2004 35th Annual IEEE Power Electronics Specialists Conference Aachen, Germany.

2W4

A Novel Modeling Method for Photovoltaic Cells

Weidong Xiao William G. Dunford Antoine Capel

University of British Columbia University o f British Columbia llniversity Rovira i Virgili
Vancouver, BC Canada Vancouver, BC Canada Tarragona,Spain
weidongx@ece.ubc.ca wgd@ece.ubc.ca antoine.capel I @wanadoo.ti

Absfracr-The mathematical description of current-voltage

characteristics for photovoltaic cells are generally represented
by a coupled nonlinear equation, which is diliicult to solve by
analytical methods. In this paper, a novel modeling process is
proposed to configure a computer simulation model, which is V
able to demonstrate the cell’s output features in terms of
environment changes in irradiance and temperature. Based on
a simplified single-diode model, the parameters are determined
in the sense of minimum model error and temperature effect.
It is tested to simulate three popular types of photovoltaic
panels made of different materials, CIS thin film, muti-
crystalline silicon, and mono-crystalline silicon. The
effectiveness of this approach is evaluated through comparison
of simulation results to the data provided by product’s
r r +

manufacturer.

NOMENCLATURE
I
A Ideality factor y Charge on an electron
(b)
AM Airmass I+, Photo current
G. insolation (W/m’) k Boltnnan wnstant Fig. I The cell equivalent circuits: (a) the single-diode model; (b) the
C, Standard insolation T Cell temperature double-diodemodel
(iOOOW/m’) (degrees Kelvin)
I, Saturation current V, Thermal voltage (volt) model mathematically, due to the exponential equation of a
STC Standard test condition T, Standard temperature
(29810
p-n diode junction. I n [I], the solar model was developed
through the coupled multi-physical processes of
I. INTRODUCTION photovoltaic energy conversion. In 121, the Levenberg-
Marquardt method was chosen to solve the double-
The performance of solar cell is normally evaluated under
exponential model equation. To avoid the modeling
the standard test condition (STC), where an average solar
sophistication, a datzi-based approach is presented in this
spectrum at AM I .5 is used, the irradiance is normalized to
paper.
1000W/m2, and the cell temperature is defined as 25%. To Some researchers [3],[4] on photovoltaic models use
satisfy the requirement of temperature and insolation in constant parameters (i.e. ideality factor A, series resistance
STC, the test usually needs specified environment and some R, and shunt resistance R,,), which result in modeling
special testing equipment, such as an expensive solar inaccuracy. In realiiy, these parameters vary with the
simulator. Simple experiments may not be sufficient to temperature change. To prevent this problem, this proposed
reproduce the electrical characteristics of solar cell modeling method considers temperature effect on the
accurately. In this study, the modeling method is based on important parameters of solar cells.
the specification data provided in the manufacturers’ The modeling process is divided into three steps. First,
datasheets. the simulation modet. is presented and the parameters are
The traditional equivalent circuits of a solar cell determined. Second, a computer simulation model is created
represented by a current source in parallel with one or two to illustrate the electrical features of a solar cell. Finally, the
diodes are shown in Fig.1. The single-diode model [ I ] accuracy of modeling method is evaluated through
includes four components: a photo current source, a diode comparison of simulation results to the practical data.
parallel to the source, a series resistor R, and a shunt resistor
Rp. In double-diode model [2] shown in Fig.l(b), an 11. MODELING
additional diode is added for better curve fitting.
In most cases, it is difficult to determine the 5 parameters This model r e q u k s four parameters derived from data
that can be obtained (iom commercial photovoltaic modules
in single-diode model and 6 parameters in double-diode
under three condition:j short circuit current (Isc), open circuit

0-7803-8399-0/04/$20.00 02004 IEEE. 1950

 2004 35th Annual IEEE Power Electronics Specialists Conference Aachen, Germany, 21394

voltage (Vac), and the current (Impp)

