Optimization of Packed Tower Inlet Design by CFD Analysis. Dana Laird Koch-Glitsch, Inc.

Transcription

1 39e Optimization of Packed Tower Inlet Design by CFD Analysis Dana Laird Koch-Glitsch, Inc. Brian Albert ExxonMobil Research and Engineering (formerly with Koch-Glitsch, Inc.) Carol Schnepper John Zink Co. Prepared for presentation at the 2002 AIChE Spring National Meeting, New Orleans, LA, March Dana Laird and Brian Albert, Koch-Glitsch, Inc, and Carol Schnepper, John Zink Co. AIChE shall not be responsible for statements or opinions contained in papers or printed in its publications.

2 Introduction The feed to crude vacuum towers is a two phase mixture that is introduced to the tower at very high velocity. The inlet must be designed to perform initial separation of the vapor and liquid and to distribute the vapor to the vacuum tower wash bed. Numerous devices have been employed, many developed through trial and error in refining vacuum towers. This approach is costly and time consuming. Koch-Glitsch initially did extensive testing of vapor horns in our 8-0 I.D. air/water simulator column. The cold flow testing was combined with the results of early commercial applications to develop the current vapor horn configuration. While simulator testing was indispensable to the development of vapor horns, it can not easily show the interaction of the vapor horn with the other column internals. Computational Fluid Dynamics (CFD) models can provide a more complete picture of the vapor horn performance by utilizing the actual tower configuration and process conditions in the analysis. Potential design improvements can be evaluated quickly using this approach. Evaluation Basis The CFD analysis compares the performance of two vapor horns in a 29 6 crude vacuum tower operating at a C-Factor of 0.35 at the bottom of the wash bed. The tower was modeled with a single 66 tangential inlet. A simplified collector was built into the tower model to simulate the effect of the overflash tray on the vapor distribution to the packed bed. The collector was modeled with 30% open area and would be expected to take approximately 0.75 in H 2 O pressure drop. The tower elevation is shown in Figure 1. The geometry of the over flash collector is depicted in Figure 2 and Figure 3. The first vapor horn represents a typical industry installation. This simple geometry, depicted in Figure 4, consists of a 90 vapor horn with no enhancements to facilitate vapor-liquid disengagement or vapor distribution. The device also does nothing to impede the natural rotational motion imparted by the tangential entry. Both of these weaknesses were identified with earlier air/water testing. As a result of these earlier tests, Koch-Glitsch developed enhancements to the simple vapor horn design. The patented Enhanced Koch-Glitsch Vapor Horn, shown in Figure 5, has a series of internal baffles designed to enhance the separation of liquid from the vapor in the feed. These baffles also improve the vapor distribution provided by the horn by "slicing" off segments of vapor as the incoming stream travels around the horn rather than allowing all of the vapor to reach the end of the horn. The design also incorporates a series of anti-swirl vanes, which project into the annular space of the vapor horn. The vanes break the swirling motion of the vapor induced by the tangential motion in the vapor horn. The vapor exits the bottom of the vapor horn and turns up through the annulus. This vapor has both vertical and radial velocity. It is crucial to break the swirling motion of the vapor before it enters the collector tray and wash zone packing above. High radial velocity at the wash bed inlet adversely effects vapor distribution to the bed resulting in localized dry out and poor wash bed reliability. Velocity profiles were compared throughout the entire three dimensional space from the bottom of the vacuum tower to the bottom of the packed wash bed. For the purposes of this discussion, four key horizontal planes are illustrated. These planes are located above the collector just below the wash bed, between the collector and the vapor horn, through the centerline of the feed nozzle, and below the vapor horn in the lower conical transition. Figure 6 illustrates the relative location of these horizontal planes. In all velocity profile figures referenced in the discussion below the feed enters on a horizontal tangent from the upper right of the vessel.

