See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/262641655

New procedure for fault detection in grid connected PV systems based on the
evaluation of current and voltage indicators

Article  in  Energy Conversion and Management · October 2014

DOI: 10.1016/j.enconman.2014.05.008

CITATIONS READS

133 1,473

5 authors, including:

Santiago Silvestre Chouder Aissa

Universitat Politècnica de Catalunya Université de M'sila
115 PUBLICATIONS   4,481 CITATIONS    114 PUBLICATIONS   3,463 CITATIONS   

SEE PROFILE SEE PROFILE

Daniel Guasch Murillo Engin Karatepe

Universitat Politècnica de Catalunya Ege University
51 PUBLICATIONS   420 CITATIONS    40 PUBLICATIONS   2,859 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Inverter control in Micro grids, View project

photovoltaic energy storage View project

All content following this page was uploaded by Chouder Aissa on 19 May 2020.

The user has requested enhancement of the downloaded file.

 Energy Conversion and Management 86 (2014) 241–249

Contents lists available at ScienceDirect

Energy Conversion and Management

journal homepage: www.elsevier.com/locate/enconman

New procedure for fault detection in grid connected PV systems based

on the evaluation of current and voltage indicators
Santiago Silvestre a,⇑, Mário Aires da Silva a, Aissa Chouder b, Daniel Guasch c, Engin Karatepe d
a
MNT Group, Electronic Engineering Department, Universitat Politécnica de Catalunya (UPC) BarcelonaTech, C/Jordi Girona 1-3, Campus Nord UPC, 08034 Barcelona, Spain
b
Univ. M’sila, Fac. Technologies, Dep. Génie Electrique, BP 166 Ichbelia, 28000 M’sila, Algeria
c
Departament d’EnginyeriaTelemàtica. Universitat Politécnica de Catalunya (UPC) BarcelonaTech. EDIFICI VG1 (EPSEVG). Avda. Víctor Balaguer, 1. 08800 Vilanova i la Geltrú, Spain
d
Department of Electrical and Electronics Engineering, Ege University, 35100 Bornova, Izmir, Turkey

a r t i c l e i n f o a b s t r a c t

Article history: In this work we present a new procedure for automatic fault detection in grid connected photovoltaic
Received 3 March 2014 (PV) systems. This method is based on the evaluation of new current and voltage indicators. Thresholds
Accepted 1 May 2014 for these indicators are defined taking into account the PV system configuration: number of PV modules
Available online 27 May 2014
included and series and parallel interconnection to form the array. The procedure to calculate the thresh-
olds that allow the identification of the faults is described. A simulation study was carried out to verify
Keywords: the evaluation of current and voltage indicators and their corresponding thresholds for a set of PV sys-
PV systems
tems with different sizes and different configurations of interconnection of PV modules. The developed
Fault detection
Current and voltage indicators
method was experimentally validated and has demonstrated its effectiveness in the detection of main
faults present in grid connected applications. The computational analysis has been reduced and the num-
ber of monitoring sensors minimized. The fault detection procedure can be integrated into the inverter
without using simulation software or additional external hardware.
Ó 2014 Elsevier Ltd. All rights reserved.

1. Introduction necessary climate data at the desired location [13–17] or meteoro-

The monitoring and regular performance supervision on the
functioning of grid-connected photovoltaic (PV) systems is neces-
sary to ensure an optimal energy harvesting and reliable power
production. The development of diagnostic methods for fault
detection in the PV systems behavior is particularly important
nowadays due to the expansion degree of grid connected PV sys-
tems and the need to optimize their reliability and performance.
Most diagnostic methods are based on the comparison of mon-
itored data with simulation results of the PV system behavior [1–
6]. Monitored climate data, irradiance and temperature, are needed
for an estimation of the energy produced and the simulation of the
PV system behavior in a determinate localization [7,8]. Further-
more, some works have been carried out using artificial intelligent
techniques [9–11] and statistical data analysis for supervision of
PV systems [12], but these techniques have not been yet optimized
for fault detection analysis and clear identification of the kind of
fault present in the system.
On the other hand, several researches were also carried out
using climate data from satellites observation to generate the
⇑ Corresponding author. Tel.: +34 93 4017491; fax: +34 93 4016756.
E-mail address: Santiago.silvestre@upc.edu (S. Silvestre).
logical data bases [18,19]. These approaches are cost-effective,
since no climate sensors are needed on the plant, although they
provide low accuracy in estimation of expected energy yield in
some specific climatic conditions.
Nowadays most inverters designed for grid connected PV appli-
cations have a wide range of interfaces, in particular sensor inputs
and communication interfaces. These sensor inputs can be used to
connect irradiance sensors as pyranometers, reference cells and PV
module temperature sensors. So, these inverters include monitor-
ing capabilities for irradiance and temperature as well as for max-
imum power point tracking (MPPT) evolution.
The fault detection procedure presented in this work is based on
previous works that evaluate the power losses present in the PV
system for fault detection [3–6,20]. In our approach the computa-
tional analysis has been reduced avoiding the use of electrical
models for simulation of the PV system behavior and the number
of monitoring sensors minimized, just an irradiance sensor and a
temperature sensor are needed to supervise each sub-array con-
nected to each inverter present in the PV system. The measure-
ment of output current and voltage of the PV array is carried out
by the inverter. The main idea is that this method of fault detection
can be integrated into the inverter without using simulation soft-
ware or additional external hardware. Two indicators of current

