Optimal design and control of trains of dividing wall columns for the separation of petrochemical mixtures

Abstract

Abstract The purification of most of the fluid mixtures of industrial relevance is performed in distillation columns trains. In spite of its low energetic efficiency, distillation remains as the most used separation process, due to its great flexibility to obtain products with high purity. Nevertheless, actual concerns about the energy use and the reduction of CO2 emissions encourage the efforts in the increasing of the energetic efficiency of these schemes. Among these schemes with major energetic efficiency, the dividing wall column allows a reduction of energy consumption along with a reduction in capital investment and space requirements. The dividing wall columns have been recently used in different applications; however, to our knowledge, they have not been used in the separation of petroleum mixtures. In this work, we study the use of trains of dividing wall columns for the separation of a typical mixture of refinery cuts. The proposed train is compared with conventional distillation sequences in terms of energy consumption and control properties. Both conventional and thermally coupled trains are designed with a multiobjective genetic algorithm, with constraints handling, coupled to Aspen Plus. As a result, Pareto front of both sequences are obtained, and some of these designs are selected to perform control studies. We also analyze the dynamic behavior of the structures under different operating points, including the one with minimum energy consumption. The basic idea is that if one changes the operation point, the control properties might change as well. The control analysis properties are analyzed with the application of the singular value decomposition technique. Results show interesting trends in the optimal designs that integrated the trains in energy consumption and that the controllability properties of integrated distillation sequences may change significantly depending on the selected operation point control properties.