Failure mode and effect analysis for photovoltaic systems

Alessandra Colli n

Brookhaven National Laboratory, Upton, 11973 NY, USA

a r t i c l e i n f o

Article history:Received 18 May 2014Received in revised form27 March 2015Accepted 2 May 2015Available online 31 May 2015

Keywords:FMEAPhotovoltaic systemsReliability

a b s t r a c t

Failure mode and effect analysis (FMEA) is an inductive and conservative system reliability analysisapproach, here applied to photovoltaic system. A system is a complex combination of components andsub-components, where technical and disciplinary interfaces apply in their mutual interactions. FMEAprocesses the individual analysis of each system's sub-component with the task to identify the variousfailure modes affecting each part, along with causes and consequences for the part itself and the entiresystem. In the proposed analysis the system's component and sub-components have been identifiedfrom the design of the Northeast Solar Energy Research Center (NSERC) photovoltaic research arraylocated at Brookhaven National Laboratory's (BNL). The complete FMEA analysis is presented, along withthe applied ranking scales and final results. The approach is discussed in its benefits and limitations, thelatter mainly identified in the limited amount of open source information concerning failure prob-abilities for the photovoltaic system parts.

& 2015 Elsevier Ltd. All rights reserved.

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8042. The FMEA process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8053. The system model and its components . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8054. The available data and the scoring system. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8065. The FMEA table . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8076. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 807References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 809

1. Introduction

Electric utilities and grid operators face major challenges froman accelerated evolution towards an extensive integration ofvariable renewable energy sources into the electric power grid,such as solar photovoltaic (PV). The integration of such a variableenergy source into the existing, sometimes weak or overloaded,electric grid requires an adequate risk-informed decision makingapproach. The ideal grid integration design for PV systems shouldoptimize the mutual benefits between the grid and the PV systemitself; this has to take into consideration the PV source variability,availability, reliability, as well as the stability of the electric grid.The aim is to reduce or promptly intervene with outages and

impairments affecting the PV system, to improve the confidencein this renewable energy source.

So far, the most of the photovoltaic-related reliability analysis hasfocused on modules [1] and balance of system (BOS) separately [2].Only in recent years the shift of focus to grid integration has requiredconsidering the entire system. The purpose of this paper is to presentand discuss the complete results of a failure modes and effects analysis(FMEA) developed for a PV system [3]. To the author's knowledge,there are no complete and detailed FMEA analyses for PV systemsincluding risk ranking information published to date. This workrepresents part of the background investigations needed to developa probabilistic risk analysis (PRA) for PV systems [4], to investigatesafety-related and energy-production-related risks. For this reason, theFMEA has been preferred to other methods, such as Taguchi [5,6].

The system under analysis is a simplified model having allthe principal components and sub-components as from the designof the Brookhaven National Laboratory's (BNL) Northeast SolarEnergy Research Center (NSERC) research array. The analysis aims

Contents lists available at ScienceDirect

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

Renewable and Sustainable Energy Reviews

http://dx.doi.org/10.1016/j.rser.2015.05.0561364-0321/& 2015 Elsevier Ltd. All rights reserved.

n Tel.: þ1 631 344.2666.E-mail address: alessandra.colli@gmail.com

Renewable and Sustainable Energy Reviews 50 (2015) 804–809

to identify the failure modes affecting the system's sub-componentsand to list possible causes and effects.

Despite the approach is now common for PV applications, FMEAanalyses have been performed in other renewable energy areas,such as wind energy [7,8].

2. The FMEA process

Failure modes and effects analysis (FMEA) is an established semi-qualitative reliability engineering approach to systematically evaluat-ing system design on a component-by-component basis to identifyfailure modes and their effects on system function and other systemcomponents. It can support fault tolerant design, testability, safety,logistic support and related functions. This bottom-up technique hasbeen an essential tool for industries such as the aerospace and aut-omobile industries, the semiconductor industry [9], and the nuclearindustry [10,11]. Government agencies (such as the Air Force and theNavy) require that an FMEA is performed on their systems to ensuresafety as well as reliability. The automotive industry has adopted theuse of FMEA to support the design and manufacturing/assembly ofautomobiles.

Kumamoto and Henley [12] recommend several uses for theFMEA:

1) Identification of critical components for fail-safe design,failure-rate reduction or damage containment.

2) Identification of components requiring particularly stringentquality control.

3) Formulation of special requirements to be included in specifi-cations for suppliers.

