Вы находитесь на странице: 1из 10

Inverter Power Sizing Considerations in Grid-Connected PV Systems

Guillermo Velasco*, Francesc Guinjoan**, Robert Piqu*, Alfonso Conesa* and J.J. Negroni**
Electronic Engineering Department Polytechnic University of Catalonia (UPC)
*
CEIB-EUETIB, C/. Compte dUrgell, 187 **ETSETB, C/. Jordi Girona, 31
Barcelona, SPAIN
Tel.: +34 / 93 413 72 89
Fax: +34 / 93 413 74 01
guillermo.velasco@upc.edu
guinjoan@eel.upc.edu
http://www.euetib.upc.edu
http://www.upc.edu

Acknowledgements
This work has been partially granted by the Spanish Ministry of Science and Technology and by the EU
(FEDER funds) through the research project DPI-2006-15627-C03-01, and was developed at the URT
EdePAE 1 installations, of the EUETIB.

Keywords
Renewable energy systems, Photovoltaic, Efficiency, Modeling.

Abstract
This work is devoted to establish preliminary criteria helping in the choice of a central inverter power
sizing in grid-connected PV systems in order to maximize the yearly energy injected to the grid. These
criteria come from an estimation of the injected energy by means of a set of Matlab-based simulations
involving a simulation-oriented model of the PV conversion chain, the environmental data of several PV
Spanish sites as well as the PV installation mounting type characteristics.
The simulation results show that the current practice of under-sizing the inverter maximum power with
respect to the PV generator nominal power may not be the best choice in terms of yearly produced energy
and also evidence the strong impact of the PV generator operating temperature in the choice of the optimal
inverter power sizing.

I. Introduction
Central inverter grid-connected PV systems are formed by a set of series-parallel electrically
interconnected PV modules, which are known as the PV generator and one central inverter ensuring the
proper transfer of the collected energy to the mains. The choice of the central inverter maximum power
(PINV) is related with the total installed peak power of the PV generator (PGF) given in Standard Test
Conditions (STC: 1000W/m2 at 25C and for a solar spectrum of AM1.5) and has to maximize the injected
energy to the mains. The ratio between these two powers is known as the Sizing Factor (SF) and is
defined as (1):

PINV = SF PGF

(1)

The SF value is normally chosen lower than one since the PV generator generally operates below the
irradiance level and above the temperature value given by the STC.
1

Technology Transfer Unit on Power Electronics and Electric Drives

In previous papers [1 - 2] a simulation-based procedure using simple PV system models neglecting the
temperature influence on the energy production was presented to estimate the SF optimal value
maximizing the yearly injected energy.
This work follows the same simulation procedure, but includes the operating temperature effects in the PV
generator model. Accordingly, the main objective is to establish preliminary criteria on the optimal SF
value choice by estimating the yearly injected energy to the grid, taking into account the following
characteristics of the PV conversion chain elements involved in the power production process, namely:
1. Irradiance and temperature level evolution at the PV installation site
2. Characteristics of the PV generator (operating temperature, PV mounting type )
3. Inverter electrical characteristics (maximum power, efficiency curve )
The paper content is organized as follows:
Section II describes, in terms of power processing, a set of simplified models of all the elements involved
in the PV conversion chain.
Section III focuses on the numerical simulation procedure leading to the yearly injected energy estimation.
This simulation procedure is applied, in Section IV, to different PV sites assuming the same PV generator,
the same inverter and considering the SF as a parameter.
Finally, Section V presents a set of criteria on the SF choice from the simulation results obtained in
Section IV.

II. PV system modeling


This Section details the simulation-oriented models of all the elements of the PV grid-connected system
conversion chain used in this work.

Solar irradiance and temperature data


The available power at the PV generator output (PDC) depends, among others, on the solar irradiance and
temperature of the PV system location. The irradiance level and ambient temperature values are dependent
on the PV site location latitude, longitude and altitude. Therefore, it is assumed that the SF optimal value
will exhibits this kind of dependence.
This work uses the hourly average solar radiation and temperature data available at the SoDa2 project
database for the 47 Spanish peninsular capitals of province. These climatologic data correspond to the
hourly average of the daily ambient temperature (TA) and the incident irradiance on a flat surface (G), in
accordance with the image processing coming from several meteorological satellites.
The hourly average irradiance and temperature values of each day have been also averaged for all the days
of each month. As a result, a set of points in both the irradiance-time and temperature-time planes have
been obtained and subsequently interpolated by means of a third order polynomial function.
At the end of this process 12 curves for the irradiance level and temperature evolution (one per month)
have been generated for the 47 selected PV locations. Fig. 1 shows the time evolution of irradiance level
and temperature for the city of Zaragoza, located at: latitude 41.63 N, longitude 0.88 O and altitude
210m above the sea level.
2

