My Problem

 Data 

Having been party to many discussions in particular about equipment reliability and availability and their contribution to the planning for and conduct of required tasks/operations, it has long been clear to me that the understanding of the terms Reliability and Availability vary dramatically among the people employed within organizations. This is quite understandable because these terms have multiple definitions that tend to be applied within discrete areas of authority, responsibility, and specialization. Therefore we should not be surprised to find that there is misconception and misunderstanding when these terms are employed routinely without definition agreement and/or qualification where necessary. I have seen these terms used in a variety of contexts and sometimes interchangeably.

The assumption seems to be that we are "comparing apples with apples". If we are then that is fine. But it is not enough because when making comparisons we need to ensure that not only are we comparing apples, but that the variety of the apples under consideration is known to enable us to draw reasonable conclusions; whether comparing Cox's with Cox's or two different varieties to understand the differences.

The matter is made more difficult because there is a felt need to express equipment reliability and availability in a precisely quantitative "measured" manner, usually Mean Time Between Failures for Reliability, or, Availability as a percentage of the fleet, that is, has been, or will be available. A single figure can speak volumes to some people without any absolute understanding of how it came to be. It is authoritative because it is deemed to have numerical certainty and a foundation in reality because it purports to reflect that which has been reported, analysed and scrutinized. Therefore it is the mathematical truth.

However, measurements are well and good. But it is unreasonable to assume that just because you start out with a measurement of something it means that you have control of it and it will never vary. If we deem a process or activity to be of sufficient importance that we would wish to control it, then it follows that we must conduct analysis to understand in absolute terms the significance and utility of the measurement(s) in relation to managing it.

In an ideal world systems would never fail. However, we live in an imperfect world where chemistry, the laws of physics and even nature are conspiring against the ideal. A simple example of this is bird strike causing an aircraft engine to fail. Whilst this is a relatively rare occurrence, we accept that equipment does fail for a variety of reasons. It is widely acknowledged that fatigue is a primary cause of equipment failure in mechanical equipment. More recently the Computer Aided Life Cycle Engineering (CALCE) Electronic Products & Systems Center of Maryland University ⁽¹²⁾ physics of failure analysis has shown that this is also the case in electronic products. Whatever the cause, in the real world we accept that equipment isn't absolutely reliable, that failures do occur and consequently that maintenance is necessary, even for relatively simple equipments, therefore it cannot be available at all times.

Stevens et al ⁽¹³⁾ talk about "coping with risk and change". They say that risk and human fallibility are constants in system development and operations. "The perfect design is beyond our reach, reason has its limitations, and life is not predictable". On the face of it, this seems to acknowledge not only the physical constraints of the natural world but also Simon's notion of bounded rationality (1957). Additionally they say "Although systems engineering is about prediction and direction, the future is only partially discernible." This seems to acknowledge the uncertainty and in so far as the user in a non-profit organization is concerned this will likely manifest itself as an inability to accomplish its mission in the manner and/or with the resources it should reasonably expect to employ; in essence, reduced availability. Interestingly, in their book, Stevens et al ⁽¹³⁾ make no specific reference in the index to operational availability or availability of any sort and therefore not surprisingly neither does it feature within the glossary of terms.

 Reality