Simulation-Assisted Event Tree Analysis
Also known as: Monte Carlo ETA, stochastic event tree analysis, simulation-enhanced ETA, probabilistic event tree simulation
Simulation-assisted event tree analysis (ETA) extends classical event tree analysis by replacing fixed point-estimate branch probabilities with Monte Carlo or discrete-event simulation. This allows analysts to propagate uncertainty through every branch of the tree and obtain full probability distributions over accident sequences and system outcomes, yielding far richer risk insights than deterministic ETA alone.
Read the full method
Sign in with a free account to read this section.
Method map
The neighbourhood of related methods — select a node to explore.
When to use it
Use simulation-assisted ETA when branch probabilities are uncertain or variable, when failure rates depend on time, load, or environment, or when regulations require a quantified uncertainty band around risk estimates (as in nuclear and process industry PRA standards). It is the right choice for complex systems where single point-estimate ETA would mask important variability. Do not use it when branch probabilities are well-established constants and the tree is small enough that analytical propagation is trivial; in that case standard ETA suffices. Avoid this method if the event tree structure itself is highly uncertain — in that case, invest first in qualitative hazard identification (HAZOP, FMEA) before quantification.
Strengths & limitations
- Propagates uncertainty through all branches simultaneously, producing confidence intervals rather than single risk estimates.
- Handles time-dependent, load-dependent, or stochastically varying failure rates that analytical ETA cannot accommodate.
- Supports importance and sensitivity analysis, revealing which components or parameters drive risk most strongly.
- Compatible with Bayesian updating — branch distributions can be refined as new operational or test data become available.
- Scales to large, complex event trees where analytical convolution would be computationally infeasible.
- Requires reliable probability distributions for branch probabilities; if data are sparse, the input uncertainty assumptions dominate the output.
- Computationally intensive — large trees with many branches and fine-grained distributions may require significant run time.
- The event tree structure itself is assumed known; structural model uncertainty (missing branches, wrong initiating events) is not captured by simulation alone.
- Results can be difficult to communicate to non-technical stakeholders accustomed to single-number risk estimates.
Frequently asked
How does simulation-assisted ETA differ from standard (analytical) ETA?
Standard ETA multiplies fixed point-estimate branch probabilities along each path to compute a single frequency for each accident sequence. Simulation-assisted ETA replaces those fixed values with probability distributions and uses Monte Carlo sampling to propagate uncertainty, yielding a distribution of frequencies for each sequence rather than a single number.
How many simulation trials do I need?
For sequences with probabilities in the range of 10^-3 to 10^-5, 10,000 to 100,000 trials are often sufficient for stable mean estimates. Rare sequences (below 10^-6) require variance-reduction techniques such as importance sampling or stratified sampling; naive Monte Carlo with standard trial counts will rarely observe these sequences and will underestimate their frequency.
Can I combine simulation-assisted ETA with fault tree analysis?
Yes — this is standard practice in full probabilistic risk assessment. Fault trees are used to compute the failure probability of each safety system (branch probability), and those probabilities are then passed — with their uncertainty distributions — into the event tree simulation. The combination is sometimes called an integrated fault tree-event tree model.
What software supports this method?
Dedicated PRA codes such as SAPHIRE (US NRC), RiskSpectrum, and CAFTA support event tree quantification with uncertainty propagation. General-purpose Monte Carlo environments (R, Python with NumPy/SciPy, MATLAB) can implement custom simulation-assisted ETA for research applications.
Is this method appropriate for small, simple systems?
Not necessarily the most efficient choice. If the event tree has fewer than five branches and the branch probabilities are well-characterized constants, analytical ETA with simple sensitivity analysis is faster and equally informative. Simulation assistance pays off when uncertainty is substantial, branches are numerous, or failure rates are time-dependent.
Sources
- Zio, E. (2009). Reliability engineering: Old problems and new challenges. Reliability Engineering and System Safety, 94(2), 125–141. DOI: 10.1016/j.ress.2008.06.002 ↗
- Event tree analysis. Wikipedia. link ↗
How to cite this page
ScholarGate. (2026, June 3). Simulation-Assisted Event Tree Analysis. ScholarGate. https://scholargate.app/en/experimental-design/simulation-assisted-event-tree-analysis
Which method?
Set this method beside its closest kin and read them side by side — the library lays the books on the table; the choice is yours.
- Bayesian Event Tree AnalysisExperimental design↔ compare
- Event Tree AnalysisReliability↔ compare
- Failure Mode and Effects AnalysisExperimental design↔ compare
- Fault Tree AnalysisReliability↔ compare
- Risk-based event tree analysisExperimental design↔ compare
- Simulation-assisted fault tree analysisExperimental design↔ compare