Distributed feed columns

FIGURE 3.2 CS breakdown for the (a) distributed feed column having five CSs and (b) side draw column with four CSs. 

FIGURE 6.4 Movement of difference points for distributed feed columns. 

It is apparent that the single feed column is the one limit on the distributed feed column. At the other end of the design spectrum, one may theoretically choose to split the feed stream in infinitely many substreams, resulting in infinitely many CSs. Although this may seem like a purely academic or impractical limit, it does have some use because the feasible region resulting from this will show us all possibilities when deciding on a column. Thus, the profiles of any feed stream division will lie in... 

For the infinitely distributed feed column the feasible region is as large as it can possibly be, bound by the locus of saddle nodes of the internal CSs TTs. Notice that roles of the two product producing CSs have been increasingly dimini ed as the number of feed points are increased. They are almost not needed at all in Figure 6.8 ... 

Try this for yourself Reproduce Figure 6.10 using DODS-DiFe and the given parameters. Try and find other minimum reflux designs for a distributed feed column by changing parameters such as a, Rai f< < product compositions. 

FIGURE 6.12 A multiple (5) distributed feed column where TTs cascade around a line. 

The reader should be aware that the minimum reflux scenarios presented here are just one of three possible ways the minimum reflux limit can be obtained in distributed feed columns. The designs shown thus far all depicted minimum reflux when the vertex of the internal CS adjacent to the topmost rectifying section lies exactly on its profile, that is, a pinch occurs on the topmost rectifying CS. It is perfectly valid for the minimum reflux condition to be determined by the bottommost stripping profile, or indeed where the TTs of the internal CSs do not overlap one another. The latter case is shown in Figure 6.12 where the column reflux has been reduced and TTs cascade around one another, thereby limiting any further column reflux reduction. The general requirement for minimum reflux is however the same as for simple columns any reflux value below the minimum reflux value will lead to a discontinuous path of profiles, and minimum reflux is therefore the last reflux where a continuous path is still maintained. 

FIGURE 6.13 A distributed feed column at minimum column reflux for acetone/ benzene/ chloroform system with a feed distribution of [0.5,0.5]. Solid profiles indicate CPM profiles while circles indicate Aspen generated compositions. 

FIGURE 6.15 Distributed feed column at (a) minimum reflux and (b) below minimum reflux. 

It should be noted that we have only looked at evenly distributed feed columns, that is, the feed stream is divided into equal substreams. Feed streams can, of course, be split into different fractions, and sometimes, different compositions, and liquid fractions too. A case where there are di erent quantities of feed distributed is discussed in the following example. 

Example 6.1 Suppose you have two incoming process streams with the same purity of [0.4,0.3] but with flowrates of Fi = 0.7 mol/s and F2=0.3 mol/s. We would like to obtain a distillate stream with a purity of [0.95,0.01 ] and the bottoms product should only contain 0.01 of lights. From purely an energy point of view, is it better to (a) combine the streams and purify them in a simple column or (b) feed a distributed feed column in their given proportions Note that to test option (b), both alternatives should be considered that is, which stream should be sent to the top of the c<4umn, F or F2 The relative volatilities can be assumed as [5,1,2. Use the DODS-DiFe package to solve this problem. 

It is blatant that both scenario 1 and 2 are feasible as both the TTs from the internal CSs overlap the product producing profiles. The vertices of both these TTs are located some way from the profiles therefore, a reduction in the reflux for both scenarios will have energetic advantages and still be feasible. However, it would seem that the second scenario may be more beneficial than the first, because its TFs vertex is located fiirth away from the profiles. We can proceed to determine what the minimum reflux conditions of the distributed feed columns are. We thus have to find the reflux values that will result in either of TTs vertices lying precisely on a profile. This is shown in Figure 6.17a and b for Scenario 1 and 2, respectively. 

---