Hybrid Distillation/Melt Crystallization Process Using Thermally Coupled Arrangements: Optimization with evolutive algorithms

Abstract

Innovative hybrid processes offer significant cost savings, particularly for azeotropic or close-boiling mixtures. Hybrid separation processes are characterized by the combination of two or more different unit operations, which contribute to the separation task by different physical separation principles. Despite of the inherent advantages of hybrid separation processes, they are not systematically exploited in industrial applications due to the complexity of the design and optimization of these highly integrated processes. In this work we study a hybrid distillation/melt crystallization process, using conventional and thermally coupled distillation sequences. The design and optimization were carried out using, as a design tool, a multi-objective genetic algorithm with restrictions coupled with the process simulator Aspen Plus™, for the evaluation of the objective function. The results show that this hybrid configuration with thermally coupled arrangements is a feasible option in terms of energy savings, capital investment and control properties.

Introduction

The chemical industry is the largest consumer of energy of the manufacturing industries; it represents 6% of all domestic energy use, and 24% of the total U.S. manufacturing energy use. Petroleum refining is the second largest consumer of energy, with a contribution of approximately 10% to the total U.S. manufacturing energy use. Separations assess for approximately the 60% of the in-plant energy usage for these two industries. Distillation operations account for near to 95% of the total separation energy used in the refining and chemical processing industries with about of 40,000 distillation columns operating in over 200 different processes [1]. This high usage rate is primarily due to distillation's flexibility, low capital investment relative to other separations technologies and low operational risk. Unfortunately, the energy efficiency of a commercial distillation column is low, with a second-law efficiency of less than 10% being typical [2]. This is reflected on high quantities of thermal energy required to achieve the desired purification. Thus, there is a major research opportunity area on the development of improvements or replacements for distillation to achieve significant energy savings, due to the large sunk capital investment in existing plants, the slow rate of plant replacements, and the diverse numbers of applications where distillation is utilized. Such enhancements to the distillation operation may have particular importance in separations with low relative volatilities, or those that operate at cryogenic or very high temperatures.

Opportunities for improving equipment in existing distillation systems include divided wall columns, improved packing designs, heat integrated distillation, and improving mass transfer efficiencies; these options have already been studied [3], [4], [5], [6]. Independent technical reviews of industrial experiences to date and pilot-plant demonstrations are needed for wide-spread implementation of divided wall columns and heat integrated distillation systems. The debottlenecking opportunities described above can not be implemented generically across the industry; they will be process flowsheet specific. Hybrid and improved equipment systems are already being used to a limited extent within the industry, and there are improved mass separating agents and process equipment on the market today that are not being used extensively by the industry for debottlenecking plants. This is in part due to lack of tools to evaluate their performance for specific applications [1].

In recent years, there has been an increasing interest on hybrid process in chemical engineering (see [7], [8], [9], [10], [11], among others). Because of inherent drawbacks of conventional processes, enhancement of such processes represents an important field of opportunity to the development of technological advances. Therefore, the focus of separation and purification studies is gradually being shifted from unit operations towards hybrid processes. One example of a hybrid processes involves the combination of the well-known crystallization operation with the ease of phase separation typical of distillation. They are applicable to solid–liquid–vapor three-phase equilibrium, and it may result useful for the separation of mixtures with pairs of components with relative volatilities close to 1. In such systems, the liquid, enriched in the impurities that have not passed into the crystals, vaporizes at a low pressure, near the triple-point pressure of the main component. Therefore, combining all these processes in one installation is expected to raise the separation efficiency without requiring any considerable extra expenses [9]. Another important hybrid process corresponds to the combination of distillation and melt crystallization for separation of close-boiling isomer mixtures. The hybrid distillation/melt crystallization process combines advantages of the distillation and the melt crystallization, in which very high separation factors per stage can be reached. Simultaneously, the combination of distillation and crystallization overcomes the shortcomings of the individual unit operations, i.e. high energy requirements at small separation factors, and limitation of yield by eutectics, respectively. Several hybrid processes for the separation of terphenyl, xylene, dichlorobenzene and diphenylmethane diisocyanate isomers are reported in literature. Berry and Ng [7] used simplified models, based on constant separation factors, for the synthesis of hybrid distillation/melt crystallization processes and proposed guidelines for flowsheet selection. Wallert et al. [12] also applied shortcut methods for the synthesis and evaluation of a hybrid distillation/melt crystallization processes. Franke et al. [10] proposed a three-step design method for hybrid distillation/melt hybrid processes. In a first step different sequences are generated by heuristic rules. These sequences are evaluated in a second step by shortcut methods on the basis of energy requirements, in order to identify the most promising alternatives. In the third and last step a reduced number of promising sequences is rigorously optimized by MINLP methods, and the best sequence on the basis of total annualized costs is chosen. It should be noted that the creation of a superstructure is not a trivial task, especially for hybrid separation processes. Marquardt et al. [13] reviewed an optimization-based framework composed of shortcut and rigorous design steps for the robust and efficient synthesis of hybrid distillation/melt hybrid processes. A multitude of hybrid processes composed of distillation and melt crystallization units are evaluated with powerful shortcut models. A selection of promising process variants is subsequently rigorously optimized by an economic objective function and discrete-continuous optimization techniques. It is shown that the design of the cost-optimal hybrid process within the systematic synthesis framework can be accomplished with robustness and efficiency. However, the main drawback of a mathematical programming method is that they cannot guarantee to find the global optimum if non-convex equations are present, and, also, in spite of their high computational effort they often fail to solve large scale process engineering problems with highly non-lineal and mixed-integer models [14], [15].

On the other hand, the presence of recycle streams for complex distillation schemes has generated the idea that control problems might be expected during the operation those arrangements with respect to behavior of conventional distillation configurations. Understanding control properties of process with distillation columns with thermal couplings is an issue of extreme importance since designs with economic incentives often conflict with their operational characteristics. However, recent publications report considerable progress in the identification of suitable control variables and control strategies for some configurations with thermal coupling [16], [17], [18], [19].

In this work, the design and optimization of a hybrid distillation/melt crystallization process with conventional and thermally coupled distillation sequences is presented. We select the use of distillation configurations with thermally coupling due to the energy savings, total annual cost reduction and good dynamic behavior of those arrangements in comparison with properties of conventional distillation structures (see [5], [20], [21], [22] among others). The design and optimization were carried out using a multi-objective genetic algorithm with restrictions handling, coupled with the process simulator Aspen Plus™, for the evaluation of the objective function, ensuring that all results considered full use of the equations of the models contained in the Aspen Plus simulator. Rudolph [23] proved that genetic algorithms, a kind of evolutive algorithms, converge to the global optimum. To the best of our knowledge, multiobjective stochastic methods have not been reported for optimal design of a hybrid distillation/melt crystallization process with conventional and non-conventional distillation arrangements. For each analyzed system a set of optimal designs, called Pareto front, is obtained. The results show that the hybrid configuration with thermally coupled arrangements is a feasible option, presenting advantages over configurations with conventional distillation systems in terms of energy savings (and, consequently, reductions in greenhouse gas emissions), capital investment and control properties.