The Reliability and Confidence Sample Size Calculator
Reliability Calculator disclaimer
The Calculator is for illustration and planning purposes for design verification testing. It is not intended to be a substitute for statistically derived samples sizes for process validation. Using the calculator is at your own risk and you are encouraged to pursue greater understanding of statistical methods, design verification and process validation from additional sources.
 |
Design Verification (Design Verification Planning and Reporting - DVP&R) requires a sufficient sample size to be able to make inferences about the performance of a design. Sample sizes of 1 are typically used because of the high cost of prototypes and long lead times for testing. It must be noted that a sample size of 1 equals a 50/50 chance that performance of the sample fits into a failure distribution that either is acceptable or not. There is no way to know with a sample size of 1 where the distribution is located or what type of distribution is indicated.
The purpose of the calculator is to provide a starting place to determine a sample size for testing based on one life. It works by selecting a Reliability Target value and picking a Confidence in that reliability that the design engineer would like to have.
For example: If the system to be tested is new or changed significantly, the design engineer may select 90% reliability as a target (lower if past history or Legacy indicates the design is currently OK with robustness considerations.) Next the Confidence that an Engineer requires is determined. Confidence is based on observation. The more I do a thing the more I am confident that I can do it again. Statistical confidence works the same way. The engineer may select 90% confidence.
Assuming no failures the sample size will be n=23. This number represents the greatest number a samples needed the prove a 90% confidence in the reliability of the product at 90%.
This is a starting place and not the end. Many ways exist to lower the sample size through intelligent Design Verification (DVP&R) planning including but not limited to:
- Past product data (product DNA)
- Computer Aided Engineering and FEA Finite Element Analysis
- Component testing
- FMEA - Failure Mode and Effects Analysis
- Testing to Failure and Weibul analysis
The 23 samples are based on a bogey being achieved without testing to failure. If tested to failure and 2 times expected life is achieved the sample size would drop to approximately 6 or ¼ of 23 samples.
Please feel free to use the calculator and contact us if you would like to improve the way you plan and execute testing and Design Verification and Validation.
Link to this page:
Need more information?
For more detailed information on this topic please email our qualified staff, contact us by phone at (248) 280-4800, or fill out a contact form.
 |
 |
 |
 |
