Apr-2009

Demystifying the zero entrainment myth for vacuum column flash zones

One of the recurring questions about refinery vacuum tower operation is the issue of flash zone entrainment and the operation of the wash section.

Mario Roza, Markus Duss, S M Wong and Mark Pilling
Sulzer Chemtech

Article Summary

As refiners push to optimise yields from the vacuum tower, limits of existing column internals are often encountered. Although the operation of the vacuum tower needs to ultimately be judged as a whole, this paper will focus on the flash zone and wash sections of the column since they are critical for high yield operations.

The flash zone serves to transition the high velocity two-phase feed from the transfer line into the vacuum column in such a manner that separates the liquid and routes it to the bottom of the column while delivering the vapour uniformly to the upper sections of the column. The feed nozzle orientation into the column can be radial or tangential. The number of nozzles can vary, depending on the column sise and heater configuration. The typical feed internal for a vacuum column is a vapour horn or other device that uses the feed inertia to redirect the stream to contact and remove dispersed liquid particles. The liquids entrained in the upward flowing feed vapour stream need to be removed because they contain disproportionately high amounts of heavy end contaminants such as metals and hydrogen deficient molecules. The metals are catalyst poisons for downstream processes. The hydrogen deficient molecules tend to form coke and also adversely affect the distillate end point and colour.

The flow configuration of the bottom of a typical vacuum column is shown in Figure 1. Above the flash zone is a relatively short packed bed called the wash section. The wash section is designed to remove heavier components in the upward flowing vapour from the flash zone by coalescing entrained liquid droplets and by condensing the heavier vaporised components. The packing provides the surface area that coalesces the entrained liquid droplets in the vapour stream. A small liquid gas oil stream is fed to the top of the wash section which serves as a kind of reflux stream and also to wet the packing to prevent it from drying out and coking. It also aids the removal of the coalesced liquid from the packing surface. Liquid leaving the bottom of the wash bed is collected in a slop stream. It consists of coalesced liquid entrainment from the flash zone, the heavy condensed components, and the heavy portion of the liquid gas oil feed that made it down through the packed bed without being vaporised. Since gas oil is more valuable than slop, process economics dictate that the gas oil feed stream to the wash section be minimised. However, the stream must be maintained at a sufficient rate so that coking is avoided. From a column product draw standpoint, as the entrainment rate to the wash bed increases, the slop draw should increase by the same amount but the contribution of the slop from the liquid feed to the wash section will remain essentially constant.
 
Optimising the Flash Zone & Wash
Section Design
When trying to optimise the feed and wash sections of the vacuum column, the goal must be clearly understood. From an operational standpoint, the true goal is to provide the best possible vapour feed to the distillation section above the wash section. The term, “best possible”, means the lowest possible amount of entrained liquids as well as the most uniform vapour distribution. Again, this is measured above the wash section. However, since the intermediate conditions between the flash zone and the wash section can affect the output of the wash section, we need to understand that region as well. Therefore, we need to look at the vapour distribution throughout these sections as well as the corresponding entrainment.

Looking first at vapour distribution leaving the wash section, this should be fairly straightforward. In order for the wash section to function properly with respect to de-entrainment and coking resistance, the vapour feed to the wash section must be very well distributed. Since a packed bed is typically a good distributor of vapour, the vapour flow from the top of a properly working wash bed, by definition, must be excellent. Therefore, when discussing vapour distribution, we need to focus on the region between the flash zone and the wash section. 

Now looking at entrained liquids, from a practical standpoint, the flash zone section behaves as a rough vapour-liquid separator and the wash section behaves as a polishing bed where droplets can be removed and fractionation takes place. Generally, it doesn’t matter how much liquid is removed in the feed section or the wash section independently; it matters how the combined feed and wash section perform as a unit. To some extent, the de-entrainment effects of the flash zone and the wash section are cumulative. In a simplistic approach, if both sections operate at 90% efficiency, the net efficiency will be approximately 99%. For example, if 100 kg of entrainment enter the tower, a 90% efficient flash zone will allow 10 kg to leave and go to the wash section, which will capture 90% of that entrainment, allowing 1 kg of entrainment to leave to the tower above. However, the efficiency of both sections is based on droplet particle sise. If the flash zone or the feed inlet device allow a bigger portion of large diameter droplets to escape to the wash section, the wash section will be extremely efficient in removing those droplets. However, if a larger percentage of very small droplets leaves the flash zone, the wash section will be less effective in removing that extra entrainment.

Nevertheless, it’s desirable to minimise the entrainment from the feed section to the wash section. However, when this comes at the expense of poor vapour feed to the wash section, improved de-entrainment from the feed section usually does not result in a better product from the wash section. The following CFD study provides a review of an industrial column configuration with two types of feed devices. From this study, it can be seen that different feed design can have unexpected results with respect to vapour distribution and de-entrainment capabilities.

CFD Study: De-entrainment Versus Vapor
Distribution
When studying the performance of various vacuum tower feed devices, it becomes apparent that vapour-liquid separation capabilities (entrainment removal) and vapour distribution quality to the wash section do not necessarily go hand in hand. In other words, a modification that decreases the entrainment to the wash section doesn’t necessarily help the vapour feed profile to the wash bed. Often, the opposite can be true. This can be seen with CFD simulations shown in Figures 2 & 3. Figure 2 shows a pure cyclonic device with a strong rotation that optimises de-entrainment. Figure 3 shows the vapour velocity profile leaving this device at an elevation just below the packed bed. As can be seen, the feed to the wash section has some distinct high velocity regions. However, from an entrainment standpoint, we know from in-house testing that a well designed feed device can have very low entrainment rates of less than 1%.¹

Figure 4 shows a vapour horn type feed device with baffles and a vortex recovery mechanism to limit the swirling effect in the bottom of the column. The CFD results show that this is not as effective as an initial de-entrainment device as the pure cyclone. When operating at typical industrial conditions, a well designed vapour horn feed device like this can have entrainment levels as low as 2%. This is still relatively low, but higher than that of the cyclone. Looking at the vapour distribution results in Figure 5, the vapour horn device clearly provides a more uniform vapour flow to the wash section above. It should be noted that the entrainment rates discussed above have been measured in laboratory conditions with column diameters of 1-3 meters. Entrainment rates from well designed industrial columns are typically in the range of 3-5%. So there is some scaling effect that is seen. However, the trend of entrainment versus vapour distribution should still hold true, regardless of scale.
            
Vacuum Tower Example
The example below shows the performance of an industrial column with a less than optimum wash section design. With this case, it can be seen that even these older, less efficient designs can provide excellent de-entrainment removal.