Technical Briefing system integration

Reliability of large-scale PV
plants and PV inverters
Inverter reliability | The renewable energy market is currently booming, with large numbers of
PV systems being installed throughout the world. However, a primary objective of any PV power
system is to ensure that the system operates continuously and reliably. As Vicente Salas from the
Universidad Carlos III de Madrid (UC3M) explains, this aspect takes on special relevance in the
case of utility-scale PV projects

• Operating correctly and safely, in compli-

ance with the relevant requirements.
• Keeping costs low and achieving a high
return on investment.

Nevertheless, a profitable, reliable PV

project is only possible if its components
are reliable.
It is clear that a PV power system consists
of many vulnerable components whose
life cycle reliability is highly sensitive to
temperature, power losses and ambient
environments. This can lead to high
electrical stress, as well as to temperatures
in PV modules as high as those in power
electronics converters; this may shorten the
operational life cycles and consequently
result in lower system reliability compared
with conventional generation sources.
Damage, defects and failures of the equip-
ment and elements therefore affect PV
plant production during the exploitation
phase (Fig. 2).

R
eliability is a key risk in any project, For large-scale PV plants, financing is Figure 1. Inverter The non-functioning of some element
including PV plants, but the risk very important and complex, and involves reliability is a vital of the plant is a sensitive issue from the
aspect of ensur-
is more significant in the case of many different parties, including develop- financial point of view. The PV inverter
ing the expected
large-scale PV plants, where the cost of the ers, landowners, utilities, grid operators, performance of a is always a critical component in the PV
project is high. In those projects, where government agencies and financing institu- PV power plant. system, and for many years the inverter was
the typical design target lifetime is around tions. PV projects are a financial investment, one of the components most responsible
25–30 years, there is a discrepancy between which means they are all about returns. for failures. Fortunately, PV inverters have
the lifetime of the PV inverters (5–10 years) Additional operations and maintenance improved, thanks above all to the advances
and of the PV modules (20–25 years). In related to PV inverters, however, can erode made in power electronics, and today these
addition, while inverter interconnection, returns. products are more reliable. To increase
performance and safety standards exist, As shown in Fig. 2, many tasks – both availability and secure maximum return
there are no well-established reliability technical and non-technical – are involved on investment, a PV system requires high
standards. A reliability evaluation (simula- in the financing process of any PV plant; PV inverter reliability in order to reduce
tion and test) must be carried out in order the plant viability is linked not just to the downtime and ensure regular power
to observe failures of PV inverters and to technical tasks but to all the tasks as a generation.
better understand their failure modes and whole. The objectives of any utility-scale PV PV module technology has also
lifetime characteristics. From this it will be project include: continued to improve: the robustness of
possible to guarantee the lifetime of the • Establishing a trade-off between risk modules is evidenced by the standard
inverter. Clearly, PV inverter reliability has management and crisis management. 20- to 25-year warranties that accompany
an impact on life cycle cost, and is therefore • Implementing a long-life power plant, most PV modules today. Thus, it is reason-
an important aspect to address. with high energy yield and availability. able to expect that the PV system inverters

www.pv-tech.org | February 2016 | 67

 system integration Technical Briefing

Figure 2. Stages In order to achieve a highly reliable

of a typical utility- system, it is important to reduce the initial
scale PV plant.
failure rate, provide a low rate of intrinsic
failures, and ensure that wear-out failures
begin to occur only after the system’s useful
lifetime ends.

