Liquefied Natural Gas (LNG) facilities are constructed according to “Liquefied Natural Gas Facilities: Federal Safety Standards”1 and regulated by the Federal Energy Regulatory Commission (FERC). FERC has worked closely with the Pipeline and Hazardous Materials Administration (PHMSA), part of the U.S. Department of Transportation (DOT), to provide interpretations and guidelines to meet these regulations. The U.S. federal regulations incorporate NFPA 59A,2 which is a prescriptive standard. The objectives of the U.S. federal regulations and NFPA 59A are to keep the fire and explosion hazards onsite (i.e., within the facility boundaries) in the event of a loss of containment event.
When liquefied, natural gas is a refrigerated cryogenic liquid that boils at -162°C. Spills of LNG from low-source pressures can be conveyed safely to impounding areas or a sump to minimize the size of resulting flammable vapor clouds as the cold liquid boils on the warmer ground. Pressurized releases may produce liquid sprays or flashing jets, which can create larger flammable vapor clouds.
In either case, natural gas vapor clouds are unlikely to produce damaging overpressures if ignited. There have been very few major incidents involving LNG terminals or shipping. The most severe incident occurred in Cleveland, Ohio, in 1944. A more recent incident occurred at the Skikda facility in Algeria in 2004 when the vapors of a flammable refrigerant release were ingested by a steam boiler. The flammable vapor cloud explosion killed 27 workers onsite.
NFPA 59A2 prescribes a series of 10-minute-duration design spills (also called single accidental leakage sources), which must be analyzed to prove that the contour of the ½ LFL methane (i.e., 2.5% concentration on a volumetric basis) vapor cloud does not cross the property boundary that can be built upon.
Practically, the property boundary that can be built upon has been treated as the line beyond which the facility no longer has administrative control. In addition, the radiant heat flux from pool fires within impoundment areas must be shown not to exceed 5 kW/m2 (1,600 BTU/hr ft2) across this boundary. The areas within these boundaries are termed “exclusion zones” where the potential fire hazard exists, and the public cannot be exposed to this hazard.
FERC specifies that only passive mitigation strategies can be applied to meet the exclusion zone requirements and does not allow for active systems to be used to meet the criteria. Thus, shorter duration releases based on detection and emergency shutdown procedures have not been acceptable, even though NFPA 59A does address this option, and such technologies are widely used.
Many of the exclusion zone analysis requirements are stated broadly in NFPA 59A and require considerable interpretation for the spill and leak scenarios that need to be considered. Over the past decade, FERC has clarified its interpretation of the federal requirements by means of formal letters, less formal precedent setting memoranda as well as data requests to specific projects requiring certain analyses to be performed. These interpretations continue to evolve over time, with the continual introduction of new analytical tools and new hazard criteria by FERC.
The 2013 edition of NFPA 59A and recent FERC interpretations, memos, and guidance have introduced risk-based analysis approaches that deviate from the original prescriptive approach in the 2001 edition of NFPA 59A. The new Chapter 15 “Performance (Risk Assessment) Based LNG Plant Siting” is not entirely consistent with traditional Quantitative Risk Assessment (QRA) approaches.3
Risk-based approaches such as QRA assign risk by aggregating the likelihood or probability of scenarios with the consequence in terms of injury or fatality of susceptible populations. QRA is required by the European code.4 FERC requires release scenarios to be selected ac cording to FERC-generated generic failure rates without consideration for the consequence portion of the risk assessment.
FERC has also expanded the consequences to be analyzed to include vapor cloud explosion hazards of flammable refrigerant releases from liquefaction processes, which were not present in import terminals because they only vaporized LNG. Although the changes to the required analyses of fire hazards are complex, some change has been necessary due to the anticipated growth of the U.S. LNG liquefaction infrastructure. This article provides a brief background of changes in the LNG industry in the U.S. and the evolution of current passive fire protection and facility siting guidelines.
