable energy conversion
technologies, this dissertation compares the energy extraction characteristics of a PV system for
different converter configurations. A detailed comparison study is conducted to investigate what
enhancements and impacts can be made by using different bypass diode schemes. It is found that
compared to micro-converter based PV systems, the central converter scheme with effective
bypass diode connections could be a simple and economic solution to significantly enhance PV
system efficiency, reliability and performance. Lastly, the development of coordinated control
tools for next-generation PV installations, along with energy storage units (ESU), provides
flexibility to distribution system operators. The objective of the control of this hybrid PV and
energy storage system is to supply the desired active and reactive power to the grid and at the
same time to maintain the stability of the dc-link voltage of the PV and energy storage system
through coordinated control of power electronic converters. This research investigates three
different coordinated control structures and approaches for grid integration of PV array, battery
storage, and supercapacitor (SC). In addition, other applications including single-phase DirectQuadrature (DQ) control and ramp rate limit control are presented in this dissertation.
iii
DEDICATION
This dissertation is dedicated to everyone who helped me and guided me through the
trials and tribulations of creating this research. In particular, the graduate school of the
University of Alabama and some knowledgeable and up-lifting professors in ECE department
who stood by me throughout the time taken to complete this research.
iv
ID
Diffusion current
IS
Drift current
IL
Photogenerated current
Rp
Parallel resistance accounting for current leakage through the solar cell
Rs
Series resistance which causes an extra voltage drop between the junction voltage
and the terminal voltage of the solar cell
I0
Elementary charge
Absolute temperature
Boltzmann's constant
Ic
Ps
Shading factor that the shaded cell is relevant to the unshaded cell
Vd
Vc
Pc
N D
N A
Eg
Ratio of the present solar irradiation over the nominal irradiation of 1000W/m2
IMPP
ISC
ksc
VMPP
VOC
Koc
I a Va
Instant conductance
I a Va Incremental conductance
ib_ref
isc_ref
Vdc_ref
psto_ref
pdc_ref
pg_ref
ppv
pf
id
iq
Rf
vi
XR
XI
Initial phase
Fundamental frequency
T-1
P(t)
vii
ACKNOWLEDGMENTS
I am pleased to have this opportunity to thank those who gave me an enormous amount of
help and guidance for this research project. My supervisor, Dr. Shuhui Li, has steered me from
the early stages of problem formulation to the clarification and careful presentation of ideas in
this dissertation. He has kept me on the right track while forcing me to discover the hard
problems for myself. His enthusiasm for my research topic and tremendous expertise is very
much appreciated and he has always made time to review my experimental objectives and
conclusions and give excellent guidance, despite his busy schedule.
I would also like to thank all of my committee members, Dr. Tim. A. Haskew, Jaber Abu
Qahouq, Dawen Li and Min Sun for their invaluable input, inspiring questions, and support of
both the dissertation and my academic progress. I would like to thank Dean David Francko and
Dr. Haskew for their assistance at the most difficult time of this journey.
In addition, I would like to thank Dr. Bharat Balasubramanian for opening up a
transformative cooperative program with practical industrials, which provided me with a
wonderful opportunity to apply knowledge to the work in Mercedes- Benz U. S. International,
Inc., Vance, Alabama.
In my long journey through the University of Alabama, the graduate school has been
supporting me all the way to my graduation. With Graduate Council Fellowship, I accumulated
professional knowledge of industrial electrical engineering and adapted myself to the colorful
campus life. With the support of Graduate Student Research and Travel Support, I was able to
viii
ix
CONTENTS
ABSTRACT .......................................................................................................... ii
DEDICATION ..................................................................................................... iv
LIST OF ABBREVIATIONS AND SYMBOLS ...................................................v
ACKNOWLEDGMENTS .................................................................................. viii
LIST OF TABLES ............................................................................................. xiv
LIST OF FIGURES ..............................................................................................xv
LIST OF ILLUSTRATIONS ............................................................................. xix
CHAPTER 1 - INTRODUCTION .........................................................................1
CHAPTER 2- ENERGY EXTRACTION CHRACTERISTIC STUDY OF
SOLAR PHOTOVOLTAIC CELLS, MODULES AND ARRAYS ......................6
2.1 Semiconductor Characteristics and Equivalent Model of a Solar Cell ............6
2.1.1 Silicon Solar Cell...........................................................................................6
2.1.2 Photogenerated Current and Voltage ............................................................8
2.1.3 Equivalent Model of a Solar Cell ..................................................................9
2.2 Energy Extraction Characteristics of PV cells under Uneven Shading
Conditions.............................................................................................................11
2.2.1 Two Series PV Cells under Uneven Shading Condition .............................11
2.2.2 PV Module under Uneven Shading Condition ............................................16
2.2.3 Model Validation .........................................................................................21
2.3 Bypassing Diode Impact to the Characteristics of Solar PV Cells................22
x
xi
xiii
LIST OF TABLES
xiv
LIST OF FIGURES
2.1. Diffusion current, drift current, and depletion zone of a p-n junction ...........7
2.2. Illustration of drift current as well as photogenerated current and voltage ....8
2.3. Solar cell equivalent circuit model .................................................................9
2.4. Solar cell I-V and P-V characteristics ...........................................................11
2.5. Two series PV cells with uneven shading.....................................................12
2.6. Characteristics of two series solar cells ........................................................14
2.7. A PV module connected to an external circuit .............................................18
2.8. Characteristics of PV module (one cell shaded) ...........................................19
2.9. Characteristics of PV module (18 cells shaded) ...........................................20
2.10. Schematics of a PV module connected with bypassing diodes, created
by NI Multisim..............................................................................................24
2.11. Characteristics of a PV module (3 cells with a bypass diode) ......................25
2.12. Characteristics of a PV module (9 cells with a bypass diode) ......................25
2.13. Characteristics of a PV module (18 cells with a bypass diode) ....................25
2.14. Bypass and blocking diodes in a solar PV generator ....................................27
2.15. PV array characteristics (without bypass diode) ...........................................30
2.16. PV array characteristics (one module with a diode) .....................................30
2.17. PV array characteristics (each cell with a bypass diode) ..............................30
2.18. Solar PV generator under an open-loop controlled dc/dc power converter.. 31
xv
xvi
xviii
LIST OF ILLUSTRATIONS
xix
CHAPTER 1
INTRODUCTION
Investment in solar photovoltaic (PV) energy is rapidly increasing worldwide [1]. A gridconnected solar PV system consists of a PV generator that produces electricity from sunlight and
power converters for energy extraction and grid interface control [2, 3]. The smallest unit of a
PV generator is a solar cell and a large PV generator is built by many solar cells that are
connected together through certain series and parallel connections [4].
Although in most power-generating systems, the main source of energy (the fuel) can be
manipulated, this is not true for solar energies [5]. Industry must overcome a number of technical
issues to deliver renewable energy in significant quantities. Control is one of the major enabling
technologies for the deployment of renewable energy systems. Photovoltaic power requires
effective use of advanced control techniques. In all, safe and effective integration of PV system
cannot be achieved without extensive use of control technologies at all levels.
Firstly, unlike a solar thermal panel which can tolerate some shading, PV modules are
very sensitive to shading. Many brands of PV modules can be affected considerably even by
shading of the branch of a leafless tree. If enough cells are hard shaded, a module will not
convert any energy and will, in fact, become a tiny drain of energy on the entire system [2, 6]. In
existing research, most shading studies of a PV system focus mainly on how the I-V and P-V
characteristics of an entire PV system are affected [7-12]. Different from the conventional
approaches, Chapter 2 investigates the characteristics of shaded PV cells, modules, and arrays by
integrating the semiconductor physics characteristics of PV cells and the electrical characteristics
1
power generation from a solar PV system. In order for effective design and development of solar
PV systems in electric power systems, it is important to investigate and compare operating
principles, performance, and advantages or disadvantages of conventional MPPT techniques
used in the solar PV industry, and develop new competent technology for fast and reliable
extraction of solar PV power. In Chapter 3, the dissertation first presents an analysis of PV array
characteristics and the impacts of temperature and solar irradiance on PV array characteristics in
Section 3.1. Section 3.2 investigates conventional fixed-step MPPT techniques used in solar PV
industry. Section 3.3 presents traditional adaptive MPPT techniques, and a proposed
proportionalintegral (PI) based adaptive MPPT approach for fast and reliable tracking of PV
array maximum power. Section 3.4 gives performance evaluation of the conventional and
proposed MPPT methods under stable and variable weather conditions through a computational
experiment strategy. Section 3.5 shows a hardware experiment evaluation of the conventional
and proposed MPPT methods under more practical conditions in a dSPACE-based digital control
environment. Finally, Section 3.6 concludes with the summary of main points.
Thirdly, to make a PV system more efficient and economic, it is necessary to analyze
different converter configurations. Many different converter structures have been developed and
used in a solar PV system. Typical configurations include a central dc/dc/ac converter [14], a
central dc/ac inverter [15, 16], multi-string dc/dc converters plus a central dc/ac inverter [14, 17],
string inverters [15, 16], dc/dc optimizers [16, 17] and microinverters [15, 17, 18]. For all the
different converter structures, the energy extraction characteristics and maximum power capture
capability for all the converter schemes under even solar irradiation are very similar. However,
under shading conditions, the energy extraction depends strongly on what converter structure is
used in a PV system. Therefore, it is important to understand what the differences of energy
extraction characteristics are when using different converter schemes. In [17, 19], it is pointed
out that the string converter system has the advantage in capturing the maximum power of each
string of PV modules separately. In [15, 17], it is commented that micro converter PV system is
effective to overcome shading impact and enhance PV system efficiency. But, no detailed
comparison studies have been conducted previously on PV array performance using different
converter structures. This research first introduces configurations of grid-connected solar PV
system in Section 4.1 and typical PV power converter architectures in Section 4.2 respectively.
PV array models for different converter configurations are discussed in Section 4.3.Section 4.4
and 4.5 investigate PV system energy extraction characteristics with and without bypass diodes,
respectively, for different converter schemes. Finally, Section 4.6 concludes with the summary of
main points.
Last but not least, the control of energy storage is a key component in improving energy
efficiency, security and reliability, which allows the desired active and reactive power delivered
to the grid and at the same time to maintain the stability of the dc-link voltage of the PV and
energy storage system through coordinated control of power electronic converters. Batteries are
the technological solution most commonly employed to help make a PV power smooth and
dispatchable [20]. A battery stores electrical energy in the form of chemical energy. Normally,
batteries perform three main functions in a grid-connected PV system: storing energy into the
batteries when the PV production is high and the grid demand is low, releasing energy to the grid
when the PV production is low or during grid peak demand intervals, and preventing large
voltage fluctuations. Except for batteries, supercapacitor (SC) is usually used in conjunction with
batteries to form an advanced PV energy storage system [20, 21]. However, unlike batteries,
where the voltage remains relatively even over most of the batterys remaining charge levels, a
SCs voltage scales linearly with the remaining energy. This means additional circuitry is
required to make the SC energy usable. In order for effective design, development, and analysis
of integrated PV and Energy storage units (ESU) systems, it is important to investigate operating
principles, performance, and disadvantages and advantages of typical coordinated control
techniques used in the PV and ESU systems. In chapter 5, this research first introduces gridconnected PV and ESU system in Section 5.1. Section 5.2 evaluates control technologies
associated with each individual PV system components. Section 5.3 investigates coordinated
control methods for the integrated PV system. Section 5.4 gives performance evaluation for
coordinated control of PV array and ESU integration with the grid. Other applications including
single-phase DQ control and ramp rate limit control are illustrated in Section5.5. Finally, chapter
5 concludes with the summary of main points in Section 5.6.
Taken as a whole, this research demonstrates some issues of PV energy generation and
conversion from devices to gird integration.
CHAPTER 2
ENERGY EXTRACTION CHRACTERISTIC STUDY OF SOLAR PHOTOVOLTAIC CELLS,
MODULES AND ARRAYS
In most of solar cells, the absorption of photons takes place in semiconductor materials,
resulting in the generation of the charge carriers and the subsequent separation of the photogenerated charge carries. Therefore, semiconductor layers are the most important parts of a solar
cell.
A solar cell is a device that converts the energy of sunlight directly into electricity by the
photovoltaic effect [2]. Although there are many kinds of solar cells developed by using different
semiconductor materials, the operating principle is very similar. The most commonly known
solar cell is configured as a large-area p-n junction made from silicon. When a piece of p-type
silicon is placed in intimate contact with a piece of n-type silicon, a diffusion of electrons occurs
from the region of high electron concentration (the n-type side) into the region of low electron
concentration (p-type side). Similarly, holes flow in the opposite direction by diffusion. This
forms a diffusion current ID from the p side to the n side (Fig. 2.1a). When the electrons diffuse
across the p-n junction, they recombine with holes on the p-type side. The diffusion of carriers
does not happen indefinitely because of an electric field which is created by the imbalance of
charge immediately on either side of the junction which this diffusion creates. The electric field
established across the p-n junction generates a diode that promotes charge flow, known as drift
current IS, that opposes and eventually balances out the diffusion current ID. The region where
electrons and holes have diffused across the junction is called the depletion zone (Fig.2.1b).
(b) Drift current IS from the n side to the p side and the depletion zone
Fig. 2.1. Diffusion current, drift current, and depletion zone of a p-n junction
When a visible light photon with energy above the band-gap energy strikes a solar cell
and is absorbed by the solar cell, it excites an electron from the valence band. With this
newfound energy transferred from the photon, the electron escapes from its normal position
associated with its atom, leaving a localized "hole" behind [2]. When those mobile charge
carriers reach the vicinity of the depletion zone, the electric field sweeps the holes into the p-side
and pushes the electrons into the n-side, creating a photogenerated drift current. Thus, the p-side
accumulates holes and the n-side accumulates electrons (Fig. 2.2), which creates a voltage that
can be used to deliver the photogenerated current to a load. At the same time, the voltage built up
through the photovoltaic effect shrinks the size of the depletion region of the p-n junction diode
resulting in an increased diffusion current through the depletion zone. Hence, if the solar cell is
not connected to an external circuit (switch in the open position in Fig. 2.2), the rise of the
photogenerated voltage eventually causes the diffusion current ID balancing out the drift current
IS until a new equilibrium state is reached inside a solar cell.
Fig. 2.2. Illustration of drift current as well as photogenerated current and voltage
8
When a solar cell is connected to an external circuit (i.e., switch in the close position in
Fig. 2.2), the photogenerated current then travels from the p-type semiconductor-metal contact,
through the wire, powers the load, and continues through the wire until it reaches the n-type
semiconductor-metal contact. Under a certain sunlight illumination, the current passed to the
load from a solar cell depends on the external voltage applied to the solar cell normally through a
power electronic converter for a grid-connected PV system. If the applied external voltage is
low, only a low photogenerated voltage is needed to make the current flow from the solar cell to
the external system. Nevertheless, if the external voltage is high, a high photogenerated voltage
must be built up to push the current flowing from the solar cell to the external system. This high
voltage also increases the diffusion current as shown in Section 2.1.2 so that the net output
current of the solar cell is reduced.
To analyze the behavior of a solar cell, it is useful to create a model which is electrically
equivalent. According to Section 2.1.2, an ideal solar cell can be modeled by a current source,
representing the photogenerated current IL, in parallel with a diode, representing the p-n junction
of a solar cell. In a real solar cell, there exist other effects, not accounted for by the ideal model.
9
Those effects influence the external behavior of a solar cell, which is particularly critical for
integrated solar array study. Two of these extrinsic effects include: 1) current leaks proportional
to the terminal voltage of a solar cell and 2) losses of semiconductor itself and of the metal
contacts with the semiconductor. The first is characterized by a parallel resistance Rp accounting
for current leakage through the cell, around the edge of the device, and between contacts of
different polarity (Fig. 2.3). The second is characterized by a series resistance Rs, which causes
an extra voltage drop between the junction voltage and the terminal voltage of the solar cell for
the same flow of current.
