[Please note that the following article — while it has been updated from our newsletter archives — may not reflect the latest software interface and plot graphics, but the original methodology and analysis steps remain applicable.]
Life data of a component, product or system can be separated into the following categories: complete data, right censored data, interval censored data, and left censored data. Each data type requires a modification to the analysis in order to correctly estimate the underlying lifetime distribution. A good life data analysis package, like ReliaSoft Weibull++, must be able to correctly handle all of these types of data. The characteristics of each data type are presented next.
Complete data, as shown in Figure 1, indicates that all of the units under the test failed and the time-to-failure for each unit is known. Therefore, complete information is known regarding the entire sample.
Right censored data, also called suspended data, is composed of units that did not fail during the test as shown in Figure 2. For example, suppose that five units are put under test. Three units fail and their observed times-to-failure (in hours) are 65, 76, and 84. The last two units are still operating when the test is stopped at 85 and 100 hours respectively. Therefore, the last two units are considered to be suspended or right censored.
Another type of censored data, shown in Figure 3, is called interval censored data or inspection data. Interval censored data contains uncertainty as to when the units actually failed. For example, if five units under test are inspected every 100 hours, then the status of each unit (failed or still running) is known only at the time of each inspection. If a unit fails, it is known only that it failed between inspections and the exact time of failure is not known. Instead of an exact time-to-failure, an interval of time (e.g., between 100 and 200 hours) would be recorded.
Left censored data is demonstrated in Figure 4. Left censored data is a special case of interval censored data in which the time-to-failure for a particular unit is known to occur between time zero and some inspection time. For example, if the inspection occurs at 100 hours, a failed unit could have failed at any time between 0 and 100 hours.
This will bring together HBM, Brüel & Kjær, nCode, ReliaSoft, and Discom brands, helping you innovate faster for a cleaner, healthier, and more productive world.
This will bring together HBM, Brüel & Kjær, nCode, ReliaSoft, and Discom brands, helping you innovate faster for a cleaner, healthier, and more productive world.
This will bring together HBM, Brüel & Kjær, nCode, ReliaSoft, and Discom brands, helping you innovate faster for a cleaner, healthier, and more productive world.
This will bring together HBM, Brüel & Kjær, nCode, ReliaSoft, and Discom brands, helping you innovate faster for a cleaner, healthier, and more productive world.
This will bring together HBM, Brüel & Kjær, nCode, ReliaSoft, and Discom brands, helping you innovate faster for a cleaner, healthier, and more productive world.
