Technical Briefing plant performance

Sampling guideline for inspection

and testing of PV modules in the field
Module performance | Testing a sample of modules at an operational solar can help identify faults
and underperformance in the wider plant, but which ones to choose? Authors from Mahindra
Teqo describe a new methodology they have developed to Satish Pandey, Preetha Pillai, Sandeep
Jadhav, Shyam Kumar, Gaurav Mishra, Rajesh Kumar Dhuriya

S
olar photovoltaic (PV) system power output (Pmax) at standard test of the module. The IR thermographic
installations are increasing by conditions and helps to evaluate the inspection of PV modules is performed to
leaps and bounds throughout the comparative analysis with the rated detect non-conformities such as hotspot
world. These systems are expected to power of the module. Flash testing is and diode failure. During thermo-
produce clean, safe and reliable electric- performed as per IS 14286/IEC 61215 graphic inspection the evaluation will be
ity over several decades of operation. and visual inspection of modules is performed on 100% of the plant modules
However, PV installations are subjected performed as per IS14286:2015/IEC or as per the respective requirement of
to extreme environmental conditions 61215:2016. Visual inspection can be the plant owner.
that could result in deteriorating effects done on a random basis and does not
on the equipment’s performance during require any equipment for inspec- Sample selection methodology at
their operational years. To ensure best tion. Hence it can be characterised as PV plant
performance and optimum ROI, these a general inspection. Similarly, a flash The sampling plan will apply to each
PV systems need periodic maintenance test and EL test are time consuming and module make respectively and the
and testing throughout their operational costly, and thus cannot be done on many bottom-line approach is to not consider
phase. These practices can help to under- samples. In IS2500/ISO-2859 there are visually observed defective modules,
stand module degradation behaviour two categories – general inspection level which would give a false interpretation
and provide essential information which and special inspection level. Based on our of average plant performance. If we have
can be used effectively to troubleshoot best practises we recommend General different module makes in the plant,
any problems arising within the system. inspection Level-II for visual inspection then the sampling plan will apply as per
Sampling for testing of PV modules and special inspection level S-4 for EL and the plant capacity but the total number
comprises the procedures involved to flash testing, as given in Table 1. In the of the samples will be distributed as per
select a part of PV modules from the case of EL testing it interprets the exist- the weighted capacity of the modules at
entire solar PV plant for inspection and ing micro-cracks, cracks and potential- the plant. For example, consider a 10MW
it should adhere to standard sampling induced degradation (PID) in the module, hypothetical plant with X make modules
methods IS2500/ISO-2859 and field- which affect the overall performance along with Y make modules and their
testing norms as per IEC 61215/61646
standards . The IS2500/ISO-2859 Sampling Plant size (MWp) Number of modules Sample size for EL Sample size for visual
sampling plan has been designed mainly bracket in plant & flash test (as per inspection sampling
for the pre-dispatch module inspection at special inspection (as per General
level S4) Inspection Level II)
manufacturing facility. However, in field
A Up to 0.0045MW 2 – 15 2 2
testing, the sampling needs to adopt the
constraints of the field environment and B 0.0045-0.008MW 16 - 25 3 5
limitation of the running plant. Accord-
C 0.0045-0.028MW 26 - 90 5 13
ingly, Mahindra TEQO has implemented
the sampling plan with the stakeholders D 0.028-0.048MW 91 – 150 8 20
for whom the testing has been carried E 0.048-0.16MW 151 - 500 13 50
out.
F 0.16-0.38MW 501 – 1,200 20 80

