Reliability testing and growth
Reliability testing is one of the most important reliability activities of a reliability program. Its main objective is to obtain information concerning failures, in particular, the equipment/product tendency to fail as well as failure consequences. This type of information is very useful to control failure tendencies along with their consequences. A good reliability test program may simply be described as the one that provides maximum information concerning failures from minimal amount of testing. Over the years, many publications on reliability testing have appeared; in particular, two such publications are listed in Refs. [1, 2].
In the design and development of new systems, the first prototypes generally contain various design and engineering-related deficiencies. Elimination of such deficiencies leads to a system's/product's reliability growth . Although the serious thinking concerning reliability growth may be traced back to the late 1950s, the frequently used reliability growth monitoring model was postulated by Duane in 1964 . A comprehensive list of publications up to 1980 on reliability growth is available in Ref. .
This chapter presents various important aspects of reliability testing and growth.
Reliability test classifications
Reliability tests may be grouped under the following three classifications [6, 7]:
• Reliability development and demonstration testing. This is concerned with meeting objectives such as to indicate whether any design changes are needed, to determine whether the design is to be improved to meet the stated reliability requirement, as well as to verify improvements in design reliability. This type of testing depends on factors such as the type of system/subsystem being investigated and the level of complexity under consideration. For example, in the case of electronic parts, the reliability development and demonstration testing could take the form of life tests to evaluate whether the part can meet its reliability-related goals; if not, what actions are needed?
In order to meet objectives of reliability development and demonstration testing effectively, the accumulated test data must be a kind that clearly allow insight into the failure effects/probabilities for a certain design under consideration. Furthermore, these data values serve as a good basis for reliability assessment and analysis for two specific items: design under consideration and subsequent related programs.
- • Qualification and acceptance testing. This is concerned with meeting two fundamental objectives: to determine whether a certain design is qualified for its projected application and to arrive at a decision whether a part/assembly/end item is to be accepted or not. These objectives differ from the objectives of other reliability tests, particularly with respect to reject/accept mechanism. Furthermore, qualification and acceptance testing incorporates the usual testing and screening the quality-control function of incoming parts. In regard to the materials and components to be used in the equip-ment/system under development, the qualification and acceptance testing begins early in the program.
- • Operational testing. This is concerned with objectives that include verifying the results of reliability analysis performed during the equipment/system design and development, providing data for subsequent activities and providing data indicating desirable changes to operating policies and procedures in regard to reliability/main-tainability. Finally, it is added that the operational testing provides the feedback from practice to theory.
This type of testing is sometimes used in receiving inspection and in engineering test laboratories where a no-failure test is stated. Normally, the main goal for this test is to ensure that a certain reliability level has been achieved at a stated confidence level.
In this case, for zero failures, the lower 100(1-0)% confidence limit on the desired reliability level is expressed as 
Rtl = 01/n (7.1)
0 is the level of significance or consumer's risk, n is the number of items placed on test.
Thus, with 100(1-0)% confidence, it may be stated thatwhere
R, is the true reliability.
By taking the natural logarithms of the both sides of Equation (7.1), we obtain
In Rtl = — In 0 n
Thus, from Equation (7.3), we get
The desired confidence level, CL,(, is expressed by
CLrf = l-e
By rearranging Equation (7.5), we obtain
- 0 = l-CLrf
Using Equations (7.2) and (7.6) in Equation (7.4), we obtain
Equation (7.7) can be used for determining the number of items to be tested for stated reliability and confidence level.
Assume that 90% reliability of an electronic item is to be demonstrated at 95% confidence level. Calculate the number of electronic items to be placed on test when no failures are allowed.
By substituting the specified data values into Equation (7.7), we get
Thus, 28 electronic items must be placed on test.