Common Tools for Risk Management and the Use of Probability*
| Tool or Analytical Process | Comment on Probability Concepts Used in These Methods |
|---|---|
| General, “high-level,” problem scoping and risk framing tools: implicit probability | |
| Hierarchical holographic modeling | Mapping possibilities in a hierarchy. Categories attempt to describe 100% of the probability for risks of concern. |
| Structured “what-if?” technique (SWIFT) | Identifying possible scenarios without probabilities. |
| Hazard analysis and critical control points (HACCP) | Hazard analysis is scenario generation—possibilities, not probabilities. Some uses might include the relative likeliness or probability for critical control points. |
| Hazard operability analysis (HAZOP) | Hazard analysis is scenario generation. Favors possibilities, not probabilities. |
| Preliminary hazard analysis (PHA) | Hazards analyses generate scenarios for risk assessments. Favors possibilities, not probabilities. |
| Root cause analysis (RCA) or cause-and-effect tools: implicit to explicit probability | |
| Ishikawa (fishbone) | Broad categories cover all possibilities lead to the event (failure) |
| Cause and effect (tree diagram) | Broad categories cover all possibilities lead to the event (failure) |
| Failure modes and effects analysis | Independence of scenarios needed; distributions useful for serial analysis; |
| Linguistically imprecise borders between probabilities are possible; | |
| Quantitative tools: explicit probability | |
| Event tree analysis (ETA) | Using initiating event and inductively maps the faults and probabilities of multiple outcomes. |
| Fault tree analysis (FTA) | Starting at the outcome, works deductively to map causes and probabilities for the specific causal pathways. Sequential event analyses (conditional probabilities) are implemented. |
| Cause-consequence analysis | Combines FTA and ETA for a single system view of multiple faults and multiple consequences. |
| Markov chain | The probability for elements of a system transitioning from one state to another (e.g., operating→failing) is used. Often combined with Monte Carlo simulations and Bayesian Monte Carlo. |
| Monte Carlo simulations | Propagation of probability distributions by modeling uncertain, linked variables in the process or system. Yields uncertainties in the overall process. Can be used to compare scenarios under and sensitivity analyses. |