GROUP ASSIGNMENT #1

KRISSANA NUSORNRAM 8808005

PROGRAM: PROCESS QUALITY ENGINEERING

CONESTOGA COLLEGE

QUAL8100: RELIABILITY METHODS, SECTION 10

COURSE INSTRUCTOR: PROF. EBRAHIM AL-FAKIH

ASSIGNMENT DUE DATE: JUNE 05, 2023

This assignment aims to study and practice determining a Non-repairable product's Time-to-
Failure and warranty period. In the case study, we experimented with paperclips fatigue testing, the
professor provided the paperclips, and the process was to bend the paperclips 45° forward and
backward until the clip broke (Fail). The data is collected and recorded with the number of cycles of
failure. We spent 30 paper clips on this event and analyzed the result using Minitab software.

Regarding the observation data. Anderson-Darling distribution testing indicates that our
collected data best fits with Lognormal Distribution. So, we use the Lognormal Distribution parameter to
analyze our data in further steps.

From the data analysis, we can conclude that the paperclip should survive, on average, up to
18.5965 cycles. And if we would like to expect a 95% survival rate of paper clips, the warranty cycle
should be at most 4.3414 cycles. Or at the same time, we can explain that, at 4.3414 cycles, the failure
can be expected up to 5%.

Please follow the sections below for the collected data and analysis details.

2. Raw data

Below is the raw data for the cycles to failure of paperclips.

There are 30 observations of data from 30 randomly selected paper clips.

Group 3
1 Krissana Nusornram
2 Mohamed Ashfaq Ahmed
3 Pavan Kumar
Data - Group 3
No No No
. Cycles to failure . Cycles to failure . Cycles to failure
1 45 11 4 21 19
2 8 12 16 22 6
3 21 13 13 23 5
4 19 14 50 24 14
5 16 15 30 25 9
6 19 16 14 26 18
7 8 17 31 27 25
8 3 18 10 28 12
9 11 19 7 29 7
10 22 20 65 30 27
3. Data Distribution Analysis

The data analysis is performed using Minitab software, consisting of 3 steps.

3.1 The Distribution ID plot: To determine which distribution best fits our collected data.

3.2 Distribution Overview Plot: To describe the best-fit data distribution in graphical
information.

3.3 Parametric Distribution Analysis: To fit the data and evaluate the reliability of the
product by estimating the parameters for the distribution.

3.1 Distribution ID plot

Figure 1 &2 depicts the test result of the different types of data distributions. In Figure 1, the
Anderson-Darling value for the Lognormal distribution is 0.545, which is the least value when compared
to the other distributions; furthermore, when looking at the probability plot in Figure 2, the distribution
of data is the best fit to the reference line in the Lognormal distribution plot. So, we concluded that our
data follow the Lognormal distribution.

Figure1: Goodness-of-Fit

Figure2: Probability plot for cycle-to-failure

Regarding step 3.1, we know that our data distribution follows the Lognormal distribution. Now
we will analyze the data in the Lognormal distribution parameter.

Figure 3: Goodness of fit table

Figure 4: Distribution overview plot

In Figure 3, The Goodness-of-fit table displays the data plotted using a Lognormal distribution
parameter.

In Figure 4, Distribution Overview Plot, Minitab also displays the Probability Density, Survival,
and Hazard function plot under the lognormal parameters, i.e., Loc is 2.6603, and the scale is 0.724765.
The Mean Time to Fail for this experiment is 18.5965 cycles, which means that the paper clips can be
expected to survive until the bending cycle reaches 18.5965 cycles on average.
3.3 Parametric Distribution Analysis:

3.3.1 Survival function plot.

Figure 5: Survival plot for cycles to failure.

Figure 6: Table of percentiles

Figures 5 and 6 show the probability of the paper clips’ 95% survival rate at 4.3414 cycles. The
survival function is based on the lognormal distribution with Loc = 2.66033 and scale = 0.724765.
3.3.2 Cumulative distribution failure plot:

Figure 7: Cumulative failure plot for cycles

From Figure 7, The probability of failure rate of paper clips is 5% at 4.3414 cycles. The cumulative
failure function is based on the lognormal distribution with Loc = 2.66033 and scale = 0.724765

3.3.3 Hazard plot:

Figure 8: Hazard plot for cycles to failure.

From Figure 8, the Hazard function is based on the lognormal distribution with Loc = 2.66033 and scale =
0.724765. Here, the hazard rate exponentially increases between 4 to 16 cycles and gradually decreases
over cycles.
3.3.4 MTTF:

Figure 9: Table of MTTF Figure10: Characteristics of distribution

Figures 9 and 10 show that the average cycle in the lognormal distribution is about 18.5965
cycles. Meaning that the paper clip is expected to survive up to 18.5965 cycles.

3.3.5 If the survival rate (R(t)) is 95%, how many cycles should the warranty cover:

If the survival rate is 95%, the number of cycles the warranty should cover is 4.3414 cycles which is equal
to 4 cycles in real-time. Only then the hardly 5% can be claimed for a warranty.

4. Conclusion

In conclusion, the data from the paperclip fatigue testing experiment follows a Lognormal
distribution according to the Anderson-Darling distribution testing. The survival function analysis
indicates a 95% probability of paper clips surviving approximately 4.3414 cycles. The hazard function
shows that the hazard rate increases exponentially between 4 and 16 cycles and gradually decreases. On
average, paper clips survive around 18.5965 cycles.

Based on these findings, it is recommended to provide a warranty that covers four cycles to
account for the 5% failure rate (95% survival). It means that paperclip failures within the first four cycles
would be eligible for warranty claims. This analysis helps evaluate the time-to-failure and reliability of
the paper clips, allowing for informed decisions regarding warranty coverage and customer satisfaction.
Reference

YouTube. (2020, May 31). Minitab : Reliability Analysis of Failure Times. YouTube.
https://www.youtube.com/watch?v=FTqrnNAIEME

YouTube. (2019, June 27). Life Data Analysis of complete data using Minitab software.
YouTube. https://www.youtube.com/watch?v=JuQ_AEtj1vE