The mathematical model of a solar cell is described by
qVd
V
I c I L I 0 e mkT 1 d ,
Rp
Vc Vd I c Rs
(2.1)
where IL is proportional to the sunlight illumination intensity, m is the diode ideality factor (1 for
an ideal diode), the diode reverse saturation current I0 depends on temperature, q is the
elementary charge, k is the Boltzmann's constant, and T is the absolute temperature [22]. For all
the studies presented in this dissertation, IL=6A, I0=610-6A, RP=6.6, RS=0.005, and T=25
which represents full sun condition used in [23]. Thus, characteristics of a solar cell can either be
simulated using a circuit simulation tool based on the equivalent circuit model or computed
directly by using MatLab based on (2.1). Important characteristics for a solar cell consist of
output current Ic and power Pc versus output voltage Vc characteristics. Figure 2.4 shows typical
I-V and P-V characteristics of a solar cell under ideal condition and with the consideration of
parallel and series resistance obtained by using a Spice simulation tool. As it can be seen from
the figure, if the external voltage applied to the solar cell is low, the net output current of the
solar cell, depending primarily on the photogenerated current, is almost constant. Therefore, as
the external voltage increases, more power is outputted from the solar cell. But, if the external
10
voltage is around the forward conduction voltage of the p-n junction diode, the net output current
drops significantly and the output power reduces.
a) I-V characteristics
b) P-V characteristics
In most conventional studies of a solar PV system, it is usually assumed that all the PV
cells and modules making up a solar PV generator are identical and work under the same
condition [24- 26]. However, in reality, the characteristics of the cells and modules are subject to
some variations. This may happen when uneven sunlight is applied to solar cells, unclean PV
cells, variation and inconsistence of the cell parameters to be expected from manufacturing
process, or other conditions [2, 4].
Figure 2.5 shows the configuration of two series connected PV cells. If both cells are
identical and operate at the same condition, then, the concentration of the photon-excited charge
11
carriers are the same in both cells. Thus, the photogenerated current in one cell can flow through
the second cell continuously and then to the external system, and the output voltage of the two
cells is the summation of the photogenerated voltage of both cells.
Rs1
Rp1
IL1
Vs
Rs2
Rp2
IL2
Fig. 2.5. Two series PV cells with uneven shading
Nevertheless, if the two cells operate at different conditions, such as one cell is at the full
sun while the other is shaded, then, the photon-excited charge carriers in the unshaded cell are
more than the photon-excited charge carriers in the shaded cell. Thus, the photocurrent of the
unshaded cell cannot completely flow through the shaded cell due to the insufficient charge
carriers, causing the rest of the photon-excited charge carriers to be accumulated on the p- and nside of the unshaded cell. Then, the output voltage of the unshaded cell rises, which causes (a)
more diffusion current through the p-n junction of the unshaded cell (Fig. 2.2) and (b) some of
the photogenerated current of the unshaded cell being pushed through the parallel resistance of
the shaded cell until an equilibrium state is reached.
If assuming that the parameters of the two cells are identical, the mathematical model of
the series PV cells under the shading condition is described by
12
qVd 1
V
I c I L I 0 e mkT 1 d 1 , Vc1 Vd 1 I c Rs
Rp
(2.2)
d2
qV
V
mkT
1 d 2 , Vc 2 Vd 2 I c Rs
I c (1 ps ) I L I 0 e
Rp
(2.3)
Vs Vc1 Vc 2
(2.4)
where ps stands for the shading factor that the shaded cell is relevant to the unshaded cell, and IL
represents the photogenerated current of unshaded cell under the full sun condition, Vd1 and Vd2
and Vc1 and Vc2 represent p-n junction diode voltages and output voltages of the unshaded and
shaded cells, respectively. Based on (2.2) to (2.4), a system of nonlinear equations can be
developed as
f1 Vd 1 , Vd 2 0
f 2 Vd 1 , Vd 2 0
(2.5)
Then, for a given voltage applied to the PV cells, voltage Vd1 and Vd2 can be solved
numerically by using Newton-Raphson algorithm in the following steps:
a)
Initial estimation:
(2.6)
f1 Vdk1 f1 Vdk2
J
k
k
f 2 Vd 1 f 2 Vd 2
c)
(2.7)
(2.8)
Error calculation:
err
f1 Vd k 1 f 2 Vd k 1
2
13
(2.9)
e)
predefined threshold).
For detailed study under shading condition, the I-V and P-V characteristics of the series
PV cells can be obtained through either simulation of Fig. 2.5 or the numerical computation
shown above. Although simulation of Fig. 2.5 is convenient to implement by using a circuit
simulation tool, numerical computation approach is more practical for a large solar PV system
14
that contains thousands of solar cells. It is necessary to point out that the study based on both
approaches can provide a cross validation mechanism.
Figure 2.6 shows the I-V and P-V characteristics under three shading conditions. The
shading factors are 0%, 50%, and 100%, where 0% represents the unshaded condition and 100%
stands for the completely shaded condition. This shading representation is applicable to the rest
of this research. Usually, the power dissipated by a shadowed cell increases cell temperature,
which changes the solar cell electrical properties by varying the values of I0 and IL slightly.
However, detailed temperature change, involving complicated heat transfer issues, is very hard
to calculate. Therefore, the temperature change caused by the power dissipation of a shadowed
cell is not considered here. According to Fig. 2.6 as well as other results, the following remarks
are obtained.
1)
When both cells operate at the same condition and under the same illumination
intensity, the photogenerated voltages are the same (Figs. 2.6c and 2.6e) and the P-V
characteristics are identical for both cells (Figs. 2.6d and 2.6f). Compared to a single cell, the
output voltage and power at the maximum power point are increased.
2)
If one cell is 100% shaded while the other is in full sun, the photogenerated current
of the unshaded cell has to pass through the parallel resistor of the shaded cell. Moreover, to
push the current through the high parallel resistance, the photogenerated voltage of the unshaded
cell must be high (Fig. 2.6c), which increases the diode drift current of the unshaded cell and
reduces the net output current significantly so that the actual output power is very low (Figs. 2.6b
and 2.6d).
3)
If one cell is partially shaded while the other is in full sun, the unshaded cell has
more photon-excited charge carriers than the shaded one. Therefore, part of the photon-excited
15
charge carriers of the unshaded cell passes through the shaded cell and part of charge carriers of
the unshaded cell has to pass through the parallel resistor of the shaded cell so that the terminal
voltage of the shaded cell is reversed. Thus, the unshaded cell generates power while the shaded
cell absorbs power (Figs. 2.6d and 2.6f), depending on the external voltage applied to the two
series solar cells. Similarly, to push the current through the high parallel resistance, the
accumulated photogenerated voltage of the unshaded cell must be high (Fig. 2.6c), which
increases the diode diffusion current of the unshaded cell so that the net current actually passing
through the parallel resistor of the shaded cell is very low (Fig. 2.6a).
4)
Under partial shading conditions, the power absorbed by the shaded cell is
influenced by the applied external voltage. The higher the external voltage, the less the current is
pushed through the parallel resistor of the shaded cell by the unshaded cell, the less the reverse
terminal voltage of the shaded cell and the less the shaded cell absorbs power. When the external
voltage is higher than the diode forward conduction voltage of the unshaded cell, the shaded cell
basically starts to generate power (Fig. 2.6f). In other words, increasing external voltage applied
to the two series of cells could prevent the shaded cell from becoming a hot spot under an uneven
shading condition. But, this special regularity cannot be seen effectively by just looking at the
overall P-V characteristics as shown by Fig. 2.6b.
Normally, solar cells are connected in series to form a module that gives a standard dc
voltage. A module typically contains 28 to 36 cells in series (Fig. 2.7), to generate a dc output
voltage of 12V in standard illumination condition. The 12V module can be used singly or
16
connected in series and parallel into an array with a large voltage and current output, according
to the power demand by an application.
The I-V and P-V characteristics of a PV module under a shading condition are more
complicated, depending on how many cells are shaded and what the shading factor of each cell
is. Assume there are N cells in a PV module and the shading factor of the ith PV cell in the
module is pi. Then, the mathematical model of a PV module under a shading condition is
described by:
qVdi
V
I c (1 pi ) I L I 0 e mkT 1 di , Vci Vdi I c Rs
Rp
(2.10)
Vs Vc1 Vc 2 Vc ( n 1) VcN
(2.11)
where pi stands for the shading factor of the ith cell relevant to the full sun condition, IL
represents the full sun photogenerated current, and Vdi and Vci are the p-n junction diode voltages
and output voltages of the ith PV cell. Similar to Section 2.2.1, a system of N nonlinear equations
can be developed as shown by (2.12).
f1 Vd 1 , , VdN 0
f N Vd 1 , , VdN 0
(2.12)
Then, for a given voltage applied to a PV module, voltage Vd1, Vd2, VdN can be solved
numerically by using Newton-Raphson algorithm in the following steps: 1) obtaining initial
estimation values of PV cell voltages, 2) computing the Jacobian matrix, 3) computing the
correction and updating PV cell voltages, 4) calculating the error, and 5) repeating steps (2) to
(4) until a stop criterion is reached [27]. After the completion of the iteration, solutions of Vd1,
Vd2, VdN for all PV cells are available for both shaded and unshaded cells. It is necessary to
point out that the initial estimation is vital for the stability and convergence of the NewtonRaphson algorithm, which is achieved based on the knowledge and estimation of a common
17
voltage range for a shaded or unshaded PV cell. In addition, before the iteration process, PV cells
with the same shading factor are regrouped together, which can greatly reduce the number of the
nonlinear equations and accelerate the numerical computation. It is worth noting that the
Bishops numerical program based on an equivalent PVNet is another approach that was
developed and used to investigate the electrical behavior of solar cell interconnection circuits as
presented in [28].
Shade
Vs
The I-V and P-V characteristics of the PV module can be obtained through either
numerical computation or simulation of Fig. 2.7. Figure 2.8 shows the characteristics of a PV
module when the shading factors of one cell are 0%, 50%, and 100%, respectively, while the
other cells are in full sun.
18
As it can be seen from the figure, if all the cells are in full sun irradiation and have the
same operating condition, the current from each cell is the same, and the output voltage and
power of the PV module are enhanced significantly due to the fact that more cells are connected
in series. But, this situation is completely different even when only one cell is shaded (Fig. 2.8a
and 2.8b). Due to the shading of one cell, part of charge carriers of the unshaded cells must go
through the parallel resistor of the shaded cell so that the terminal voltage of the shaded cell is
reversed (Fig. 2.8e).Thus, the unshaded cells generate power while the shaded cell absorbs
19
power (Fig. 2.8d and 8f). Similarly, to push the current through the high parallel resistance of the
shaded cell, the accumulated photogenerated voltage of each unshaded cell must be high (Fig.
2.8c) so that the net series voltage of all unshaded cells causes a high current through the parallel
resistor of the shaded cell (Fig. 2.8a) and a high reverse terminal voltage on the shaded cell (Fig.
2.8e), which results in a high absorbing power by the shaded cell especially when the external
voltage applied to the PV module is low (Fig. 2.8f). This high absorbing power may damage the
shaded PV cell.
20
Figure 2.9 shows the characteristics of the PV module when 18 out of the 36 cells are
shaded. The shading factor, identical for all the 18 shaded cells, is 100%, 50% and none.
Compared to Fig. 2.8, when there are more cells shaded in a PV module, the net output voltage
of the unshaded cells is smaller and is applied to the shaded cells in a distributed manner. Hence,
the reverse voltage applied to the parallel resistor of each shaded cell is lower (Fig. 2.9e) and the
absorbing power by each shaded cell is decreased (Fig. 2.9f). Compared to Fig. 2.8f, the chance
for a shaded cell to become a hot spot is reduced, implying that a single shaded cell condition is
more hazardous to affect proper function of a PV module.
convenient to build by using professionally developed circuit components. The procedure for the
PSpice-based simulation includes: 1) drawing circuit schematics, as illustrated by Figs. 2.3, 2.5
and 2.7; 2) setting up circuit parameters of the PV system; 3) simulating the circuit; 4) plotting
the results. According to Fig. 2.3, each PV cell has four components, including two resistors, one
diode, and one ideal current source. For a PV module containing 36 cells, there would be 144
components.
The model validation involves the development of computer program using the NewtonRaphson algorithm and the building of the PV simulation system using NI Multisim. For the
PSpice-based simulation, each circuit component of a PV cell is treated as a different simulation
element. Therefore, solar PV system simulation using NI Multisim is extremely expensive in
terms of computing speed and memory requirements. However, for the computer program
especially developed for the PV system study, the PV cells having the same operating conditions
are first regrouped automatically before the simulation. Therefore, both the computing speed and
memory requirement are much more efficient, particularly for a large PV array. The results
generated using the two different approaches are compared for different case studies, including
PV cells (Fig. 2.6), PV modules (Figs. 2.8 and 2.9), and small-scale PV arrays. The comparisons
always show the same results generated by both approaches (Figs. 2.6, 2.8, 2.9, 2.11, 2.12 and
2.13), demonstrating that it is effective and accurate to use the models and algorithm developed
in this chapter for small- and large-scale PV system studies (Section 2.4).
In photovoltaic industry, external bypass diodes in parallel with a series string of cells are
normally utilized to mitigate the impacts of shading on P-V curves. The polarity of the bypass
22
diode is reversed with respect to the PV cells [2]. Consequently, reverse bias of the cells
corresponds to the direct bias of the bypass diode which provides a bypass for the current
generated by other cells. With bypass diodes, the I-V and P-V characteristics of a PV module are
more complicated [36].
Normally, a bypass diode is applied to a PV module or a group of series PV modules [712]. For research purpose, however, different bypass diode schemes within a PV module will be
studied in this dissertation. Figure 2.10 shows a bypass diode arrangement, in which a bypass
diode is applied to each three series PV cells. For a general case, it is assumed that there are M
bypass diodes with each bypass diode being applied to L=N/M series PV cells. Then, the current
and voltage relations of the PV cells connected with the ith bypass diode and overall system
current and voltage are described by
I ci (1 pij ) I L I 0 e
qVdij
mkT
1 Vdij Rp
I s I ci I 0 e
qV pdi
mkT
(2.13)
(2.14)
(2.15)
where pij stands for the shading factor relevant to the full sun condition for the jth PV cell within
the ith bypass diode group, Vdij and Vcij represent p-n junction diode voltage and PV cell output
voltages of the jth PV cell within the ith bypass diode group, Ici is the output current of the series
PV cells within the ith bypass diode group, and Vpdi represents the voltage applied to the ith
bypass diode.
Then, similar to (2.12), a system of N nonlinear equations can be developed and voltage
Vd1 to VdN can be solved numerically by using Newton-Raphson algorithm for a given external
23
voltage Vs applied to the PV module. If some of the PV cells within the PV module operate at the
same condition, the numerical computation could be simplified considerably.
Fig. 2.10. Schematics of a PV module connected with bypassing diodes, created by NI Multisim
Figures 2.11- 2.13 show the characteristic of the PV module when one PV cell in the
module is shaded for three different bypass diode arrangement schemes: three series cells with a
bypass diode, nine series cells with a bypass diode, and eighteen series cells with a bypass diode.
From the figures, other case studies, and comparison with Section 2.2, it is concluded that:
1)
When a PV cell is shaded, there are two possible paths for the current generated by
other unshaded cells to pass through. One is through the shaded cell and parallel resistor of the
shaded cell; the other is through the bypass diode. The condition for the current passing through
the bypass diode is that the resultant reverse voltage of the series cells in parallel with the bypass
diode must be larger than the forward conduction voltage of the bypass diode.
24
a) I-V characteristics of PV
a) I-V characteristics of PV
a) I-V characteristics of PV
module
module
module
b) P-V characteristics of PV
b) P-V characteristics of PV
b) P-V characteristics of PV
module
module
module
c) Shaded cell terminal voltage c) Shaded cell terminal voltage c) Shaded cell terminal voltage
characteristics
characteristics
characteristics
Fig. 2.11.Characteristics of
a PV module
a PV module
a PV module
25
2)
When the bypass diode turns on, the voltage applied to the shaded cell equals to the
photogenerated voltages of the unshaded cells within the bypassing cell group plus the bypass
diode forward conduction voltage. Therefore, the less the PV cells within a bypassing cell group,
the smaller the reverse voltage which is applied to a shaded cell (Figs. 2.11c, 2.12c, and 2.13c)
and the less the shaded cell absorbs power (Figs. 2.11d, 2.12d, and 2.13d). In other words, to
prevent a shaded cell from becoming a hot spot, the number of series PV cell within a bypassing
cell group should be properly designed.
