Global optimization of distillation columns using surrogate models

Area of Science:

• Chemical Engineering
• Optimization Techniques

Background:

• Distillation column optimization is crucial for process efficiency.
• Deterministic global optimization is needed to avoid local minima in complex problems.
• Mixed-integer nonlinear programming (MINLP) challenges arise in ideal and non-ideal column design.

Purpose of the Study:

• To investigate surrogate-based optimization of distillation columns.
• To employ an iterative Kriging approach for deterministic global optimization.
• To determine optimal setups and operating conditions for distillation processes.

Main Methods:

• Utilized an iterative Kriging approach as a surrogate model.
• Applied deterministic global optimization to avoid local optima.
• Conducted case studies on ideal and non-ideal distillation columns formulated as MINLP problems.

Main Results:

• The adapted Kriging approach achieved results comparable to direct global optimization for ideal columns.
• Significant improvements were observed for non-ideal distillation columns compared to multistart local optimization.
• The method effectively addressed MINLP challenges in distillation column design.

Conclusions:

• Iterative Kriging is an effective surrogate-based optimization strategy for distillation columns.
• This approach offers a robust solution for complex, non-ideal distillation systems.
• The Kriging method enhances the reliability of finding global optima in chemical process optimization.