Full Support for Traditional Reliability Growth Analysis
Since there are many different ways that reliability growth data might be collected, the RGA software provides a comprehensive array of options for data entry and analysis using traditional reliability growth methods.
Discrete Data (Also Called Attribute, One-Shot or Success/Failure Data)
When you have data from one-shot (pass/fail) reliability growth tests, you can use the Standard Gompertz, Modified Gompertz, Lloyd-Lipow, Logistic, Crow-AMSAA (NHPP) or Duane models for data analysis. The available model(s) will depend on the type of data sheet. Data types include:
- Sequential – for data sets where a single unit is tested for each successive configuration.
- Sequential with Mode – for sequential data sets where the mode is identified for each failure.
- Grouped per Configuration – for data sets where multiple units are tested for each successive configuration.
- New in Version 7!Mixed – for data sets where a single unit is tested for some configurations and multiple units are tested for other configurations.
Reliability Data
When you have the calculated reliability values for different times/stages within developmental testing, you can use the Standard Gompertz, Modified Gompertz, Lloyd-Lipow or Logistic models for data analysis.
Times-to-Failure Data
When you have data from developmental testing in which the units were operated continuously until failure, you can use the Crow-AMSAA (NHPP), Duane or Crow Extended models for data analysis. Data types include:
- Failure Times – for data sets with exact times-to-failure.
- Grouped Failure Times – for data sets where the exact failure times are not known but the number of failures within each given interval has been recorded.
- Multiple Systems – Known Operating Times – for data sets where multiple units were tested and you recorded the operating times of the other non-failed units at the time of each failure.
- Multiple Systems – Concurrent Operating Times – for data sets where multiple units were tested but you did not record the operating times of the other non-failed units. The analysis is based on an "equivalent" system that combines the operating hours of all systems.
- Multiple Systems with Dates – for data sets where the calendar dates of the failures can be used to create the "equivalent" system that will be analyzed.
Parameter Estimation, Confidence Bounds and Statistical Tests
Depending on the data type and model, RGA may use Maximum Likelihood or Least Squares for parameter estimation. The Chi-Squared or Cramér-von Mises (CVM) statistical tests are used to check the model’s goodness-of-fit. Confidence bounds may be calculated with the Fisher Matrix, Crow or Least Squares methods.
For multiple system data, the Common Beta Hypothesis (CBH) test checks whether all of the systems exhibit similar behavior and the Laplace Trend test is used to determine whether there is a trend in the combined data.
Calculated Results for Traditional Reliability Growth Analysis
For developmental test data analyzed with the Standard Gompertz, Modified Gompertz, Lloyd-Lipow or Logistic models, you will be able to:
- Calculate the reliability at a specified time/stage.
- Determine the amount of testing that will be required to demonstrate a specified reliability.
For developmental test data analyzed with the Crow-AMSAA (NHPP) or Duane models, you will be able to:
- Calculate the MTBF or failure intensity for a given time/stage.
- Determine the amount of testing that will be required to demonstrate a specified MTBF or failure intensity.
- Estimate the expected number of failures for a given time/stage.
