To obtain aromatic compounds from a crude mixture such as reformate or pyrolysis gasoline, three different processes are simulated with the realistic composition of reformate and product specification. Simulations were performed by Aspen Plus supported with COSMO-RS method to predict the physical and thermodynamic properties of ionic liquid. Furthermore, utility analysis and economic evaluation are presented. Conventionally, aromatic compounds are extracted from a crude mixture either by extraction or by extractive distillation using a solvent such as sulfolane and separated by a series of distillation columns. In this study, the sulfolane-based commercial process is first introduced, and two novel processes that use 4-methyl-N-butylpyridinium tetrafluoroborate ionic liquid as solvent are proposed. The second process shows that energy consumption has successfully decreased, but the high price of ionic liquid offset the cost advantage. The third process is therefore proposed to reduce the amount of ionic liquid using two extractions. A similar level of energy saving is achieved with reduced costs.

No abstract

No abstract

No abstract

No abstract

No abstract

No abstract

As the digital transformation of the bioprocess is progressing, several studies propose to apply data‐based methods to obtain a substrate feeding strategy that minimizes the operating cost of a semi‐batch bioreactor. However, the negligent application of model‐free reinforcement learning (RL) has a high chance to fail on improving the existing control policy because the available amount of data is limited. In this article, we propose an integrated algorithm of double‐deep Q‐network and model predictive control. The proposed method learns the action‐value function in an off‐policy fashion and solves the model‐based optimal control problem where the terminal cost is assigned by the action‐value function. For simulation study, the proposed method, model‐based method, and model‐free methods are applied to the industrial scale penicillin process. The results show that the proposed method outperforms other methods, and it can learn with fewer data than model‐free RL algorithms.