SoDa: Solar Radiation Database for Environment: www.soda-is.com

Fig. 1: Monthly irradiance [W/m2] and temperature [C] time evolution at the city of Zaragoza

Photovoltaic generator model


The available power at the PV generator output (PDC) depends on the incident irradiance (G), the PV
modules operating temperature (TM) and its efficiency (1) [3 - 4], and can be modelled as (2):

PDC = 1 G [1 (TM TR )] SGF ,


where:

(2)

1: PV generator efficiency at 25 C.
G: Incident irradiance on the PV generators plane.
: Thermal power coefficient of the PV generator material.
TM: PV generator operating temperature.
TR: Reference temperature (25 C).
SGF: PV generator surface.

A constant value for both PV generator efficiency and thermal power coefficient of the PV generator
material has been assumed, and fixed to the typical values of silicon mono-crystalline PV modules,
namely 1 = 0.12 and = 0.5 %/C. On the other hand, this work considers a static PV generator (i.e. with
no sun tracking capability) oriented to the South, with a surface of 10 m2. Taking into account the
previous assumptions, the nominal power of this PV generator will be of 1.2 kWp.
Finally, the PV generator operating temperature (TM) simulation model is defined in [5 - 6] and is given
by (3):

TM = TA + G ,
where:

(3)

TA: Ambient temperature.


G: Incident irradiance on the PV generators plane.
: Thermal coefficient of the PV generator according to mounting type.

As shown in [7] from an analysis considering several mounting types of grid-connected PV systems
located in different countries, the thermal coefficient value () exhibits a strong dependency with the PV
generator ventilation capability (natural ventilation by convection, or forced ventilation by wind or
airflows).
According to this work, the typical values of the thermal coefficient are fixed to = 0.025 for PV systems
on flat surface (flat roof) with good ventilation, and = 0.050 for PV systems integrated in buildings

(habitually on faade or sloped roof) where the PV generator ventilation is worse. The simulations of this
work will use these typical values.
Fig. 2 shows the daily operating temperature for a PV generator located at Zaragoza, during the month of
August and for the PV generator mounting types previously mentioned.

Fig. 2: PV generator operating temperature for = 0.025 (flat roof) and = 0.050 (building integrated)

Inverter model
The available power at the inverter output (PAC) will depend on both the input power (PDC) and the
inverter efficiency (2). This work has considered the model given by [8], which is applicable for input
power ranges lower than the maximum inverter power (PINV). This model is given by the equation (4):

2 =

p
,
p + k0 + k1 p + k 2 p 2

(4)

2: Inverter efficiency.
p: PDC / PINV ratio.
k0: Losses coefficient at not load.
k1, k2: Linear and quadratic current losses coefficients.

where:

For input power ranges greater than the inverter maximum power, this work assumes the inverter
operating mode suggested in [9 - 10], i.e., the input power (PDC) is limited to its maximum value without
output power interruptions until overload conditions are no longer present.
In this case, the new inverter efficiency function can be properly modelled by the following hyperbolic
dependence (5):

2 =

2 ( p =1)
p

(5)

Fig. 3 shows the inverter efficiency curve for the set of parameters k0 = 0.014, k1 = 0.001 and k2 = 0.081,
which closely match the efficiency curve of a commercial grid-connected inverter with low frequency
isolation transformer (SB1100 inverter from SMA).

Fig. 3: Inverter efficiency curve used in this work

III. Simulation procedure


Fig. 4 shows the block-diagram of the simulation procedure used in order to obtain the Sizing Factor (SF)
optimal value of the PV system for the sites considered in this work. As it can be seen, both the value of
the available power at the inverter input (PDC) and the annual energy delivered by the PV generator (EDC)
are computed from the irradiance and temperature data and the PV generator model. Taking into account
the inverter efficiency model, the inverter output power (PAC) and the value of the yearly injected energy
to the grid (EAC) can be estimated.

Fig. 4: Simulation procedure block-diagram

Finally, the energy efficiency of the PV system defined by (6) can be evaluated.

