API RP 754: Process Safety Incident Classification (Tier 1 to 4)

API RP 754 is a recommended practice from the American Petroleum Institute that sets out process safety performance indicators for the refining and petrochemical industries. It sorts events and indicators into four tiers, so a plant can measure process safety itself, not just personal injury rates.

What Is API RP 754?

A site can have a low injury rate and still carry serious process safety weaknesses, such as leaks, failing protective systems, or degraded equipment. API RP 754 gives a common way to count and classify these, so trends can be tracked inside a plant and compared across plants. Its second edition was published in 2016, so always check which edition applies to your site. For a broader introduction to fire safety in industrial facilities, see our guide, What is Fire Safety? A Practical Guide for Industrial Facilities.

It is an industry recommended practice. It does not replace laws, statutory reporting requirements, or a plant’s own emergency response and incident investigation procedures.

The Center for Chemical Process Safety (CCPS) publishes a free guide, Process Safety Metrics, that was aligned with API RP 754 so that either document can be used for the top-tier event definitions. It is a useful companion when classifying events.

The Four Tiers at a Glance

TierFocusTypeTypical use
Tier 1Process safety events with the greatest consequenceMost laggingIndustry-wide reporting
Tier 2Process safety events with lower consequenceLaggingIndustry-wide reporting
Tier 3Challenges to safety systemsLeadingInternal use at the site
Tier 4Operating discipline and management system performanceMost leadingInternal use at the site
API RP 754 four tiers pyramid from most lagging Tier 1 to most leading Tier 4

Lagging indicators count what has already happened. Leading indicators show whether the barriers that prevent incidents are healthy. The four tiers form a continuum from one to the other.

The line between leading and lagging is not sharp. Tier 3 near-miss data are retrospective in form, yet many companies use them as leading signals, and Tier 2 events also act as precursors to Tier 1.

Tier 1 and Tier 2: Process Safety Events

Tier 1 and Tier 2 cover process safety events, which are based on a loss of primary containment (LOPC) that meets defined criteria. Tier 1 has the higher thresholds, and Tier 2 captures qualifying events below them.

Tier 1 consequence criteria include a lost-time injury or fatality, an officially declared community evacuation or shelter-in-place, fire or explosion damage of US$100,000 or more in direct cost, and a release above the threshold quantity within any one-hour period. Tier 2 captures lower consequences, such as a recordable injury, fire or explosion damage from US$2,500 up to US$100,000, and releases between the Tier 2 and Tier 1 threshold quantities. These figures follow the second edition. The exact criteria and exclusions must be taken from the edition in force.

Threshold quantities depend on the material hazard category. Two examples, for releases in any one-hour period:

MaterialTier 2 range starts atTier 1 threshold
Toxic inhalation hazard, Zone A0.5 kg5 kg
Flammable gases50 kg500 kg

These are examples only. The full table, other material categories, and indoor release values must be taken from the edition in force. To compare performance across sites and companies, process safety events are often expressed as a rate: Tier 1 event rate = (number of Tier 1 events ÷ total work hours) × 200,000. The same calculation can be done for Tier 2 events. Using work hours puts plants of different sizes on a common basis.

Tier 3: Challenges to Safety Systems

Tier 3 looks at how often safety barriers are being tested by real conditions. Typical Tier 3 indicators include:

  • Demands on safety instrumented systems and other protective systems
  • Relief device activations
  • Failures found during safety system tests
  • Safe operating limit excursions
  • Piping or vessel thickness found below acceptable limits
  • Loss of containment smaller than the Tier 2 threshold

Not every small leak or alarm is automatically a Tier 3 event. Check the definition used by the standard and by your own indicator methodology.

Tier 4: Operating Discipline and Management Systems

Tier 4 measures whether the activities that prevent incidents are actually being done. Common examples are:

  • Process hazard analysis recommendations closed on time
  • Safety-critical inspections and tests completed on schedule
  • Management of change reviews completed and of good quality
  • Operating procedure reviews and competency checks completed

These are examples, not a mandatory list. Each organisation selects and defines its own Tier 4 indicators.

Loss of Primary Containment (LOPC)

Primary containment is the equipment that holds process material: pipes, vessels, tanks, pumps, compressors, and reactors. LOPC is an unplanned or uncontrolled release from it. Common causes are corrosion and erosion, seal failures, gasket and flange leaks, overpressure, material degradation, vibration, and operating or maintenance errors.

Not every LOPC is a Tier 1 or Tier 2 event. Each one must be tested against the thresholds and exclusions.

How to Classify an Incident

  1. Establish the facts. Record date, location, equipment, material, operating conditions, and sequence of events.
  2. Confirm loss of containment. Identify the source and document the quantity.
  3. Estimate the release in the worst 60 minutes. Thresholds apply to the quantity released in any one-hour period, so find the peak hour of the event and do not average over its full duration. If the duration of a release cannot be determined, assume it was one hour.
  4. Record the consequences. Injuries, fatalities, fire or explosion, damage cost, and off-site effects.
  5. Check scope and exclusions, then test Tier 1 first, then Tier 2. Never assign a tier from how serious the event looks.
  6. Review Tier 3 and Tier 4 separately, and document the decision. Record the criteria used, the evidence, and the reason for the final classification.

