Comparison Of Reliability Prediction Methods Using Life Cycle Cost

Analysis
Adithya Thaduri, Lulea University of Technology
A K Verma, PhD, Stord/Haugesund University College
Uday Kumar, PhD, Lulea University of Technology
Keywords: 217+, BJT Transistor, Instrumentation Amplifier, Life cycle cost, Physics of Failure, Reliability prediction
SUMMARY AND CONCLUSION
In this paper, it was discussed on the several reliability
prediction models for electronic components and comparison
of these methods was also illustrated. A combined
methodology for comparing the cost incurring for prediction
was designed and implemented with an instrumentation
amplifier and a BJT transistor. By using the physics of failure
approach, the dominant stress parameters were selected on
basis of research study and were subjected to both
instrumentation amplifier and BJT transistor. The procedure
was implemented using the methodology specified in this
paper and modeled the performance parameters accordingly.
From the prescribed failure criteria, mean time to failure was
calculated for both the components. Similarly, using 217 plus
reliability prediction book, MTTF was also calculated and
compared with the prediction using physics of failure. Then,
the costing implications of both the components were
discussed and compared them. From the results, it was
concluded that for critical components like instrumentation
amplifier though the initial cost of physics of failure prediction
is too high, the total cost incurred including the penalty costs
were lower than that of traditional reliability prediction
method. But for non-critical components like BJT transistor,
the total cost of physics of failure approach was too higher
than traditional approach and hence traditional approach was
much efficient. Several other factors were also compared for
both reliability prediction methods.

concentrates on this issue on which reliability prediction

method was to be selected for calculation of replacement
human factors, time and reputation costs.
The efficient reliability prediction is needed before the
installation of the components and appropriate changes can be
made at the design stages. Conventional reliability prediction
methods such as Mil-Hdbk 217F, Telcordia, Bell Core,
PRISM etc implements constant failure methods and believed
to be true in the era of 1980s and 1990s [3]. But due to the
advancements in the technology, these methods are no longer
adequate to define the characteristics as there are so much
variability in the design and fabrication of devices. Especially,
electronics spreads out rapid developments in all the aspects
and for control aspects, miniature and cost effectiveness.
Some of the devices are used in safety, security and military
areas where the availability is the major concern and incorrect
operation leads to the unsafe shutdown. Moreover reliability
aspects and prediction is critical for these components and this
paper provides advanced physics of failure methodology for
finding failure characteristics and reliability indices. The
following Table 1 demonstrates various traditional prediction
methods the differences between the values of time to failures
of DC-DC converter constraints the ambiguity and risk in
selecting appropriate figure [1].
Table 1: Comparisons of different reliability prediction
models [1]

1 INTRODUCTION
The important considerations for the customer to select an
item depend on the reliability, cost, availability and
maintainability. To deal with the perfect repair/replacement
costs, selection of the time to repair or time to replace was
properly calculated or else inaccuracy in reliability prediction
of this values leads to increase in the excessive costs even the
actual component has higher reliability. Hence, reliability
prediction of these components along with life cycle costs was
needed to be considered for effective working of the system.
Accordingly, there would not be same replacement times for
all the components with variable reliability and costs due to
the variability in criticality of the each component. This paper

978-1-4673-4711-2/13/$31.00 2013 IEEE

Physics of failure prediction methodology lay emphasis

on the root cause of failure following fundamentals of physics
of materials also considered as white box testing. Electronic
devices were fabricated of different materials like Silicon,

Germanium, Copper, Aluminum etc., with predefined

technology and processes like deposition, etching and masking
under controlled environment [5]. Most of failures happened
due to changes in thermal activation, changes in defects of
molecules, activation energy in the materials and
correspondingly there was degradation in the technology and
performance factors. Depends on the failure characteristics,
several failure mechanisms were modeled and categorized
according to the cause, material, and failure point location etc.,
produced in the literature. This method requires sophisticated
tools for failure analysis and advanced tools for analyzing the
simulated data. Still, this methodology also has challenges like
insufficient data from the manufacturer, needs expert
judgment and also time taking process.

cycle costs of each stage.

2 METHODOLOGY
The block diagram shown in Figure 2 describes the flow
of prediction mechanism adopted in this paper.
2.1 Component Description
The first step in this methodology was to describe the
component with all the necessary and essential information for
analyzing the failure and calculating the reliability indices.
The sources required for the information on the component
were: materials, processes, layout diagrams, technology,
architecture, design, criticality, cost, datasheet, manuals, field
data and any similar item analysis that was analyzed earlier.

