RIAC 217plus Reliability Prediction Model in Photovoltaic Systems

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 5

RIAC 217Plus Reliability Prediction Model

in Photovoltaic Systems
G. Graditi, G. Adinolfi, A. Pontecorvo
ENEA - Italian National agency for new technologies, Energy and sustainable economic development
Research Center, Portici (NA), Italy
corresponding author - email: giorgio.graditi@enea.it - phone: +390817723400

Abstract – Systems reliability evaluation is an important fact, Italy was the top market for the 2011, with 9.3 GW
task in industry. In fact, reliability indices suggest connected, followed by Germany with 7.5 GW. In fact,
information about equipments quality level, maintenance considering the globally installed capacity, PV represents,
and investments. Reliability performances can be estimated after hydro and wind power, the third most used
in different ways. Most used methods are reliability
prediction models. In spite of their diffusion, their
renewable energy source. PV modules are characterized
applications and results are often discussed, since each of from a highly reliable components in the range of 15-20
them provides a different reliability evaluation for the same years, with good performance and efficiency in the early
system under investigation. In addition, these models years. After several years of operation, their performance
evaluates a system behaviour in a specific working decrease can affect reliability. A PV plant is a complex
condition. This aspect introduces supplementary problems system constituted by many electronic and mechanical
in case of photovoltaic systems characterized by continuous
changing operating conditions. The aim of this paper is to
devices, such as PV modules, converters, tracker systems,
use the RIAC 217Plus model to a qualitative comparison optics and so on. It works in continuous changing
between different Distributed Maximum Power Point operating conditions depending on meteorological
Tracking converters to identify the most reliable one. A new phenomena (irradiance, ambient temperature) and
index to characterize the reliability behaviour of systems installation site position (shadow, dust, etc). The majority
used in continuous changing environmental conditions is of PV systems failures are attributed to inverter failures.
proposed.
In fact, DC-AC converters are characterized by Mean
Index Terms - Mean Time Between Failure, failure rate, Time to First Failure (MTFF) of about five years.
reliability performance, photovoltaic system Designers’ aim is to improve these systems
performances increasing efficiency and reliability and
I. NOMENCLATURE reducing costs. In fact, a high efficiency PV plant allows
to make shorter the energy payback time, while a reliable
ȜOB base failure rate, Operating PV plant permits to make lower the maintenance service
ȜEB base failure rate, Environmental number. In this scenario, the reliability evaluation of PV
ȜTCB base failure rate, Temperature Cycling plants devices becomes a high importance matter to be
ȜSJB base failure rate, Solder Joint considered from the project start. The method to Mean
ȜEOS failure rate, Electrical OverStress Time Between Failure or Failure Rate calculations for
ʌG reliability growth failure rate factor these devices is not a direct and unique issue, since many
ʌDCO failure rate factor for Duty Cycle, operating reliability prediction models, methods and tests are
ʌTO failure rate factor for Temperature, available. In addition, considering a specific system, the
operating application of different methods provide different
ʌS failure rate factor for Stress reliability performances evaluations. Many literature
ʌDCN failure rate factor for Duty Cycle, non- works discuss reliability prediction models problematic
operating aspects and limitations [1-2] underlining these models
ʌTE failure rate factor for Temperature effectiveness for qualitative analysis and systems
environmental behaviour comparative studies. The fitness of these
ʌCR failure rate factor, Cycling Rate models to correctly express the reliability behaviours of
ʌDT failure rate factor, delta Temperature PV plants subsystems in continuous changing operating
ʌSJDT failure rate factor, Solder Joint Delta conditions is another nodal point to the choice of the most
Temperature suitable one to use. In detail, in section 2 the most used
MTBF Mean Time Between Failure reliability prediction models are analyzed. In section 3
Distributed Maximum Power Point Tracking (DMPPT)
II. INTRODUCTION boost converters design are considered and their
The number of grid-connected PhotoVoltaic (PV) reliability performances are evaluated by means of the
systems has increased enormously in the last years. In RIAC Handbook 217Plus. In addition, a new weighted
Mean Time Between Failure formula is proposed to best

