Energy and Power E ngineering, 2013, 5, 230-236
doi:10.4236/epe.2013.54B045 Published Online July 2013 (http://www.scirp .o rg/journal/epe)
Comprehensive Modulation and Classification of Faults
and Analysis Their Effect in DC Side of
Photovoltaic System
Mehrdad Davarifar, Abdelhamid Rabhi, Ahmed El Hajjaji
Universi ty of Picardie “Jules Verne”, Laboratory MIS (Modeling, Information & Systems),
33Rue Saint Leu, Amiens, 80039, France
Received April, 2013
ABSTRACT
The first step in automatic supervision, condition monitoring and fault detection of photovoltaic system is recognition,
exploratio n and classification of all possible faults that maybe happen in the system. This pap er aims to perceive, classi-
fied, simulate a nd discus all electrical faul ts in DC side of photovoltaic s ystem, regarding electrical voltage a nd current
inspections. For that, simplified hybrid model of photovoltaic panel in MATLAB environment is used. Investigation
and classification of each type of faults is down and the effects of the faults are illustrated in this paper. Flash test are
applied to improved electrical model. Current-Voltage curves signature are interpreted and investigated in simulation
environment.
Keywords: Photovoltaic Systems; Modeling; Electrical Fault; Fault Detection
1. Introduction
Nowadays photovoltaic (PV) panels are used in several
sector of industry. They could be found, on building
roofs, illuminated highway signs [1], etc. where rural
electrification is embryonic [2], even in infrastructure
industries such as oil industry [3], huge power plant in
desert and different aspects of life. PV markets are
growing fast because of their advantages such as: pollu-
tion free, safety, noiseless, easy installation, and short
construction period. Also tariff of the solar electricity is
enormous compared to the traditional power especially in
peak load time [4]. Enquiries for lower cost and high-
efficiency-devices motivate the researchers to increase
the reliabilit y o f P V systems.
On the other hand, this wide diffusion has of the dis-
tributed generation (DG) such as photovoltaic panels
have not been supported by monitoring, fault detection
and diagnosis equipment.
In fact, monitoring systems represent an additional
cost and it weighs appreciably on Residential Photovol-
taic System (RPS). For this reason, small PV plants are
not regularly checked and partial system faults can occur,
leading to energy losses that are dificult to be observed
and causing financial losses [5-7]. Without proper fault
detection, non-cleared faults in PV arrays not only cause
po wer losses, but also might lead to safety issues and fire
hazards [8].
Photovoltaic system are subjected to different sort of
failures, thus befo re st arti n g sup erviso r y syst em a nd fa ult
diagnosis methods, it is necessary to identify what kind
of failures can be found in the real system. The first step
in this challenge is to cognition and classification all
possible faults, in the first.
On the other hand, the fault detection methods for the
PV array are varied and are classified in the Table 1.
Amon g these metho ds suc h as: vi sual, the rmal, a nd elec-
trical, the visual and the thermal methods need to look
down the PV array and observe the color changes of the
modules or observe the thermal properties such as hot
spots. These methods need thermal cameras or other
equipment in front of the array, while the electrical me-
thod s need o nly t he outp ut ter minal of the array to meas-
ure the volt age, c urre nt, and signa l da ta such as te mpera-
ture and irrad ia tion.
This characteristic of electrical methods is important
for the fault diagnosis technologies to install them into
the power conditioners or into the system inspection
equipment. Then, electrical methods are promising for
PV system fault diagnosis [10, 11]. This paper aims to
perceive, simulate and discuss about electrical fault in
DC side of photovoltaic system and has been focused on
electrical voltage and current inspections. For that sim-
plified hybrid model of PV panel in MATLAB environ-
ment is proposed in third part [12-15].

