Efficiently solving large-scale heat exchanger networks (HENs) remains a changing task due to the non-convex and nonlinearity characteristics. A novel methodology for optimal design of HEN combined structure design model based on pinch analysis and automatically converted the structure into associating topology in the superstructure is developed, subsequently as the initial point to optimize the superstructure-based model. Various InHEN structures obtained by changing âˆ†Tmin, and results of initial HEN (InHEN) model satisfy theoretical utility. The conversion model uses a temperature comparison-based method to locate the exchanger position, and distance-based method proposed in this study achieve updating the location. The reduced MINLP to NLP model through the strategy of replacing binary variables with absolute values in the superstructure is presented. A weakening strategy is proposed to simplify the computation of temperature differences. The TAC as the objective is then solved with various initial points, and the near globally optimal solution is obtained. The framework of optimization with initial points is named PinNLP model. The complex large-scale case is investigated for proving the methodâ€™s applicability and advantages in solving quality and time, and the results show that the optimal is close to the global optimal solution. The proposed methodology can be applied to industrial HEN synthesis and thus has a wide space of applications.
Keywords heat exchanger networks, pinch analysis, mathematical modeling, NLP, design optimization