Figure 1: Short idea of Modified approach

On the other hand, statistical methods were widely
available in order to find out the reliability indices from the
test data. This method was also considered as black box
testing which concentrate on available data and proper model
was selected depends on the application. There were
possibilities to analyze the data and generated model to extract
enormous amount of information to characterize the
performance parameters. Some of them include design of
experiments, accelerated testing, regression analysis, etc.
Even, there were several tools available for model selection,
mathematical formulation and model analysis. This
methodology has some advantages like time consuming, no
need for manufacturer data and parameter analysis.
Inclusion of multidisciplinary science and engineering
approaches was very effective in solving of real life problems
and our modified approach was combination of both physics
of failure (deterministic) and statistical (probabilistic)
approaches in Figure 1. This advancement methodology first
starts with the proper understanding of basic failure physics of
the component and process the physics of failure
methodology. This knowledge was fed to the statistical
approach to further refining of data for accurate models.
Finally, we get three faces of models; history and literature,
white box and black box models and these were sent to
decision support system. Simultaneously, the cost of each
prediction method was also compared for the decision on
which method needed to be implemented on the criticality of
the component. Other hidden costs include the amount of time
that was spent on the entire processing of the prediction as it
also takes care of the entire decision strategy. These details
were also incorporated in this paper. The next section provides
the entire block diagram for the reliability prediction with life

Figure 2: Reliability prediction methodology for comparisons

2.2 Literature Survey and Failure Survey
To study the failure behavior of the component, the
literature survey was required to understand the behavior of
several factors that affecting the performance. The information
essential for this study were: expert reviews, stress factors,
failure criteria, failure mechanisms, failure modes, failure
analysis, degradation analysis and other factors.
2.2.1 Traditional Reliability Methodology
Based on the operational stresses and field environment,
reliability indices were calculated using MilHdbk 217F and
other traditional methods.
2.2.2 Physics of Failure Methodology
The advanced methodology, physics of failure lay
emphasis on the root cause of failure inherently depends on
the operational stress factors, environment and physical
characteristics of the device [5]. From the information on
component description and literature survey, appropriate
failure phenomenon and failure criteria were selected.

2.3 Failure Mechanisms

There was several failure mechanisms reported in the
literature characterized on operational environment, stress
parameters, level of approach, technology etc. Some of list of
the failure mechanisms that was not limited to
electromigration, hot carrier defects, time dependent dielectric
breakdown, negative bias temperature instability, corrosion,
fatigue, solder reliability, stress migration, soft errors
(radiation effects) etc [5,6]. There were several failure time to
failure models associated with each mechanism and
appropriate model was picked for the application. According
to the selected component, the appropriate failure mechanism
or degradation mechanisms were studied.
2.4 Failure Analysis
From the literature, an appropriate failure analysis was
selected to examine and illustrate the failure of the component
and root cause of failure by electrical characterization or by
using non-destructive testing by making use of sophisticated
instruments like scanning electron microscope, infrared
spectroscopy, thermal analysis etc [9]. For some of the
components where there was no information on the failure
mechanism, this step was need to be implemented beforehand
to acquire information on failure characteristics.
2.5.1 Experimental Planning
From the acquired data, the next process of
experimentation was planned for testing and reliability
prediction. The desired circuitry was designed and fabricated
using printed circuit board. This step also includes the number
of samples, stress parameters and experimental setup for
further testing of the component.
2.5.2 Simulations
There are several tools were available to carry out
simulations depends on finite element analysis, circuit
simulations and parametric analysis. This step was performed
simultaneous to the experiment testing to reduce the time
effort for the procedure.
2.6 Design of Experiments
Design of Experiments was very advanced and efficient
methodology to find the prominent factors, component
selection and variability analysis of the component. The
prominent approach, Taguchi method was implemented here.
In order to get best out of design of experiments, a modified
methodology was designed as two-step DOE. In general, there
was uncertainty in selecting the stress factors for design of
experiments and accelerated testing. Hence, at first screening
step, the test was designed to know variability of stress on the
effect of performance parameters. In the second testing step,
the levels of the stress were aggressive which defines the
degradation of the performance parameters.
2.7 Accelerated Testing
The input pattern obtained for degradation from modified

