Rainflow Algorithm Based Lifetime Estimation of
Power Semiconductors in Utility Applications
Lakshmi GopiReddy1, Leon M. Tolbert1,2,
1 2
EECS Department, PEEMRC,
The University of Tennessee, Oak Ridge National Laboratory,
Knoxville, USA. Oak Ridge, USA.
Email: lgopired@utk.edu
Abstract—Rainflow algorithms are one of the best counting
methods used in fatigue and failure analysis popularly used in
semiconductor lifetime estimation models. However, the rainflow
algorithm used in power semiconductor reliability does not
consider the time dependent mean temperature calculation. The
equivalent temperature calculation proposed by Nagode et al is
applied to semiconductor lifetime estimation for the first time in
this paper. A month long arc furnace load profile is used as a test
profile to estimate temperatures in IGBT in a STATCOM for
reactive compensation of load. The degradation in the life of the
IGBT power device is predicted based on time dependent
temperature calculation.
Index Terms—Cycle counting, rainflow algorithms, power
semiconductor reliability, lifetime estimation, STATCOM
I. INTRODUCTION
With the integration of Smart Grid (SG) technology into the
bulk power system, the issue of reliability of the system is a
matter of concern. North American Electric Reliability
Corporation (NERC), in its report on “Reliability
Considerations from Integration of Smart Grid,” has
emphasized all the existing devices and systems such as
Phasor Measuring Units (PMUs), Flexible AC Transmission
Systems (FACTS), etc., need to be considered when assessing
reliability of the bulk power system [1]. According to an
industry based survey presented in [2] the power
semiconductor device failures are a major concern for
reliability of the power converter. According to MIL-std 757,
the IGBTs are the second most unreliable devices causing
failure of the inverter, after capacitors [2].
Reliability of power electronics has gained importance with
the European automotive initiatives for traction like LESIT
(Leistungs Elektronik Systemtechnik und Informations
Technologie), a Swiss government funded research program,
and the German project, RAPSDRA (Reliability of Advanced
Power Semiconductor Devices for Railway Traction
Application) [6][7]. Ciappa et al applied simplified thermo-
The submitted manuscript has been prepared by the Oak Ridge National
Laboratory, Oak Ridge, Tennessee 37831, managed by UT- Battelle for the
U.S. Department of Energy under contract DE-AC05-00OR22725.The
manuscript is authored by a contractor of the U.S. Government under contract
DE-AC05-00OR22725. Accordingly, the US Government retains a non-
exclusive, royalty-free license to publish from the contribution, or allow
others to do so, for U.S. Government purposes.
978-1-4799-2325-0/14/$31.00 ©2014 IEEE
Burak Ozpineci2, João O. P. Pinto3
3
Department of Electrical Engineering
Universidade Federal de Mato Grosso do Sul
Campo Grande -MS, Brasil
mechanical analysis to semiconductors. Recent IGBT
reliability applications included inverters integration for photo
voltaic, and wind applications, adjustable speed drives, matrix
converters, electric vehicle applications, aerospace
applications etc. [3]-[22]. Little research has been found in
reliability of IGBTs and the power converters in FACTS
applications.
Fig.1 demonstrates the steps involved in lifetime estimation
for semiconductor applications. In power semiconductor
applications, cycle counting algorithms are applied to
temperature data obtained from the loss calculation and
thermal model of the semiconductors, and by using lifetime
models based on temperature obtained from manufacturer’s
test data, a degradation model can be estimated assuming
linear accumulation of degradation with Miner’s rule.
To predict the life of a component subjected to a variable
load history, cycle counting methods are applied to reduce the
complex history into a number of events which can be
compared to the available constant amplitude test data [1].The
most common cycle counting methods are a) Peak Counting,
b) Level Counting, c) Hayes method, d) Racetrack method and
e) simple range counting [23]. Rainflow algorithm was
developed to incorporate the cycles formed by the hysteresis
loop characteristic of the stress-strain plot, shown in Fig. 1.
Rainflow algorithm has gained popularity mainly because it
provides the average value [15] and has little relative error
[12]. The analogy of rainflow counting first described by
Matsuishi and Endo [24] is derived from the rain flowing
(dripping) off the pagoda roofs, demonstrated in Fig. 2 (a).
The current state of power semiconductor reliability studies
is presented in section II. Section III presents Nagode and
Hack’s equivalent temperature calculation. Application of
rainflow algorithm to an IGBT based STATCOM providing
reactive compensation to an arc furnace based industrial load
Figure 1. Block diagram of the steps involved in lifetime estimation of power
semiconductors.
2293

