LIFE EXPECTANCY OF PV INVERTERS AND OPTIMIZERS IN

RESIDENTIAL PV SYSTEMS

Dr. Christof Bucher1, Dr. Jasmin Wandel2, David Joss1,

Bern University of Applied Sciences / Department of Engineering and Computer Science
1Institute for Energy and Mobility Research / Laboratory for Photovoltaic Systems
2Institute for Optimisation and Data Analysis

Jlcoweg 1, 3400 Burgdorf, Switzerland, [email protected]

The average Time To Failure (TTF) of PV inverters and PV optimizers is investigated in this paper. The focus is on
residential and small commercial systems. The data used in this paper includes 1195 PV systems consisting of 2121
inverters and 8542 optimizers. The data is obtained by means of an online survey sent to the system operators (343
systems) and by analysing system service databases of professional PV system installers and operators (852 systems).
Since most of the systems examined are still in operation and have never failed so far, the results are presented using
Kaplan-Meier survival curves. After 15 years, 34.3 percent of inverters show a first failure. The most important factors
influencing the TTF are the installation location of the inverter (indoor installations have a lower TTF than outdoor
installations), the manufacturer and the inverter topology (installations with optimizers have a lower TTF than
installations without optimizers).
Keywords: Life expectancy of PV inverters, Time to Failure (TTF), survival curve, Kaplan-Meier estimator