Sampling selection criteria as per G 0.38-1MW 1,201 to 3,200 32 125

IS2500/ISO-2859 G 1-2MW 3,201 to 10,000 32 200
This sampling plan is a result of our
H 2-8MW 10,001 to 35,000 50 315
expertise of handling a plus-3GW portfo-
lio since 2012. The below mentioned J 8-35MW 35,001 to 150,000 80 500
sampling plan has been designed for
J 35-120MW 150,001 to 500,000 80 800
electroluminescence (EL) testing, flash
testing and visual inspection. Flash K >120MW 500,001 & above 125 1,250
testing signifies the PV module maximum Table 1. Sampling plan for field testing in solar PV plant as per IS2500/ISO 28591-1

www.pv-tech.org | November 2019 | 77

 plant performance Technical Briefing

Sample selected as per sampling plan module assessment criteria is to bring it

into alignment with the standard guide-
Plant Capacity Samples Module make Proportion of Bracket Make-wise lines of ISO-2859. In field testing Mahin-
selected modules in number of
dra Teqo has absorbed the AQL criterion
plant modules
selected primarily to validate the outlier selection
10MWp 32 X 40% E 13/32 during the assessment process. The
outlier selection should be made through
Y 60% F 19/32 following the AQL 2.5 guidelines for
Table 2. Sample selection at PV plant with different module make major non-conformity as per ISO-2859.
The AQL and the sample size code letter
Plant Capacity Number of modules Sample size as per Table 1 Acceptable Outlier
shall be used to obtain the sampling
1MW 3,200 32 2 3 plan from Tables 1, 2, 3 or 4 (ISO-2589-1)
attached at the end of the document.
Table 3. Example for AQL
For a specified AQL and a given capacity
proportion in the plant is 40:60. Then, can reach to the PV module supplier/ of plant, the same combination of AQL
as per the sampling standard, the total manufacturer as PV modules accounts for and sample size code letter shall be used
number of modules to be selected for EL/ the 60% capex of the plant assets. This to obtain the sampling plan from the
flash testing will be 32 but these will be practice should be performed in accord- table for normal, tightened and reduced
divided as per the weighted capacity of ance with the warranty agreements of inspection.
the manufacturer; thus, we must select 13 the supplier/manufacturer. As per AQL 2.5 of ISO-2589 two major
modules from X and 19 modules from Y. conformities will be allowed for each
To select modules from the plant Acceptance quality limit to be module in acceptable range and if it is
Mahindra Teqo recommends following followed in compliance with more than two it will be considered an
methodology: ISO-2859 outlier. Therefore, it will be removed from
1. If the PV plant is operational then the Acceptance quality limit (AQL) is an average calculation. The AQL process will
module selection should be made as assessment criterion as per ISO-2589 in be followed by the sampling process as
per the inverter performance. pre-dispatch statistical sampling plans. proposed by Mahindra Teqo. For example,
2. If the plant is not operational then The notion behind including AQL in PV as given in Table 3.
the sample should be selected from
a random pallet or module mounting Correlating energy yield data with
structure/table. field data
Mahindra Teqo has correlated the energy
For operational plants, the weighted yield assessment (EYA) and samples
numbers of each module make should tested on a PV plant to get the overall
be divided into least performing, average performance of the plant. This correlation
performing and maximum performing is representative of the entire plant which
inverter. is validating the sampling of modules.
• The selection of these inverters will Data from tested modules using
be performed on a random basis this sampling methodology has been
with a stipulation of maximum three validated with the degradation obtained
locations for each module make. from the performance ratio (PR). A few
• After selection of the inverter, the next examples of plants are shown in Figure
stage is to select the modules from 2. Plant A with 1.2MWp capacity was first
the mounting table, which should be analysed using daily generation data,
picked from the positive and negative where the module degradation based on
end equally, and from the middle of the PR value is calculated. Then based on
the table. This helps to detect PID the plant capacity and performance of
problems more accurately if they exist. the inverter and watt peak rating of the
module, flash testing is performed on
IEC standards 61215 and 61646 set out modules. Based on the plant capacity the
special testing requirements for crystal- number of samples is selected as given in
line silicon and thin-film modules respec- Table 1. It has been observed that in Plant
tively. Performance of a module at a site A the degradation of modules obtained
can be determined with the help of these from flash testing is essentially the same
standards. The flash test results should as the yearly degradation obtained from
be interpreted as per the expected/ PR, hence the sample selected for testing
guaranteed performance of the module is representative of plant performance.
make from the respective manufacturer/ The PR calculation has the added
supplier. Also, if the corresponding results uncertainty of other equipment such
are not aligned with the expected perfor- as inverters, cables etc., so calculation
mance values then a plant developer Figure 1. Sample selection to correlate EYA and field test data of the module degradation in the plant

78 | November 2019 | www.pv-tech.org

 Technical Briefing plant performance

Authors
Satish Pandey is the head of solar plant performance
analysis and PV testing at Mahindra Teqo. He holds
both an M.Tech and MSc in physics and has over 10
years’ experience in the solar industry, working in
PV modules strategy, plant performance analysis,
design and engineering, R&D, process engineering, production
and quality control. He worked with PV manufacturer Moser
Baer Solar from 2008-2013 before joining Mahindra in 2013.

