Reliability

Reliability and quality

The French electrotechnical Commission, on recommendation of the international electrotechnical committee, proposed the following definition: reliability is the aptitude of a device to achieve a function necessary under conditions given for a given period of time (quoted by Pierre Chapouille, reliability, Which I know?).

It is measured by a probability (from 0 to 1, or 0 to 100%).

One should not confuse reliability (function of time) and it quality control (static function)

For example, one tests Integrated circuits into leaving the Line production, and one notes that 3% of them do not function, or incorrectly: one can say that the " qualité" of this chain is 97% (3% of defects).

Once these circuits inserted in a system, it is noted that their average time of correct operation before breakdown (MTTF, for Mean-time To Failure) is 100.000 hours. Their rate of Panne (many breakdowns by Unit of time) will be thus 1/MTTF. This one notes λ.

If one notes moreover than these breakdowns are not prédictibles and occur in a completely random way, then the reliability of these circuits according to time will be given by the function R (T) = exp (- t/MTTF) or exp (- λt)

It is noted that, whatever the MTTF:
- for t=0, reliability is worth always 1
- for T tightening towards the Infinite , reliability tends towards 0.

Note: the fall of the value of reliability with time should not be confused with a phenomenon of wear.

Whatever the duration of good performance already achieved, at any moment the probability of breakdown of a circuit between the moment T and the moment (t+dt) remains constant, and equal to dt/MTTF (essential property of the exponential Distribution).

Reliability and probability

The Prédiction S of reliability are necessarily probabilistic , because they require the knowledge of the failure rate of each component.
These failure rates being obtained on samples inevitably limited in the face, their value are controlled by the laws of the statistics (confidence intervals in particular).
The mathematical theory of reliability will thus consist of a particular application of the theory of probability to the problems of operation life without incidents.
The most current approximation, especially of electronics, consists in supposing the exponential distribution of the breakdowns of the components, which involves the law of addition of the failure rates for a subset non-redondant.
The reliability and the availability of the redundant groupings of not-redundant subsets then being able to be calculated using the Process of Markov.
A new method of forecast of the reliability of the electronic systems named FIDES is the concrete example.
Foot-note: in practice the Distribution of the failure rates moves away sometimes from the exponential one: it is the case, for certain equipment, at the beginning of life (grinding) and in the end-of-life (wear).

Safety, quality, durability, tolerance with the faults

The questions of Sécurité are relating to the prevention of the serious accidents: cost in human lives, physical injuries, important material damage.

The studies of reliability are not limited to the questions of security but include/understand also the studies of quality: many products can achieve the same function but some do it better than others, they get more satisfaction with their users, they are of better quality. To envisage the satisfaction degree gotten by a product belongs to the studies of reliability. Durability is at the same time a question of security and of quality. It is necessary to guarantee the durable safety of way, but one cannot always await from a product which it functions eternally, and one is all the more satisfied as he lasts longer.

Often one cannot force the device to always function without failures but it is wanted only that the probable dysfunctions cause only moderate damage. This tolerance with the faults (operation in degraded mode) is one of the aspects of reliability.

Reliability and decompartmentalization of information

In many cases of serious accidents, certain people knew that there was a problem. Either they were not listened, or they did not even seek to be made listen because they knew that they would not be taken with the serious one. In general for the complex systems person is able to prove in an infallible way only it will not have there failures. The returned conclusions are provisional: “Taking into account information of which we lay out, here all that we can say.” All new information source must be taken into account because it is likely to call in question the conclusions previously selected.

Humbler of the employees to most eminent of the scientists, all can have their word to say on the studies of reliability.

The decompartmentalization (to open doors and windows) information is a guarantee of reliability.

Extended from the studies of reliability

All the human activities are directed by intentions. For any activity one can pose the problem of the reliability of the means implemented: are the means sufficient to reach fine the aimings? The potential field of the studies of reliability thus includes/understands all the human activities: all products and all services.

Reliability and reliability

Reliability is an essential component of the reliability. Reliability takes part in the availability of equipment. In order to consider a study of exhaustive Reliability, it will be necessary to make complementary studies in the fields of maintainability, the safety and probabilistic calculations of the availability.