Group contribution method

Group contribution methods (also referred to as a fragment methods) are in technical chemistry widely used method for the estimation of material data.

Method

Chemical properties that are required for instance in the process simulation, are always properties of a substance or mixture of substances. Since there are an almost endless and exponentially further increasing number of pure substances and mixtures, group contribution methods have been developed, no longer assign the material properties throughout the material, but fragments.

The final effect is that of a few group properties, typically a dozen to a few hundred, the material data for many thousands of substances and their mixtures can be determined.

These fragments ( the groups ) are generally the functional groups of a molecule such as the hydroxyl group (-OH), amino group (-NH2) or carboxyl (-COOH). Often, as a group, other molecular characteristics should be taken such as ortho-/meta-/para-Stellungen aromatics, ring sizes and chain lengths.

Pure component properties

A property, as an example, the critical pressure is calculated on the sum of the contributions.

Typical pure component parameters such as critical temperature, critical pressure, critical volume, normal boiling point, heat capacities, viscosities, phase transition heats are estimated from this simple relationship.

Mixture properties

For models used to estimate properties of mixtures, not only the sum of the group contributions are often used but group interaction parameters and used.

A property that is typically calculated through group interaction models such as UNIFAC or ASOG, is the activity coefficient.

A negative effect of the use of group interaction is the massive increase in the required parameters. For 10 groups interaction parameters are, for example, already requires. Therefore, group interaction models are usually not fully parameterized.

Determination of group contributions

The group contributions are usually directly matched to experimentally determined material data using multi linear or nonlinear regression. Non-linear regressions provide a rule is multi-modal optimization problems, ie, optimization problems with more than one optimum in the solution space considered. For adaptation of group interaction parameters, therefore, are often Evolutionary algorithms (eg (nested ) evolution strategies, genetic algorithms, etc. ) are used as deterministic optimization algorithms are not usually in a position to the global optimum ( in regressions: minimum) to find.

As a data base of experimentally determined material data used, for example Factual databases such as Beilstein, the Dortmund Data Bank or the DIPPR 801 database. Often experimental measurements are carried out to supplement if there are gaps in the considered group interaction matrix or additionally describe group contribution methods, a temperature and / or pressure dependence.