4) Formulation of special procedures, safeguards, protectiveequipment, monitoring or warning systems.

5) Distribution of project funds across these areas.

Although there are various types of FMEA (design, manufactur-ing process, equipment, system) and for different applications(hardware to software), the principal aim of this approach is tosupport the early identification of potential problems and addressthem before accidents happen.

The FMEA presented in this work has the task to identify failuremodes along with possible causes and effects for a grid-connectedPV plant. The FMEA process followed along this study is shown bythe block diagram in Fig. 1. It requires to identify the systemmodel, its components, sub-components, requirements, descrip-tions, and, when useful, also functional diagrams. Failure modesare investigated at the system's sub-component level, according tothe desired level of depth in the analysis. For each failure mode aseverity (S), occurrence (O) and detection (D) rating is defined andrated according to subjectively defined scales, based on availableinformation and supported by expert opinion and evaluation. Therating system involves expert opinion and a level of subjectivitywhich is typical of rating systems based on a scales defined bythe user.

The combination of the three ratings defines an overall riskmeasure, the risk priority number (RPN), which indicates therelevance of each failure mode in affecting the PV system.Villacourt [13] describes this approach in relation to the semi-conductor industry. The RPN is calculated for each failure modeaccording to the following equation [5,6, 13, 14]:

RPN¼ S � O� D ð1ÞA high RPN is a critical indicator for corrective action considerationson identified sub-components. The RPN simplifies the computationof the criticality number adopted in the failure mode effect andcriticality analysis (FMECA) by requiring only the probability of

failure (occurrence) and the severity classification; however, theRPN extends the criticality number approach by incorporating thedetection likelihood rating. This is crucial in evaluating PV systemssince system downtime directly leads to power supply interruptionand financial losses when energy purchase agreements or feed-intariffs are in place. Thus, quick, efficient detection of failures is cri-tical, and the RPN is implemented such that the detection of failuresis a conscious goal of the FMEA application.

The FMEA is a systematic, inductive, and conservative techni-que for failure analysis and it is here performed ahead of thedevelopment of more complex system-level methods such as faulttrees (FT) and event trees (ET) analysis, combined into theprobabilistic risk analysis (PRA). In further research developmentsat BNL, we will use the FMEA primarily as an investigation tosupport the development of a PRA model and identify elementsand failures to be represented in the PRA in relation to the rest ofthe system. A fundamental difference between the FMEA and PRAis actually that the former is focusing on individual components,while the latter is modeling the interactions between componentsin the entire system, thus providing a holistic overview.

3. The system model and its components

To perform the FMEA analysis, the PV system will be repre-sented by a simplified model reporting all the components as bydesign. Fig. 2 shows the simplified model used for the FMEA andbased on the BNL's NSERC photovoltaic research array configura-tion. The diagram shows that the system is mainly built in 3 blocks:(i) source system, (ii) string combiner, and (iii) power conditioningsystem.

The NSERC array design is in real much more complex than thesimplified model shown in Fig. 2, which has the only purpose ofidentifying the sub-components to consider in the FMEA. In itspresent development the NSERC array reaches a rated power of518 kWp and includes a total of 1672 PV modules, rated 310 Wp

each. The modules are arranged in strings of 19 modules each.Combiner boxes merge 11 strings to reach the input of a singleinverter's module. The 3 plant inverters are actually modular, andallow the independent management of each set of 11 strings. All

Fig. 1. Block diagram representing the FMEA process followed along this study.

A. Colli / Renewable and Sustainable Energy Reviews 50 (2015) 804–809 805

the inverter outputs finally merge into a 1MVA transformer toreach a second step-up transformer at the network interconnec-tion point. The plant is grid-connected to the electric grid of BNLand operated independently by the laboratory. The need forindependence from the utility in terms of plant management isrequired by the reconfigurable characteristic of the plant to meetvarious research needs in the field of grid integration and smartgrids. An additional purpose of the array is to evaluate the PVmodule's performance and degradation in the specific NortheastUS climate conditions.

However, the FMEA approach looks at the failure modes of thesingle components and is less interested in the combinations ofthem. The interest in the complexity of the design comes up whenthe combined system analysis is performed, which is the case theprobabilistic risk assessment (PRA) approach.

