Renewable Energy Focus 41 (2022) 160–178

Contents lists available at ScienceDirect

Renewable Energy Focus

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

Research Paper (blue)

A monitoring system for online fault detection in multiple photovoltaic

arrays
Siva Rama Krishna Madeti
Faculty of Engineering, Department of Electrical and Electronics Engineering, SRKR Engineering College (Autonomous), Bhimavaram, India

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

Article history: In this paper, a low-cost comprehensive Zigbee-based wireless monitoring system with fault detection
Received 17 January 2022 technique is developed for online monitoring of a multiple photovoltaic (PV) array configurations.
Revised 1 March 2022 Different electrical parametric attributes were utilized to investigate the susceptibility of various com-
Accepted 1 March 2022
monly used PV array topologies to partial shade and electrical faults. A prediction model is used to com-
Available online 5 March 2022
pute a number of attributes for a given set of operating conditions - solar irradiance and PV module
temperature. The predicted qualities are then compared to the ones acquired from field measurements,
resulting in the identification of potentially defective operating conditions. In addition, a user-friendly
web application is being built for quick Internet access to monitored data. The suggested approach is
meant to have minimal processing needs and hence, a single microcontroller is employed for data trans-
fer and fault detection. The proposed technique has been tested experimentally. The properties are
shown to be significantly dependent on the shading/faults in the arrays and varied for various topologies,
and may be utilized to detect/predict faults. Also, it has been shown that at varied shading levels the par-
allel topology produced higher power. However, with 90% system failure, all the topologies produced
almost identical power.
Ó 2022 Elsevier Ltd. All rights reserved.

Introduction voltage events, avoid islanding, and report on PV condition (e.g.,

current, voltage, power). Traditional over current protection sys-
The need for renewable energy production is rapidly rising, tems, on the other hand, may overlook significant output power
owing to rising electricity consumption and the need to address loss owing to concealed faults or partial shading of PV modules,
environmental problems such as pollution, global warming, and resulting in PV module damage or a violation of safety regulations.
the use of fossil fuels. Solar photovoltaic (PV) is the most promising The study in [5] shows that the power loss due to faults present in
renewable energy source owing to its plentiful source availability, PV modules is up to 18.9%. The output power of a PV system can
simplicity of installation, static, minimal maintenance, ecologically also decrease considerably when mismatch occurs as a result of
benign, and noiseless generating characteristics [1]. PV technology partial shading, soiling and non-uniform irradiation [6]. Partial
has grown rapidly in recent decades as a result of the development shade is the most major cause producing a mismatch among these
of PV technologies and numerous financial incentives granted by elements, and partial shading situations are unavoidable, particu-
government to the electrical energy generating industry. This is larly in PV systems deployed in metropolitan districts and in places
evident by the increase in the cumulative installed PV capacity where low moving clouds are widespread. [7]. When buildings,
(MW) of countries, the use of PV energy requires system analysis trees, or other things in close proximity to PV modules shade them,
and optimization to improve the performances of PV installations partial shading occurs. Because the PV module’s output current is a
[2,3]. These standards emphasize the need of providing effective function of solar irradiance and module temperature, a decrease in
monitoring equipment and tools for PV systems in order to enable solar irradiance will impact the PV module’s performance. The
for data analysis, comparison, identification, and problem detec- change in output power for various PV array layouts during the
tion [4]. irradiance transition induced by moving clouds was examined in
Monitoring systems are often integrated into inverters and are [8]. Manually identifying and solving defects is common, but it is
used to connect and disengage from the grid during low and high not advised since it is time-consuming, incorrect, and possibly dan-
gerous to the operator, particularly in large-scale PV arrays.
Instead, automatic fault monitoring for PV systems that is fast
E-mail address: [email protected] and efficient is required.

https://doi.org/10.1016/j.ref.2022.03.001
1755-0084/Ó 2022 Elsevier Ltd. All rights reserved.
S.R.K. Madeti Renewable Energy Focus 41 (2022) 160–178

Nomenclature

ANN artificial neural network S Standard deviation

BL bridge-link n no. of observations
CSS cascading style sheets C refcell cost of reference cell to measure irradiance
ECM earth capacitance measurement C install labor cost to install PV plant
ESR equivalent series resistance C stor cost of data storage
FER fixed electrical reconfiguration C temp cost of k type thermocouple sensor
G group IHT i ith HT instrument current
GHI global horizontal irradiance Idev i ith developed I
V tracer current
HC honey-comb Immr current at the maximum power point under STC
HTML hypertext markup language Imp current at the maximum PowerPoint
IB inverted bypass diode Isc short-circuit current
IEC international electrotechnical Iscm short-circuit current of PV module
INR Indian rupee Iscmr short-circuit current of PV module at STC
MW megawatt Nsc number of solar cells in series
OC open-circuit Pdccal calculated dc power
OS operating system Pdc PU PV array output dc power
P parallel Pdc meas measured DC power
PHP hypertext pre-processor Pmp power at the maximum PowerPoint
PV photovoltaic Rsm series resistance of PV module
RMSE root mean square error Tr temperature under STC
Rx receive VT thermal voltage
S series V mp voltage at the maximum PowerPoint
SC short-circuit V oc open-circuit voltage
SDS shade dispersion scheme V ocm open-circuit voltage of PV module
SFBD shading with-out bypass diode V scmr open-circuit voltage of PV module at STC
SIBD shading with inverted bypass diode nis number of current sensors
SNBD shading with normal bypass diode nv s number of voltage sensors
SP series-parallel r1 standard deviation of P dc PU
STC standard test conditions r2 standard deviation of module power
TCT total cross-tied r3 standard deviation of module voltage
TDR time-domain reflectometry r4 standard deviation of module current
TFT Thin-film transistor r5 standard deviation of module open circuit voltage
Tx Transmit r6 standard deviation of module short circuit current

no effect " increase
B1 ; B2 circuit breakers # decrease
C X
beemod cost of X-bee modules
# no-change or decrease
C dcisec cost of each DC current sensor #" decrease or increase
A measured data

Many diagnostic techniques have been developed to detect lematic modules, but offline fault diagnostic techniques are unable
faults present in PV systems [9]. Some of these approaches are to provide real-time fault information. In [24] presents a detailed
based on simulations of the PV module electrical circuit [10,11], comparison and analysis of several fault detection techniques.
while others rely on statistical analysis of various PV module out- PV array configurations, on the other hand, are one of the meth-
put terminal electrical characteristics, such as current and voltage ods for reducing mismatch and partial shading losses in PV plants.
[12]. In [13] proposes using earth capacitance measurement (ECM) Series (S), Parallel (P), Series-Parallel (SP), and Total-Cross-Tied
to detect PV module separation in a string. Without needing cli- (TCT) are the various PV array topologies evaluated in this study
matic data, the time-domain reflectometry (TDR) approach was (TCT). Several studies are analyzing the impact of shadowing on
presented in [14], which monitors the electrical properties of various PV array configurations in order to decrease mismatch
transmission lines to identify failure points, defects, and impe- losses and offer maximum output power production have been
dance change owing to deterioration. These approaches, on the developed in the past [25]. In [26] describes a comparison of sev-
other hand, can only be used in offline mode. In practice, online eral PV array configurations using MATLAB/Simulink software.
diagnostic approaches that can capture measurements while the The results reveals that the TCT design delivers the best power pro-
tested item is in use are greatly desired. Based on power loss anal- duction under most shade circumstances. In [27,28] provide a
ysis, [15] proposes automated supervision and fault detection. mathematical study of the TCT configuration, as well as compar-
However, it is unable to locate the malfunctioning module. Based isons to other configurations such as S, P, SP, bridge-linked, and
on the PV string operation voltage and the ambient temperature, honey-comb. It was concluded that the TCT arrangement outper-
an operating voltage-window is then produced [16]. It can discover forms the other PV array options. [29-31] presents a detailed
open and short faults, but it can’t find the problematic module in assessment of different PV array topologies under mismatch or
the array. The method proposed in [17-19] uses statistical meth- partial shade. A novel approach was presented in [32] to increase
ods, whereas optimization techniques are used in [20-23]. Many the output power from a PV array by uniformly distributing the
defect diagnostic approaches have been developed thus far, both shade impact in each PV row. Another technique [33] proposes
online and offline. However, online fault diagnosis approaches to enhance the output of PV arrays under partial shadowing situa-
have a fundamental drawback in that they are unable to find prob- tions by changing the placement of PV modules linked in TCT con-

