PROBABILISTIC NATURE OF RELIABILITY

We cannot say exactly when a particular product will fail, but we can say what percentage of the products in use will fail by certain times. This is analogous to the reasoning used by insurance companies in defining mortality. We can state reliability in various ways:

• The probability that a product will be performing its intended function at 5000 hours of use is 0.95.

• The reliability at 5000 hours is 0.95 or 95%.

• If we place 1000 units in use, 950 will still be operating with no failures at 5000 hours.

Or to cite another example:

• The reliability at 8000 hours is 0.80.

• The unreliability at 8000 hours is 0.20.

From a service point of view, we may be interested in repair frequency and then we say that 20% of the units will have to be repaired by 8000 hours. Or the repair per hundred units (R/100) is 20 at 8000 hours. The important point is that the reliability is a metric expressing the probability of maintaining intended function over time and is measurable as a percentage.