Abstract
Green ammonia (NH3) produced by hydrogen from water electrolysis using renewable electricity is considered to have an important role in the ambitions to decarbonize the maritime sector. With a boiling point of -33.4 °C, much closer to ambient temperatures than LH2, NH3 is less complicated to handle and to store. While the low flammability of NH3 represents a challenge as a fuel, this is an advantage regarding handling, and the volumetric energy density of NH3 is almost 50% higher than for LH2. A disadvantage with NH3 is its toxicity and odour, already at 5-50 ppm most people will recognize its characteristic odour, and at concentrations of 1% in air NH3 may be fatal within minutes. For a successful introduction of NH3 as a maritime fuel, safe handling is critical. To shorten the construction time and increase flexibility related to scale-up to meet possible future increasing demands, AZANE Fuel Solutions has developed a bunkering concept with NH3 stored in floating barges permanently moored at the bunkering site. With partial support from the Research Council of Norway the bunkering concept has been further developed, one of the key tasks was to ensure a best possible safety level of the terminal. To achieve this the NH3 is stored refrigerated near its boiling point with all penetrations on top and redundant cooling systems. The tank is further well protected against impact with B/5 separation from the hull of the barge. For the siting studies the NH3 toxicity is the primary concern, and as the terminal will store more NH3 than 200 tonnes, the Seveso-III upper tier major accident legislation applies. To develop required risk contours for the approval, all possible loss of containment scenarios must be evaluated considering expected leak rates and durations from statistical rupture and hole size frequency distributions. These scenarios must be evaluated considering the actual geometry and terrain and actual distribution of wind speed and direction. The toxic dose fatality probability contribution must then be assessed using probit functions considering exposure concentration and duration for each leak scenario, wind direction, wind speed and distance. In addition to defining risk contours temporary safety zones during bunkering must be established using a consequence-based approach estimating the maximum distance to 1% fatality risk in case of hose rupture with expected safety systems working as intended. The leak and dispersion dynamics of liquid NH3 strongly depend on storage temperature and pressure. For storage near the boiling point there is no overpressure to push liquid NH3 out of the tank through top penetrations, while with increasing temperature and saturation pressure a larger fraction of the tank content can be pushed out as liquid NH3 in the tank will boil when the pressure reduces. Leaks of refrigerated and semi-refrigerated NH3 (< 0 °C) will generally form buoyant plumes. Evaporation rates will be limited by heat transfer and cooling of substrate. When spilled onto water a major fraction of liquid NH3 will dissolve in water, while the spill dynamics will decide whether the evaporation of the remaining 30% or more is immediate or spread over minutes. For releases of warm liquid NH3 (> 10 °C) most NH3 is expected to flash-boil and form denser than air plumes with extensive hazard distances. The complexity of the NH3 siting studies is high. Even with leaks grouped into 15 scenarios of primary concern, the wind conditions simplified to 16 wind directions and 4 wind speed categories, consequences from 1000 scenarios would need to be assessed, considering transient concentration predictions at 100 different distances from leaks with duration of seconds, minutes or hours. Only a limited number of scenarios is feasible to model in detail using computational fluid dynamics (CFD). If simpler plume models are used for dispersion, important geometry details near the site and terrain will generally be ignored. The article will describe the assessment and illustrate how a limited use of CFD modelling can help improve the precision of the study.