Stress
σ B
D
E
H
C Strain, ε
A
I
a) b)
Figure 2. "Rainflow" on a) a load sequence and b) its stress-strain plot.
is presented in section IV. The analytical loss calculation
methodology is discussed in section V and the results are
presented.
II. APPLICATION OF RAINFLOW ALGORITHM TO
SEMICONDUCTOR LIFETIME ESTIMATION
The application of rainflow algorithms to semiconductor
reliability involves understanding the basics of thermal
expansion, stress and strain. The strain is the ratio of change in
length of a material to its actual length as a result of change in
temperature (ΔT). Thus, in order to obtain the strain in the
different packaging materials of the power semiconductor
module, it is important to consider the temperature gradient
(ΔT). For power semiconductor devices, the ambient
temperature is usually 25 oC, and the temperature data above
25 oC results in a strain in the different materials of the
module. Instead of considering the temperature gradient for
the complete dataset, the temperature gradient of only the
closed cycles determined by a rainflow algorithm is
considered. Hence, the rainflow algorithm can be easily
applied to temperature data.
The other cycle counting methods, unlike rainflow
algorithm, do not consider the stress-strain hysteresis
characteristic and can be easily extended to temperature data.
Ciappa et al specifically used range counting for temperature
data [8]. Bryant et al were the first to apply rainflow
algorithms to temperature data for semiconductor lifetime
estimation in traction applications [5]. Most of the research
that followed used rainflow algorithms which were written
using C and MATLAB software code by Nieslony available in
MATLAB Central for temperature data [26]. Fig. 3
demonstrates the application of rainflow algorithm by
Nielsony to a sample load profile.
Held et al proposed a relation between temperature gradient,
and mean temperature to lifetime or number cycles it can
withstand [6]. Choosing the lifetime estimation models are an
important step in lifetime estimation as seen in Fig.1. In the
past few years, extensive research in this area has led to
various lifetime models [22]. Bayerer’s model describes the
dependence of all operating conditions of a power
semiconductor and would be a good fit. However, the
constants of the model can be obtained only by experimental
validation which is time consuming [22]. Most manufacturers
follow the Coffin-Manson and/or the Coffin-Manson
2294
2
Figure 3. Rainflow algorithm of a signal with cycle’s information from code
available in MATLAB [26].
Arrhenius model [6] for the power cycling test data they
conduct and hence the constants are readily available. In this
paper, the Coffin-Manson-Arrhenius model is used for
estimating lifetime in conjunction with a rainflow algorithm
for cycle counting.
The drawbacks of conventional rainflow algorithms as
applied to power semiconductor lifetime estimation are
1) Conventional rainflow can be applied to extreme points
(peaks and valleys) only and so the load data is
transformed to a data with only peak and valley
information. Hence it cannot be applied to real-time data.
2) Half-cycles are counted only at the end of data [14]. It is
difficult to calculate remaining useful life in between the
load points.
3) The mean temperature calculation is independent of the
time period.
The first two drawbacks are addressed by Mussallam and
Johnson in [25]. An online real-time rainflow algorithm for
half-cycle counting was proposed and applied to a metro-train
application. The issues of online half-cycle counting and
temperature bins were addressed.
The equivalent temperature calculation described in section
IV gives a time-dependent relation to lifetime estimation. It is
followed in fatigue analysis but has not been considered for
semiconductor analysis.
III. EQUIVALENT TEMPERATURE-TIME DEPENDENT FATIGUE
MODEL
In conventional rainflow algorithms, stress or strain data
over time are input to the algorithm. The data is reconsidered
with only positive and negative peak values. After the
criterion for cycle counting is met, the corresponding
amplitude and mean values of stress or strain are calculated
for each cycle. The output is thus a matrix with cycle
information, mean value and amplitude of stress or strain that
forms a cycle. Usually the mean and amplitude values are
plotted as a histogram. The stress and strain plots are
temperature dependent as shown in Fig. 4 for aluminum and
 363K
403K
443K
523K

a) b)
Figure 4. Stress-strain graph variation with temperature of a) solder [32] b) aluminum specimen [31].