M. DAVARIF AR ET AL.
231
Flas h test is sp anni ng of the I-V curve range s fro m the
short circuit current (Isc) at zero volts, to zero current at
the open circuit voltage (Voc). The measured and pre-
dicted curve shapes may disagree to some extent even
when the PV string (or module) under test is performing
perfectly. This could be cause d by errors in irradiance or
temperature measurement, or effect of the fault that has
been classified in section two.
2. Classification of Faults in DC Side of PV
System Based on Location and Structures
Generally speaking, faults in PV system could be oc-
curred into t wo side o f the s ystem: DC sid e and AC side,
the interface between this to part is DC/AC inverter that
connected to grid. Internal maximums point power
tracker algorithm with some method must be applied for
rising efficiency of the system.
In AC side of PV system, typically two types of faults
could be happened: total black out which considered as
exter ior faul t for s yste m, l ight ing a nd u nbal anced volt age
or grid outage for AC part defect such as switch frailer,
over current or over voltage and etc. This sort of defect is
not investigated in this work. Besides, it is considered
that the PV array is the only source of fault, since most
PV inverters contain transformers that could provide
good galvanic isolation between PV arrays and utility
grids and perfect electrical protections.
2.1. Typical Faults in PV Array
Typical faults in PV arrays consist of two main groups,
PV panel fault and cabling. In large scale system, some
method of cabal testing could be applied such as earth
capacitance measurement (ECM) and the time-domain
reflect (TDR) [16], but for small PV and domestic appli-
cation these equipment test are so costly. In addition,
number of PV and length of transfer power line in resi-
dential photovoltaic system is finite, and then in this
study has been just focused on integrated panel.
Four of the most co mmon types of fault in PV system
(Earth Fault, Bridge Fault, Open Circuit Fault and Mis-
match Fault) are described and simulated.
2.1.1. PV panel/Module Faults
1) Earth Fault
Earth fault is the most co mmon fault in PV and occurs
when the circuit develops an unintentional path to ground.
According to NEC Article 690.5, two types of grounding
shall be provided for PV system. The first one is system
grounding: (the negative conductor usually is grounded
via the Earth fault protection device (GFPD) in the PV
inverter (Figure 1). The other one is the equipment
grounding: the exposed non-current-carrying metal parts
of PV module frames, electrical equipment, and conduc-
tor enclosures should be grounded.
Two types of Earth faults with zero impedance at dif-
ferent locations have been studied a nd their fault currents
are predicted in simulation
a) Lower Earth fault: the potential fault point is upper
than half o f the maximum voltage po wer point .
b) Upper Earth fault: This fault will cause large backed
current and very high Earth-fault current. This case of
faults is easily detectable by a change of the sign of the
monitored pr imary current o f the solar inverter. An add i-
tional sensor is not necessary. If the primary current be-
comes negative, some modern solar inverters initiate a
controlled internal short circuit [17].
2) Bridging One or More Panels:
A fault bridging in PV system is happened when low-
resistance connection established between two points of
different potential in string of module or cabling. Bridg-
ing faults in PV arra ys may be caused by insulation fail-
ure of cables such as an animal chewing through cable
insulation, mechanical damage, water ingress or corro-
sion.
3) Open Circuit Fa ult:
When one of the current-carrying paths in series with
the load is unintentionally broken or opened, an open
circuit fault can be created. Some examples of this are
poor connections between cells, plugging and unplugging
connectors at junction boxes, or breaks in wires. In gen-
eral, a series arc fault has less energy than a parallel
bridging fault, but it has a much higher probability of
occurring due to the large number of connections in PV
systems.
4) Mismatch Fault:
Mismatches in PV modules occur when the electrical
parameters of one or group of cell are significantly
changed from other. In addition, mismatch faults are
caused by interconnection of solar cells or modules,
which experience different environmental conditions (i.e.
irradiance or temperature) from one another. Mismatch
faults are the most common type of fault compared with
Upper
Ground
Fault
Open ci rcuit Fault
Lower Earth Fault
Bridge Fault
Hot sp ot
Negative
Earth Fault
Shadow
Bypass
diod e fa i l
Delamination
Fault
Hot sp ot
Industrial
MPPT
Figure 1. F ault schematic in DC side of PV system.

