EURO 2024 Copenhagen
Abstract Submission

EURO-Online login

962. A development of a nonconvex and combinatorial bi-objective programming for molecular design problems

Invited abstract in session TC-37: Multiobjective Mixed-Integer Nonlinear Optimization, stream Multiobjective Optimization.

Tuesday, 12:30-14:00
Room: 33 (building: 306)

Authors (first author is the speaker)

1. Lauren Ye Seol Lee
Department of Chemical Engineering, University College London
2. Amparo Galindo
Department of Chemical Engineering, Imperial College London
3. George Jackson
Department of Chemical Engineering, Imperial College London
4. Claire Adjiman
Chemical Engineering, Imperial College London

Abstract

Multi-objective optimization (MOO) is a technique widely used in engineering to balance different, often conflicting decision criteria that are difficult to compare directly. However, common methods for solving MOO problems, including the weighted sum, normal boundary intersection (NBI), and sandwich methods, often struggle in addressing the complexity of nonconvex and discrete decision variables in engineering designs.

In our research, we introduce a robust approach, referred to as SDNBI, that blends the sandwich algorithm with a modified version of the NBI method (mNBI). This new method is particularly effective in navigating the complex, non-linear parts of the decision space, and in quickly identifying areas where no further optimal solutions can be found. Our study explores the theoretical interplay between mNBI and the sandwich algorithm, focusing on three key aspects: the accuracy of its approximations, the decomposition of objective space based on Pareto front convexity, and its efficiency in circumventing redundant searches in disconnected Pareto segments. The effectiveness of this combined approach is benchmarked against existing MOO methods using standard literature problems and further validated in real-world scenarios such as solvent design for CO2 capture and the design of working fluids in Organic Rankine Cycle processes.

Keywords

Status: accepted

Back to the list of papers