solder [31][32]. As the temperature increases, the stress the

metals can withstand decreases. Hence, a temperature
dependent rainflow algorithm was proposed by Nagode and
Hack.
Nagode et al [26] presented the online model for
temperature dependent fatigue, based on the four point
rainflow algorithm and damage estimation for moderate but
variable temperatures. The algorithm is based on equivalent
temperature calculation and recalculation. It consists of four
variables as input, the temperature, stress or strain, time, and
equivalent temperature. The equivalent temperature is
calculated after every step and updated. The temperature
dependent rainflow algorithm does not take creep and other Figure 5. Time considerations for equivalent temperature model.
effects into account, but it enables improved fatigue life
estimates [26].
temperature amplitude (gradient), ∆T, and mean temperature,
Consider the 4-point temperature profile shown in Fig.
Tm, even a slight difference in temperatures for large number
5.Ti+1 and Ti+2 forms a closed cycle according to four-point
of load points, would affect the overall lifetime, that can be
rainflow algorithm. The equivalent temperature of a closed
seen from temperature dependent stress-strain plots [31][32].
loop for 4-point algorithm starting at Ti+1, reaching Ti+2, and
arriving back at Ti+1 shown in Fig. 5 is calculated as
IV. LIFETIME ESTIMATION OF IGBT IN A STATCOM FOR AN
T ei +1 > T t , or , T ei + 2 > T ⎫
INDUSTRIAL LOAD
⎧ max( T ei +1 , T ei + 2 )
(1)
⎪ t p − t i+2 ⎪ A STATCOM is a voltage source converter connected to
T ei = ⎨ t i + 2 − t i +1 ⎬
T + T otherwise
⎪ t p − t i +1 ei +1 t p − t i +1 ei + 2 ⎪ the grid to provide reactive compensation and harmonic
⎩ ⎭
elimination, as shown in Fig. 6. In order to estimate the
tp- is the time at which the half cycle is closed, as described lifetime of an IGBT in a STATCOM, a long term load
by Fig. 5. The equivalent temperature calculation is preferred variation for a power system should be obtained. In this case,
over the mean temperature calculation previously used an arc furnace based industrial load in Brazil, whose power
because it considers the dependence of time on the lifetime. factor values are monitored every 15 minutes throughout the
Table I illustrates the difference in the mean calculations for day over a period of a month, shown in Fig. 7, is considered
the signal in Fig. 3 for conventional and equivalent for analysis and lifetime estimation of IGBT. Every day
temperature based calculations. Four closed cycles are during the weekdays from 19:30 to 23:30 hrs the arc furnace
determined by the four-point algorithm criterion. The mean load is switched off and the capacitor banks are used to
temperatures using conventional calculation method and provide reactive compensation of the grid current. During
equivalent temperatures are shown in the 6th and 9th columns these off-peak hours, the output power is low but reactive
of Table I. It also displays the equivalent time values from the power is injected into the grid, resulting in low leading power
calculations. A major difference in the mean temperature factor. The loss calculation model and the thermal model
described in [28][29] are used in this paper. It is assumed that
calculation can be observed for the second and third cycles
STATCOMs connected in parallel compensate for overall
where the mean temperatures by conventional calculation are
reactive power of the load and a single STATCOM can
57.0oC and 54.2oC, while that by equivalent temperature compensate the maximum rated power, 180 kVA.
method are 58.7oC and 55.3oC. Since lifetime is dependent on