The performed analysis includes all the plant components up tothe grid interconnection point: PV modules and their rack supportstructures, DC subsystem with string combiner, and power con-ditioning system with inverter and transformer. From the per-spective of the FMEA development, it is important to know thedetails at the level of the sub-components. Table 1 lists the systemcomponents and the associated sub-components. Functional dia-grams have been consulted when more complex elements areconsidered, such as the case of the inverter or the transformer.

4. The available data and the scoring system

The qualitative aspects of the FMEA analysis, including theidentification of failure modes, causes and consequences, arebased on a large amount of literature easily available in the areaof PV reliability and degradation studies, as well as studies on theelectric components. The major problems encountered along thedevelopment are actually related to the quantitative information,namely failure rates or failure probabilities, and they are: (i) thelack of PV specific open-source databases or data collections ofquantitative information on failures of specific sub-componentsand (ii) the sometime outdated quantitative information reportedin available databases for electric elements. Assuming that weshould not expect extreme variation in the failure rates reported

by specific databases or report for electric sub-components (suchas [7]), the first limitation specified above has been the mostdifficult to overcome. PV operation and maintenance companiesare not prone to release their data, thus the quantitative informa-tion on the failures of PV specific components has been based on asingle publication in the PV reliability field reporting failurestatistics [15]. Table 1 shows the failure rates, expressed in failuresper unit hour, considered for each sub-components.

The scoring system for the failure modes has been developed onand adapted to the available data and information, expressing bothquantitative values and subjective evaluations from expert opinions,as typical in this kind of approach. Tables 2–4 show the rankingcriteria developed respectively for the severity, the occurrence andthe detection ratings. Each ranking system follows a scale from 1 to 5.In all the three cases 1 denotes the best situation, while 5 is ass-ociated to the worst situation. The scales have been implemented onthe basis of [9], following 5 basic criteria. However, differently from

Fig. 2. The simplified PV system diagram showing the components and sub-components considered in the analysis.

Table 1Component and sub-components of the PV system.

Component Sub-component Failure rate (failures perunit-hour)

PV module Module 1.35E-06a

Junction box/bypass diode 6.77E-07a

Connectors 4.51E-07 a

Encapsulantion 4.06E-06 a

Rack Rack structure 2.44E-05 a

Grounding/lightningprotection system

1.62E-05 a

Cable Aerial cables 1.05E-06b

Underground cables 7.00E-07 b

Stringcombiner

Fuse 2.17E-07 b

Disconnect 6.96E-07 b

Powerconditioning

Reverse polarity diode 2.26E-07 a

Fuse 2.17E-07 b

Breaker 4.00E-07 b

Inverter 1.75E-04 a

Disconnect 6.96E-07 b

Transformer 4.22E-07 b

Protective relays 2.28E-07 b

a Value extrapolated by elaboration of the data available in Ref. [15].b Value from Ref. [13].

A. Colli / Renewable and Sustainable Energy Reviews 50 (2015) 804–809806

[9], the numerical scale has been set 1 to 5 instead of 1 to 10; thisbecause the additional sensitivity given by a range of scoring valuesfor a selected single criterion has not been considered necessary inthis work, given the limitations already expressed on the quantitativedata available. Consequently, the RPN values are ranked on a scalebetween 1 and 125; also in this case, the smaller the RPN the better,the larger the worse.

5. The FMEA table

The details of the FMEA analysis are shown in Table 5. Potentialfailure modes, causes, effects and ratings up to the overall riskpriority number are listed for each sub-component of the system.

The identification of causes has been performed by looking tothe PV system in a holistic way and trying to expand the view topossible uncommon events, based on the outcome of discussionswith different experts.

The potential effects have been focused on the fundamentalpurposes of this analysis, which involve: (i) security of electricitysupply (energy output), (ii) technical damages and (iii) humansafety.

The risk priority numbers show three major contributors to thePV system reliability: in first place the inverter, and the grounding/lightning protection system; in a second position the modules,intended as active components, such as cells and contacts.

However, it is interesting to notice that even components withlow risk priority number could present a high detection ratingvalue, which indicates a minimal probability that the problem canbe detected during normal operation. This calls for the importanceof a regular maintenance routine, to avoid unexpected problemswhen less wanted. In this framework, an existing extension of this

FMEA work has been done in collaboration with George Washing-ton University [16]. The work points at the use of a surprise index(SI). The surprise index is based on the information score of thefailure mode probability. By weighting the risk priority by thefailure mode's information score, we are increasing the influenceof extremely unlikely, yet extremely catastrophic, events in riskmanagement decision contexts. This also decreases the amount ofprominence placed on relatively likely events in the decisioncontext. Thus, if an increasing RPN indicates a higher priority forredundancy investments, the SI should be used to prioritize thedevelopment of contingency plans [16].