If release estimates change during the investigation, review and update the classification.

How We Estimate the One-Hour Release in Practice

In my work, we determine the release within the one-hour window by cross-checking process data, such as flow rate drops, control valve position history, and pressure trends, against physical evidence like liquid pool dimensions or containment levels.

For borderline events, classification relies on engineering consensus. A multidisciplinary panel of process safety specialists, operations managers, and area engineers reviews the data and agrees the final tier. In one case, a high-pressure flange leak sat right on the Tier 1 and Tier 2 mass threshold, until a dynamic flashing model showed that the vapour fraction took the release over the limit within a 60-minute window. It was classified as Tier 1.

Case Study: The Chattering Control Valve Gland

At a pressurised unit, the packing gland of a control valve began leaking hazardous material intermittently. Because the process pressure was fluctuating, the leak sprayed heavily for about 10 minutes, stopped for about 20 minutes while the valve repositioned, and then started again.

The trap. The operations team looked at the clock: the leak started at 08:00 and stopped at 09:15. Since the total duration was more than 75 minutes, they proposed averaging the total mass over the whole duration, which kept the hourly rate just below the Tier 1 threshold.

The reality check. During the process safety review, the DCS trend logs were pulled and the worst continuous 60-minute window was isolated, from 08:15 to 09:15, when pressure peaked and the valve chattered most. Integrating the flow rate over that window alone showed the mass released exceeded the Tier 1 threshold.

API RP 754 worst 60 minute release window compared with a time-averaged release

The lesson. The classification looks at the worst hour, not at an average over the life of the incident. If a leak lasts three hours, the total mass cannot simply be divided by three. The single most severe 60-minute block must be located, and if the threshold is crossed in that hour, the incident is classified at the higher tier.

Practical takeaway. In incident investigation procedures, require the instrumentation team to export raw, high-resolution data for the peak pressure period, and do not accept simple time-averaging to bring a borderline event down to a lower tier.

Screening Examples

ScenarioScreen first asNote
A flammable gas pipeline leaks, ignites, and causes qualifying damageTier 1Confirm material, quantity, and damage against the criteria
A pump seal fails and releases a chemical above the Tier 2 threshold, with no Tier 1 consequenceTier 2Check the material-specific threshold and exclusions
An intermittent valve gland leak, with a peak 60-minute release above the Tier 1 thresholdTier 1Classify on the worst hour, not the average
A slow leak of about 10 bbl of gasoline over two weeks, never above the hourly thresholdNot Tier 1 or 2The threshold is checked per hour; the company may record it as a Tier 3 loss of containment
A small flange leak is controlled quickly and stays below Tier 1 and Tier 2Possible Tier 3May be captured by an internal indicator; check the definition
A safety-critical relief device is overdue for inspection, with no releaseTier 4 concernA leading-indicator issue, not a process safety event

Common Mistakes

  • Averaging a release over the full duration instead of finding the worst 60 minutes
  • Treating every fire or leak as a Tier 1 event
  • Assuming every small release is automatically Tier 3
  • Classifying on quantity alone without checking all the criteria
  • Confusing occupational safety indicators with process safety indicators
  • Not documenting the basis for the classification and the exclusions applied
  • Ignoring repeated small events because none caused major damage

Frequently Asked Questions

Is API RP 754 mandatory in India?
It is an industry recommended practice. Whether it is required for a facility depends on applicable laws, regulatory directions, contracts, and corporate standards.

Are Tier 3 and Tier 4 incidents?
They are better understood as performance indicators than as lower-severity incidents. Their exact definitions depend on the edition and on the indicator methodology.

Where can I find the official criteria?
In the official API RP 754 publication for the edition in force. Use it for precise definitions, thresholds, scope, and exclusions, rather than relying on summaries.

Key Takeaways

  • API RP 754 measures process safety performance in four tiers, from most lagging (Tier 1) to most leading (Tier 4)
  • Tier 1 and 2 classify loss-of-containment events by consequence and by the quantity released in any one-hour period
  • Classify on the worst 60 minutes of an event, never on a time average
  • Tier 3 and 4 are internal indicators of barrier challenges and management system health
  • Test Tier 1 first, then Tier 2, and document the basis for every classification

Conclusion

API RP 754 is useful not just for assigning an incident category, but for using reliable, consistent data to find weaknesses early, strengthen safeguards, and prevent repeat events.

This article is a general summary for educational purposes. It is not a substitute for the official API RP 754 publication, site-specific procedures, or applicable legal requirements. Verify every classification against the current applicable standard and your organisation’s approved methodology.

Scroll to Top