design of experiments was applied in the accelerated testing

step [13]. As from the analysis, this particular pattern leads to
further degradation over the accelerated time.
2.8 Regressions and Failure Modeling
The data collected from both design of experiments and
accelerated testing was used for statistical data and modeling
analysis using various methods such as response surface
regressions, regression methods, support vector regression
and other tools to quantify the stress parameters and its
behavior on the performance of the device. This data was also
useful for failure models obtained from failure mechanisms.
2.9 Compare prediction models
The reliability indices calculated from traditional
approaches, physics of failure models, models generated from
statistical analysis and simulations were compared in this step.
In accordingly, the reliability growth techniques were
implemented and appropriately the repair/replacement
mechanism was discussed. The actual decision was selected
on basis of available inventory, importance of the component,
resources available, selection of vendor and several other
factors such as risk, uncertainty and human factors.
2.10 Factor Calculations
In parallel, the factors like time taken, human factors and
risk was calculated at each step. With regarding the cost, this
cost only provides the prediction cost of the entire block
diagram in Fig 2. The total cost thus calculated including the
penalty cost that was incurred by the repair/replacement costs.
In the most of electronic components, the probability of
replacement was more than the repair. In this case study, we
are considering the replacement cost of the component.
3 FAILURE STUDIES OF COMPONENTS
3.1 Instrumentation Amplifier
An instrumentation amplifier is a gain differential device
that provides the output with high common mode rejection
ratio and high accuracy. This differential amplifier is widely
used in several electronic applications even precise
environments and also considered as most versatile amplifier.
This device (INA118) require very high input impedance, low
bias and offset currents, low noise and balanced inputs in
order to minimize common mode gain. The basic building
blocks of the instrumentation amplifier are op-amps.
Technically, it is equivalent to standard op-amp, the each of
the inputs of it is driven by two another op-amps to buffer the
inputs and to produce desired output for impedance matching.
The gain of the amplifier is defined as the output voltage
divided by input voltage [10]
/ 1
2
1 2 1/
3/ 2 (1)
Since, the voltage across Rgain equals Vin (V1-V2), the
current through Rgain will equal (VIn/Rgain). Amplifiers A1
and A2 on left side will operate with gain and amplify the
input signal. Rgain can be useful to tune for desired gain
without effecting common mode gain and error. The

Fig 3: Test circuit for Instrumentation Amplifier, INA118

differential signal will be increased by gain, but the commonmode error will not increases, so the ratio (Gain
(VDIFF)/(VError CM)) will increase. Thus, CMRR will
theoretically increase in direct proportion to gaina very
useful property. CMMR is the tendency of the devices to
reject the input signals common to both input leads. Ideally, a
differential amplifier takes the voltages V1 and V2 on its two
inputs and produces an output voltage

1
2
(2)
where Ad is the differential gain. But the output of a real
differential amplifier is described as

1
2
1
2
(3)
where Acm is the common-mode gain, which is typically
much smaller than the differential gain. Thus CMRR is
defined as the ratio of the powers of the differential gain over
the common-mode gain, measured in decibels
10

20

(4)

3.1.1 Effect of Input Voltage Difference

The both gain and CMMR equations were considered as
performance parameters and sensitive to the input voltage
difference that further degrades the amplifier. Hence, the input
voltage was selected as stress parameter.
3.1.2 Effect of Temperature
The temperature dependence of bipolar transistors
depends on a multitude of parameters affecting the bipolar
transistor characteristics in different ways. Important effect is
the temperature dependence of the current gain. Since the
current gain depends on both the emitter efficiency and base
transport factor [15].
The emitter efficiency depends on the ratio of the carrier
density, diffusion constant and width of the emitter and base.
As a result, it is not expected to be very temperature
dependent. The carrier densities are linked to the doping
densities. Barring incomplete ionization, which can be very
temperature dependent, the carrier densities are independent of
temperature as long as the intrinsic carrier density does not
exceed the doping density in either region. The width is very
unlikely to be temperature dependent and therefore also the
ratio of the emitter and base width. The ratio of the mobility is
expected to be somewhat temperature dependent due to the
different temperature dependence of the mobility in n-type and

p-type material.
Leakage current ICO and increase with temperature.
The DC hFE increases exponentially. The AC hfe
increases, but not as rapidly. It doubles over the range of -55o
to 85oC. As temperature increases, the increase in hfe will
yield a larger common-emitter output, which could be clipped
in extreme cases. The increase in hFE shifts the bias point,
possibly clipping one peak. The shift in bias point is amplified
in multi-stage direct-coupled amplifiers. The solution is some
form of negative feedback to stabilize the bias point. This also
stabilizes AC gain [11].
As from the studies from BJT technology, temperature
and radiation is selected as stress parameters. The emitter and
collector current of npn BJT is given as Equation (5) and (6).
1
1
The output voltage VCE is given as in Equation

