Downloaded from https://www.everyspec.com AIXP 706-17 any time maintenance of any kind is per- might take. See the chapteis that follow and formed, there is the real possibility and dan- the Bibliography at the end of thi3 chapter for ger that some part of the system has been some sources. aumapd unknowingly. There is a short period Roughly speaking, the lower the level at of "infant mortality" immediately after any- which redundancy is applied, the more effec- one fuses with any complicated system.
One tive it is (if switching is perfect and failures illusttation of this fact is that, at least during are s-independent) and the more it costs (in World War II, the repair crew chief for aircraft evring ) w.s supposed to go along on the checkout everything). flight after a repair. 7-4 METHOD OF SWITCHING The state of a complicated real system is not an easy thing to determine. Many analyses In virtually all systems, some kind of make the blithe assumption of perfection af- "switching" is necessary for redundancy to be ter repair, replacement, or checkout. Real effective. A fluid flow system might require a equipment is rarely like that.
check-valve on each redundant pump; an elec- tronic system might have to be disconnected. 7-3 SYSTEM LEVEL FOR REDUNDANCY The three main categories discussed here are APPLICATION automatic, manual, and repair. I, automatic switching, the operator need c:nia system, at what level ought redundan- not do anything in case of a unit failure. He cy to be applied? In principle (in the mathe- nodoayhgincsofautfilr.H may not even be aware that anything has gone matics anyway), one could make every piece- wrong. This is the easiest kind of redundancy part redundant, or one could just have several to analyze, although it is difficult to imple- systems. All of the factors listed in par.
7-1 ment in hardware. If periodic checkout is not apply to this decision. The question of switch- performed, the failed umt might not be dis. ing is especially important, simply because so covered until system failure. often it is assumed (in the mathematics) to be perfect: zero cost, instantaneous, no informa- Manual switching and repair/replacement tion lost, no size or weight, no design time, are different degrees of the same thing. An etc.
operator might have only to turn a switch or valve handle; or he may merely release some The lower the level at which redundancy catches or quick disconnects, pull out the is applied, the more likely are common-mode faulty unit, and shove in a good one. The time failures to be important. The question of con- it takes for removal/installation and the time ditional s-independence needs to be investi- for acquiring the spare are usually matters of gated very carefully. This question is allied degree, rather than of kind, in the analysis. In closely with the level at which repair parts a fixed ground installation, the whole thing ought to be stocked.
What about throw-away might be accomplished in a few minutes for a maintenance? At what level ought it be per- radio-receiver. The transmission in a tank formed? might take hours to remove/install and days In practice, an analysis barely can hope to to fix or acquire another. scratch the surface. Some rough guidelines The method that the designer finally can be developed, but pilot projects are the chooses depends on the system specifications places where knowledge is really gained. It is and constraints, on what he is familiar with, easy for the proposed system to be intractable and on what he thinks will really happen in for anything but a Monte Carlo simulation. the field.
A lot depends on the kind of logistic Therefore, the design engineer and his staff system in use for that equipment. analysts need to know what simulation a,:- guages are available on their computer. Often, a Monte Carlo simulation of the Many analyses are scattered in the litera- system is the only practical way to analyze ture. Rarely will the one be there that you what will happen. In such an analysis it often want. They can, however, give an idea about pays to be aware of some of the "paths" a what to analyze and what direction the results system takes during the failure/repair se- '. 7-2