Product life cycle

Product defects and failures can be
anticipated by managing the product life
cycle (PLC). All manufactured products
have a limited lifetime, and during this
lifetime they will pass through four PLC
stages: introduction, growth, maturity and
decline. In each of these stages manufac-
turers face a different set of challenges. PLC
Figure 3. The management is the application of different
bathtub curve, strategies to help meet these challenges
describing the
and ensure that, whatever stage of the
correlation
between failure cycle a product may be going through,
rate and time. the manufacturer can maximise sales and
profits for their product.
Historically, the quality and reliability of
products has been approached in different
ways. For many years manufacturers paid
little attention to historical failures; they
assumed that quality and reliability groups
were responsible for quality and reliability.
Moreover, manufacturers assumed that the
product design did not significantly affect
quality and reliability, and that quality
and reliability failures were not caused by
manufacturing and suppliers. However,
have a comparable service life. However, result in a significant amount of economic that approach has now changed, and a
although inverters have made progress losses, and are a potential risk (financial risk). revised reliability concept is already begin-
over the same period of time, it has only The bathtub curve, shown in Fig. 3, expresses ning to be applied. As mentioned earlier, PV
been modest: manufacturers today offer the correlation between failure rate and inverter reliability affects life cycle cost, and
inverter warranties of only 10–15 years, time. therefore needs to be dealt with [1].
which means that replacement is necessary In addition, from the point of view of time,
long before any other components of the failures can be classified (according to the Reliability management process
PV system. time of occurrence) into three regions: The reliability of a PV inverter depends
• Early failures (initial or ‘infant mortality’ on the reliability of each of its compo-
Failures in PV inverters failures) nents (for example, semiconductor and
Three types of inverter failure can be • Intrinsic failures (random failures) soldering failures lead to inverter failure),
distinguished: unplanned failure (where the • Wear-out failures which is illustrated in Fig. 4. Unfortunately,
equipment has failed in normal operation Early failures are failures that occur
and was not expected to fail), planned relatively soon after the beginning of opera-
failure and repeat failure. The most critical, tion; the main causes of these initial failures
“PV inverter reliability affects life
and the most difficult to predict, is the are manufacturing or material defects. The cycle cost, and therefore needs to
unplanned failure, which can happen at failure rate in this phase decreases over time.
any time. Different factors can cause such Intrinsic failures are failures that occur at be dealt with”
failure: a fairly constant rate after the initial-failure
• Latent internal causes that existed in the period, until wear-out failures start to occur. in general a PV inverter has no parallel
product from the beginning (predisposi- The majority of electronic components fail at redundancy built into it, which means that
tions). a constant failure rate during this random- a failure in any one of its components will
• External stressors, such as heat and failure stage. lead to an outage of the entire inverter.
humidity of the installation environment Wear-out failures are failures that are It must also be taken into consideration
(external causes). caused by wear and fatigue, and occur that a PV inverter may handle a high level
• Degradation with time. because of the physical limits of the materi- of power flow and operate under high-
als. The failure rate in this phase increases temperature conditions, which degrades
These unanticipated interruptions will over time. the inverter reliability and increases the risk

