Failure Mechanism-Based Reliability Assessment Framework: A Case Study in An Automotive Component
Failure Mechanism-Based Reliability Assessment Framework: A Case Study in An Automotive Component
Failure Mechanism-Based Reliability Assessment Framework: A Case Study in An Automotive Component
1007/s12206-022-0903-2
Correspondence to:
Daeil Kwon Abstract This paper presents a framework for failure mechanism-based reliability as-
[email protected]
sessment, starting from collecting failed field samples. The framework includes failure analysis
for identifying the failure mechanism and accelerated life tests (ALTs) for reproducing the failure
Citation: mechanism. We consider that relevant information, such as warranty data and qualification test
Kang, M., Kwon, D. (2022). Failure
mechanism-based reliability assessment
results, enables to be utilized in the framework. The proposed framework is validated by using
framework: a case study in an automotive it to identify the failure mechanism and estimate the lifetime of an automotive component under
component. Journal of Mechanical Sci- normal use conditions. The case study shows that the framework enables to quantitatively as-
ence and Technology 36 (10) (2022)
4895~4903.
sess reliability based on the failure mechanism.
http://doi.org/10.1007/s12206-022-0903-2
4895
Journal of Mechanical Science and Technology 36 (10) 2022 DOI 10.1007/s12206-022-0903-2
4896
Journal of Mechanical Science and Technology 36 (10) 2022 DOI 10.1007/s12206-022-0903-2
Stress Effects
·Physical expansion
·Thermal aging, oxidation
High temperature ·Viscosity reduction
·Evaporation of lubricant
·Electrical parameter change
·Embrittlement of materials
Low temperature
·Physical contraction
·Loss of mechanical strength
·Structural collapse
Shock, vibration
·Cracking and rupture
·Looseness of interconnects
·Oxidation
High humidity ·Corrosion
·Electrical leakage
Fig. 2. Flow chart of accelerated life tests.
·Moisture generation
Temperature change
·Mechanical stress between coupled materials
·Scratch and abrasion on surface of material formed to obtain failure data at elevated stress levels based on
Sand, dust the knowledge of the identified failure mechanism. Therefore,
·Contamination of lubricant
the flow chart assumes that the failure mechanisms are identi-
fied prior to ALTs.
Table 2. Typical failure mechanisms [24].
are called overstress mechanism. Wearout mechanisms refer 2.2.2 Accelerated life tests planning
to failure due to the accumulation of damage via repeated During test planning, number of stress levels, levels of stress
stresses. Typical failure mechanisms are summarized in Table and number of test units at each stress level need to be con-
2 [24]. Engineers should deduce the failure mechanisms based sidered. These variables need to be determined according to
on the identified failure mode and cause to understand the various considerations, such as limited time, budget and capa-
process of failure under use condition. Furthermore, variables bility of the test equipment [26]. It is notice-able that all stress
of ALTs such as types of stress, level of stresses should be levels have to be selected within the range that induces the
determined based on the knowledge of the identified failure same failure mechanism observed under normal use condi-
mechanisms. tions. It is recommended to select stress levels as close as
possible to normal use condition considering available test time
to avoid excess extrapolating lifetime data at the elevated
2.2 Accelerated life tests (ALTs)
stress levels to use condition level. The larger the number of
A flow chart of reliability assessment based on ALTs is stress level and test units, the more accurate the results of
shown in Fig. 2. The objective of this process is to describe the lifetime estimation under normal use condition, while the test
relationship between stress and lifetime, and to analyze reliabil- takes longer [27]. ALTs are generally conducted at two stress
ity including lifetime of products under normal use conditions. levels or three stress levels. It is recommended to allocate a
To accomplish the objectives, ALTs are designed and per- higher number of test units to a lower stress level [28]. Two-
4897
Journal of Mechanical Science and Technology 36 (10) 2022 DOI 10.1007/s12206-022-0903-2
4898
Journal of Mechanical Science and Technology 36 (10) 2022 DOI 10.1007/s12206-022-0903-2
4899
Journal of Mechanical Science and Technology 36 (10) 2022 DOI 10.1007/s12206-022-0903-2
Parameter Value
Shape parameter β 7.7
Constant A 0.3846
Activation energy Ea 0.4807
4900
Journal of Mechanical Science and Technology 36 (10) 2022 DOI 10.1007/s12206-022-0903-2
Fig. 12. Weibull probability plot of field failure data in two countries.
Fig. 11. Weibull distribution of the lifetime (cycles) estimated at the use
condition level (e.g., 25 °C).