1 INTRODUCTION reached after 99 years (Germany 47 years).

An IEEE publication from Aalborg University [4]
PV inverters are typically said to have a life deals with the influence of PV module degradation on the
expectancy of 15 years and must therefore be replaced lifetime of inverters. The publication assumes inverters to
once in the service lifetime of a typical PV system [1]. have a higher life expectancy if a lower PV power capacity
Accordingly, the warranties for inverters usually only is connected to the inverter. A system in Denmark is
cover about half the time that the performance warranties compared with a more degrading one in Arizona
for PV modules are valid for. However, optimizer (Denmark: 0.15%/a, Arizona: 1%/a). The lifetime of the
manufacturers usually provide longer warranties than inverter in Arizona is underestimated by 54% (7 years).
inverter manufacturers. In [5] from Aalborg University it is shown that DC
The cost of maintaining and replacing inverters oversizing can have an effect on the lifetime, depending
accounts for the largest share of the operating and on the location, while the effect is more serious at other
maintenance costs (O&M) of a PV system. Therefore, it is locations. In Arizona, for example, the lifetime hardly
advantageous for investors and operators of PV systems to changes due to the persistently high irradiation values,
have an accurate estimate of the expected lifetime of an while in Denmark it deteriorates significantly with
inverter and to know the factors that influence its lifetime. increasing oversizing. The main cause is the increasing
thermal load.
1.1 Study to assess life expectancy of PV inverters According to Jordan [6], inverters are still the most
In a study by the Bern University of Applied Sciences, frequently failing components of PV systems, accounting
the life expectancy of PV inverters is investigated. Due to for 4–6 % of all cases evaluated. The failure frequency of
various practical limitations such as a lack of information residential PV systems is slightly lower than that of
on inverter repairs, the study uses the time to energy- commercial or industrial systems, possibly due to more
related failures (called TTF in this paper) as the relevant reliable inverters. On the other hand, the average
parameter. production loss of large-scale industrial systems is much
lower than that of commercial and residential PV systems.
1.2 Literature research According to this study, the failure frequency of module
P. Hacke et al. [2] makes a detailed inventory of the inverters and power optimizers, which together constitute
PV industry in terms of reliability and safety. The a device class, is underestimated.
publication assesses inverter reliability in relation to Herz et al. [7] focus mainly on the detailed type of
existing regulations and standards. The study shows that failure, but do not investigate correlations of specific
there are big differences in the information on the quality situations or the overall probability of a failure occurring.
of inverters and that these can have an impact on O&M Häberlin et al. [8] analysed the inverter reliability of
costs. The recommendation is made to collect financial PV systems in Switzerland between 1992 and 2009. He
data for O&M. To improve quality standards, it is noted that the average age of the inverters has stagnated at
recommended that design validation tests should be around 110 months. However, many of the inverters had
conducted under realistic environmental and operation been repaired in the first years after commissioning. The
conditions including end-use field conditions. TTF definition used in this paper is therefore assumed to
In Gatla [3], the lifetime of a single-phase 3kW be lower than the life expectancy calculated by Häberlin.
inverter (600V/30A IGBT) is simulated with PLECS. The The discontinued data collection does not allow any direct
differences in the top ten PV sales countries are compared. conclusions to be drawn about current inverter
This reveals that not only the maximum temperature at the technologies.
site but also the temperature and irradiation cycles have an
influence on the service life. At the system level, in
Australia an average of 10% of the devices failed after 11
years, while this rate for England is theoretically only
2 METHODOLOGY inverter was running again. This is not classified
as a malfunction.
2.1 General approach o Inverter was making strange noises, so it was
The aim of this study is to quantify inverter failures in replaced. This was classified as a malfunction.
the field. The focus is on systems in residential buildings.
The aim is not to specify the details of failures, but rather 2.4 Survival curves
their year of occurrence and their relative frequency In order to assess the reliability of the inverters, the
compared to inverters without failures. duration until the first energy-relevant fault occurs is
In addition, local conditions such as sizing ratio (SR), considered. This duration is defined as Time to Failure
installation location, type of PV system and others are (TTF).
evaluated to investigate their influence on the inverter The TTF is calculated and plotted using the concept of
lifetime. the nonparametric Kaplan-Meier estimator together with
the corresponding Kaplan-Meier survival curve. The
2.2 Data collection survival curve shows what percentage of the population is
Data is gathered from three sources between March still alive after how many years. Applied to this study, this
and May 2022: means what percentage of the inverters has still not had
1. By means of an online survey, mainly filled in by any yield-relevant faults after the specified number of
private owners / operators of PV systems (343 years, while it adjusts for so-called censored data points.
systems). This means that the Kaplan-Meier estimator accounts for
2. Data of the PV system portfolio of a professional the fact that a lot of inverters did not show a failure up to
PV system operator (83 systems). the time point of investigation. However, those inverters
3. Data of the PV system portfolio from are still included in the analyses with the given observation
professional PV system installers and operators time used as lower bounds for the actual failure time
(769 systems). sometime in the future. With the help of the survival curve
method, various other information can be derived:
2.3 Data gross error detection and correction • The Kaplan-Meier estimator can be adjusted for
In the context of this survey, a large part of the data different influencing factors, e.g. to the year of
was collected with the help of people who know the commissioning, the inverter manufacturer or the
history of the PV systems well. These are, in particular, the nominal power ratio.
system operators. However, they are often not familiar • Statistical measures can be read from the survival
with the technical details. For this reason, the survey gave curves. For example, it can be calculated what share
respondents the opportunity to add their own comments to of inverters has a first fault after five years. Or it can
the technical data. With the help of these comments, the be seen what percentage of inverters reach their tenth
data was then manually corrected where possible. The year without any energy-relevant error.
corrections made are described below. The survival curve also takes into account how many
inverters are still under observation in the corresponding
2.3.1 Deletion of incomplete or unrealistic data-sets: year of operation (“samples at risk”). This means that,
• Incomplete and unanalysable data records were although the majority of the inverters have never had a
deleted. If data-sets are incomplete but can still be malfunction, it can be estimated in which year of operation
evaluated (e.g. inverter manufacturer is missing) a certain percentage of the inverters will have their first
they were not deleted completely, but they were only expected malfunction. However, because only little data is
included in the evaluations for which the missing available for old inverters, the uncertainty in the survival
information is not relevant. Therefore, not all curve increases with increasing inverter life. The
investigations are based on the same number of data- confidence interval is shown by the coloured area in the
sets. diagrams.
• Unrealistic entries deleted (e.g. if the first
malfunction occurred before the year of 2.5 Avoidance of cross-correlations between dependent
commissioning). variables
In the survey, a set of system parameters (variables) is
2.3.2 Creation of a consistent nomenclature gathered for each inverter. Several of these variables
• Comma replaced with full stop. cannot be assumed to be independent. For example,
• Obviously wrong units replaced / corrected. preliminary results have shown that both large inverters
• Spelling of products and companies standardised. and inverters installed outside a building have a shorter
TTF than small indoor inverters. In addition, small
2.3.3 Correction of obvious errors inverters are more likely to be installed indoors, while
• Inconsistent data corrected or deleted (e.g. obvious large inverters are predominantly installed outdoors.
factor 1000 for power values). Comparing the TTF of inverters installed indoors and
• Incorrect category assignments of inverters outdoors therefore shows results that do not reflect
corrected. For example, the category “hybrid causality.
inverter” is not listed, but recorded as “string or The qualitative dependency of variables is determined
multistring inverter”. by means of mosaics. As shown in Figure 1, protected
• Uniform power information used for known outdoor inverters are by far the biggest category for large
products (e.g. AC power of inverters). inverters (> 20 kVA), whereas the share of protected
outdoor inverters is a relatively small category for small
• Faults considered or not based on description. Two
inverters (< 20 kVA). Figure 2 shows, that manufacturer
fictitious examples:
m1 is mainly represented with recent inverter installations,
o Fuse has blown, after switching back on the
whereas old installations are dominated by manufacturer
m5 and to some extent m4. Table I: Data sources (abbreviations: sys. = system; inv.
In order to exclude the influence of dependent = inverter, priv. = private, prof. = professional, instal. =
variables, the method proposed in [9] is applied. installer)
According to Terry M Therneau et al., data is adjusted for