(5)
(6)

(7)
Where Reff is effective output resistance at the output, IES
= reverse saturation current at base-emitter diode, T =
common base forward short circuit gain, VT = Thermal
Voltage kT/q, VBE = base-emitter Voltage, VCE = basecollector Voltage, VCC = Source Voltage typically 5V/10V.
In Eber-Moll Model, IC grows at about 9%/0C if you hold
VBE constant and VBE decreases by 2.1mV/ 0C if you hold IC
constant with the temperature.
Since both the currents depend on temperature parameter
VT, the raise in the temperature leads to vary these parameters
which finally lead to degrade the performance of op-amps in
turn the instrumentation amplifier. Hence, temperature was
considered as another stress parameter which leads to reduce
the gain and CMMR of instrumentation amplifier. In this
paper, we selected IN128 for failure investigation. Since there
was no possible failure mechanism associated with it in the
literature, the failure model was not considered. But from this
root cause analysis, the appropriate information was drawn out
to make further testing.
The failure mechanism involved in the instrumentation
was the degradation of the device parameters like gains,
thermal voltage and intrinsic voltage and currents. This failure
could be assessed using the standard failure analysis method
which is electrical characterization.
3.2 BJT Transistor
In this study we selected 2N2222, a normal BJT transistor
for failure study. At the field, this component was exposed to
nuclear radiation and hence we interested to test the IC for
radiation. As BJT technology was sensitive to the temperature
as described in case of instrumentation amplifier, we are
considering it as another stress parameter. Even in this case
also, there were no failure models for failure mechanisms
observed from the literature, the methodology was continued
on the information gathered from the root cause analysis and
field environment failed data. For further hypothesis testing,
experimentation was conducted to achieve the results of

failure data. The test circuit of 2N2222 is given below in Fig

4. The stress parameters of BJT transistor was selected as
temperature and radiation.
3.2.1 Effect of Radiation
Another stress parameter which degrades the BJT devices
is -radiation. Degradation of many types of bipolar transistors
and circuits is known to depend strongly on dose rate. For a
given total dose, degradation is more severe in low dose rate
exposure than high dose rate exposure [12]. This effect has
been attributed to space charge effects from trapped holes and
hydrogen related species through oxygen vacancies in base
oxide. There are several hardness assurance tests and most
popular has been high dose rate irradiation at elevated
temperatures.

Figure 4: Test circuit for 2N2222 BJT transistor

4 RELIABILITY PREDICTIONS
By applying the above modified methodology for
reliability prediction, reliability indices of both the devices
were calculated using MILHDBK 217+ and by using
experimental analysis considering the accelerated testing since
there were no standard physics of failure models available for
developing failure model. The work also carried out the
simulation results to verify the behavior of both the stress
parameters on the subject of interested component.
4.1 Instrumentation Amplifier
Using RIAC 217 + [7], the failure rate equation for plastic
encapsulated integrated circuit was:
(8)
Where P = Predicted failure rate, G = Reliability growth
rate multiplier, OB = base failure rate, DCO = failure rate
multiplier of duty cycle, TO = failure rate multiplier for
temperature, EB = environmental base failure rate, DCN =
failure rate multiplier of duty cycle for non operating, RHT =
failure rate multiplier for temperature-humidity, TCB = base
failure rate for temperature cycling, DCO = failure rate
multiplier for cycling rate, DT = failure rate multiplier for
delta temperature, SIB = base failure rate for solder joint, , DT
= failure rate multiplier for solder joint delta temperature and
EOS = failure rate overstress.

