A novel structured packing design was developed for this column. Sheets of expanded metal with appropriate openings (6 mm x 3.5 mm) were cut into strips (300 mm wide) and rolled into cylinders that fitted snugly inside the column. The rolled cylinders provided a substantial contact area and a complicated flow path, which suggests an adequate HETP. Air-water tests indicated that these packings also provide adequate capacity, holdup and radial dispersion and a low pressure drop. It was not possible to rigourously test these packings in a distillation environment before the column was built, but the significant cost saving compared with commercial packings (e.g. S334/L for 6 mm raschring rings compared with S16/L, excluding labour, for the rolled cylinders) provided the incentive to pursue the more novel (and therefore riskier) approach.
Figure 11.1 - Reactive Distillation Column Design
The structured packings were installed in the column on a support grid constructed from 5 mm x 5 mm stainless steel mesh mounted on a ring welded inside the column. A similar hold down grid was also manufactured and can be secured at the top of the packed bed to prevent movement of the packing due to rapidly expanding vapour. However, the weight of the packing (2.05 kg/L), which would be a disadvantage for commercial installations, is sufficient to prevent internal movement in this case.
The reactive section was required to accommodate packings for mass transfer and sufficient catalyst for the ETBE reaction to approach chemical equilibrium. Several commercial cation exchange resins are suitable to catalyse etherification reactions and Amberlyst 15™ was used based on availability and cost. SpeedUp simulations incorporating a full kinetic model of the reaction (Chapter 3, Section 3.2) indicated that the isobutene conversion would be within 0.1% of the equilibrium conversion and that the variation in stage-to-stage reaction rates would be acceptable if a total of 11.5 kg of catalyst was used. This is equivalent to 2.5 L of catalyst at a bulk density of 0.61 g/cm3 (Zhang and Datta, 1995).
The catalyst has a mean particle size of approximately 0.7 mm (Zhang and Datta, 1995) and is, therefore, too small to be packed directly in the column because of hydraulic restrictions alone. A suitable catalyst packing arrangement was, therefore, required to provide adequate contact between liquid and catalyst to allow the reaction to progress normally and adequate vapour space to prevent flooding. Several packing arrangements were tested using cloth to contain the catalyst particles. The packing arrangements tested fall into two categories: (a) arrangements based on circular sachets of catalyst (Figure 11.2a); and, (b) arrangements based on rolled cloth strips containing catalyst (Figure 11.2b). Three types of material were tested: (a) interfacing; (b) 'Chux' wipes; and, (c) fine terylene weave (a lightweight curtain material). The materials were chosen to offer minimal resistance to both liquid and vapour but provide adequate strength to prevent tearing. Of the materials investigated, interfacing had the greatest strength and workability while the nylon mesh offered least resistance to flow.
Tests were performed on the various catalyst packing arrangements using the same packed column rig that was used to test the separation section packings. Graded sand (0.50-0.85 mm) was initially used in place of the catalyst. The catalyst sachets were tested with 40 ml of graded sand (full) and 20 ml of graded sand (half-full) per sachet. However, none of the test materials provided adequate capacity for the proposed operating conditions, either when full or half-full, and severe flooding was experienced at low liquid and vapour loadings. Catalyst bales, constructed to the dimensions given in Figure 11.2b, and filled with approximately 4 ml of graded sand per pocket were also tested. A relatively high pressure drop was observed but flooding was not detected at the proposed vapour and liquid flows. The test results were corrected for differences in physical properties between the air-water system and the hydrocarbon mixture using the generalised pressure drop correlation (GPDC) for packings (Kister, 1992). On this basis, it was considered that flooding was still possible in the pilot plant for the proposed operating conditions, and a second catalyst bale was constructed using a 50 mm strip of mesh interleaved with the catalyst bale. This arrangement had a significantly lower pressure drop and, based on the GPDC, is considered unlikely to flood at the operating conditions expected in the pilot plant column.
Six catalyst bales (two per reactive stage) were constructed using the cloth strip design, shown in Figure 11.2b, interleaved with 2.4 m of 50 mm mesh strip. Each roll contained approximately 160 pockets so that the total volume of catalyst in the reactive bed (three stages) was approximately 4.0 L. This is equivalent to around 2.4 kg of catalyst and should be sufficient to ensure that the reaction progresses close to equilibrium. The overall structure of the reactive bed consisted of catalyst rolls interspersed with randomly packed pieces of expanded metal (approximately 25 mm x 25 mm) of the same grade used in the separation section beds. A greater packed height per stage was used in the reactive section as the configuration specified was estimated to have a higher HETP.
As with the stripping and rectifying sections, a support grid and hold down grid were manufactured. The support grid was welded to a ring which was welded inside the column while the hold down grid attached to another ring welded into the column in order to prevent disruption of the packed bed.
Lateral mixing is usually sufficient to eliminate the need for liquid distributors if the ratio of the column diameter to packing diameter (D/dp) is less than 20 (Kister, 1994). Most pilot scale columns meet this criterion as the packing size is usually large compared with the column diameter. However, structured packing is more sensitive to maldistribution and the design of the separation section packings is susceptible to channelling so that the use of distributors might be beneficial.
A ring distributor was selected for the feed. A drip point analysis (Kister, 1994), Figure 11.3, suggested that a six point distributor would produce the most consistent liquid distribution without leaving the centre dry. The ring diameter was selected to provide adequate flow at the walls using the empirically derived formula shown in equation (11.1).
10 Dnp Points
12 Drip Points
15 Drip Poinls
Figure 11.3 - Ring Distributor Designs
10 Dnp Points
12 Drip Points
15 Drip Poinls
The diameter of the drip holes (1.0 mm i.d.) was selected to guarantee a large ratio of exit pressure drop to transport losses which is conducive to good distribution of liquid around the distributor ring. The additional pressure drop created by the small hole diameter is compensated by surplus pump head. The distributor dimensions are shown in Figure 11.4.
The reflux distributor was based on the same basic design in order to provide good radial distribution. However, an identical design would have created an excessively large pressure drop within the reflux circuit that relies on gravity flow. Therefore, larger holes (3.0 mm i.d.) were specified.
Was this article helpful?