COMPUTER-AIDED MIXTURE DESIGN WITH SPECIFIED PROPERTY CONSTRAINTS

This paper presents an algorithm that determines the identity and composition of solvent mixtures with desirable properties and also ranks the mixtures with respect to cost. This is done by integrating optimization algorithms with computer-aided mixture design (CAMD). Thus, a non-linear optimization problem with linear and non-linear constraints is solved. The optimization variables are compounds in a data bank and the mixture compositions. The object function is a cost function, and the constraints are different thermodynamic properties such as boiling point, density, and solubility parameters. Because of the nature of the derivatives of the non-linear constraints with respect to the optimization variables, it has not been possible to employ standard non-linear optimization algorithms. Instead, a two-loop optimization algorithm, called successive regression and linear programming (SRLP), has been developed. This paper will describe the SRLP algorithm and provide examples of application. The numerical examples will include processes of industrial interest where minimum cost solvent blends need to be determined.