2295
 TABLE I
A STEP BY STEP ANALYSIS OF RAINFLOW ALGORITHM WITH AND WITHOUT EQUIVALENT TEMPERATURE CALCULATION
APPLIED TO THE EXAMPLE PROFILE IN FIG. 2.
Conventional rainflow
Start and Cycle formed Equivalent temp calculation
end Points (Range ) using
extreme considered Four-point Start and Mean Equiv. start Mean
Amplitude Amplitude
points rainflow end times temp and end times Temp
1-2 77.96-36.93 - 3-9 - - 3-9 - -
2-3 36.93-72.03 No 9-18 - - 9-18 - -
3-4 72.03-42.71 No 18-21 - - 18-21 - -
4-5 42.7-70.7 Yes 21-24 28 56.7 21-24 28 56.58
3-6 72.04-42.17 Yes 18-27 29.87 57.10 18-21.05 29.87 58.7
2-7 36.9-76.46 No 9-30 - - 9-19.15 - -
7-8 76.46-39.82 No 30-36 - - 19.15-36 - -
8-9 39.82-60.8 No 36-42 - - 36-42 - -
9-10 60.8-47.58 Yes 42-48 13.22 54.19 42-48 13.22 55.3
8-11 39.82-71.54 No 36-51 - - 36-43.34 - -
11-12 71.54-57.48 Yes 51-57 14.06 64.51 43.34-57 14.06 64.82
12-13 57.48-75.8 No 57-60 - - 57-44.04 - -

TABLE II
LIST OF PARAMETERS AND DESCRIPTIONS
Il load current of the system
Ia inverter Phase current
Φ power factor angle of the load
Rfd on state resistances of diode
Roni on state resistances of IGBT
Von_d forward voltage drop across the diode
Von_i forward voltage drop across the IGBT
turn on switching losses from test
E on_test
conditions
E off_test turn off switching losses from test
DC voltage across the switch during the
VDC
current operation
DC voltage across the switch with the
VDC_test
test conditions from the datasheet
current across the IGBT during test
Ictest
conditions
fsw switching frequency

TABLE III.
Figure 6. STATCOM model connected to grid. LIST OF IGBT AND DIODE PARAMETERS FOR LOSS
CALCULATION
Analytical model for power calculation is discussed in section Ron Von Eon Eoff (mJ)
IGBT 4 mΩ 1.8 V 22 mJ 15
V. Diode 11 mΩ 1.0 V 32(reverse recovery)

V. ANALYTICAL POWER LOSS CALCULATION

⎛ V on _ d I 1 × sin( φ ) R on _ d ⎞
Table II is a list of symbols, and parameters used in the Pcond _d = (I 1 × sin φ )⎜⎜ + ⎟
⎟
(3)
equations in this paper. ⎝ 2Π 8 ⎠
a. Power Factor Correction Conduction losses are dependent on the square of the
For a lagging/leading power factor current drawn by the current and sine of the power factor angle. The switching
load, the real power is provided by the source while the losses can be approximated by the turn-on and turn-off losses
reactive power is provided by the STATCOM, i.e. the from the switching test values listed in the datasheet. In
magnitudes of the source and inverter currents are iactive=is=il general, switching losses are linearly dependent on operating
× cos(Φ) and inonactive=ic=±il × sin(Φ) respectively. current and voltage.
For inverter current, Ia=Il×sin (Φ), and power factor, pf = 0 In power factor correction, the inverter current is
(since the voltage and current are 90o apart) the equations are ic=il × sin(Φ), and the switching losses can be written as
transformed to [28] [29]. P =
f ⎡
× E
sw
×
V
×
I sin( Φ )
+E ×
V
×
I sin( Φ ) ⎤ (4) DC
⎢
DC l l
⎥
sw
π ⎣⎢
on _ test
V DC _ test ICtest
off _ test
V DC _ test ICtest ⎦⎥
⎛ Von _ i I1 × sin( φ ) Ron _ i ⎞
Pcond _i = (I1 × sin φ )⎜⎜ + ⎟
⎟
(2) From (4), it can be deduced that as the power factor decreases,
⎝ 2Π 8 ⎠ the switching losses increase. The parameters required for
calculation of losses are listed in Table III.

