What question did this study set out to answer?

This research aims to improve chemical separation processes by integrating molecular and process design methodologies.

April 28, 2026Computers & Chemical Engineering0 citationsOpen Access

Integrated Product and Process Reverse Engineering Design combining Adaptive Molecular Search and Rigorous Separation Process Validation

EPEdoardo ParascandoloUniversité Fédérale de Toulouse Midi-Pyrénées GNGIOVANA CORREIA DE ASSIS NETTOUniversidade de São Paulo IRIvonne Rodriguez-Donis

Key Points

This research aims to improve chemical separation processes by integrating molecular and process design methodologies.
Developed a Computer-Aided Molecular and Process Design (CAMPD) framework for limonene extraction from linalool.
Used a hierarchical fallback strategy for property estimation and enhanced a genetic algorithm with an Adaptive Strategy Manager.
Conducted simulations to compare CAMPD with traditional Computer-Aided Molecular Design (CAMD) methodologies.
CAMPD identified five diol-based alternatives outperforming the benchmark 1,3-butanediol in optimization trials.
Confirmed reductions of 23–29% in Total Annual Costs (TAC), 25–32% in CO2 emissions, and 22–33% in Eco-Indicator (EI99) life cycle index through simulations.

Abstract

Chemical separation process optimality requires addressing suitable molecular properties in addition to process-level parameters. The solvent molecular level challenge is typically handled using traditional Computer-Aided Molecular Design (CAMD). However, CAMD may neglect process-inherent trade-offs by prioritizing chemical properties, it may result in suboptimal decisions. To overcome this, we present a Computer-Aided Molecular and Process Design (CAMPD) framework for the extraction of limonene from linalool in citrus waste valorization. The in-house InBioSynSolv (IBSS) CAMD framework was extended with python models for a Liquid-Liquid Extraction (LLE) column and a Solvent Regeneration (SR) distillation column. Property estimation relied on a hierarchical fallback strategy combining Group Contribution (GC) methods, a neural network model for vapor pressure prediction and a caching strategy to reduce the computational time. Moreover, IBSS genetic algorithm was substituted with an augmented version featuring an Adaptive Strategy Manager (ASM) and Synthetic Accessibility (SA) screening, respectively to avoid population collapse and to orient search towards chemically accessible molecules. A comparison between CAMD and CAMPD rankings revealed that property-based screening alone failed to differentiate among candidates at the process level. Five diol-based alternatives to the industrial benchmark 1,3-butanediol were identified by the CAMPD approach and subsequently optimized through stochastic evolution. External validation through closed-loop Aspen Plus® simulations confirmed that all candidates outperform the benchmark, achieving reductions of 23–29% in Total Annual Costs (TAC), 25–32% in CO2 emissions, and 22–33% in the Eco-Indicator (EI99) life cycle index, proving the practical applicability of the integrated design approach for thermodynamically complex systems.

Ask AI

Helpful

Bookmark

View Full Paper