Heat Exchanger Network Design/Retrofit with Partitioning Technique for Linearization of Specific Heat Capacity-Temperature Relation

Abstract

Heat Exchanger Network (HEN) design has been studied for 40 years starting from heuristic technique; pinch analysis and design method to mathematical programming. The stage-wise superstructure by mathematical programming from Yee and Grossmann (1990) was one of the famous models suitable for industrial HEN design under assumption of constant average specific heat capacity (Cp) which generally should be fitted by empirical form of cubic equation. Zhu and Asante (1999) and Ayotte-Sauvé et al. (2017) used piece-wise linearization called stream segment model, to approximate Cp. Their Cp approximation techniques used high number of piece-wises to calculate Cp accurately. This paper presents novel technique called partitioning technique; using less number of piece-wise to approximate temperature-dependent Cp more accurately along with stage-wise superstructure model. The concept is to linearize polynomial cubic equation of Cp as a function of temperature with different Cp approximation techniques from Zhu and Asante (1999) and Ayotte-Sauvé et al. (2017). Cp is approximated as a linear equation for each partition of temperature range. Our technique in weighted average Cp calculation at stage in stage-wise superstructure model is novel, giving more accurate Cp approximation and HEN synthesis at validation step. Our model synthesizes HEN with less total annualized cost (TAC) and exchanger area calculation error between our model and Pro/II simulated HEN compared to other models; constant Cp model and cubic equation technique from Kim and Bagajewicz (2017) represented by four examples. First, crude preheat train example from Pro/II library is used to validate new concept. Normally, constant specific heat capacity directly affects outlet temperature which increase error of area calculation. It shows that constant heat capacity flow rate can reach error of heat exchanger area 30 % but using the new model can reduce the error to less than 1 %. Next, two examples from Kim and Bagajewicz (2017) get better result when compared to previous solution of cubic equation technique. Forth example is to show new model in retrofit case. Therefore, new model has many advantages that it increases accuracy of HEN design and better solution can be obtained from variable of specific heat capacity.