LNG IN THE U.S. BACK THEN: IMPORT TERMINALS
The North American LNG industry experienced a surge in growth in about 2003 when the industry believed that existing North American natural gas production was going to be overtaken by increasing demand from power generation, chemical feedstock applications, and domestic use. At that time, there were only four operating LNG import terminals to provide the gas supply: Cove Point, MD; Everett, MA; Elba Island, GA; and Lake Charles, LA.
The worldwide shipment of LNG occurs via ocean-going tanker ships. Import terminals receive the LNG, store it in large cryogenic tanks, and vaporize it into the nation’s gas pipeline network. In 2004-2005, the need to import gas into the U.S. prompted a major effort to develop the terminal infrastructure to receive imported LNG.
At its peak, around 2006, dozens of terminals on the West, East and the Gulf Coasts sought to receive permits for construction. The first to be constructed was the Cheniere LNG terminal in Sabine Pass, LA, and others followed on the East and Gulf Coasts. Unfortunately for the owners, the expected gas demand did not materialize, causing many of the projects to stall and new terminals to remain underutilized.
Import Terminal Fire Protection Considerations
During this time period, FERC applied the 2001 edition of NFPA 59A to identify single accidental release scenarios that needed to be analyzed as part of the application process. Two types of hazardous outcomes were analyzed: radiant heat flux from LNG pool fires and flammable vapor dispersion. LNGFIREIII is the PHMSA-approved software package for modeling LNG pool fires.5 The software calculates the radiation heat flux for LNG pool fires based on a prescribed surface emissive power (SEP) and a cylindrical flame geometry that is based on the impoundment area. FERC recently confirmed that this approach adequately represents the radiant heat from LNG pool fires based on recent large-scale LNG pool fire tests conducted by Sandia National Laboratories.6 LNGFIREIII remains the primary code for calculating heat flux, but the requirements for calculating vapor dispersion have undergone many changes.
In mid-2005, at a time when only import terminals were being considered on U.S. shores, FERC required evaluations of vapor dispersion from full cross-section pipe breaks at the tanker ship unloading line and from high-pressure flashing jets at small-diameter attachments to the transfer piping for instrumentation or pressure relief, at flanges, and at valves or other equipment connections. Based on these requirements, a wide variety of single accidental leakage sources, ranging from valve packing and flange leaks to full cross section ruptures of ship unloading lines, were analyzed by applicants in their FERC submittals.
The primary analytical tool used at that time for the analysis of vapor dispersion was the integral vapor dispersion model, DEGADIS.7 DEGADIS was used to compute the vapor dispersion from evaporating LNG that was spilled into sumps or impoundment areas. The practice was to calculate the source term based on a rate of evaporation that was determined by transient heat conduction from the concrete surface. This vapor source was then input into a code called SOURCE5 that accounted for vapor hold-up within the impoundment area.8 This gave a time delay and a rate of spill of vapors out of the impoundment area, which was input into DEGADIS to calculate the extent of the ½ LFL cloud.
LNG IN THE U.S. TODAY: EXPORT TERMINALS
Around 2010, industry began to develop plans for natural gas liquefaction facilities to export LNG as a result of the natural gas surplus from recent production of natural gas from shale formations. Many of the proposed liquefaction facilities were put forth by previously approved LNG import terminals, the first being Cheniere’s Sabine Pass terminal. It was followed by Freeport LNG and Cameron LNG, among others.
Previously, North America only had one LNG export facility. It was in Alaska on the Kenai Peninsula, approximately 100 km from Anchorage. The Kenai LNG plant began operating in 1969, and was recently taken offline.
Refrigeration processes and the associated plants that are used to liquefy natural gas are considerably more complicated than import regasification terminals. FERC’s limited experience with liquefaction and industry’s rush to develop this new infrastructure forced FERC and DOT (PHMSA) to re-evaluate their requirements. Over a period of two to three years, FERC issued a sequence of new interpretations for required fire and explosion hazard analyses. The most significant changes required a new approval methodology for vapor dispersion software tools, a new method of identifying single accidental leakage sources, and the introduction of vapor cloud explosion calculations for flammable refrigerants.