3 Results The vertical vapor velocity at the plane below the collector can be seen in Figure 7 for the simple vapor horn and Figure 8 for the enhanced vapor horn. Both horn designs have areas of low vertical velocity (< 3 ft/s). However, with the simple vapor horn the area of low velocity is much larger and bisects the column diameter. In addition the simple vapor horn design creates an area of extremely high vertical velocity (> 60 ft/s) at a point located about 270 from the entrance. By comparison, the maximum velocity above the enhanced vapor horn is approximately 30% lower and the areas of low velocity are more evenly distributed across the column. The uniformity of vertical vapor velocity at the plane above the liquid collector can be seen in Figure 9 and Figure 10. For both designs the velocity varies from approximately 3-13 ft/s. However, with the simple vapor horn the velocity profile is biased much more to one side of the column bounded on the outside with an annulus of low velocity. In addition, the most concentrated area of high velocity has moved to the other side of the tower from the plane 80 inches lower. With the enhanced vapor horn design the velocity profile is much more evenly distributed across the tower with the areas of low and high velocity corresponding almost entirely to the velocity distribution expected from the design of the chimney tray. This is demonstrated by Figure 11, which shows the vertical velocity at the same plane when the collector is fed with a uniform velocity profile. The enhanced vapor horn significantly reduces the rotational velocity in the flash zone. This is shown in Figure 12 and Figure 13, which show the magnitude of the rotational velocity in the plane below the collector. The high rotational velocity caused by the simple vapor horn produces an effect similar to a tornado in the flash zone, increasing entrainment. In the case of the simple vapor horn the rotation continues even above the collector tray as shown in Figure 14. The enhanced vapor horn significantly reduces the rotation above the collector tray as demonstrated in Figure 15. The vapor distribution action of the internal vanes can be seen in Figure 16 through Figure 19 which show the vertical velocity in the plane through the centerline of the feed nozzle and the plane below the nozzle. This lower plane is in the conical transition below the column flash zone so the velocity profile does not cover the complete cross section. These figures clearly show the advantage the enhanced horn has in establishing initial vapor distribution, critical to minimizing entrainment. As shown in Figure 16 with the simple vapor horn there is a large region at the nozzle centerline where the local velocity is approximately 19.5 m/s (64.0 ft/s) which translates to a local C-Factor of about 1.2. Although there is a region where the peak velocity reaches the same magnitude with the enhanced horn the area is much smaller. In addition, the velocity profile is much more evenly distributed across the center of the vessel rather than being concentrated on one side. The velocity contours below the horn are important because they indicate a region of extremely high velocity at the vessel wall where the vapor will tend to reentrain liquid that is running down the sides of the vessel. With the enhanced horn the peak velocity below the horn is much lower and the region of high velocity is smaller. This results in less reentrainment of liquid that has already disengaged from the vapor. Conclusions All of the figures illustrate the importance of the feed device in establishing uniform flow to the wash bed. Despite the presence of a collector tray that has a pressure drop typical of vacuum tower installations flow irregularities persist at the entrance to the wash bed. This non-uniform flow can translate to premature flooding, poor wash zone performance, and poor wash bed reliability. Large variations in vapor velocity entering the wash bed will almost certainly result in bed coking since

4 sections with high vapor velocity (therefore high vapor mass flux) will have a greater tendency to dry out. Shortened run lengths and expensive bed replacement will be the result. This work also illustrates the benefits of CFD for designing vacuum tower inlet devices. Since each tower is different the interaction between the feed device, transfer line, vessel shell, and other internals will vary. CFD offers a way to analyze these interactions and optimize the feed device design for a given unit. Used in conjunction with historical operating and inspection data CFD can help to eliminate chronic problems due to feed maldistribution from a poorly designed feed device. Figure 1 Tower Elevation Figure 2 - Collector Underside Figure 4 - Simple Vapor Horn Geometry Figure 3 - Collector Cross Section

5 Bottom of Packing Support, 155" Plane Above Collector, 153" Collector, 113" Plane Below Collector, 73" Nozzle Centerline, 0" Plane Below Vapor Horn, -105" 19' 0" Figure 5 - Koch Proprietary Vapor Horn Figure 6: Elevation of Planes Shown for CFD Analysis Figure 7 - Simple Vapor Horn Vertical Velocity Figure 8 - Enhanced Vapor Horn Vertical Velocity Figure 9 - Vertical Velocity Contours Above Collector - Simple Vapor Horn Figure 10 - Vertical Velocity Contours Above Collector - Enhanced Vapor Horn

6 Figure 11 - Vertical Velocity Contours Above Collector with Perfect Initial Distribution Below Figure 12 - Rotational Velocity Below Collector - Simple Vapor Horn Figure 13 - Rotational Velocity Below Collector - Enhanced Vapor Horn Figure 14 - Rotational Velocity Above Collector - Simple Vapor Horn Figure 15 - Rotational Velocity Above Collector - Enhanced Vapor Horn

7 Figure 16 - Simple Vapor Horn - Vertical Velocity at Nozzle Centerline Figure 17 - Enhanced Vapor Horn - Vertical Velocity at Nozzle Centerline Figure 18 - Simple Vapor Horn - Vertical Velocity Below Nozzle Figure 19 - Enhanced Vapor Horn - Vertical Velocity Below Nozzle