http://dx.doi.org/10.1016/j.enconman.2014.05.008
0196-8904/Ó 2014 Elsevier Ltd. All rights reserved.
242 S. Silvestre et al. / Energy Conversion and Management 86 (2014) 241–249

and voltage are used as benchmarks and both relationships can be Considering a PV array composed of Np parallel branches of PV
calculated by the inverter himself. modules, including a number of Ns PV modules in series per
branch, Eqs. (5) and (6) can be scaled to calculate the Voc and Isc
of the entire PV array as follows:
2. Theory/calculation
   
Iscmr dIscm
2.1. Current and voltage indicators for fault detection
Isc ¼ N p Gþ ðT cell  T r Þ ð7Þ
1000 dT
    
Chouder and Silvestre [3] presented a diagnosis method for PV dVocm Iscm
systems based in the evaluation of the power losses present in the V oc ¼ N s V ocmr þ ðT cell  T r Þ þ V T ln ð8Þ
dT Iscmr
PV system. In this work the fault detection procedure is based in
the current and voltage ratios given by the following equations: Then the ratios NRc and NRv defined by Eqs. (3) and (4) can be
estimated from Eqs. (7) and (8) once the coordinates of the MPP
IPV sim of the PV array are known.
Rc ¼ ð1Þ
IPV meas On the other hand, we define the ratios NRco and NRvo as the
expected values of NRc and NRv, in normal operation of the PV sys-
V PV sim tem (fault-free). These indicators are given by:
Rv ¼ ð2Þ
V PV meas
Imo
NRco ¼ ð9Þ
where IPV_meas and VPV_meas are the current and voltage measured at Isc
the output of the PV array respectively, and IPV_sim and VPV_sim the
results obtained for those parameters in the simulation of the PV V mo
NRv o ¼ ð10Þ
system behavior by using real irradiance and temperature moni- V oc
tored profiles. where Imo and Vmo are the current and voltage at the maximum
This method was successfully experimentally evaluated and power point of the output of the PV array in absence of faults.
allows identifying main faults in grid connected PV systems. How- The values of Imo and Vmo for a PV array of arbitrary series–par-
ever, this method requires the use of simulation tools that imply allel (Ns  Np) connection of PV modules can be calculated for any
important computational effort to evaluate the losses in the PV condition of G and Tcell by using the following equations that
system and also to calculate the current and voltage ratios. In order include parameters of the PV modules forming the PV array [21]:
to completely avoid the use of modeling and simulation of the PV    
system in the fault detection procedure, we defined two new cur- Immr dIscm
Imo ¼ Np Gþ ðT cell  T r Þ ð11Þ
rent, NRc, and voltage, NRv, indicators as follow: 1000 dT
   
NRc ¼
Im
ð3Þ Iscm  Imm  NVscocm
Isc V mo ¼ Ns Nsc V T ln 1 þ e VT  1  Imm Rsm ð12Þ
Iscm

