Risk Estimation 1.
Why do it?
Is this risk much bigger than that one?
Is it justifiable to invest this much money into improved risk control measures?
Facing Criticism!
You don’t really know what these numbers should be so how can I believe what you are telling me?
A typical “Risk Matrix”
Yet these same people are happy to take a stab at a point in this table – the so-called Risk Matrix and defend their undefendable assertion that they are correct!
Lots of experience of the same process producing adverse
Consequences the value of which we know precisely.
A statistician would say the data is correct with a very high
level of confidence.
Example – road accidents.
Derek Viner
Version 1 Revision 0
Synthesised probability
7
Machine Breakdown Masses of component failure data, reliability engineering principles Personal Injury (OHS, public & product safety)
Small amount of occurrence data. Some risk analysis tools Plant and Property Damage E.g. Fires and Explosions Little occurrence data.
Synthesis of risk scenarios. Good design guidelines Catastrophes
E.g High technology And high hazard system Failures Little or no occurrence data.
Synthesis of risk scenarios
Probability or Frequency Consequence Value Availability of Occurrence Data, and Methods Used to Manage Risk
Synthesised probability
8
Quantified Risk Analysis (QRA): QRA uses judgement,
experience or indicative values to synthesise a real
number value of the risk variables using logic diagrams.
Risk is estimated as a Cost per year value.
9
The Risk Matrix is a typical example of the use of subjective
scales of values, using word choices (a nominal scale)
associated with a set of numbers (an ordinal scale)
Conclusion
10
Subjective estimation – the least reliable (and the most used!)
Objective estimation – limited probably to the two domains of
plant failure and vehicle crashes
Modelled Risk estimation – the opportunity to do something a
lot better than Subjective estimation
Modelled Risk
11
The use of Probability estimates based on the use of order of
magnitudes known to be relevant for similar generic failures
(equipment or people)
Direct estimates of the position
of the line on the Risk Diagram
space, achieved through the
application of experience and
judgement (as well as statistical
method) involving the
parameters of Exposure and
Frequency.
Derek Viner
Version 1 Revision 0
Refuting the Critics
13
Estimates of any parameter, whether Exposure, Frequency,
Probability or Consequence Value is likely to centre on a
mean value of the estimate.
Practical numerical tools exist to replace these single point
estimates with a range of possible values and to assign a
distribution of the probability of a value existing within the
range.
Numerical approaches like the Monte Carlo technique enable
a multitude of possible calculations to be made in a short
time to produce a final result (how big is the Risk?) which
itself is a mean and a range of possible values distributed
about that mean.
The post Risk Estimation 1 appeared first on My Assignment Online.
