Impact of Column Geometry and Internals on Gas and Particle Flows in a Fluidized Bed with Downward Solids Circulation

Fluid Coking^TM is a process employed by the petroleum industry for thermal conversion of heavy hydrocarbon molecules, such as bitumen from oil-sands, into distillate products. It is particularly important in Canada as synthetic crudes produced in Fluid Cokers represent about 15% to 20% of total oil production. In Fluid Cokers, heavy oil is upgraded into synthetic crude oil through a thermal, non-catalytic cracking process. Industrial Fluid Cokers are composed of two vessels, a fluidized reactor and a burner: in the reactor, heavy oils are sprayed on hot coke particles where it cracks to solid coke and vapors that leave the reactor from the top. Coke particles flow down from the spray region through a stripping section where hydrocarbon vapors are displaced by steam. Cold coke is then conveyed from the bottom of the reactor to the burner where it is reheated by combustion before reinjection in the reactor. A variation of the Fluid Coking technology, FlexiCoking, uses a third vessel to gasify coke.

A major operability issue in industrial Fluid Cokers is the fouling of stripper sheds in the lower reactor section. This fouling is attributed to wet particles and agglomerates that move too quickly from the liquid spray zone to the stripper section, thus carrying unreacted liquid to the stripper section. To minimize the presence of unreacted liquid in the stripper section, Fluid Cokers are currently operated at temperatures that are significantly higher than the optimum temperature for the conversion of bitumen to the desired liquid products, leading to overcracking and a lower yield of valuable products. Therefore, the objective of this study is to determine how reactor geometry and operating conditions can be modified to minimize the presence of unreacted liquid in the stripper section. Different aspects of Coker design and operation are studied, such as column geometry, number and location of spray nozzle banks; and number and location of baffles.

The study is conducted in a cold recirculated fluidized bed, with a circular cross-section. The bed of coke particles is fluidized with air and operated at room temperature. The hydrodynamics of Fluid Cokers are greatly influenced by the hydrocarbon vapors evolving from the lateral injections of heavy oil feedstock, at various heights. In this study, the vapors are simulated with air injection, distributed over 5 banks of 8 nozzles each. The design of the bed was chosen to match key parameters of industrial Fluid Cokers, such as the heights of the 5 nozzle banks, the diameter/bed height ratio, and the solids recirculation flowrate. Since industrial units are often tapered, to moderate the increase in superficial gas velocity caused by the feedstock injections, two different, typical taper shapes were studied.

Ring baffles have been proposed to reduce stripper shed fouling, by minimizing the bypassing of wet particles from the injection zone to the stripper region. The impact of ring baffles on the Residence Time Distribution (RTD) of wet agglomerates and particles was, thus, studied with different column geometries and feedstock distribution schemes. Studies were also conducted to evaluate the potential fouling of ring baffles.

The main experimental tool of this study is non-intrusive Radioactive Particle Tracking. A radioactive tracer is encapsulated to achieve the required size and density. The unit is surrounded by 12 scintillation detectors at known locations. For each experiment, the system tracks a single tracer during a long time (18 to 60 hours). These long recording times allow statistically significant analysis of the data to extract information about agglomerates or particles behavior. Geiger counters at the solids inlet and outlet also provide the solids residence time distribution for the whole bed. Radioactive Particle Tracking measurements were analyzed to provide the following information:

• The overall residence time distribution (RTD) of the tracer. These results were compared to the results obtained by injecting dyed coke particles in the solids inlet and sampling particles from the solids outlet. Because, in this system, dyed particles sampled from the solids outlet may have gone more than once through the column, the dyed particles experiments cannot provide the true RTD. In this study, they were used to determine which radioactive tracer best simulates individual particles.
• The distribution of time-spans between the agglomerate formation zones (located in front of each lateral injection) and the stripper. This information is essential to develop strategies that will reduce stripper fouling in Fluid Cokers.
• The flow patterns of agglomerates and particles in the Coker model. This information helps design more effective baffles and nozzle positioning.
• The flow patterns of the gas bubbles. These results were compared to measurements performed with invasive triboelectric probes. This information provides information on vapor flows in cokers.

Experiments to date have shown that the addition of ring baffles, i.e. one baffle below each of the 4 top lateral injection banks, decreases the proportion of short residence time over the whole bed, from the solids inlet to the column to the solids outlet from the column to the recirculation line. For instance, the proportion of tracer going through the whole bed in less than 10 s would decrease from about 7 to 5 % with the addition of the baffles. However, the baffles increase the proportion of agglomerates that go from the zone where agglomerates are produced to the stripper in less than 10 s, in average over the 5 banks, from 14 % to 20 %. Further experiments will investigate how the number, location and geometry of the baffles can be modified to enhance Fluid Coker performance.

Preliminary results have shown that redirecting all the liquid that normally flows to the lowest nozzle banks to the other 4 nozzle banks reduces the proportion of agglomerates that go from the zone where agglomerates are produced to the stripper in less than 10 s. Further experiments will confirm these results for different baffle configurations for publication at the conference.

Future studies, to be presented at the conference, will investigate the effect of changing the column taper, when adjusting the feed distribution between the various nozzle banks. The impact of baffles with a different column taper will also be studied.

Additional work is ongoing to better understand the phenomena at play through the analysis of tracer trajectories and bubble flow patterns: for example, it seems that the right combination of baffles and injection banks promotes the formation of zones where wet agglomerates that would otherwise move too quickly are captured and held for a while before being released.