161
S.R.K. Madeti Renewable Energy Focus 41 (2022) 160–178

figuration and PV arrays connected in ‘‘Su Do Ku” puzzle pattern. module level, dedicated paradigms are utilized to examine changes
However, the solutions presented in [32,33] suffer from a large in output powers as well as to estimate likely reasons of PV system
increase in wiring needs as well as inefficient shadow dispersion. functioning perturbation. Furthermore, the created approach is
Several data analyses on the attribute’s change among partial totally self-contained and low-energy, i.e., it is not reliant on com-
shade and defective PV situations applied to diverse PV array puter or grid power. Various partial shade and defective PV situa-
topologies have been published in the literature. As a result, the tions were evaluated in order to compare the performance of each
article’s major contribution is to propose a novel monitoring sys- PV array arrangement. Under extreme environmental circum-
tem with a fault detection approach for the isolation and identifi- stances, the suggested method has been experimentally verified.
cation of defects in a PV system. As a replacement for the Finally, all of the essential experimental findings have been pre-
traditional cable-based monitoring system, the designed monitor- sented to illustrate that the proposed defect detection approach
ing system employs an X-based wireless communication network is efficient.
for data transfer. This system makes use of a multi-sensor design The structure of this paper is as follows: Section 2 describes the
with voltage and current sensors installed in each module. At the PV plant layout as well as the proposed monitoring. In Section 3,

Figure 1. PV plant layout with monitoring system.

162
S.R.K. Madeti Renewable Energy Focus 41 (2022) 160–178

the proposed fault detection methods for multiple PV array config- ies with other string of cells (groups). Thus, the PV module consists
urations are presented. Section 4 contains the experimental results, of 3 groups of cells G1 ; G2 and G3 as depicted in Figure 2. The main
uncertainty analysis of measuring instrument, and discussions. components used in this experimental circuit are:
Section 5 has a cost analysis. Finally, Section 6 outlines the
conclusion. (a) ATMega8535 microcontroller;
(b) X-Bee-Pro Series 1 modules;
(c) K-type thermocouple and reference cell are used to measure
Materials and methods the module temperature and Global Horizontal Irradiance
(GHI) irradiance respectively;
System architecture (d) Voltage divider is used for voltage measurement;
(e) Hall-effect based sensor is used for current measurement;
The PV system under study is a 1 kWp PV system located on 2nd
floor of Electrical and Electronics Engineering (EEE) department, a The ATMega8535 microcontroller collects data to determine the
building owned by Velagapudi Ramakrishna Siddhartha Engineer- PV characteristic. The main functions of the controller are (i) con-
ing College (VRSEC), India. Figure 1 illustrates the overall architec- trol breakers ðB1 ; B2 Þ; (ii) reading data from sensors; (iii) display
ture of the PV system, as well as available devices and sensors on module characteristic with set of attributes data on Thin Film Tran-
each location. The PV modules are located on the 2nd floor of EEE sistor (TFT); and (iv) transfer data. The PV module output voltage
building, while the developed main server is on the 1st floor. measurement is performed by a simple resistive voltage divider
Because various parts of the system are located at different places and the output current of PV module is measured and amplified
because of this reason wireless monitoring system is preferred using LA 55-P Hall-Effect sensor. HT304N reference cell is used to
over wired system. sense the solar radiation. The temperature measurement is carried
The core elements of this system are low cost I
V curve tracer, out using a numerical sensor (Type-K thermocouple) with a resolu-
data transfer mechanism and monitoring method. The developed tion of 0.25°C.
I
V curve tracer is installed on the rear side of each PV module An external electrolytic capacitor with low Equivalent Series
to measure important operational parameters such as voltage at Resistance ðESRÞ is used for I
V curve tracing. In the experimental
maximum power point ðV mp Þ, current at maximum power point setup, capacitor is connected to PV module through breaker B1 and
ðImp Þ, power at maximum power point ðP mp Þ, open circuit voltage to discharge the capacitor after the test, it is connected to a dis-
ðV oc Þ and short circuit current ðIsc Þ. Each module is checked for charge resistor via breaker B2 . The length of wires in a capacitor
any abnormalities based on the measurements. charging method is limited to a few meters to make the voltage
drop negligible. For easy power conditioning and low voltage drop
Experimental I
V curve tracer for PV modules (<1% of rated voltage) in the power wires the measuring circuit
which connects the capacitor to terminals (positive and negative)
Figure 2 depicts the layout of PV system with capacitor charging of PV module are placed close to each other. The important opera-
based I
V curve tracing method [34]. A 100 Wp PV module is tional parameters such as V mp , Imp , P mp , V oc and Isc are easily
used in the experiment which is composed of 36 series connected obtained from the characteristic I
V curve by the developed tra-
cells. One bypass diode is placed across 12 string of series con- cer. The traced I
V curve is displayed on the TFT display as
nected cells called as a group ðGÞ, which is in turn connected in ser- depicted in Figure 3.

Figure 2. Setup to measure the characteristic curve of the PV module.

163
S.R.K. Madeti Renewable Energy Focus 41 (2022) 160–178

Figure 3. I
V curve traced by the developed tracer and displayed on TFT display.

Theory/calculation of operational parameters for fault detection MPP and the temperature both at standard test conditions ðSTCÞ,
respectively, N sc is the number of solar cells in series forming the
By using simulation tools various diagnosis methods based on PV module, V T is the thermal voltage,Rsm is the series resistance of
the evaluation of expected values of operational parameters for PV module, Iscm and V ocm are the short circuit current and open cir-
PV system have been proposed in the past. The authors in cuit voltage of PV module respectively given by [38]:
[35,36], predicted the expected DC power by using LabVIEW and

Iscmr dIscm
MATLAB simulation model. Actual measured environmental data Iscm ¼ Gþ ðT cell
T r Þ ð3Þ
1000 dT
is used as simulation model inputs for precise prediction of PV sys-
tem yields. Other approaches to calculate the expected DC power


dV scm Iscm
are investigated in [37]. The main disadvantages of the above- V ocm ¼ V ocmr þ ðT cell
T r Þ þ V T ln ð4Þ
mentioned methods are the dependency on a personal computer
dT Ismr
(PC), use of commercial software for simulation, and requirement where Iscmr and V scmr are the short circuit current and open circuit
of grid power supply, which increases the cost of the system and voltage of PV module at STC; respectively. A simulation study was
limited its spread and use. The operational parameters of the PV conducted to determine the efficacy of the estimated values. It is
system are determined analytically to completely eliminate the based on a five-parameter solar cell model. In this work, the funda-
usage of modeling and simulation in a defect detection technique. mental model for PV system modeling proposed by [17] is applied.
The PV module maximum current and voltage (Imp , V mp Þ are calcu- The approximated values produced using (1) and (2) are fairly sim-
lated for an arbitrary value of G and T cell are given by [38]: ilar to the numbers reported in the manufacturer data sheet. For


  any condition of G and T cell , the values of Im and V m for a PV array
Immr dIscm with any series-parallel (N s  N p ) connection of PV modules may
Imp ¼ Gþ ðT cell
T r Þ ð1Þ
1000 dT be determined using (5) and (6) [38].


