(43d) Dynamic Behavior of a Multiple Dividing Wall Column – a Theoretical and Experimental Study

The world’s first pilot-scale multiple dividing wall column (mDWC) was designed recently at Ulm University (Germany) and was finished in late 2021. Depending on the number of dividing walls, mDWCs are able to separate mixtures into four or more pure components and offer significant potential savings in both, OPEX and CAPEX, in comparison with conventional distillation sequences. The plant configuration realized in Ulm has two instead of three partition walls for the separation of quaternary mixtures. Despite this simplification, the column can separate a very large number of potential mixtures at the thermodynamic optimum. During the design phase of the plant, one of the challenges was finding a way to start up the column from the cold and dry state. The start-up of distillation columns is an intensively discussed topic in research. Most research in this field is aimed at minimizing the production costs as well as generated waste by shortening the time needed for the start-up. There are already several studies on time-optimal start-up strategies for distillation columns and also the start-up of dividing wall columns has been dealt with in some works. However, to the best of the author’s knowledge, the start-up of mDWCs has neither been studied in theory, nor in practice yet.

In this contribution, a first study on the start-up process of the pilot mDWC is described, that combines theoretical and experimental investigations. The goal of the study was to generate a start-up procedure, which is fast, reliable, and universal.

For the development and investigation of possible start-up strategies, a close-to-reality dynamic simulation model of the plant was set up in Aspen HYSYS. As an initial step, start-up strategies for simple three-product DWCs from literature are adapted to be used for mDWC and tested with the help of the simulation model. The strategies, which are found feasible for the pilot column, are looked at in greater detail. The influence of different operating parameters on the start-up time is studied and the most relevant factors affecting the start-up time are pointed out. The start-up procedure was then optimized based on the obtained theoretical findings. Through the optimization, the start-up time could be significantly reduced and the simulation results suggest that the cold, dry plant can reach the desired operating point within a few hours. The start-up strategy was transferred to the real column and the results of the simulation could be validated against experimental data. The agreement of the simulations with the real plant behavior will be presented.