Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
API Publ 2376-1998: Guidelines for LNG Vapor Dispersion Model Evaluation and Application
Technical requirements for demonstration of vapor dispersion modeling for LNG facilities and risk assessment
Scope and Purpose
API Publication 2376 (1998) establishes a framework for evaluating and demonstrating the suitability of vapor dispersion models for liquefied natural gas (LNG) applications. The document addresses the need for standardized procedures to assess model predictions against experimental data, ensuring that models used in LNG facility siting, risk assessment, and emergency response planning produce reliable and conservative results. The scope includes both dense-gas and passive dispersion regimes, covering typical release scenarios such as instantaneous spills, continuous releases, and jet releases from LNG operations.
Originally developed by the American Petroleum Institute (API) in response to regulatory requirements and industry best practices, this publication remains a key reference for engineers, safety analysts, and regulatory reviewers involved in LNG vapor dispersion analysis. It does not prescribe a specific model but defines a consistent methodology for model selection, validation, and application.
Key Insight: API Publ 2376 emphasizes that no single model is universally applicable; the selection must consider the release scenario, atmospheric conditions, and the desired level of conservatism. The publication provides a systematic approach to identify model limitations and uncertainties.
Technical Requirements and Methodology
Model Evaluation Protocol
The core technical requirement of API Publ 2376 is a structured evaluation protocol that includes:
- Source term characterization: Definition of release rate, duration, liquid pool formation, flash evaporation, and aerosol behavior.
- Atmospheric input data: Wind speed, stability class, ambient temperature, humidity, and surface roughness.
- Comparison metrics: Quantitative measures such as peak concentration, cloud arrival time, cloud duration, and distance to lower flammability limit (LFL).
- Validation data sets: Reference experimental trials (e.g., Burro, Coyote, Maplin Sands) used to assess model performance.
| Model Category | Typical Models | Dispersion Regime | Key Input Parameters | Validation Focus (API 2376) |
|---|---|---|---|---|
| Gaussian-based | SLAB, HGSYSTEM, ALOHA | Passive / slightly dense | Release rate, wind speed, stability | Far-field concentration, LFL distance |
| Lagrangian particle | SCIPUFF, CFD-Lagrangian | Dense to passive | Release geometry, turbulence parameters | Cloud meander, peak concentration |
| Computational Fluid Dynamics (CFD) | FLACS, ANSYS CFX, OpenFOAM | All regimes, complex geometries | 3D mesh, boundary conditions, turbulence model | Obstacle interaction, near-field detail |
Performance Metrics and Acceptance Criteria
API Publ 2376 defines several statistical measures to quantify model accuracy:
- Fractional bias (FB): Indicates systematic over- or under-prediction.
- Normalized mean square error (NMSE): Overall spread of model predictions relative to observations.
- Geometric mean bias (MG) and geometric variance (VG): Used for log-normal concentration distributions typical of dense gas dispersion.
The publication provides suggested acceptance criteria for these metrics based on historical validation studies. Models must demonstrate predictive capability within defined limits to be considered acceptable for LNG risk assessments. The criteria are not absolute pass/fail but are used to build confidence in the model’s application for a specific scenario.
Important Consideration: The validation data sets referenced in API Publ 2376 primarily involve non-cryogenic releases (isothermal dense gases) and small-to-medium scale LNG spills. Users should be aware of potential scale-up uncertainties when applying validated models to large, catastrophic scenarios.
Implementation Highlights
Application in LNG Facility Siting
The primary application of API Publ 2376 is in LNG facility siting studies, where dispersion modeling determines the exclusion zones for public safety. The publication’s evaluation protocol helps analysts choose a model that:
- Represents the dominant physical processes (gravity spreading, air entrainment, heat transfer from ground).
- Has been validated against field data of comparable scale and atmospheric stability.
- Provides output suitable for comparison with regulatory thresholds (e.g., ½ LFL distance).
Many regulatory jurisdictions (e.g., US DOT PHMSA, NFPA 59A) implicitly or explicitly reference the methodologies outlined in API Publ 2376 when reviewing dispersion modeling studies.
Use in Risk Assessment
Beyond siting, the publication supports quantitative risk assessment (QRA) by defining how model uncertainties should be handled. The guidance encourages the use of multiple models or sensitivity runs to capture variability in atmospheric conditions, release orientations, and model physics.
Best Practice: When implementing API Publ 2376, consider establishing a model validation matrix that compares candidate models against field trials for the specific release scenarios (e.g., instantaneous spill vs. pressurized jet). Document all assumptions, input ranges, and performance metrics to justify model selection to regulators or third-party reviewers.