2296
 The IGBT current and temperature distribution while
maintaining unity power factor of the power system is shown
in Fig. 8 and Fig. 9, respectively. The temperature varies from
40oC to a maximum of 100oC.
This temperature profile of a semiconductor is applied to
fast rainflow algorithm [30] in conjunction equivalent
temperature calculation [26]. The remaining useful life and
lifetime estimation methodology are described in detail in
[30].
The relation between the life in terms of number of cycles
and the mean temperature, and the amplitude of temperature is
given by the Arrhenius-Coffin-Manson’s lifetime model
proposed by Held et al [6] as shown in (5)
Q temperature calculation, and the histogram plots are shown in
( )
α
N f (Tm , ΔT j ) = A × ΔT j × exp RTm
(5)
where A, α are constants and are module dependent, R is the
gas constant (8.314 J1mol.K), Tm is the mean junction
temperature in one power cycle in Kelvin, and the internal
energy is Q = 7.8 × 104 JMol-l (or about 0.8eV). ΔTj is the
variation of the junction temperature. By curve fitting the
supplier’s power cycling data to that of (4), the constants are
derived as α equals -6.14, for mean temperature, Tm of 25 oC,
and A=1300. The constant values were obtained by curve
fitting the manufacturer’s data for power cycling to (5).
Linear accumulation of the fatigue damage is considered.
The degradation is given by [1]
N (ΔT , Tm )
D=∑ (6)
N f (ΔT , Tm )
Using (5) and (6), the degradation is 6.8×10-3 for the load
profile shown in Fig. 7. The total cycles for the load profile
are 533 using four-point rainflow algorithm [25].T
Degradation is calculated with and without equivalent
Fig. 10 a) and b) respectively. For conventional rainflow
algorithm calculation, the degradation accounts for 0.029%
degradation for the month while the degradation calculation
with equivalent temperature calculation accounts for 0.033%.
This emphasizes the importance of equivalent temperature
based lifetime calculation for semiconductors in long-term
calculations.

Figure 7. Arc furnace based industrial load power factor variation monitored for every 15 minutes for a period of month.

Figure 8. Inverter current in amperes for unity power factor of the load.

Figure 9. IGBT temperature variation for the reactive compensation of the load.