Vm where a value of n = 1 is used for the diode ideality factor, Immr and
NRv ¼ ð4Þ Tr are the current of the PV module at the maximum power point
V oc
and the temperature at standard test conditions respectively, Nsc
where Vm and Im are the voltage and current of the maximum power is the number of solar cells in series forming the PV module, Rsm
point (MPP) at the PV array DC output respectively, Isc is the short is the series resistance of the PV module, Iscm and Vocm are the short
circuit current of the PV array and Voc the open circuit voltage of circuit current and open circuit voltage of the PV module given by
the PV array. Eqs. (5) and (6), and Imm is the maximum current of the PV module
The coordinates of the MPP, Im and Vm, are available at the given by:
inverter input, considering that the inverter includes MPPT, which  
Immr dIscm
is the case of most inverters used in grid connected applications. Imm ¼ Gþ ðT cell  T r Þ ð13Þ
On the other hand, for an arbitrary value of the irradiance and tem- 1000 dT
perature, the short circuit current of a PV module, Iscm, is given by In normal operation of the PV system, fault-free operation, the
[21]: values of the indicators NRc and NRv should be very similar to the
  values of NRco and NRvo given by Eqs. (9) and (10) and maintain
Iscmr dIscm
Iscm ¼ Gþ ðT cell  T r Þ ð5Þ values over specific thresholds. The thresholds that confirm a
1000 dT
fault-free operation of the system must be fixed depending on
where Iscmr and Tr are the short circuit current of the PV module and the specific characteristics of the PV array, mainly the configura-
the temperature at standard test conditions (STC: G = 1000 W/m2 tion of series and parallel connection of the PV modules forming
and T = 25 °C) respectively, G is the actual irradiance on the PV mod- part of the array. When one of the values of the indicators, NRc
ule and Tcell is the operating cell temperature. or NRv, is below the threshold, a fault is detected in the PV system.
The open circuit voltage of the PV module, Vocm, can be written Table 1 shows the most probable faults present in the PV system
as: based on the values of the ratios NRc and NRv.
    Table 1
dV ocm Iscm
V ocm ¼ V ocmr þ ðT cell  T r Þ þ V T ln ð6Þ Possible faults based on the values of the ratios NRc and NRv.
dT Iscmr
Possible fault NRc NRv
where Vocmr is the open circuit voltage of the PV module at STC and No fault OK OK
VT the thermal voltage. Faulty string Below threshold OK
The values of Iscm and Vocm can be easily estimated for any con- Bypassed modules in a string OK Below threshold
dition of temperature and irradiance, by using Eqs. (5) and (6) tak- Other faults: Below threshold Below threshold
Partial shadow,
ing into account the PV module parameters given by
aging, grid fault. . .
manufacturers at STC.
 S. Silvestre et al. / Energy Conversion and Management 86 (2014) 241–249
When a fault is detected in the PV system an internal data log-
ger interface of the inverter can be used for the transmission of the
measured data and the alarm event to a server or a local network
through a standard RS485, an Ethernet connection or optionally
with a GSM Modem.
On the other hand, the inverter requires a minimum input volt-
age (start-up voltage) to start working. A minimum level of irradi-
ance on the PV array is necessary to enable the proper operation of
the inverter. Taking into account this fact we considered a mini-
mum level of G = 200 W/m2 to start the fault detection evaluation
procedure in the PV system and calculate the corresponding cur-
rent and voltage indicators.
The distributed maximum power point tracking (DMPPT) sys-
tems optimize the energy generated by the PV system by reducing
energy losses caused by shadowing and mismatch effects respect
to conventional central maximum power point tracking (MPPT)
PV systems. In these systems the monitoring of all parameters is
done at module level. So, all input parameters used in our work
are then available at module level. In these cases the proposed fault
detection algorithm could be able to detect the fault and also to
identifying PV module causing the fault. A possible improvement
of the proposed algorithm could be to include the evaluation and
permanent tracking of the optimal range of the bulk inverter volt-
age for a proper operation of DMPPT systems [22] optimizing the
energy harvesting in these kind of PV systems.
2.2. Analysis of the thresholds for indicators of voltage and current
Two of the most common faults in PV systems are due to the
appearance of short circuits and open circuits in PV modules form-
ing part of the strings. When an open circuit appears in one string
of the PV array the whole string is disconnected and then the out-
put current of the PV array is reduced. On the other hand, when
bypass diodes are activated as effect of some solar cells working
in dark conditions, these PV modules are bypassed and the output
voltage of the PV array is reduced [23]. The thresholds for current
and voltage indicators can be fixed attending to these two situa-
tions: a faulty string in the PV array and one PV module bypassed
in a string.
2.2.1. Faulty string
A PV array is formed by connection of a number of Np strings of
PV modules, having a number of Ns PV modules connected in series
per string. In fault-free operation the output current of the PV array
at the MPP, Im, is the product of Np and the maximum output cur-
rent of one of the PV modules, Imm. So, Eq. (10) can be rewritten as
follows:
Np Imm
NRco ¼
Isc
When an open circuit is present in one of the strings, the reduc-
tion in the output current of the PV array is equal to the current of
the faulty string, Imm. Then, in case of a faulty string the value of the
ratio of current, NRcfs, will be lower than its value in free fault oper-
ation mode NRco. The expected value of NRcfs can be calculated as
the following equation shows:
ðNp  1ÞImm Np Imm
NRcfs ¼ ¼a
Isc Isc
where a is the relation between the ratios of current in faulty string,
NRcfs, and fault-free operation, NRco, given by:
NRcfs
a¼
NRco
1 using this base model has been implemented in different software
a¼1
Np
243
Taking into account the above Eqs. (15)–(17), the threshold for
the ratio of current in case of fault string, TNRcfs, can be evaluated.
When one or more faulty strings are present in the PV system the
value of NRc must be below the value indicated by TNRcfs to allow
the identification of this kind of fault. In order to avoid false fault
detections we have fixed an offset of a 2% respect the NRco value.
Then, the threshold for the ratio of current can be written as
follows:
TNRcfs ¼ 1:02 a NRco ð18Þ
2.2.2. Bypassed PV module
In fault-free operation the output voltage of the PV array at the
MPP, Vm, is the product of Ns and the maximum output voltage of
one of the PV modules, Vmm. Then, Eq. (10) can be rewritten as
follows:
Ns V mm
NRv o ¼ ð19Þ
V oc
When a PV module is bypassed in one of the strings the new
value of the voltage ratio, NRvbm, will be lower than its value in
fault-free operation mode, NRvo. Considering that the voltage drop
across the by-pass diode can be neglected, the expected value of
NRvbm in that situation can be calculated using the following
equation:
ðNs  1ÞV mm Ns V mm
NRv bm ¼ ¼b ð20Þ
V oc V oc
where b is the relation between the voltage ratios in case of one
bypassed PV module and fault-free operation, given by:
NRv bm
b¼ ð21Þ
NRv o
1
b¼1 ð22Þ
Ns
The threshold for the voltage ratio in case of one bypassed PV
module in a string, TNRvbm, can be evaluated following the same
criteria that we used in the previous case of faulty string as
follows:
TNRv bm ¼ 1:02 b NRv o ð23Þ
When one or more PV modules are bypassed in some of the
strings of the PV system, the value of NRvbm will be below the value
indicated by TNRvbm given by Eq. (23), allowing the identification of
this kind of fault.
The expressions obtained for a and b in Eqs. (17) and (22)
ð14Þ respectively, show that both parameters do not depend on the
PV module parameters neither of the total power of the PV array.
Both parameters depend only on the array configuration, Ns and
Np. If the PV array configuration is known a and b can be evaluated
and then the thresholds for the current, TNRcfs, and voltage, TNRvbm,
indicators that evidence the presence of faults in the PV system can
be fixed by using Eqs. (18) and (23).
2.3. Simulation study
ð15Þ
A simulation study was carried out to analyze the potential of
the new current and voltage indicators in fault detection as well
as the effectiveness of the values estimated in the above section
for parameters a, b, TNRvbm and TNRcfs.
ð16Þ
The simulation of the PV system is based in the well-known five
parameter model for the solar cell. The simulation of PV systems
ð17Þ
environments as: Matlab [3,24,25], Pspice [20,21,26,27] or
244 S. Silvestre et al. / Energy Conversion and Management 86 (2014) 241–249