70 | February 2016 | www.pv-tech.org

 Technical Briefing system integration

of age-related component failures [2]. and is even more critical in the case of a PV
In a reliability management process, the inverter, which may be required to endure
PV inverters should be designed to last both extremely hot and extremely cold
the entire life cycle (up to 30 years) of the ambient temperatures and daily tempera-
product; this process should begin with an ture variations of 30°C or more. Thermal
initial checklist of requirements, and finish management in commercial PV-powered
with an evaluation of operation in the field. inverters is accomplished by means of a
As shown Fig. 5, the reliability manage- fully integrated mechanical design that
ment process must <AQ4>take place in is simple and reliable and which delivers
parallel with other company processes, exactly the cooling that is required to each
such as product definition, development, part of the system. Forced convection
manufacturing and customer service (field cooling is used because it provides superior
deployment). cooling performance at a lower cost, and
The reliability management process with less mechanical complexity, than other
utilises individual sub-processes from other types of cooling (e.g. liquid cooling).
processes, but at the same time adds or
superimposes unique and challenging Reliability evaluation
elements (e.g. stringent qualification and Figure 4. Failure percentages of the most fragile components A reliability analysis during the design and
test procedures for materials, products and of electronic power systems. (Taken from an industry survey development of such complex equipment
processes, as well as advanced methods [2].) as a PV inverter is important in order to
and tools for failure analysis). teristics. The procedure involves a focused detect and eliminate reliability weaknesses
examination of failure point locations as early as possible and to perform
Reliability approach which takes into account the fabrication comparative studies. Different reliability
The reliability approach involves a physics- technology, process, materials and circuit evaluation techniques exist for PV systems:
based multi-level analysis and identification layout obtained from the manufacturer. they can be classified as either theoretical
of the failure points. The ultimate goal is This methodology is capable of providing (simulation tools) or practical (experimental
a system operational lifetime with a low recommendations, using intuitive analysis, tests). The simulation tool category includes
failure rate, and the only way to achieve this for increasing the reliability of components. the Markov process method, Monte Carlo
is to utilise a combination of: simulation, state enumeration method,
• Reliability-oriented design rules. Design for reliability reliability block diagram and fault tree
• Selection of top-tier suppliers and Reliability should be designed-in from the analysis.
acceptance testing of their components. very beginning of the design phase; this As regards reliability tests, those must be
• Manufacturing in well-controlled process is referred to as design for reliability carried out at each stage of development
environments. (DFR). The DFR process therefore starts and mass production. When a product
• Accelerated lifetime testing of the from topology selection, circuit design, and is developed, a reliability test will be
system (and its components) up to the component selection and application, and performed to check the design, material
wear-out point in order to determine uses a highly accelerated testing method and process; then, during mass produc-
when the product will fail, at what rate to discover design flaws in the early devel- tion, a reliability test will be performed as
and which failure mechanisms are at opment stages. The major DFR aspects a quality-assurance inspection or a failure-
fault. that should be borne in mind during a PV rate test for predicting the reliability of the
inverter design include topology selec- product. The purpose and type of reliability
Reliability prediction methodologies tion and design, and thermal design and test therefore greatly depends on the
The newest reliability prediction methodol- management [3]. device manufacturing stage.
ogy, the so-called physics-of-failure (PoF), There is a distinction between quality
emphasises the root cause of failure, failure Thermal management and reliability control. Traditional quality
Figure 5. Reliabil-
analysis, and failure mechanisms as the Thermal management is an essential part ity management control assures that the product will work
basis of an analysis of parameter charac- of the reliability of any electronic system [4] process. after assembly and as designed, whereas
reliability provides the probability that an
item will perform its intended function for
a designated period of time without failure
under specified conditions. In other words,
reliability looks at how long the product will
work as designed, which is a very different
objective from that of traditional quality
control. Therefore, certain tools and models
can be applicable to reliability but not
necessarily to quality, and vice versa.
The reliability test generally has associ-
ated time and cost implications. Testing
under normal operating conditions
requires a very long time, especially for

www.pv-tech.org | February 2016 | 71

 system integration Technical Briefing

products with long expected lifetimes. mitigating ‘infant mortality’ in PV modules,

