The effective and secure supply of pure substance and miscibility properties represents the essential basis for development, design and optimization of engineering plants. In this context the supply of phase equilibrium properties plays a special role. In general it will be performed by means of excess models and equations of states, whose model parameters have to be fitted on experimental data in an extensive process of modelling. Especially in consideration of multiple liquid phases (e. g. Liquid-Liquid-Equilibrium, Liquid-Liquid-Vapour-Equilibrium) problems can occur, which cannot be solved by means of standard methods in a closed manner. In view of this background, the concept for an heuristic-numerical computer system was developed, which is based on a systematical and generalized investigation of the whole process of modelling. In detail, it will be characterized by the the supply of pure substance and miscibility properties, prediction of properties (estimation of miscibility behaviour of binary solubility systems), fitting of parameters, equilibrium calculation and visualization of data. Concerning these points known methods from different sources and results from our own investigations were summarized and evaluated. Strategies were derived for a generalized and optimized procedure for modelling of phase equilibrium. The knowledge based system POLPERT was developed by means of the logical programming language Prolog, so that the principles for modelling of multicomponent-multiphase-equilibrium presented in this work are applicably in practice.