3)
With bypass diodes, the I-V and P-V characteristics of a PV module is more
complicated and different from the traditional understanding of the photovoltaic I-V and P-V
characteristics. An important issue, as it can be seen from Figs. 2.11b, 2.12b, and 2.13b, is that
the P-V characteristics of a PV module may contain multiple peaks. Hence, using traditional
maximum power point tracking approaches, one may get into a local peak point so that the
efficiency of the PV module would be reduced greatly.
By comparing Figs. 2.8 and 2.9 with Figs. 2.11-2.13, it can be been that bypass diodes of
a PV module can reduce absorbing power of shaded cells within the PV module and improve the
performance of PV system.
There are generally two ways to connect PV modules into an array. The first approach
connects modules in series into strings and then in parallel into an array. The second approach
first wires modules together in parallel then combines those units in series. Both connections are
equivalent if all the cells and modules are identical and work at the same condition. But, if
sunlight is applied unevenly to different PV cells as well as shading or other impacts, the second
26
connection approach could cause many very bothersome problems [2]. Figure 2.14 shows a
series-parallel PV array connection with a dc/ac power converter, in which the converter handles
both maximum power point tracking (MPPT) and grid interface control of the PV array [24]. At
the top of each string in Fig. 2.14, a blocking diode is used to prevent a shaded or malfunctioning
string from withdrawing current from the rest strings that are wired together in parallel.
For the series-parallel connected PV array, the voltage applied to each string of the PV
modules is the same. However, the P-V and I-V characteristics of each string could be different
depending on how many cells in a string are shaded and how much the shading factors are. For
each string, the mathematical procedure to obtain P-V and I-V characteristics is very similar to
Section 2.3 except that the external voltage applied to each string equals to the sum of
photogenerated voltages of all series connected PV modules. Then, with the consideration that
27
the output current of the PV array is the sum of currents of all parallel strings, characteristics of
the PV array can be achieved quickly through numerical computation. It is necessary to point out
that for any PV cells having the same operating condition within a string, combining those PV
cells into one mathematical equation could significantly accelerate the numerical computation
speed.
Figures 2.15 to 2.17 show a comparative study of PV array characteristics for three
different bypass diode conditions, i.e., no bypass diode employed, one bypass diode for each PV
module, and one bypass diode for each PV cell. The PV array has a configuration of 10 parallel
strings with each string containing 20 modules. Assume there are 19 shaded modules in the 1st
string, 17 in the 2nd string, 15 in the 3rd string and 1 in the last string. In each shaded module,
there is one shaded cell only, which is the worst condition that would damage a PV cell
according to Section 2.2. The shading factors are 0%, 50% and 100%, respectively. From the
figures, other case studies, and comparison with Section 2.3, the following properties are
obtained:
1) If no bypass diodes are applied, the PV array characteristics can be shifted
significantly by shaded cells (Fig. 2.15a and Fig. 2.15b). The degree of the shift depends on how
many strings contain shaded cells, how many shaded cells are in each string and how much the
shading factors are. When there is only one shaded cell in a string, all the photogenerated
voltages of the unshaded cells in that string are applied to the shaded cell (Fig. 2.15c), which
would cause a high risk to damage the shaded cell due to the high absorbing power of the shaded
cell (Fig. 2.15d).
2) If each PV module has one bypass diode, it is found that there is an improvement in
the PV array characteristics under shading conditions depending on the distribution of the shaded
28
cells in the PV array. For each string if the number of the shaded cells is the same, the best
situation is that all the shaded cells appear in one module. However, if the shaded cells are
distributed evenly in different modules in a string, the enhancement of the PV array
characteristics is trivial (Figs. 2.16a and 2.16b). If there is only one shaded cell in a module,
then, all the photogenerated voltages of the unshaded cells in that module are applied to the
shaded cell (Fig. 2.16c). Compared to Fig. 2.15d, the absorbing power of the shaded cell under
100% shading condition is reduced a lot but changes very little for 50% shading condition.
Another impact of the bypass diodes is that multiple peaks would result in the P-V characteristics
of the PV array. The extent of the multiple peaks depends on the distribution of the shaded cells
in the PV array as well as the number of parallel strings and the number of series modules in
each string. For Fig. 2.16b, multiple peak impact can be seen clearly when the figure is enlarged.
Hence, using traditional MPPT approaches [37-40], one may get into a local peak power point so
that the efficiency of the PV module would be reduced greatly.
3) If each PV cell has a bypass diode, the influence of the shaded cells to the PV array
characteristics is significantly reduced. Compared to both Figs. 2.15d and 2.16d, the absorbing
power of the shaded cell is very small (Fig. 2.17d). Under the condition that the number of the
shaded cells is significantly less than that of the unshaded cells, the P-V characteristics of the PV
array is very close to the unshaded condition no matter how the shaded cells are distributed in the
PV array (Fig. 2.17b). Therefore, with a bypass diode for each PV cell, it is more convenient to
manage the MPPT control of the PV array even under shading conditions, implying that a new
solar PV cell design with a bypass diode would be a significant benefit for extraction and
management of solar PV energy.
29
None
50%
100%
20
100
200
300
400
40
None
50%
100%
20
0
500
Current (A)
Current (A)
Current (A)
40
60
60
60
100
200
300
400
40
None
50%
100%
20
0
500
100
200
Vs (V)
Vs (V)
300
400
500
Vs (V)
10
5
10
5
0
100
200
300
400
None
50%
100%
15
500
100
200
300
400
None
50%
100%
15
Power (kW)
Power (kW)
Power (kW)
None
50%
100%
15
20
20
20
10
5
0
500
100
200
Vs (V)
Vs (V)
300
400
500
Vs (V)
Voltage (V)
Voltage (V)
0
-10
-20
None
50%
100%
-30
-40
100
200
300
400
-20
500
None
50%
100%
0.5
-10
-30
0
None
50%
100%
Voltage (V)
10
-0.5
100
200
300
400
500
100
200
300
400
500
Vs (V)
Vs (V)
Vs (V)
c) Shaded cell terminal voltage c) Shaded cell terminal voltage c) Shaded cell terminal voltage
characteristics of the last string characteristics of the last string characteristics of the last string
100
-100
None
50%
100%
-200
0
-50
-100
-150
100
200
300
400
500
Vs (V)
None
50%
100%
Power (W)
Power (W)
Power (W)
-300
50
0
-2
-4
100
200
300
400
500
Vs (V)
None
50%
100%
100
200
300
400
500
Vs (V)
characteristics
characteristics
characteristics
The behavior of the solar PV system is further examined under more realistic transient
conditions through a virtual experiment by using MatLab SimPowerSystens, which includes: 1)
actual circuit connection of the solar PV array, 2) open-loop controlled power converter
including inductors and capacitors, and 3) losses of the system. Figure 2.18 shows the transient
simulation system. The dc voltage source stands for the dc-link voltage between the dc/dc
converter and the dc/ac inverter (Fig. 2.14). The dc/dc converter is a boost converter, i.e., power
flows from the PV array to the dc voltage source. The PV array is represented by a subsystem
containing all the PV modules in series and parallel. At each time instant, the Newton-Raphson
algorithm is used to find the current and voltage of each solar cell. The parameters of the solar
PV system are the same as those used in the characteristic study (Figs. 2.15-17). The number of
series and parallel PV modules are 20 and 10, respectively. Major measurements include current,
voltage and power of PV cells, modules, and array under test. For power measurement, generator
sign convention is used, i.e., power generated by a PV cell, module, or array to the dc source is
positive.
600
200
Voltage (V)
300
-25
Power
-10
-50
-15
-75
-20
-100
-25
-125
100
0
-5
Power
400
Power (kW)
Voltage (V)
500
0
Voltage
Power (W)
Voltage
4
Time(s)
0
8
-30
4
Time(s)
-150
8
last string
Fig. 2.19. Transient simulation results of a PV array relevant to
the 100% shading condition applied in Fig. 2.16
In Fig. 2.18, the average power converter model [41] is used, in which the duty ratio is a
ramp function of time, which causes the voltage applied to the PV array increases with the time
until the full dc source voltage is reached. Figure 2.19 shows the transient results corresponding
to the 100% shading condition used in Fig. 2.16. The dc source voltage is 500V. As it can be
seen from Fig. 2.19a, the voltage applied to the PV array increases with time. The output power
of the PV array increase, reaches maximum output power, and then decreases, a phenomenon
similar to Fig. 2.16b. The terminal voltage of the shaded cell is around -20V before the bypass
diode turns on (Fig. 2.19b) and the absorbing power of the shaded cell is about 120W (Fig.
2.19b), which is consistent with the steady-state characteristics shown in Fig. 2.16d. Under the
uneven shading condition and a bypass diode for each PV module, the output power of the PV
array also shows the multiple peaks (Fig. 2.19a) in the transient environment, which is consistent
with Fig. 2.16b. For all the other conditions, the results obtained through the transient simulation
experiment agree with stead-state characteristic results, demonstrating that the models and
32
Newton-Raphson algorithm are suitable for transient analysis of power converter controlled solar
PV systems.
2.6 Conclusions
how many shaded cells are in each string. If there is only one shaded cell in a string, the shaded
cell would be in the worst condition due to its high absorbing power. If each PV module has one
bypass diode, the improvement of the PV array characteristics depends on the distribution of the
shaded cells in the PV array. The best situation is that all the shaded cells appear in one module.
However, if the shaded cells are distributed evenly in different modules in a string, the
enhancement of the PV array characteristics is trivial. If each PV cell has a bypass diode, the
influence of the shaded cells on the PV array characteristics is significantly reduced in various
aspects no matter how the shaded cells are distributed in the PV array, implying that a new solar
PV cell design with a bypass diode would be a significant benefit for energy extraction and
management of solar PV energy (Chapter 4).
The models developed in this chapter as well as the NewtonRaphson algorithm
applications are suitable for transient analysis of power converter-controlled solar PV systems,
making it possible to develop and test advanced MPPT control strategies for solar PV systems
under shading conditions through virtual computer experiments.
34
CHAPTER 3
A FAST AND RELIABLE APPROACH FOR MAXIMUM POWER POINT TRACKING
PV generation systems have two major problems: the conversion efficiency of electric
power generation is very low (9-17%) [42], especially under low irradiation conditions; the
amount of electric power generated by solar arrays changes continuously with weather
conditions. The power delivered by a PV system of one or more photovoltaic cells is dependent
on the irradiance, temperature, and the current drawn from the cells. In general, there is a unique
point on the I-V and P-V curve, called the maximum power point (MPP), at which the entire PV
system operates with maximum efficiency and produces its maximum output power. The
location of the MPP is not known, but can be located, either through calculation models or by
searching algorithms. To maximize the output power of a PV system, continuously tracking the
maximum power point of the system is necessary.
There are many different approaches to maximizing the power from a PV system. These
range from using simple voltage relationships, to more intelligent and adaptive based algorithms.
Typical MPPT techniques that have been proposed in the literature include the Short-Circuit
Current method [43], Open-Circuit Voltage method[44], Perturb and Observe (P&O) methods
[45, 46], Incremental Conductance (IC) methods [47, 48, 49], and Adaptive P&O method [50,
51], and Intelligent and Fuzzy Logic methods [52, 53]. These techniques vary between each
other in many aspects, including simplicity, convergence speed, system stability, and MPP
tracking effectiveness. The primary challenges for maximum power point tracking of a solar PV
array include: 1) how to get to a MPP quickly, 2) how to stabilize at a MPP, and 3) how to
35
smoothly transition from one MPP to another for sharply changing weather conditions. In
general, a fast and reliable MPPT is critical for power generation from a solar PV system.
A grid-connected solar PV system consists of three parts (Fig. 3.1): an array of solar
cells, power electronic converters, and an integrated control system [44, 45]. The control system
of a solar PV array contains two parts: one for MPPT and the other for grid interface control [4952]. Both control functions are achieved through power electronic converters. In general, the
dc/dc converter implements the MPPT function while the dc/ac converter performs the gridinterface control.
The power extracted from a PV array Pa is determined by the terminal voltage Va and
output current Ia of the array. The terminal voltage Va depends on the control of the dc/dc
converter while the output current Ia depends on temperature, irradiation level, and the PV array
36
terminal voltage. During a day, solar irradiation and temperature fluctuates over time [54] (Fig.
3.2), causing the MPP of the PV array changes continuously. Consequently, the PV system
90
Temp
1000
Irra
800
80
600
70
400
60
200
50
12
Time (Hour)
16
20
Solar Irradiation(W/m2)
Temperature (F)
0
24
Temperature affects solar cell characteristics primarily in the following two ways:
directly via T in the exponential term in (3.1) and indirectly via its effect on the reverse-diode
saturation current I0 and the photo-generated current IL. The dependence of the reverse-diode
saturation current I0 on temperature for a silicon solar cell is:
I0 K T e
3
q Eg
mk T
(3.1)
where K is the approximate constant with respect to temperature, Eg is the band-gap energy of
the semiconductor (eV), m is the diode ideality constant, k is Boltzmann constant, and T (K) is
the temperature of the pn junction. The photo-generated current IL is also influenced by the
temperature too as the following:
37
I L ( I L,n K I T )
E
En
(3.2)
where IL,n is the photo-generated current at the nominal condition, T is the difference between
actual temperature T and nominal temperature Tn (25C), respectively, E is the irradiation on the
device surface (W/m2), and En is the nominal irradiation (1000 W/m2).
25
20
P(kW)
15
10
5
0
100
75
50
25
0
T ( oC)
100
200
300
Vs (V)
400
500
700
600
0.1
T=0
T=20
T=40
T=60
T=80
T=100
dP/dV (kW/V)
0.08
0.06
0.04
0.02
0
-0.02
-0.04
100
200
300
400
Vs (V)
500
600
700
Then, based on (3.1) and (3.2), characteristics of a PV array versus the PV cell
temperature and the terminal voltage of the array can be obtained as shown by Fig.3.3.
38
According to Fig. 3.3, as the temperature increases, the maximum power production of the PV
array drops greatly and the MPP voltage reduces, indicating that PV arrays perform better on
cold days than hot ones. Figure 3.4 shows the derivative of the output power versus PV array
terminal voltage for several constant temperatures. For each constant temperature, the voltage
corresponding to the zero derivative represents the required voltage for MPP. The derivative has
a stable positive value before reaching the MPP, and then drops sharply around the MPP. As
temperature increases, the zero derivative point shifts to the left.
(3.3)
39
(3.3). Figure 3.6 shows the derivative of PV array power versus terminal voltage of the PV array
for several constant irradiation levels. Compared to Fig. 3.4, as the irradiation level changes, the
P (kW)
25
20
15
10
5
0
700
0
0.25
0.5
600
500
400
300
0.75
200
100
Vs (V)
0.08
s=0.2
s=0.4
s=0.6
s=0.8
s=1
dP/dV (kW/V)
0.06
0.04
0.02
0
-0.02
-0.04
100
200
300
400
Vs (V)
500
600
700
There have been reported many traditional MPPT methods for solar PV generators. In
general, the MPPT control is achieved by varying the terminal voltage applied to the PV
generator. Typical MPPT techniques include fixed-step MPPT: Short-Circuit Current method
40
[43], Open-Circuit Voltage method[44], Perturb and Observe (P&O) methods [45, 46],
Incremental Conductance (IC) methods [47, 48, 49].
The SCC method is based on the observation that IMPP is about linearly proportional to
ISC of a PV array (Fig. 3.7a) , i.e.,
I MPP ( S ) k SC I SC ( S )
(3.4)
where kSC is a constant. According to (3.4), the SCC controller (Fig. 3.8a) generates a control
signal to the dc/dc converter based on the error signal between the actual current of the PV array
Ia and the IMPP calculated from (3.4).