Total power
several confounding factors by a (inverse probability)

inv. (kVA)
No. of sys.

No. of inv.
No. of inv.

sys (kVA)
Power per

Power per
reweighting of the observed inverters before survival

per sys
(kVA)
curves are computed. For example, it is known that the
Source of
place of installation (indoor vs. outdoor) depends on the
data
confounders power and manufacturer of the inverter. To
adjust for these two confounders, the observations from Priv. owner 343 6172 546 1.6 18 11.3
the comparison indoor vs. outdoor are reweighted such Prof. owner 83 7050 314 3.8 84.9 22.5
that the two comparing groups are balanced with respect
to the confounding variables power and manufacturer of Prof. instal. 769 13443 1261 1.6 17.5 10.7
the inverter.
The inverter manufacturers in this survey (Figure 3)
are not representative of today’s PV market. The
manufacturers with most inverters analysed in this paper
are Fronius, Huawei, Kostal, SMA, SolarEdge and
Sputnik (alphabetic order, not corresponding to m1 to m5).
The inverter types are predominantly string and multi
string inverters (Figure 4). Central inverters are
underrepresented in a global context. In addition, several
of the central inverters examined in this study are old, low-
power models that are now usually replaced by
transformerless multistring inverters.

Figure 1: Mosaic showing inverter manufacturer, place of

installation and inverter power.

Figure 3: Number of inverters by manufacturer.

Figure 2: Mosaic showing inverter manufacturer, place of

installation and inverter power. Figure 4: Number of inverters by type of inverter.

The oldest inverters investigated in this study are from

3 RESULTS the early 1990ies (Figure 5). However, most of the
The quantitative results of the survey and data inverters were commissioned between 2008 and 2013
gathering are presented using pie charts, histograms and (first big installation boom in Switzerland due to the feed-
other descriptive graphics. The TTF of the inverters are in tariff introduced in 2009) and after 2017 (mainly driven
shown using Kaplan-Meier survival curves. by self-consumption).
As this paper is concerned with residential systems,
3.1 Description of the data most inverters have power ratings below 50 kVA (Figure
In the following figures (Figure 3 to Figure 10), the 6). The biggest share of inverters has a power rating of
numbers of inverters are listed in different categories. 10 kVA to 15 kVA.
Table I gives an overview of all data available in this The distribution of Sizing Ratios SR (defined in this
paper. It shows the clear tendency, that data systems from paper as AC power divided by DC power) is shown in
professional owners have more inverters per system and Figure 7, peaking at roughly SR=0.9. Although it was
higher power ratings than data from private owners and assumed SR for new systems would change over time, the
data from professional installers. distribution stayed fairly uniform throughout the observed
period.
 Figure 8: Number of inverters by place of installation.