The results are only useful for an operat- but improvements to these standards are
ing environment which is similar to that in ongoing. They are necessary for ensur-
which the tests were conducted; they may ing the overall reliability and durability of
not be suitable for predicting the reliability products going into the field.
of units operating in significantly different The recently published standard IEC TS
conditions. Alternative methods therefore 62941:2016, Ed. 1.0 (“Terrestrial photovol-
need to be investigated for ‘predicting’ the taic (PV) modules – Guideline for increased
reliability metrics using data and test condi- confidence in PV module design qualifica-
tions other than normal operating condi- tion and type approval”) is a collection of
tions. The main objective of these methods best practices from across the industry.
is to induce failures or degradation of the It refers to the basic requirements of ISO
components, units and systems in a much 9001, and focuses on PV-specific manufac-
shorter time, and to use the failure data and turing processes and procedures to ensure
degradation observations for these acceler- to improve the product’s design in order to Figure 6. quality and consistency, and the key
ated conditions in order to estimate the eliminate the cause(s) of failure. Example of a metrics and capabilities required for PV.
utility inverter
reliability in normal operating conditions. Modules produced in accordance with this
burn-in cycle [7].
Careful reliability testing of systems, ESS and burn-in standard will be more likely to perform as
products, and components during the first The second type of accelerated test is ESS warranted (25+ years).
stage of the product’s life cycle (design and burn-in testing. ESS is a process involv- A dedicated reliability standard for PV
stage) is crucial for achieving the desired ing the application of environmental stimuli inverters, however, does not yet exist; the
reliability in subsequent stages. During to products on an accelerated basis; the standards that do exist – such as ANSI/UL
this early stage, the elimination of design stimuli can include thermal cycling, random 1741 and IEC 62109 Part 1 and 2 – focus
weaknesses inherent to intermediate proto- vibrations and electrical stresses. The goal of primarily on the safety of PV inverters.
types of complex systems is conducted the test is to expose, identify and eliminate Although Ed. 1 of IEC 62093 discusses
via the ‘test, analyse, fix and test’ (TAFT) latent defects which cannot be detected by inverter qualification, it includes all the
process. This process is generally referred to visual inspection or electrical testing, but balance of system (BOS) components. In a
as reliability growth [5]. which will cause failures in the field. ESS is new edition of this standard, a well-accept-
performed on the entire population and ed design qualification standard is being
Types of accelerated test does not involve sampling. developed specifically for PV inverters that
Each type of test that has been designated Burn-in (Fig. 6) is a test performed for the will significantly improve the reliability and
an accelerated test provides different infor- purpose of screening or eliminating margin- performance of these devices. 
mation about the product and its failure al devices, and can be regarded as a special
mechanisms [6]. Generally, accelerated tests case of ESS. Marginal devices are those with
can be divided into three types: qualitative inherent defects, or defects resulting from References
tests, environmental stress screening (ESS) manufacturing aberrations, that cause time- [1] Margolis, R. 2006, “A review of PV inverter technology cost and
and burn-in, and quantitative accelerated and stress-dependent failures. As with ESS, performance projections”, National Renewable Energy Laboratory,
NREL/SR-620-38771.
life tests. burn-in is performed on the entire popula- [2] Yang, S. et al. 2011, “An industry-based survey of reliability in power
tion. Readers interested in learning more electronic converters”, IEEE Trans. Ind. Appl., Vol. 47, pp. 1441–1451.
Qualitative tests about the subject of ESS and burn-in are [3] Sintamarean, C. et al. 2015, “Reliability oriented design tool for the
new generation of grid connected PV-inverters”, IEEE Trans. Power
Qualitative tests yield failure information (or referred to Kececioglu and Sun [8,9]. Electron., Vol. 30, No. 5, pp. 2635–2644.
failure modes) only, and have been referred [4] Catelani, M., Ciani, L. & and Simoni, E. 2012, “Thermal analysis of
critical components in photovoltaic inverter”, Proc. IEEE I2MTC, Graz,
to by many names, including elephant Quantitative test Austria.
tests, torture tests, highly accelerated life In the quantitative test category, the typical [5] Elsayed, E.A. 2012, “Overview of reliability testing”, IEEE Trans. Rel.,
Vol. 61, No. 2, pp. 282–291.
testing (HALT), and shake and bake tests. tools are: [6] Weller, S.D. et al. 2015, “The role of accelerated testing in reliability
In the qualitative category, the typical tools • Life data analysis (a.k.a. distribution analy- prediction”, Proc. 11th Europ. Wave Tidal Energy Conf., Nantes,
are: sis or Weibull analysis) France.
[7] Vidano, R. 2015, “Accelerated reliability testing for commercial and
• Failure modes • Reliability growth analysis utility PV inverters”, NREL PV Inverter Workshop, Golden, Colorado,
• Effects and criticality analysis (FMEA/ • Accelerated testing (a.k.a. life-stress analy- USA.
[8] Kececioglu, D. & Sun, F.-B. 1995, Environmental Stress Screening: Its
FMECA) sis) Quantification, Optimization and Management. Englewood Cliffs,
• Reliability-centred maintenance (RCM) • System modelling using reliability block NJ: Prentice Hall.
[9] Kececioglu, D. & Sun, F.-B. 1997, Burn-in Testing: Its Quantification
• Failure reporting, analysis and corrective diagrams (RBDs) and Optimization. Englewood Cliffs, NJ: Prentice Hall.
action systems (FRACAS) • Simulation
• Root cause analysis (RCA) • Fault tree analysis (FTA)
Qualitative tests are performed on • Design of experiments (DOE) Author
small samples, with the specimens being • Standards-based reliability predictions Vicente Salas is an associate professor at the
subjected to a single severe level of stress, (e.g. MIL-217) Universidad Carlos III de Madrid (UC3M), and the
to a number of stresses, or to a time-varying CEO of the UC3M PV Lab. He has several years’
experience in the fields of PV plant qualification
stress (i.e. stress cycling, cold to hot, etc.). Standards
and monitoring, laboratory PV inverter and module
If the specimen survives, it passes the test; Design qualification test protocols – such as investigation, and performance measurement.
otherwise, appropriate actions will be taken IEC 61215 and IEC 61730 – have been key to

72 | February 2016 | www.pv-tech.org