Repository logo
 

Solvent selection methods in aromatic extraction processes towards optimal and sustainable design choices.

Thumbnail Image

Date

2023

Journal Title

Journal ISSN

Volume Title

Publisher

Abstract

This study presents a novel process decision-making framework that amalgamates Computer-Based Molecular Design (CAMD) of solvents using group contribution methods and artificial neural network Quantitative Structure Property Relationship (QSPR) models with Health, Safety, and Environmental (HSE) impacts using a rating-based risk assessment method. These then culminate in integrated solvent selections and process designs with relatively fixed configurations that capture the interaction of solvent choice on design alternatives, with the incorporation of sustainability measures. The objectives are to provide a platform based on systematic methods on which engineers can optimize solvent extraction processes by simultaneously considering all interlinking steps of process systems engineering from molecular design and selection to the optimization of the economics of large-scale processes, in the larger context of sustainable development. In applying the proposed framework to the liquid-liquid extraction of aromatics from aliphatics, a selection of organic chemicals was proposed for further study as replacement solvents that meet the technological requirements for the process of aromatics extraction from alkanes and have not been conventionally considered previously for this application. Organic chemicals were screened with the use of a search algorithm based on process requirements in terms of physical properties, capacity, selectivity, and performance index, using UNIFAC-LLE. A risk assessment was constructed to further screen the identified new potential solvents using a ratings-based structure using HSE data. Thus, solvent impact in terms of sustainability were systematized in a manner that related thermodynamic theoretical predictions using molecular knowledge to that of HSE risk assessments using a benchmarking rating system. Limitations in the group contribution models led to the need for concurrent screening methods to be used as a supplement to the predictions of UNIFAC-LLE in order to improve the robustness of the approach. As a result, a QSPR model was developed using artificial neural networks to estimate the binary interaction parameters for the temperature-dependent form of the NRTL model with the objective of using it to complement group contribution methods in the screening of potential solvents for liquid-liquid extraction processes. Parameters were regressed using experimental LLE and VLE data and checked for consistency. Molecule structures were drawn, and descriptors determined with the use of Materials Studio. The QSPR model uses 31 descriptors as input and produced absolute average deviations of 0.23 and 0.19 for each pair of the NRTL dimensionless interaction parameterss τij and τji respectively. A novel method of solvent screening using these a priori NRTL interaction parameters is also presented in which a ratio is defined that incorporates values of the interaction parameters based on extent of ideality in relation to phase equilibria for the various binary pairs in a ternary mixture. Validation of CAMD solvent screening results were done with the use of novel experimentally measured liquid-liquid equilibrium (LLE) phase compositions that were thermodynamically modelled for the ternary systems n-heptane + toluene + (butane-1,4-diol or glycerol) and n-nonane + o-xylene + (butane-1,4-diol or glycerol) at 298.2, 313.2 and 333.2 K and 0.1 MPa. n-Heptane represents the alkane component and toluene the aromatic component. The selectivity for the solvents studied were found to be comparable or superior to conventional solvents, but results of the ternary systems indicated that solvent capacities were poor, motivating for the use of a cosolvent to reduce solvent-to-feed ratios. The use of 2-methyl-2,4-pentanediol as a potential co-solvent to butane-1,4-diol and glycerol was then studied in order to investigate its impact on solvent capacity in the separation of toluene from n-heptane via liquid-liquid extraction. To this end, quaternary liquid-liquid equilibrium (LLE) data was experimentally measured for the system n-heptane + toluene + (butane-1,4-diol or glycerol) + 2-methyl-2,4-pentanediol at 298.2 and 313.2 K and 0.1 MPa. All measurements were conducted in a double-walled glass cell using the direct analytical method, and phase compositions were analysed via gas chromatography. The data was correlated using the NRTL and UNIQUAC models, which were able to suitably represent the tie-line compositions. At different molar ratios of solvent to co-solvent, the selectivity and solvent capacity were calculated and compared to that of using pure butane-1,4-diol or glycerol. It was ascertained that the pseudo ternary systems possess type I or type II behaviours depending on the molar ratio. It was observed that solvent capacity is not appreciatively improved for molar ratios that exhibit type II behaviour. However significant increases in capacity were noted for high molar ratios producing a type I system. Process designs were developed with the use of ASPEN Plus V10 in order to ascertain the effects of solvent choice on process economics via the use of total annual costs. The screening process in this work produced significant insights due to its holistic approach. The incorporation of factors such as solvent price, solvent loss, utilities, capital costs, health and environmental impact, showed that several of the solvents identified may be sustainable and cost-effective alternatives to conventionally used solvents. Conversion of energy sources and consumptions to equivalent carbon emissions closes the sustainability loop and associates solvent selection directly with largescale environmental impact. In this manner, all steps of Process Systems Engineering (PSE) are combined in an approach that relates molecular solvent design with sustainable development and large-scale process optimization in a decision-making framework that may be effectively used to determine optimal design choices and inform efficient process retrofitting.

Description

Doctoral Degree. University of KwaZulu-Natal, Durban.

Keywords

Citation

DOI