Many companies already use a risk matrix for HAZOP ranking. A matrix can also give a first SIL target: each cell holds the SIL needed to bring that severity–likelihood combination to the tolerable region.
Making it consistent
- Define likelihood categories numerically (frequency bands per year), not only in words.
- Define severity categories with the same consequence classes as the TMEL table.
- The SIL in a cell follows from (upper frequency of the likelihood band) / (tolerable frequency of the severity class): an RRF of 10–100 gives SIL 1, and so on.
- Mark cells where a single SIF is not enough ("b") and cells where no special safety requirement exists ("a").
Limits
A matrix is coarse: one band can span a factor of ten. Use it for screening and send anything above SIL 1 (or any multiple fatality case) to LOPA.
In First Out AI
Risk calibration tab: choose 3 to 6 levels each way, edit labels and cells, and assess a scenario by choosing its severity and likelihood. The example matrices show typical layouts; they are not calibrated.
Written for First Out AI as engineering guidance. It supports engineering work and does not replace it; check every value against your standards, certificates and company rules with a competent functional safety engineer.