IMPP(A)
60
40
20
0
10
20
30
40
50
60
70
ISC (A)
VMPP (V)
600
500
400
300
200
300
350
400
450
500
VOC (V)
550
600
650
700
This method requires measurements of ISC. Therefore, a static switch in parallel with the PV
array is needed in order to create the short-circuit condition for each solar irradiation level
change, which could cause a large oscillation of PV array output power. Another disadvantage is
that the computation of IMPP is very sensitive to kSC, and the relation between IMPP and ISC is not
100% linear. Thus, a small deviation of IMPP, calculated by (3.4), can easily reduce the output
power of the PV array greatly.
I SC
k SC
*
I MPP
ia
VOC
*
VMPP
kOC
va
The OCV method is based on the observation that VMPP is about linearly proportional to
VOC of a PV array (Fig. 3.7b), i.e.,
V MPP ( S ) k OC VOC ( S )
42
(3.5)
where kOC is a constant. Based on (3.5), a close-loop control scheme is developed to bring the PV
array voltage to VMPP calculated from (3.5) (Fig. 3.8b). Similar to the SCC method, the OCV
method requires measurements of VOC. Thus, a static switch in series with the PV array is
necessary in order to create the open-circuit state for each weather condition change, which
would also cause a large oscillation of PV array output power. In addition, since VOC varies with
temperature and other factors and the relation of VMPP and VOC is affected by shading, actual
VMPP for a practical PV application is difficult to get.
The P&O method is the most commonly used MPPT technique for PV arrays. It operates
by periodically perturbing the array terminal voltage or current and comparing the PV output
power with that of the previous perturbation cycle.
In general, if an increased perturbation of PV array operating voltage causes an increase
of output power, the control system moves the operating point in the same direction; otherwise
the perturbation is changed to the opposite direction. The process continues until the MPP is
reached [55-57]. There are many different P&O methods available in the literature. In the classic
P&O technique [41] (Fig. 3.9), the perturbations of the array operating point have a fixed
magnitude. In the optimized P&O technique [55, 56], an average of several samples of the array
power is used to adjust the perturbation magnitude.
43
Va (k ) & I a (k )
Va (k ) & I a (k )
Va (k 1) & I a (k 1)
Pa (k ) Va (k ) I a ( k )
Pa (k 1) Va (k 1) I a (k 1)
Pa Pa (k ) Pa (k 1)
Va Va (k ) Va (k 1)
Pa 0?
Va 0?
Va 0?
Vref
Vref
Vref
Vref
The IC method is based on the principle that ideally the following equation holds at the
MPP [47, 51]:
I a Va I a Va 0
(3.6)
Also, when the operating point in the P-V plane is to the right of the MPP,
I a Va I a Va 0 ; when the operating point is to the left of the MPP, Ia Va I a Va 0. Thus,
the direction in which the MPP operating point must be perturbed can be determined by
comparing the instant conductance I a Va to the incremental conductance I a Va . Using the IC
44
method (Fig. 3.10), it is theoretically possible to know when the MPP is reached and when the
perturbation should be stopped.
Va (k ) & I a (k )
Va (k ) & I a (k )
Va (k 1) & I a (k 1)
I a I a (k ) I a (k 1)
Va Va (k ) Va (k 1)
Va 0?
Vref
I a Va I a / Va ?
I a 0?
I a Va I a / Va ?
I a 0?
Vref
Vref
Vref
In a PV system, the maximum power point tracking speed and tracking accuracy are the
key factors for a MPPT control algorithm. These factors directly relate to the duty ratio
adjustment of the dc/dc converter. When the PV system operating point is away from the MPP,
the tracking speed must be accelerated, i.e., the regulation of the duty ratio should be large.
When the operating point is around the MPP, the regulation of the duty ratio must be small to
45
In conventional adaptive MPPT methods, the perturbation value changes during the hill
climbing process [50, 55]. Typical adaptive P&O methods use power derivative information to
determine the next perturbation operation. It is based on the observation that the derivative is
positive on the left side of the MPP, zero at the MPP, and negative on the right side of the MPP
(Fig. 3.4 and 3.6). Thus, a scaling factor (SF) perturbation strategy is developed as shown by
46
(3.7), in which M is a constant coefficient and the multiplication of M with the derivative
determines an adaptive adjustment of the duty ratio in the next perturbation cycle [55]. Hence,
the duty ratio adjustment is scalable rather than fixed. Similar to the IC method, the perturbation
stops theoretically when the MPP is reached.
d (k ) d (k 1) M
dPa
dVa
(3.7)
Va
Ia
dPa
dVa
D
0
The proposed adaptive MPPT strategy has adopted the advantage of the inner close-loop
control mechanism for the duty-ratio regulation. But, it introduces a hyperbolic function (3.8)
into the MPPT design.
47
(3.8)
where k is a constant value which is tuned to meet a fast and reliable MPPT requirement for a
typical solar PV array. The output of the sigmoid function (Fig. 3.12) is close to 1 if |dPa /dVa| is
large while around dPa dVa =0, the output reduces greatly. This hyperbolic function enables a
more accurate and stable, and much faster tracking properties under dynamic condition. The
control diagram of the proposed H-PI method is shown by Fig. 3.13, the measured voltage and
current are first processed through a low-pass filter. Then, the power over voltage derivative is
processed by a hyperbolic function and the adjustment amount of the duty-ratio is determined by
a PI controller. Finally, a new duty-ratio computed from this block is applied to the power
y=tanh(4*x)
1
0.5
0
-0.5
-1
-1
-0.5
0
x
0.5
Va
Ia
dPa
dVa
tanh(k
dPa
)
dVa
D
0
48
One issue for the proposed MPPT is the computation associated with the tanh()
function. In general, the tanh() can be calculated very quickly in a digital control system.
According to a large number of experiments performed over a 2GHz PC, the average
computation time of tanh() in MatLab is about 10ns. Compared to the controller sampling time,
the computation time of tanh() is much smaller and ignorable. For tanh() implementation in a
DSP chip, the additional computational effort is even more insignificant.
49
fc
fc
50
Modern MPPT algorithms are usually implemented by using the digital control
technology. The MPPT control module of this paper is developed with detailed consideration of
digital control system natures, including the sample and hold, digital signal processing, and time
delays (Fig. 3.15). The measured voltage and current signals first pass through sample and hold
blocks, which converts measured continuous signals to discrete signals. Then, a digital
filtering mechanism is applied to remove high frequency components caused by noises or rapid
switching of power converters. A time delay block is applied to account for potential delay
between digital and physical systems. The comparison in this section focuses mainly on IC fixed
step, traditional scaling factor (SF) adaptive, and the proposed hyperbolic-PI based (H-PI) MPPT
methods.
The temperature during a day normally does not change sharply, but solar irradiation
levels could vary quickly from one value to another. To test and compare MPPT algorithms
under abrupt changes of solar irradiation levels, a solar irradiation curve with step and ramp
changes is generated (Fig. 3.16). The irradiation increases suddenly from 400W/m2 to 1000W/m2
at 1.5s, stays at 1000W/m2 between 1.5s to 2.2s, and drops to 600W/m2 at 2.2s. At 2.9s, there is a
ramp change of solar irradiation levels. The solar irradiance level increases gradually until it
reaches 900W/m2 at 3.2s, maintains at this value for a period of 0.6s, and drops slowly to
700W/m2 at 4s.
The PV array maximum power, along with the captured power by using IC, SF and H-PI
methods under the step and ramp changes of solar irradiation levels, is presented by Fig. 3.17a.
The sampling rate of the MPPT control module is 0.1ms.
51
Irradiation (k W /m 2)
1
0.8
0.6
0.4
1.5
2.5
3
Time (s)
3.5
4.5
320
14
60
50
300
12
10
40
280
Max
1.5
2.5
IC
3
Time(s)
HPI
S-PI
SF
3.5
(HPI)
260
1.5
4.5
2.5
3
Time(s)
3.5
30
4.5
16.6
16.4
16.2
16
Max
1.5
1.51
1.52
IC
1.53
Time(s)
HPI
S-PI
SF
1.54
1.55
15
14.5
14
1.56
3.14
O u tp u t P o w e r(kW )
detD
0
-2
-4
IC
1.5
3.18
Time(s)
HPI
S-PI
SF
3.2
3.22
3.24
16
-6
3.16
IC
-3
Max
13.5
HPI
S-PI
SF
2.5
3
Time(s)
3.5
4.5
IC
SF
HPI
S-PI
PV curve (S=0.6)
PV curve (S=0.9)
Max Power Points
0
0
50
12
8
100
150
200
250
300
350
Vs (V)
Fig. 3.17. Comparison of MPPT under step and ramp changes of solar irradiation levels
52
400
C u rre n t (A )
Voltage(HPI)
(S-PI)
Voltage (SF)
V o lta g e (V )
340
16
The current and voltage waveforms of the proposed MPPT are shown by Fig. 3.17b.
Figures 3.17c and 3.17d are the zoom-in plots of Fig. 3.17a. Figure 3.17e presents the duty-ratio
adjustment during the MPPT control. Figure 3.17f shows, for the three MPPT methods, the
power vs. voltage locus for a slope change of the solar irradiation level from 0.6kW/m2 to
0.9kW/m2 around 3sec (Fig. 3.16).
For the IC method, it is quite stable under sharp and gradient solar irradiation changes.
The primary issue of the IC method is a continuous perturbation in duty ratio (Fig. 3.17e) even
when the solar irradiance level is stable. The extent of the oscillation depends on the perturbation
step selected. The smaller the perturbation step, the smaller the oscillation. However, if the
perturbation step is too small, the MPPT speed would be affected.
For the SF method, there is a very small oscillation when the irradiation level remains at
a stable level, at which the power over the voltage derivative is close to zero (Fig. 3.17 e). But,
for changing irradiation levels, the output power of the PV array oscillates a lot as demonstrated
by time-domain waveforms (Figs. 3.17c and 3.17d) and the power vs. voltage locus plot (Fig.
3.17f). This results from a sharp change of around the MPP (Fig. 3.4 and 3.6), causing unstable
variation in duty ratio.
The proposed H-PI approach shows the best performance (Figs. 3.17a, 3.17c, 3.17d, and
3.17f). This is due to the fact that the duty ratio adjustment of the H-PI method is tuned based on
the power and voltage derivative that is preprocessed through a hyperbolic function as shown by
(3.8). As it can be seen in Fig. 3.17e, the change in duty ratio has a smoothly continuous value
during an abrupt or ramp change of solar irradiation, and is around zero when the solar
irradiation is stable.
53
DC link V oltage (V )
1,250
1,225
1,200
1,175
1,150
1.5
2.5
3
Time(s)
3.5
4.5
3.58
3.6
1,000
500
0
-500
-1,000
3.5
3.52
3.54
3.56
Time(s)
Grid P ower (k W )
50
25
0
-25
-50
1.5
2.5
3
Time(s)
3.5
4.5
The PV voltage and current oscillate continuously (Fig. 3.17b), particularly under
changing solar irradiation conditions. This causes more oscillation of the instantaneous power of
the PV array. This issue is critical and must be considered in the design of the low-pass filters
54
(Fig. 3.15) to assure fast and robust MPP tracking, particularly for the adaptive MPPT techniques
(Figs. 3.11 and 3.13). The power vs. voltage locus as shown by Fig. 3.17f illustrates more clearly
how the maximum power is tracked by using three different MPPTs approaches. As it can be
seen from the figure, the proposed adaptive MPPT is more reliable and efficient in tracking the
MPP than the conventional adaptive MPPT.
The dc-link voltage is very stable (Fig. 3.18) under the direct-current vector control
strategy applied to the dc/ac inverter, which is an important factor for the MPPT. The threephase current waveform on the grid side is shown by Fig. 3.19 and the instantaneous grid power
is shown by Fig. 3.20. As shown by Fig. 3.20, the grid power follows the captured PV power.
However, due to the existence of harmonics and unbalance in the grid three-phase currents, there
are oscillations in the grid power, which is similar to the instantaneous grid power in other
renewable energy applications [65, 66].
When designing a digital control system, sampling rate is usually predefined. After that,
the perturbation rate for each MPPT technique should be designed independently until an
acceptable performance is obtained. Figure 3.21 shows the maximum power tracking by using
the three different MPPT techniques under the sampling rate of 1ms per sample and 10ms per
sample, respectively.
As shown by Fig. 3.21a, all the MPPT methods can track MPP when the sample time is
1ms. However, when the sample time is 10ms, there will be a big notch in the captured power by
IC and SF method (Fig. 3.21b). An examination of power vs. voltage locus (Fig. 3.21c) reveals
more detailed information about the MPP tracking using the three different MPPT approaches.
55
The figure, consistent with Fig. 3.17f, demonstrates that the proposed adaptive MPPT is more
reliable. Overall, the proposed method responds much faster and is more stable under different
irradiation conditions.
16
14
12
10
8
Max
1.5
IC
2.5
SF
3
Time(s)
H-PI
3.5
4.5
8
1.5
2.5
IC
3
Time(s)
SF
3.5
H-PI
4
4.5
O u tp u t P o w e r(k W )
16
IC
SF
HPI
S-PI
PV curve (S=0.6)
PV curve (S=0.9)
Max Power Points
0
0
50
12
8
100
150
200
250
300
350
400
Vs (V)
c) Power vs. voltage locus at the increasing slope change (10ms)
Fig. 3.21. MPPT comparison under different sampling rates
56
It is needed to point out that the sampling rate determines the waiting time for the next
perturbation. From this point of view, the sampling rate concept is not exactly equivalent to the
cutoff frequency notion normally used in the digital signal processing field. In a MPPT algorithm
for a PV array, the low-pass filters shown in Fig. 3.15 help to remove noises while the sampling
rate determines how fast to conduct the next perturbation. The impact of the sampling rate can be
seen from Fig. 3.21. In general, as the sampling time increases, it is slower to track the maximum
power.
In reality, solar irradiation level changes constantly over time [67]. Therefore, it is
important to evaluate and compare MPPT performance under variable irradiance conditions. For
20
Output P ower (kW )
Irra d ia tio n (k W /m 2 )
0.8
0.6
0.4
0.2
0
0.5
1.5
Time (s)
2.5
Max
IC
SF
HPI
S-PI
15
10
5
0
0.5
1.5
Time(s)
2.5
Figure 3.22b compares the MPP tracking using different MPPT algorithms and the
parameters of the MPPT algorithms are the same as those used in Fig. 3.17. As shown by the
57
figure, among the three algorithms, the proposed H-PI method is the most effective to track the
MPP. For the IC method, the fixed step perturbation disables the fast changing requirement in
duty ratio to track the MPP. For the SF method, a stable adaptive adjustment based on the
derivative information is hard to obtain in tracking the MPP under fast changing weather
conditions.
In this chapter, not only the software simulation but also the hardware experiment has
been applied for the comparison of different MPPT schemes.
A hardware laboratory test system of Fig. 3.14 is built for further investigation of the
conventional and proposed MPPT algorithms. Figure 3.23 shows the testing system with the
following setups. 1) An Agilent E4360A solar simulator is used to represent an actual PV array
[68]. The solar simulator can generate real output voltage and current with relation that is
equivalent to a practical PV panel or array. By using the solar simulator, it is possible to repeat
the same solar irradiation condition to test and compare different MPPT algorithms through this
hardware experiment which is otherwise impossible. Another advantage is that the maximum
output power of the simulated PV array can be calculated based on the experiment settings so
that one can determine whether a MPPT algorithm is effective or not in a hardware experiment.
Due to these reasons, solar simulators have been widely used by many researchers around the
world for evaluation of a PV control system [69]. 2) The dc/dc converter is built by using a
58
LabVolt MOSFET power converter. 3) The capacitor connected to the output terminal of the
simulator is formed by several LabVolt capacitors in parallel. 4) A smoothing inductor is
employed for the dc/dc converter. 5) The solar simulator is controlled by a dSPACE digital
control system [70].
The control system collects output voltage and current signals of the solar simulator, and
sends a control signal to the converter based on control demands generated by different MPPT
algorithms. Although the dSPACE system is not a digital device used for practical applications,
it is a digital control system based on modern DSP chips [71]. Using the dSPACE system, a
MPPT digital controller can be quickly built and tested before converting it to a practical digital
control device.