IMPROVED VAPOR DISPERSION MODEL REQUIREMENTS
An absence of consistent guidelines on the performance of vapor dispersion software prompted a study sponsored by the Fire Protection Research Foundation.9 The final report of this study proposed a formal process for the approval of analytical tools for vapor dispersion at LNG facilities. The resulting Model Evaluation Protocol (MEP) requires prospective models to be compared to a database of spill tests on ground and water, and associated vapor dispersion measurements that were conducted over the past decades. The National Association of State Fire Marshals (NASFM) commissioned an independent review of the MEP to assist local and state emergency response officials.10 This review in part concluded that the MEP was unnecessarily long and complex.
Four years after publication of the MEP, two commercial software products were approved by DOT (PHMSA) in 2011. These were the PHAST Version 6.6/6.711 and FLACS Version 9.112 computer codes. The MEP review process was elaborate, and it took considerable time for the respective software developers to compile their MEP cases and for the regulators to approve them. In addition to DEGADIS, these two software packages are approved for vapor dispersion analyses today. PHAST is commercial software that uses the Unified Dispersion Model (UDM) to calculate vapor dispersion following a two-phase pressurized release or an unpressurized release. It models near-field and far-field jet dispersion, droplet evaporation in the air, rainout (droplets hitting the ground), liquid pool spread, vaporization and subsequent heavy gas dispersion. These features have been essential for analyzing FERC-required pressurized LNG or liquid refrigerant jetting and flashing scenarios. PHAST only accounts for flat ground, and therefore cannot accommodate complex geometries such as tanks, buildings, and walls that are typically present at LNG facilities.
The FLACS software can model vapor dispersion scenarios and vapor cloud explosions in three dimensions. This CFD model discretizes the domain using a rectangular grid. FLACS has a routine called FLASH that can be used to model high pressure jetting and flashing releases. It also contains a liquid spill model to calculate the spread of LNG or refrigerants over the ground. The model calculates heat transfer to the liquid and its evaporation. Currently, FLACS is the only model approved by FERC that can be used to model the vapor clouds resulting from liquid spills into trenches. FLACS also is the only approved model that can be used to determine the effect of structures and vapor fences on the flammable vapor cloud dispersion.
The Latest Single Accidental Leakage Requirements
In 2010 and 2011, FERC’s single accidental leakage scenarios were prescriptive in that the hole size had to be chosen based on pipe size and later pipe length. These criteria were superseded in 2012 by the requirement that single accidental leakage sources be selected for analysis if the likelihood of failure is greater than 3x10-5 failures per year.14 A detailed discussion of the criterion’s development and application is provided elsewhere.14
FERC staff provided a table of yearly failure rates for piping and other equipment. All single accidental leakage sources that need to be considered are now selected based upon the length of the piping system and the resulting failure rate for a given hole size. Once selected, the scenarios are analyzed using the approved commercial software.
This latest change was a paradigm shift from a strict prescriptive approach to one that is based on a probabilistic criterion, even though the consequences remain prescriptive: the exclusion zones must remain within the boundaries of the facility. This paradigm shift constitutes a step closer towards the European Standard,7 which is entirely based on Quantitative Risk Analysis (QRA).
Vapor Cloud Explosion Hazards
Unlike LNG regasification - only facilities, liquefaction plants contain flammable refrigerants in significant volumes. Common refrigerants include chlorofluorocarbons, ammonium, carbon dioxide, and non-halogenated hydrocarbons. In most refrigeration cycles, the mixed refrigerant may include varying concentrations of nitrogen, methane, ethane, ethylene, propane, and iso-pentane.
Some of the refrigerants are generally more reactive than natural gas. That is particularly the case with ethylene, which can undergo vapor cloud detonation. As a result, refrigerants introduce the risks of vapor cloud explosions that did not previously exist with import terminals. The Jan. 19, 2004, Skikda Algeria liquefaction plant accident involved a refrigerant vapor cloud explosion that killed 27 workers.15
NFPA 59A does not address this risk. FERC16 now requires applicants to analyze vapor cloud explosions associated with worst-case flammable gas releases, to identify the 1 psi (7 kPa) over-pressure boundary and to analyze the associated offsite consequences of 1 psi (7 kPa) and greater overpressures.