Table 2
Main parameters of commercial PV modules included in the simulation study.

Model parameter Isofoton 106W-12 BP Helios Siemens

5160S 1500 SP150
n 1.14 1.12 1.2 1.13
Io (A) 3  1010 1.56  109 2.83  108 2.6  109
Rs (X) 0.33 0.587 0.182 0.932
Rsh (X) 199 1946 39.21 348.2
Isc (A) 6.54 4.7 7.88 4.8
Solar cells (Ns  Np) 36  2 72  1 36  1 72  1
Voc (V) 21.6 44 21 43.4
P (W) 106 150 120 150
Vcoeff. (mV/K) 144.7 160 90 77
Icoeff. (mA/K) 2.5 3.05 0.2 2.06

LabView [8,28], and experimentally validated. In this study, we

Table 5
have selected the Matlab environment for the simulations set.
Values obtained for parameters a and b.
The inputs for the simulation of the PV system are the main
parameters of the solar cell model: The cell photocurrent (Isc), Case study a (%) b (%)
the diode saturation current (Io), the ideality factor of the diode 1  (1/Np) NRcfs/NRco 1  (1/Ns) NRvbm/NRvo
(n), the series resistance (Rs) and the shunt resistance (Rsh). Other 0 50.00 50.00 80.00 79.95
parameters included as inputs are: The number of solar cells per 1 50.00 50.00 93.33 93.33
PV module (Nsc, Npc), the configuration of the PV array: Ns and Np 2 50.00 49.99 93.33 93.33
3 50.00 50.00 95.45 95.44
and the data corresponding to the irradiance (G) and temperature
4 66.66 66.66 93.33 92.32
profiles. 5 88.88 88.89 90.90 90.90
Table 2 shows the commercial PV modules selected in this sim- 6 83.33 83.34 93.33 93.88
ulation study and their main parameters at STC. The PV modules: 7 75.00 74.99 97.50 97.49
Isofoton 106W-12 and Helios 1500 have open circuit voltages 8 92.85 92.85 90.00 89.99

around 21 V and short circuit currents over 6 A, while PV modules:

