If LNG is spilled, it vaporizes. The natural gas vapors are initially heavier than air and they form a cloud close to the ground, which is pushed downwind and eventually dissipates. If a viable ignition source is present where a vapor cloud exists at a 5%–15% concentration in air, the vapor cloud can ignite and burn. Therefore, the permitting of LNG terminals requires thermal radiation and vapor dispersion hazard distances to be quantified and demonstrated to pose no threat to people or property outside the plant.
Exponent has considerable expertise in modeling LNG flammable vapor cloud dispersion, using both integral models (DEGADIS, SLAB) and CFD models (Fluent). Recent discussions on the fate of LNG spills into impoundments have led to the conclusion that the commonly used combination of SOURCE5 and DEGADIS to predict dispersion distances of flammable vapor is not physically accurate, because it does not account for air entrainment into the evaporating gas, and it does not allow for heating of the LNG vapor cloud within the impoundment. An alternative approach to predicting the flammable vapor dispersion distance is to use a computational fluid dynamics (CFD) model. As an industry-sponsored project, Exponent is in the process of completing the validation of Fluent, a widely used commercial CFD code, to predict flammable vapor cloud dispersion from an LNG spill into an impoundment, using experimental data from the Falcon Test Series (Livermore National Labs 1982).
Comparison between CFD model (left) and Falcon Test 5 video image (right) of visible vapor cloud when it begins to overflow the impoundment after approximately 12.5 seconds (Lawrence Livermore National Laboratory)
Comparison between CFD model (left) and Falcon Test 5 video image (right) of the visible cloud approximately 30 seconds into the spill, showing the depression in the visible cloud at the center of the impoundment, which remains even after the cloud has begun overflowing the downwind and crosswind edges of the impoundment (Lawrence Livermore National Laboratory)
The possibility of large releases from LNG tankers has raised the need to analyze larger spills than have been investigated previously. Models have been developed to address these scenarios; however, these models are based on the extrapolation of small-scale experimental data and/or the adaptation of light hydrocarbon spill models. The lack of experience with large LNG spills continues to challenge the engineering community to develop better fire and atmospheric dispersion models.
Exponent staff is active in LNG safety-related research and scholarly activities. As part of an industry-sponsored project, Exponent staff developed an experimental facility to perform controlled and repeatable experiments to study the behavior of LNG spills on water. Currently available models for the behavior of a large LNG spill over water are based on simplified assumptions and coarse estimates of the relevant parameters. Detailed experimental data on the behavior of cryogenic-liquid spills on water are sparse, inconsistent, and inadequate. The spreading of an LNG pool over water is a dynamic process, influenced by the following factors:
-
The spreading speed of the leading edge of the LNG pool
- The variable thickness of the LNG pool as it spreads
- Friction between the LNG pool and water
- The evaporation rate of LNG on water
Exponent has provided quantitative data and improved physical models for the factors described above.