The rated values of the hardware experiment system (Fig.3.23), including the power
converter and the PV simulator, are different from those used in the computational experiment
(Fig.3.14). In general, the rating of the hardware experiment system is lower than the rating of a
practical PV array. Therefore, parameters of the MPPT controllers must be retuned. To ensure
that the controllers work properly, the retuned MPPT algorithms for both the conventional and
proposed techniques are evaluated in simulation first before the hardware experiment, where the
simulation time step for the controllers is the same as the sampling time used in the dSPACE
digital control system. Another big challenge, that is different from the simulation, is that noises
are more significant than expected. One strategy to reduce the noises is to increase the strength
of the measured signals.
Because of the noises, it is very hard to tune MPPT parameters for IC and SF algorithms,
especially for the SF algorithm. This is due to the fact that a noise can result in a high notch in
the calculated power during the next sampling time, causing a large variation in power derivative
and thus affecting the stability of the SF algorithm. However, for the proposed H-PI algorithm, a
stable MPPT algorithm is much easier to obtain. The test sequence is scheduled as the following
with t=0s as the starting point for data recording. Around t=20s, there is an increase of the solar
irradiation. A small increase of the solar irradiation appears near t=40s. Close to t=60s, there is a
large decrease of the irradiation. At about t=80s, the sequence repeats itself. The PV simulator
voltage and current are not only collected by the dSPACE system but also monitored by
oscilloscopes and/or meters. Figure 3.24 shows the captured maximum power by all the three
algorithms. Again, the proposed H-PI approach has the best performance because for the
proposed H-PI approach, the power derivative is smoothly processed before it is applied to the PI
controller. In addition, PI controller can responds much faster than an open-loop scheme.
60
10
12.2
IC
SF
S-PI
12
8
6
0
20
40
60
80
100
Time
12
11.8
11.6
11.4
11.2
IC
40
45
50
SF
S-PI
55
Time
3.6 Conclusions
This chapter proposes a fast and robust MPPT technique and compares it with typical
conventional MPPT algorithms used in solar PV industry (Table 3.1).
Among the three most popular conventional MPPT algorithms (fixed step P&O, IC and
adaptive P&O), the fixed step P&O and IC methods have continuous oscillation even when the
solar irradiance level is constant in the power converter switching environment; the adaptive
P&O method has very small oscillation if the solar irradiance level is stable. For the proposed
MPPT approach, it has the least oscillation and the highest stability.
The sampling rate affects the design of the perturbation rate. This result indicates that a
match between the sampling rate and the perturbation step is important. If the sampling rate is
too slow, a stable and reliable MPPT would be hard to achieve. Again, the proposed method is
more stable and reliable under different sampling rate conditions.
Under the variable irradiance levels, the proposed H-PI approach has better performance
than conventional methods, indicating that the power derivative information is valuable in
61
tracking and capturing maximum PV array power under changing weather conditions. The
comparison between the traditional and proposed adaptive methods shows that the hyperbolic
processing of the derivation is important for high performance of a solar PV system.
Sensed
around MPP
Parameters
Low
No
Current
No
Low
No
Voltage
P&O
Yes
Low
Yes
V/C
IC
Yes
Medium
Yes
V/C
SF
Yes
Low
No
V/C
RCC
Yes
Low
No
V/C
Fuzzy/Neural
Yes
High
No
Varies
Proposed H-PI
Yes
Low
No
V/C
MPPT
True
Technique
MPPT
SCC
No
OCV
Complexity
In the hardware experiment, the unexpected noises would drastically influence the power
increment or power derivative calculation in the next perturbation step. Because of the noises, it
is very hard to tune the MPPT parameters for IC and SF algorithms, especially for the SF
algorithm. However, for the proposed H-PI approach, the power derivative is smoothly
processed before it is applied to the PI controller; in addition, PI controller can respond much
faster than an open-loop scheme. The comparison demonstrates that the proposed H-PI approach
is much easier to tune and has the best performance.
62
CHAPTER 4
PV ENERGY EXTRACTION CHARACTERISTICS STUDY UNDER SHADING
CONDITIONS FOR DIFFERENT CONVERTER CONFIGURATIONS
A solar PV energy conversion system requires power converters for maximum power
extraction and grid integration. At present, many different converter structures have been
developed and used in a solar PV system. For all the different converter structures, the energy
extraction characteristics and maximum power capture capability for all the converter schemes
under even solar irradiation are very similar. However, under shading conditions, the energy
extraction depends strongly on what converter structure is used in a PV system.
A grid-connected solar PV system consists of three parts: an array of solar panels, power
electronic converters, and an integrated control system [13, 72]. Normally, solar cells are
connected in series to form a module that gives a standard dc voltage. For an application,
modules are connected into an array to produce sufficient current and voltage to meet a demand
[23]. There are generally two ways to connect PV modules into an array. The first approach
connects modules in series into strings and then in parallel into an array. The second approach,
first wires modules together in parallel and then those units are combined in series. Ideally, both
series-parallel and parallel-series connections are equivalent if all the cells and modules are
63
identical and work at the same condition. But, if sunlight is applied unevenly to different PV
cells, the second connection approach could cause many very bothersome problems [23].
The control system of a solar PV system contains two parts: one for maximum power
point tracking (MPPT) and the other for grid interface control [38, 39]. Both control functions
are achieved through power electronic converters. On the dc side, MPPT optimizes the power
output by varying the closed loop system voltage. On the ac side, these inverters ensure that the
sinusoidal output is synchronized to the grid frequency (60Hz). Illustration 4.1 shows a
residential grid connected solar system.
This PV array structure consists of a dc/dc converter for maximum power extraction of a
whole PV array, a dc/ac converter for interface of the PV array to the grid, and a dc-link between
the two converters (Fig. 4.1). The MPPT control is achieved through the regulation of the dc
voltage applied to the array and the grid interface control stabilizes the dc-link voltage and
adjusts the reactive power sent to the grid [38, 39].Hence, the control of the two converters is
decoupled. It is a cost-effective approach to develop large MW-scale solar PV systems. The
primary disadvantage of the central converter configuration is that a large power loss may occur
in the energy harvest when mismatch or varying shading happens among the PV panels within
the array. Normally, a bypass diode is applied to a PV module [23]. At the top of each string, a
blocking diode is used to prevent a shading or malfunctioning string from withdrawing current
from the rest strings that are wired together in parallel.
Fig. 4.1. PV array with central dc/ac and dc/dc converter structure
65
This PV array structure consists of a central dc/ac converter for interface of the PV array
to the grid and multiple dc/dc converters with each dc/dc converter responsible for MPPT of one
PV string [14, 17] (Fig. 4.2) while the dc/ac inverter stabilizes the dc-link voltage and adjusts the
reactive power sent to the grid by the whole array. The key difference between Figs. 4.1 and 4.2
is that each dc/dc converter performs MPPT of a string rather than a whole PV array. The string
converter structure is likely to provide better power harvest than a central converter structure
because each string is tracked independently. Multiple self-contained string converters may offer
some advantage by eliminating the single point of failure, but will have a higher installation cost.
Fig. 4.2. PV array with central dc/ac inverter and string dc/dc converters
Dc/dc optimizers place a dc/dc converter to each PV module and provide one, large
66
central inverter that aggregates the power from all optimizers [16, 17] (Fig. 4.3). The optimizers
provide MPPT at the module level while the dc/ac inverter accomplishes the same control
functions as shown in Sections 4.2.1 and 4.2.2. The dc outputs of each module are joined in
series and parallel to be collected and then converted to ac power by a central inverter. With
additional equipment to purchase and install, dc/dc optimizers add to the initial cost of a PV
system. The added module-level hardware also imposes a penalty on overall system-level
efficiency by introducing an additional stage of loss power conversion. Furthermore, some dc/dc
optimizer systems also require a separate command-and-control device to operate.
the dc voltage applied to the module and 2) control of the reactive power sent to the grid by each
module. Similar to dc/dc optimizers, detached microinverters offer enhanced energy harvest by
performing MPPT at the module level, therefore, a single failure of panel or inverter will not take
the entire string offline. Also, they produce grid-matching power directly at the back of the
panel, and different ratings of solar panels can be added to an array even if they don't match the
original types. In addition, a micro-inverter on each module as the inverters anti-islanding
capability will shut down the module energy production when the system is disconnected, which
can significantly limit the risk to firefighters under emergency conditions according to proposals
of 2014 NEC [74].
However, the MPPT and reactive power control functions of a module are not decoupled,
which is a disadvantage for this PV system architecture because the two control requirements
may not be met at the same time as reported in [14]. Also, microinverters need to be installed
individually on the racking near each module. Todays microinverters have a five- to fifteen-year
warranty, much shorter than the life of PV modules. As such, it would be necessary to replace all
of the nonintegral microinverters at least once during the lifetime of the system.
68
For both central and string inverter configurations, there is no dc/dc converter in the PV
system [15-17]. Thus, the inverters need to handle both MPPT and grid interface control
functions at the string or array level. Similar to Section 4.2.4, a disadvantage for these PV system
architectures is that the MPPT and reactive power control requirements may not be met at the
same time. In the central inverter PV structure, a single and large inverter is connected to many
PV modules wired in series to form strings with up to 600V of open-circuit voltage (1,000V in
Europe) [16] (Fig. 4.5a). In the string inverter PV structure, each string of series PV modules is
connected to one string inverter. Then, inverters for all the strings are connected in parallel
before feeding into the grid (Fig. 4.5b).
a) Central inverter
In summary, from the maximum power extraction standpoint, there are basically three
69
different configurations: 1) one dc voltage applied to the whole PV array for MPPT control
(Figs. 4.1 and 4.5a), 2) multi-independent dc voltages applied to multiple strings of series PV
modules (Figs. 4.2 and 4.5b), and 3) one independent dc voltage for each PV module (Figs. 4.3
and 4.4). For detailed evaluation of the three different configurations, it is important to be able to
assess power and voltage characteristics at individual cell, module, and array levels. However,
such a study is difficult to be conducted experimentally because the existing commercial PV
modules are not built in such a way that the current or voltage of each individual cell can be
measured. To overcome the challenge, a numerical algorithm developed in Chapter 2 for
simulation study of PV cell, module, and array characteristics under different converter schemes.
It is also important to indicate that a large PV array usually consists of thousands of cells. Thus, a
rigorous computational approach is very important.
Detailed mathematical and simulation model of a PV cell, module and array have been
described Section 2.2. The PV Array model with consideration of different converter
configurations will especially be analyzed and validated.
For the micro inverter structure, the energy extraction characteristics of each module is
computed separately based on the PV module model (Sections 2.2.2 and 2.3) while the total
energy extraction of a PV array is the summation of energy from all the modules. For the string
converter structure, the voltage applied to each string is different (Sections 4.2.2 and 4.2.5). For
each string, the mathematical model and procedure to obtain P-V and I-V characteristics is very
similar to Sections 2.2.2 and 2.3 except that the external voltage applied to each string equals to
the sum of photogenerated voltages of all series connected PV modules. The output power of the
70
array is the sum of the output powers of all strings. For the central converter structure with the
standard series-parallel connected PV array (Section 4.2.1), the voltage applied to each string of
the PV modules is the same. However, the P-V and I-V characteristics of each string could be
different depending on how many PV cells in a string are shaded and how much. The model and
procedure for energy extraction computation of each string is similar to that of the string
converter structure while the total output current of the PV array is the sum of currents of all
parallel strings.
Based on the analysis shown in Section 4.2, PV system characteristic studies are
conducted for the following three converter configurations, i.e., a power converter for an entire
array, for each string, and for each panel.
For a typical series-parallel connected PV array with central converter configuration, the
voltage applied to each string of the PV modules is the same. However, the P-V and I-V
characteristics of each string could be different depending on how many PV cells in a string are
shaded and how much. The mathematical procedure to obtain P-V and I-V characteristics is
based on Sections 2.2. The output current of the PV array is the sum of currents of all parallel
strings.
71
Power (kW)
20
None
50%
100%
15
10
5
0
100
200
300
400
500
Vs (V)
a)
10
Voltage (V)
0
-10
-20
None
50%
100%
-30
-40
100
200
300
400
500
Vs (V)
Power (W)
0
-100
None
50%
100%
-200
-300
100
200
300
400
500
Vs (V)
c)
Figure 4.6 shows the PV array characteristics. The PV array has a configuration of 10
parallel strings with each string containing 20 series modules. Assume there are 19 shaded
modules in the 1st string, 17 in the 2nd string, 15 in the 3rd string and 1 in the last string. In
72
each shaded module, there is only one shaded cell, which represents the worst condition that may
damage a PV cell [75, 76]. The shading factor is 0, 50% and 100%, respectively. From the
figures and other case studies, it is found that the PV array characteristics can be shifted
significantly under shading conditions (Fig. 4.6a). The degree of the change depends on how
many strings contain shaded cells and how many shaded cells are in each string. When there is
only one shaded cell in a string, all the photogenerated voltages of the unshaded cells in that
string are applied to the shaded cell (Fig. 4.6b), which would cause a high risk to damage the
shaded cell due to the high absorbing power of the shaded cell (Fig. 4.6c).
For a PV array with the string converter configuration, the voltage applied to each string
of PV modules is different. Therefore, the P-V and I-V characteristics of each string are
independent from those of other strings. Under the MPPT control strategy, each string captures a
different power. The total output power of the array is the summation of the maximum powers
captured by all strings.
shade factor=100%
shade factor=50%
2
Average max
strings 1-10
1.5
P o w e r (k W )
P o w e r (k W )
1
0.5
0
100
200
300
400
500
Vs (V)
Average max
strings 1-10
1.5
1
0.5
0
100
200
300
Vs (V)
73
400
500
Figure 4.7 shows the P-V characteristics of all strings under the same shading conditions
used in Fig. 4.6. In the figure, the constant line represents the maximum power of the PV array
divided by the number of all strings. From the figure and other analysis, the following properties
are found.
1) When there is no shaded cell in the PV array, all the strings have the same P-V
characteristics. The output power of the PV array under the MPPT control strategy is the
maximum power of one string multiplied by the number of all strings.
2) When there are shaded cells in the PV array, the P-V characteristics of each string
would be different from other strings depending on the distribution of the shaded cells and the
shading factors of shaded cells. The average maximum power lies between the maximum
captured powers of the least and the most shaded strings.
3) Similar to Fig. 4.6, the most damaging condition for a PV cell is when there is only
one shaded cell in a long string, which would cause a high risk to damage the shaded cell due to
the high absorbing power of the shaded cell.
For the micro-inverter configuration, each PV module is connected to the grid via its own
inverter. Hence, the P-V and I-V characteristics of each module are independent from those of
other modules. Under the MPPT control strategy, each module captures a different power. The
total output power of the PV array is the summation of the maximum powers captured by all the
PV modules.
74
P o w e r (k W )
0.1
None
50%
100%
0.075
0.05
0.025
0
10
15
20
25
Vs (V)
0
-25
None
50%
100%
-50
-75
-100
10
15
20
25
Vs (V)
V oltage (V )
5
0
-5
None
50%
100%
-10
-15
-20
10
15
20
25
Vs (V)
Figure 4.8 shows the P-V characteristics of a PV module with one shaded cell for shading
factor of 0, 50%, and 100%, respectively. As it can be seen from the figure, shading of even one
single cell could cause significant drop of PV module output power (Fig. 4.8a). Under the
shading condition, a high reverse terminal voltage would appear on the shaded cell (Fig. 4.8b),
75
which results in a high absorbing power by the shaded cell especially when the external voltage
applied to the module is low (Fig. 4.8c). Similarly, the most demanding situation is when there is
only one shaded cell in the PV module.
Table 4.1 compares maximum power that can be captured by the PV array for the three
power converter configurations under the same shading condition used in Fig. 4.6. In general, if
there is no shaded cell, the captured maximum power is the same for all the three converter
configurations. However, under shading conditions, the micro-converter configuration based PV
system has the highest energy yield while the central converter configuration one has the lowest
energy production.
