Control of the Divided-Wall Column

The second column has 28 stages and is fed on stage 14. The reflux drum operates at 0.13 atm, giving a reflux-drum temperature of 322 K. Base pressure is 0.31 atm, which gives a base temperature of 378 K. The reflux ratio is 1.53, and the reboiler heat input is 24.53 MW. The column diameter is fairly large (8.22 m) because of the vacuum operation. 

3 CONTROL OF THE DIVIDED-WALL COLUMN 12.3.1 Control Structure 

The control objectives are to maintain stable on-specification operation in the face of disturbances in throughput and feed composition and to minimize energy consumption. We limit our study to conventional PID control structures. 

Conventional distillation control wisdom says that it is usually more effective to control impurity levels than to control purity levels. The use of impurity instead of purity is a 

Standard process control principle because you want to control a variable that is sensitive to the manipulated variable. A change in impurity from 1 to 1.5 mol% is much greater (on a relative basis) that the corresponding change in purity from 99 to 89.5 mol%. The principle is particularly important in distillation control where changes in trace amounts of other nonkey components can make it impossible to maintain a key-component purity, but maintaining an impurity of the other key component is still possible. 

---

The dynamic control of the divided-wall column has been explored in a relatively small number of papers. Control is more difficult than with a conventional two-column separation sequence because there is more interaction among controlled and manipulated variables since the four sections of the column are coupled. The vapor spht is fixed at the design stage and cannot be changed during operation, but the Uquid split can be manipulated to achieve some control objective. 

M. Serra, A. Espufia, and L. Puigjaner. Control and optimization of the divided wall column. Chem. Eng. Process., 38 549-562, 1999. 

Energy reductions of up to 30% have been reported in some systems for Petlyuk and divided-wall column configurations compared with the direct-separation sequence. Figure 12.1 gives the flowsheet of a divided-wall column for the numerical benzene/toluene/xylene separation example considered later in this chapter. The material presented in this chapter is based on the paper that studied the control of divided-wall columns. ... 

The divided-wall column has many degrees of freedom at the steady-state design stage. The number of stages in the four different sections of the column, the locations of the feed and sidestream withdrawal points, and the location of the wall are seven of the parameters that must be specified and are all fixed by the physical equipment at the time of construction. They cannot be changed during operation. The location of the wall fixes how the vapor splits between the two sides of the wall, so the vapor split is not adjustable during operation for control purposes. 

The control objectives for the conventional direct-separation sequence are quite similar to those of the divided wall. Each column has two manipulated variable reflux and reboiler heat input. Therefore, two variables can be controlled in each column. Figure 12.26 gives the proposed control structure for this conventional two-column process. 

Industrial applications of the divided-wall (Petlyuk) column have expanded, so a new chapter has been added that covers both the design and the control of these more complex coupled columns. The use of dynamic simulations to quantitatively explore the safety issues of rapid transient responses to major process upsets and failures is discussed in a new chapter. A more stmctured approach for selecting an appropriate control structure is outlined to help sort through the overwhelmingly large number of alternative stmctures. A simple distillation column has five factorial (120) alternative structures that need to be trimmed down to a workable number, so that their steady-state and dynamic performances can be compared. 

In this chapter, we discuss both the steady-state design and the dynamic control of divided-wall columns. Aspen simulation tools are used. The industrially important ternary separation of benzene, toluene, and o-xylene (BTX) is used as anumerical example. The normal boiling points of these three components are 353, 385, and 419 K, respectively, so the separation is a fairly easy one with relative volatilities aa/ax/ax of about 1. I2.2I. The feed conditions are a flow rate of 3600 kmol/h, a composition of 30/30/40 mol% B/T/X, and a temperature of 358 K. Chao-Seader physical properties are used in the Aspen simulations. Product purities are 99mol%. All simulations use rigorous distillation column models in Aspen Plus. 

All the spheres in a layer were supported by two spheres of the layer below and the column wall, creating a stable packing structure. As the tube-to-particle diameter ratio of the bed was only four, the entire packing structure was controlled by the influence of the wall. Nevertheless, the packing was divided into an immediate wall layer and a central section, but this should not be taken to imply that the central structure was not wall influenced. Although a three-sphere planar structure would almost fit within the nine-sphere wall layer, there was just not enough room at the same axial coordinate. When, however, the... 

We have designed and implemented a reactive divided wall distillation column for the production of ethyl acetate from acetic acid and ethanol. Important aspects derived from steady state simulation were considered for instance, a side tank was implemented in order to split the liquid to both sides of the wall and a moving wall inside the column that allows to fix the split of the vapor stream. The dynamic simulations indicate that it is possible to control the composition of the top and bottoms products or two temperatures by manipulating the reflux rate and the heat duty supplied to the reboiler, respectively. The implementation of the reactive divided wall distillation columns takes into account important aspects like process intensification, minimum energy consumption and reduction in Carbon Dioxide emission to the atmosphere. 

---