978-1-4673-4430-2/13/$31.00 ©2013 IEEE 343


characterize continuous changing operating conditions Fixed Uncontrolled, Ground Mobile, Airborne
systems. An analysis is carried out to compare reliability Commercial, and Space) plus additional alternatives
performances of Synchronous Rectification and Diode useful in the military environment. It is based on
Rectification power optimizers. pessimistic failure rate assumptions. This model does not
consider other factors that can contribute to failure rate
III. RELIABILITY PREDICTION MODELS such as burn-in data, lab testing data, field test data,
Reliability R(t) is defined in Eq.1, it is the probability designer experience, wear-out, etc.
than an item will perform a required function without ƒ The Telcordia prediction model - Reliability
failure under stated conditions for a stated period of time. Prediction Procedure for Electronic Equipment SR-332
[4]- was developed by AT&T Bell Labs in 1997 and it is
R (t ) Pr^T ! t` (1) focused only on electronic equipments. This model
(previously known as Bellcore) modified the MIL-
where:
HDBK-217 Prediction Model to better represent the
Pr = probability
equipment of the telecommunication industry by adding
T = random variable = lifetime of the unit
the ability to consider burn-in, field, and laboratory test
t = mission time
data. Although the Telcordia standard was developed
specifically for the telecom field, it is used to model
The reliability value R is a number in the range (0,1)
products in a number of other industries. A disadvantage
and its graph is shown in Fig.1.
is that the predictions are limited to environment that
work with temperature between 30 ºC and 65 ºC.
ƒ The RIAC Handbook 217Plus model [5], published
in 2006, has been developed by the Reliability
Information Analysis Center (RIAC) and pointed out by
United States Department of Defence as the successor of
the MIL-HDBK-217 and the PRISM® methodology. The
form of this model is quite different from MIL-HDBK-
217 and Telcordia SR-332. In fact, 217Plus considers
different base failure rates for each generic class of
Fig. 1. Reliability function R(t) graph. failure mechanism. These process factors are determined
by a qualitative assessment of process criteria with
In detail, at the beginning the probability of a device weighting factors applied.
proper functioning is high, it decreases with t and, for ƒ Fides: the reliability methodology FIDES Guide
high mission time, the probability of device proper 2004 [6] has been created by FIDES Group, a consortium
functioning is low. Indices used to express devices of French industrialists from the fields of aeronautics and
reliability performances are : Defence (Airbus France, Eurocopter, GIAT Industries,
ƒ the Failure Rate (FR) or hazard function represents MBDA and THALES). The FIDES methodology is based
the frequency with which a component or a system fails on the Prediction Model physics of failures supported by
ƒ the Mean Time Between Failure (MTBF) is a the analysis of test data, so it is different from traditional
measure of how reliable a product is. It is usually given in prediction methods which are exclusively based on the
units of hours. High MTBF values characterize high statistical analysis of historical failure data collected in
reliability products. the field, in-house or from manufacturers.
These indices can be calculated by means of many
different reliability prediction models. Some of them - IV. RELIABILITY EVALUATION: CASE STUDY
Military Handbook 217, PRISM® and 217Plus – were In spite of the number of applicable reliability
developed for military applications, while others-Fides, prediction models, the identification of the “best” one is
Telcordia,etc- for avionics and telecommunication not easy. In fact, many of them permit to calculate
sectors. The most used ones are: systems MTBF in a specific working condition [7-11].
ƒ MIL-HDBK-217: this reliability model, published These methods are suitable for a great number of
by the United States Navy in 1965, was the only application fields characterized by a nominal or worst
reliability prediction method available at the time, case operating point. In PV applications is not possible to
therefore the reliability communities adopted this tool for identify a nominal or worst condition. In fact, PV systems
their own use. It is probably the most internationally and subsystems work in continuous changing conditions
recognized empirical prediction method. As a result, in terms of irradiance, ambient temperature, wind
MIL-HDBK-217 [3] became and still is one of the most velocity, humidity and so on. A solution could consider
widely known and used reliability prediction method. It the worst case and to estimate the PV plant reliability
includes models for a broad range of part types and performance in that condition, but a similar result cannot
supports the five most commonly used environments in correctly characterize PV plants relaibility behaviour,
the telecom industry (Ground Fixed Controlled, Ground