Figure 5: Number of inverters by commissioning year.

Figure 9: Number of inverters by tilt angle of the PV

modules (typically corresponding to rooftop tilt angle).
Figure 6: Number of inverters by nominal inverter power
(AC power in kVA).

Figure 10: Number of inverters by type of installation.

3.2 Survival curves

The survival probability of all inverters is given in
Figure 11. Because only a few inverters are older than 18
years, the survival curve is only plotted until year 18. In
Figure 7: Number of inverters by sizing ratio. year 18, only 15 inverters were still in operation, resulting
in a large confidence interval in the survival curve. After
The number of inverters by place of inverter 18 years, the survival probability of all inverters is still
installation is given in Figure 8. About half of the inverters 59.1% with a confidence interval of [52.1%, 67.0%].
are installed inside buildings. Only a minor share is Optimizers could only be observed during 8 years.
installed outdoors, mostly protected from rain and direct Figure 12 shows the survival curve of inverters with and
sunshine. For a third of systems, the place of installation is without optimizers. Throughout the observation time
unknown. inverters with optimizers show roughly twice as many
The tilt angles of the PV modules connected to the energy-relevant failures than inverters without optimizers.
inverters are given in Figure 9. The type of the system For individual optimizers, the survival curve is
(installation) is given in Figure 10. However, it was found relatively flat (Figure 13). However, as optimizers are
that neither of these categories have a measurable impact normally used in larger quantities even for small systems,
on the TTF of the inverters. their TTF should rather be calculated for a whole fleet of
optimizers than for a single device. Figure 14 shows the
TTF after 5 and 10 years respectively for a given number
of optimizers. Comparing the survival probability of a fleet
of optimizers in Figure 14 with inverters without
optimizers in Figure 12, a set of roughly 20 optimizers
shows the same TTF after 5 and 10 years respectively as
an inverter without optimizers. However, this comparison
is only indicative, as optimizers cannot be operated
without inverter.

Figure 14: Survival probability after 5 and 10 years for a

given number of optimizers (without inverter).

The survival curves of inverters split into three power

categories are shown in Figure 15. A trend of more failures
in large inverters can be observed. However, if these
Figure 11: Survival probability of all inverters. curves are adjusted for inverter manufacturer and place of
installation (indoors/outdoors), this trend is not confirmed
(Figure 16). This is most likely based on the fact that
inverters installed outside a building have a lower TTF
than inverters installed inside a building (Figure 17).

Figure 12: Survival probability for systems with and

without optimizers.

Figure 15: Survival probability for inverters of different

power categories.

Figure 13: Survival probability of all optimizers (without

inverter).

Figure 16: Survival probability for inverters of different

power categories adjusted for manufacturer and place of
installation.

Figure 17 shows the difference between the survival

curves of inverters installed inside and outside a building.
Because these variables are correlated to the inverter
manufacturer and the nominal inverter power, an adjusted
graph is shown in Figure 18. However, both figures show
higher survival probabilities for systems installed inside a
building. In the adjusted survival curve, this effect only
becomes apparent after eight years.

Figure 19: Survival probability of inverters installed

inside the building.

Figure 17: Survival probability of inverters installed

inside and outside of a building.

Figure 20: Survival probability of inverters installed

outside the building.

No clear conclusion is possible from the survival

curves of inverters installed with different sizing ratios
(Figure 21). In the data-set, it is not the heavily loaded
inverters with sizing ratios < 0.9 that have the lowest
Figure 18: Survival probability of inverters installed survival curve, but rather inverters with sizing ratios
inside and outside of a building, adjusted for manufacturer between 0.9 and 1.0. However, the confidence intervals
and inverter power. overlap. This can indicate that too little data might be
available to make a definite statement about the influence
Both inside and outside a building, inverters can face of the sizing ratio on the TTF. However, a clear conclusion
different situations. The survey asked whether installation is not possible.
rooms were heated or not (Figure 19). Outside the
building, it was asked if the installation was protected
against rain and direct sunlight (Figure 20). Both indoors
and outdoors, stable conditions (heated room, protected
installation site) seem to have a positive effect on the TTF
of inverters. However, only little data is available to
quantify these effects. The confidence intervals are
overlapping in the outside case. Especially in this case, a
clear conclusion therefore cannot be drawn from the
available data.