6. Conclusions

The application of the FMEA approach has been discussed anddemonstrated for PV systems. The methodology proved the inverterand the ground system of the PV field to show the highest values ofthe RPN, calculated according to Eq. (1). This is in line with whatreported in existing literature [15] and with the experience person-ally discussed with some PV plant operators. However, the FMEAshows also the importance of maintenance activities for the earlydetection of some hidden failure modes that could not affectimmediately the plant, but could degenerate into a system problemif not promptly handled.

Despite the use of FMEA and risk analysis techniques in the PVindustry [17], the lack of publically available FMEA analysis for PVsystems makes it difficult to validate the results. Interactions withindustry, working groups and researchers in the PV field have beenused to support the development and understand the proper levelof details to be considered for a meaningful evaluation in respectto the available numerical information. Future analyses along with

Table 3Occurrence ranking criteria.

Rank Description

1 Unlikely – failure rate per unit-hour in the order of E-72 Remote probability – failure rate per unit-hour in the order of E-63 Occasional probability – failure rate per unit-hour in the order of E-54 Moderate probability – failure rate per unit-hour in the order of E-45 High probability – failure rate per unit-hour in the order of E-3 and E-2

Table 4Detection ranking criteria.

Rank Description

1 Almost certain that the problem will be detected (chance 81–100%)2 High probability that the problem will be detected (chance 61–80%)3 Moderate probability that the problem will be detected (chance 41–60%)4 Low probability that the problem will be detected (chance 21–40%)5 None/minimal probability that the problem will be detected (chance 0–20%)

Table 2Severity ranking criteria.

Rank Description

1 Minor failure/degradation, hardly detected, no influence on the system performance2 Failure/degradation will be detected by plant owner/operator and/or will cause slight deterioration of parts or system performance3 Failure/degradation will be detected by plant owner/operator, will create dissatisfaction, and/or will cause deterioration of parts or system performance4 Failure/degradation will be easily detected by plant owner/operator, will create high dissatisfaction, and/or will cause extended deterioration of parts and system

relevant non-functionality/loss of performance5 Failure/degradation will result in non-operation of the system or severe loss of performance

A. Colli / Renewable and Sustainable Energy Reviews 50 (2015) 804–809 807

Table 5FMEA table.

Potential failuremode

Potential causes Potential effects Severityrating

Occurrencerating

Detectionrating

Riskprioritynumber

Module (active components – cells and contacts)Loss of electricfunction

Shorts, arcs, open contacts. No energy output, safety, fire 5 2 3 30

Impairment ofelectricfunction

High series resistance, low shunt resistance, aging,shading, soiling.

Reduced energy output, hot spot damage 4 2 4 32

Junction box/bypass diodeOpen contacts Disconnections, improper installation, corrosion No energy output 5 1 3 15Short, arc incontacts

Damaged insulation, aging, animals, lightning No energy output, safety, thermaldamages, fire

5 1 2 10

Poor contact/intermittent

Material defects, oxidation, aging Reduced energy output, no energy output,thermal damage

4 1 4 16

Shorted diode(end-to-end)

Material defects, aging, thermal stress, mechanicalstress, electrical stress, material contamination,processing anomaly

Reduced energy output, loss of modulepower, overcurrent

4 1 4 16

Open diode Very high resistance, material defects Reduced energy output, thermal damagesin module, fire, safety

3 1 5 15

Parameter changein diode

Material defects, aging, continuous thermal stress Reduced energy output, improperintervention. loss of module power,overcurrent

3 1 5 15

ConnectorsOpen Damage, disconnection, animals, vandalism, strong

wind, pulled cablesNo energy output 5 1 2 10

Poor contact/intermittent

Corrosion, improper installation, lightning damage Reduced energy output, no energy output,thermal damage

5 1 4 20

Short Damages, improper installation, disconnections,animals, vandalism

No energy output, safety, thermaldamages, fire

4 1 5 20

EncapsulationLoss of airtightness

Bad lamination, high voltage stress, hot spots, high cell/module temperature, corrosive effects in the modulestructure, aging, damage from frame distortion, cleaningactions, extreme wind, snow load, vandalism, animals,lightning, earthquake, accidental impacts

Humidity/water/contaminant entrance,increased degradation, reduced energyoutput, no energy output

2 2 5 20

Rack structureLoss ofconfiguration

Improper installation, damages, extreme weatherconditions, excessive thermal expansion/contraction,earthquake

Front glass breakage, cell damages, framedistortion, reduced energy output, noenergy output, safety

4 3 1 12

Bracketsdetachment

Strong wind, improper installation, earthquake,accidental impacts

Unstable configuration, loss of modules 3 3 2 18

Grounding/lightning protection systemOpen orineffective

Corrosion, improper installation, lightning, mechanicaldamage, too high resistance

Safety, module damage, reduced energyoutput.