344
since they do not always (24 hours/ day for 365 In detail, the SR boost is able to guarantee a better
days/year) operate in that particular condition. An efficiency since its HS FET switching device is
alternative method was introduced in [9] in which the characterized by a lower voltage drop and so lower losses
MTBF formula reported in Eq.2 was proposed for DC- than the DR ones. In addition, the SR converter always
DC converters used in PV fields. The Eq.2 weights and works in Continuous Conduction Mode (CCM) without
partial load percentages are the same of the European requiring high inductance value to avoid Discontinuous
efficiency formula used to evaluate inverters Conduction Mode. Table I reports the 217Plus formulas
performances. In this paper, the PV power optimizers to calculate the failure rates of the MPPT converter
design procedure proposed in [12,16] is completed with devices under investigation. Since the thermal stress is
the converters reliability estimation since their design one of the most invalidating factor [16], the attention is
start. The developed tool, after the identification of a set here focused on MTBF variations depending on
of feasible commercial electronic components able to temperature increase. Referring to the reported failure
works as a specific DMPPT converter topology, provides rate formulas, ʌTE and ʌTO factors and converter devices
suitable indices to characterize the obtained solution temperature rise TR for some ambient temperature values
efficiency, cost and reliability performances. In this case, are shown in Figs. 3-5. In this analysis, typical
the RIAC Handbook 217Plus reliability prediction model meteorological data recorded at South Italy installation
is chosen. It is the updated version of MIL-HDBK-217F sites are used. In Fig.6 the considered temperature
and it is compatible with its predecessor permitting to distribution is represented by means of a bar graph. The
keep unchanged industry practice for products reliability most relevant temperature ranges and how many hours
predictions. It also considers components operating and the PV system worked in those conditions during the total
non-operating conditions (operating and non-operating operating hours are reported. In Table II, the MTBF
temperatures, Duty Cycles, Cycling Rate, etc). values of a considered SR DMPPT converter, referring to
the previous mentioned temperature ranges, are
presented. In a similar way, applying RIAC 217Plus
failure rate expressions, MTBF data for a DR boost are
obtained and shown in Table III. In the reported tables is
possible to underline the power optimizers reliability
dependence on temperature and how their performances
deteriorate with thermal stress increase. This is a crucial
aspect for systems operating in continuously changing
conditions. In detail, in case of the SR boost under
investigation, there is a great difference between the
MTBF value estimated at 15°C and that calculated at
Fig. 2. (a) SR boost converter; (b) DR boost converter. 35°C. Also in case of the DR converter, this difference is
considerable. So it is evident that the reliability
As many others prediction models, it evaluates performance analysis in a unique operating point is not
components failure rates as the product of a base failure sufficient to represent the system reliability behavior in
rate Ȝb and some ʌi factors representing the i possible every working condition. To obtain a more realistic
stresses influencing devices reliability behaviour. The evaluation the above mentioned weighted approach is
purpose is to calculate the reliability performances of shared and the new MTBF formula reported in Eq.3 is
Distributed Maximum Power Point Tracking (DMPPT) defined. The MTBF weights Įi are, in this case,
boost converters [13-16]. determined on the percentages of hours the DMPPT
In detail, the RIAC 217Plus is used to calculate the converter worked at an ambient temperature in the i
reliability performances of Synchronous (SR) and Diode temperature range. As shown in Fig.6, the synchronous
Rectification (DR) boost converter shown in Fig.2. The boost operates for 760 hours at a temperature value in the
SR power optimizer is constituted by an input filter, an 15°C centered range corresponding to 18% of the
inductor, a Low-Side (LS) MOSFET Q1, a High-Side DMPPT converter total operating hours. In Table IV, Įi
(HS) MOSFET Q2, an output capacitor. values for all the considered temperature ranges are
Particularly, considered converters components [17- listed. The MTBFnew values for the considered power
20] are: optimizers are calculated in Eqs.4. Qualitative results
- Ceramic capacitor obtained for the two analyzed solutions point out better
- Low frequency, switching diode DR reliability performances respect to the SR ones. In
- Field effect, low frequency silicon transistor fact, the DR weighted MTBF is about 70% higher than
- Choke inductor the SR estimated value. As a matter of fact, repeating this
The DR power optimizer has a similar topology study on the same SR topology using converters
characterized by a diode used as HS switching device. characterized by different sets of commercial components
These two DMPPT converters assure different constituting their electrical parts lists (Bills Of Material),
performances in terms of efficiency as shown in [14-15].