and voltage ( VmPp)at R. I
maximum power points. Temperature coefficients are also
necessary in this modeling method to consider the +
temperature effect on the important parameters of solar
V
cells.
A. Important data used in the photovol!aic modeling L I 0
The data of photovoltaic panel essential for this modeling
Fig. 2 The equivalent circuit of a simplified singlediode model
approach is listed in Table 1.
B. Simplified one-diode model !’_(T)= V,, + A V _ ( T - T ) (4)
Based on the single-diode model, the equivalent circuit where V,, is the open circuit voltage in STC, and AV, is
(Fig.2) used in this study is further simplified by neglecting the temperature coefficient on open circuit voltage.
the shunt resistor. As a function of voltage, the current of a According to the short circuit situation in equivalent circuit
photovoltaic cell is given by shown in Fig.3 (b), the value of photo current is equal to the
current going through the diode:
I, ( G J ) = 1, ( G J ) (5)
Based on Shockley equation, the diode current can be
where Iph represents the photo current, I‘! is the thermal found as
voltage, and Isat is the diode saturation current.
C. Short Circuit Condition
According to the short circuit situation in equivalent
circuit shown in Fig.3 (a), the current relationship can be where ~ ( T ) = Ais~ Tthe
thermal voltage, A is the
expressed as 4
is diode saturation current that is a
ideality factor, and lso,
I , = I” flu: (2)
function of irradiance and temperature. From ( 5 ) and (6),
The value of diode voltage I’D, caused by the short circuit the saturation current I,, can be evaluated as
current I, through &, is very small at short circuit
condition, so the diode current I D is negligible. It is well
known that the photo current Iph is directly proportional to
solar irradiance. Considering the effects of irradiance and
temperature on photo current, the photo current can be E. A i the peakpower point
appoximated as The peak power point (Vmpp imp,,) under different testing
environment can be known according to the photovoltaic
panel’s specification. At the peak power point,
where G, and T represent irradiance and temperature
respectively. Other parameten in (3) are constants: I,, is
short circuit current on standard test condition (STC), G, is
standard irradiance (/OOOW/rn’), AI,, is the temperature Substitute (7) into (S), we get
coefficient on short circuit current, and T, is the standard
temperature (298K).
D. Open Circui! Condition
The open-circuit voltage can be computed as
TABLE I
IMPORTANT DATA USED IN THE PHOTOVOLTAIC
MODELING

>
Geneer01 Specijcations at
standDrd tesr condition (SK)
Peak power: Pme
Short circuit current: I,
Peak power voltage: V-
Open circuit Voltage: V,,
Number of cells in series: N.
Number of Cells in parallel: No
* Temperature coefficients may be represented
ways.
I Temperature coe$icients* v,=v,
I
ITemperature effect on peak + +
power: A Pmpn
Temperature effect on peak
power voltage: A V,,
Ternperam effect on open
circuit voltage:d V,
Temperalure effect on short
circuit current: A I- different
by other
Fig. 3 (a) The equivalent circuit in short circuit condition; (b) the
equivalent circuit in open circuit condition

1951
 2004 35th Annual IEEE Power Electronics Specialists Conference Aachen, Germany, 2604

dP dl I
-= 0 c;)-+-= 0
dV dV V
From (I), we get
dl
By re-organizing (9), we obtain the following relationship dV
between R, and V,
Therefore,

'w
where V, is the thermal voltage, a function o f ideality
factor A .
F. Determining the diode idealiiy factor and series
resistance
The typical power-voltage curve of solar cell is illustrated
in Fig.4. It is clear that the peak powers are the points,
wheredP = 0. Because of P = V I , we know
dV
With (12), the value of dl/dV at the peak power point can
be calculated. By comparing the dl/dV with ~,,p,,~V,,pps,
the
values of A and R, in the model that best represent the
output characteristics of the solar cell can be derived by
computer software. Fig.5 shows the software flow chart to
determine the final vnlues of A and R,. The estimation of A
and R, is done when the absolute value of

[ f,,;nw +?] is minimized.

G. Temperature e@
The parameters o f A and R, determined in Fig.5 only
represent the characteristics of a solar cell at a certain
temperature level. Repeated calculations are necessary to
generate a group of data of A and R,, which illustrate the

voltage (wlt)
(a)

IiI
5 io 15 20
Voltage (wlt)
(b)

Fig. 4 The lypical power-voltage characteristicswith influence of

insolation and temperature; (a) power-voltage characteristicsof a
photovoltaic module for different insolation levels at uinstanl
temperature (25°C): (b) power-voltage characteristicsof a photovoltaic
module for different temperature levels at constant solar insolation
(1OOOw/m') Fig. 5 The software flow chart in determining the final values ofA and
Q

1952
 2004 351h Annual IEEE Power Elecrronics Specialisrs Conference Aachen, G e m m y , 2004

temperature effects. Fig.6 shows the specific characteristics

of ideality factor A and series resistance R, in term of
temperature. In this solar module (Shell ST40), the ideality
factor decreases and the resistance increases, while 1.7-
temperature increases. This model is named as adaptive
parameter model (APM).
;; 1.8~