4 3 4 48

Aerial cablesOpen Faulty cabling, material aging, animals, vandalism,

extreme weather conditions, earthquakeNo energy output, safety 5 2 2 20

Short, arc Cracks/ruptures on cables, insulation failure, aging,animals

No energy output, safety, fire 5 2 2 20

Excessive wear Improper system design Reduced energy output, thermal damages 3 2 4 24Underground cablesOpen Faulty cabling, material aging, animals, vandalism,

extreme weather conditions, earthquakeNo energy output, safety 5 1 2 10

Short, arc Cracks/ruptures on cables, insulation failure, aging,animals

No energy output, safety, fire 5 1 2 10

Excessive wear Improper system design Reduced energy output, thermal damages 3 1 4 12FuseFails to open Bad system configuration, construction defect,

mechanical defects, improper maintenanceExcessive increase of current in thesystem, overheating, safety, arcs, fire

4 1 4 16

Slow to open Bad system configuration, construction defect,mechanical defects, improper maintenance

Excessive increase of current in thesystem, overheating, safety, arcs, fire

4 1 4 16

Premature open Bad system configuration, construction defect,mechanical defects, improper maintenance

No energy output 5 1 2 10

DisconnectOpen withoutstimuli

Bad system configuration, construction defect,mechanical defects, improper maintenance

No energy output 5 1 1 5

Does not open Faulty switch, damages to structural parts, flashover/arc,improper maintenance, aging

No disconnection, safety, fire, arcs 4 1 4 16

Reverse polarity diodeShort (end-to-end)

Material defects, aging, thermal stress, mechanicalstress, electrical stress, contamination, processinganomaly

No protection against reverse currents 2 1 4 8

Open Very high resistance, material defects No energy output 5 1 1 5Parameter change Material defects, aging, continuous thermal stress Activation with different variable range 3 1 5 15Breaker

A. Colli / Renewable and Sustainable Energy Reviews 50 (2015) 804–809808

the access to system information and data for a statistically rele-vant time span for the system operation could provide the vali-dation of the results. This is a first attempt to provide a completeFMEA analysis for PV systems and it required a substantial work incollecting information. However, despite the limitations given bythe lack of validation of the results with published work, the workhas highlighted a large set of failure modes along with causes andeffects that will feed a probabilistic risk analysis for safety-relatedand production-related issues.

The application of traditional reliability, hazard analysis andrisk analysis techniques into new environments, such as PV, arepossible [18] and desirable for reliability improvements and risk-informed decision making.

References

[1] Review of Failures of Photovoltaic Modules, IEA-PVPS T13 report; 2014, ⟨http://www.isfh.de/institut_solarforschung/files/iea_t13_review_of_failures_of_pv_modules_final.pdf⟩.

[2] Flicker, J., Kaplar, R., Marinella, M., Granata, J., PV inverter performance andreliability: what is the role of the bus capacitor? In: Proceedings of the IEEE38th photovoltaic specialists conference (PVSC), volume 2, 2012.

[3] Colli, A., An FMEA analysis for photovoltaic systems: assessing differentsystem configurations to support reliability studies – introduction to PRAanalysis for PV systems. In: Proceedings of society for risk analysis annualmeeting. San Francisco, CA; 2012.

[4] Colli, A., Extending performance and evaluating risks of PV systems failureusing a fault tree and event tree approach: analysis of the possible application.In: Proceedings of the 38th IEEE photovoltaic specialist conference. Austin TX;2012.

[5] Su C-T, Lin H-C, Teng P-W, Yang T. Improving the reliability of electronic paperdisplay using FMEA and Taguchi methods: a case study. Microelectron Reliab2014;54:1369–77.

[6] Mariajayaprakash A, Senthilvelan T. Failure detection and optimization ofsugar mill boiler using FMEA and Taguchi method. Eng Fail Anal2013;30:17–26.