There is no specific assigned model for the

instrumentation amplifier. Based on its internal diagram, there
are 2 portions for analysis to be needed: precision amplify and
overload-protection. Comparatively, overload protection was
smaller than the precision amplifier and hence precision
amplifier was only analyzed.
Selecting the required
parameters for the INA118 instrumentation amplifier, the
MTTF was calculated as 67x106 hrs.
The test circuit for instrumentation amplifier in Fig 3 was
subjected to both temperature and voltage simultaneously
using set of runs depicted by design of experiments. From the
experimentation, the stress levels of temperature 550C and
voltage 11V produces higher degradation than other set of
stress levels. These values were considered as input to the
accelerated testing for extended period of time. The output
voltage in terms of temperature, input stressed voltage and
accelerated time was generated by response surface regression
and calculated by using equation (9)
4.4588 0.00029
4.035
0.033
5.546 10
3.797 10
6.624 10
4.459 10
0.00012
(9)
By considering the 5% degradation as failure in the gain
and normal operating temperatures of temperature, the MTTF
was calculated as 52x107 hrs. There was reduction in MTTF
figure when compared to traditional prediction because this
experimentation considers another stress parameter as input
voltage which reduces the degradation of the output variable.
The CMMR of the instrumentation was also degraded when
applied to the stress parameters within 5-10% with respective
stress levels. The simulation studies verified the behavior of
both stress parameters on degradation of output variables.
4.2 BJT Transistor
Using RIAC 217+ [7], the failure rate equation for
transistors was:
(10)
Where, S = stress failure rate multiplier, TE = failure rate
multiplier, temperature environment and IND = induced failure
rate. The remaining factor representation was same as in Eqn
.By selecting and calculating the above equation, the MTTF
was calculated as 1.5x106 hrs.
The test circuit for BJT transistor was subjected to
temperature and then radiation as these parameters cannot be
applied simultaneously. The run for stress levels for maximum
degradation were temperature as 900C and radiation exposure
of 1M rad. The voltage was generated by using response
surface regression and shown below in equation
2.321 0.00027
3.035
0.0452
5.235 10
2.535 10
5.644 10
4.487 10
0.00026
(11)
Under the normal conditions with 5% degradation as
failure, MTTF was calculated as 1.2x107 hrs. The gain over
the traditional methodology was that the physics of failure
considers the actual stress parameter in this case, the radiation.

4.3 Life Costing Calculations

For the instrumentation amplifier, the initial costs of both
prediction methodologies were produced (in Indian equivalent
of dollars). The initial cost of traditional methodology was
(12)
|
Where Cit = initial cost for traditional methodology and
Chdbk|eqv = cost of handbook for equivalent component. The
actual cost of RIAC 217 plus handbook was 200$. The
assumption was that the figure of each calculation was 200.
Hence, Cit was assumed as 1$. The cost of physics of failure
methodology was calculated as

(13)
Where Cip = initial cost for physics of failure
methodology, Cer = expert reviews as consultation fee, Crs =
cost of research papers, Cfa = cost of failure analysis, Cexp =
cost of experimentation includes design, fabrication of PCB
boards, sources and measurement instruments and other
miscellaneous and Cmodel = cost of tools required for modeling.
By considering the figure of each calculation as 200, the total
cost consumed by physics of failure methodology was Cip =
50+120+50+250+30 = 500$ (approximately).
Similarly, the initial or prediction costs of BJT transistor
was calculated as Cit = 1$ and Cip = 500$.
For calculating the penalty costs, it was needed to
consider costs of the instrumentation amplifier as CIA = 15$
and cost of BJT transistor as CBT = 1$, time period and also
the number of replacements over time. Hence, penalty costs
were computed for both the components. For instrumentation
amplifier, the cost of penalty cost by using traditional method,

4.4 Other Factors

In the traditional reliability prediction, there was a certain
amount of risk associated with it and uncertainty over the
selection of several standard books available. There was no
possible inclusion of human factors, needed less amount of
time, space and man power. But in the case of physics of
failure reliability prediction methodology, it requires large
amount of time, space, man power and human factors
regarding the implementation of experimentation and
modeling. Also, it provides the approximate figure of the
failure time and had higher accuracy than previous prediction
models. The difference between both the prediction methods
was tabulated as in table [2].
Table 2: Comparison of Reliability Prediction methods
Factor
Initial Cost
Penalty Cost
Total Cost for Critical
Total Cost for non-critical
Time
Space
Expertise
Human factors
Risk
Uncertainty
Man Power

(14)

Where Cc = cost of the component, MTTFt = time to

failure using traditional methodology, Cf = 2*Cc = failure cost
as it was assumed as doubled for traditional methods, i =
interest rate, t0 = age of replacement =1(assumed) and t is the
design life. Similarly the cost of penalty using physics of
failure methodology was shown in equation.

replacements. But for the BJT transistor, the total cost

expended on traditional methodology (88) was less than the
proposed methodology (508) for n replacements.

(15)

Where MTTFp = time to failure by physics of failure.