E =

PAC
PDC

(6)

IV. Simulation results: optimal SF and energy production maps


The numerical simulations have been carried out by means of the Matlab software and are organized into
two categories, namely:
a) Simulation results corresponding to the environmental variables processing (average irradiance and
temperatures for the 47 PV sites)
b) Simulation results leading to the optimal SF value and the corresponding yearly produced energy
according to the simulation procedure described in Section III. These simulations are also performed
for the 47 PV sites.
These results and the corresponding coordinates of the PV sites are subsequently combined by means of a
regressions computing software (DataFit from Oakdale Engineering) to generate a set of continuous
functions which estimates with the minimal error the geographical distribution of the computed variables
values. These geographical distributions are finally represented on a Spain map for further exploitations.
The following paragraphs give more details of the simulation procedure for each of the computed
variables.

Irradiance and temperature average values


For each of the 47 PV sites, the incident irradiance and the ambient temperature have been averaged only
during the daylight hours. Fig.5 and Fig.6 show the geographical distribution of the average irradiance and
the average ambient temperature values.

Fig. 5: Geographical distribution of the average


irradiance, in W/m2

Fig. 6: Geographical distribution of the ambient


temperature, in C

Optimal Sizing Factor (SF) value


The optimal SF maximizing the yearly injected energy (or equivalently the yearly energy efficiency, E) is
obtained for each PV site from the simulation procedure given in the flowchart of Fig.4.

As regards the PV generator operating temperature effect on the optimal SF value, the resulting
simulations have considered two installation mounting types, the first one mounted on a flat surface with
good ventilation (i.e. = 0.025), and the other one integrated in buildings with a worse ventilation
capability (i.e. = 0.050).
Fig. 7 and Fig. 8 show the SF optimal values maps obtained from the regression analysis for both values
of the thermal coefficient ().

Fig. 7: Geographical distribution of the SF optimal


value for flat surface mounting systems ( = 0.025)

Fig. 8: Geographical distribution of the SF optimal


value for integrated mounting systems ( = 0.050)

Table I summarizes the most significant results obtained with respect to the SF optimal value, the
irradiance levels and the average temperature for the 47 studied cities.

Table I: SF optimal value, irradiance average level and average temperature


Location
Cdiz
Zaragoza
Almera
Huelva
Lrida
Granada
Zamora

Vitoria
La Corua
Santander
Oviedo

Optimal SF value
Flat surface
Integrated
1,03
0,93
1,03
0,93
1,02
0,92
1,02
0,93
1,01
0,92
0,99
0,90
0,98
0,89

0,76
0,71
0,75
0,70
0,74
0,69
0,73
0,68

Irradiance
average [W/m2]

Temperature
average [C]

396,8
390,2
396,7
405,6
382,9
390,9
355,9

269,4
262,1
263,1
255,4

18,2
16,6
20,1
19,9
17,3
17,4
14,2

13,0
15,7
15,6
14,4

Energy production
In order to compare the yearly produced energy of different sites assuming that their corresponding PV
installations exhibit the optimal SF, the energy value of each site is normalized with respect to the

maximum value of the obtained energies and is referred as EAC (%). Fig. 9 and Fig. 10 show the EAC
values map for the previous values of the thermal coefficient ().

Fig. 9: EAC normalized value (%). Geographical


distribution for flat surface mounting installations
( = 0.025)

Fig. 10: EAC normalized value (%). Geographical


distribution for integrated mounting installations
( = 0.050)

Table II summarizes the most significant results with respect to the annual energy production EAC for the
47 cities.
In order to enhance the comparative, two different normalized values of produced energy are computed
for building integrated installations, namely: the first column corresponds to a normalization with respect
to the maximum energy value of this mounting type, whereas the second one considers the maximum
energy value of the flat surface installations mounting type.
By this way given a PV site, the loss of produced energy depending on the installation mounting type can
be evaluated.

Table II: Normalized annual energy production obtained results

Location

Cdiz
Huelva
Zaragoza
Almera
Lrida
Granada
Jan

Lugo
Santander
La Corua
Oviedo

Flat Surface (FS)

Energy [%]
Building Integrated (BI)

E AC ( FS )

E AC ( BI )

E AC ( BI )

E AC ( FS ) max

E AC ( BI ) max

E AC ( FS ) max

100,0
99,4
99,1
98,5
96,9
96,7
95,7

71,5
69,7
69,6
67,9

100,0
99,1
99,0
98,5
96,9
96,9
95,8

73,2
71,6
71,4
70,0

92,2
91,3
91,2
90,8
89,3
89,3
88,3

67,5
66,0
65,8
64,5

Finally, the grid injected energy, in terms of the SF, has been computed for different PV sites and for the
two installation mounting types, as plotted in Fig.11 for the city of Barcelona.

