This paper presents the local thermodynamic models networks (LTMN), a new technique aimed at reducing the time spent in simulating processes that are based on phase equilibrium. The LTMN shown in this work approximate K-values as generated by a conventional property method, with a sensible reduction of the computer burden. The LTMN combine the local thermodynamic models technique and the concepts of multiple modeling to provide a valid approximation on the entire domain of the independent thermodynamic state variables. The paper discusses topics from the local models employed in the LTMN and the division of the space to the interpolation algorithm. Additionally, a new polynomial-time algorithm for the calculation of the bubble point temperature is developed on the basis of the form of the thermodynamic local models. The methods presented are tested in the dynamic simulation of distinct separation processes, for two distinct multicomponent mixtures, with time savings Lip to 90%.