Sustainability assessment using local lazy learning: The case of post-combustion CO 2  capture solvents

Shavalieva, Gulnara; Παπαδόπουλος, Αθανάσιος Ι.; Badr, Sara; Seferlis, Panos; Papadokonstantakis, Stavros

doi:10.1016/b978-0-444-64241-7.50132-4

Cited by 3 publications

(1 citation statement)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Some of the most known examples of these models are COSMO-RS and CODESSA. − More recently, machine learning approaches have been combined with the QSPR approach with improvements achieved in terms of predictive capability. − However, the high computational cost of QSPR methods utilizing quantum-based descriptors, combined with the “black box” nature of machine learning algorithms, still presents a barrier to the scalability of such models for computer-aided molecular design (CAMD) to very large search spaces. An alternative approach is the group contribution method (GCM), which is based on the assumption that the structural functional groups that make up the chemical species make defined contributions toward the overall properties. , As these functional groups can also be descriptors in a QSPR model, GCMs can be considered to be a special case of a QSPR model . Because they rely only on simple structural information, rather than complex descriptors, GCMs are much less computationally intensive and much more flexible in their application.…”

Section: Introductionmentioning

confidence: 99%

Ionic Liquid Melting Points: Structure–Property Analysis and New Hybrid Group Contribution Model

Mital

Nancarrow

Ibrahim

et al. 2022

Ind. Eng. Chem. Res.

View full text Add to dashboard Cite

Melting point (T m) is one of the defining characteristics of ionic liquids (ILs) and is often one of the most important factors in their selection for applications in separation processes, lubrication, or thermal energy storage. Due to the almost limitless number of theoretically possible ILs, each with incrementally different physiochemical properties, there is significant scope for designing ILs for specific applications. However, the need for extensive synthesis and experimental characterization to find the optimum IL is a major barrier. Therefore, it is essential that predictive tools are developed for estimating the physiochemical properties of ILs. The starting point for any such approach should be the prediction of T m since most other property models will be based on the assumption that the IL is in the liquid phase at the application temperature. While several attempts have previously been made at developing group contribution methods (GCMs) for estimating IL T m, the complex relationship between the IL structure and T m has resulted in only limited success. In this study, an extensive database of IL T m has been compiled and used as the basis for a top-down structure–property analysis. Based on the findings, a new hybrid GCM has been developed, which combines functional group parameters with simple, indirect structural parameters derived from the structure–property analysis. The new hybrid GCM has a mean absolute percentage error (MAPE) of 8.6% over the dataset of around 1700 data points and performs quantitatively and qualitatively better than the standard GCM approach.

show abstract