Fig. 11: PV system annual energy production at Barcelona versus SF value for the two types of
installations under consideration
This plot shows on one hand the values of the optimal SF, and the energy production difference depending
on the installation mounting type. In particular a reduction of a 5% is observed in the case of building
integrated PV installations, this being attributable to the greater value of the PV generator operating
temperature of this type of installations due to the worse ventilation with respect to the flat surface
mounting ones.

V. Conclusions
This work has presented a set of simulations based on simple simulation-oriented models of all the
elements of the generation and conversion chain involved in PV grid-connected systems, namely:
irradiance and temperature evolution, PV generator and inverter models. These simulations are oriented to
calculate the optimum SF value, defined as the ratio between the maximum inverter power and the
nominal power of the PV generator, which maximizes the yearly energy production.
The obtained results, carried out for 47 Spanish PV sites, evidence the existence of a SF optimal value,
maximizing the installation yearly energy production and show that this value can be greater than one for
PV installations located at low latitude sites. Accordingly, for these latitudes the results suggest to
oversize the inverter maximum power with respect to the PV generator nominal power, in spite of the
under-sizing current practice.
On the other hand, a geographical distribution of the SF optimal values and the yearly produced energy
has also been obtained and lead to the following conclusions:

With respect to the geographical distribution of the SF optimal value:

The SF optimal value has exhibits a stronger dependency with the irradiance than the temperature.
Therefore, PV sites with higher average irradiance levels present higher optimal SF values.
For PV sites with similar levels of irradiance, the greater SF optimal values correspond to those
sites with lower average temperature levels.

With respect to the value of the optimal SF in a PV site:

The SF optimal value is strongly dependent with the PV generator operating temperature. The
lower the PV generator operating temperature (i.e. the better PV generator ventilation), the larger
the SF optimal value is.

Finally, with respect to the yearly injected energy to the grid in the different PV sites:

PV sites with a greater SF optimal value are also the PV sites with a greater yearly energy
production.
The yearly injected energy to the grid is greater when the PV generator operating temperature is
lower.

References
[1] G. Velasco, F. Guinjoan and R. Piqu: Consideraciones sobre la Relacin entre las Potencias del Generador
Fotovoltaico y del Inversor en Sistemas Fotovoltaicos Conectados a Red, XIII SAAEI. Gijn, September 2006. (In
Spanish)
[2] G. Velasco, F. Guinjoan, R. Piqu and J.J. Negroni: Sizing Factor Considerations for Grid-Connected PV
Systems Based on a Central Inverter Configuration, 32nd IECON. Paris, November 2006.
[3] N.J.C.M. van der Borg and A.R. Burgers: Inverter Under-sizing in PV Systems, 3th WCPEC pp. 2066-2069.
Osaka, May 2003.
[4] David L. King, Jay A. Kratochvil and William E. Boyson: Temperature Coefficients for PV Modules and Arrays:
Measurement Methods, Difficulties and Results, 26th PVSEC pp. 1183-1186. California, September-October 1997.
[5] T. Nordmann and L. Clavadetscher: Understanding Temperature Effects on PV Systems Performance, 3th
WCPEC pp. 2243-2246. Osaka, May 2003.
[6] H.G. Beyer and O. Prignitz: Modeling the PV-Output, PVSAT 2 Workshop on Satellite-Based Performance
Check of Photovoltaic Systems. Barcelona, June 2005.
[7] T. Nordmann and L. Clavadetscher: Understanding Temperature Effects on PV System Performance, 3th WCPEC
pp. 2243-2246. Osaka, May 2003.
[8] E. Lorenzo: Retratos de la Conexin Fotovoltaica a la Red V. De la AIE a los Inversores, ERA SOLAR Vol
126 pp. 52-59, 2005. (In Spanish)
[9] L. Keller and P. Affolter: Optimizing the panel area of a photovoltaic system in relation to the static inverter
Practical results, Solar Energy Vol 55 no 1 pp 1-7, 1995.
[10] B. Burger and R. Rther: Inverter sizing of grid-connected photovoltaic systems in the light of local solar
resource distribution characteristics and temperature, Solar Energy Vol 80 no 1 pp 32-45, 2006.

Вам также может понравиться