M. DAVARIF AR ET AL.
232
Earth fault and bridging faults, among PV arrays. Mis-
match faults may lead to irreversible damage on PV
modules and large power loss. However, they are diffi-
cult to detect using conventional protection devices,
since the y generall y do not l ead to la rge fault currents .
These faults can be categorized into two groups, per-
manent and temporary. Their causes are listed below:
a) Temporary Mismatches: are divided in two groups:
• Partial shading:
Shading effect occurs when a part of the panels array
are shaded which can be caused by a number of different
reasons, like shade from the building itself, light posts,
chimneys, trees, clouds, dirt, snow and other light-
blocking obstacles [18].
Non- uniform temperature: Snow covering,
b) Permanent Mismatches:
• Hotspot:
Hot spot heating occurs when a module’s operating
current exceeds the reduced short circuit current of a
shadowed or faulty cell or group of cells within the mod-
ule [19]. To create a Hot spot fault, a variable resistor in
series with the Rsn of each defective cell could be added
in Simulink. Value of this resistor is considered approx-
imately one unt i l five ohm.
• Soldering: this defective appears in resistive solder
bond between cell and contacted ribbons.
• between cells and contact ribbons, Degradation:
o Discoloration;
o Delaminatio n;
o Transparent layer crack.
Typical Chart of fault in (Figure 2) is illustrated:
2.1.2. Cabling Fault
Such as PV panel three principal type of fault is occur in
power line carrier and cabling system.
1) Bridging Fault:
A very typical location of a possible bridging fault
could be an aged connection box at the back side of a
solar panel or in the corner and bend aria of cable [20].
2) Open-Circuit fault;
3) Earth Fault:
a) Upper Earth Fault
b) Lower Earth Fault
These common faults occur between panels and
ground. Earth fault res ults in lo wered output voltage and
power, and can be fatal if the leakage currents are run-
ning thr o ug h a pe r so n. T he se fa u lts have almost t he same
effect on PV array and PV panel.
Faults
DC side
PV Panel fau lt
Photovoltaic
array
Earth
fau lts
Upper Ground
fau lts
Lower Earth faults
Bridging
faults
Open Circuit
fau lts
Mismatch
faults
Temporary
Mismatches
Partial Shading
Shading
Bird Droppings or
Tress Leav es
Snow
covering
Non-uniform
temperature
Snow covering
Permanent
Mismatches
Soldering
Hot Spot
Degradation
fau lts
Discoloration
Delamination
Crack
Cabling
Bridging
faults
Open Circuit
fau lts
Earth fault s
Upper Earth
fau lts
Lower Earth
fau lts
MPPT
AC si de
(grid)
Grid
outage
Lighting
Inv erter
Total black
out
Not interested in this work
Interested in this work
Figure 2. Classification of Faults i n DC Si de o f PV system bas ed on locat ion and structures.