 


 
Iscm
Imp  NVscocm
Immr dIscm
Im ¼ N p Gþ ðT cell
T r Þ ð5Þ
V mp ¼ Nsc V T ln 1 þ e VT
1
Imp Rsm ð2Þ 1000 dT
Iscm


 
where V mp and Imp are the PV module’s output voltage and current Iscm
Im  NVscocm
V m ¼ Ns Nsc V T ln 1 þ e VT
1
Im Rsm ð6Þ
at MPP, respectively, Immr and T r are the PV module current at the Iscm

Table 1
Comparison among manufacturer specifications; real time measured data and calculated values

SP-36P1600108 / SolTek
Parameters STC (at irradiance 1000 W/m2and temperature 25°C) at irradiance 569.79 W/m2 and temperature 35.23°C
Datasheet Calculated Calculated Measured
V mp ðVÞ 18.70 18.68 17.82 17.48
Imp ðIÞ 5.35 5.32 3.07 3.03
P mp ðW Þ 100 99.38 54.71 52.95
V oc ðVÞ 22.50 22.45 21.58 21.15
Isc ðIÞ 5.80 5.76 4.79 4.52

164
S.R.K. Madeti Renewable Energy Focus 41 (2022) 160–178

The PV array’s predicted DC power may be calculated using the for-

V oc (V)
mulae given as:

22.42
22.42
22.42
22.42
22.42
22.42
21.36

21.23
21.10
20.95
21.3
21.6
Pdc cal ¼ V m  Im ð7Þ
Over a three-year period, the maximum power depreciation
rate of individual PV modules employed in this research was inves-

Isc (A)
5.72
5.72
5.72
5.72
5.72
5.72
2.63
3.16
3.94
3.23
2.58
2.25
tigated. The annual growth rate was calculated to be 0.5%, which is
well within the limitations set by the IEC-61215 standards [39]. As
a result, after 11 years of operation, the power output of the mod-
ules used in this study had degraded to 90% of its original nominal

P mp (W)

43.84
51.62
62.25
51.67
41.51
99.05
99.05
99.05
99.05
99.05
99.05

36.03
power and to 80% after 22 years. Table 1 presents the actual and
estimated values of Imp , V mp , Isc ; V oc and P mp based on equations
(1), (2), (3), (4) and (7) for STC and real climatic conditions,
respectively.

V mp (A)
18.62
18.62
18.62
18.62
18.62
18.62

17.48
17.41
17.32
18.08

17.10
17.7
PV array configuration under STC and real working conditions

PV module
In this part, the variation of the operational parameters at STC
and real working conditions for PV array configurations are given

Imp (A)
in Table 2. From the obtained results the major outcomes are: for

5.32
5.32
5.32
5.32
5.32
5.32
2.42
2.92
3.64
2.97
2.39
2.08
series configuration the dominant parameters are V mp and V oc
whereas for parallel configuration the dominant parameters are
Imp and Isc . But SP and TCT configurations have a similar value

n21.23

n20.95
across all parameters.

m22.4

m22.4

m21.6

m21.0
n22.4

n22.4
V oc (V)
224.2
22.42

213.6
21.3
Partial shading and fault conditions applied to PV array configurations
To assess the behavior of each PV array configurations under
partial shading and fault conditions, different shading scenarios

m5.72

m5.72

m3.23

m2.25
n5.72

n5.72

n3.94

n2.58
and faulty conditions were tested. The considered faults are

31.61
Isc (A)
5.72
57.2

2.63
applied at 33.33%, 66.67% and 100% in the system. For example,
33.33% short circuit fault refers the scenarios where group 1 ðG1 Þ
number of cells in PV module ðM1 Þ are short circuit. In partial shad-
ing condition the pattern used to evaluate the effect on PV module
P mp (W)

437.95
516.29
622.59
516.78
415.16
361.72
990.58
990.58
990.58
990.58
990.58
990.58

is depicted in Figure 4.
Table 3 shows the operational parameter variations in the P, S,
SP, and TCT PV array topologies under partial shadowing and for all
considered faulty PV conditions. Changes in the properties of vari-
ous operational parameter readings are related in order to identify
m17.41
n17.48

n17.32
m18.6

m18.6

m17.1

the faulty module. The amount of faulty PV modules has an effect

n18.6

n18.6
V mp (V)
186.2
18.62

180.6
17.7

on the anomalies. To represent whether the value of an indicator

has ‘‘ ;” decreased, ‘‘ "” increased, ‘‘–” no change in its value, ‘‘
;"” decline or increase in the value of the indicator, and ‘‘ –;” no
change or decrease in its value, three distinct symbols are
PV Array

m5.32

m5.32

m2.97

m2.08

employed. The obtained results show that the P configuration pro-

n5.32

n5.32

n3.64

n2.39
Imp (A)

29.17
5.32
53.2

2.42

duces the most output power under various partial shade and fault
Operational parameters values for all examined PV array Configurations

conditions, compared to all other configurations. For varied partial

shade circumstances, the S, SP, and TCT configurations provide the
same output power.
mn

mn

mn

mn
nm

nm

nm

nm
PV Configuration

Proposed method for fault detection and diagnosis

Many Researchers have developed fault detection techniques

TCT

TCT

and compared the performances of various PV array configurations

SP

SP
P

P
S

and analyzed its behavior under partial shading scenario. Authors

in, [40,41] analyzed and compared five PV Configurations S, SP,
TCT, BL and HC under six shading conditions. Authors in [42] mod-
T=29.84°C
T=34.59°C
T=41.88°C
T=38.55°C
T=36.03°C
T=35.71°C

eled and analyzed six PV Configurations S, P, SP, TCT, BL and HC

G=1000W/m2, T=25°C
Under STC conditions @

under various shading patterns. Authors in [26,43] investigated

on output power of various PV Configurations by conducting
@G=450.72W/m2,
@G=538.82W/m2,
@G=666.88W/m2,
@G=548.16W/m2,
@G=438.38W/m2,
@G=382.93W/m2,

experiments under uniform and PSCs. In [44], the authors pre-

sented the mathematical analysis of the TCT PV Configuration
under PSCs and compared its performance to that of other PV Con-
figurations such as SP, BL, and HC. The authors of [45] investigated
Table 2

at S, P, SP, TCT, and BL PV array configurations in the presence of

defective PV and partial shade. The impact of partial shadowing
165
S.R.K. Madeti Renewable Energy Focus 41 (2022) 160–178

on SP, BL, and TCT PV configuration was examined by authors in things. The suggested approach is based on the examination of a
[46,47]. MATLAB/SIMULINK software was used to compare the set of I-V characteristic features (such as current, voltage, and peak
power produced and mismatch loss for the aforementioned PV number) that indicate normal and faulty operations.
configurations. However, the use and reliability of these defect The main objective of the fault detection method is to identify
detection methods for conventional PV array configurations have when and where a fault occurred in the PV system. This proposed
not been investigated. Furthermore, these methods keep track of fault detection technique may be applied to any type of PV array
the PV system at the inverter or string level. They are unable to configurations considered in this work. The operational parameters
identify problematic PV modules or assist in their maintenance of PV module are calculated by taking real time measured irradi-
for optimal performance. Based on the findings of the literature ance and module temperature profiles as an input to the predicted
review, there is potential to design a monitor-level PV monitoring model, so that the parameters are calculated in real time. Variation
system with defect detection for various PV array configurations. between the theoretically computed ðP dc cal Þ and measured
The suggested fault detection technique introduces a novel method ðP dc meas Þ values of PV array output power are compared with the
for isolating and identifying defects in a PV system, as well as its threshold to begin the fault detection operation.
integration into the same microcontroller that handles data trans-
Pdc cal
Pdc meas
fer. The approach may locate and diagnose problems in PV mod- Pdc PU ¼ ð8Þ
Pdc cal
ules, PV strings, partial shading, and bypass diodes, among other