Compliance and Regulatory Notes
Regulatory Status
API Publ 2376 is a voluntary industry publication, not a mandatory standard. However, it is frequently cited in regulatory guidance because it provides a defensible, science-based procedure for vapor dispersion analysis. Agencies such as the Pipeline and Hazardous Materials Safety Administration (PHMSA) and the U.S. Coast Guard may require that dispersion models used in LNG applications meet the validation criteria described in API Publ 2376 or equivalent protocols.
Limitations and Updates
Users must recognize that API Publ 2376 was published in 1998 and reflects the state of modeling and experimental data available at that time. Since then, significant advances have been made in CFD capabilities, field-scale experiments (e.g., tests with larger LNG spills), and uncertainty quantification techniques. Therefore:
- Models validated solely against the original data sets should be reassessed against more recent, large-scale experiments (e.g., Falcon, Joule-Brayton).
- Additional validation requirements may be imposed by local regulators, particularly for complex terrain or congested environments.
- The publication’s framework remains valid, but the specific acceptance criteria may need revision based on updated statistical methods (e.g., the use of the F-test for model acceptance).
Warning: Relying exclusively on API Publ 2376 without considering newer validation data or advanced modeling techniques may lead to non-compliance with some international standards (e.g., ISO 16924, EN 1473). Always verify the current regulatory expectations in your jurisdiction.
FAQs
Q: What is the main difference between API Publ 2376 and other dispersion model evaluation guidelines (e.g., ASTM E2582, CCPS)?
A: API Publ 2376 focuses specifically on LNG vapor dispersion and provides evaluation criteria tailored to dense gas behaviors (gravity spreading, air entrainment). It emphasizes comparison with field trial data, while other guidelines may offer broader model validation protocols that include chemical and neutrally buoyant releases. The publication is often considered the de facto reference for LNG dispersion model validation in the United States.
A: API Publ 2376 focuses specifically on LNG vapor dispersion and provides evaluation criteria tailored to dense gas behaviors (gravity spreading, air entrainment). It emphasizes comparison with field trial data, while other guidelines may offer broader model validation protocols that include chemical and neutrally buoyant releases. The publication is often considered the de facto reference for LNG dispersion model validation in the United States.
Q: Can I use API Publ 2376 for models applied to other cryogenic fuels (e.g., liquid hydrogen, ammonia)?
A: The evaluation methodology is physically based and can be adapted to other cryogenics, but the specific validation data sets and acceptance criteria are optimised for LNG (methane-dominated). The source term behavior (flash evaporation, pool spreading) differs significantly for other cryogens, so additional validation would be required. Users should develop a separate validation plan following the general framework but using appropriate experimental data for the substance of interest.
A: The evaluation methodology is physically based and can be adapted to other cryogenics, but the specific validation data sets and acceptance criteria are optimised for LNG (methane-dominated). The source term behavior (flash evaporation, pool spreading) differs significantly for other cryogens, so additional validation would be required. Users should develop a separate validation plan following the general framework but using appropriate experimental data for the substance of interest.
Q: Is it necessary to perform a model validation study every time a new scenario is analyzed?
A: Not necessarily. API Publ 2376 advocates that models should be validated for a range of scenarios similar to the intended application. If a model has shown acceptable performance for a set of conditions that encompasses the new scenario (e.g., similar release rates, atmospheric stability, and terrain), and the model is applied within its validated parameter space, a new full validation is not mandatory. However, sensitivity analysis and documentation of the applicability justification are essential.
A: Not necessarily. API Publ 2376 advocates that models should be validated for a range of scenarios similar to the intended application. If a model has shown acceptable performance for a set of conditions that encompasses the new scenario (e.g., similar release rates, atmospheric stability, and terrain), and the model is applied within its validated parameter space, a new full validation is not mandatory. However, sensitivity analysis and documentation of the applicability justification are essential.
Q: What are the principal limitations of API Publ 2376 noted by modern practitioners?
A: The most cited limitations include: (1) reliance on small-to-medium scale experiments (<20 m³ spills) with limited validation for very large releases; (2) absence of guidance for CFD model validation metrics and grid sensitivity; (3) minimal discussion of uncertainty propagation in risk assessments; and (4) no incorporation of multi-phase jet dispersion models that have become common in LNG design. Many practitioners supplement API Publ 2376 with additional protocols (e.g., ERM/Colebrander or NORSOK) to address these gaps.
A: The most cited limitations include: (1) reliance on small-to-medium scale experiments (<20 m³ spills) with limited validation for very large releases; (2) absence of guidance for CFD model validation metrics and grid sensitivity; (3) minimal discussion of uncertainty propagation in risk assessments; and (4) no incorporation of multi-phase jet dispersion models that have become common in LNG design. Many practitioners supplement API Publ 2376 with additional protocols (e.g., ERM/Colebrander or NORSOK) to address these gaps.
This technical article is prepared for informational purposes and reflects general understanding of API Publ 2376 (1998). Users should refer to the original publication for authoritative requirements. All year references are as of 2026.