Background
The reliability requirement for a new roller assembly is reliability = 95% for 50,000 cycles at a 70% lower one-sided confidence level. The objective is to demonstrate this reliability goal through testing. The test lab at the manufacturing facility can test only five units at a time, so a test is set up for a sample size of 20 components (four groups of five units each), to be run sequentially until the first failure in each group. In other words, the first group of five units is tested until one of the five units fails. After that, the second group is tested until the first failure and so forth. 

Experiment and Data
The following data are recorded during the experiment:
 

Analysis Discussion
Even though the test is performed sequentially, the analysis can include all data as follows:
 

Analysis
Step 1: Using Weibull++ 7, the first step is to create a new Data Entry Spreadsheet to hold grouped times-to-failure with suspensions data.

Step 2: The next step is to enter the life data from the test, as shown in the second table above. The data set is then analyzed using the 2-parameter Weibull distribution and MLE. The results are displayed next.

Step 3: The remaining task is to determine the demonstrated reliability at 50,000 cycles and at a 70% confidence level. The QCP in Weibull++ is utilized to compute this value as shown next. In this case, the demonstrated reliability at the desired confidence level is found to be 96.41, which is greater than the requirement, thus demonstrating the requirement.

Alternatively, a probability plot can be obtained, which shows the same result. This is shown next. Note that the X-axis scale has been changed.

Supplemental Discussion
It should be noted that since the test is run sequentially, the data can be analyzed once the second step is reached. When the data set is analyzed after the second step, it can be seen that the requirement is met by the second step, thus requiring no additional testing.