Figure 4. PV array under different shading conditions. (a) All the modules are uniformly illuminated (b) partially shaded array with 33.33% of M1 is shaded (c) partially
shaded array with 66.67% of M 1 is shaded (d) partially shaded array with 100% of M 1 is shaded.

166
S.R.K. Madeti Renewable Energy Focus 41 (2022) 160–178

Table 3
Change in operational parameter of each PV array configuration. A) Open Circuit (OC) Fault B) Short Circuit (SC) Fault. C) Inverted Bypass Diode (IB) Fault. D) Partial Shading with
Normal Bypass Diode (SNBD) Fault. E) Partial Shading with-out Bypass Diode (SFBD) Fault. F) Partial Shading with Inverted Bypass Diode (SIBD) Fault

Under OC conditions @ G=562.15W/m2, T=37.8°C for SolTek

PV Array Unhealthy PV module
PV Configuration Imp (A) V mp (A) P mp (W) Isc (A) V oc (V) Imp (A) V mp (A) P mp (W) Isc (A) V oc (V)
S – ; ; – ; ; ; ; –; ;
P ; – ; ; ; ; ; ; –; ;
SP nm ; ; ; – – ; ; ; – ;
mn ; ; ; – – ; ; ; – ;
TCT nm ; ; ; ; – ; ; ; – ;
mn ; ; ; ; – ; ; ; – ;

Under SC conditions @ G=250.68W/m2, T=25.3°C for SolTek

PV Array Unhealthy PV module
PV Configuration Imp (A) V mp (A) P mp (W) Isc (A) V oc (V) Imp (A) V mp (A) P mp (W) Isc (A) V oc (V)
S – ; ; – ; – ; ; – ;
P ; ;" ; ; ; – ; ; – ;
SP nm ; ; – – ; ; ; ; – ;
mn ; – – – ; ; ; ; – ;
TCT nm ; ; ; ; – ; ; ; – ;
mn ; – ; ; – ; ; ; – ;

Under IB diode conditions @ G=919.26W/m2, T=40.3°C for SolTek

PV Array Unhealthy PV module
PV Configuration Imp (A) V mp (A) P mp (W) Isc (A) V oc (V) Imp (A) V mp (A) P mp (W) Isc (A) V oc (V)
S – ; ; – ; – ; ; – ;
P ; – ; ; ; ; ; ; – ;
SP nm – ; ; – – ;" ; ; – ;
mn – ; ; – – ;" ; ; – ;
TCT nm – ; ; ; ; ;" ; ; – ;
mn – ; ; ; ; ;" ; ; – ;

Under SNBD conditions @ G=632.35W/m2, T=40.4°C for SolTek

PV Array Unhealthy PV module
PV Configuration Imp (A) V mp (A) P mp (W) Isc (A) V oc (V) Imp (A) V mp (A) P mp (W) Isc (A) V oc (V)
S ; ; ; – – – ; ; – ;
P ; ; ; ; ; ; ; ; ; ;
SP nm ; ; ; – – ; – ; – –
mn ; ; ; – – ; – ; – –
TCT nm ; ;" ; ; – ; ;" ; ; ;
mn ; ;" ; ; – ; ;" ; ; ;

Under SFBD conditions @ G=620.58W/m2, T=36.5°C for SolTek

PV Array Unhealthy PV module
PV Configuration Imp (A) V mp (A) P mp (W) Isc (A) V oc (V) Imp (A) V mp (A) P mp (W) Isc (A) V oc (V)
S ; ; ; – ; ; ; ; ; ;
P ; – ; ; ; ; ; ; ; ;
SP nm ;; ;" ; – ; ; ;" ; ; ;
mn ; ;" ; – ; ; ;" ; ; ;
TCT nm ; ; ; ; – ; ; ; ; ;
mn ; ; ; ; – ; ; ; ; ;
Under SIBD conditions @ G=297.76W/m2, T=32.7°C for SolTek
PV Array Unhealthy PV module
PV Configuration Imp (A) V mp (A) P mp (W) Isc (A) V oc (V) Imp (A) V mp (A) P mp (W) Isc (A) V oc (V)
S ; " ; – ; ; ; ; ; ;
P ; – ; ; ; ; ; ; ; ;
SP nm ; ; ; – ; ; ; ; ; ;
mn ; ; ; – ; ; ; ; ; ;
TCT nm ; ; ; ; – ; ; ; ; ;
mn ; ; ; ; – ; ; ; ; ;

} # } indicated decrease, } " } indicates an increase, }
} indicates no effect in the value, } #" } indicates decrease or increase in the value of the indicator, and }
# } no
change or decrease in its value.

To avoid false or incorrect fault detection, a specified threshold Pdcref PU
r1  PdcPU  Pdcref PU þ r1 ð9Þ
for PV array output dc power reading ðPdc PU Þ is required. Therefore,
in this research establishes thresholds for various parameters. If
the results do not exceed a certain threshold, the PV system is con- where Pdcref PU is PV array output dc power per unit during fault free
sidered normal; otherwise, an issue is detected. After statistical operation.
examination of the development of the P dc PU These reference The diagnostic signal is set to low state when the power differ-
thresholds were established when the PV system was fault-free. ential is within the preset boundary: If the diagnostic signal is zero,
Then, the standard deviation ðr1 Þ of Pdc PU is derived. When the val- the system is considered to be in working order; otherwise, it is
ues are within the following threshold, the PV array is treated to be assigned to a high state: Diagnostic signal = 1, and an anomaly in
fault-free: the PV array has been discovered. Figure 5 shows the flow chart
167
S.R.K. Madeti Renewable Energy Focus 41 (2022) 160–178