100%
Central converter
13.65 kW
7.34 kW
String converter
13.79 kW
7.76 kW
Micro converter
15.71 kW
11.39 kW
Configuration
In the PV industry, bypass diodes are normally adopted in central and string converter
configurations [38]. For the micro converter configuration, it is reasonable to assume that bypass
diodes are not used.
76
the number of the unshaded cells, the P-V characteristics of the PV array is very close to the
unshaded condition no matter how the shaded cells are distributed in the PV array. In addition, it
is more convenient to manage the MPPT control of the PV array even under shading conditions
when there is a bypass diode for each PV cell.
3) As fewer cells are in parallel with a bypass diode, the peak power of the PV array
increases and the multi-peak impact to the P-V characteristics of the PV array reduces (Fig.
4.10).
20
None
50%
100%
15
10
P o w e r (k W )
P o w e r (k W )
20
5
0
100
200
300
400
10
5
0
500
None
50%
100%
15
100
200
Vs (V)
a) PV array characteristics
P o w e r (W )
P ow er (W )
-50
-100
-150
100
200
300
400
500
None
50%
100%
2
0
-2
-4
500
None
50%
100%
400
b) PV array characteristics
300
Vs (V)
100
200
300
400
Vs (V)
Vs (V)
78
500
shade factor=50%
20
full-sun
n=1
n=2
n=3
4
n=6
n=9
n=12
n=18
n=36
P ow er (k W )
15
10
100
200
300
400
500
Vs (V)
full-sun
n=1
n=2
n=3
4
n=6
n=9
n=12
n=18
n=36
P ow er (k W )
15
10
50
100
150
200
250
Vs (V)
300
350
400
450
500
With the same bypass diode layouts, the only difference between the string and central
converter configurations is that the voltage applied to each string of series PV modules is
different. Due to the bypass diode impact, the characteristics of PV strings would be different
from Fig. 4.7. Figure 4.11 shows the P-V characteristics of all strings under the same shading
79
conditions used in Fig. 4.6 for different bypass diode schemes. Again, the constant red dot line
represents the maximum power of the PV array divided by the number of all strings.
If each module has one bypass diode, the characteristics of each string has multiple
peaks, requiring that the MPPT algorithm must be designed to be able to locate the maximum
power production of each string. If there are more bypass diodes within a module, the multiple
peaks moves from low voltage range to high voltage range in general, resulting in more power
production from each string under the same shading conditions. If one bypass diode is applied to
every one or two PV cells, the energy yield of each string is significantly increased and the peak
power of each string is closer to ideal peak power under the nonshading condition, making it
easier to develop and design MPPT algorithm.
4.5.3 Comparison of Maximum Power Using Central, String and Micro Converter Configuration
One important issue for the evaluation and comparison of the power production using
micro, string, and central converter configurations is how much power can be extracted for
different converter structures. Tables 4.2 and 4.3 compares the maximum power of the PV array
that can be captured by using different converter configurations for varying bypass diode
arrangements under the same shading condition used in Fig. 4.6. The comparison shows:
1)
When the number of PV cells of each bypass diode is small, the difference among
the maximum power extracted by using the three different converter structures is very small.
2)
The micro-converter based PV system has the highest power capture capability
while the power production of central converter based PV system is the lowest. The difference
between the highest and the lowest power is more obvious as the number of PV cells contained
in each bypass diode increases.
80
shade factor=100%
Power (kW)
Average max
strings 1-10
1.5
1
0.5
0
100
200
300
400
500
Vs (V)
Average max
strings 1-10
Power (kW)
1.5
1
0.5
0
100
200
300
400
500
Vs (V)
Average max
strings 1-10
Power (kW)
1.5
1
0.5
0
100
200
300
400
500
Vs (V)
Average max
strings 1-10
1.5
1
0.5
0
100
200
300
400
500
Vs (V)
Average max
strings 1-10
1.5
1
0.5
0
100
200
300
400
500
Vs (V)
Table 4.2. Comparison of maximum power extraction under 50% shading factor (kW)
Configuration
Central converter
String converter
Micro converter
19.0610
19.1124
19.1127
18.7270
18.8392
18.8395
18.3700
18.5661
18.5664
17.9912
18.2929
18.2932
17.1742
17.7466
17.7470
15.8272
16.9274
16.9277
12
14.7374
16.1083
16.1087
18
13.9460
14.7711
15.7179
Table 4.3. Comparison of maximum power extraction under 100% shading factor (kW)
Configuration
Central converter String converter
Micro converter
19.0475
19.1012
19.1014
18.7120
18.8277
18.8280
18.3535
18.5543
18.5546
17.9730
18.2808
18.2811
17.1517
17.7340
17.7343
15.7945
16.9137
16.9140
12
14.3236
16.0934
16.0938
18
11.3022
14.4531
14.4535
82
3)
based PV systems. Thus, when the number of PV cells of each bypass diode is small, such as 1 to
4 as shown by Tables 4.2 and 4.3, the central converter based PV system is much more efficient
than micro-converter based PV system. In addition, the reactive power compensation ability of
central-converter based PV system is higher [14, 77].
4.6 Conclusion
This chapter compares the energy extraction characteristics of a solar PV system for
different converter schemes, including central, string and micro converter configurations. The
chapter particularly focuses on how energy extraction characteristics of a PV array are affected
by uneven shadings using different converter structures. Without shading, the PV system has the
same energy yield for all the three converter schemes. Under shading and without bypass diodes,
the micro converter based PV system has the highest energy yield while the central converter
based PV system has the lowest energy production. However, the micro converter based PV
system is costly and is still unable to prevent energy loss caused by the shading within PV
panels.
It is found that with bypass diodes, the efficiency of a PV system can be improved
significantly especially when the number of PV cells within a bypass diode is small. Simulation
studies show that the difference among the maximum extracted power is trivial under shading
conditions for the three different converter structures for PV modules with large number of
bypass diodes. The study shows that the central converter based PV system with properly built-in
bypass diodes is an effective and economic approach to improve efficiency, performance, and
reliability of a PV system.
83
CHAPTER 5
COORDINATED CONTROL FOR GRID INTEGRATION OF PV ARRAY, BATTERY
STORAGE, AND SUPERCAPACITOR WITH RELATED ISSUES
The intermittent nature of PV energy and quick fluctuations of load demanding require
energy storage units (ESU) which generally consists of storage battery and supercapacitor (SC)
[78, 79]. Batteries are the technological solution most commonly employed to help make a PV
system [80-84], whose power output cannot be controlled, smooth and dispatchable. A battery
stores electrical energy in the form of chemical energy. For a PV-battery system to function
effectively the electrochemical processes must work in both directionsin other words, the
system must be rechargeable. Normally, batteries perform three main functions in a gridconnected PV system: storing energy into the batteries when the PV production is high and the
grid demand is low, releasing energy to the grid when the PV production is low or during grid
peak demand intervals, and preventing large voltage fluctuations.
Except for batteries, SC is usually used in conjunction with batteries to form an advanced
PV energy storage system [85-87]. However, unlike batteries, where the voltage remains
relatively even over most of the batterys remaining charge levels, a capacitors voltage scales
linearly with the remaining energy. This means an additional circuitry is required to make the SC
energy usable.
For the PV array, a boost converter is usually connected to the dc link to raise the input
voltage and then transfer the high voltage dc energy to ac energy through a grid-connected
converter [88]. Bidirectional dcdc converters are required to interface the ESU, including
84
batteries and SC, to the dc link for controlling the power flow. The coordinated control of the
entire system is critical. Currents and voltages of PV array, SCs, battery, dc link, and grid are
input signals used to modulate PWM signals to power electronic converters. The modulation
algorithm has to be specified for each converter control topology and has to generate control
signals to fulfill basic functions such as dc link voltage control, ESU control, grid active and
reactive power control, and MPPT control, which has resulted in many different control
strategies for the integrated PV and ESU system [87-94].
A typical grid-connected solar PV and energy storage system consists of five parts (Fig.
5.1): an array of solar cells, battery storage, SC, power converters, and an integrated control
system.
ig
v pv
vdc
i pv
v g ig
vb
ib
vsc
isc
Normally, solar cells are connected in series to form a module that gives a standard dc
voltage. Modules are connected into an array to produce sufficient current and voltage to meet a
demand for a gird-connected application [90, 92, 95]. Usually, the PV modules are first
connected in series into strings and then in parallel into an array. The power produced by a PV
array is dependent on the irradiance and temperature. In general, there is a unique point on the PV curve of the PV array, called the maximum power point (MPP), at which the entire PV system
operates with maximum efficiency and produces its maximum output power. To maximize the
output power of a PV system, continuously tracking the MPP of the system is necessary (Chapter
3). This is accomplished through the control of the dc/dc converter connecting the PV array to
the dc-link capacitor.
A rechargeable battery is one or more electrochemical cells that convert stored chemical
energy into electrical energy during a discharge process or convert electrical energy into
chemical energy during a charge process [95]. The electrical energy storage system in a PV
system is expected to be designed with adequate energy capacity and output peak power to
satisfy grid integration needs [88, 89]. In an integrated PV and battery storage system, the battery
is connected to the dc-link also through a dc/dc converter which controls the charge and
discharge of the battery.
5.1.3 Supercapacitor
86
Unlike batteries, SCs store their energy in an electrostatic field. The most significant
advantage of SCs over batteries is that they are capable of very fast charges and discharges [81,
82, 87]. Disadvantages are that their power is available only for a very short duration and their
self-discharge rate is much higher than that of batteries. Thus, SCs can supply power when there
are surges or energy bursts, while batteries can supply the bulk energy over a longer time period
[81, 82, 87].The combination of the two is crucial for diverse energy storage needs of both fast
and slow fluctuating solar power [81, 87, 89].
The control system of an integrated PV system has three levels: device level at PV array,
battery and SC, PV system central control level (PVCC), and distribution management system
(DMS) level [79, 93, 97]. At the device level, the energy is either captured from the PV array or
generated from ESUs. At the PVCC, the power production is determined based on the optimal
overall profits to operate the PV, ESUs and GCC. The central controller sends out power
references to PV, ESUs and GCC based on commends from DMS, while each individual device
control system ensures that the power reference from the PVCC is reached. At the DMS level,
87
the power production of the integrated PV system is managed to meet the overall grid demands
and stability and reliability needs.
Unlike the PV array which requires only one-directional power flow, ESUs require bidirectional power flow. Due to its simplicity and robustness, the bi-directional buck/boost
converter is used in this dissertation to interface the SC or battery ESUs with the dc voltage bus.
In general, the dc/dc converter acts as a boost converter during ESU discharge mode and as a
buck converter during ESU charge mode.
Both constant-current and constant-voltage control schemes are developed. The two
control techniques are then integrated together to meet ESUs charge and discharge requirements.
88
Figure 5.2 shows the block diagram for battery control. Similar current control strategy is also
used to control the SC.
vb
vb _ ref
ib
ib _ ref
Fig. 5.2. Block diagram of nested-loop battery control strategy
For battery control, the switch block passes through the top input or the bottom input
based on the value of SOC in the middle. If the switch is in the upper position, the system
operates in the constant-voltage control mode. If the switch is in the bottom position, the system
operates in the constant-current control mode. The reference current can be either positive or
negative. Assume that the current flowing into the dc-link is positive. Then, if the reference
current is positive, the system implements constant-current discharging control function; if the
reference current is negative, the system implements constant-current charging control function.
SOC is the main factor to decide the battery status from current control to voltage control. When
SOC is below 70%, constant-current charging mechanism is applied; when SOC is above 70%,
the system switches to constant-voltage charging mechanism. In order for a smooth transition
from current to voltage control, the same value of reference current before transition should be
the initial value of the voltage controller. While the battery is in discharging mode, the constantcurrent control mechanism is applied and the power provided by the battery to the grid can be
regulated by adjusting the reference discharging current.
89
Figure 5.3 shows the schematic of the GCC, in which a dc-link capacitor is on the left,
and a three-phase voltage source, representing the voltage at the Point of Common Coupling
(PCC) of the ac system, is on the right. The following dynamic equations can be obtained:
va
ia
ia va1
d
vb R ib L dt ib vb1
vc
ic
ic vc1
(5.1)
where L and R are the inductance and resistance of the grid filter and vabc , iabc , and vabc1 are
instantaneous space vectors of the PCC voltage, line current, and converter output voltage,
respectively.
By using the abc-dq transformation with its d-axis aligned to the voltage vector of the
PCC, full equation in Eq. (5.1) be described in the synchronously rotating reference frame as
follows:
vd
id
iq vd1
d id
L
v
i
i s v
dt
q
q
q
id q1
90
(5.2)
where s is the angular frequency of the grid's PCC voltage, and. Using space vectors, (5.2) is
expressed by the complex equation in (5.3), in which vdq, idq and vdq1 are instantaneous space
vectors of the PCC voltage, line current, and converter output voltage, respectively. In the
steady-state condition, (5.3) becomes (5.4), where Vdq, Idq and Vdq1 stand for the steady-state
space vectors of PCC voltage, grid current, and converter output voltage, respectively.
vdq R idq L
d
idq js L idq vdq1
dt
Vdq R I dq js L I dq Vdq1
(5.3)
(5.4)
In the grid's PCC voltage-oriented frame [98-103], the instantaneous active and reactive
powers absorbed by the GCC from the grid are proportional to the grid's d- and q-axis currents,
respectively, as shown by (5.5) and (5.6):
p(t ) vd id vq iq vd id
(5.5)
q(t ) vq id vd iq vd iq
(5.6)
In terms of the steady-state condition, Vdq=Vd+j0 if the d-axis of the reference frame is
aligned along the PCC voltage position. Assuming that Vdq1=Vd1+j Vq1 and neglecting the grid
filter resistance, the current flowing between the PCC and the GCC according to (5.4) is:
(5.7)
91
(5.8)
The basic principle of the direct-current vector control is to use d- and q-axis currents
directly for active and reactive power or dc-link voltage and reactive power control of the GCC
[96]. Unlike the conventional standard vector control approach that generates a d- or q-axis
voltage from a GCC current-loop controller, the direct-current vector control structure outputs a
current signal at the d- or q-axis current-loop controller (Fig. 5.4). In other words, the output of
the controller is a d- or q-axis tuning current, while the input error signal tells the controller how
much the tuning current should be adjusted during the dynamic control process. The
development of the tuning current control strategy has adopted intelligent control concepts [104],
e.g., a control goal to minimize the absolute or root-mean-square (RMS) error between the
desired and actual d- and q-axis currents through an adaptive tuning strategy.
_
_
PI
_
_
PI
PI
PI
+_
+
PI
_
_
PI
Due to the nature of a voltage-source converter, the d-and q-axis tuning current signals,
id and iq, generated by the current-loop controllers must be transferred to d- and q-axis voltage
signals v*d1 and v*q1 to control the GCC. This is realized through (5.11), which is equivalent to
92
the transient d-q equation (5.2), after being processed by a low pass filter in order to reduce the
high oscillation of d and q reference voltages applied directly to the converter.
(5.11)
The initial values of the GCC PI current-loop controllers are tuned by minimizing the
RMS error between the reference and measured values.
In grid integration of the integrated PV system, the PVCC receives active and reactive
power commends from the DMS and then determines how to control PV array, ESUs and GCC.
Normally, the PV array is controlled for maximum power extraction. Thus, depending on the
active power or dc-link voltage control for the GCC (Fig. 5.4), three coordinated control
strategies are implemented.
In this control scheme, the GCC operates in PQ control mode by maintaining the PCC
active and reactive power output according to the grid control commend. Hence, the dc-link
voltage must be controlled via ESUs [79, 89]. In general, if the PV generated power is more than
the GCC output power, the dc-link voltage goes up so that the ESUs must operate in charging
mode to reduce the voltage; if the PV power is less than the GCC output power, the dc-link
voltage drops so that the ESUs must operate in discharging mode to increase the dc-link voltage.
