New Quantitative Risk Criteria for U.S. LNG Facilities

The recent increase in natural gas production has expanded the outlook for exporting domestically-produced natural gas.  As a result, many existing LNG import facilities in the U.S. are proposing to construct liquefaction facilities collocated with existing import terminals.  The Federal Energy Regulatory Commission (FERC) reviews and approves applications for construction.  As part of this process, LNG facilities must perform specific hazardous release consequence analyses to demonstrate flammable vapor dispersion exclusion zones, thermal radiation exclusion zones, and explosion overpressure exclusion zones to meet criteria established in 49 CFR 193 and NFPA 59A.

Over the last few years, the FERC has refined the risk analysis criteria for LNG facility construction projects.  The current regulations do not specify quantitative risk assessment (QRA) as a necessary hazard analysis tool.  However, FERC has recently issued interpretations for leak size determination that are based upon failure frequencies for piping systems.  The underlying studies that FERC relied upon to derive the failure frequency criterion for single accidental leak source size determination (i.e., 3.0x10^-5 per year to 5.0x10^-5 per year) are largely based on generic industry data not LNG-specific data.  The net result of this criterion has been to require elements of QRA as part of LNG facility permitting process to identify the scenarios that need to be considered and those that don’t.  This is a shift in the paradigm away from prescriptive leak sizes and single accidental release scenarios.  

This paper aims to provide an analysis and discussion of the FERC failure frequency arguments grounded in the strength and weaknesses of the underlying databases.  An open question is whether this new criterion provides a reasonable improvement in safety over the prior requirements.  The underlying historic databases and studies do not reflect the major differences on the safety records between the LNG industry and general industry.  More specifically, they do not necessarily take into consideration the effects of process services, management systems, materials of construction, or failure modes on the failure rate data.  Given the basis of generic industry failure rates, the question of whether the QRA approach provides a reasonable approach is also viable.