Mamdani Fuzzy Modeling of Flash Vaporization Using a New Concept: Fuzzy Composition Variable

In this paper, a novel modeling method is presented using Mamdani fuzzy approach by introducing a new fuzzy concept, named Fuzzy Composition Variable. Fuzzy Composition Variable tries to combine the mole (or mass) fraction variables of the model in one variable. This variable tackles the problem of a high number of fuzzy rules when using Mamdani approach. The proposed method is capable of modeling complex systems with a large number of variables (including molar composition) with an acceptable performance, bypassing solving various types of mathematical equations governing the system and calculating several parameters. To demonstrate the capabilities of the proposed approach, it was implemented to modeling the equilibrium flash separation of crude oil. The overall prediction accuracy of the model (with a manageable number of rules) was obtained more than 85%, without using an optimization procedure.