Sunsetting API RP 581, Part 2: RBI Quantitative POF Model Examples

Why It Matters

Quantitative RBI transforms reliability programs by unlocking data value, improving risk accuracy, and cutting inspection costs, driving stronger ROI for asset owners.

Key Takeaways

• Quantitative RBI integrates historic CML data, reducing uncertainty
• Component-level risk yields precise equipment and circuit assessments
• Tailored inspection recommendations cut waste and improve safety
• Data-driven models enhance ROI for piping RBI programs
• Transition supports SME expertise while leveraging analytics

Pulse Analysis

The shift from deterministic API RP 581 to data‑driven quantitative RBI marks a pivotal evolution in asset integrity management. Traditional models often dismissed older corrosion‑monitoring location (CML) data, labeling it unreliable and defaulting to generic SME‑derived rates. Modern probabilistic frameworks, however, treat historical measurements as probabilistic inputs, allowing the model to weigh data confidence and age. This nuanced handling not only preserves valuable field information but also reduces the uncertainty that has long plagued risk assessments, fostering more informed decision‑making across the plant lifecycle.

Component‑level risk modeling addresses the long‑standing bottleneck of overly simplified piping circuits. By evaluating each CML individually, the quantitative approach aggregates risks to equipment and system levels without sacrificing detail. This granularity yields clearer insight into high‑risk zones, enabling owners to prioritize interventions where they matter most. Early adopters report improved return on investment, as the refined analysis justifies targeted mitigation strategies and minimizes unnecessary inspections, ultimately extending asset life while controlling costs.

Specificity in inspection recommendations is another critical benefit. Instead of generic "A‑level" thinning inspections applied to entire circuits, the new methodology prescribes actions at the exact CMLs that pose the greatest risk. This precision eliminates redundant work, optimizes resource allocation, and enhances safety by ensuring critical locations receive appropriate attention. As the industry embraces these quantitative tools, reliability programs become more agile, data‑centric, and cost‑effective, setting a new standard for risk‑based inspection practices.

Sunsetting API RP 581, Part 2: RBI Quantitative POF Model Examples

By Lynne Kaley, Director of Reliability Strategy at Pinnacle; Andrew Waters, PhD, Director of Data Science at Pinnacle; and Davis Baker, Solutions Engineer at Pinnacle · Appears in the November/December 2025 issue of Inspectioneering Journal

Introduction and Context

The recent Inspectioneering article, “Sunsetting API RP 581 Risk‑Based Inspection Semi‑Quantitative Methodology,” generated a lot of discussion and interest. Since the article was released, requests have been received for examples of what a next‑generation quantitative RBI might look like [1]. We don’t have space in one article to walk through detailed examples for each of the seven categories discussed in the Sunsetting article. For the purposes of this article, we have focused on three limitations referenced in the article to start:

• Limitation 2 of 7 (CML/TML Data Management): Large quantities of corrosion monitoring location (CML) data are available at most facilities, and yet, with the current model, we often hear “our data is not good,” “we don’t believe it,” or “let’s ignore the data more than x years old.” Because of the API 581 model’s deterministic nature, it can be difficult to harness the value of the CML data sets and engineer out the value provided by the old data.

  Additional Note: As a result of these historical data‑handling practices, most RBI programs rely almost exclusively on SME‑assigned corrosion rates and the expectation of localized/generalized thinning. The intent of API 581 was always to use SME data during a period of transition to using reliable historical corrosion‑rate data.

• Limitation 3 of 7 (Piping Modeling): Managing piping in current models is a cumbersome task, with current piping model complexity, including circuitization methodology, representative components, and deterministic values, over‑simplifying the risk calculation across a group of piping. As a result, owner‑operators have often elected to keep piping time‑based. For those who have implemented piping RBI, returns on investment from the analysis and management effort have not been widely published.

  Additional Note: By assessing risk for individual CMLs, component risks can be combined to calculate equipment and/or circuit risks.

• Limitation 4 of 7 (Inspection Recommendation Specificity): Current models provide inspection plans on the circuit or equipment level, with a recommendation to do an “A‑level inspection for thinning.” This leaves room for interpretation, waste, and risk, as the specific inspection plan is conducted on a user‑by‑user basis, sometimes without a complete understanding of the implications of selecting one location over another [2].

  Additional Note: By assessing risk for individual CMLs, inspection recommendations can be specified by CML. In addition, the potential for increased or decreased coverage is feasible.

In this article, we will walk through four equipment‑thinning examples that illustrate how next‑generation models can address these limitations.

---

About the Authors

Lynne Kaley – Director of Reliability Strategy at Pinnacle

Lynne Kaley has over 30 years of refining, petrochemical, and midstream gas‑processing experience. She was a plant metallurgist/corrosion and corporate engineer for over 10 years and has 30+ years consulting with plant management, engineering, and inspection departments, including Risk‑Based Inspection (RBI) development leadership.

Andrew Waters, PhD – Director of Data Science at Pinnacle

Andrew Waters has been with Pinnacle since 2019, where he works on developing the next generation of data‑driven algorithms for reliability. Dr. Waters also specializes in utilizing machine‑learning methods to improve and augment human decision‑making. He holds a doctorate in Electrical and Computer Engineering from Rice University and has over 30 publications in the areas of data science.

Davis Baker – Solutions Engineer at Pinnacle

Davis is an experienced and innovative engineer with over a decade in product development, specializing in mechanical integrity and reliability. He is analytical, process‑driven, and passionate about simplifying complex problems.