2297
 a)
Figure 10. Rainflow histogram of load profile in Fig. 7 with a) conventional calculation and b) equivalent temperature calculation.
VI. CONCLUSION
Rainflow algorithm is one of the popular counting
algorithms used in calculating lifetime, specifically for
semiconductor lifetime in power converters. This paper
presents a comparison of conventional rainflow algorithms
used in lifetime calculation. For the first time, the applicability
of rainflow algorithm for semiconductor lifetime in power
converters and the drawbacks of using conventional rainflow
algorithms for semiconductor lifetime are discussed in detail.
Equivalent temperature calculation proposed by Nagode and
Hack is introduced to semiconductor lifetime calculation in a
STATCOM. An arc furnace based industrial load data,
monitored every 15 minutes for a month, is used as mission
profile to estimate temperature of IGBT in STATCOM when
compensating reactive power of the load. For the load profile,
the degradation using conventional and equivalent temperature
based rainflow algorithm is used to calculate lifetime and a
difference of 4% in degradation calculation was observed.
Thus, it is important to accurately calculate lifetime using
equivalent temperature calculation.
ACKNOWLEDGMENT
This work made use of Engineering Research Center Shared
Facilities supported by the Engineering Research Center
Program of the National Science Foundation and DOE under
NSF Award Number EEC-1041877 and the CURENT
Industry Partnership Program.
REFERENCES
[1] North American Electric Reliability Corporation “Reliability
Considerations from Integration of Smart Grid,” Report, Dec. 2010.
[2] S. Yang, A. T. Bryant, P. A. Mawby, D. Xiang, L. Ran, and P. Tavner,
“An industry-based survey of relibility in power electronic converters,”
IEEE Transactions on Power Electronics, Nov. 2010, vol. 25, pp. 2734-
2752.
[3] J. Liu and N. Henze, “Reliability Consideration of Low-Power Grid-tied
Inverter for Photovoltaic Application,” 24th European Photovoltaic
Solar Energy Conference and Exhibition, Hamburg/Germany, 21-25
Sept. 2009, pp. 3688-3694.
[4] D. Hirschmann, D. Tissen, S. Schröder, and R. W. De Doncker,
“Reliability Prediction for Inverters in Hybrid Electrical Vehicles,” in
Proc. IEEE Power Electronics Society Conference, Jeju, Korea, Jun.
2006, pp. 1---6.
2298
b)
[5] T. Bryant, P. A. Mawby, P. R. Palmer, E. Santi, and J. L. Hudgins
“Exploration of Power Device Reliability Using Compact Device
Models And Fast Electro-Thermal Simulation,” IEEE Ind. Appl
Conference, Oct. 2006, vol. 3, pp. 1465-1472.
[6] P. Held, G. Jacob, P. Nicoletti, P. Scacco, M.-H. Pooch; “Fast Power
Cycling Test for IGBT Modules in Traction Application,” Proc. Power
Electronics and Drive Systems, 1997, pp. 425-430.
[7] H. Berg, E. Wolfgang, “Advanced IGBT Modules for Railway Traction
Applications: Reliability Testing,” Microelectronics Reliability, vol. 38,
no. 6-8, 8 June 1998, pp. 1319-1323.
[8] M. Ciappa, F. Carbognani , and W. Fichtner, “Lifetime Prediction and
Design of Reliability Tests for High-Power Devices in Automotive
Applications,” IEEE Trans. Device Material. Reliability, Dec. 2003,
vol. 3, no. 4, pp. 191–196.
[9] L. Wei, R. J Kerkman, R. A. Lukaszewski, “Evaluation of Power
Semiconductor Power Cycling Capabilities for Adjustable Speed Drive,”
Proc. IEEE Industrial Application Society Conference, Edmonton,
Canada, 2008, pp.1-10.
[10] L. Wei, T. Lipo, R. A. Lukaszewski, “Analysis of Power Cycling
Capability of IGBT Modules in a Conventional Matrix Converter,” Proc.
IEEE Industrial Applications Society conference, Edmonton, Canada, 5-
9 Oct. 2008, pp. 1 – 8.
[11] M. Aten, G. Towers, C. Whitley, P.Wheeler, J. Clare, K. Bradley,
“Reliability Comparison of Matrix and Other Converter Topologies,”
IEEE Transactions on Aerospace and Electronic Systems, July 2006, no.
3, vol. 42, pp. 867-875.
[12] K. Mainka, M. Thoben, O. Schilling, “Lifetime Calculation for Power
Modules, Application and Theory of Models and Counting Methods,”
European Conference on Power Electronics and Applications (EPE),
Aug. 30- Sept. 1, 2011, pp. 1-8.
[13] M. Musallam, M. Johnson, C. Yin, C. Bailey, and M. Mermet-Guyennet,
“Real-time Life Consumption Power Modules Prognosis Using On-line
