The prediction of physical properties of pure components using group contributions methods.

  • K. S. Lim

Student thesis: Doctoral ThesisPhD


The implementation of a physical property prediction procedure in a microcomputer for use in computer-assisted education has been investigated. A literature survey of existing group contribution methods for the prediction of a few selected properties has been performed resulting in the selection of an appropriate model for the prediction of each of the discussed properties. This selection is based both on the accuracy of prediction of the methods and the extent of commonality of the fragments used by them. Improvement in the accuracy of prediction for the heat capacities of ideal gases using Doraisamy's model has been achieved. The extent of commonality among the accepted models is also discussed. A compilation and statistical analysis of the pure component vapour pressure data has been undertaken in order to relate the basic parameters in the AMP kinetic model of the liquid state to molecular structures. Attempts to develop a general model fail to predict the vapour pressures of the compounds to the desired accuracy and as such, models are based on individual homologous series. The general model is accurate to a factor of 2 which is of the same order of accuracy as Edward's. A general model based on the Antoine equation has been derived but this would require the user to input the experimental normal boiling point of the compound. An APPLESOFT BASIC program it has been developed to process the IUPAC name of the compound in order to construct a 2-dimensional matrix representation of the molecule. The program then proceeds to construct a Boolean matrix which relates the connections between nodes with each bond as a bi-directional edge. Nodes separated by any number of bonds may then be identified by adapting the concept of the adjacency matrix used to identify recycle streams in steady-state process analysis. Fragments/subgroups are then identified by comparing the features of each node with the unique characteristics of those predetermined fragments.
Date of Award1985
Original languageEnglish
Awarding Institution
  • University of Bath

Cite this