sensitivity-analysis

LOPA Sensitivity Analysis

Purpose

Perform structured sensitivity analysis on LOPA results to assess the robustness of risk decisions. This skill evaluates how variations in input parameters (initiating event frequencies, IPL PFDs, conditional modifiers) affect the mitigated scenario frequency and risk tolerance outcome.

Dependencies

• ccps-2001-checker — Verified frequency ranges and PFD ranges from Tables 5.1, 6.3–6.5
• lopa-report-generator — Scenario data and baseline calculations

When to Apply Sensitivity Analysis

Sensitivity analysis is warranted when:

• The mitigated frequency is within one order of magnitude of the risk tolerance criterion
• Screening values have been used rather than site-specific data
• IPL PFDs are at the boundary between ranges (e.g. close to SIL boundary)
• The scenario has few IPLs (risk decision is sensitive to any single IPL's PFD)
• Stakeholders require ALARP demonstration
• There is uncertainty in conditional modifier values

Analysis Types

Type 1 — Single Parameter Variation

Vary one input parameter at a time across its credible range while holding all others constant.

Procedure:

1. Identify the baseline LOPA result (mitigated frequency and risk decision)
2. For each parameter, apply the upper and lower bound from the CCPS range:
   • Initiating event frequency: use full range from Table 5.1
   • IPL PFDs: use full range from Tables 6.3–6.5
   • Conditional modifiers: use reasonable upper and lower estimates
3. Recalculate mitigated frequency for each variation
4. Determine whether the risk decision changes

Output table format:

SENSITIVITY ANALYSIS — SCENARIO [number]
Baseline mitigated frequency: [value] per year
Risk tolerance criterion: [value] per year
Baseline decision: [Met / Not Met]

| Parameter | Baseline Value | Low Estimate | High Estimate | Mitigated Freq (Low) | Mitigated Freq (High) | Decision Changes? |
|-----------|---------------|-------------|---------------|---------------------|----------------------|-------------------|
| IE Frequency | [val] | [val] | [val] | [val] | [val] | [Yes/No] |
| IPL 1 PFD | [val] | [val] | [val] | [val] | [val] | [Yes/No] |
| IPL 2 PFD | [val] | [val] | [val] | [val] | [val] | [Yes/No] |
| P_ignition | [val] | [val] | [val] | [val] | [val] | [Yes/No] |

Type 2 — Worst-Case / Best-Case Bounding

Calculate mitigated frequency using:

• Best case: All parameters at their most favourable (lowest frequency, lowest PFD) credible values
• Worst case: All parameters at their least favourable (highest frequency, highest PFD) credible values
• Screening case: All parameters at CCPS screening values (the standard conservative LOPA)

Type 3 — Risk Tolerance Gap Assessment

When the mitigated frequency exceeds the risk tolerance criterion, calculate the required additional risk reduction:

Required additional PFD = Risk tolerance criterion / Current mitigated frequency

Then identify candidate IPLs that could provide the required PFD, referencing Tables 6.3–6.5.

Type 4 — IPL Removal Analysis

For each credited IPL, calculate the mitigated frequency if that IPL were removed (PFD set to 1.0). This identifies which IPLs are critical to the risk decision.

Calculation Verification

All sensitivity calculations must be presented to the user for verification before inclusion in any report. Present:

1. The baseline calculation (restated)
2. Each variation with full working
3. Summary table of results
4. Clear statement of which parameters the risk decision is sensitive to

Presentation in Reports

When sensitivity analysis results are included in a LOPA report (Section 8), present:

• Purpose and scope of the sensitivity analysis
• Parameters varied and their ranges (with source references)
• Summary results table
• Key findings: which parameters most influence the risk decision
• Implications for risk management (e.g. importance of maintaining specific IPL performance)

Language and Tone

• Use factual, neutral language when describing sensitivity results
• Avoid characterising results as "concerning" or "alarming" — state the numerical outcomes
• Use Australian English spelling throughout
• Present ranges as factual data, not as qualifications of the base case