Rainflow Algorithm in Metro Applications,” IEEE Energy Conversion
Congress &Expo–ECCE, Atlanta, Georgia, USA, September 12–16,
2010, pp. 970-977.
[14] M. Mussallam, C. M. Johnson, “An Efficient Implementation of the
Rainflow Counting Algorithm for Life Consumption Estimation,” IEEE
Transactions on Reliability, Dec. 2012, vol. 61, no. 4, pp. 978-987.
[15] X. Perpina, X. Jorda, M. Vellvehi, J. Rebollo, M. Mermet-Guyennet,
“Long-Term Reliability of Railway Power Inverters Cooled by Heat-
Pipe-Based Systems,” IEEE Transactions on Industrial Electronics, July
2011, vol. 58, no. 7, pp. 2662-2672.
[16] Y. Xiong and X. Cheng, “Prognostic and Warning System for Power-
Electronic Modules in Electric, Hybrid Electric, and Fuel-Cell
Vehicles,” IEEE Transactions on Industrial Electronics, June 2008, vol.
55, no. 6, pp. 2268 - 2276.
[17] G. Petrone, G. Spagnuolo, R. Teodorescu, M. Veerachary, M. Vitelli,
“Reliability Issues in Photovoltaic Power Processing Systems,” IEEE
Trans. on Industrial Electronics, Jul. 2008, vol. 55, no. 7, pp. 2569-
2580.
[18] B. Abdi, A.H. Ranjbar, G.B. Gharehpetian, J. Milimonfared, “Reliability
Considerations for Parallel Performance of Semiconductor Switches in
 High-Power Switching Power Supplies,” IEEE Trans. on Industrial
Electronics, Jun 2009, vol. 56, no. 6, pp. 2133-2139.
[19] F. Chan, H. Calleja, “Reliability Estimation of Three Single-Phase
Topologies in Grid-Connected PV Systems,” IEEE Trans. on Industrial
Electronics, Jul 2011, vol. 58, no. 7, pp. 2683-2689.
[20] K. Nguyen-Duy, Tian-Hua Liu, Der-Fa Chen, J.Y. Hung, “Improvement
of Matrix Converter Drive Reliability by Online Fault Detection and a
Fault-Tolerant Switching Strategy,” IEEE Trans. on Industrial
Electronics, Jan 2012, vol. 59, no. 1, pp. 244-256.
[21] J. Liu and N. Henze, “Reliability Consideration of Low-Power Grid-
Tied Inverter for Photovoltaic Application,” 24th European Photovoltaic
Solar Energy Conference and Exhibition, Hamburg/Germany, 21.-25.
September 2009.
[22] C. Busca, R. Teodorescu, F. Blaabjerg, S. Munk-Nielsen, L. Helle, T.
Abeyasekera, P. Rodriguez, “An Overview of the Reliability Prediction
Related Aspects of High Power IGBTS in Wind Power Applications,”
Proc. of European Symposium on the Reliability of Electron Devices,
Failure Physics and Analysis, September–November 2011, vol. 51, no.
9–11, pp. 1903–1907.
[23] J. A. Bannantine, J. J. Comer, J. L. Handrock, “Fundamentals of Metal
Fatigue Analysis,” Englewood Cliffs (NJ): Prentice-Hall, 1990.
[24] M. Matsuishi, T. Endo, “Fatigue of Metals Subjected to Varying Stress,”
Japan Society of Mechanical Engineers, Fukuoka, Japan, March 1968.
[25] C. Amzallag , J. P. Gerey, J. L. Robert, J. Bahuaud , “Standardization of
the Rainflow Counting Method for Fatigue Analysis,” International
Journal of Fatigue, 1994, vol. 16, no. 4, pp. 287–93.
2299
[26] M. Nagode, M. Hack, “An Online Algorithm for Temperature
Influenced Fatigue Life Estimation: Stress–Life Approach,”
International Journal of Fatigue, 2004, vol. 26, pp. 163–171.
[27] A. Nieslony, Rainflow Counting Algorithm, MATLAB File Exchange
Central, 17 Feb, 2003.
[28] L. GopiReddy, L. M. Tolbert, B. Ozpineci, Y. Xu, D. T. Rizy, “Impact
of Power Factor Correction and Harmonic Compensation by
STATCOM on Converter Temperature,” IEEE Energy Conversion
Congress and Exposition, 17-22 Sept. 2011, pp. 1928-1934.
[29] L. GopiReddy, L. M. Tolbert, B. Ozpineci, Y. Xu, D. T. Rizy,
“Reliability of IGBT in a STATCOM for Harmonic Compensation and
Power Factor Correction,” IEEE Applied Power Electronics Conference
and Exposition, 5-9 Feb. 2012, pp. 783-788.
[30] L. GopiReddy, L. M. Tolbert, B. Ozpineci, “Lifetime Prediction of
IGBT in a STATCOM Using Modified-Graphical Rainflow Counting
Algorithm,” IEEE Annual Conference of Industrial Electronics Society,
Oct.25-29, 2012, Montreal, pp. 3425 – 3430.
[31] D. Sharma, A. Yadav, N Hasen, “Strength Determination of Aluminium
at Different Temperatures,” International Journal of Computer Science
& Management Studies, August 2011, vol. 11, no. 2, pp. 83-87.
[32] X. Chen, L. Zhang, M. Sakane, H. Nose, “Tensile Stress-Strain Curves
Modeling of Four Kinds of Solders with Temperature and Rate
Dependence,” Key Engineering Materials, vol. 297-300, pp. 905.