(152e) Validation of PHAST Dispersion Model As Required for USA LNG Siting Applications

PHMSA in consultation with FERC has issued guidance relating to approval in the USA of atmospheric dispersion models for LNG siting applications. This guidance includes a Model Evaluation Protocol (MEP), and an associated experimental database against which the model needs to be validated. Final approval was obtained in October 2011 for the PHAST dispersion model UDM, and this paper summarises the submission of this model according to the above PHMSA guidance.

The UDM model is the main model in the hazard assessment software package PHAST. It is a Unified Dispersion Model (UDM) for two-phase jet, heavy and passive dispersion including droplet rainout and pool spreading/evaporation. The UDM model can deal with a wide range of scenarios including both pressurised jet releases and unpressurised releases.

The paper first summarises the overall verification and validation of the UDM (both for LNG and other chemicals), which includes validation against experiments included in the REDIPHEM database (as produced as part of the EU project SMEDIS) and the MDA database (as produced by Hanna et al.). Subsequently the paper outlines UDM validation against experiments as required by the PHMSA for the LNG model evaluation protocol (MEP). This includes large-scale LNG field experiments involving dispersion from a liquid pool (Maplin Sands, Burro, Coyote) and  large-scale Freon/Nitrogen field experiments involving dispersion  from ground-level vapour area sources (Thorney Island). Overall very good agreement was obtained for the field experiments.

The results of the validation were submitted to PHMSA including required statistics for model accuracy (mean ratio observed to predicted, variance, etc.). The PHMSA submission also included detailed technical documentation for the UDM dispersion model (theory, verification and validation), and a report relating to conformance of the UDM against the model evaluation protocol (MEP).  The authors believe that this type of model evaluation exercise is essential to ensure sufficient model accuracy and to gain community support.