Risk matrices are widely used in the process industries. Details vary considerably from company to company, particularly with regard to the size of a risk matrix, but generally a process such as the following is used.
1. A hazard is identified.
2. The consequence of that hazard is determined. The most important consequence is usually do with safety, but environmental, economic and public relations impact can also be considered.
3. The projected frequency (not probability) of occurrence is determined, usually in events per thousand years, but sometimes - as in the case of batch processes - events per thousand operations. Sometimes two levels of frequency are considered: with and without safeguards.
4. A value for overall risk is assigned. Steps 2 through 4 are often managed with three matrices such as those shown below.
Risk levels are divided into categories shown. Level 'A' (red) is "Stop the bus right now". A hazard of high consequence and high frequency has been identified. Immediate corrective action is needed. Level 'B' (orange) shows a high level of risk; action must be taken within say 90 days. Level 'C' (yellow) represents moderate risk. Action must be taken within say 18 months. (In practice, the difference between 'B' and 'C' is that items at the 'C' level can wait for the next scheduled turnaround.) Level 'D' represents a low level of risk - action must be taken, but time is not of the essence. There are other categories. These include:
- 'O' - Operational;
- 'L' - Low Hanging Fruit; and
- 'S' - Standards
In spite of their widespread use, risk matrices do have limitations, some of which are described below.
- 'A' Risk not Realistic Red square risks have such a low likelihood that, were they to exist, we would be having major events on an on-going basis. Since such events occur very rarely that part of the matrix is hardly used.
- 'D' Risk Ignored Green square events tend also to be ignored.
- Upper Right Corner High consequence/low predicted frequency events are also mostly ignored. "Could an airplane crash into our facility? It's not going to happen - move on" (but see the comment below about being ready for black swans).
- ALARP Risk management professionals constantly struggle with the concept of "acceptable risk". Some companies place a diagonal line across the matrix and require that all items on the unacceptable side of the line must be modified so that they become acceptable. But the decision as to where the line is to be located is a subjective one.
- Estimates of Frequency/Probability The value that people assign to the likelihood of an event occurring varies enormously and is particularly dependent on whether they have actually experienced such an event. A related difficulty, particularly when matrices with many rows and columns are used, is the potential for confusing precision with accuracy.
So we are left with the 'B' and 'C' yellow/orange risks in the middle. The matrix does not provide much practical guidance. In practice, risks tend to be categorized on the following lines.
- Low-hanging fruit Just do it. Write that procedure, paint the yellow line. It's not worth spending time on risk ranking. Some of these items may be quite costly, but the response is the same. For example, on one refinery the block valves in a highly corrosive service were chronic leakers. The solution was simple but not cheap: replace all the valves with those from another manufacturer that were known to be effective.
- Be ready for black swans There is no need to analyze rare, catastrophic events such as the airplane crash. But it important make sure that the facility has the necessary response systems in place to handle worst-case scenarios such as explosions, leaks of toxic gas or blowouts.
- Expensive good ideas For most of the 'B' and 'C' items management has neither the funds nor the people to do them all at once. Each item develops its own life and trajectory. Realistically the ones selected are those that have the strongest support or champions. In all cases the necessary decisions without consulting the matrix. The difficulties just discussed may help explain why some companies are giving an increasing amount of attention to just the consequence square.