Risk Achievement Worth (RAW)
Risk Achievement Worth (RAW) measures how much worse a system gets if a basic event is forced to fail. It's the ratio of P(top) with the event nailed at probability 1 to the baseline P(top). RAW answers a defensive question: "if this component degraded badly, what would it do to my safety target?" — useful for prioritising surveillance and proof-test intervals.
Formula
RAW(i) = P(top | event i = TRUE) / P(top)
The result is always ≥ 1. RAW = 1 means the event is irrelevant; high RAW means the event is currently being well-defended (its low probability is suppressing significant risk) — and thus deserves close attention.
Worked example
Tree: TOP = AND(a, b, c) with all events at probability 0.01.
P(top) = 0.01·0.01·0.01 = 1e-6 P(top | a = 1) = 1·0.01·0.01 = 1e-4 RAW(a) = 1e-4 / 1e-6 = 100
RAW = 100 means failing a alone makes the system 100× more risky. In a 3-out-of-3 redundant configuration like this, every leaf has the same RAW — they're all 100% essential to the redundancy.
How operators use it
In nuclear PRA and IEC 61511 SIS verification, RAW guides:
- Tech-spec surveillance frequencies — high-RAW components get tested more often.
- Maintenance scheduling — don't take a high-RAW component out of service while another is degraded.
- Configuration risk management — risk-informed decisions about online maintenance windows.
RAW is the dual to RRW: where RAW asks "what's the upside risk if this component degrades?", RRW asks "what's the downside risk if I make it perfect?" Together they bracket the achievable risk envelope around current performance.