of the mentioned procedure to detect type of PV configuration and where, r2 , r3 ; r4 , r5 ; and r6 are the standard deviation of module
malfunction in operational PV system. power, maximum voltage, maximum current, open-circuit voltage,
The diagnostic process described above allows for the identifi- and short circuit current, is calculated in daily basis of experimental
cation of the kind of PV array configuration as well as the detection and predicted model given in case of clear sky conditions. The stan-
of any abnormalities in the PV system. Theoretical boundary values dard deviation values for operational parameters are derived, as
were established for this reason, with the measured operational depicted in Table 4. Since the PV modules considered in this work
parameters of each module not exceeding any of them; otherwise, are from the same manufacturer the obtained standard deviation
the system would be considered faulty. The introduction of clear values are the same for each PV system.
sky in-plane irradiance data to the analytically computed model The most probable faulty PV array configuration can be found
evaluates these upper and lower boundaries of the PV module by using the flowchart in Figure 6.
operating parameters. Thus, the upper and lower boundaries are Figure 7 shows the diagnostic procedure used to find out the
determined using a statistical technique, and the measured mod- most likely faults that caused excessive losses in particular PV
ule operating parameters stay within the theoretical boundaries array configuration.
in the event of normal PV system operation (fault-free), as pro- The level (percentage) of fault existent in PV module can be
vided by the following expression: detected by following the flowchart depicted in Figure 8.
DPmodref
r2  DPmod  DPmodref þ r2 ð10Þ
Table 4
DV mp modref
r3  DV mpmod  DV mpmod þ r3 ð11Þ Standard deviation for operational parameters.
ref
Standard deviation (r)
DImp modref
r4  DImpmod  DImpmod þ r4 ð12Þ r1 0.071
ref
r2 0.066
r3 0.054
DV oc modref
r5  DV ocmod  DV ocmod þ r5 ð13Þ r4 0.042
ref
r5 0.053
r6

r6  DIscmod  DIscmod þ r6
0.044
DIsc modref ð14Þ
ref

Figure 5. Flowchart of the proposed PV detection procedure.

168
S.R.K. Madeti Renewable Energy Focus 41 (2022) 160–178

Figure 6. Flowchart of the proposed fault detection procedure in various PV array configurations.

Experimental test results sor’s accuracy dictating the measurement’s accuracy rather than
the ADC’s. Should the need arise, the high-performance ADC
Experimental test results of I
V tracer enables for future sensor upgrades with increased sensor accuracy.
For the clock, a 20 MHz crystal was used. The microcontroller is
Experiments were conducted to trace the accurate I
Vcurves of programmed in the C programming language. The ‘‘Five parame-
PV module under real operating conditions. Two low cost ters” methodology was utilized to determine the PV module’s
ATMega8535 microcontrollers are used to measure various moni- key 5 parameters, which were then used to compute the operating
tored parameters, data transmission and for fault detection. The parameters analytically using SolTek. MATLAB/Simulink software
sensor circuit has a resolution of up to 1.22 mV as a result of this. was used at first to extract the five parameters. Netbeans, which
This results in a high-precision ADC measurement, with the sen- uses Java programming, was selected for the web-based solution.

169
S.R.K. Madeti Renewable Energy Focus 41 (2022) 160–178

Figure 7. Flowchart of the proposed fault detection procedure for (a) parallel (b) series (c) series-parallel or cross-Tied.

Figure 8. Flowchart for the level of fault present in PV module (a) partial shading with bypass diode (b) partial shading with faulted bypass diode (c) short circuited (d) open
circuited (e) partial shading with inverted bypass diode (f) inverted bypass diode.

This is due to the fact that Java is cross-platform, meaning it can The generated model’s performance was compared to the Solar
run on any operating system (OS), whereas.NET can only run on I-V of HT instruments. The current-voltage characteristics of the PV
Windows. Furthermore, as compared to.NET, the Java platform is module are measured using this device. The data collected by the
more secure. instrument is sent to a server through the Universal Serial Bus

170
S.R.K. Madeti Renewable Energy Focus 41 (2022) 160–178

Table 5 Uncertainty analysis

HT Instruments’ ’Solar I-V’ curve tracer technical specifications.

Description Range Using statistical analysis of a series of recorded observations,

Measuring range of I-V curve/V oc
Isc 1000 V/15 A measuring devices such as voltage, current, K-type thermocouple,
Photovoltaic testing measuring range 1000VDC/265V AC and reference cell sensors were examined for uncertainty analysis.
Memory capacity on board 200 curves I-V 8 days The following steps are used to determine the instruments’ uncer-
Recording 5s–60m
tainty and standard deviation.
I-V curve detection on PV modules and strings 128 Points
Size in L *W* H 235 * 165 * 75 (mm)
Step 1: A certain device was employed to create a set of mea-
Weight 1200 g sured data under the same conditions, namely A1, A2, A3,....., An,
where n is the number of observations.
Step 2: The average of the measured data (A) was determined
using (16), where ‘i’ and ‘n’ are the values and number of observa-
tions, respectively.
Pn
i¼1 Ai
A¼ ð16Þ
n
Step 3: Eq. (17) was used to compute the standard deviation (S)
of experimental results, where Ai is the measured data, A is the
average, and n is the number of observations.
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
uP
un
u ðAi
AÞ2
ti¼1
S¼ ð17Þ
n
1

Uncertainty analysis of voltage sensor

A constant voltage source of 20V DC provided by a Regulated
Power Supply device was utilized to examine the uncertainty of
a voltage sensor module (25V) deployed in the experiment. The
device specifications and five measurement values have been given
Figure 9. Simulated and measured I
V characteristics under different test
in the Table 6.
conditions for SolTek PV module SP-36P100. So, using (17), the standard deviation of the voltage sensor
module was determined to be 11.32 mV or ±0.05%.
(USB) (USB-RS232). Table 5 lists the technical requirements for the
Solar I-V curve tracer. Uncertainty analysis of current sensor
For a single day, outside experiment measurements were Similarly, LEM LA 55-P current sensor has been used to measure
obtained at one-hour intervals from 10:00 a.m. to 17:00 p.m. the PV current in the experiment. It has been tested for using a
(12th Aug 2021). Figure 9 displays the I-V characteristics of a PV power supply that carries the current at a constant current and
module measured using a designed I-V tracer and PV analyzer that has low noise (ripple) and five measurements have been
’Solar I-V’ of HT instruments for the SolTech (SP-36P100) PV mod- taken. The specifications and measurement values are shown in
ule under various irradiation and temperature levels. The findings Table 7. The current sensor’s standard deviation was calculated
showed that the developed I-V tracer is capable of reproducing the and determined to be 8.95 mA, or ±0.03%
PV module’s I-V characteristics. The root mean square error (RMSE)
was utilized to assess the I-V tracer’s accuracy. The RMSE is calcu- Uncertainty analysis of reference cell
lated as follows: Similarly, the reference cell (Si-V-1.5TC) was tested using six
separate observations, with the solar irradiance (G) being mea-
"Pn  2 #0:5
IHT i
Idev i sured six times at a period when the module’s irradiance was
i¼0
RMSE ¼ ð15Þ approximately 600 W/m2. Table 8 lists the observation values as
n
well as device specifications. The device’s standard deviation was
calculated and determined to be ±0.21%.
th th
where IHT i and Idev i is the i HT instrument current and the i devel-
oped I
V tracer current, respectively and n is no.of measured Uncertainty analysis of temperature sensor
points. Comparing the developed I
V current values with HT Five measurements of the module temperature operating at a
instrument values, it is noted that the results are very closer. The particular point of time of 38°C were taken for the uncertainty
RMSE has been calculated using (15) and found to be 0.0224. analysis of the K-type thermocouple temperature sensor. Table 9

Table 6
Voltage sensor (in V) module specifications and measured values

Measurement Value (in V) Deviation from Average Specifications

A1 20.00 +0.00 Trademark Walfront
A2 20.00 +0.00 Input voltage 0 to 25V
A3 19.99 -0.01 Voltage detection range 0.02445 to 25V
A4 19.98 -0.02 Analog Voltage Resolution 0.00489V
A5 20.01 +0.01 Origin country China
Average (A) 19.99 Estimated Standard Deviation = 0.01224

171
S.R.K. Madeti Renewable Energy Focus 41 (2022) 160–178

Table 7
Current sensor (in A) specifications and measured values.

