US Army Materiel Command AMCP 706-197 — AMCP 706-197 Reliability Prediction Handbook — US Army Engineering Design Guide PDF

Downloaded from https://www.everyspec.com -M 70&-197 The Laplace transform i very closely related rep.,nera n (renewal) point. Statistically to the Lapiace-Stieltes transform and to the -t. the system (when in a particular Fourier transform. The Moment Generating a no memory as to .ow long it has function and the Characteristic function ar bee.. n that state; each instant is just like also related to the Laplace transform, al- every other instant. though statistics texts seem rarely to point If general statistical distributions are used, this out. (The Characwistic function is, for- this i no longer simply the caw.

The trick in maly, the Fourier transform; and the Mo- ian analysis is to imd (or invent) some time ment Generating function is, formally, the instants which have this regeneration ip- Laplace transform.) The Stities form of the ertyn ch have this regerat phop Laplace transform has fewer difficulties with erty; once you know that the system at this than does the Laplace transform, time instant, its past history can be forgotten. "existence" OnetwaydoefindingLaitabce rgeneratio although in practical reliability work, "exist- One way of finding suitable regeneraion ence" of integrals and pdf'rs is rarely a diffi- points is to introduce an extra time variable culty.

In the remaining discussion, the phrase, to help describe the tate of the system. Laplace transform, includes all the related For example, suppoie a system of two transforms and functions, units is in one of the follo-ving three states" The Laplace transform cLhanges differenti- 1. One unit operating, other in-standby ation and integration into multiplication and 2. One unit operating, other in-repair division by the transform-variable. In reli- 3. One unit in-repair, other waiting-for- ability analysis, another of its properties is repair. even more important.

The Laplace transform of the sum of several s-independent random The unit is in state two at time = t; intro- variables is the product of the individual duce the time = u at which the in-repair unit Laplace transforms of the random variables, fails; at time = 0 the operating unit was put Thus convolution is transformed to multipli- into operation. With u as an extra variable, cation. time = u is a regeneration point; the state When the equations of the system are ex- probabilities do not depend upon the history pressed in Laplace transforms, the steady of the system prior to u.

state (t -* as) behavior can be found easily Of course, the introduction of extra vari- without inverting the transforms. ables complicates the analysis, but, at least, The Laplace transform of the answer in a some equations can be written down. This reliability problem often can be obtained in a supplementary variable technique is used in closed form, albeit usually unwieldy. The dif- the literature, e.g., Ref. 5, in order to "solve" closedyformeslbeitausuall unvsiiely ea- reliability problems where random vaiiables ficulty arises because inversion is rarely fea- have unspecified distributions.

Virtually all sible in closed form; then numerical inversion polm hnsae hswywl nov must be used.problems when stated this way will involve the sums of s-independent random variables; REGENERATION POINTS so Laplace transforms will ordinarily be used in the solution of the proulem (see par. 5-3). The big advantage of assumir.g constant Ref. 7 discusses renewal theory in detail. transition rates, is that every time-instant is a