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

net/publication/242190872

Mean time before failure of photovoltaic modules (MTBF-PVm)

Article · January 2002

CITATIONS READS

32 2,940

4 authors, including:

Antonella Realini N. Cereghetti

University of Applied Sciences and Arts of Southern Switzerland University of Applied Sciences and Arts of Southern Switzerland
17 PUBLICATIONS 222 CITATIONS 30 PUBLICATIONS 316 CITATIONS

SEE PROFILE SEE PROFILE

Domenico Chianese
University of Applied Sciences and Arts of Southern Switzerland
54 PUBLICATIONS 673 CITATIONS

SEE PROFILE

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

TISO 35 + View project

PV grid-connected pilot plant in Nepal View project

All content following this page was uploaded by Domenico Chianese on 01 April 2014.

The user has requested enhancement of the downloaded file.

 Energy research Active solar energy Swiss Federal Office of Energy
Photovoltaic programme SFOE

Annual Report 2002

Mean Time Before Failure of

Photovoltaic modules
(MTBF-PVm)
Author and Co-Authors A. Realini, E. Burà, N. Cereghetti, D. Chianese, S. Rezzonico
Institution / Company SUPSI, DCT, LEEE-TISO
Address C.P. 110, 6952 Canobbio, Switzerland
Telephone, Fax Tel.: + 41.91.935 13 55, Fax: + 41.91.935 13 49
E-mail, Homepage antonella.realini@dct.supsi.ch, http://www.leee.dct.supsi.ch
Project- / Contract Number BBW 99.0579
Duration of the Project (from – to) 17.04.2000 – 16.04.2003

ABSTRACT
In May 1982, the first European grid-connected PV plant came into operation. The objective of this array was to
provide a technologically advanced facility of medium size giving practical information for the planning of future
photovoltaic plants.
In April 2000, in proximity to the 20-year design life of the plant, a collaboration between the LEEE-TISO and
the ESTI Laboratory (JRC Ispra) started to determine the Mean Time Before Failure (MTBF) of the system and
to investigate the physical degradation mechanism in action and to correlate field reliability with accelerated
lifetime tests (IEC/CEI 61215).
Results show that, although it is not looking good from a visual aspect, the system is working in a very
satisfactory manner.
Concerning IR analysis, hot-spots have been detected on 24% of modules. Hot-spots represent the principal
cause of power degradation.
Outdoor strings measurements have been performed and the obtained data have been compared with the one
of the same measurements executed in 1983. It has been a general comparison, as the plant configuration and
the electrical wiring were changed in 1992 (inverter substitution). Results indicated no relevant strings
performance losses.
Since June 2000, individual string energy production has been continuously monitored and recorded, allowing
analysis and comparison of string behaviour. These data also show the seasonal variation of plant energy
production, which is strongly influenced by ambient temperature.
Regarding the determination of the Mean Time Before Failure of the plant, it is reasonable to assume, on the
basis of results obtained from accelerated lifetime tests, that the modules could continue to provide useful
electrical power for another 10-15 years.
 S-2
Mean Time Before Failure of Photovoltaic modules (MTBF-PVm)

Introduction

Durability of PV modules represents an important concern both for module manufacturers, interested
in producing reliable and cost-competitive devices, and for consumers, willing to invest in this quite
expensive technology in exchange of a guarantee of quality. Regarding c-Si technology, today’s PV
market offers modules qualified to survive 20-25 years, with guaranteed power production varying for
different manufacturers. At present, one of the aims of PV industries is to produce commercial
modules with lifetimes of 30 years or more.

The study of modules’ failure mechanisms can aid this drive towards higher levels of durability. With
this aim, several investigations have been carried out on field-aged and laboratory-aged modules;
utilising destructive and non-destructive techniques to analyse the degradation of various module
components.

The collaborative research program (Mean Time Before Failure Project) between the LEEE-TISO and
the ESTI laboratory of the JRC, in Ispra, aims to study these effects through the analysis of the 10 kW
PV plant located on the roof of the University of Applied Sciences of Southern Switzerland, also seat
of the LEEE-TISO. It was installed in May 1982 and it represents the first grid connected PV plant in
Europe.

The aim of this study is the determination of the Mean Time Before Failure of this plant, through the
analysis of the physical degradation mechanisms acting on the modules and by means of accelerated
ageing tests, according to the International Standard IEC 61215.

Project description

Since its installation, the plant has been continuously monitored by TISO, with the scope of detecting
possible safety and technical problems involved in the connection of a PV system to the public
electrical grid. Hence, a useful batch of data and information is available for reconstructing the history
of the plant and enabling a greater understanding of its behaviour during its “first” 20-year of life.