BP 5160S and Siemens SP150 have higher open circuit voltages, per branch (Ns) and number of parallel branches (Np), were studied
around 44 V, and lower short circuit currents, around 4.7 A. These to analyze if the size of the PV array affects the values of the
differences in open circuit voltage and short circuit current were thresholds for the current and voltage indicators when a fault
selected to study the influence of the PV module parameters in appears in the PV array.
the thresholds of the current and voltage indicators when a fault Table 3 describes the different PV arrays included in the simu-
appears in the PV array. lation study as well as the PV modules forming part of the array
On the other hand, PV arrays with DC output powers ranging and the configuration of the PV array: Ns  Np in each of the case
from 1 kW up to 21 kW, including different types of the selected studies. These PV arrays have been simulated, considering the
PV modules and different configurations of PV modules in series same set of real irradiance and temperature profiles, along a day
in three cases: Normal operation (no faults present in the PV sys-
tem), faulty string (one string in open circuit) and bypassed mod-
Table 3 ule (one PV module bypassed in one string). The evolutions of the
Description of the PV arrays included in the simulation study. indicators NRco, NRvo, NRcfs and NRvbm have been calculated in each
Case study PV module Ns Np P (kW) one of the simulations carried out as well as mean values and stan-
dard deviations.
0 Isofoton 106W-12 5 2 1.06
1 Helios 1500 15 2 3.3 Table 4 shows the average values obtained for those indicators
2 BP5160S 15 2 4.5 and their standard deviations. The values of the thresholds: TNRcfs
3 Helios 1500 22 2 5.28 and TNRvbm calculated by using Eqs. (18) and (23) are also given in
4 BP5160S 15 3 7.2
Table 4.
5 Isofoton 106W-12 11 9 10.5
6 SP150 15 6 15
As can be seen in Table 4, in case of normal operation the values
7 Helios 1500 40 4 19.2 obtained for the voltage ratio, NRvo, vary from 77.89% to 81.05%,
8 BP5160S 10 14 21 while for the ratio of current, NRco, these values vary from 85.20%
to 92.23%.

Table 4
Values obtained in the simulations for the ratios and thresholds of current and voltage.

Case study Normal operation (No faults) Faulty string Bypassed PV module
NRco (%) r NRco NRvo (%) r NRvo NRcfs (%) r NRcfs TNRcfs NRvbm (%) r NRvbm TNRvbm
0 90.78 4.70  103 81.05 1.12  102 45.39 2.31  103 46.29 64.80 8.41  103 66.13
1 85.20 2.21  102 77.95 1.46  102 42.60 1.04  102 43.45 72.80 1.31  102 74.20
2 92.23 2.99  103 80.03 1.30  102 46.11 1.44  103 47.03 74.70 1.30  102 76.18
3 85.20 2.05  102 77.95 1.47  102 42.60 1.03  102 43.45 74.40 1.30  102 75.89
4 92.23 2.99  103 80.03 1.38  102 61.48 1.39  103 62.71 74.69 1.31  102 76.18
5 90.78 4.63  103 81.05 1.05  102 80.69 4.11  103 82.30 73.68 9.56  103 75.15
6 92.23 2.93  103 79.63 1.45  102 76.86 2.43  103 78.39 74.76 1.35  102 75.80
7 85.46 1.90  102 77.89 1.49  102 64.09 1.43  102 65.37 75.94 1.41  102 77.46
8 92.23 2.92  103 79.98 1.44  102 85.64 2.72  103 87.35 71.98 1.30  102 73.42
 S. Silvestre et al. / Energy Conversion and Management 86 (2014) 241–249 245

In case of a faulty string in the PV system the values obtained of

the ratio of current, NRcfs, are lower than the values of NRco and
below the thresholds given by TRNcfs in all cases.
When a PV module is bypassed in one string the results
obtained for the voltage indicator, NRvbm, are also lower than the
values obtained for NRvo and below the thresholds, TNRvbm, in all
cases.
We have calculated the values of parameters a and b in all cases
included in this simulation study. First the values of both parame-
ters were evaluated using Eqs. (17) and (22) and the values of Ns
and Np shown in Table 3, then we used Eqs. (16) and (21) and
results shown in Table 4 for: NRco, NRvo, NRvbm and NRcfs. A compar-
ison of the set of values obtained for a and b by using both proce-
dures is shown in Table 5.
As can be seen in Table 5, a good agreement between values
Fig. 1. a parameter values as a function of Np.
obtained from the two procedures for both parameters is achieved.
These results are in accordance with the assumption that those
parameters depend only on the array configuration, Ns and Np,
and can be evaluated by using Eqs. (17) and (21).
Fig. 1 shows the values obtained for the parameter a as a func-
tion of Np. As can be seen in the figure the values of a vary from 50%
in case of a PV array having 2 strings to 92.85% for a PV array
formed by 14 strings.
On the other hand, Fig. 2 shows the values obtained for the
parameter b, the minimum threshold as a function of Ns. The values
obtained for b are ranging from 80% for a PV system formed by
strings of 15 PV modules connected in series to 97.5% for a PV sys-
tem formed by strings containing 40 PV modules in series.

