Everyspec (implied, based on the URL) AMC PAMPHLET 31a, AMCP 706.198 — Everyspec AMC PAMPHLET 31a: US Army Reliability Development Guide for Military Equipment (PDF)

Downloaded from https://www.everyspec.com AMCP Me-IN ment pts out In the field; don't lose sight a. For mean-life requirements, of that goal-even at the expense of annoy- Ing a statistician. v a 9 o/OA (3-3) -In acceptance sampling, the probability of b. For failure-rate requirements, a Type I error a is commonly called the producer risk, since it represents the risk that 'Y 0 *i/xo (34) a product conforming to the specification will be rejectr -; The probability of a Type c. For survival-probability requirements, II error # Is r-led the consumer risk, since it represents the risk of accepting a product v " (3-5) that ought to be rejected [(I - i) is known as the power of the test).

where Reliability can be measured by various 9 a mean life (exponential parameter) parameters, such as probability of survival, mean life, and failure rate. X a 1/0 = constant failure rate The following nomenclature is employed: - s-unreliability 1. Reliability Level (RL) is the level of the "0" subscript signifies the ARL the reliability-measure which the lot or sys- tem actually has. the "1" subscript signifies the URL 2. Acceptable Relability Leol (ARL) is The ARL has a high probability of accep. "the RL considered to be acceptable, and tance and the URL a low probability of represents the null hypothesis, acceptance.

It Is Important that specified values of the ARL and URL be consistent A -: AL - ARL . (3-1) with operational requirements. Generally, the URL Is that value which is minimum for S(The Acceptable Quality Level, AQL, Is the satisfying operational requirements, while analogous term for aereptance tests based on the ARL Is a highly acceptable value; the fraction defective.) latest revision of Ref. 25 ought to be con- sulted for contractual situations. 3. Unacceptable Reliability Levl ( URL) Is the RL considered to be unacceptable, This method of describing a sampling plan and represents the alternate hypothesis, can be very misleading to a nonstatistlcian.

It Is probably beat to consider this procedure Hi: RL - URL (3-2) as a means of picking two points on the Operating Characteristic (see par. 3-2.4) of (The Lot-Tolerance-Percent-Defective (LTPD) the sampling plan, In order to generate the Is the analogous term for acceptance tests complete Operating Characteristic. Once the based on fraction defective.) entire Operating Characteristic Is generated, one can forget the names of the two points 4. Discriomatlon Ratio y, Is a ratio in- used to generate it, volving ARL and URL, and alwaya is defined , so that it is greater than one. Many pain of points will lead to the sane 11 ...... A V... ... I" 'In'II