(381bb) Computer-Aided Design of Crystallization Solvents for the Recovery of High-Purity 2- Mercapotobenzothiazole Conference: AIChE Annual MeetingYear: 2019Proceeding: 2019 AIChE Annual MeetingGroup: Separations DivisionSession: Poster Session: Separations Division Time: Tuesday, November 12, 2019 - 3:30pm-5:00pm Authors: Du, J., Dalian University of Technology Chai, S., Dalian University of Technology Liu, Q., Dalian University of Technology Liang, X., Dalian University of Technology Guo, Y., Dalian University of Technology Zhang, S., China Sunsine Chemical Holdings Xu, C., China Sunsine Chemical Holdings Zhang, L., Dalian University of Technology Yuan, Z., Tsinghua University Gani, R., Technical University of Denmark 2-Mercapotobenzothiazole (MBT) is an important vulcanization accelerator, which is widely used in rubber industry. The production of high purity MBT mainly includes acid-base method and solvent-based method. In recent years, due to the restriction of environmental protection policies, the application of acid-base method has been greatly restricted. In solvent-based method, it is important to select a suitable crystallization solvent to ensure the refined MBT meets the industrial requirements. Therefore, a Computer-Aided Molecular Design (CAMD) model for designing crystallization solvents is proposed. The proposed CAMD problem is express as a Mixed-Integer Non-Linear Programming model, which contains objective functions, structural constraints, property constraints and process constraints. The objective functions are the potential yield and the purity of the MBT product. Other solvent-related properties are taken as property constraints, including molecular weight, normal melting point, normal boiling point, flash point, toxicity, solubility parameters, and SLE (solid-liquid equilibrium), etc. Among them, the activity coefficient related to the SLE is predicted by the COSMO-SAC model, and the other properties are predicted by the Group Contribution method. The model is then solved by the decomposition-based approach to obtain the optimal solvents, which are verified by experiments. Through the crystallization experiments, the effects of solvent ratio, initial crystallization temperature and the number of mother liquor cycles on the purity and yield of MBT product are investigated. Meanwhile, the experimental results are compared with the model results. Finally, optimal solvents are selected, which perform better results than the currently reported solvents for crystallization. The effectiveness of the model is verified, and it also provides guidance for the design of other crystallization solvents.