Sandeep Jadhav is head of O&M at Mahindra Teqo.

He has 22 years of experience in O&M, quality con-
trol, and testing and commissioning in the Indian
solar industry. He leads a workforce of 600 person-
nel and is responsible for imparting continuous on
Figure 2. Module degradation obtained from PR versus tested module samples job training for accomplishing greater operational effectiveness/
efficiency.

Preetha Pillai is an assistant manager in the solar

plant performance analysis and PV testing depart-
ment at Mahindra Teqo Pvt Ltd. She holds an
M.Tech. in electronics & communication and has
four years of experience in the PV module R&D and
quality control. She worked with the National Centre
for Photovoltaic Research and Education (NCPRE) at IIT Bombay
as a senior research fellow from 2015-2018.

Rajesh Kumar is an assistant managemer in the

solar plant performance analysis and PV testing
department at Mahindra Teqo. He worked at
Jodhpur Institute of Engineering & Technology
as assistant professor from 2013-2015 and joined
Mahindra Teqo in 2015.

Shyam Kumar is an assistant site engineer in the

solar plant performance analysis and PV testing
department at Mahindra Teqo. He holds an ITI in
electronics and has had over 14 years’ experience
Figure 3. EL image of healthy module Figure 4. EL image with crack on working in the solar Industry. He has worked with
module Moser Baer Solar Ltd. from 2008-2013, Hindustan
Power Projects 2013-2015 and joined Mahindra in 2015.

is done by flash test, visual inspection, Conclusions Gaurav Mishra is an assistant site engineer in
thermography and EL test of modules. This sampling methodology can be solar plant performance analysis and PV testing
department at Mahindra Teqo. He has a diploma in
This methodology is validated in the used to ascertain the overall perfor-
mechanical engineering and has 10 years’ experi-
course of seven years of Mahindra Teqo’s mance of a plant by testing sampled ence working in the solar Industry. He worked with
experience in the solar industry. Figures 3 modules that represent the entire Moser Baer Solar Ltd. from 2008-2014 and joined
& 4 show the EL testing of solar modules plant. There is no concrete guideline Mahindra in 2014.
done on site, where the healthiness of in a single standard available for field
solar module is checked. The samples for testing of PV modules in the market;
testing are selected as per the sampling to our knowledge, we are the first to Testing a meaningful sample of modules from a PV
guidelines recommend in this paper. standardise the whole process, and power plant can prevent potentially large financial
losses
Testing of sampled modules enables have prepared these guidelines based
us to identify faults in the plant, apply on our consultation with key stakehold-
corrective action and increase genera- ers such as independent engineers,
tion. If a 1MWp plant generates 1.70 lenders, financial institutions, develop-
million kWh/yr, then 1.5% extra module ers, EPC, manufacturer etc. This method-
degradation can cause a loss in genera- ology is aligned with IS 2500/ISO 2859
tion of 25,500kWh/yr. Based on a tariff sampling standards, which are defined
of US$0.07/kW, this would result in a primarily for pre-dispatch module
revenue loss of US$1,785/yr. Hence for testing; here IS standards have been
a 100MW plant, which is quite common incorporated as per field constraints.
nowadays, the revenue loss will be These guidelines will bring a coherency
178,500 USD/yr – a significant amount. to field testing for PV modules, helping
Therefore, identifying faulty modules to standardise the process and will
through testing of selected samples can provide a common platform for every
save revenue loss. stakeholder to compare the results. 

www.pv-tech.org | November 2019 | 79