3. Experimental validation

Fig. 2. b parameter values as a function of Ns. The procedure of fault detection in grid connected PV systems
based on the current and voltage indicators described in the previ-
ous sections was tested in a grid connected PV system located in

Fig. 3. Monitored irradiance profile in fault-free operation.

Fig. 4. Monitored cell temperature profile in fault-free operation.

246 S. Silvestre et al. / Energy Conversion and Management 86 (2014) 241–249

Fig. 5. Indicators of current in fault-free operation.

Fig. 6. Indicators of voltage in fault-free operation.

Table 6
Average values and standard deviations for current and voltage indicators measured
in case of fault-free operation.

PV array normal Parameter values Standard deviation

operation (mean) (r)
NRvo (%) 81.05 10.51  103
NRv (%) 82.23 17.94  103
TNRvbm 78.11 10.20  103
NRco (%) 90.77 4.65  103
NRc (%) 89.56 39.65  103
TNRcfs 46.41 23.74  103
Fig. 8. Monitored temperature profile for PV system in case of faulty string.

The PV system includes a monitoring system using an Agilent

34970A for the data acquisition as well as two pyranometers (Kipp
& Zonen CM 11 type) and a reference solar cell to measure irradi-
ance at different planes. For the measurement of the temperature
the monitoring system includes k type thermocouples. A more
detailed description of this monitoring system can be found else-
where [7].

Fig. 7. Monitored irradiance profile for PV system in case of faulty string.

3.1. PV Array in normal operation (fault-free operation)

the Centre de Développement des Energies Renouvelables (CDER),
Algerie. This PV system of 9 kWp is divided in three sub-arrays of
3 kWp each one, which are connected to 2.5 kW (IG30 Fronius) sin-
gle phase inverters. Each sub-array is formed by 90 PV modules
(Isofoton 106 W–12 V) in a configuration of two parallel strings,
Np = 2, of 15 PV modules in series, Ns = 15.
The values of parameters a and b for the sub-array calculated
using Eqs. (17) and (21) are 50% and 93.33% respectively.
A sub-array of the PV system described above was monitored
along a day in fault-free operation. Figs. 3 and 4 show respectively
the irradiance and temperature profiles monitored at the PV instal-
lation along that day. The evolution of Voc and Isc of the PV array
was calculated by using Eqs. (7) and (8) and data shown in Figs. 3
and 4.
Then the current, NRco, and voltage, NRvo, indicators were calcu-
lated according to Eqs. (9) and (10) and taking into account the
 S. Silvestre et al. / Energy Conversion and Management 86 (2014) 241–249 247

Fig. 9. Current indicators: NRc and NRco, and threshold of current: TNRcfs and in case of one faulty string.

The evolution of the voltage and current indicators was calcu-

lated following the same procedure than in the case of fault-free
Table 7 operation described above using monitored data. Fig. 9 shows
Average values and standard deviations for current and voltage indicators measured the evolution of NRc, NRco as well as the evolution of TNRcfs, calcu-
in case of faulty string. lated using Eq. (13). The values for NRco are higher than the values
PV array faulty string Parameter values (mean) Standard deviation (r) of NRc that is below the threshold given by TNRcfs, as can be seen in
NRvo (%) 81.85 10.52  103 Table 7. This situation indicates the fault of a string in the PV array
NRv (%) 83.59 28.48  103 as table 2 shows.
TNRvbm 77.87 10.02  103 On the other hand the voltage indicators show a good behavior
NRco (%) 90.99 7.43  103 in Fig. 10. The values of NRv and NRvo are of the same order and the
NRc (%) 45.34 2.06  103
TNRcfs 46.82 3.80  103
threshold TRNvbm presents a lower value than the voltage ratios.
Mean values and standard deviations are given in Table 7.
The combination of the evolution of current and voltage ratios
monitored profiles of the voltage and current at the maximum
results in the detection of a faulty string in the PV array.
power point (MPP) of the PV sub-array DC output. Figs. 5 and 6
show the evolution of the indicators of current and voltage respec-
tively. As can be seen in the figures, NRc and NRv are over the cor- 3.3. Bypassed PV module
responding thresholds: TNRcfs and TRNvbm, indicating that the PV
system is in case of fault-free operation. The activation of by-pass diodes was simulated by manually
Finally, Table 6 shows the mean (average) values and standard jumping the PV module with cables. One of the PV modules form-
deviations obtained for these parameters in this case of fault-free ing part of a string of the sub-array was bypassed and then the
operation. The mean values of the indicators of current are around
90% and 81% in the case of voltage. In both cases the thresholds:
NTRcfs and NTRvbm show mean values below the corresponding
indicators.