-BE 1 . 5
111. MODEL
SIMULATION 72 1 . 4 ~
Based on the equations ( I , 3, 4, 7) and the determined
parameters, it is easy to formulate a simulation model with
most computer simulation tools. The model shown in Fig.7
is designed with the Sirnulink" software package. The
1'2L
1.10 10

photovoltaic output current is represented as a function of

voltage (input I), and voltage-current characteristics
configured inside the box is influenced by insolation (input
2) and temperature (input 3). Look-up tables are used to
represent the relationship between temperature and
parameters (A or R,). -E o'042
0.04

0.038-
9
IV. EVALUATION E" 0.036

There are several types of photovoltaic modules made of 5e, 0.034~

various materials. Three types are used to evaluate the 'P

effectiveness of the proposed modeling method. Because the

2 0.032~

open circuit point (Vac, 0) and short circuit point (0, Isc) are
derived directly from the data given by photovoltaic
datasheet, we only need to evaluate the matching accuracy
at the different maximum power points.
A. CIS thinfilm
(bl
An alternative solar cell technology is thin film, which
reduces the material's cost. One of the active materials Fig. 6 The temperalure characteristics of imponant parameters: (a) the
feahues of ideality factor (A) in tern oftemperature:(b) the feahms
based on this technology is copper-indium-diselenide(CIS). of series resistance in terms of temperature
Fig. 8 shows the simulated characteristics of a typical CIS

4IPh

Q
T(k)
T(K) Voc kV.x 1.a.-

voc Vt

Q
V K1.-

1953
 2004 35th Annual IEEE Power Electronics Specialists Conference Aachen, Germany, 2004

thin film solar module (Shell ST40). TABLE II

The comparison between the simulation and practical data SIMULATION ERRORS ON THE M A X I M U M POWER POINT A T
is illustrated in Table 11. The results show that the average -
ENT TEMPERATURE (SHELL ST40)
Data~rovrdedby I Irnulatron I Relarrva
relative error on peak power voltage is 0.34% and the
average relative error on peak power is 0.02%. The Temperature ',P = 0.00% on
simulation error plot is shown in Fig.9 in a certain 50°C v, = 14.1ov
I insolation:
temperature range from -25°C to 49°C. Based on error IOOoW/m' - 1416V V,,
analysis, the simulation accuracy can be compared between Temperature: P, := 40W .,
p I= 0.00%on
the constant-parameter model (CPM) and the adaptive- 23T V=
, 16.60V 40.00W ,.P
parameter model (APM). The parameters of CPM are Insolation: .V, = 0.06%on
I OOOW/m' 16.39V
determined with the data at STC. The results show the Temperature 0.00% on
relative simulation deviation for CPM is 1.57% and 0.04% 0°C
for APM. This demonstrates the proposed model is more Insolation:
IOOOW/m'
accurate in term of simulation results. Temperature 0.12%on
E. Muti-crystalline Silicon -25°C
Iinsolation:
Fig. I O shows the simulated characteristics of a typical I OOOW/m'
muti-crystalline solar module (Shell S36). The comparison
between the simulation and practical data is illustrated in
Table 111. The results show that the average relative error on
peak power voltage is I .44% and the average relative error
on peak power is 0.15%. Again, the simulation error plot in
term of variable temperature level is shown in Fig. 1 I . The
results show the relative simulation deviation for CPM is
0.56% and 0.16% for APM.
C. Mono-cryslalline silicon photovoltaic module
Fig. 12 shows the simulated characteristics of a typical
mono-crystalline solar module (Shell SP70). The
comparison between the simulation and practical data is
illustrated in Table I V . The results show that the average
relative error on peak power voltage is 0.34% and the
average relative emir on peak power is 0.02%. The
simulation errors in term of variable temperature level are
shown in Fig.13. The results show the relative simulation
vonage *a) deviation for CPM is 1.19% and 0.03% for APM.
(a)

0.041
1

'0 5 10 15 20 25
Vobge (vdt)

(b) Temperature ('C)

Fig. 8 The current-voltage characteristicsofClS thin film solar module Fig. 9 Relative simulation error in the comparison of APM to CPM in
(Shell ST40) (a) the characteristics for different insolation levels at term of temperature change (CIS thin film, Shell ST40)
constant temperatun (23'C) (b) the characteristics for different
temperature levels at constant solar insolation ( IOOOw/m')