Figure 21: Survival probability of inverters with different

sizing ratios.
 The commissioning year has an influence on the 4.2 Inverter as the system boundary
survival probability of the inverters under investigation The relevant unit of this study is one inverter. For PV
(Figure 22). Old inverters seem to have less energy- systems with optimizers, all optimizers are counted as part
relevant failures in the beginning of their life span. The of one single inverter. Few systems in the scope of this
authors of this study cannot say whether this observation study have many inverters with optimizers (>5); one case
is biased by an effect not taken into account in this study, even had very many (>20) inverters with optimizers. In the
e.g. repair measures forgotten by the data providers. case of the reported optimizer failures, it is not clear which
inverter they were connected to. However, since the
failures are assigned to one specific inverter in this study,
it is unclear whether the reported defective optimizers are
to be considered multiple failures on one inverter or many
individual failures on several inverters. It was assumed
that the optimizers with failures were randomly distributed
within the PV systems. However, since the majority of the
reported systems only involve one inverter, this error can
be assumed to be insignificant.

4.3 Cause of failure unknown

The causes of the failures and defects are unknown.
For example, it is not always possible to distinguish
between a minor fault (such as the failure of a display) and
a total loss (such as a full inverter breakdown including the
failure of the display).
Figure 22: Survival probability of inverters with different
commissioning years. 4.4 Rating of optimizer and inverter failures
The failure of an optimizer usually results in less yield
Finally, the survival probability of inverters from loss than the failure of an inverter. However, in the
different manufacturers is shown in Figure 23. Although survival curve (TTF), both failures are represented in the
the confidence intervals of the curves are partly same way. If the failure of an optimizer triggers a
overlapping, the percentage of inverter failures observed maintenance service, the financial impact for the system
varies by about a factor of three. The differences between owner is close to the financial impact of an inverter failure.
the TTF varies thus remarkably between different However, if only the power of one module is lost and no
manufacturers. maintenance is done, the impact of an optimizer failure is
lower than that of an inverter failure. In the course of this
study, the tendency was noted that individual defective
optimizers were replaced in each case. Optimizers that are
defective and have not been replaced were probably not
detected within the scope of this study.
Under this consideration, however, some
communication failures would also have to be evaluated
as TTF as they can trigger maintenance. In this study, this
is neglected, because the reason for communication errors
is normally outside the PV system (e.g. poor mobile phone
reception or security updates on the internet connection).
While inverters are usually repaired after the first fault
occurs, even relatively young power optimizers with
defects are not repaired but disposed of.
The number of optimizers resulting in the same TTF
as one inverter is biased because the possible inverter
Figure 23: Survival probability of inverters from different failure is neglected in the case of optimizers. In reality, the
manufacturers. TTF of inverters with optimizers would therefore be
slightly lower than indicated in this study.
Several other questions have been asked in the survey In general, the system comparison between plants with
to classify the PV systems, such as tilt angle of the roof or and without optimizers is critical because the monitoring
type of roof. However, the resulting data does not show and maintenance concept is different. Monitoring concepts
relevant differences for such properties. Therefore, the with optimizers have the potential to detect more failures,
corresponding survival curves are not shown in this paper. which can trigger more maintenance activities, but also
provide input for optimised maintenance concepts.
Inverter and optimizer technology has evolved greatly
4 DISCUSSION in the past. While the inverter industry is diverse and has
decades of experience, optimizers are relatively new to the
4.1 Shortcomings of this study mass market and dominated by only a few manufacturers.
Although the method chosen in this study allows the Therefore, the learning curve of inverters can be assumed
TTF of inverters to be quantified, there are several to be more advanced than the learning curve of optimizers.
influences on the data that may negatively affect the The results observed in this study might change with future
results. Possible sources of error are discussed in the inverter and optimizer generations.
following sections.
4.5 Inverter replacement without failure the scope of this study.
Especially in older systems, inverters are sometimes When optimizers are replaced in larger systems, it is
replaced pre-emptively without any immediate technical not always clear how many optimizers were replaced. If
reason. In this study, this is counted as an inverter failure. no information is available, it is assumed that only one
This consideration is correct with regard to the economic optimizer was replaced.
calculation. However, it underestimates the life The data has shown some large (45 inverters, SR>1.5)
expectancy of the inverters. and small (27 inverters, SR<0.5) sizing ratios. It is not
known for sure if this data is realistic or not. However, it
4.6 Repair vs. end of life seems unrealistic and has therefore been deleted for the