A major work item within the MTBF project has been to define the state of the plant, through the
execution of detailed visual inspections, infra-red (IR) analysis and indoor/outdoor performance
measurements. This was followed, where possible, by comparison between the current and historical
data, to define the evolution of the modules degradation mechanisms and to try to predict future
trends. For this purpose, additional information has been acquired by the execution of repeated
accelerated ageing tests, according to the International Standard IEC 61215.

At present, the main goal of the project is to quantify the effects of the various defects, detected on the
modules, with their performance. This could yield a greater understanding of the plant behaviour and
maybe, supply useful information to module manufacturers. Having measured, in 2001, the I-V
characteristics of each module of the plant, it has been possible to make a correlation between these
data with the results of visual inspection and IR analysis.

Results

Plant / module power

Since the plant realization in 1982, the system configuration has been changed three times because of
inverters substitutions.

The initial nominal plant power was equal to 10.7 kW. In 1983, the outdoor I-V characteristics of the 24
strings were performed, giving a total power output, after STC correction and comparison with indoor
 S-3
Mean Time Before Failure of Photovoltaic modules (MTBF-PVm)

measurements of a batch of modules, equal to 9.8 kW [1]. Having no data of initial strings
measurement or measured maximum power value of each module, it has not been possible to state if
the –8.4% difference from the nominal power existed at the beginning, or it is attributable to module
degradation and/or wiring losses.

In January 2002, measurement of the 12 strings was performed. The sum of the power output of all
strings, after correction at STC, was equal to 8.33 kW. Due to the change of the plant configuration,
which occurred in 1992, a comparison between the current results and the ones obtained in 1983 was
impossible. However, a rough estimate of the annual degradation rate of the module power, since
1983, is shown in Table I.

Measured plant power Module power (calculated)

Year N° of modules
[kW] [W]

1983 9.8 288 34.0

2002 8.3 252 32.9

∆ module power2002-1983 - 3.2%

Module power degradation/year (since 1983) - 0.2%

Table I: estimate of module power degradation based on outdoor string

measurements performed in 1983 and 2002.

In March 2001, the entire plant was dismantled to perform indoor electrical characterization of all 252
modules. The mean maximum power was equal to 33.1 W, which is in good agreement (∆ = +0.6%)
with the 32.9 W obtained through power estimation of the 12 strings (see Table I).

Visual defects
Several types of defects were detected during intensive visual inspection of all plant modules. A
comparison with previous visual analysis results showed that some defects were present during the
first years of module exposure. For example, a 1985 TISO publication stated that 20% of modules had
cracked cells and that this aspect did not affect the plant performance [2]. At present, this percentage
has increased to 22%; with the new breakages corresponding to hot-spotted cells.

The browning of cell electrical grid was already detected in 1983; in particular, this phenomenon was
associated with the cells sited in front of the junction box. As of today, this colour change has also
affected the cells close to the edges of the modules. A TISO internal report of 1984 referred to a
document, edited by Arco Solar, in which the browning of electrical grid, previously detected in other
plants, was attributed to a PVB encapsulant oxidation accompanied with a catalytic reaction with
decomposition products of poly-vinyl-alcohol (PVA). The manufacturer affirmed that this reaction did
not involve the silver grid contacts; hence the cell electrical characteristics were not modified. No
significant effects were mentioned in regard to PVB optical transmittances.

The most evident visual defect is the yellowing of encapsulant; present, in 1985 [2], on 50% of
modules, it currently affects, with different intensities, the 98% of the plant. The reason why some
modules have remained completely white is unknown. The correlation between electrical
characteristics and encapsulant discoloration showed that completely yellowed modules present
higher loss in Isc (10-13% less than the nominal value) with respect to the white or partially yellowed
ones (6-8% less than the nominal value). Effect of PVB yellowing on module power degradation is less
precisely quantifiable, due to the presence of additional defects, which could affect module efficiency.
 S-4
Mean Time Before Failure of Photovoltaic modules (MTBF-PVm)

Spectral response measurements on white and yellow modules were performed; even if results gave
no significant differences in mismatch factor (0.9994 non-yellowed and 0.9993 heavily yellowed), the
comparison of the two absolute spectral responsivity graphics, showed a lower response of
discoloured module between 400nm and 700nm (see Figure 1). The I-V characteristics of the same
two modules are compared in Figure 2 (∆Pmax = 4.5%, ∆Isc = 1.3%).

0.6

0.5
Absolute SR [A/W]

0.4

0.3

White
0.2
Yellow

0.1

Figure 1: comparison of the absolute

0 SR of a white module and a heavy
300 400 500 600 700 800 900 1000 1100 1200

Wavelength [nm]
yellowed one.

2.5

1.5
Current [A]

White
1

Yellow

0.5

Figure 2: I-V characteristics of a

0
0 3 6 9 12 15 18 21 24 white module and a heavy yellowed
Voltage [V] one.

