Downloaded from https://www.everyspec.com APACP 74*10 implies the value before a 2. Evaluate the relatio.ddpe between shift of the origin, see applied environments and stresses and rel- pf., 218 ability (evaluation tests) implies an estimated value 3. Verify that an item meats a prescribed of the parameter minimum reliability (tests of verification) 2-1 INTRODUCTION 4. Select the moe.reliable unit or approach from several alternatives (tests of The main advantage of statistics is that it comparison). can provide a good measure of the uncer- tainty involved In a numerical analysis.
The Reliability-measurement tests must be con- secondary advantage is that it does provide ducted under controlled conditions that methods for estimating effects that might approximate those to which the equipment otherwise be lost in the random variations will be subjected in the field (Ref. 2). Op. in the data. Wherever possible In the ex- crating times and number of failures are amples in this chapter, the uncertainty in the accumulated and used to estimate the under- numerical result& will be emphasized. lying failure distribution, the reliability, and level of a-confidence of the remults. Rarely is an engineer Interested only in the results of analyzing his model.
The Evaluation tests provide estimates of the engineer must solve a real-world, not a math- relationships between failures, and applied ematical, problem. The answer to the math- environments and stresses. Numerical rela- ematical problem must be tempered by all tionships between failure rate (and rell- the other important considerations that never ability) and specific stream can be derived. found their way into the model; this is why In addition, the relative effects of each en- the estImation of uncertainty is so important.
vironment in a multienvironment situation The engineer needs to know how much he can be estimated using techniques such as can afford to be swayed by those other con- Analysis of Variance and Multiple Regression. siderations. Tests of vedlfication are used to verify that A well designed and properly executed a desired result has been obtained (Ref. 2). reliability test program provides useful data A hypothesis such as "the reliability is equal for system designers and managers. The to or greater than 0.95 for 500 hr of opera.
statistical tests described in this chapter can tion" or "the failure rate is equal to or less be used to help ensure that the system de- than 0.02 per 1000 hr" is tected. The test sign metts reliability requirements. A de- hypothesis is then verified at some level of scription of the basic concepts of statistical s-significance by the test results. A wide testing during system design and develop- variety of tests can be designed--depending ment is presented in this chapter. on the number of units tested, the time allowed for testing, and the level of risks Reliability test and measurement are among taken in accepting the results.
the most important parts of a design and development program (Ref. 1), During de- Frequently, alternate design approaches sign and development, tests are performed are available to the system designer. Tests to; of comparison permit the designer to com- pare their reliability. The basic hypothesis of I. Measure the reliability of equipments this kind of test is that "no difference exists and subsystems (measurement tests) between the reliabilitics". This hypothesis is 14 J r