Identifying Heat-Integrated Energy-Efficient Multicomponent Distillation Configurations

Abstract

We present a tractable nonlinear programming (NLP) formulation that models a given multi-component distillation configuration and searches for its global minimum heat duty. The novelty in the current model is that it can explore feasible heat integrations with a pre-specified desired minimum approach temperature between various condensers and reboilers while simultaneously optimizing the operating conditions within the configuration. We do not use cumbersome thermodynamic models for the equilibrium temperature calculation of a saturated multicomponent mixture. Instead, we propose a modified version of the well-known Antoine equation that reduces the calculation of the temperature at a given pressure to a simple function of component mole fractions and relative volatilities while retaining the fidelity of more complex models. We explore possible heat integrations by creating a heat exchange network between column condensers, reboilers, and side draw product locations. Considering these integrations along with the heat duty minimization is essential because it is often possible to alter the operating conditions of the columns and reduce energy consumption by admitting more heat integration possibilities. Finally, we demonstrate the power of our framework in identifying optimal configurations that yield large energy savings for several four- and five-component zeotropic distillation systems.

Relaxing the constant molar overflow assumption in distillation optimization

“…To demonstrate the effect of wide differences in latent heats, we discuss a second case study for a five‐component paraffins separation, shown in Table 2, where . We use the shortcut model of Mathew et al., 43 with the temperature constraints of Nogaja et al 44 . and the modification mentioned in Section 3, to identify an energy‐efficient heat‐integrated configuration.…”

“…Column pressures were allowed to vary between 1 and 2 atm for adjusting temperatures of the condensing and boiling fluids when needed. Fitted relative volatilities were determined to capture the VLE, , 43 and to capture temperatures, 44 using the Peng–Robinson thermodynamic package 42 . The pseudo‐Antoine constants 44 for this separation were determined to be 9.48, 11,994.7 K, and 376.4 K. All sidedraws were restricted to be liquid only.…”

“…shown in Table 2, where maxðλÞ ¼ 2:01 minðλÞ. We use the shortcut model of Mathew et al, 43 with the temperature constraints of Nogaja et al 44 and the modification mentioned in Section 3, to identify an energy-efficient heat-integrated configuration. We solved this shortcut model to global optimality for each of the 6128 different regularcolumn configurations possible for this separation.…”

“…Plotting data used in Figures 3,4 The results of Section 4 were either cited from an earlier publication, 20,41 or were produced using the optimization models of Nallasivam et al, 30 Mathew et al, 43 Nogaja et al, 44…”

“…Fitted relative volatilities were determined to capture the VLE, , 43 and to capture temperatures, 44 using the Peng–Robinson thermodynamic package 42 . The pseudo‐Antoine constants 44 for this separation were determined to be 9.48, 11,994.7 K, and 376.4 K. All sidedraws were restricted to be liquid only.…”