Studies on divided wall distillation column

Abstract

Distillation is the most employed, but also the most energy-intensive separation method being used in chemical process industry. Therefore, the efficient design of the distillation process is important so as to reduce the energy requirement without compromising with the product quality. Divided wall column (DWC) is an innovative design in this direction where two or more simple columns are thermally coupled into one column by providing a vertical wall, dividing the core in two parts that work as the pre- or post-fractionator column and the main column. Thermodynamically, a DWC can be considered as an equivalent of Petlyuk column; the only difference being that a DWC is built in only one shell, whereas a Petlyuk column consists of two or more columns. Simulation of a DWC is a challenging task, as it involves many connecting streams making the problem highly non-linear. Further, simulation of a DWC requires simultaneous solution of material, momentum and energy balance equations. For a simple distillation column, the momentum balance equations are important only at the mechanical design stage for the estimation of effects of pressure drop whereas these are usually neglected during theoretical design (simulation) stage as its impact on vapour-liquid equilibrium is not appreciable. The optimal design of a DWC is obtained by minimizing the energy duty, as it accounts for the major fraction of the operating cost. The design variables such as reflux ratio and split ratios interact with each other and need to be optimized simultaneously for optimal operation and design. The optimization of the DWC operation cannot be handled with the existing commercial simulation software without coupling them with an external sub-routine/ software for optimization. The structural and process parameters of a DWC are generally optimized separately and there are no reports in the open literature on the simultaneous optimization of the structural and operating parameters of a DWC. The present thesis reports about the modelling of a DWC, the rationalization of the degree of freedom analysis, the effect of column pressure drop, the effect of feed quality and the optimization of energy requirement for a ternary system. A mathematical model was developed for a DWC by applying the conventional MESH equations plus the pressure drop (ΔP) equation, taken together as MESHD equations. Rigorous simulation of this DWC was carried out using Multifrac model of ASPEN Plus™ software (for solving MESH equations). For calculating the pressure drop, ‘Tray Rating’ feature of ASPEN Plus™ was used. Box–Behnken design (BBD) under response surface methodology (RSM) was used for the optimization of the design and operational parameters and to evaluate the effects of these parameters and their interactions on the energy efficiency of a DWC. The importance of pressure drop in a DWC was analyzed theoretically by attempting to relate the ‘natural’ or ‘feasible’ vapour split as a function of pressure drop, reflux ratio and the liquid split ratio of a DWC. Benzene-toluene-para-xylene (BTX) separation was taken as the case study. The results are interesting in that they show virtually no effect of liquid split ratio on the feasible vapour split ratio. Significantly different pressure drops are obtained at the same internal vapour flow rate in the two sides of the DWC, and even a slight variation in the vapour flow rate leads to a large variation in the concentration profile of the components. The tray hydraulics plays an important role in the performance of a DWC. It was concluded that the pressure drop should be taken as one of the key parameters for the optimal design of a DWC. The effect of feed (thermal) quality on the performance of a DWC was also studied. The feed quality was varied from the superheated vapour to sub-cooled liquid feed. It was observed that the feed quality influences the pressure drop across the feed side of the divided wall which in turn affects feasible vapour split ratio. However, the vapour split ratio is almost independent of liquid split ratio and reflux flow rate for sub-cooled and saturated liquid feed. For saturated vapour and superheated feeds, liquid split ratio has appreciable influence on the feasible vapour split ratio. It was also observed that the equalization of pressure drop leads to nearly equal vapour flow rate on both the sides of the DWC when a sub-cooled or a saturated liquid feed is introduced in the DWC. The BBD under RSM was used to optimize the parametric values for obtaining the optimum energy efficiency for a specified separation using a DWC for different thermal quality of feed. Optimization of a DWC was carried out to study the product quality and energy efficiency as a function of reflux rate, liquid spilt and vapour split. It was also targeted that the difference between the pressure drops across the two sections of the divided wall remains zero. The reflux rate and vapour split are highly significant for the reboiler duty of a DWC and for the pressure drop across the dividing wall as compared to liquid split. The structural and process parameters of a DWC were optimized simultaneously for the energy efficiency, for the separation of a specified ternary mixture. The system has nine structural variables and four process variables including the location of the feed tray, the location of the side stream tray, and the number of trays in the main column and in the post-fractionator, location of the divided wall, number of trays in the divided wall section, reflux rate, liquid spilt, vapour split and the feed quality. These variables were used for the optimization of the product purities and the reboiler and condenser duties. The process variables were found to have significant effect on the energy efficiency of a DWC as compared to the effect of structural variables. The predictions from the BBD optimization agreed well with the results of the rigorous simulation. Optimization by BBD under response surface methodology (RSM) vividly underscores the effects of different parameters and their interactions on the energy efficiency of a DWC. This method may prove to be very useful for the industrial operation of a DWC, as it requires a fewer number of simulation runs and can be carried out in a shorter time.