345
MTBFwgt 0.03MTBF5%  0.036 MTBF10%  0.13MTBF20%  0.1MTBF25%  0.48 MTBF30%  0.2 MTBF100% (2)

TABLE I
217Plus FAILURE RATES FORMULAS
Device Failure rate by [RIAC-HDBK-217Plus]

MOSFET OMOS S G (OOBS DCOS TOS S  OEBS DCN S TE  OTCBS CRS DT )  OSJBS SJDT  OEOS
Diode Odiode S G (OOBS DCOS TOS S  OEBS DCNS TE  OTCBS CRS DT )  OSJBS SJDT  OEOS
Capacitor OC SGSC (OOBS DCOSTOS S  OEBS DCNSTE  OTCBSCRS DT )  OSJBS SJDT  OEOS

Inductor
OInductor S G (OOBS DCOS TO  OEBS DCNS TE  OTCBS CRS DT )  OEOS

Fig. 3. Converter devices ʌTE vs ambient temperature-


Fig. 6. South Italy typical temperature distribution ( year 2009).

TABLE II
SR DMPPT CONVERTER MTBF
Temperature MTBF
(°C) (106 operating hours)
15 20.55
20 20.41
25 19.82
30 15.95
35 9.70
Fig. 4. Converter devices ʌTO vs ambient temperature.

TABLE III
DR DMPPT CONVERTER MTBF
Temperature MTBF
(°C) (106 operating hours)
15 35.74
20 35.48
25 34.48
30 28.94
35 18.29
Fig. 5. Converter devices temperature rise TR vs ambient
temperature.

MTBFnew D15MTBF15qC  D 20MTBF20qC  D 25MTBF25qC  D30 MTBF30qC  D35MTBF35qC (3)

