Simple, analytical criteria for the sequencing of distillation columns

Abstract

A quantitative criterion for the selection of simple distillation sequences is derived for ideal mixtures. A simple cost model, along with a shortcut solution of Underwood's equations, gives an analytical form for the total vapor rate, which is the key design variable. The results for column sequencing that are based on the analytical criterion agree well with more exact solutions, but they indicate that in numerous situations the commonly accepted heuristics are incorrect. SCOPEThe synthesis of a distillation separation sequence is often a key step in the invention of a process flowsheet. Heuristics are normally used for this step in the process synthesis. However, some heuristics, such as "remove the lightest component first," depend only on volatility, whereas other heuristics, such as "remove the most plentiful component first," depend only on feed composition. Thus, it should not be surprising that the previously published heuristics can be contradiktory.Since we expect that the selection of the best separation sequence must depend both on relative volatility and feed composition, we need to have a criterion available that includes both quantities. A few functions of this type have been proposed (Lu and Motard, 1982; Tedder and Rudd, 1978), but the functional forms chosen are based on intuition. As an alternate approach, we show how an appropriate functional form can be derived that is based on approximate solutions of the design equations and an order of magnitude analysis of the cost equations. The results indicate that the new analytical approach gives better predictions than the qualitative heuristics. CONCLUSIONS AND SIGNIFICANCEConflicts and inaccuracies in heuristic or rank-order approaches to the selection of a distillation sequence can be eliminated using the analytical criterion developed here. The approximate, but accurate, solution of the design equations, along with a simple cost model, are sufficient for process synthesis. For ternary mixtures separated in a simple column sequence, the optimal sequence and, more importantly, the sensitivity of this decision to the composition and thermodynamic information are shown by Eq. 23. Mixtures containing more components can be easily investigated with similar expressions.

Towards computer‐aided separation synthesis

“…For completeness, feed compositions in regions III, IV, and V requiring but one binary column, and the product regions α, β, and δ without the need for separations are indicated. The accurate results of the complete solution map for varying feeds agree with simplified heuristics proposed by Malone et al (1985). The complete solution maps computed with the novel approach can be used to decide on optimal column configurations when feed compositions are changing from day to day, as is the case in batch processes campaigns used in pharmaceutical manufacturing sites (e.g., Chakraborty and Linninger, 2002, 2003).…”

Automatic design of conventional distillation column sequence by genetic algorithm

“…Although the difference is only some 1.0%, the DCS shown in Figure 6a is better than the one shown in Figure 6b. This minor difference has also been confirmed in other studies (Malone et al, 1985;Gadkari and Govind, 1988). The TACs corresponding to all arrangements shown in Figure 1 are presented in Table 6 in increasing order.…”

“…The TAC corresponding to this DCS is 3.159 × 10 6 $/year. This DCS is also reported as the optimum simple DCS by other researchers (Malone et al, 1985;Gadkari and Govind, 1988;Glinos and Malone, 1989;Porter and Momoh, 1991;Rong et al, 2001). There is also a slightly different configuration proposed by some of the researchers (Floquet et al, 1994;Jobson, 1997;Shah and Kokossis, 2001) and is shown in Figure 6b.…”

“…The first test problem is widely used by other researchers (Malone et al, 1985;Gadkari and Govind, 1988;Glinos and Malone, 1989;Porter and Momoh, 1991;Floquet et al, 1994;Jobson, 1997;Rong et al, 2001;Shah and Kokossis, 2001). Table 3.…”

Separation vector formulation for the synthesis of multicomponent separation sequences

“…Another intuitive measure is based on the commonly-used six-tenth rule for equipment cost correlation: n-I Objective Function = (L,D,)o.6 (1) , = I for the separation of an n-component mixture, where L, and Dj are the separation mass load and the cost index of separation of the i-th simple separator, respectively. Malone et al (1985) proposed the total annualized cost (TAC), which is a function of the number of trays, N, and the vapor rate, V , as the basis to determine the optimal sequence. The problem of optimal separation sequence synthesis is further complicated by the introduction of justifiable options of stream splitting and blending that require little energy, but can offer considerable improvement in process economics.…”