One of the major disadvantages of PVB as a module encapsulant was its propensity for water
absorption, which complicated handling procedures and occasionally produced voids on lamination.
Moisture sensitivity dictated the use of a metal foil backsheet, as was made for ASI 16-2300 modules.
Such a precaution did not prevent PVB delamination; since 1986, the presence of water infiltration and
sealant penetration in some modules was documented [3]. At present, 233modules (92% of the plant)
show encapsulant delamination along edges and 191 (76% of the plant) sealant diffusion.

Even if in 69 modules (27% of the plant) delamination represents a major defect, as it forms a
continuous path between frame and circuit (as defined by IEC 61215), no insulation failures was
detected both in wet and dry conditions.

Electrical characterization of differently delaminated cells showed that performance losses of single
cells are proportional to their affected area; the presence of partially delaminated cells does not
necessarily significantly affect the overall module efficiency. Particular attention has to be paid to
modules where delamination affects the entire area of hot-spotted cells. In 20 years, 3 modules (1.2%
of 252) presented such a defect: 1 of them was replaced in 1997, while 2 are still working and have a
maximum power respectively of 26.4W and 28.2W (-20.2% and –14.8% with respect to the actual
mean module power).

A quite frequently detected defect is the bad seal of the junction box on the tedlar backsheet with the
consequent risk of detachment when opening, and loss in insulation resistance.
 S-5
Mean Time Before Failure of Photovoltaic modules (MTBF-PVm)

This last aspect is important, as it causes terminal oxidation that leads to higher electrical resistance
and could provoke detachment when wiring.

For example, the IV curves shown in Figure 3

were obtained after a sequence of
measurements on one module, executed one
after the other, without apply any change.
Differences were due to a partially detached
terminal connector.

Figure 3: results of a series of I-V measurements

on a module with one partially detached
connector

In 2001, the comparison between the defects appeared after repeated accelerated ageing tests (damp
heat test and thermal cycles according to the International Standard IEC 61215) [4] and the ones
provoked by ageing in field conditions, evidenced two significant differences:

1. thermal cycles and damp heat did not provoke delamination

2. no naturally aged modules exhibited tedlar detachment [5].

After one year, the backsheet detachment in the bottom part of two modules has been detected. Even
if tedlar is still attached along the edges, the eventual exposition of the aluminium foil has to be kept
under control, as it could represent an electrical safety hazard.

Infra-Red analysis
The infrared analysis of the plant enables the thermal uniformity of the modules to be checked at
normal operating conditions and, consequently, to detect the presence of hot-spots. The hot-spot
resistance of a PV module is a quality feature of the device and is essential to its lifetime.

In all modules of the plant, the cell in front of the junction box always exhibited an increase of about
4°C, in respect to the other cells of the module, due to the insulating effect of the junction box. Infrared
analysis of the plant detected the presence of hot-spots (in excess of the “normal” 4°C increase) on 67
modules (26% of 252). The hot-spots are likewise always found on the cell in front of the junction box
and show an overheating of about 10°C with respect to the rest of the module surface, whose
temperature distribution is very regular.

Modules with hot-spots show the highest power degradation; the worst affected module of the plant,
whose maximum power is equal to 26.1W (–21.1% with respect to the actual mean module power),
has a hot-spot.

Summary
The main defects affecting ASI 16-2300 modules, and their effects on module efficiency were
identified and indoor performance measurements of all 252 modules of the plant were executed.
 S-6
Mean Time Before Failure of Photovoltaic modules (MTBF-PVm)

These data, together with the visual inspection and IR results are summarized in Table II. Having no
initial measured power value for all of the modules, the data have been compared to the
manufacturers nominal value (Pn) of 37W.

∆Pmax_Pn <10% ∆<20

10%<∆ >20%

N° of modules 143 92 17

YELLOWING 97% 99% 100% Table II: correlation between module

efficiency and defects distribution.
DELAMINATION 94% 90% 94%
DEFECTS

Percentages in italic refers to the n° of

nd
HOT-SPOT 7% 43% 100% modules in the 2 row of the table. An
example: 100% of the 17 modules which
TERMINAL actual Pmax differs more than 20% with
42% 52% 29% respect to Pn show PVB YELLOWING and HOT-
OXIDATION
SPOTS.

