API Publication 331-1994: Definitive Analysis of Aboveground Storage Tank Failures

Leveraging Historical Incident Data to Enhance Risk Management and Integrity Programs

Scope and Purpose of API Publication 331-1994

API Publication 331-1994, Aboveground Storage Tank Failure Survey, Update and Review of Recent Incidents, represents a landmark effort by the American Petroleum Institute (API) to systematically collect and analyze failure data from atmospheric aboveground storage tanks (ASTs). Prior to this publication, the industry lacked a statistically robust, centralized repository of failure incidents, making it difficult for operators to prioritize inspection resources or accurately quantify the risk of a loss of containment.

The scope of the survey was specifically defined to include atmospheric storage tanks—encompassing fixed-roof, external floating-roof, and internal floating-roof designs—typically used in the refining, petrochemical, and bulk liquid storage sectors. The survey explicitly excluded pressurized storage spheres, refrigerated cryogenic tanks, and process piping, thereby concentrating the analysis on the most prevalent type of large-scale bulk liquid containment. The outcome was not a prescriptive standard but a technical report that provided the foundational failure statistics which would later drive the development of risk-based inspection (RBI) methodologies and the proactive integrity management approaches outlined in API 650 and API 653.

Technical Data and Failure Analysis Methodology

Data Collection and Classification

The methodologies employed in API Publ 331 involved collating incident data from member companies, regulatory submissions, and historical archives. Incidents were categorized by failure mode (leak vs. catastrophic rupture), cause, and equipment type. The survey specifically analyzed the frequency of failures and the relative contribution of various root causes.

Key Methodology Note: The survey distinguished between leaks (slow loss of containment) and catastrophic failures (rapid structural collapse with major spill or fire). This distinction is critical, as the leading root causes differed significantly between the two failure modes.

Statistical Findings: Root Causes of Failure

The data collected provided a revelatory breakdown of failure causes. The table below summarizes the primary categories and their typical contribution to the overall incident count as documented in the survey.

Root Cause CategoryShare of Leak IncidentsShare of Catastrophic FailuresPrimary Mechanism
Corrosion (Internal/External)~75%~10%Floor pitting, shell thinning, under-deposit corrosion
Operational Error~5%~40%Overfilling, overpressure, vacuum collapse
Lightning / Fire~2%~25%Rim seal ignition, vent flashback
Mechanical / Structural~10%~15%Welding defects, brittle fracture, foundation failure
Unknown / Other~8%~10%Various

Table 1: Distribution of failure modes by root cause category, derived from the analysis in API Publ 331-1994.

Implementation Highlights and Industry Impact

Redefining the Risk Landscape

The survey’s most profound impact was redefining the risk landscape for tank operators. It clearly demonstrated that the lifecycle of a tank presented two distinct risk profiles: the high-probability, low-consequence risk of corrosion leaks (primarily in the tank floor) and the low-probability, high-consequence risk of operational ruptures.

This dual nature of risk directly informed the development of Risk-Based Inspection (RBI) protocols in API RP 580 and API RP 581. The statistical failure frequencies published in the survey serve as the generic baseline frequencies which analysts adjust using facility-specific design and inspection factors.

Enhancing Integrity Programs

  • Bottom Inspection: The dominance of floor corrosion drove the widespread adoption of advanced NDT techniques (magnetic flux leakage, ultrasonic mapping) and stringent minimum thickness requirements in API 653.
  • Overfill Prevention: The significant share of catastrophic failures attributed to operational error reinforced the need for high-integrity overfill protection systems, independent level alarms, and robust management of change (MOC) procedures.
  • Lightning Protection: The data on rim seal fires led to improved designs for floating roof shunts, gap minimization, and enhanced lightning diversion paths.
Implementation Milestone: Many industry programs use the findings of API Publ 331 to justify the implementation of Integrity Operating Windows (IOWs) for pressure, level, and temperature, specifically targeting the operational error risks highlighted in the survey.

Compliance Notes, Limitations, and Modern Relevance

Compliance and Audit Considerations

While API Publ 331 is a publication and not a mandatory code, its content is deeply embedded in the technical fabric of industry best practices. Auditors and regulatory bodies (such as the US EPA or OSHA) may reference the failure statistics within this document when evaluating the adequacy of an operator’s risk assessment or Spill Prevention, Control, and Countermeasure (SPCC) plan. Demonstrating that an integrity program effectively addresses the specific failure modes identified in the survey—balancing corrosion management with operational safeguards—is considered a hallmark of a robust program.

Acknowledging the Limitations

It is essential for technical professionals to apply the data from API Publ 331-1994 with a clear understanding of its context. The data is primarily historical, covering incidents from the mid-20th century up to the early 1990s. Industry practices have evolved substantially since that time. The widespread adoption of cathodic protection, improved coating systems, more rigorous inspection per API 653, and enhanced operational procedures have demonstrably reduced the generic failure frequencies reported in the survey. Modern RBI programs must update this legacy data using site-specific data, Bayesian analysis, and newer industry joint project databases to maintain an accurate risk profile.

Caution: Relying solely on the unmodified failure frequencies from API Publ 331-1994 without adjusting for modern design improvements, inspection history, and current operational management practices can lead to overly conservative or inaccurate risk assessment outcomes. It is a baseline, not a current snapshot.

Despite its age, API Publication 331-1994 remains an indispensable piece of technical literature. It provides the fundamental understanding of why tanks fail, moving the industry from reactive maintenance to proactive, risk-informed integrity management.

Frequently Asked Questions

Q: Is API Publication 331-1994 a mandatory code like API 653?
A: No. It is a technical report and survey of historical incidents. However, its statistical findings provide the risk-based justification for many requirements found in mandatory codes and recommended practices like API 653 and API RP 580.
Q: Should my tank integrity program directly cite the failure rates from this 1994 publication?
A: Yes, particularly when performing qualitative or quantitative risk assessments. The data is a valid generic baseline. For higher-tier risk assessments, these generic rates should be modified by equipment factors (based on actual inspection results, corrosion rates, and protective systems) to reflect the current realistic risk level.
Q: What was the most surprising finding of the API Publ 331 survey?
A: The most impactful finding was the clear bifurcation of risk. While leaks were overwhelmingly caused by corrosion, catastrophic ruptures and major fires were most often triggered by operator error (overfill, overpressure). This forced the industry to broaden its focus from purely material and corrosion engineering to include robust operational integrity systems.

This article is provided for technical reference. For the latest regulatory requirements and industry standards, please refer to the current edition of API 653 and facility-specific regulations.
© 2026 Technical Standards Analysis. All rights reserved.

📥 Standard Documents Download

🔒
Please wait 10 seconds, the download links will appear after the ad loads

Leave a Reply

Your email address will not be published. Required fields are marked *