M. DAVARIF AR ET AL.
233
3. Simulation and Experimental Results
3.1. MATLA B/ Psp ice Cir cuit Ba sed Mode l of PV
Panels for Fault Diagnosis Application
Photovoltaic system are subject to different sort of fail-
ures. Therefore, before stating health monitoring and
fault diagnosis in PV system, finding general and real
time model of PV, with good and fast performance is
inevitable. This general and universal model must have
following specification:
• Applicable for almost all co mmercial PV panel;
• Capable to work in real-time for fault diagnostic
delibera tion and monitoring system;
• Capable to simulate partial shading conditions
(PSC) effect and degradation to investigate pa-
nels fault;
• Possible to connect easily and interfaced to the
electronic devices and power converters model in
MATLAB Simulink for maximum power point
tracking and fau lt diagnosis st ud ie s.
Hybrid model is used and adopted according to the PV
datasheet value for better I-V curve estimation. In fact,
equivalent circuit of a practical PV device includes the
cells connected in series and parallel.
With this modification it is possible to change input
data (irradiating and temperature) for each cell and si-
mulate it individually. Of complying with superposition
rules the simulation result of each cell combine together
to form I-V characteristic of PV module as the output.
In our model (Figure 3):
• Dependent current source to irradiation and te mper-
ature for each cell are considered separately, in or-
der to convert solar energy to electrical form;
• Adopted diode has been superseded instead of nor-
mal silicon dio de;
• Variable resistor is used to appearance effect of
solar irradiation (especially for Amorphous tech-
nology);
• Adopted diode connected in series form to demon-
strate number of P-N junction and became justifica-
tion for high-level ideality factors a.
Using Spice diode block, it is possible to consider en-
vironment temperature effect in the simulation. In fact,
the diode satura tion current Isat and b and -gap energy Eg is
corrected regarding datasheet value, and then parameters
are manipulated in diode block. This procedure could be
extended for PV arrays, output of each PV panel/module
connected together according array configuration to for m
total I-V curve.
3.2. Simulation of Fault in Real Time
The model presented in this work is implemented and
tested on the a-Si:H triple layer amorphous. The panels
are installed in MIS laboratory energy renewable plat-
form in university of Picardie Jules Verne (Fiugre 4).
Solar irradiation data is captured by pyrometer CS300.
This pyrometer connects directly to our data loggers.
PV temperature is sensed with a type K thermocouple
(Silicone rubber patch with self-adhesive aluminum foil
backing) that mounted on panels, also RMS value of
voltage and current are captured by national instrument
data acquisition device (NI DAQ 6212 USB).
IL/n
IL/np
Voc
Rs
Voc/ns
Voc
Voc/ns
Rshnp/nsl
Rsnp/nsl
Rshnp/nsl
Rsnp/nsl
Rshnp/nsl
Rsnp/nsl
Rshnp/nsl
Rsnp/nsl
IL
Rs
V
I
I
L
I L
R
s
Rsh
Eg, an, Isatn
Amorphou
Cell model
Adopted diode
( )
3
ITa
scTmax n
a .k.T .T
IT n max
max satn n
E= -Ln.
gVq.T- T
n max
ocTmax
exp- 1
a.VTm

 

 

 









I+K . T
scn i
I=
sat V+K. T
ocn v
exp- 1
a.Vtn
∆
∆




Figure 3. MATLAB/Pspice circuit based model of PV pa-
nels for fault diagnosis application.
Figure 4. Evaluated Simulink model in MIS laboratory
Energy ren ewable platform.
05 10 15 20 25
-5
0
5
10
15
20
25
30
35
40
45
Time (h)
Po wer ( w)
-- Power of model
-The real power
Figure 5. Comparison between the actual and the power
provided by the model.

M. DAVARIF AR ET AL.
234
Temperature and irradiation captured as input data.
The voltage and current predicted by simulation in real
time. According to this figure, it appeared that there was
a good agreement between the real data given by mea-
surement sensors and the results obtained by simulation.
This Comparison is fundamental task to Feasibility of
error by considering power lowering.
4. Flash Test Result of Faults in DC Side of
PV System (I-V Characteristic)
Flash test is a fundamental method for measuring the
performance I-V characteristics o f photovolta ic panels b y
spanning the PV voltage from zero to open circuit in
short time. The output of measuring is a set of data,
which are determined by output peak power, open circuit
voltage, short circuit current, operating voltage, current,
or power and efficienc y [21].
Flash test has been applied to PV models for different
type of fault scenario. Results of fore principal of fault,
Open circuit (Figure 6), Mismatched fault (Figure 7),
Bridging Fau lt (Figure 8) and Erath Fault (Figure 9) are
illustrated.
00.5 11.5 22.5 33.5
0
10
20
30
40
50
Voltage
Power & Curr ent
I-V Normal
P-V Normal
I- V Open circ iut
P-V Open c ircuit
Figure 6. Open circuit fault affect ed on ISC.
00.5 11.5 22.533.5
0
10
20
30
40
50
Voltage
Power & Curr ent
I-V M ismat ched
P -V M i sm ached
I-V Normal
P -V Norm a l
P -V Hot Spot
I-V Hot S pot
Figure 1. I-V curve ha s notc hes or s teps in M is matc hed an d
Hot spot fault.
00.5 11.5 22.5 33. 5
0
10
20
30
40
50
Voltage
P ower & Current
I-V Normal
I-V Normal
I- V Bridging Fault
P-V Bridging Fault
Figure 2. Bridging Fault Loses power and is effe cted on Voc.
00.51 1.52 2.53 3.5
0
10
20
30
40
50
V oltage
P ower & Current
I-V Normal
P-V Normal
I-V Lower Earth Fault
P- V Lower Eart h F ault
I- V Upper Earth Fault
P- V Upper Earth Fault
Figure 3. Earth Fault affected Voc, Upper Earth fault is
more effective in out-put characteristic.
For simulation initialize three strings in parallel and
each string is included five cells or panel. (It is men-
tioned before that it is possible to consider PV module
instead of cell, because the result is consequence of su-
perposition rules).
Inference and investigated I-V and P-V Curve are in-
terprete d at the end part o f this study:
• The I-V curve shows higher or lower current than
predicted, which is caused by following faults:
- PV array is soiled (especially uniformly).
- PV modules are degraded.
• The slope of the I-V curve near Isc does not match
the prediction, if:
- Shunt paths exist in PV cells (H ot Spot)
- Shunt paths exist in the PV cell interc onnects
- Module Isc mismatch
• The slope of the I-V curve near Voc does not match
the prediction, in cases below:
- PV wiring has excess resistance or is insufficiently
sized
- Electrical interconnections in the arra y are resistive
- Series resistance of PV modules has increased
• The I-V curve has notc hes or steps, if:

M. DAVARIF AR ET AL.
235
- Array is partially shaded
- PV cells are damaged
- Bypass diode is short-circuited
• The I-V curve has a higher or lower Voc value than
predicte d in the following cases:
- PV cell temperature is different than the modeled
temperature
- One or more cells or modules are completely
shaded
- One or more bypass diodes is conducting or s horte d
- One or more PV modules were not included in the
circuit as-built
5. Conclusions
A comprehensive classification of faults in DC Side of
PV system based on location and structures is proposed
for the first time. Flash test applied to validated model
and deferent type of faults are simulated. Inference and
investigated I-V and P-V curves are interpreted.
The future research plans to address some intelligent
algorithm, which is used for fault detection and localiza-
tion i n P V sys tem r egar din g the inference of I-V and P -V
curves characteristics.
REFERENCES
[1] C. J. Huang, L. Y. Kang, Y. C. Jen, and C. S. Jen, “De-
sign and Implementation of a Photovoltaic
High-Intensity-Discharge Street Lighting System,” Power
Electronics, IEEE Transactions, Vol. 26, No. 12, 2011,
pp. 3464 -3471. doi:10.1109/TPEL.2011.2116806
[2] M. Kolhe, “Techno-Economic Optimum Sizing of a
Stand-Alone Solar Photovoltaic System,” Energy Con-
version, IEEE Transactions on, Vol. 24, No. 2, 2009, pp.
511-519. doi:10.1109/TEC.2008.2001455
[3] N. Groom, “Analysis: Oil companies go solar to tap
hard-to -get supplies,” Ed: USA Edition; Reuters, 2011.
[4] D. Solyali and M. A. Redfern, “Why Should Cyprus Ex-
ploit the Solar Power to Match its Peak Demand?” Uni-
versities' Power Engineering Conference (UPEC), Pro-
ceedings of 2011 46th International, 2011, pp. 1-6.
[5] P. D ucange, M. Fazzolari , B. Lazzerini and F. Marcelloni,
“An Intelligent System for Detecting Faults in Photovol-
taic Fields,” In Intelligent Systems Design and Applica-
tions (ISDA), 2011 11th International Conference on,
2011, pp . 1341-1346.
[6] S. Sho jaee, M . Davari far, M. H. Varahram and A. Davari,
“Implementation of Remote Meter Reading Via Power
Line,” In RCEE 2006,Gonabad Azad University, Iran,
2006.
[7] M. Davarifar, Z. Daneshi-Far and H. Fayyaz, “I mpl e-
mentation of Automate Remote Meter Reading (AMR)
via BPL in Kish Island,” In International Power System
Conference (PSC), 2009, Tehran, Iran., 2009.
[8] Y. Zhao, B. Lehman, J. F. DePalma, J. Mosesian and R.
Lyons, “Fault Evolution in Photovoltaic Array during
night-to-day Transition,” In Control and Modeling for
Power Electronics (COMPEL), 2010 IEEE 12th Work-
shop on, 2010, pp. 1-6.
[9] T. Takashima, J. Yamaguchi and M. Ishida, “Disconnec-
tion Detection using Earth Capacitance Measurement in
Photovoltaic Module String,” Progress in Photovoltaics:
Research and Applications, Vol. 16, No. 8, 2008, pp.
669-677. doi:10.1002/pip.860
[10] S. Buddha, H. Braun, V. Krishnan, C. Tepedelenlioglu, A.