Measurement Value (in A) Deviation from Average Specifications

A1 4.99 -0.01 Trademark LEM
A2 5.00 +0.00 Primary nominal r.m.s. current 50A
A3 5.01 +0.01 Current measuring range 0 to ±70A
A4 4.99 -0.01 Accuracy ± 0.65%
A5 4.99 -0.01 Origin country USA
Average (A) 4.99 Estimated Standard Deviation = 0.008

Table 8
Irradiance (in W/m2) sensor specifications and measured values

Measurement Value (in W/m2) Deviation from Average Specifications

A1 600 0.5 Trademark IMT Solar
A2 597 -2.5 Current consumption Vdc = 24 (4 to 28 Vdc) typical 15mA
A3 601 1.5 Output signal 0 to 1.5V for 0 to 1500W/m2
A4 598 -1.5 Accuracy ± 2.5%
A5 602 2.5 Operating temperature -35°C to 80°C
A6 599 -0.5 Origin country USA
Average (A) 599.5 Estimated Standard Deviation = 1.654

Table 9
Temperature (in °C) sensor specifications and measured values.

Measurement Value (in °C) Deviation from Average Specifications

A1 34.8 +0.2 Trademark ReoTemp
A2 35.1
0.1 Voltage detection range
270 °C to 1260 °C
A3 35.0 +0.0 Accuracy ± 2.2%, ± 0.75%
A4 34.9 +0.1 Special Limits of Error +/
l°C or0.4%
A5 35.2
0.2 Origin country India
Average (A) 35 Estimated Standard Deviation = 0.0132°C

Table 10
The tested fault cases in detail.

August 13, 2021 August 14, 2021 August 15, 2021 August 16, 2021 August 17, 2021
Day 1 Day 2 Day 3 Day 4 Day 5
PV S, P, SP, TCT Parallel Series Series-Parallel Total Cross-Tied
Array Configuration
Case 1: No-Fault Case 2: SC Fault Case 2: SC Fault Case 2: SC Fault Case 2: SC Fault
Tested configurations: 8 AM to 9 AM 8:30 AM to 9:30 AM 8:10 AM to 9:10 AM 8:15 AM to 9:25 AM
Parallel: Case 3: OC Fault Case 3: OC Fault Case 3: OC Fault Case 3: OC Fault
9:30 AM to 11 AM 9:30 AM to 10:30 AM 9:56 AM to 10:46 AM 9:50 AM to 11:0 AM 9:55 AM to 10:55 AM
Series: Case 4: IB Fault Case 4: IB Fault Case 4: IB Fault Case 4: IB Fault
11:15 AM to 12:45 PM 11 AM to 12:30 PM 11:15 AM to 12:30 PM 11:41 AM to 12:30 PM 11:15 AM to 12:18 PM
Series-Parallel: Case 5: SNBD Fault Case 5: SNBD Fault Case 5: SNBD Fault Case 5: SNBD Fault
1:30 PM to 3 PM 1:40 PM to 2:40 PM 1:30 PM to 2:45 PM 1:40 PM to 2:40 PM 1:15 PM to 2:20 PM
Total Cross-Tied: Case 6: SFBD Fault Case 6: SFBD Fault Case 6: SFBD Fault Case 6: SFBD Fault
15:15 PM to 16:45 PM 3:10 PM to 4:10 PM 3:15 PM to 4:18 PM 3:04 PM to 4:00 PM 3 PM to 4:00 PM
Case 7: SIBD Fault Case 7: SIBD Fault Case 7: SIBD Fault Case 7: SIBD Fault
4:43 PM to 5:33 PM 4:45 PM to 5:35 PM 4:30 PM to 5:30 PM 4:15 PM to 5:20 PM

* SC Fault: short circuit fault.

* OC Fault: open circuit fault.
* IB Fault: inverted bypass diode fault.
* SIBD Fault: shaded with inverted bypass diode.
* SNBD and SFBD Fault: shaded with bypass diode and shaded with fault bypass diode.

172
S.R.K. Madeti Renewable Energy Focus 41 (2022) 160–178

shows the results of the temperature sensor (T) measurements. arrow of I
V tracer tab on the home page real-time I
V and
The standard deviation of the temperature sensor was calculated P
V characteristics of an individual PV module can be observed,
and found to be ±0.32%. as shown in Figure 11. The tracing curve and operational parame-
As a result of the measurement equipment’s uncertainty analy- ters of the PV module can be viewed by using the corresponding
sis, the measured values have maximum standard deviations of module button. The ‘‘Display Plot” button may be used to view
less than 1%. As a result, the effect of measurement devices on real-time monitoring parameter values in the form of graphs on
the system study is negligible and can be ignored for future particular data, as shown in Figure 12. Because the sample interval
investigation. of data measures is 10 min, the website refresh rate is set to 10
min.
Experimental test results of the fault detection technique
Normal operation
The proposed fault detection and diagnosis approach is evalu- The different PV array configurations were monitored in fault-
ated in this part to ensure that it works for different PV array free operation under clear skies. On Day 1 (August 13th, 2021),
designs. Specific tests on the system were conducted to assess all of the evaluated configurations are made with a one-hour inter-
the capabilities of the developed approach in identifying the kind val. On that particular day, Figure 12 depicts the observed solar
of PV array setup and fault detection. Measurements taken every radiation, temperature profiles, and predicted and measured DC
10 min during a five-day period (August 13–17, 2021) were exam- power of several PV array topologies. It was determined that mea-
ined for this purpose. Six case studies with various fault configura- sured and estimated powers are extremely similar. P dc PU is lower
tions were examined. Table 10 lists the faults that were manually than the set threshold for DC power differential. As can be seen,
created. The created monitoring system can automatically display the observed data and estimated outcomes are in excellent agree-
and store the type of PV configuration as well as the PV system’s ment. The status of the diagnostic signal determines the system’s
status (normal/fault). It also includes a feature that allows it to normalcy or abnormalities. It is in a low ’0’ condition in this scenar-
visually depict the degree of shading existing on the PV module io, suggesting that there is no malfunction in the PV system. The PV
in real-time. The fault diagnostic method has been tested on a vari- system seems to be in good working order. In addition, as shown in
ety of PV array configurations. Figures 10, 11, and 12 show the Figure 10, information on the system’s setup and performance is
complete web page. provided on the created web page.
The home screen, shown in Figure 10, displays real-time solar
radiation, module temperature, PV array configuration, diagnostic Fault detection during various PV array configurations
signal, and fault status. It also shows that how much percentage Different fault studies on various PV array configurations were
shading occurs on the PV module pictorially. By clicking the down conducted to assess the performance of the proposed technique.

Figure 10. Online web page view of various PV array configurations on 13th August 2021

173
S.R.K. Madeti Renewable Energy Focus 41 (2022) 160–178

Figure 11. Real time I
V and P
V characteristics of the PV module.

Figure 13a–d shows the meteorological data as well as the differ- fault detection approach when the PV array is in a parallel config-
ences in measured and predicted PV power corresponding to dif- uration. On day 2, the connection of one of the modules (M 1) was
ferent fault conditions at various PV array configurations, shorted between 8:00 AM and 9:00 AM for this purpose, and the
respectively. shading effect was achieved by partly shading module (M 3) with
translucent gelatin paper. 16:43 to 17:33, also inverted the bypass
Parallel PV array configuration. Short circuit and partial shading diode on the same day, respectively. Figure 14 (c) and (d) illustrate
with inverted bypass diode faults are investigated during this the recorded, measured, and estimated values of variables during
mode of operation to demonstrate the efficiency of the suggested these tests. The most likely defect causing these significant power

Figure 12. Graphical view of complete monitored data on 13th August 2021.