30 °C and 35 °C. The distributions of lifetime under various Fig. 13. Weibull probability plots of field data with a constant shape parame-
ter.
temperature conditions were estimated by the obtained rela-
tionship between temperature and lifetime. The shape parame-
ter remained constant ( β = 7.7) and the scale parameters at and 1.83 for country B data. Hypothesis testing was conducted
the specified temperatures were estimated using the parame- for verifying the assumption of a constant shape parameter of
ters in Table 5 and Eq. (2). For example, the Weibull distribu- two countries field data. P-value was 0.5482, so that the as-
tion of lifetime at 25 °C was estimated as shown in Fig. 11. sumption of a constant shape parameter was statistically vali-
Based on the estimated distributions, several life characteris- dated. In addition, Fig. 13 shows the data points are located
tics such as B10 lifetime were estimated at various tempera- within the 95 % confidence intervals except only a few points.
ture-based use conditions such as 25 °C, 30 °C and 35 °C as Consequently, AF for field data (AFfield) calculated as 1.2903.
presented in Table 6. The actual values were replaced with AF was also calculated from the life-stress relationship. AF
alphabets for security reasons of the data provider. The higher due to changes in temperature is defined as Eq. (3) derived
the temperature, the smaller the estimated lifetime (a > b > c). from the Arrhenius model equation Eq. (2).
Furthermore, the estimated lifetime had a lower confidence
limit (a*, b*, c*) and an upper confidence limit (a’, b’, c’). LA ⎛ E ⎛ 1 1 ⎞⎞
AF = = exp ⎜ a ⎜ − ⎟ ⎟
⎜ k T T ⎟
(3)
LB ⎝ ⎝ A B ⎠⎠
4901
Journal of Mechanical Science and Technology 36 (10) 2022 DOI 10.1007/s12206-022-0903-2
4902
Journal of Mechanical Science and Technology 36 (10) 2022 DOI 10.1007/s12206-022-0903-2
analysis, Reliability Engineering and System Safety, 53 (1) film on flexible substrate under high cyclic bending fatigue con-
(1996) 73-83. ditions, IEEE Transactions on Components, Packaging and
[21] J. B. Guinee et al., Life cycle assessment: past, present, and Manufacturing Technology, 2 (12) (2012) 2007-2016.
future, Environ. Sci. Technol., 45 (1) (2011) 90-96. [33] K. Alzoubi et al., Factor effect study for the high cyclic bend-
[22] S. Mathew et al., Virtual remaining life assessment of elec- ing fatigue of thin films on PET substrate for flexible displays
tronic hardware subjected to shock and random vibration life applications, Journal of Display Technology, 7 (6) (2011) 348-
cycle loads, Journal of the IEST, 50 (1) (2007) 86-97. 355.
[23] R. W. Johnson et al., The changing automotive environment: [34] Y. S. Lee et al., Tensile and fatigue behavior of polymer sup-
high-temperature electronics, IEEE Transactions on Electron- ported silver thin films at elevated temperatures, Materials Let-
ics Packaging Manufacturing, 27 (3) (2004) 164-176. ters, 193 (2017) 81-84.
[24] S. Menon et al., Optimization of PHM system for electronic
assemblies using maintenance aware design environment
software, Australian International Aerospace Congress (2011).
[25] E. Suhir, Accelerated life testing (ALT) in microelectronics and Moonsik Kang received the Bachelor’s
photonics: its role, attributes, challenges, pitfalls, and interac- degree in Mechanical Engineering from
tion with qualification tests, J. Electron. Packag., 124 (3) (2002) Sungkyunkwan University, South Korea.
281-291. He is currently studying a Master’s de-
[26] S. Limon, O. P. Yadav and H. Liao, A literature review on gree in Industrial Engineering from Sung-
planning and analysis of accelerated testing for reliability as- kyunkwan University, South Korea. His
sessment, Quality and Reliability Engineering International, 33 research interests are focused on prog-
(8) (2017) 2361-2383. nostics and health management, reliabil-
[27] M. Park and S. Rhee, A study on life evaluation & prediction ity engineering and design of accelerated life test based on
of railway vehicle contactor based on accelerated life test data, failure mechanism.
Journal of Mechanical Science and Technology, 32 (10) (2018)
4621-4628. Daeil Kwon received the Bachelor’s
[28] T. J. Kielpinski and W. Nelson, Optimum censored acceler- degree in Mechanical Engineering from
ated life tests for normal and lognormal life distributions, IEEE POSTECH, South Korea, and the Ph.D.
Transactions on Reliability, 24 (5) (1975) 310-320. degree in Mechanical Engineering from
[29] W. Q. Meeker and G. J. Hahn, How to Plan an Accelerated Life the University of Maryland, College Park,
Test: Some Practical Guidelines, ASQC Quality Press (1985). MD, USA. He was a Senior Reliability
[30] Y. W. Kown and B. J. Kim, Mechanical fatigue lifetime of Engineer with Intel Corporation, Chan-
metal electrode for flexible electronics under high temperature dler, AZ, USA, where he developed use
and high humidity condition, Journal of the Microelectronics condition-based reliability models and methodologies for as-
and Packaging Society, 27 (2) (2020) 45-51. sessing package and system reliability performance. He is
[31] K. Alzoubi et al., Bending fatigue study of sputtered ITO on currently an Associate Professor with Sungkyunkwan Univer-
flexible substrate, Journal of Display Technology, 7 (11) (2011) sity, Suwon, South Korea. His research interests are focused
593-600. on prognostics and health management of electronics, reliabil-
[32] M. M. Hamasha et al., Reliability of sputtered aluminum thin ity modeling, and use condition characterization.
4903