It cannot be said without doubt whether an inverter sizing ratio survival curve evaluation.
failure indicates the end of the life of an inverter. Because It was reported that many inverters were repaired in
installers and manufacturers often install second-hand the years between 1992 and 2010 [8]. It is not clear if
replacement devices, even an inverter replacement does repairs of old inverters are always reported in this survey.
not necessarily mean the end of the inverter’s service life.
Due to this uncertainty, the TTF is defined as “time to first 4.10 Misinterpretation due to correlated variables
energy-relevant fault” in this study. As described in section 2.5, some survival curves are
adjusted for possible confounders. However, not all
4.7 Non-representative data set possible influencing factors could be identified and
It can be assumed that the inverter brands and eliminated. Therefore, it cannot be said for the graphs that
manufacturers that feature in the online survey are the correlations found in the results correspond to the
represented roughly proportionally to their market shares causes for the observed TTF.
in German-speaking Switzerland. In the data from the
institutional plant operators, however, individual products
are naturally overrepresented. Because the inverters of 5 FUTURE RESEARCH
different manufacturers have different TTFs, this leads to The results of this study clearly show that the TTF of
a distortion of the TTF of the entire inverter fleet. inverters depend on many different factors. It was possible
It is not known whether one group of system operators to identify and quantify several of these factors. However,
felt particularly addressed by this survey. It is conceivable many questions remain open and many shortcomings
that system operators with defective systems would tend could not be excluded (see chapter 4). Therefore,
to fill out this survey more often than operators of systems researchers are motivated to repeat this study and expand
without inverter defects. However, the comparison of the the data set to improve the quality and quantity of the
data from the online surveys with the data from results. The questionnaires should also be improved in
institutional operators does not reveal a strong bias in this order to reduce possible misinterpretations.
respect. On the other hand, it can be assumed that
professional plant operators tend to pay more attention to
quality and optimum operating conditions and therefore 6 CONCLUSIONS
tend to have lower inverter failure rates. Data on inverter and power optimizer failures in PV
Institutional system operators sometimes own systems systems were collected from PV system operators.
that were predominantly built in certain time periods (for The inverters and optimizers in this study represent a
example, in the years 2008–2014 when Swiss feed-in broad, albeit not representative, cross-section of the
tariffs, or FIT, were introduced). As a result, installations inverters and optimizers used in PV systems on buildings
from these periods are overrepresented among these data in Switzerland and Europe. Inverters from the
providers. This leads to correlations in the statistics that manufacturers Fronius, Sputnik and SMA are most
have no causal link with the data providers. widespread, while most optimizers are produced by
SolarEdge. Compared to the European inverter market, the
4.8 Unknown initial failures products from Sputnik (SolarMax) are probably
It is assumed that initial inverter failures are often not overrepresented. In an international comparison, the
known to the building owner because they are either fixed European manufacturers are overrepresented.
by the installation companies or the inverters are replaced The results across all inverters and power optimizers
during installation. In the case of optimizers in particular, examined show that over 65% of inverters do not have a
it must be assumed that initially defective devices were not yield-relevant fault by their 15th year of operation.
reported in this study, as their replacement during plant Furthermore, the investigation shows that the TTF is
commissioning causes only minor additional work and dependent on various factors, in particular:
costs. • Manufacturer: Different manufacturers have different
TTF.
4.9 False information • Power: More powerful inverters have shorter TTFs;
It must be assumed that not all the information individually considered power optimizers have the
provided for this study was correct. The resulting errors longest TTFs. However, the overall reliability of a PV
cannot be quantified. However, from the personal system increases significantly with the reduction in the
comments made in the survey it can be assumed that most number of power electronic components.
people correctly understood the questions. Obvious • Installation location: Outdoor power electronics have
misunderstandings were corrected manually. a shorter TTF than indoor power electronics.
Certain errors are interpreted by the system operators • Topology: In PV systems with inverters with power
as inverter errors, though fact they are not inverter errors. optimizers, the first fault occurs earlier than in PV
One example is earth faults in the module field, which lead systems that do not have power optimizers.
to an error message on the inverter. As far as possible,
these faults were masked as such and not evaluated within
Care should be taken not to overinterpret the results.
Various factors play a role in the reliability of a PV system
that are not taken into account in this study. It is also
possible that certain influencing factors and correlations
were not found within the scope of this study and were
therefore wrongly neglected.

