Process intensification has become an important concept to meet the challenges of an increasingly cost driven chemical industry. The integration of two or more unit operations into a single apparatus can lead to significantly lower cost of investment as well as cost of operation. While in dividing-wall columns two distillation columns are integrated into a single shell, reactive distillation columns combine reaction and distillation in just one apparatus. Furthermore, these two integrated processes can be combined to the reactive dividing-wall column (RDWC).
The design of reactive dividing-wall columns has to handle more degrees of freedom compared to less integrated process alternatives. This is due to the liquid split and the vapor split, which distribute the liquid flow at the top and the vapor flow at the bottom on both sides of the dividing-wall. In total, this highly integrated process shows a strongly non-linear behavior, making its design complex. Many local optima exist and various design variables proof to be highly sensitive on each other.
Therefore, an effective approach to determine an (e.g. cost) optimal column design is the application of suitable optimization algorithms. But also these algorithms cannot guarantee to find the best possible design. Especially the combination of local optima and high dependency of design variables complicate the search in the solution space.
To overcome these limitations, we will show in this work how fundamental process understanding can be used to transform the solution space and thereby facilitate the search for the global (cost) optimal column design. By combining fundamental process understanding with optimization algorithms the process design becomes not only more effective, but also more efficient. Furthermore, the application of this method to a wide variety of process alternatives and reaction systems allows developing heuristics regarding the optimal process integration level, and hence, to speed up the conceptual design phase drastically.
In the presentation we will deduce firstly fundamental process understanding of the reactive dividing-wall column by extensive and systematic simulation studies. Furthermore, we will explain the optimization approach, and finally how its performance can be leveraged by fundamental process understanding.