Simultaneous synthesis and optimization of refrigeration cycles and heat exchangers networks

• Published in: Applied thermal engineering 2022-04, Vol.206, p.118052, Article 118052
• Main Authors: Martinelli, Matteo, Elsido, Cristina, Grossmann, Ignacio E., Martelli, Emanuele
• Format: Article
• Language: English
• Subjects: Algorithms; Compressors; Design optimization; Heat exchangers; Heat transfer; Mixed integer; Optimization; Power consumption; Real variables; Refrigeration; Superstructures; Synthesis; Throttling
• ISSN: 1359-4311; 1873-5606
• DOI: 10.1016/j.applthermaleng.2022.118052

•Refrigeration cycle structure, pressures/temperatures and HEN are optimized.•An ad hoc decomposition algorithm for the problem solution is devised.•Results indicate considerable cost (−40%) and power consumption (−52%)•Major advantages found for case studies with multiple hot and cold streams.•Even if the problem size is large, the required computational time is practicable. This work proposes a simultaneous approach for the synthesis and design optimization of industrial refrigeration cycles integrated with heat exchanger networks. Given the process hot and cold streams, the methodology determines the economically optimal refrigeration cycle configuration (number of evaporation/condensation levels, compressor intercooling, multiple throttling, etc.),cycle variables and heat exchanger network. The methodology includes a novel refrigeration cycle superstructure capable of reproducing a wide range of cycle architectures and an effective solution algorithm (based on the decomposition of the problem on two levels) to tackle the challenging Mixed Integer Non-Linear Program. The application to four literature case studies indicates that the proposed approach returns solutions which are considerably better both in terms of efficiency and economics than those published in literature. The application to four literature case studies indicate that the required computation time is tractable even for problems with thousands of variables and constraints (up to 87,000 real variables, 29,000 binary variables and 129,000 equations). Moreover, compared to previous literature studies, the optimized solutions feature a total annual cost reduction up to 40% and a decrease in compression power consumption up to 52%.