The total cost of the reliability prediction was summation
of initial and penalty costs. For traditional prediction
methodology,
(16)
Similarly for physics of failure prediction, the total cost was
(17)
To compare the total cost incurred by both the prediction
methods for both the components, it was assumed that the
design life was 10 years and ratio of number of replacements
for a component was equal to the ratio of individual time to
failures.

By considering these assumptions, for an instrumentation

amplifier, the total cost expended on traditional methodology
(1052) was more than the proposed methodology (650) for n

Traditional
Low
High
High
Low
Less
Less
Low
Less
High
High
Few

PoF
High
Low
Low
High
Large
Large
High
More
Low
Low
Large

REFERENCES
1.
2.
3.
4.
5.
6.
7.
8.

European power supply manufacturers association,

MTBF Report, June 2005
Military Handbook, Reliability prediction of electronic
component, MIL-HDBK-217F
Jeff Jones, Joseph Hayes, A Comparison of ElectronicReliability Prediction Models, IEEE Transactions on
Reliability, vol. 48, no. 2,1999 june
B. Foucher, J. Boullie, B. Meslet, D. Das, A review of
reliability prediction methods for electronic devices,
Microelectronics Reliability 42 (2002) 11551162
Failure Mechanisms and Models for Semiconductor
Devices, JEDEC Publication, JEP122E, Originally
published as JEP122D.01 March 2009
Semiconductor Device Reliability Failure Models,
International SEMATECH, May 31, 2000
Handbook Of 217Plus Reliability Prediction Models,
Reliability Information Analysis Center (RIAC), 26 May
2006
Lloyd W. Condra, Reliability improvement with design
of experiments, 2nd edn, Marcel Dekker, 2001.

9.

Perry L Martin, Electronic Failure Analysis Handbook,

Techniques and Applications for Electronic and Electrical
Packages, Components, and Assemblies, McGraw Hill,
1999
10. Charles Kitchin, Lew Counts, A Designer's guide to
Instrumentation Amplifiers, Analog Devices, 2006
11. Guijie Wang, Gerard C.M. Meijer, "The temperature
characteristics of bipolar transistors fabricated in CMOS
technology", Elsevier, Sensors and Actuators 87 2000 81
89.
12. S.C. Witczak, R.D. Schrimpf, D.M. Fleetwood, K. F.
Galloway, R.C.Lacoe, D.C. Mayer, J.M. Puhl, R.L.Pease,
J.S. Suehle, "Hardness assurance testing of bipolar
junction transistors at elevated irradiation temperatures",
IEEE transaction on Nuclear Science, Vol 44, No. 6, Dec
1997
BIOGRAPHIES
Adithya Thaduri: He studied electronics and instrumentation
engineering in bachelors and reliability engineering in
masters. Presently, he is doing PhD at IIT Bombay India and
Lulea University of Technology, Sweden in association with
Bhabha Atomic Research center India. His core area of
research on predicting the life of critical electronic
components using physics of failure approach.
A.K. Verma received the B.Tech (Hons) and Ph.D. (Engg.)
degrees from Department of Electrical Engineering, IIT
Kharagpur. He has been with IIT Bombay as a faculty since

1988. He is a professor in the Department of Electrical

Engineering at IIT Bombay. He has over 180 research papers
to his credit and has supervised twenty eight Ph.D. thesis and
eight four Masters Thesis at IIT Bombay. He is the Editor in
Chief of the International Journal of Systems Assurance and
Engineering Management. He has been a guest editor of
special issues of various international journals He is a senior
member of IEEE and life fellow of IETE. His research
interests on reliability engineering include interdisciplinary
applications in software engineering, computing, maintenance,
and power systems.
Uday Kumar obtained his B. Tech from India during the year
1979. After working for 6 years in Indian mining industries,
he joined the postgraduate program of Lulea University of
Technology, Lulea, Sweden and obtained a PhD degree in
field of Reliability and Maintenance during 1990. He worked
as a Senior Lecturer and Associate Professor at Lulea
University 1990-1996. In 1997, he was appointed as a
Professor of Mechanical Engineering (Maintenance) at
University of Stavanger, Stavanger, Norway. Since July 2001,
he has taken up the position as a Professor of Operation and
Maintenance Engineering at Lulea University of Technology,
Lulea, Sweden. Currently he is the Director of Lulea Railway
Research Center. His research interests are equipment
maintenance, reliability and maintainability analysis, etc. He is
also member of the editorial boards and reviewer for many
international journals. He has published more than 150 papers
in International Journals and Conference Proceedings.