3.2. Faulty string

One of the strings of the sub-array was disconnected from the

system and then the evolution of the MPP, the corresponding volt-
age and current, was monitored as well as the irradiance and cell
temperature for the whole day. Figs. 7 and 8 show the irradiance Fig. 11. Monitored irradiance profile for PV system in case of one bypassed diode in
and temperature profiles monitored respectively. one string.

Fig. 10. Voltage indicators in case of one faulty string.

248 S. Silvestre et al. / Energy Conversion and Management 86 (2014) 241–249

Fig. 12. Monitored temperature profile for PV system in case of one bypassed diode in one string.

Fig. 13. Voltage indicators: NRv and NRvo, and voltage threshold: TNRvbm in case of one bypassed diode in one string.

Fig. 14. Indicators of current in case of one bypassed diode in one string.

evolution of the MPP, voltage and current, was monitored as well Table 8
as the irradiance and cell temperature along the experiment. Average values and standard deviations for current and voltage indicators measured
in case of one bypassed module in a string.
Figs. 11 and 12 show the corresponding irradiance and tempera-
ture profiles monitored. PV array Parameter values Standard deviation
The evolution of the current and voltage was calculated follow- One bypassed PV module (mean) (r)
ing the same procedure described in previous sections from the NRvo (%) 81.81 10.61  103
monitored data. Fig. 13 shows the evolution of NRv as well as the NRv (%) 74.65 12.02  103
TNRvbm 77.88 10.1  103
evolution of NRvo and TNRvbm, calculated using Eq. (18). The evolu-
NRco (%) 91.28 4.81  103
tion of the ratios of current NRc and NRco as well as the threshold NRc (%) 90.15 37.77  103
TRNcfs are given in Fig. 14. TNRcfs 46.55 2.45  103
Mean values obtained for current and voltage indicators and
thresholds, and standard deviations are given in Table 8.
As can be seen in Fig. 13 the values of NRv are below the thresh- Table 8. On the other hand, the evolution of the indicator of current
old indicated by TNRvbm showing a fault present in the PV array. NRc shows a similar behavior to the case of fault-free operation and
The main values obtained for these parameters are shown in remains over the threshold TRNcfs. Main values around 91% have
 S. Silvestre et al. / Energy Conversion and Management 86 (2014) 241–249
been obtained for NRc and NRco while the mean value of TRNcfs is
46.55%, as can be seen in Table 8.
From the results obtained for NRc and NRv the fault detected in
this experiment corresponds to one bypassed module in a string of
the array as expected.
4. Conclusions
In this work we have presented a new procedure for automatic
detection of main faults in PV systems. The method is based on the
evaluation of current and voltage indicators and in the definition of
thresholds for both parameters that can help to identify the system
faults. The definition of the thresholds is determined by the num-
ber of PV modules forming the PV array and its configuration in
parallel and series connection.
A simulation study was used to verify the definition of the cur-
rent and voltage indicators as well as the definition of the corre-
sponding thresholds for fault detection.
Finally we have verified the fault detection procedure experi-
mentally in real conditions of work of a grid connected PV system.
The results obtained were the expected in all cases and faults pres-
ent in the PV system were detected by using this method.
The proposed method is simple but effective while considering
the minimum sensors and minimizes the monitoring and supervi-
sion system, which can be included in the inverter. Moreover, the
supervision of the PV system could be carried out in real time by
the inverter itself.
References
[1] Yagi Y, Kishi H, Hagihara R, Tanaka T, Kozuma S, Ishida T, et al. Diagnostic
technology and an expert system for photovoltaic systems using the learning
method. Sol Energy Mater Sol Cells 2003;75:655–63.
[2] Chao KH, Hob SH, Wang MH. Modeling and fault diagnosis of a photovoltaic
system. Electric Power Syst Res 2008;78:97–105.
[3] Chouder A, Silvestre S. Automatic supervision and fault detection of PV
systems based on power losses analysis. Energy Convers Manage
2010;51:1929–37.
[4] Gokmen N, Karatepe E, Celik B, Silvestre S. Simple diagnostic approach for
