A New Shortcut Design Method and Economic Analysis of Divided Wall Columns

Abstract

Divided wall columns (DWCs) can save energy and capital costs compared with traditional distillation columns; however, the design of DWCs is more difficult because there are more degrees of freedom. This paper describes a novel short-cut method that can be used to rapidly determine near-optimal values of important design parameters, including the reflux ratio, number of stages in all sections, and split liquid and vapor ratios for the three most common types of DWCs. The method is based on the development of a rational and efficient net flow model and the application of the methods of Fenske, Underwood, and Gilliland and the Kirkbride equation. The method is applied to two real systems, and the results are compared with results from rigorous simulations and optimization. The results show that the shortcut method leads to a process similar to a feasible actual process, and the total annual cost (TAC) based on the design method is close to the minimum (optimum) total annual cost. The results also show that the method also provides good initial values for rigorous optimization. The method is also applied to a fictitious process consisting of three components with constant relative volatilities, for different values of the ease of separation index (ESI), overall split difficulty, and feed composition. The results indicate that the method works well for a variety of process conditions and that the minimum vapor flow rate is a good surrogate for the total cost of process operation.

Effects of the ease of separation index (ESI) and feed composition on the selection of distillation configuration

“…In their work, they proposed equations to approximate the compositions of interconnecting streams. K. T. Chu et al [5] divided DWC system into five sections. Firstly, the component net flow model can determine the recoveries of component in the prefractionator.…”

“…, ′ , " are roots of Underwood's equation at the minimum reflux condition. They are determined by solving the equations (4), (5), and (6).…”

“…NOMENCLATURE A, B, C = ternary mixture (A is the most volatile component and C is the least volatile component). F [kmol.h -1 ] = feed flowrate D [kmol.h -1 ] = Top product flowrate W [kmol.h -1 ] = Bottom product flowrate S [kmol.h -1 ] = Side product flowrate [kmol.h -1 ] = liquid flowrate in the rectifying section ̅ [kmol.h -1 ] = liquid flowrate in the stripping section [kmol.h -1 ] = vapor flowrate in the rectifying section ̅ [kmol.h -1 ] = vapor flow rate in the stripping section N = number of stage R = reflux ratio R L = liquid split between prefractionator and main column R V = vapor split between prefractionator and main column K = volatile of component z = mole fraction at the feed flowrate x = molefraction at the product flowrate F − factor = the factor is defined by Kaibel et al[5] q = quality of the stream.…”

A comprehensive economic optimization methodology of divided wall columns for biopolyol separation

“…DWCs, in which both columns are located in a single-shell, and three-product streams are collected by the insertion of a dividing wall, have been studied since 1949 [11]. DWCs can not only save space and investment requirements because of the reduced number of columns and heat exchangers (figure 1) but also reduce energy costs by reducing the remixing effect [12] of streams; the economic savings can reach 30% [13]. Therefore, the application of DWCs to separate near-volatile biopolyol would save substantial costs relative to that of conventional approaches.…”

“…However, fully heat-integrated DWCs have more degrees of freedom than conventional columns, and the design of such units is complex. Many researches about the short-cut design of DWCs were based on the Fenske-Underwood-Gilliland-Kirkbride (FUGK) method aimed at minimum vapour flow rate [12,14,15]. Besides, recent articles have reported other systematic methods and models for DWC design: a rigorous model validated in the experiment was used for the scaled design [16]; coordinate descent methodology with a random research [17] and optimization-based design approach with an automated initialization procedure were employed to determine variables of DWCs during the initial design [18].…”