Abstract
When nearly identically processed materials/devices are placed under the same set of stress conditions, they will not fail exactly at the same time. An explanation for this occurrence is that slight differences can exist in the materials’ microstructure, even for materials/devices processed nearly identically. This means that not only are we interested in time-to-failure but, more precisely, we are interested in the distribution of times-to-failure. Once the distribution of times-to-failure is established, then one can construct a probability density function f(t) which will permit one to calculate the probability of observing a failure in any arbitrary time interval between t and t + dt, as illustrated in Fig. 7.1.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Notes
- 1.
Note that the lognormal distribution has the same general form as does the normal distribution in Chap. 5. The major differences are: (1) the natural logarithm of time ln(t) is used rather simply the time t; and (2) σ now represents the logarithmic standard deviation σ = ln(t50/t16). Also, the (1/t) in the prefactor of the lognormal distribution is needed to ensure that f(t)dt will continue to represent the probability of failure. This is due to the fact that dln(t) = (1/t)dt.
- 2.
Note that any cumulative fraction F, and its corresponding failure time, may be used in Eq. (7.8) to determine the Weibull slope. The author’s preference is to use F = 0.1 and t10. However, this is only a preference, not a requirement.
- 3.
A lognormal distribution was used here but a Weibull distribution could have been used and would show similar results.
- 4.
A lognormal distribution could also have been used and would produce similar results.
Bibliography
Dhillon, B. and C. Singh: Engineering Reliability, John Wiley & Sons, (1981).
McPherson, J.: Reliability Physics. In: Handbook of Semiconductor Manufacturing Technology, Marcel Dekker, 959 (2000).
Miller, I. and J. Freund: Probability and Statistics for Engineers 2nd Ed., Prentice Hall, (1977). Nelson, W.: Accelerated Testing, John Wiley and Sons, (1990).
Author information
Authors and Affiliations
Problems
Problems
-
1.
Time-to-rupture data (from a creep study) are shown for steel rods that were held at a fixed level of stress at very high temperatures until rupture occurred. The time-to-failure data are shown in hours. Find the lognormal t50 and σ that describes the data.
44.0 | 43.3 | 49.0 | 36.0 | 70.8 |
---|---|---|---|---|
50.3 | 45.2 | 47.4 | 36.3 | 68.3 |
65.6 | 51.4 | 43.2 | 39.4 | 70.7 |
54.5 | 58.3 | 42.5 | 52.2 | 56.5 |
41.1 | 60.2 | 42.7 | 63.6 | 40.0 |
Answers: t50 = 49.8 h, σ = ln(t50/t16) = 0.22
-
2.
Given the lognormal distribution (t50 = 49.8 h, σ = 0.22) from Problem 1,
-
(a)
What is the expected time for 0.1 % of the steel rods to rupture?
-
(b)
What is the expected time for 99.9 % of the steel rods to rupture?
Answers: (a) t0.1 % = 25.2 h, (b) t99.9 % = 98.3 h
-
3.
Given the lognormal distribution (t50 = 49.8 h, σ = 0.22) from Problem 1, what fraction of failures occur between 35 and 55 h?
Answer: 0.620
-
4.
Using the time-to-rupture data in problem 1, find the Weibull distribution that gives the best fitting to the data. What are the values for t63 and the Weibull slope β?
Answer: t63 = 55.2 h, β = 5.4
-
5.
Given the Weibull distribution (t63 = 55.2 h, β = 5.4) from Problem 4,
-
(a)
What is the expected time for 0.1 % of the steel rods to rupture?
-
(b)
What is the expected time for 99.9 % of the steel rods to rupture?
Answers: (a) t0.1 % = 15.4 h, (b) t99.9 % = 79.0 h
-
6.
Given the Weibull distribution (t63 = 55.2 h, β = 5.4) from Problem 4, what fraction of the failures occurred between 35 and 55 h?
Answer: 0.543
-
7.
Using the normal distribution in Chap. 5, fit the data shown in Table 7.1.
-
(a)
What are the values of t50 and sigma for the normal distribution?
-
(b)
Compare your normal fit to the lognormal-fit shown in Fig. 7.3. Which distribution gives the better fitting, normal or lognormal?
Answers:
-
(a)
t50 = 1,573 h, σ = 576 h,
-
(b)
Lognormal distribution gives a better fitting to this data set.
-
8.
The following time-to failure data was collected and found to have two failure mechanisms in the time-to-failure data.
Cum fraction F | Time-to-failure (h) |
---|---|
0.05 | 16 |
0.09 | 20 |
0.15 | 25 |
0.22 | 30 |
0.3 | 35 |
0.38 | 42 |
0.45 | 104 |
0.51 | 110 |
0.63 | 118 |
0.68 | 127 |
0.75 | 135 |
0.82 | 143 |
0.87 | 150 |
0.9 | 155 |
-
(a)
Perform a lognormal plot of the above data.
-
(b)
Find the point of inflection which separates the two mechanisms.
-
(c)
Replot the data for the two mechanisms.
-
(d)
What are the (t50, σ) values for the two mechanisms?
Answers:
(b) Point of inflection: F = 0.42,
(d) Mechanism A: t50 = 27.3 h, σ = 0.4,
Mechanism B: t50 = 131.1 h, σ = 0.17.
9. Using the time-to failure data (shown in the table in Problem 8):
-
(a)
Perform a Weibull plot of the above data.
-
(b)
Find the point of inflection which separates the two mechanisms.
-
(c)
Replot the data for the two mechanisms.
-
(d)
What are the (t63, β) values for the two mechanisms?
Answers:
(b) Point of inflection: F = 0.42,
(d) Mechanism A: t63 = 32.3 h, β = 2.99 Mechanism B: t63 = 139.6 h, β = 6.29.
Rights and permissions
Copyright information
© 2019 Springer Nature Switzerland AG
About this chapter
Cite this chapter
McPherson, J.W. (2019). Time-to-Failure Statistics. In: Reliability Physics and Engineering. Springer, Cham. https://doi.org/10.1007/978-3-319-93683-3_7
Download citation
DOI: https://doi.org/10.1007/978-3-319-93683-3_7
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-319-93682-6
Online ISBN: 978-3-319-93683-3
eBook Packages: EngineeringEngineering (R0)