[7] Arabian-Hoseynabadi H, Oraee H, Tavner PJ. Failure Modes and Effects Analysis(FMEA) for wind turbines. Electr Power Energy Syst 2010;32:817–24.

[8] Shafiee M, Dinmohammadi F. An FMEA-based risk assessment approach forwind turbine systems: a comparative study of onshore and offshore. Energies2014;7:619–42.

[9] Villacourt, M., Failure Mode and Effects Analysis (FMEA): a guide for contin-uous improvement for the semiconductor equipment industry. SEMATECH,Austin, TX, Technology Transfer #92020963B-ENG; 1992.

[10] Vesely, W.E., Goldberg, F.F., Roberts, N.H., Haasl, D.F., Fault tree handbook,NUREG-0492, 1981.

[11] Betancourt, L., Birla, S., Gassino, J., Regnier, P., Suitability of fault modes andeffects analysis for regulatory assurance of complex logic in digital instru-mentation and control systems, NUREG/IA-0254, 2011.

[12] Kumamoto H, Henley EJ. Probabilistic risk assessment and management forengineers and scientists. 2nd ed.. NY: IEEE Press; 1996.

[13] Power systems reliability subcommittee of the power systems engineeringcommittee, IEEE industry applications society, IEEE Std. 493-2007 recom-mended practice for the design of reliable industrial and commercial powersystems, IEEE-SA standards board, New York, NY, IEEE 493-2007.

[14] Šolc M. Applying of Method FMEA (Failure Mode and Effects Analysis) in thelogistics process. Advanced Research in Scientific Areas, Section 12, Industrialand Civil Engineering 2012:1906–11.

[15] Golnas A. PV system reliability: an operator's perspective. IEEE J Photovolt2013;3(1):416–21.

[16] Francis R, Colli A. Information-based reliability weighting for failure modeprioritization in photovoltaic (PV) module design. Honolulu, Hawaii: Prob-abilistic Safety Assessment and Management PSAM 12; .

[17] Ramu, G., Overview of the proposed PV quality management system. In: Solarpower international workshops. NREL, Golden, Colorado, 2014. ⟨http://www.nrel.gov/pv/performance_reliability/pdfs/2014_spi_wkshp_ramu.pdf⟩.

[18] Colli A, Serbanescu D, Ale B. PRA-type study adapted to the multi-crystallinesilicon photovoltaic cells manufacture process. In: al ME, editor. Safety,reliability and risk analysis: theory, methods and applications. London: Taylorand Francis Group; 2008.

Table 5 (continued )

Potential failuremode

Potential causes Potential effects Severityrating

Occurrencerating

Detectionrating

Riskprioritynumber

Open withoutstimuli

Bad system configuration, construction defect,mechanical defects, improper maintenance

No energy output 5 1 1 5

Does not open Faulty switch, damages to structural parts, flashover/arc,improper maintenance, aging

No disconnection, safety, fire, arcs 4 1 4 16

InverterFails to transfer Contact damage, card/board problem, software failure

(within working conditions), ventilation obstruction,extreme weather conditions, vandalism.

No energy output 5 4 1 20

Degraded output MPPT unbalance, extreme weather conditions Reduced energy output 4 4 3 48Transformer

Open Extreme weather conditions (including lightning),flooding, earthquake, explosion, exposure to non-electric fire/burning, shorting, aging

No energy output 5 1 1 5

Short Insulation breakdown, damages to structural parts,water/particles in oil, transient overvoltage disturbance,continuous overvoltage, shorting, lack of protectivedevice, improper maintenance, aging

Reduced energy output, no energy output,safety, fire

5 1 1 5

Parameter change Failure of tap changer, damages to structural parts,improper maintenance, aging

Loss of efficiency, improper energy output 3 1 2 6

Protective relaysFails to trip Inadequate protective device, improper setting of

protective device, improper maintenance, agingLoss of protection resulting in electrical/structural damages with reduced energyoutput, no energy output, safety, fire,explosion

4 1 2 8

Spurious trip Bad system configuration, corrosion, aging, lack ofmaintenance or improper maintenance

No energy output 5 1 1 5

Short Inadequate protective device, improper setting ofprotective device, improper maintenance, aging

Loss of protection resulting in electrical/structural damages with reduced energyoutput, no energy output, safety, fire,explosion

4 1 3 12

A. Colli / Renewable and Sustainable Energy Reviews 50 (2015) 804–809 809