Trajectory optimization of an oscillating industrial two-stage evaporator utilizing a Python-Aspen Plus Dynamics toolchain

•A model-based, dynamic optimization of an industrial evaporator system is presented•Optimization performed with Python toolchain; system modeled in Aspen Plus Dynamics•The SciPy implementation of deterministic derivative-free algorithm COBYLA utilized•Steam consumption trajectory found to minimize oscillations of evaporator system•Variance of oscillations reduced by 99.7 % (including bang-bang penalty) Evaporators are integral parts of many separation processes across production industries, and they need to be well understood in order to be operated well, thereby enabling high resource-efficiency and productivity. In a previous investigation, the effects of disturbing oscillations in a two-stage evaporator system were quantified. In the current study, these oscillations were reduced through trajectory optimization using steam consumption as a temporally discretized decision variable, taking advantage of a dynamic process flowsheet model in Aspen Plus Dynamics (APD) employed as if it were a black-box model. The optimization was performed utilizing a Python-APD toolchain with the SciPy implementation of COBYLA. The optimal trajectory was able to successfully reduce the objective function value (including the product stream mass flow variance and a bang-bang penalty on the trajectory itself) to slightly less than 0.3 % of that of the nominal case, in which a time-invariant steam consumption was employed. This in turn grants opportunities to increase throughput of the process, leading to significant financial gains.