Aromatics Column

In his book Random Packings and Packed Towers, Design and Applications R. Strigle (Gulf, 1987) describes a heavy naphta reformer aromatics column separating the toluene and lighter components from the xylenes and heavier aromatics. The column is 6 ft 6 inches in diameter and contains 60 valve trays (of a type unspecied) that are 20 inches apart. The column is being considered for retting with packing. The feed consists of benzene through C-10 aromatics and the column operates at a top pressure of 21.4 psia. The feed is superheated and undergoes a 12-weight % ash on entry to the column. Strigle states that the distillate ow is 18,580 lb/hr containing 0.40 wt % xylenes, the bottoms ow is 43,170 lb/hr with 0.25 wt % toluene, and the reux ow is 70,600 lb/hr. Strigle reports that the column is equivalent to 42 theoretical stages. The feed location is not specied but when discussing the packed column revamp Strigle states that the rectifying section has 25 theoretical stages and the stripping section has 22. This is more than the 42 stages mentioned above but we have used the ratio of 25/47 to determine where the relative feed location in all of our calculations with this system. With this information we are able to conduct a preliminary simulation based on the equilibrium stage model in ChemSep. We assumed constant pressure operation (at the top pressure of 21.4 psia specied by Strigle). Thermodynamic properties were calculated from the Peng-Robinson equation of state. For simplicity the feed pressure was assumed to be the same as the column pressure and the feed vapor fraction was specied to be 12 mole percent rather than 12 wt % that Strigle reports (This was done in order to avoid carrying out a number of simulations to nd the correct vapor fraction in molar units because at present ChemSep does not permit the user to specify the feed vapor fraction in weight units). Specifying the bottoms ow rate in lb/hr and the reux ratio (calculated from the reux and distillate ows given above) completed the column specications. The feed and product ows from this calculation are shown below. Numbers shown as 0 are less than 0.001 lb/hr. Feed specications and product compositions Component wt% Feed Distillate Bottoms Benzene 0.3 1 0 n-Heptane 0.6 2 0 Toluene 25.5 84.6 0.1 n-Octane 4.2 12.4 0.7 p-Xylene 9.9 0 14.2 m-Xylene 26.7 0 38.2 o-Xylene 19.2 0 27.5 n-Nonane 1.3 0 1.9 Indene 12.0 0 17.2 Naphthalene 0.3 0 0.4 Total mass ow (lb/h) 61750 18580 43170

The liquid composition proles are shown below. 1

(a) Liquid mole fractions proles equilibrium column

(b) The McCabe-Thiele diagram between the key components Toluene

We now model the column as it really was: with 60 valve trays. We know only that the column was 6.5 ft in diameter. Design specications and options used here are summarized in the table below. All other details of the trays were left for ChemSep to determine using its built in design procedure. The material balance from this simulation is very similar to those obtained from the equilibrium stage calculation (as, for this kind of system, they should be). The only difference is that the n-Octane percentage is 12.3 instead of 12.4 mole%. The tray design produced by ChemSep is summarized below (recall that Strigle provided no details of the tray design so ChemSep was allowed to design them as part of the simulation). The composition proles also dont look very different from their equilibrium stage counterparts (again, we should be surprised if there was a signicant difference). Valve tray models and computed layout Mass transfer coefcient Liquid phase resistance Vapour ow model Liquid ow model Pressure drop Entrainment Column diameter (ft)

AIChE Included Plug Plug Estimated None 6.5

K closed (-) Eddy loss coefcient (-) Valve density (lb/ft3)

2.76004 4.26509 25751.3

Tray spacing (ft) Number of ow passes Liquid ow path length (ft) Tray spacing (ft) Active area (%total) Total hole area (%active) Downcomer area (%total) Hole diameter () Hole pitch () Weir type Weir length (ft) Weir height () Downcomer clearance () Deck thickness () K open (-) Weight ratio valve (-) Valve thickness (ft)

2 2 2.23 1.7 76 15 12 0.0156 0.03842 Segmental 11.325 2 1.5 0.1 0.401854 4.23228 0.00984252

The fact that the trays do not operate at equilibrium is emphasized in the plot that shows the Murphree efciencies. The average efciency in the rectifying section is found to be around 65% and approximately 80% in the stripping section. The trayed column easily meets both product specications and is also well balanced: the top product contained 0.51 mole% C4s and the bottoms 0.48 mole% C3s. From the tray efciency plot we see that the tray efciencies are around 2

90 %, which is normal for this kind of operation. Inspecting the fraction of ooding prole we observe that the bottom section is limiting the capacity of the column as it peaks at tray 20 (from the top). Turndown is a factor 2.7 for the trays in the rectifying section versus about 1.9 for the trays in the stripping section. This can be improved by using different tray layouts in the rectifying and stripping section. Observe that combining the two sections caused the design method to come with a tray layout with a maximum of 83% of ood. But the actual column is only 5 ft in diameter, not 5.37 ft! This implies that at this reux ratio the stripping section would be ooding (83% 5.372 /52 = 96%) and can explain why Strigle used a low tray efciency of just 14/22 = 64%. When the reux ratio is lowered down to 0.92 the 5ft double pass sieve tray can operate at around 90% of ood (stripping section) and tray efciencies remain high at 90+% resulting in 1.94 mole% C3 in the bottoms and 0.67 mole% C4 in the distillate. Strigle rets his column with #40 IMTP packing which is not available in ChemSep and we have selected an alternative. If we pack the 102ft tall tower with 35mm NORPAC packing, ChemSep estimates that a column only just over 6 feet in diameter is needed while delivering a slightly improved separation. Our column, however, is 6.5 ft in diameter. So the next step is to specify the actual diameter. The simulation clearly shows that the column now is underperforming as there are a great many warnings that the packing in insufciently wet. It does, however, converge to give a column with a signicantly improved separation. This allows us to (again following Strigle) cut the reux ratio from 3.8 to 3.2 with still a signicant improvement in performance over the original tray column design. Clearly, therefore, we can increase column throughput should we wish to do so. Strigle, in fact, proposes to increase the feed ow rate by 30%. We can do this easily in ChemSep by double clicking on the total ow cell in the feed ow panel (to make sure the contents of the cell are not highlighted) and typing *1.3 after the number that is in that cell. Click out of the cell and the calculation is done and the new ow rate recorded. It is important also to change the desired bottoms mass ow rate by 30% as well. ChemSep converges this new design quite easily with no warnings indicating the hydrodynamic design is satisfactory. The stream table for this nal design is the same as before except that the Toluene mole fraction of the distillate is 84.7 instead of 84.6 mole%.

Conclusion
We have used ChemSepT M to determine if an existing tray column can be retted with #50 IMTP packing. This design exercise was posed in Random Packings and Packed Towers, Design and Applications by R. Strigle. Although not all details of the exercise are in agreement with Strigle we nd that a packed column can meet the purity specications but that it provides no capacity advantage over a convential double pass sieve tray. We have used ChemSep to determine if an existing tray column can be ret with packing and provide an increase in capacity. The design exercise was posed in Random Packings and Packed Towers: Design and Applications by R. Strigle (Gulf, 1987). Although not all details of the exercise are in exact agreement with Strigle (due to a lack of information in most cases) we nd, in agreement with him, that it is possible to equip the tower with packing, gain an improvement in separation and increase capacity by (at least) 30%.

ChemSep Case Book http://www.chemsep.com/downloads/index.html Copyright (2006) Ross Taylor and Harry Kooijman 3