Simultaneous Solvent Selection and Process Design for Continuous Reaction–Extraction–Crystallization Systems

Abstract

Solvent selection is a crucial decision in many high valueadded chemical manufacturing processes. Computational approaches for solvent selection may substantially reduce the experimental burden during early process development. Furthermore, the selection of optimal operating conditions is closely related to the solvent selection. Computational approaches for simultaneous solvent selection and process design need to balance various trade-offs between solvent-intensive unit operations, which is especially important for continuous processes. This work presents a computational framework involving a generalized thermodynamic framework based on the electrolyte perturbed-chain statistical associating fluid theory (ePC-SAFT) equation of state for the simultaneous selection of solvents and optimization of the operating conditions of continuous processes involving the common sequence of reaction−extraction− crystallization steps with possible recycling of solvents. A predictive activity coefficient model based on group contributions is used for the estimation of the PC-SAFT pure component parameters. The proposed framework is illustrated for the continuous manufacture of dalfampridine. The optimization problem can be solved successfully with a mixed-integer nonlinear programming relaxation strategy, followed by either continuous mapping or a branchand-bound approach for solvent identification. The computational tractability of the proposed computational framework indicates the good potential for applications to industrially relevant cases featuring similar thermodynamic equilibria and complexity.

A Workflow for Crystallization Process Design with Simultaneous Process Optimization and Solvent Selection based on the Perturbed‐Chain Statistical Associating Fluid Theory

“…The number of variables is only slightly higher than the number of equations, which means there is only a small number of free variables. Usually, global NLP solvers are ineffective for models with such characteristics [29], and, thus, a suitable local NLP solver has to be used such as CONOPT [30], which is generally well suitable for NLP models with the above properties [29] and has been reported to be effective for PC-SAFT-based process optimization problems [10][11][12][13][14]. It is common to couple local NLP solvers with multi-start initialization strategies [10][11][12][13][14] in which the same NLP problem is solved multiple times with the model variables being initialized differently on each occasion to identify better solutions.…”

“…The continuous-molecular targeting MINLP solution strategy [17], which has been proven to be suitable for crystallization-related applications [10][11][12][13][14], is applied in Step 4. Three steps need to be followed: 1) relax all the integer variables, i.e., treat the pure component parameters of the solvents as continuous variables and solve the resulting NLP problem, 2) identify a solvent(s) from the solvent database by Taylor series-based mapping, 3) optimize the crystallization conditions for the identified solvent(s).…”

“…An MINLP-based solvent blend design framework for crystallization processes was developed by Adjiman and co-workers [6][7][8], which was based on the statistical associating fluid theory (SAFT)-γ Mie [9] as the unified thermodynamic model. Our group [10][11][12][13][14] developed a simultaneous solvent selection and process optimization framework for crystallization-based processes based on the perturbed-chain statistical associating fluid theory (PC-SAFT) [15,16]. The applicability was demonstrated for several continuous processes involving cooling, evaporative, antisolvent or reactive crystallization steps with downstream solvent purification and recycling.…”

“…Our aim is that a user with a need to purify a certain chemical with crystallization can easily navigate the workflow without having to commit substantial resources in either extensive experimental screening or programming yet can obtain a crystallization process design with good performance. The workflow is based on the solvent selection and process optimization framework that our group previously developed [10][11][12][13][14]. The workflow considers cooling, antisolvent, and evaporative crystallization, which are the most common crystallization methods without reaction.…”

N-Formylsaccharin: A Sweet(able) Formylating Agent in Mechanochemistry

“…It is the major reaction component [ 37 ], representing a high economic cost for the industry and a relevant global issue for public health [ 38 , 39 ]. Unfortunately, solvent removal from a process is not a trivial task as it often seriously affects the reaction outcome [ 40 ]. Mechanochemistry [ 41 ], which IUPAC recently recognized as one of the ten chemical innovations that will change the world [ 42 ], has seriously contributed to overcoming these issues by greening many classical solvent-based procedures [ 43 , 44 , 45 , 46 , 47 , 48 , 49 ].…”