7 REFERENCES

[1] Peter Toggweiler, Betriebskosten von

Photovoltaikanlagen: Solarstromanlagen effizient
betreiben. [Online] Available:
www.bundespublikationen.admin.ch.
[2] P. Hacke, S. Lokanath, P. Williams, A. Vasan, P.
Sochor, G. TamizhMani, H. Shinohara, and S.
Kurtz, “A status review of photovoltaic power
conversion equipment reliability, safety, and
quality assurance protocols,” Renewable and
Sustainable Energy Reviews, vol. 82, pp. 1097–
1112, 2018.
[3] R. K. Gatla, S. S. Kshatri, P. Sridhar, D. S.
Malleswararao, D. G. Kumar, A. S. Kumar, and J.
Lu, “Impact of Mission Profile on Reliability of
Grid-Connected Photovoltaic Inverter,” JESA, vol.
55, no. 1, pp. 119–124,
https://www.researchgate.net/profile/Ranjih-
Gatla/publication/359465305_Impact_of_Mission_
Profile_on_Reliability_of_Grid-
Connected_Photovoltaic_Inverter/links/623f434a5
7084c718b6a0b85/Impact-of-Mission-Profile-on-
Reliability-of-Grid-Connected-Photovoltaic-
Inverter.pdf, 2022.
[4] A. Sangwongwanich, Y. Yang, D. Sera, and F.
Blaabjerg, “Lifetime Evaluation of Grid-Connected
PV Inverters Considering Panel Degradation Rates
and Installation Sites,” IEEE Trans. Power
Electron., vol. 33, no. 2, pp. 1225–1236, 2018.
[5] A. Sangwongwanich, Y. Yang, D. Sera, and F.
Blaabjerg, “Impacts of PV array sizing on PV
inverter lifetime and reliability,” in 2017 IEEE
Energy Conversion Congress and Exposition
(ECCE), Cincinnati, OH, 102017, pp. 3830–3837.
[6] D. C. Jordan, B. Marion, C. Deline, T. Barnes, and
M. Bolinger, “PV field reliability status—Analysis
of 100 000 solar systems,” Prog Photovolt Res
Appl, vol. 28, no. 8, pp. 739–754, 2020.
[7] Magnus Herz et al., “Quantification of Technical
Risks in PV Power Systems,” Report IEA-PVPS
T13-23:2021, IEA PVPS Task 13, 2021. [Online]
Available: https://iea-pvps.org/wp-
content/uploads/2021/11/Report-
IEA%E2%80%93PVPS-T13-23_2021-
Quantification-of-Technical-Risks-in-PV-Power-
Systems_final.pdf.
[8] Prof. Dr. Heinrich Häberlin und Philipp Schärf,
Ed., Langzeitverhalten von PV-Anlagen über mehr
als 15 Jahre, 2010.
[9] Terry M Therneau, Cynthia S Crowson, Elizabeth J
Atkinson, Adjusted Survival Curves. [Online]
Available: https://cran.r-
project.org/web/packages/survival/vignettes/adjcur
ve.pdf.