determining of faulted PV modules in string based PV arrays. Sol Energy
2012;86:3364–77.
[5] Silvestre S, Chouder A, Karatepe E. Automatic fault detection in grid connected
PV systems. Sol Energy 2013;2013(94):119–27.
[6] Trillo-Montero D, Santiago I, Luna-Rodriguez JJ, Real-Calvo R. Development of a
software application to evaluate the performance and energy losses of grid-
connected photovoltaic systems. Energy Convers Manage 2014;81:144–59.
[7] Forero N, Hernández J, Gordillo G. Development of a monitoring system for a
PV solar plant. Energy Convers Manage 2006;47:15–6.
View publication stats
249
[8] Chouder A, Silvestre S, Taghezouit B, Karatepe E. Monitoring, modeling and
simulation of PV systems using LabVIEW. Sol Energy 2013;91:337–49.
[9] Ducange P, Fazzolari M, Lazzerini B, Marcelloni F. An Intelligent system for
detecting faults in photovoltaic fields. In: Proc. of the 11th international
conference on intelligent systems design and applications (ISDA), 2011, p.
1341–6.
[10] Mellit A, Kalogirou SA. ANFIS-based modelling for photovoltaic power supply
system: a case study. Renew Energy 2011;36:250–8.
[11] Syafaruddin, Karatepe E, Hiyama T. Controlling of artificial neural network for
fault diagnosis of photovoltaic array. In: IEEE 16th international conference on
intelligent system application to power systems, 2011, p. 1–6.
[12] Vergura S, Acciani G, Amoruso V, Patrono GE, Vacca F. Descriptive and
inferential statistics for supervising and monitoring the operation of PV plants.
IEEE Trans Ind Electronics 2009;56:4456–63.
[13] Muselli M, Notton G, Canaletti JL, Louche A. Utilization of meteosat satellite-
derived radiation data for integration of autonomous photovoltaic solar energy
systems in remote areas. Energy Convers Manage 1998;39:119.
[14] Stettler S, Toggweiler P, Remund J. Spyce: satellite photovoltaic yield control
and evaluation. In: Proc. of the 21st European photovoltaic solar energy
conference, 2006, p. 2613–2616.
[15] Drews A, de Keizer AC, Beyer HG, Lorenz E, Betcke J, Van Sark WGJHM, et al.
Monitoring and remote failure detection of grid- connected PV systems based
on satellite observations. Sol Energy 2007;81:548–64.
[16] Fernández-Pacheco DG, Molina-Martínez JM, Ruiz-Canales A, Jiménez M. A
new mobile application for maintenance tasks in photovoltaic installations by
using GPS data. Energy Convers Manage 2012;57:79–85.
[17] Tadj Mohammed, Benmouiza Khalil, Cheknane Ali, Silvestre Santiago.
Improving the performance of PV systems by faults detection using GISTEL
approach. Energy Convers Manage 2014;80:298–304.
[18] Quesada B, Sánchez C, Cañada J, Royo R, Payá J. Experimental results and
simulation with TRNSYS of a 7.2 kWp grid-connected photovoltaic system.
Appl Energy 2011;88:1772–83.
[19] Hernandez J, Gordillo G, Vallejo W. Predicting the behavior of a grid-
connected photovoltaic system from measurements of solar radiation and
ambient temperature. Appl Energy 2013;104:527–37.
[20] Chine W, Mellit A, Pavan AM, Kalogirou SA. Fault detection method for grid-
connected photovoltaic plants. Renew Energy 2014;66:99–110.
[21] Castañer L, Silvestre S. Modelling photovoltaic systems using Pspice. Wiley;
2002.
[22] Vitelli M. On the necessity of joint adoption of both distributed maximum
power point tracking and central maximum power point tracking in PV
systems. Prog Photovoltaics Res Appl 2014;22:283–99.
[23] Silvestre S, Boronat A, Chouder A. Study of bypass diodes configuration on PV
modules. Appl Energy 2009;86:1632–40.
[24] Tsai HL. Insolation-oriented model of photovoltaic module using Matlab/
Simulink. Sol Energy 2010;84:1318–26.
[25] Chouder A, Silvestre S, Sadaoui N, Rahmani L. Modeling and simulation of a
grid connected PV system based on the evaluation of main PV module
parameters. Simul Model Pract Theory 2012;20:46–58.
[26] Moreno A, Julve J, Silvestre S, Castañer L. Spice macromodelling of photovoltaic
systems. Prog Photovolt: Res Appl 2000;8:293306.
[27] Di Vincenzo MC, Infield D. Detailed PV array model for non-uniform irradiance
and its validation against experimental data. Sol Energy 2013;97:314–31.
[28] Yu TC, Lin YB, Chang FS. Establishment of the photovoltaic simulation system
using mixed programming with LabVIEW and simulink. In: Intelligent
technologies and engineering systems. Springer 2013, p. 383–394.