Superstructure modeling and optimization of dynamic processes applied to high-performance liquid chromatography with recycling

Optimal design of dynamic processes is far more challenging than steady-state optimization due to the added complexity of time, leading to a highly complex optimization problem. If more than one possible design or operation is also to be considered, a dynamic superstructure approach is more efficient than considering a series of individual dynamic optimizations. This work proposes a four-step methodology to optimize a dynamic chemical process, here applied to high-performance liquid chromatography (HPLC), based on a superstructure approach. HPLC is commonly used to separate valuable pharmaceutical products, normally based on a single-column elution process, although the basic operation can be improved by considering recycling. A superstructure model for a single recycling HPLC column is introduced, capable of handling the conventional elution policy as well as three recycling policies – conventional recycling, peak shaving (PS), and peak shaving with multiple feed injection (PS-MFI). The superstructure methodology is found to be capable of identifying the optimal operating policy for different objective functions, and can save over 60% of the CPU time when compared to the total time needed to carry out individual optimizations for each operating policy.