1954
2004 35rh A n n u l IEEE Power Elecrronics Specialisls Conference Aachen, Germany, 2094

TABLE 111

Insolation: v*, = 2.31% on

1OOOW/m’ 14.94V V,
Temperature: P,= 36.00W Pw * = 0.15%on
2ST V, = I6.50V 36.06W Pw
Insolation: V,*= 1.55%on
1000Wlm’ 16.76V V,
Temperature: Pw=4O.0SW PW* = 0.011% on
0°C V,= 18.40V 40.09W P,
01 Y Insolation: vm* = 1.18%on
0 5 10 15 20
VOhge (volt) Temperature: P,=44.1OW .P, = 0.012%on
(4
Insolation: V *= 1.05%on
10Oow/m’ 20.51v

ACKNOWLEDGEMENTS
The authors would like to acknowledge the very useful
discussions with Magnus Lind, at the department of
electrical and computer engineering, University of British
Columbia.

20 25
Vormtage (volt)
(b)
Fig. 10 The cunent-voltage characteristics of muti-crystalline silicon
photovoltaic module (Shell S36) (a) the characteristics for different
insolation levels at constant temperature (25‘C) (b) the characteristics
far different temperature levels at constant solar insolation ( 1000wlm2)
REFERENCES
[I] S . Liu and R. A. Dougal, “Dynamic multiphysics model for solar
array”, IEEE Tram. Energv Conversion, Vo1.17, No.2, pp.285-294,
June 2002
[2] 1. A. Cow and C.D. Manning, “Development of a photovoltaic array
model for me in power-electronics simulation studies”, IEE Proc.
Elecrr. Power Appl., Vol. 146, Na.2, pp.193-200, March 1999
H. Yamarhita, K.Tamahashi, M. Michihim., A. Tsuyoshi, K. Amako,
[3]
and M. Park, “A novel Simulation technique of the PV generation
system using red weather conditions”, in 2002 Proc. Power
Conversion Con!, V01.2, pp. 839 4 4 4 , April 2002
141 G.A. Vokas, A.V. Machias, and J.L. Souflis, “Computn modeling
and parameters estimation for solar cells”, h 1991 Proe
Medirerranean Elecrrorechnicol Con/, v0l.l. pp. 206 -209, May
1991
[SI Shell Solar Product Information Sheets

V. CONCLUSION
In this paper, a general approach on modeling
photovoltaic modules is presented. The points chosen for
the parameter determination are the short circuit current
point (0, L), open circuit voltage point (Vx, 0), and the
maximum power point (V,,, Impp). The data needed for the
model are from either the products’ data sheet or
experimental testing results. Three types of solar module
(CIS thin film, m-Si and c-Si) were modeled and evaluated.
The model accuracy is also analyzed through comparison
between product’s data and simulation results. Compared
with constant parameter model (CPM), the deviation of
simulation results from the actual data is very much
I
reduced. The evaluations prove the effectiveness of this
modeling method based on a simplified one-diode model.
This allows efficient use of SimulinkQ to model
photovoltaic power systems.

I955
2004 35th Annual IEEE Power Electronics Specialists Conference Aochen. Germany, 2W4

TABLE Iv

v ,- 7
....................... ........................
4 .............. i............... i...............t ..... .......!...
4s
Condrtiom
i rnmufizcturerr5] i y$= i ;??%;
SIMULATION ERRORS ON THE MAXIMUM POWER POINT AT
DIFFERENT TEMPERATURE (SHELL SP70)
I
[ Doto provided by Sirnulotion Relalive I
Temperature: Pw=62.13W
=,V 14.60V 62.13W
Insolation; vm* = 0.04% on

.............. i .............. :
............. .........
0 1.5 ..............ij ..............
4oOWh 2
............... j .............
I Insolation:
, .V = 16.50V
v,* =
16.50V
0.00% on

0.Wh on
.............. ..................
Insolation: v-* = 0.05% on
5 10 15 20
I 18.31V I v,,
0.06% on
Voltage (volt) -25°C V,=20,3OV 85.8OW ,P
(4
Insolation.
Iooow/m~ I V,'
20IIV
= 0.94% on
v.,

M 25
votage Wt)

(b)

Fig. 12 The cunent-voltage characteristics ofmono-crystalline silicon

photovoltaic module (Shell SP70)(a) the characteristics for different
insolation levels at constant temperature (25'C) (b) the characteristics
for different temperature levels at constant solar insolation (IOOOw/m')

1956