Downloaded from https://www.everyspec.com PCHAPTER 5 SOME ADVANCED MATHEMATICAL TECHNIGUES 5-0 LIST OF SYMBOLS tions: Good, Degaded, Failed waiting for re- pair, In repair. Further suppose that the state n - number of states of the system is characterized adequately by s- = denotes statistical definition giving the states of each of the three subsys- S, - system-state i tems. Then there are 4 X 4 X 4 - 64 possible t - time states of the system. A state of this system u - time at which in-repair unit fails; re- consists of the specification of the states of generation point each of its three subsystems, e.g., Good, In A = transition rate from 8, to 8, repair, Good.
When the state of a subsystem changes, the state of the system will change. 5-1 INTRODUCTION 5-2.2 MARKOV CHAINS The approach to reliability wherein transi- tion distiLutions from one state to another Suppose the states of the system are speci- are all general is not tractable, because there fied, e.g-, Si, ..., S,, then there are r, states. It are no simple instants of time at which past is prerumed that the probAbility of going histury can Ie ignored. The best that can be from one state to another depends only on done in the general case is to give a compli- those twc Atates, and no others; past history is cated algorithm for calculating probability of wiped out.
For any two states, the transition transition at any tine. Therefore, everyone rate is a constant. The transition rate XU from uses simplifying assuiptions of some sort. A state S, to state S, corresponds to a failure few of the mathematical techniques tnat are rate for an exponential process in that it is a useful in the simplification process are men- ratio of a probability density function to a tioned here. Ncne were discovered or invent- Survivor functicn. Many of the XU for a sys- ed for reliability analysis; they are well-known tem are usup';y zero, because certain transi- (to mathematicians) wchniques Refs.
1 and 2 tions are not possible, by the very nature of give more details cn many of them. Hand- the particular system. In the example in par. books such as Ret:. 6 also show these and 5-2.1, just repaired subsystems might always other techniques; Ref. 7 is an example of a be Good, never Degraded. Then, a subsystem textbook which teaches some of these tech- could never go from "In-repair" to "Degrad- niques. ed", but it could go from "In-repair" to "Good" or from "Degraded" to "In-repair". 5-2 MARKOV PROCESSES The X,, car be put in a matrix form. Many special cases have been worked out There are se-jeral kinds and generalizations in the literature. Refs.
3-5 are likely sources of Markov processes, out only the most sim- of material. pie process will be discussed here. For more detil, se efs 1and2 nd heBibliography Considerable simplification of thc theory details, see Refs. 1 and 2 and the is possible when only the steady-state behav- ior of the system is of concern, not the tran- 5-2.1 SYSTEM STATE sient (start-up) behavior. 4 In practice, the number of system states The system is presumed to be in one of a must be severely limited i~1 order for the anal- set of states and can go from one state to ysis to be tractable. another. The state of a system is a description of its condition.
The analyst can choose the 5-3 LAPLACE TRANSFORMS way a state is characterized. Consider this example. Suppose a system consists of three The Laplace transform is perhaps the subsystems, each of which. can be adequatAly most popular transform for E-ngineers; they described by one of the following four condi- use it often in solving differential equations. 5-1