While the need for more efficient and integrated processes for the separation of non-ideal and azeotropic mixtures mandates model-based design methods, the resulting mixed-integer problems are highly nonlinear and particularly hard to solve without proper initialization. In order to handle such complex design problems, two hybrid optimization methods are presented in the current contribution. Both methods rely on a nested approach, which integrates a polylithic modeling and solution strategy, based on the solution of successively refined nonlinear programming problems, with an upper level metaheuristic for the initialization and optimization of discrete design decisions. Both methods are evaluated for two case studies. The optimization of a simple column design problem allows for the direct comparison with an available global deterministic optimization solver for identifying the global optimal solution. The optimization of an extractive distillation process with simultaneous solvent selection and energy integration illustrates the potential complexity that can be handled effectively.

尽管需要更有效，更集成的方法来分离非理想混合物和共沸混合物，这需要基于模型的设计方法，但由此产生的混合整数问题是高度非线性的，如果没有适当的初始化，则很难解决。为了处理这样的复杂设计问题，在当前的贡献中提出了两种混合优化方法。两种方法都依赖于嵌套方法，该方法基于逐步精炼的非线性规划问题的解决方案，结合了多级建模和解决方案策略，并具有用于离散设计决策的初始化和优化的高层元启发式方法。两种方法均经过两种案例评估。简单列设计问题的优化允许与可用的全局确定性优化求解器进行直接比较，以识别全局最优解。同时进行溶剂选择和能量整合的萃取蒸馏工艺的优化说明了可以有效处理的潜在复杂性。