Spanias, T. Yeider, et al., “Signal processing for photo-
voltaic applications,” in Emerging Signal Processing Ap-
plications (ESPA), 2012 IEEE International Conference
on, 2012, pp. 115-118.
[11] Z. Ye, Y. Ling, B. Lehman, J. de Palma, J. Mosesian and
R. Lyons, “Decisi on tree-b ased fault d etection an d classi-
fication in solar photovoltaic arrays,” in Applied Power
Electronics Conference and Exposition (APEC), 2012
Twent y-Seventh An nu al IEE E , 2012, pp. 93-99.
[12] M. Davarifar, A. Rab hi and A.Hajjaji, “Circuit Model Of
PEM Fuel Cell For Energy Conversion Uses Renewable
Energy System In Matlab,” in The Internat ional Renewa-
ble Energy Congress (IREC), Hammamet, Tunisia, De-
cember 2011.
[13] M. Davarifar, A. Rabhi and A. E. Hajjaji, "Modeling of
Solar Photovoltaic Panels in Matlab/Simscape Environ-
ment," in International Conference on Renewable Energy:
Generation and Applications” ICREGA’12, Al Ain,
Unite d Ar ab E m ir at e s , March 2012.
[14] M. Davarifar, A. Rabhi, A. E. Hajjaji and J. Bosche,
“Real Time Modeling of Triple layers Amorphous Pho-
tovoltaic Panels,” in 2d International Conference on Sys-
tems and Control (ICSC12),Marrakesh, Morocco , April
2012.
[15] M. Davarifar, A. Rabhi, A. Hajjaji, J. Bosche and X.
Pierre, “Improved Real Time Amorphous PV Model for
Fault Diagnostic Usage,” in Sustainability in Energy and
Buildings, Vol. 22, A. Hakansson, M. Höjer, R. J. How-
lett, and L. C. Jain, Eds., ed: Springer Berlin Heidelberg,
2013, pp . 179-188.
[16] T. Takashima, J. Yamaguchi, K. Otani, T. Oozeki, K.
Kato, and M. Ishida, “Experimental studies of fault loca-
tion in PV module strings,” Solar Energy Materials and
Solar Cells, Vol. 93, No. 6-7, 2009, pp. 1079-1082.
doi:org/10.1016/j.solmat.2008.11.060
[17] C. Strobl and P. Meckler, “Arc Faults in Photovoltaic
Systems,” in Electrical Contacts (HOLM), 2010 Pro-
ceedings of the 56th IEEE Holm Conference on, 20 10, pp.
1-7.
[18] F. Ancu ta and C. Cep isca, “Fault analysis possibilities for
PV panels,” in Energetics (IYCE), Proceedings of the
2011 3rd International Youth Conference on, 2011, pp.
1-5.
[19] A. D. S. Wendlandt, T. Buseth, S. Krauter, P. Grunow,
“Hot Spot Risk Analysis on Silicon Cell Modules,” in
25th European Photovoltaic Solar Energy Conference
and Exhibition / 5th World Conference on Photovoltaic
Energ y Conversio, Valencia, Spain, 2010, pp. 4002-4006.
[20] H. Haeberlin and M. Real, “Arc Detector for Remote
Detection of Dangerous Arcs on the DC Side of PV

M. DAVARIF AR ET AL.
236
Plants,” in 22nd EU PV Conf., Milano, Italy, 2007.
[21] P. Mastny, L. Radil and Z. Mastna, “Possibilities of PV
Panels Defects Identification and Determination of its
Effect on the Economy of Photovoltaic Power Plants Op-
eration,” in Communication and Management in Tech-
nological Innovation and Academic Globalization, 2011,
pp. 233-238.