Lifetime: Evaluation vs. Measurement
Guest Post by Oleg Ivanov
How can we tell whether an iron is hot enough? The answer is obvious: We can measure temperature by using a thermocouple and a meter. But, in practice, we lick our finger and touch the iron. Sizzle…. Yes, it’s hot!
We know a priori the boiling temperature of water and we can evaluate the temperature of the iron. This method has a lower cost.
How can we know whether a product has a long enough lifetime? The answer is obvious too: We can test 100 products till their failure and find out that we have one mechanism of failure with a lognormal lifetime distribution, an expectation value of 7.6, and a standard deviation of 0.3. We can be sure that the reliability of this product is R = 0.99 during T = 1000 h of operation.
OK; the reliability of a product is high enough: It is the measurement that is expensive.
We can reduce testing cost by using the information we may already have available. Any information can be useful, including the results of testing of the materials and elements, as well as operation experience with similar products. If we know a priori that the product has one mechanism of failure with a lognormal lifetime distribution and a standard deviation of 0.3, we can test two products for three lifetimes (3000 h). It is an application of Six Sigma method for the life time (Fig.1). If these test yield satisfactory results, we can be sure that the reliability of this product is R = 0.99 during T = 1000 h of operation with CL = 0.99.
Fig.1. The application of Six Sigma method for the life time.
This is the evaluation. The results of evaluation and measurement are the same, but the cost of evaluation is less than the cost of measurement.
What is the payment for such a cost reduction?
In the case of the hot iron, we can burn a finger or our favorite shirt. In the case of the product lifetime, the product should have large durability reserves to pass this strong of a test (Fig.2). This is the drawback of evaluation.
Fig.2. The large risk of the supplier for the evaluation of the lifetime.
Note that evaluation is the foundation for certification of aircraft engines now. We can make an aircraft engine 10% lighter if we can measure (not evaluate) the Engine Critical Parts durability.
Michael Dashuta says
http://www.zvei.org/Publikationen/Robustness-Validation-Semiconductor-2015.pdf