346
TABLE IV [5] Handbook of 217PlusTM Reliability Prediction Models
MTBFNEW FORMULA WEIGHTS [6] FIDES Guide 2004 Issue, A reliability methodology for
Temperature(°C) Įi electronic systems.
15 0.18 [7] H. Calleja, F. Chan, I. Uribe, “Reliability-Oriented
20 0.27
Assessment of a DC-DC Converter for Photovoltaic
25 0.20
Applications”, IEEE Power Electronics Specialist
30 0.28
35 0.07
Conference 2007, pp.1522-1527, June 2007
[8] X. Tian, “Design for Reliability and Implementation on
Power Converters”, Reliability and Maintainability
MTBF SR _ new 18.32 * 10 6 operating hours Symposium Proceedings, pp.89-95, January 2005.
(4) [9] G. Graditi, D. Colonnese, N. Femia “ Efficiency and
MTBF DR _ new 32 .30 * 10 6 operatig hours Reliability Comparison of DC-DC Converters for Single
Phase Grid Connected Photovoltaic Inverters”, Symposium
the MTBF values can change, but it is in the range (18- on Power Electronics, Electrical Drives, Automation an
Motion Proceedings, pp.140-147, June 2010.
20) millions operating hours. The above reported data
[10] S. Dhople, A. Davoudi, A. Domínguez-Garcia, P.
confirm[14-15] the less reliable SR boost behaviour due Chapman, “A Unified Approach to Reliability Assessment
to the presence of two MOSFETs used as HS and LS of Multiphase DC–DC Converters in Photovoltaic Energy
switching devices. In fact, MOSFETs failure rates are Conversion Systems”, IEEE Transactions on Power
higher than the other components ones, so affecting the Electronics, vol. 27, No. 2, pp. 739-751, February 2012.
whole converters reliability performances. [11] A. Verma, V. Gopika, R. Gopinath, U. Kumar “Reliability
prediction of semiconductor devices using modi¿ed
physics of failure approach”, International Journal of
V. CONCLUSION
Systems Assurance Engineering and Management, vol. 4,
In this paper, the problem of the reliability evaluation No. 1, pp. 33-47, March 2013.
of PV systems working in continuous changing [12] N. Femia, G. Petrone, G. Spagnuolo, G. Adinolfi, M.
conditions is carried out. Among all the available Vitelli, “Design of dc/dc Converters for DMPPT PV
Applications Based on the Concept of Energetic
reliability prediction models, the RIAC Handbook
Efficiency”, Journal of Solar Energy Engineering, vol.132,
217PlusTM is chosen for its better feasibility to express May 2010.
the reliability behaviour of a DMPPT converter in [13] G. Graditi, G. Adinolfi, “Energy performances and
operating and non operating modes, also considering its reliability evaluation of an optimized DMPPT boost
duty cycle and its cycling rate. Then the attention is converter”, IEEE International Conference on Clean
focused on the thermal stress, one of the most Electrical Power, vol. 1, p. 69-72, June 2011.
[14] G. Graditi, G. Adinolfi, N. Femia, M. Vitelli,
invalidating cause. A temperature distribution typical of
“Comparative Analysis of Synchronous Rectification
South Italy installation sites is considered and a new Boost and Diode Rectification Boost Converter for
weighted MTBF formula is proposed to suitably calculate DMPPT Applications” IEEE International Symposium on
reliability performances at different working conditions Industrial Electronics, vol. 1, p.1000-1005, June 2011.
for the SR and DR boost topologies. In detail, different [15] G. Graditi, G. Adinolfi, “Performances analysis of different
solutions are analyzed in terms of MTBF. SR boost DMPPT boost converters,” 26th European Photovoltaic
Solar Energy Conference, vol.1, pp.3703-3707, September
converters are considered and their MTBFs are compared
2011.
with a DR ones. The carried out analysis demonstrates [16] G. Graditi, G. Adinolfi, “Temperature Influence on
that SR power optimizers are characterized by worse Photovoltaic Power Optimizer Components Reliability”,
reliability performances than the DR ones. The cause of International Symposium on Power Electronics, Electrical
SR power optimizers worse reliability performances is Drives, Automation and Motion vol. 1, pp.1113-1118, June
related to MOSFETs employment as LS and HS 2012.
[17] M. Held, and K. Fritz. “Comparison and evaluation of
switching devices, but this is also the reason of their
newest failure rate prediction models: FIDES and RIAC
better efficiency behaviour respect to the DR boost. In 217Plus”, Microelectronics Reliability vol.49, pp.967-971,
conclusion, since an "optimum" DMPPT converter September 2009.
represents a trade-off solution in terms of efficiency, cost [17] D. Nicholls,"An Introduction to the RIAC 17PlusTM
and reliability, the designer, on the base of the tighter Component Failure Rate Models", Journal of the
constraint, chooses the more appropriate solution. Reliability Information Analysis Center, First Quarter
2007.
[18] D. Nicholls, "The 217PlusTM Capacitor and Diode Failure
REFERENCES Rate Models", Journal of the Reliability Information
[1] J. Jones, J. Hayes “A Comparison of Electronic Analysis Center, Second Quarter 2007 .
ReliabilityPrediction Models”, IEEE Transactions on [19] D. Nicholls, "The 217PlusTM Integrated Circuit and
Reliability”, vol. 48, No. 2, pp. 127-134, June 1999. Inductor Failure Rate Models", Journal of the Reliability
[2] M. Economou, “The Merits and Limitations of Information Analysis Center, Third Quarter 2007.
ReliabilityPredictions”, Reliability and Maintainability, [20] D. Nicholls, "An overview of the 217PlusTM System
Annual Symposium - RAMS, 26-29 Jan. 2004 Reliability Assessment Methodology", Journal of the
[3] MIL-HDBK-217F. Reliability Information Analysis Center, Fourth Quarter
[4] Telcordia SR.332 Issue 2, reliability prediction procedure 2006.
for electronic equipment.

347

You might also like