292
E Concepts of Probability Analysis
Risk reduction will be significantly overestimated if common cause failure events
are not identified. To illustrate this, consider a protection layer consisting of two
temperature switches A and B, which perform the same function (i.e., they are
redundant safeguards and using an AND connection). Assuming that the failures
of A and B are independent, the probability (P ) that both safeguards fail can be
calculated using the following expression:
P = P A × P B
(E.14)
However, if the failures of A and B are not independent, the probability that both
safeguards fail will be higher, as shown by:
P > P A × P B
(E.15)
There can be many possible sources of dependency within chemical processes.
The CCPS distinguishes the following six types of functional dependency (CCPS,
1999): (1) on common support systems such as a single electricity source, (2)
on common hardware (e.g., hardware failures affecting shared equipment), (3)
on equipment similarity (e.g., systematic repeated human errors related to a
common design, operating procedures, etc. of similar equipment), (4) on a common
location (e.g., different equipment in one location can be affected by the same
external events, such as environmental conditions, etc.), (5) on a common internal
environment for multiple safeguards (e.g., water in an emergency cooling system,
air in an instrument air system, etc.), and (6) on common operating/maintenance
staff and procedures (e.g., human or procedural errors could occur such as the
miscalibration of multiple safeguards).
Some of these dependencies can be addressed through the use of a modified
fault tree and event tree analysis known as a common cause failure analysis. This
is a technique that identifies common cause events and estimates the probability of
dependent events using parametric models (CCPS, 1999). In the chemical process
industry, the emphasis of common cause failure analysis is usually on systems that
rely on multiple layers of protection and high redundancy to achieve process safety.
The impact of a common cause failure on the probability of a critical top event is
discussed in the next section.
E.4
Probability Analysis for the Grinding Process of Cyanuric
Chloride
The example of the wet grinding of cyanuric chloride was introduced in Chap. 7 for
the application of process risk assessment. The loss event within this process was
identified as the decomposition of cyanuric chloride, and this section will show how
to estimate (1) its probability of occurrence using a fault tree analysis (FTA) and (2)
the probability of the reaction further advancing into a thermal runaway through the
use of an event tree analysis (ETA) and considering existing safeguards.
E Concepts of Probability Analysis
Risk reduction will be significantly overestimated if common cause failure events
are not identified. To illustrate this, consider a protection layer consisting of two
temperature switches A and B, which perform the same function (i.e., they are
redundant safeguards and using an AND connection). Assuming that the failures
of A and B are independent, the probability (P ) that both safeguards fail can be
calculated using the following expression:
P = P A × P B
(E.14)
However, if the failures of A and B are not independent, the probability that both
safeguards fail will be higher, as shown by:
P > P A × P B
(E.15)
There can be many possible sources of dependency within chemical processes.
The CCPS distinguishes the following six types of functional dependency (CCPS,
1999): (1) on common support systems such as a single electricity source, (2)
on common hardware (e.g., hardware failures affecting shared equipment), (3)
on equipment similarity (e.g., systematic repeated human errors related to a
common design, operating procedures, etc. of similar equipment), (4) on a common
location (e.g., different equipment in one location can be affected by the same
external events, such as environmental conditions, etc.), (5) on a common internal
environment for multiple safeguards (e.g., water in an emergency cooling system,
air in an instrument air system, etc.), and (6) on common operating/maintenance
staff and procedures (e.g., human or procedural errors could occur such as the
miscalibration of multiple safeguards).
Some of these dependencies can be addressed through the use of a modified
fault tree and event tree analysis known as a common cause failure analysis. This
is a technique that identifies common cause events and estimates the probability of
dependent events using parametric models (CCPS, 1999). In the chemical process
industry, the emphasis of common cause failure analysis is usually on systems that
rely on multiple layers of protection and high redundancy to achieve process safety.
The impact of a common cause failure on the probability of a critical top event is
discussed in the next section.
E.4
Probability Analysis for the Grinding Process of Cyanuric
Chloride
The example of the wet grinding of cyanuric chloride was introduced in Chap. 7 for
the application of process risk assessment. The loss event within this process was
identified as the decomposition of cyanuric chloride, and this section will show how
to estimate (1) its probability of occurrence using a fault tree analysis (FTA) and (2)
the probability of the reaction further advancing into a thermal runaway through the
use of an event tree analysis (ETA) and considering existing safeguards.