174
S.R.K. Madeti Renewable Energy Focus 41 (2022) 160–178

Figure 13. Graphical view of faults occurring on (a) parallel configuration on day 2 (b) series configuration on day 3 (c) series-Parallel configuration on day 4 (d) total-cross-
tied on day 5.

Figure 14. (a) Online web page of monitored data on day 2 (14th Aug 2021) during SC fault (8:28 AM) and (b) Online web page of monitored on same day during SIBD fault
(17:00 PM) (c) Graphical view during SC fault (d) Graphical view during SIBD fault.

175
S.R.K. Madeti Renewable Energy Focus 41 (2022) 160–178

Figure 15. (a) Online web-page view of monitored data on day 3 (15th Aug 2021) during open circuit fault (10:18 AM) and (b) corresponding graphical view (c) Online web-
page view of on-going monitored data on same day during SNBD (13:47 PM) and (d) corresponding graphical web-page view.

Figure 16. (a) Online web-page view of monitored data on day 4 (16th Aug 2021) during inverted bypass diode fault (12:03 PM) and (b) corresponding graphical view (c)
Online web-page view of monitored data on day 5 (17th Aug 2021) during SFBD fault (15:26 PM) and (d) corresponding graphical view.

176
S.R.K. Madeti Renewable Energy Focus 41 (2022) 160–178

Table 11
Characteristics of the components of the monitoring system in terms of cost.

Cost per module that is required Cost for PV system that is required
C dcv sen
ðINRÞ C dci sen
ðINRÞ C ref cell ðINRÞ C temp ðINRÞ C X
bee mod ðINRÞ C stor ðINRÞ C misc ðINRÞ

150 200 20,000 100 600 2000 5000

Table 12
Results comparing the proposed method to the one described by Chine et al (2014) [17].

Author Investment cost (INR) Hardware/ Automatic error recognition Monitoringmethod Payback period
Software required (module level) (months)
China et al. 1,73,713 + sensors cost/module Limited to locate fault PC 20
Proposed system 385.32/module  Location and type of fault identified Web 8

losses may be identified by an examination of the operational where, nis is the number of current sensors, C dcisec ðINRÞ is the cost of
parameter values. The created tool reports the kind of configura- each DC current sensor, nv s is the number of voltage sensors, C dcv sec
tion, defect, and shading level in detail, as shown in Figure 14 (a) ðINRÞ is the cost of each DC voltage sensor, C temp ðINRÞ is the cost of k
and (b). type thermocouple sensor, C refcell ðINRÞ cost of a reference cell for
irradiance measurement, C X
beemod ðINRÞ is the cost of X-bee modules,
Series PV array configuration. To detect the problem of the faulty C stor ðINRÞ is the cost of data storage, C misc ðINRÞ is the cost of various
module in the series PV array configuration, one module ðM 2 Þ items like wire, circuit boards, enclosures, and so on. The economic
rd
has been open circuited and 23 cells of module ðM 6 Þ is partially aspects of commercially available components based on selling
shaded with bypass diode. These faults were created on 9:56 AM prices in the market are summarized in Table 11.
to 10:46 AM and 13:25 PM to 14:15 PM on day 3 respectively. It has been demonstrated that the use of X-bee modules for data
Power flow in the PV system is decreased during the faults period, transmission avoided the additional cost of wires which is used in
as indicated in Figure 15 (a) and (c). In both circumstances, the sta- case of wired based PV monitoring system. Using typical industrial
tus of the diagnostic signal is set to ’1’. The most likely defect may communication interfaces (RS232, RS485) need extra cabling,
be found by examining the PV array’s terminal voltage. The bypass which costs INR 55,000.
diode preserves circuit continuity during an open circuit failure. To see whether the suggested monitoring system with diagnos-
However, these errors result in a drop in the voltage of the faulty tic procedure is cost-effective, the cost of the created system was
string. The voltage differential ðDV dc Þ is more than the predeter- compared to the method described by Chine et al (2014, 2016).
mined value. This kind of issue may be discovered by examining Chine et al. designed a technique in which (2014), two different
each module’s measured voltage. The system performance is data loggers were used, the first one (Danfoss ComLynx Monitor)
shown in Figures 15 (b) and (d). The collected findings suggest that is used to record the meteorological data, while the second one
the proposed method is capable of appropriately identifying prob- to record the electrical data (Mastervolt QS Data Control Premium).
lems in the PV array under investigation. The cost of two data loggers are INR. 82,068 and INR. 91,645,
respectively inclusive taxes.
Series-Parallel and Total Cross-Tied PV array configuration. In this A cumulative continual 10% undiscovered loss of income of INR.
scenario, an inverted bypass diode fault was formed on module 2,85,262/year in a 50-kW PV that may theoretically invoice INR.
(M5), while partial shading was induced on module (M9) without 28,52,620/year would imply a loss of revenue of INR. 2,85,262/year
a bypass diode fault. These two problems occurred on SP and TCT in a 50-kW PV that can potentially invoice INR. 28,52,620/year. The
PV array configurations on Days 4 and 5, respectively, from estimated cost of the developed monitoring system is about INR.
11:41 AM to 12:30 PM and 15:04 PM to 16:00 PM. Figure 16 (a) 385.32 (which is 8.03 percent of a 100 Wp PV panel), and deploy-
and 16 (b) illustrate the recorded and computed values of variables ing the monitoring system in a 500 PV panels plant would cost
during these testing (c). The condition of the fault diagnostic signal around INR. 1,92,660. The investment is completed in around
becomes ’1’ since the power differential ðP dc PU Þ exceeds the 0.67 years (eight months) of PV plant operation. In case of other
threshold value in both circumstances. Analyzing the observed system developed by [17], need separate data loggers. This system
voltage and current of each module may reveal the sort of issue. also requires a skilled technician that would have to analyze each
More information about these flaws may be found in the created single module to identify the defective PV module in a string which
tool shown in Figure 16 (b) and (d). is a time-consuming activity causes 2-3% more undetected loss
(monetary loss of INR. 85,578.6/year), with an additional invest-
Cost analysis of the monitoring system ment cost of INR. 1,73,713. It will take 1.65 years (20 months)
to get the rate of investment back. The comparative study of two
The initial investment of a PV monitoring system for a PV plant monitoring systems is summarized in Table 12.
is the total sum of the expenses of the individual components
included in the PV monitoring system C mon is given as:
Conclusion
C mon ¼ nis C dcisec þ nv s C dcv sec þ C temp þ C refcell þ C X
beemod þ C stor þ C misc
ð18Þ In the proposed work, the effect of various partial shading and
faulty conditions on multiple PV array configurations have been
177
S.R.K. Madeti Renewable Energy Focus 41 (2022) 160–178