93
Usually, a battery bank has a high energy density whereas it has relatively slow response
speed. On the other hand, SC has a low power density but fast response speed. Therefore, in the
control design of ESUs, SC should be responsible for fast transient energy exchange whereas
battery should take care of relatively steady-state energy charge or discharges. This results in a
control design as shown by Fig. 5.5, in which the error signal between measured and reference
dc-link voltage generates a current reference iref through a PI controller and then a low-pass filter
is applied to obtain the battery reference current ib_ref while the rest is used as the SC reference
current isc_ref . This strategy is simple but may result in unstable dc-link voltage when the SC
and/or battery reach their lower or upper energy storage limits.
Vdc _ ref
iref
ib _ ref
ib
vdc
isc _ ref
vdc
vb
vdc
1/ vdc
1/ vdc
In this control scheme, the GCC operates in PQ control mode too, requiring that the dclink voltage must be controlled by ESUs. However, different from Section 5.3.1, the reference
current signals to the battery and SC are determined through a power balance relation as shown
in Fig. 5.6. In the figure, pf represents the power losses in the grid filter which is calculated by:
94
p f R f id2 iq2
(5.12)
where, id and iq are grid d- and q-axis currents, respectively, and Rf is the resistance of grid
filter. The error signal between measured and reference dc-link voltage generates a dc-link
capacitor power reference, pdc_ref. The summation of grid reference power pg_ref, pdc_ref and ppv is
the total instantaneous generated power in the power system. By subtracting loss power from the
filter pf , the charging or discharging power to ESUs is obtained. Similarly, through a low-pass
filter, the battery reference current ib_ref is obtained while the rest is used as the SC reference
current isc_ref.
The primary issues associated with this scheme is that the variation of system parameters,
such as Rf, or the inaccuracy of the power balance calculation, such as neglect of power
converter losses, could affect control effectiveness of the dc-link voltage. In addition, the slow
response speed of the power balance computation and the lower or upper energy storage limits of
the SC and/or battery could affect the dc-link voltage control too.
Vdc _ refdc
pg _ ref
pdc _ ref
pf
v pv
i pv
psto _ ref
pb _ ref
vb
vsc
p pv
psc _ ref
95
ib _ ref
isc _ ref
In this control scheme, the GCC is responsible for dc-link voltage control and the grid
reactive power or bus voltage control. Although this GCC control strategy can maintain a very
stable dc-link voltage, power management of ESUs must be determined through power balance
computation. Similar to the ESU control method described in Section 5.3.1, the summation of
grid reference power pg_ref with pf and actual dc-link capacitor power pdc is the power that should
be outputted by the combined PV array and ESUs before the dc-link capacitor (Fig. 5.7). Then,
the SC and battery reference currents can be obtained in the same way as described in Section
5.3.2. Due to the variation of system parameters and the inaccuracy of the power balance
calculation, the actual output power at PCC could be slightly different from the grid reference
power pg_ref by using this control scheme.
pg _ ref
pf
v pv
i pv
p dc
psto _ ref
pb _ ref b
ib _ ref
vsc
p pv
To evaluate different control schemes, a computational experiment platform of the gridintegrated PV system with ESUs is developed. The experiment system mainly includes: a PV
array module, a battery bank, a SC, four power converter modules, and four corresponding
96
control modules for these converters (Fig. 5.8). The parameters used in the simulation study are
shown in Table 5.1.
The PV array has a series-parallel connection configuration having 10 parallel strings
with each string consisting of 50 series panels. The converter modules are from Opal-RT RTEDrive toolbox. These converter modules can be integrated with the RTE PWM signal generation
function from the Opal-RT RT-EVENTS toolbox to generate drive pulses for very fast and
accurate simulation of power converters [64]. Figure 5.9 shows I-V and P-V characteristics of
the PV array. The control modules are developed with detailed consideration of digital control
system natures, including sample and hold, digital signal processing, and time delays. The
measured voltage and current signals first pass through sampling blocks, which converts
measured continuous signals to discrete signals. Then, a digital filtering mechanism is
applied to remove high frequency components caused by noises or rapid switching of power
converters [13].
Fig. 5.8. Solar PV generator under the control of a dc/dc power converter
using SPS and Opal-RT RT-LAB
97
60
60
40
30
20
0
200
400
600 800
Voltage (V)
1000
Power (kW)
Current (A)
80
0
1200
Table 5.1
Value
Parameters
Value
L1
0.3mH
C1
0.3mF
RL1
0.5 +1.2mH
RL2
1m +33mH
RL3
0.9 +1.2mH
RL4
1m +33mH
Cdc
16mF
Lf1
0.55mH
Lf2
0.55mH
Cf
2.5F
Figures 5.10-12 show results of a case study for the three control schemes presented in
Section 5.3. The solar irradiation is initialized at a value of S=0.75 (Fig. 5.13), where S stands
for the ratio of the present solar irradiation over the nominal irradiation of 1000W/m2. Then, it is
simulated as a parabola curve starts at t=6s with a highest value of S=1 at t=12s. Finally, S
maintains at a value of 0.75 after t=18s.
98
PV
Grid
Q
-50
10
15
25
0
PV
Grid
Q
-25
-50
20
10
Time (s)
15
25
0
PV
Grid
Q
-25
-50
20
10
15
20
Time (s)
Time (s)
reactive power
reactive power
reactive power
1550
1500
1450
1400
10
15
1550
1500
1450
1400
20
10
Time (s)
battery
SC
320
20
battery
SC
5
10
20
69.5
Time (s)
Current (A)
70
Current
SOC
battery
SC
5
10
30
70
15
Current
SOC
5
10
15
15
20
70.5
45
20
Time (s)
60
SOC (%)
30
15
310
20
70.5
45
15
15
320
Time (s)
60
15
330
20
69.5
Time (s)
60
Current (A)
15
330
310
Current (A)
10
340
Time (s)
10
b)Dc-link voltage
SOC (%)
320
1400
Time (s)
330
10
1450
20
340
1500
b)Dc-link voltage
340
15
1550
Time (s)
b)Dc-link voltage
310
1600
Dc-link Voltage(V)
1600
Dc-link Voltage(V)
Dc-link Voltage(V)
1600
70.5
45
30
70
15
0
Current
SOC
5
10
15
20
69.5
Time (s)
5.3.1
5.3.2
5.3.3
99
SOC (% )
50
P(kW)and Q(kVar)
25
-25
50
P(kW)and Q(kVar)
P(kW)and Q(kVar)
50
Irradiation
1
0.9
0.8
5
10
15
Time (s)
20
Power balancing between supply and demand is the most critical task in the system. To
check the demand-following services, the initial grid reference active power is given as 15kW,
which has a positive value when gird power inflow to the dc-link direction (Fig. 5.1).Then grid
demand changes to -35 kW and 5kW separately at t=8s and t=15s. The reactive power reference
is set at 0kVar practically. The simulation results of three control schemes include: PV and gridside generated power and grid-side reactive power (Figs. 5.10a, 5.11a and 5.12a), dc-link voltage
(Figs. 5.10b, 5.11b and 5.12b), battery and SC voltage profiles (Figs. 5.10c, 5.11c and 5.12c),
and battery SOC and current profiles (Figs. 5.10d, 5.11d and 5.12d).
The simulation shows that by using any of the coordinated control schemes, the active
and reactive power can follow the reference well with no error at steady state (Figs. 5.10a, 5.11a
and 5.12a). However, the methods described in Section 5.3.1 and 5.3.2 present a power
overshoot of tens of kilowatts when a change of reference active power occurs at t=8s and t=15s,
which causes voltage fluctuations, distribution losses, power quality and power balancing
reduction, and increase wear and tear on grid hardware. The proposed coordinated control has
the ability to provide the grid with the desire power at both transient and steady state
environments. PV generation maximum power point can be achieved in this situation due to the
roundness and fast response of applied adaptive MPPT control scheme (Section 3.3.2). The
100
reactive power generated in the grid side keeps at 0Var as designed through GCC direct-current
vector control.
Compared the dc-link voltage results (Figs. 5.10b, 5.11b and 5.12b), the proposed control
method which controls dc-link voltage using direct current control has the most accurate
response to the dc-link voltage. This is due to the reason that direct-current vector control
strategy is applied to the dc/ac inverter. Dc-link Voltage control through ESUs will have a delay
on the dc-link voltage performance because the chemical storage process needs a certain period
of time. Additionally, dc-link voltage control through power-balancing control of ESUs has a
delay response not only due to chemical storage process but also a result from calculation error
of the variation of system parameters and the neglect of power loss in the system.
The charging and discharging-voltage profiles present how battery and SC respond to
power regulations. The charging voltage in SC is continuously increasing because power inflows
to ESU in the whole process. However, the low frequency component in the surplus power
decides the discontinuousness of battery charging voltage and current. In Figs 5.10d, 5.11d and
5.12d, there is a point where SOC is larger than 70%, the charging current is switched to
constant-voltage control which is regulated at 335V. The charging current is smooth because the
initial value is set to be same as constant-current control reference current for the constantvoltage controller.
The three-phase current waveforms on the grid side are shown in Fig. 5.14. Reactive
power is regulated at 0Var, which indicates that iq =0, therefore, the amplitude of three-phase
current is proportional to the value of id, which is decided by the grid-side power (Fig. 5.14a and
5.12a). A zoomed-in of three-phase currents (Fig. 5.14b) shows that the currents are balanced
and smooth in Opal-RT simulator. This is because RT-Event allows multiple events in a single
101
simulation time-step, and it is used to generate fast and accurate PWM pulses for high-frequency
switches.
50
25
0
-25
-50
13
Time (s)
18
22
50
25
0
-25
-50
12.95
12.975
13
Time (s)
13.025
13.05
So far, this chapter studies the configuration and detailed components of the gridconnected PV system with energy storage units. It also investigates and compares three
coordinated control designs. In general, the PV array is controlled for MPPT, the battery is used
for slow charging or discharging control, and the suppercapacitor is used for fast charge or
discharge control. However, there are differences in detailed ESU control design depending on
whether the GCC is used for dc-link voltage control or not. If the GCC is used for dc-link
voltage control, the ESUs must be controlled in such a way to meet grid power control demand.
102
If the GCC is used for active power control, the ESUs must provide dc-link voltage control either
directly or through a power balance control mechanism. The simulation results using three
different control schemes shows that when the GCC is used for dc-link voltage and grid reactive
power control, a stable dc-link voltage can be obtained and the integrated PV system can follow
the grid reference power properly.
This dissertation also discusses other applications of coordinated control including the
coordinated control of single-phase system and considering about the ramp rate limit.
With the increasing need for electric power, small distributed generation (DG) systems
are becoming more common. Small DG systems are usually built close to the end-users and they
take advantage of using different energy sources such as wind and solar [105-108]. A few
examples are hybrid cars, solar houses, data centers, or hospitals in remote areas where providing
clean, efficient and reliable electric power is critical to the loads [105-108]. The configuration of
above mentioned systems are similar to Fig.5.8, and the only difference is that there is a singlephase inverter, which is the only interface between sources connected to DC bus and loads
connected to an AC bus. By using the control method proposed in Section 5.3.3 to the singlephase grid-connected PV system, the single-phase inverter is responsible for the control of dcbus voltage and the AC bus reactive power. AC bus active power management is achieved by
charging and discharging ESUs.
103
The DQ rotating frame transformation used in the direct-current vector control of GCC
makes all time-varying state variables become DC variables, thus making the analysis easier
because the GCC can be treated as a dc-dc converter. Because of the limitation of only one
available phase in single-phase converters, this transformation cannot be realized unless a second
phase is created for every state variable in the circuit. Therefore, based on the real circuit model
of the inverter, an imaginary orthogonal circuit is created [105, 107]. Assuming the steady state
real circuit variable is expressed as
X R X M cos t
(5.13)
where, XR is the peak value of sinusoidal waveform which may represent either the voltage or the
current in the rotating frame, is the initial phase and is the fundamental frequency. Ideally
the corresponding imaginary orthogonal circuit variable would be:
X I X M sin t
(5.14)
Equation (5.15) defines the transformation from stationary to rotating frame and from
rotating frame to stationary frame
Xd
XM
X T
XI
q
and
X
XM
1 d
X T X
I
q
(5.15)
where,
cos(t ) sin(t )
T
sin(t ) cos(t )
(5.16)
cos(t ) sin(t )
T 1
sin(t ) cos(t )
(5.17)
It is important to notice that the variables in the rotating frame become constants (DC
values), as shown in (5.15). Those DC values define the DC operating point of the single-phase
104
converters in the rotating DQ frame. All of the control methods developed for DC/DC converters
can be applied.
Fig. 5.15. Single-phase grid connected solar PV generator under the control of a dc/dc power
converter using SimPowerSystems and Opal-RT RT-LAB
P(kW)and Q(kVar)
4
0
-4
-8
PV
-12
0.5
1.5
Load
2
2.5
Time (min)
Q
3.5
Dc-link Voltage(V)
a) Active power of PV system and single-phase load active power and reactive power
500
450
400
0.5
1.5
2
2.5
Time (min)
3.5
b) DC-bus voltage
150
200
50
0
-50
-200
-400
0.5
1.5
2
2.5
Time (min)
3.5
Current (A)
Voltage (V)
400
-150
4
200
50
0
-50
-200
-400
1.55
1.5505
1.551
Time (min)
1.5515
Current (A)
Voltage (V)
400
-150
1.552
Simulation results of the proposed coordinated control method applications in singlephase inverter and ESUs are shown in Fig. 5.16. In fig. 5.16a, it is shown that the proposed
control method can provide the ac-bus load with the demanded active power when the PV
system outputs a same value of generated power. This is achieved by charging and discharging
battery and SC.
Also, it shows that the GCC vector control effectively maintains the reactive power of
system at 0kVar and regulates the dc-bus voltage at 440V steadily as requested (Fig. 5.16b).
With the steady performance of the DC-link voltage, MPPT enables the system to capture the
maximum power of the PV array, which stays the same value because of the same solar radiation
during this period of time.
Ac-bus load single-phase voltage and current are shown in Fig. 5.16c. The amplitude of
voltage is 220 V and it is noted that the amplitude of current is proportional to the amplitude of
active power. The reason is that the regulated reactive power is 0kVar, which determines that the
q-axis component of current is zero (Eq. 5.6), and then according to Eq. 5.5, the d-axis
component of current is proportional to the active power. Figure 5.16d is a zoomed-in of Fig.
5.16c, where a balanced and smooth profile of single-phase current is presented to demonstrate
the effective real-time simulation of OPAL-RT. As it can be seen from simulation results, the
proposed decoupled vector control can control power feeding into the grid, maintain reactive
power of grid and dc-link voltage.
In conclusion, Section 5.5.1 demonstrates that the proposed coordinated control method
presents its effectiveness in controlling load demand and maintaining dc-bus voltage in a singlephase system. The control method allows the single-phase inverter for reactive power and dc-link
voltage control, and makes use of ESUs to control the active power flow. The method uses the
107
reliability and improving the effectiveness of the renewable resource, thus providing a solution to
the integration of distributed renewable energy sources to the electric grid.
Ill. 5.1 Measured solar irradiance profiles (blue areas) for each day in August 2012 [109]
One-min average data are shown from a POA at the Tennessee plant
In this application, the hybrid system of battery and SC are used to smooth the output
power of a PV array. The proposed control is designed to provide active power support with
power smoothing and power ramp control by charging and discharging ESUs (Fig. 5.17) while
maintaining dc-link voltage by GCC direct vector control shown in Section 5.3.3.
In the control structure for ESUs, the difference between the desired and the actual PV
output power can be either injected or drawn from ESUs. Therefore, this energy becomes the
reference generated/ absorbed power from storage units, Psto_ref.. Reference charging or
109
discharging power of battery Pb_ref is obtained after applying a low-pass filter on Psto_ref , and
then the reference current of battery ib_ref is calculated. The rest is used as the SC reference power
Psc_ref and to calculate reference current of SC, isc_ref.
i pv
p pv
vb
psto _ ref
pb _ ref
vsc
v pv
psc _ ref
ib _ ref
ib
i
isc _ ref sc
1/ vdc
1/ vdc
ESU control strategy shown in Fig. 5.17 is implemented in the computational experiment
platform (Fig. 5.8) for simulation. To evaluate the control schemes, solar radiation data of two
random days in Adair Casey [54] is used as two simulation scenarios (Fig. 5.18). Because the
data is provided in minute-scale, a linear interpolation is used for prediction of data between
minutes.
