As shown in the above case studies, the liquid viscosity along with the size of the bubble distribution are key parameters in predicting degassing within this model- ing framework. As the liquid phase viscosity increases, good bubble size management becomes increasingly important in reducing what would otherwise be unavoidably long separator residence times required for good separation. For non-supersaturated solutions, increasing the bubble diameter mode by even a few millimeters could make
the difference between adequate and poor separation. Inlet conditioning devices that increase the total size of the inlet bubble distribution have the potential to signifi- cantly improve separation performance. Inlet conditioning devices that serve primar- ily to increase the inlet bubble density closer to the bulk gas-liquid interface, on the other hand, are unlikely to improve separation performance by a significant amount.
Other than the overall size of the bubble distribution, the sensitivity analysis re- sults indicate that the liquid viscosity, not the liquid density, is the main driver for separation performance.
Traditional horizontal gas-liquid separators were found to be ill-suited for sep- arating excess solution gas in supersaturated solutions. The driving force for mass transfer will increase the rate of entrained gas separation, however, removing the excess solution gas is more challenging. Liquid residence times of several hours are required to remove the majority of excess solution gas for the base case condition. To remove more than 50 % of the excess solution gas from a liquid with a supersaturation ratio ofξ= 0.3, the entrained gas fraction has to approach 50 % of the liquid volume and the bubble distribution mode must be less than 1.5 mm. The liquid swelling due to bubble growth during supersaturated conditions was, however, not found to be significant for a range of different supersaturation ratios. The result also highlight the need for bubbles in solution to provide additional interfacial area required for adequate excess solution gas separation. For conditions where the starting bubble sizes are large, the total volumetric interfacial area drops off quickly, ensuring the rate of gas-liquid mass transfer similarly decreases. If solution gas is present in an incoming gas-liquid stream, it could be advantageous to divert the stream to a gas- liquid contactor and removed as much of the solution gas as possible prior to entering a gas-liquid separator.
CHAPTER IX
CONCLUSION
The goal of this study was to explore gas evolution in the context of gas- liquid separation. An new experiment was developed to measure gas evolution in hydrocarbon systems at high pressure. The effect of mixing and liquid viscosity was explored using methane and air in model oils. Additionally, the rates of absorption and desorption (gas evolution) were measured and compared toone anotherusing a reference methane-dodecane system. Once mass transfer was quantified for this ref- erencesystem,the experimentalresultswereusedtoconstructahorizontal separator degassing model that calculates gas carry-under due to both entrained bubbles and excesssolutiongas.
9.1 Experimental
Developing a new experiment capableof measuring gas evolution at pressure was the primary objective of this study. The initial experimental design utilized a rapid depressurization step to generate excess solution gas within the liquid. This method was soon abandoned due to the inability of the experiment to generate an approximate step change in the pressure as the total experimental pressure was in- creased. The experiment was then modified in favor of a gradual depressurization technique. Usingadual-cylindersyringepump,asaturatedliquid wasinsteadgradu- allydepressurized,supersaturatingtheliquidintheprocess. Fromthesupersaturated solution,gasevolutionwasinitiatedbyturningonthepressurevesselmixerandmea- suringthe resulting increasein pressure.
Using the developed gas-liquid mass transfer experiment, rates of gas evolution were measured in a reference methane-dodecane hydrocarbon system. The saturation pressure of the experiment was varied from 500 to 1,500 psia (3.45 to 10.3 MPa), while the mixing speed was varied from 100 to 250 rpm. The maximum mixing speed was limited to values that maintained a flat gas-liquid interfacial area, allowing the area available for mass transfer to be quantified for all trial conditions.
In order to ensure that the measured gas evolution rates were not significantly effected by bubble nucleation, both the rates of absorption and desorption were mea- sured for each trial condition. Within the measurement error, the absorption and desorption mass transfer coefficients were found to be symmetric. The symmetry between the two mass transfer coefficients confirms that bubble nucleation was not significantly affecting the measured gas evolution rates. All measured absorption and desorption mass transfer coefficients were within 17% of one another for the same trial conditions. The mixing speed was found to be the most significant variable affecting the rate of mass transfer while the saturation pressure within the range tested here had minimal effect.
In an attempt to generalize the mass transfer results beyond the stirred tank experimental setup, theoretically derived mass transfer expressions were evaluated for their ability to quantify the data. The surface renewal theory in the form of the small eddy model was found to be a good fit to the experimentally measured data. The solid and fluid surface eddy cell models were applied to the experimental conditions and resulted in a reasonable fit for both cases. The solid surface model was found to better fit the experimental results, yielding an averaged absolute error of 12.3%.