6.3 Effect of Past Inspection Effectiveness (Categories A through E) on POF Determination

Fundamental Role of Inspection Effectiveness in RBI

Under API Recommended Practice 580 (Section 13) and API Recommended Practice 581 (Part 2), non-destructive examination (NDE) does not physically alter, repair, or stop active equipment degradation. Instead, the fundamental purpose of inspection in Risk-Based Inspection (RBI) is to reduce uncertainty regarding the physical state, remaining wall thickness, and flaw presence within pressure-containing equipment.

In quantitative Probability of Failure (POF) modeling, structural condition is expressed as a probability density function. Prior to inspection, statistical uncertainty regarding true corrosion rates and remaining wall thickness is broad (high variance $\sigma^2$). Performing a high-quality inspection updates the probability distribution using Bayesian probability updating rules, narrowing the distribution spread and shifting the mean. This directly lowers the calculated probability that remaining thickness ($t$) is less than the minimum required structural thickness ($t_{min}$), thereby reducing the calculated Damage Factor ($DF$).

---

API RP 581 Inspection Effectiveness Categories (A through E)

API RP 581 categorizes past inspection history into five standardized Inspection Effectiveness Categories, defined by the confidence level ($P_{confidence}$) that the NDE method will successfully detect, characterize, and size the specific active damage mechanism:

1. Category A: Highly Effective ($80\% - 100\%$ Confidence Credit)

• Technical Requirement: The NDE method will correctly detect, locate, and size the active damage mechanism in $>90\%$ of occurrence cases. Requires comprehensive spatial coverage (typically $100\%$ scan coverage of high-susceptibility zones) and high resolution.
• NDE Examples:
  • 100% Automated Ultrasonic Testing (AUT) wall mapping grid for localized HCl or naphthenic acid thinning.
  • 100% Internal Wet Fluorescent Magnetic Particle Testing (WFMT) on all un-PWHT internal welds for Wet $\text{H}_2\text{S}$ cracking.
  • Advanced Phased Array Ultrasonic Testing (PAUT with TFM/FMC) for High-Temperature Hydrogen Attack (HTHA) microfissuring.

2. Category B: Usually Effective ($60\% - 80\%$ Confidence Credit)

• Technical Requirement: High probability of detecting damage with moderate to good spatial coverage ($20\% - 50\%$ inspection coverage of suspect locations) or minor detection limits.
• NDE Examples:
  • UT thickness grid readings covering $20\% - 50\%$ of high-turbulence piping elbows.
  • Profile Radiographic Testing (RT) targeted at water-wash dew point injection locations.
  • 50% WFMT examination of internal weld heat-affected zones.

3. Category C: Fairly Effective ($40\% - 60\%$ Confidence Credit)

• Technical Requirement: Moderate probability of detecting damage; limited spatial coverage ($10\% - 20\%$) or lower resolution tools.
• NDE Examples:
  • Spot Ultrasonic Thickness (UT) measurements at fixed Thickness Monitoring Locations (TMLs) for uniform corrosion.
  • Dry Magnetic Particle Testing (MT) or Liquid Penetrant Testing (PT) on $20\%$ of external welds.
  • Internal Visual Inspection (VT) of accessible pressure vessel shell surfaces.

4. Category D: Poorly Effective ($20\% - 40\%$ Confidence Credit)

• Technical Requirement: Low probability of detecting active damage due to unsuitable NDE technique selection, minimal spatial coverage ($<10\%$), or severe access restrictions.
• NDE Examples:
  • External visual examination (VT) without insulation removal to assess Corrosion Under Insulation (CUI).
  • Spot UT thickness measurements for highly isolated localized pitting corrosion.
  • Single-point UT reading on a large diameter column shell.

5. Category E: Ineffective ($<20\%$ Confidence Credit / Zero Credit)

• Technical Requirement: The inspection technique or coverage has negligible capability to detect the specific active damage mechanism. Provides zero Bayesian credit in API RP 581 algorithms.
• NDE Examples:
  • Visual inspection through a 2-inch sight glass to evaluate stress corrosion cracking.
  • Performing straight-beam UT thickness measurements when the active mechanism is environmental micro-cracking (e.g., Caustic SCC).
  • Omitting an active damage mechanism entirely during RBI screening.

---

NDE Method Selection Matrix by Damage Mechanism

Assigning an effectiveness category requires matching the NDE technique to the specific damage morphology defined in API RP 571:

---

Bayesian Mathematical Impact of Inspection on $DF_{thin}$

In API RP 581 (Part 2), the calculation of the Thinning Damage Factor ($DF_{thin}$) incorporates past inspection history through parameters $N_A, N_B, N_C, N_D$, representing the cumulative number of Category A, B, C, and D inspections performed on the component.

The Bayesian update modifies the structural variance parameter ($\sigma_{art}$) of the remaining wall thickness distribution:

$$\sigma_{art} = \frac{\sigma_{prior}}{\sqrt{1 + k_A N_A + k_B N_B + k_C N_C + k_D N_D}}$$

Where $k_A > k_B > k_C > k_D > k_E = 0$ are weighting constants reflecting inspection effectiveness (e.g., $k_A \approx 5.0$, $k_B \approx 2.0$, $k_C \approx 0.5$, $k_D \approx 0.1$, $k_E = 0$).

Key Mathematical Insights:

1. Variance Reduction: As high-effectiveness inspections ($N_A$) are recorded, $\sigma_{art}$ decreases rapidly. A narrower posterior distribution reduces the area under the probability curve where $t < t_{min}$, suppressing $DF_{thin}$.
2. Compound Inspection Credit: Performing two Category B inspections ($N_B = 2$) compounds statistical confidence, yielding a variance reduction equivalent to a Category A inspection.
3. The Ineffective Penalty ($N_E$): Because $k_E = 0$, conducting a Category E inspection adds zero to the denominator. The calculated $DF_{thin}$ continues to grow exponentially over operating time, driving up total POF as if no inspection had ever occurred.

---

Technical Worked Example: Comparative POF Impact of Cat A vs. Cat E Inspection

Baseline Operating Scenario:

• Component: Heavy Gas Oil Absorber Column Shell (Carbon Steel, nominal thickness $t_{nom} = 12.7\text{ mm}$, $t_{min} = 8.0\text{ mm}$).
• Active Damage Mechanism: Localized Ammonium Bisulfide ($\text{NH}_4\text{HS}$) corrosion (estimated corrosion rate $CR = 0.25\text{ mm/yr}$).
• Operating Time: $15\text{ years}$ since commissioning ($3.75\text{ mm}$ expected wall loss, $t_{art} = 8.95\text{ mm}$).
• Baseline $gff_{total}$: $3.06 \times 10^{-5}$ failures/yr.

Scenario Comparison:

Conclusion

By executing a Category A inspection instead of a Category E inspection, the facility reduced the calculated Probability of Failure by over 40-fold (4300%), demonstrating the decisive impact of inspection effectiveness selection on RBI risk management and turnaround scheduling.