Plant production monitoring

The continuous monitoring of the individual string’s energy production enables analysis of the string
behaviour with time and in different weather conditions, so aiding fault detection. For example, in July
2002, one string was found to be not working (Figure 5). After checking of all 21 modules, one burned
cable in the connector used to wire the modules together was detected (Figure 4). This fact clearly
shows that not only module ageing could affect the plant performance, but also the deterioration of all
system components.
1000

Irradiance

800

Strings power
Power [W], Irradiance [W/m²]

600

400

200

String with damaged connector

Figure 4: burned cable and Figure 5: string fault detected through the analysis of the
damaged connector daily individual string’s power production

At this point, particular attention has to be paid to the inverter unit. The analysis of the daily production
data recorded since June 2000, revealed a considerable difference between the two sub-fields of the
plant (positive and negative – 6 strings each), starting from November 2001 (Figure 6). Due to the
noticeable difference between the Performance Ratios, I-V performance measurements of the positive
and negative arrays were executed. The results, showing no relevant differences between the sub-
fields output powers (4.30 kW and 4.24 kW), indicated that discrepancy between the two Performance
Ratios has to be attribute to an inverter anomaly.
 S-7
Mean Time Before Failure of Photovoltaic modules (MTBF-PVm)

INVERTER ANOMALY

0.9

0.8
PR [-]

0.7

Indoor measurements

0.6

0.5
06/00 12/00 06/01 12/01 06/02

Date (mm/yy)

Figure 6: Performance Ratio of positive and negative sub-fields of the 10 kW plant.

Conclusions

After 20 years of service, the Arco Solar ASI 16-2300 modules of the 10 kW TISO plant show several
signs of degradation. Yellowing of the PVB encapsulant and hot-spots have affected the module
efficiency; in particular, hot-spot represents the principal cause of power degradation.

During the last year tedlar backsheet detachment has been detected in two modules. The same defect
appeared on modules subjected to repeated damp heat test (accelerated ageing tests, International
Standard IEC 61215).

Nevertheless, the results of the indoor performance measurements of all plant modules indicate that
the ASI 16-2300 modules are still working in a very satisfactory manner (only 6% of the modules show
a power loss higher than 20% with respect to the nominal value).

By means of the results obtained during outdoor I-V measurements of the strings executed in 1983
and 2001, the annual power degradation of the modules has been estimated to be equal to 0.2%.

As well as the analysis of failure mechanisms affecting the modules, an important aspect to consider
in the plant management, is the maintenance of other system components, like the inverter unit. Since
the inverter anomaly, which occurred in November 2001, the plant performance has been strongly
influenced. As the mean lifetime of an inverter is about 8-10 years, the LEEE-TISO is now considering
the possibility of the inverter substitution.

It is reasonable to assume, on the basis of results obtained from accelerated lifetime tests [4] and
plant analysis that the modules could continue to provide useful electrical power for another 10-15
years. This estimate significantly changes the economy of the system, as it greatly extends the mean
lifetime of the plant.
 S-8
Mean Time Before Failure of Photovoltaic modules (MTBF-PVm)

Perspectives for 2003

The MTBF project will finish on April 2003; within this deadline the inverter substitution is foreseen.
Having enough stocked modules to reach the initial number of devices constituting the plant (288), it is
intention of the LEEE-TISO to change again the system configuration and to use three 2.5 kW
inverters (one for each field) instead of a 10 kW one.

A final detailed visual inspection and an infra-red analysis will be executed before the end of the
project; the outdoor performance measurement of the strings it is also foreseen.

References and publications

[1] G. Blaesser, H. Ossenbrink, J.Verbaken; “Power measurements at the 10 kW photovoltaic

demonstration field TISO”; Technical Note No. I.07.05.83.06 (January 1983).

[2] M. Camani, P. Ceppi, D. Iacobucci; “Operational characteristics of the grid connected

photovoltaic plant TISO 15”; Mediterranean Electrotechnical Conference IEEE, Madrid (1985).

[3] P. Ceppi, M. Camani, D. Iacobucci; “Behaviour of the modules of the photovoltaic plant TISO
th
15”; 7 EPVSEC, Sevilla (1986).

[4] J. Bishop, L. Rigolini, W.J. Zaaiman, A. Realini; “Repeated Accelerated Lifetime Testing of
Exposed PV Modules”; Technical Note No.I.00.42 (February 2000).

[5] A. Realini, E. Burà, N. Cereghetti, D. Chianese, S. Rezzonico; “Study of a 20-year old PV

th
plant (MTBF project); 17 EPVSEC, Münich (2001).

2002 – Conferences and publications:

th
§ EuroSun 2002 - The 4 ISES Europe Solar Congress (Bologna, 23-26 June 2002): "The
oldest grid-connected PV plant in Europe: study and first results (MTBF project)”; paper and
oral presentation.

§ PV in Europe Conference and Exhibition - From PV Technology to Energy

SolutionsTechnology to Energy Solutions (Rome, 7-11 October 2002): "TISO 10 kW plant: the
oldest grid-connected PV system in Europe"; paper and oral presentation.

Publications concerning the MTBF project are available on the TISO Internet site:
http://www.leee.dct.supsi.ch

Acknowledgements

This project is financially supported by the Federal Office for Education and Science (BBW, Bern) and
the European Union (Fifth Framework Programme).

View publication stats