studied using predicted attributes and real-time experiments. The [4] B.P. Kumar, G.S. Ilango, M.J.B. Reddy, N. Chilakapati, IEEE J. Photovolt. 8 (1)
(2018) 257–265.
effect of partial shading and faults on various operational parame-
[5] S.K. Firth, K.J. Lomas, S.J. Rees, Sol. Energy 84 (4) (2010) 624–635.
ters obtained from the developed I
V tracer have been analyzed. [6] K. Hu, W. Li, L. Wang, F. Zhu, Z. Shou, IET Generat. Transmiss. Distribut. 12 (1)
The variations for all parameters across all PV array configurations (2018) 62–71.
have been reported and compared briefly. These parameters may [7] A. Dolara, G.C. Lazaroiu, S. Leva, G. Manzolini, Energy 55 (55) (2013) 466–475.
[8] K. Lappalainen, S. Valkealahti, Appl. Energy 190 (2017) 902–910.
be utilized as indicators in small and large PV arrays to forecast [9] B. Andò, S. Baglio, A. Pistorio, G.M. Tina, C. Ventura, IEEE Trans. Instrum. Meas.
or detect partial shading and defects. Role of specific thresholds 64 (8) (2015) 2188–2199.
for set of attributes in fault detection procedure have proven to [10] K.H. Chao, S.H. Ho, M.H. Wang, Electr. Power Syst. Res. 78 (1) (2008) 97–105.
[11] M. Hamdaoui, A. Rabhi, A. El Hajjaji, M. Rahmoun, M. Azizi, Monitoring and
be effective in preventing false fault detections. The performance control of the performances for photovoltaic systems, in: International
of proposed monitoring system has been demonstrated in four Renewable Energy Congress 2009.
cases of PV system operation as follows: [12] M. Davarifar, A. Rabhi, A. El-Hajjaji, M. Dahmane, Real-time model base fault
diagnosis of PV panels using statistical signal processing, in: 2013
International Conference on Renewable Energy Research and Applications
a) Fault free operation. (ICRERA) 2013 October 20. IEEE, pp. 599–604.
b) Parallel PV array configuration with one module SC and SIBD [13] T. Takashima, J. Yamaguchi, K. Otani, T. Oozeki, K. Kato, M. Ishida, Sol. Energy
Mater. Sol. Cells 93 (6–7) (2009) 1079–1082.
diode faults. [14] L. Schirone, F.P. Califano, M. Pastena, Prog. Photovoltaics Res. Appl. 2 (1) (1994)
c) Series PV array configuration with one module OC and SNBD 35–44.
faults. [15] A. Chouder, S. Silvestre, Energy Convers. Manage. 51 (10) (2010) 1929–1937.
[16] N. Gokmen, E. Karatepe, S. Silvestre, B. Celik, P. Ortega, Energy Convers.
d) S-P and TCT PV array configuration with IB and SFBD faults
Manage. 73 (2013) 350–360.
respectively. [17] W. Chine, A. Mellit, A.M. Pavan, S.A. Kalogirou, Renew. Energy 66 (2014) 99–
110.
The functionality and performance of the proposed system are [18] M. Dhimish, V. Holmes, Sol Energy 137 (2016) 236–245.
[19] P.R. Satpathy, R. Sharma, Energy 156 (2018) 569–585.
found to be satisfactory. Finally, this research provides important [20] S.R. Madeti, S.N. Singh, Energy 1 (134) (2017) 121–135.
data on the key characteristics that may be utilized to diagnose dif- [21] S.R. Madeti, S.N. Singh, Solar Energy 1 (173) (2018) 139-51.
ferent faults and partial shading circumstances in all PV array [22] Y. Zhao, R. Ball, J. Mosesian, J.F. de Palma, B. Lehman, IEEE Trans Power Electron
30 (5) (2014) 2848–2858.
designs studied. Furthermore, it has been discovered that the P, [23] A. Hazra, S. Das, M. Basu, J Cleaner Prod 154 (2017) 220–232.
S, and TCT topologies are the most efficient during partial shading, [24] S.R. Madeti, S.N. Singh, Sol. Energy 1 (158) (2017) 161–185.
while the SP topology is the least efficient. Similarly, for fault situ- [25] P.R. Satpathy, R. Sharma, S. Jena, Energy 139 (2017) 350–365.
[26] F. Belhachat, C. Larbes, Sol. Energy 120 (2015) 399–418.
ations, S, P, and SP have the maximum fault tolerance, whereas TCT [27] S. Mohammadnejad, A. Khalafi, S.M. Ahmadi, Sol. Energy 133 (2016) 501–511.
performs mediocrely. To limit the influence of partial shading and [28] P.R. Satpathy, R. Sharma, Energy Convers. Manage. 219 (2020) 113018.
faults in PV arrays, the aforementioned analysis may be utilized to [29] K. Ishaque, Z. Salam, Renew. Sustain. Energy Rev. 19 (2013) 475–488.
[30] P.R. Satpathy, R. Sharma, Energy Convers. Manage. 223 (2020) 113480.
find the most appropriate topologies and detection signs. [31] P.R. Satpathy, R. Sharma, Energy Convers. Manage. 192 (52–70) (2019) 2019.
[32] S. Pareek, R. Dahiya, Energy 95 (2016) 561–572.
[33] B.I. Rani, G.S. Ilango, C. Nagamani, IEEE Trans. Sustainable Energy 4 (3) (2013)
Declaration of Competing Interest
594–601.
[34] F. Spertino, J. Ahmad, A. Ciocia, P. Di Leo, A.F. Murtaza, M. Chiaberge, Sol.
The authors declare that they have no known competing finan- Energy 1 (119) (2015) 461–473.
[35] Silvestre, Santiago, Chouder, Aissa, Karatepe, Engin, Solar Energy 94 (2013)
cial interests or personal relationships that could have appeared
119–127.
to influence the work reported in this paper. [36] S.R. Madeti, S. Singh, Online modular level fault detection algorithm for grid-
tied and off-grid PV systems, Sol Energy 157 (2017) 349–364.
[37] Koutroulis, Eftichios, Kalaitzakis, Kostas, Renew. Energy 28(1) (2003) 139–152.
Acknowledgement [38] S. Silvestre, M.A. da Silva, A. Chouder, D. Guasch, E. Karatepe, Energy Convers
Manage 1 (86) (2014) 241–249.
The authors would like to thank the SRKR Engineering College [39] P.R. Satpathy, T.S. Babu, S.K. Shanmugam, L.N. Popavath, H.H. Alhelou, IEEE
Access (2021).
(Autonomous), Bhimavaram, FONDECYT-REGULAR 1160690 [40] Standard IE, 61215, Crystalline Silicon Terrestrial Photovoltaic (PV) Modules
research project and postdoctoral support of the University of Talca Design Qualification and Type Approval, International Electro technical
and VRSEC (Autonomous), Vijayawada. Commission, 2005.
[41] O. Bingol, B. Ozkaya, Sol. Energy 160 (2018) 336–343.
[42] P.R. Satpathy, S.B. Thanikanti, A.H. Mahmoud, R. Sharma, B. Nastasi, IEEE Trans.
References Sustainable Energy (2021).
[43] Sangram Bana, R.P. Saini, Energy 127 (2017) 438-453.
[44] Mohammad nejad, khalafi, Ahmadi, Solar Energy 133 (2016) 501-511.
[1] S.R. Madeti, S.N. Singh, Renew. Sustain. Energy Rev. 1 (2017) 67-1180-207.
[45] P.R. Satpathy, R. Sharma, S. Dash, Energy 175 (2019) 182–194.
[2] A. Jäger-Waldau, PV Status Report 2019, Publications Office of the European
[46] M. Dhimish, V. Holmes, B. Mehrdadi, M. Dales, B. Chong, L. Zhang, Renew
Union, Luxembourg, 2019.
Energy 113 (2017) 438–460.
[3] J. Ma, X. Pan, K.L. Man, X. Li, H. Wen, T.O. Ting, IEEE Trans. Ind. Appl. 54 (6)
[47] P.R. Satpathy, S. Jena, R. Sharma, Energy 144 (2018) 839–850.
(2018) 6279–6289.

178