1.5
0.5
12
Time (h)
16
20
24
a) Scenario 1
1.5
0.5
12
Time (h)
16
b) Scenario 2
Fig. 5.18. Hourly solar radiation data of two random days in Adair Casey
110
20
24
20
40
PPV
30
PPV +PESU
P (k W )and Q (k V ar)
P (k W )and Q (k V ar)
50
Grid Q
20
10
0
10
12
14
Time (h)
16
18
20
5
0
P(kW)and Q(kVar)
P(kW)and Q(kVar)
30
20
PPV +PESU
13.5
14
14.5
15
Time (h)
15.5
16
15
10
12
14
Time (h)
16
18
20
PPV +PESU
7.5
8.5
9
9.5
Time (h)
10
10.5
Battery
SC
Battery
SC
2.5
P(kW)
P(kW)
16.5
0
-2.5
-20
4
10
12
14
Time (h)
16
18
-5
20
10
12
14
Time (h)
16
18
20
30
10
Original
Controlled
15
P(kW/min)
P(kW/min)
PPV
10
20
40
13
Grid Q
50
PPV
PPV +PESU
10
10
PPV
15
0
-15
-30
10
12
14
Time (h)
16
18
0
-5
-10
20
Original
Controlled
10
12
14
Time (h)
16
18
20
111
Figures 5.19 and 5.20 present simulation results in both solar irradiation scenarios, based
on which the following conclusions are obtained:
1) Figures. 5. 19a and 5.20a present the PV generated power, active power injected to
grid and grid reactive power. The reactive power reference can follow the reference at 0kVar.
The power injected to gird consists of PV generated power and ESUs charging and discharging
power. It is noted that the ramp rate refers only to real power, and that the reactive power
capabilities of the ESUs can be dispatched simultaneously and independently to achieve other
power system goals. According to Fig 5.19a, 5.20a and their corresponding zoomed-in in Fig.
5.19b and 5.20b, the power captured by PV array is nearly proportional to the solar irradiation.
Therefore, large amount of solar irradiation fluctuations produce relatively large changes in the
captured PV power. Also, it is observed that the proposed ramp rate control can smooth real
power fluctuations from an associated POA, following the moving average of trend of the PV
output. Therefore, the solar PV system with ESUs inject less fluctuated power into the network.
Different low-pass filter parameters are used in these two scenarios for desired performance.
Therefore, two necessary considerations for the control design are indicated: 1) using dynamic
ramp control instead of fix ramp control; and 2) predicting solar radiation profile.
2) Figure 19c and 20c are the charging and discharging power profiles for two scenarios,
the power profile of SC has a much larger frequency than that of battery. The power output of a
POA does not necessarily consist of an equal number of upward and downward deviations, so
the ESUs could have a tendency to discharge or charge outside of its desired operating range.
Therefore, it is observed that one important factor that needs to be considered when
implementing a ramp rate control, is keeping the battery at an appropriate level of charge. The
historical solar irradiation and its probabilities should be applied to design an economic ESU
112
systems associated with the PV generation. Actually, in this real-time simulation, it is remarked
that those factors that can limit the real power output, including the maximum power generation/
absorption of the ESUs, ESUs current limit, battery state of charge, which are satisfied and
simplified in this dissertation for testing the effectiveness of the proposed control method.
3) 1-min ramp rate of original PV generated power and power after ramp rate control are
presented in Fig. 19d and 20d. 1-minute ramp rate is defined as the absolute value of the
difference between the instantaneous power at the beginning and at the end of a 1-minute period
[111]. It should be noted that ramp rate values are highly dependent on the time range used to
determine the scalar ramp rate value. The 1-second ramp rates are far different than the 10minute ramp rate. Formally, ramp rates over different time scales (t) can be calculated using the
equation [111]
Ramp Rate
P(t0 ) P(t0 t )
t
(5.18)
where, P(t0) and P(to+t) are the instantaneous power at t0 and after t. Both comparisons of 1minute ramp rate in Fig. 19d and 20d depict the operation of ramp rate control smooth the
volatile ramp rate of PV array under different irradiation profiles. This behavior translates to a
significant reduction in wear and tear on the electrical grid, and helps to maintain system
electrical frequency.
In order to see the performance of ramp rate control, Table 5.2 shows a comparison of
mean and maximum values of ramp rate before and after ramp rate control in two scenarios. In
the table, ramp rate under scenario 1, which has more severe variations in the solar irradiation,
presents a larger improvement in controlled ramp rate value. It is concluded that integration of
ESUs into the grid to manage the real power variability of PV generation by providing ramp rate
113
control can optimize the benefits of PV resources because of the evident reduction of ramp rate
under both scenarios.
Figure 5.21 is profiles of dc-link voltage for two scenarios, which proves that GCC vector
control effectively regulates and maintains the dc-bus voltage at 1500V steadily as desired.
Table 5.2. Comparison of ramp rate value before and after designed
ramp rate control in two scenarios
Study Case
Scenario 1 (kW/min) Scenario 2 (kW/min)
3.00
0.12
0.72
0.07
27.47
4.95
6.67
1.28
1520
1510
1500
1490
1480
10
12 14
Time (h)
16
18
20
Dc -link V oltage(V )
Dc -link V oltage(V )
Ramp Rate
a) Scenario 1
1520
1510
1500
1490
1480
10
12 14
Time (h)
b) Scenario 2
114
16
18
20
5.6 Conclusion
Energy storage systems are a promising solution regarding the integration of fluctuating
renewable energy into grids. This chapter focuses on control designs of grid-connected
photovoltaic system with ESUs and how to coordinate all the electrical devices in the whole
system by the control of power electronics. In the proposed method, GCC is operated at VQ
mode to maintain a stable dc-link voltage and adapt to the reactive power command, PVCC is
designed to implement the MPPT of PV array, and ESUs connected converters control are
conceived to achieve the power balance of the whole system. A comprehensive computational
simulation study demonstrates that the proposed control structure can effectively supply the
desired reactive and active power to the grid and achieve the stability of dc link voltage through
power electronic converters coordination.
115
The proposed coordinated control has also been modified and tested in other applications
considering single-phase DQ control and ramp rate control. In single-phase grid-connected PV
system with ESUs, it is demonstrated that the proposed method allows inverter for active and dclink voltage control with active power buffered through ESUs. An imaginary orthogonal circuit
has been successfully created and applied into the DQ control. In ramp rate control, the proposed
system control scheme can be used to provide active power support with power smoothing,
power ramp control, and dc-link voltage support. Coupling PV generation and storage devices
will drastically increase reliability of the grid, enabling more effective grid management. It is
indicated that parameters of ESUs and desired level of smooth should be taken into consideration
to make control economical.
116
CHAPTER 6
CONCLUSIONS AND FUTURE WORK
This dissertation demonstrates solar photovoltaic energy generation and conversion from
devices to grid integration.
Firstly, this dissertation investigates solar PV system performance under uneven shading
and dissimilar conditions, especially using both simulation tools and Newton-Raphson algorithm
to study and cross-verify the I-V and P-V characteristics of shaded and unshade cells. It is found
that a traditional PV module with one single shaded cell is the most hazardous condition to affect
proper function of a PV module. It is observed that, with bypass diodes, the performance of PV
device is more complicated and different from the traditional understanding of the PV I-V and PV characteristics and each PV cell with a bypass diode will have the most significant
improvement in the performance of a PV module/array under uneven shading.
Secondly, this dissertation provides a fast and robust MPPT technique and compares it
with typical conventional MPPT algorithms used in solar PV industry. Through both software
and hardware simulation, it is concluded that the proposed MPPT approach has the least
oscillation and the highest stability. Both the sampling rate affects and variable solar irradiance
levels are considered in the comparison. The comparison between the traditional and proposed
adaptive methods shows that the hyperbolic processing of the derivation is important for high
performance of a solar PV system.
117
In ramp rate control of gird-connected PV system with ESUs, it is noted that a dynamic
ramp control design instead of fix ramp is necessary for a practical control in the future work
because of volatile solar radiation.
118
Also, it is not yet considered how to keep ESUs at an appropriate level of charging/
discharging when implementing ramp rate control.
The parameters of ESUs in this dissertation are given ideally to verify the effectiveness of
ramp rate control without optimizing the cost. It is important to design economically considering
factors that can limit the real power output including the maximum power generation/ absorption
of the ESUs, ESUs current limit, ESUs state of charge. With great certainty, we can say that the
demand for development and improvement of grid-integrated renewable energy will remain.
119
REFERENCES
[1]
[2]
[3]
[4]
J. Nelson, The Physics of Solar Cells, London: Imperial College press, 2003.
[5]
[6]
[7]
[8]
[9]
120
[10]
R. Ramaprabha and B.L. Mathur, Modelling and Simulation of Solar PV Array under
Partial Shaded Conditions, in Proceedings of IEEE International Conference on
Sustainable Energy Technologies (ICSET 2008), Singapore, November 24-27, 2008.
[11]
[12]
[13]
[14]
[15]
R. Mallwitz, B. Engel, Solar Power Inverters, CIPS 2010, March, 16-18, 2010,
Nuremberg/Germany.
[16]
[17]
[18]
[19]
[20]
[21]
121
[22]
G. W. Neudeck, PN Junction Diode, 2nd ed., Prentice Hall, ISBN-10: 0201122960, 1988.
[23]
[24]
[25]
[26]
[27]
[28]
[29]
C. Zhang, J. Zhang, Y. Hao, Z. Lin and C. Zhu, A simple and efficient solar cell
parameter extraction method from a single current-voltage curve, Journal of Applied
Physics, Vol. 110, Iss. 6, pp. 064504-064504-7, Sep. 2011.
[30]
M. Ye, X. Wang and Y. Xu, Parameter extraction of solar cells using particle swarm
optimization, Journal of Applied Physics, Vol. 105, Iss. 9, pp. 094502-094502-8, 2009
[31]
[32]
H. Zheng, S. Li, T. Haskew and Y. Xiao, Impact of uneven shading and bypass diodes
on energy extraction characteristics of solar photovoltaic modules and arrays,
International Journal of Sustainable Energy, ISSN. 1478-6541, Aug. 2012.
[33]
[34]
M. H. Rashid, Introduction to PSpice Using OrCAD for Circuits and Electronics, Upper
Saddle River, NJ: Prentice Hall, 3rd ed., 2004.
[35]
[36]
Bypass
diodes,
available
from
Photovoltaic
Education
Network:
http://pveducation.org/pvcdrom/modules/bypass-diodes , last viewed: Nov. 10, 2011.
[37]
[38]
[39]
[40]
[41]
[42]
[43]
T. Noguchi, S. Togashi and R. Nakamoto, "Short-current pulse-based maximum-powerpoint tracking method for multiple photovoltaic-and-converter module system", IEEE
Trans. on Industrial Electronics, Vol. 49, pp. 217-223, Feb 2002.
[44]
[45]
R. Faranda, S. Leva and V. Maugeri, MPPT Techniques for PV Systems: Energetic and
Cost Comparison. Proceedings of IEEE Power and Energy Society General MeetingConversion and Delivery of Electrical Energy in the 21st Century, 2008, pp-1-6.
[46]
[47]
123
[48]
[49]
W. Wu, N. Pongratananukul, W. Qiu, K. Rustom, T. Kasparis and I. Batarseh, DSPbased Multiple Peak Power Tracking for Expandable Power System, Proc. APEC,
2003, pp. 525-530.
[50]
[51]
[52]
M. Veerachary, T. Senjyu and K. Uezato, Neural-network-based maximum-powerpoint tracking of coupled-inductor interleaved-boost-converter-supplied PV system
using fuzzy controller, IEEE Trans. Ind. Electron., Vol. 50, pp. 749-758, Aug. 2003.
[53]
N. Khaehintung, K. Pramotung, B. Tuvirat, and P. Sirisuk, RISC-microcontroller builtin fuzzy logic controller of maximum power point tracking for solar-powered lightflasher applications, in 30th Annual Conf. of IEEE Ind. Electron. Society, 2004, pp.
2673-2678.
[54]
Air
temperature
and
solar
radiation
data,
available
http://www.schoolnet8.com/dl/index.phtml , last time viewed: Sep. 19, 2011.
[55]
[56]
[57]
R. Faranda, S. Leva, and V. Maugeri, MPPT techniques for PV systems: energetic and
cost comparison. IEEE Power and Energy Society General Meeting-Conversion and
Delivery of Electrical Energy in the 21st Century, 2008, pp-1-6.
[58]
C. Otieno, G. Nyakoe and C. Wekesa, A Neural Fuzzy Based Maximum Power Point
Tracker for a Photovoltaic System, IEEE AFRICON, 2009, pp.1-6.
[59]
124
from
[60]
[61]
[62]
[63]
[64]
Real-Time Modeling
rt.com/node/292.
[65]
S. Li, T.A. Haskew, K.A. Williams and R.P. Swatloski, Control of DFIG Wind Turbine
with Direct-Current Vector Control Configuration, IEEE Trans. on Sustainable Energy,
Vol. 3, No. 1, Jan. 2012, pp. 1-11.
[66]
K. Bao, S. Li, and H. Zheng, Battery Charge and Discharge Control for Energy
Management in EV and Utility Integration, Proceedings of 2012 IEEE PES General
Meeting, San Diego, CA, July 22-26, 2012.
[67]
[68]
[69]
[70]
[71]
tools:
RT-EventsTM,
125
available
from
http://www.opal-
available
from:
[72]
K. Kobayashi, H. Matsuo and Y. Sekine, "A novel optimum operating point tracker of
the solar cell power supply system," in 35th Annual IEEE Power Electron. Specialists
Conference, 2004, pp. 2147-2151.
[73]
Evergreen
Renewable
Energy,
available
from:
http://www.evergreenrenewableenergy.co.uk/index.php, last viewed: Nov. 10, 2011
[74]
Solar
Novus
Today,
available
from:
http://www.solarnovus.com/index.php?option=com_content&view=article&id=4138:nat
ional-electrical-code-auto-shutdown-vote-this-week&catid=76:guest-blogs&Itemid=353
, last viewed: Oct. 15th, 2012.
[75]
[76]
H. Zheng, S. Li, T. Haskew and Y. Xiao, Impact of uneven shading and bypass diodes
on energy extraction characteristics of solar photovoltaic modules and arrays,
International Journal of Sustainable Energy, ISSN. 1478-6541, Aug. 2012.
[77]
S. Luo, A. Luo, Z. Lv, Y. Shen, G. Lu, and W. Jiang, Power quality active control
research of building integrated photovoltaic, Power Electronics for Distributed
Generation Systems (PEDG), 2010 2nd IEEE International Symposium, 16-18 June
2010.
[78]
[79]
[80]
[81]
[82]
126
[83]
A. A. Hussein, N. Kutkut, Z. J. Shen, and I. Batarseh, Distributed Battery MicroStorage Systems Design and Operation in a Deregulated Electricity Market, IEEE
Trans. on Sustainable Energy, Vol. 3, No. 3, 2012.
[84]
[85]
[86]
H. Seo, C. Nat, G. H. Kim, S.Y. Kim, N. Kim, H. G. Lee, C. Hwang, M. Park, and I. K.
Yu, Power quality control strategy for grid-connected renewable energy sources using
PV array and supercapacitor, International Electrical Machines and Systems (ICEMS)
Conference, 2010.
[87]
[88]
[89]
[90]
[91]
[92]
[93]
127
[94]
[95]
S. Li and K. Bao, Study of Battery Modeling using Mathematical and Circuit Oriented
Approaches, 2011 IEEE PES General Meeting.
[96]
S. Li, H. T. Haskew and S. Richard, Control of DFIG Wind Turbine with Optimal and
Direct- Current Vector Control Configuration, IEEE Trans. on Sustain. Energy, Vol. 3,
Issue 1, 2012
[97]
[98]
R. Pena, J.C. Clare, and G. M. Asher, Doubly fed induction generator using back-toback PWM converters and its application to variable speed wind-energy generation,
IEEE Proc.-Electr. Power Appl., Vol. 143, No 3, May 1996.
[99]
128
129