342 / 648
CDOIF
Chemical and Downstream Oil
Industries Forum
CDOIF is a collaborative venture formed to agree strategic areas for
joint industry / trade union / regulator action aimed at delivering
health, safety and environmental improvements with cross-sector
benefits.
Guideline – Demonstrating Prior Use v4
Page 17 of 30
A.2.1 Calculation based on Mean Time Between Failures
The example provided below may produce distorted results if the sample size is small, or
detection of failures that have occurred are not identified and recorded in a timely
manner. Further, the accuracy of any failure rate calculation is dependent on failures
being revealed, and replaced immediately (failures revealed only after a proof test may
also distort the calculated results).
Using the data provided by the maintenance management system
4
, or derived from other
sources, the Mean Time Between Failures (MTBF) of the component can be derived as
follows:
MTBF = (number of hours of operation) ÷ (number of failures)
The failure rate can be calculated as follows:
λ
= 1/MTBF
For example, a barrier has been in operation for 10 years (87,600 hours). During that
period, 5 failures have been recorded, therefore:
MTBF = (87,600) ÷ (5) = 17,520 hours
This calculation may also be applied where the number of samples is increased (i.e. the
number of the same component in the same application and environment) but the
sample period is over a shortened period of time. For example, there are 10 barriers that
have been in operation for 1 year (8,760 hours), during that period, 5 failures have been
recorded, in this instance:
MTBF = (10 x 8,760) ÷ (5) = 17,520 hours
The failure rate (
λ
) would be:
λ
= 1/17,520 = 0.000057, or 5.7x10
-5
failures per hour
Note: this calculation will not work where the number of failures is zero. In this instance,
consideration should be given to an approximation to the Poisson distribution curve.
A.2.2 Calculation based on failure data analysis
Where failure rates are available for the component, a more rigorous calculation can be
performed to determine the components PFD and Safe Failure Fraction (SFF).
The types of failure that can be attributed to a component can be described as follows:
•
Safe Failure – a failure that when it occurs causes the system to perform the
function which puts the system into the safe state, this is performed without a
demand from the process and is often referred to a nuisance or spurious trip.
Safe failures can further be categorised as either:
4
When using maintenance records for prior use evidence, the end user should be able to demonstrate that
the records are sufficiently robust and statistically significant