Passive Mitigation Techniques
Passive mitigation techniques that are often used to contain the ½ LFL cloud within the property include the following:
- Relocation of LNG and refrigerant storage and piping elements to increase the distance to the property boundary.
- Changes to the LNG and refrigerant flow design, by changing the size of piping, capacity and number of pumps, and process conditions. These can reduce the worst case release flow-rate.
- Changes to the refrigerant storage capacity and the amount of refrigerant that can be released.
- The use of vapor fences and other obstacles to contain the LNG and refrigerant vapor cloud during a release.
Terminals have adopted various vapor fence strategies in the past, including long and tall fences, placing fences near the source to reduce its momentum, as well as using short fences to increase turbulence and mixing the cloud with air.
In conditions where an impoundment area, sump, or conveyance trench is located near a property boundary, the extent of the vapor cloud from a spill into this area can be addressed by selecting a concrete mixture that has a low thermal conductivity. Cryogenic liquid spills on concrete evaporate due to heat conduction from the substrate to the cold cryogenic pool. This is the dominant mode of evaporation in the early stage when evaporation rates are at their highest. Therefore, by selecting low density, heat capacity, and thermal conductivity concrete, the ½ LFL clouds can be shortened considerably.
Harri Kytomaa and Trey Morrison are with Exponent, Inc.
- Title 49, Transportation, Part 193, Code of Federal Regulations, Government Printing Office, Washington, DC, 2013.
- NFPA 59A, Standard for the Production, Storage, and Handling of Liquefied Natural Gas (LNG), National Fire Protection Association, Quincy, MA 2001.
- Williams T. and Assing N. “Risk-Based LNG Facility Siting and Safety Analysis in the U.S.: Recent Developments.” 17th International Conference & Exhibition on Liquefied Natural Gas (LNG 17), American Gas Association, Washington, DC, 2013.
- EN-1473, Installation and equipment for liquefied natural gas – Design of onshore installations, British Standards Institution, London, 2007.
- LNGFIRE3: A Thermal Radiation Model for LNG Fires. Gas Technology Institute, Des Plaines, IL, 2004.
- Recommended Parameters for Solid Flame Models for Land-Based Liquefied Natural Gas Spills, Federal Energy Regulatory Commission Office of Energy Projects, Docket No. AD13-4-000, Washington, DC, 2013.
- DEGADIS 2.1: Dense Gas Dispersion Model for LNG Vapor Dispersion, Gas Technology Institute, Des Plaines, IL, 2004.
- SOURCE5, Trinity Consultants, Dallas, TX, 2004.
- “Evaluating Vapor Dispersion Models for Safety Analysis of LNG Facilities Research Project,” Fire Protection Research Foundation, Quincy, MA, 2007.
- “Review of the LNG Vapor Dispersion Model Evaluation Protocol,” Prepared by AcuTech Consulting Group, Alexandria, VA, 2009.
- Phast, Det Norske Veritas, Hovik, Norway, 2011.
- FLACS, Gexcon AS, Bergen, Norway, 2010.
- “Meeting Summary, Corpus Christi Project, Docket No. PF12-3-000, LNG Engineering Conference Call, May 7, 2012,” FERC Document Accession No. 20120507-4014, Federal Energy Regulatory Commission, Washington, DC, 2012.
- Kohout A. “U.S. Regulatory Framework and Guidance for Siting Liquefied Natural Gas Facilities – A Lifecycle Approach.” Proceedings of Mary Kay O’Connor Process Safety Center, 15th International Symposium, College Station, Texas, 2012.
- “Report of the U.S. government team site inspection of the Sonatrach Skikda LNG plant in Skikda, Algeria,” FERC, Washington, DC, 2004.
- Title 40, Environmental Protection Agency, Code of Federal Regulations, Government Printing Office, Washington, DC, 2013.