Circuitry optimization using genetic programming for the advancement of next generation refrigerants

In this study, a new evolutionary method, which can handle the implementation of genetic operators with unrestrained number and locations of splitting and merging nodes for the optimization of heat exchanger circuitries, is developed. Accordingly, this technique expands the search space of previous optimization studies. To this end, a finned-tube heat exchanger simulator is structured around a bijective mathematical representation of a refrigerant circuitry (the tube–tube adjacency matrix), which is used in combination with traversing algorithms from graph theory to recognize infeasible circuitries and constrain the evolutionary search to coherent and feasible offspring. The performance of three refrigerants, namely R32, R410A, and R454C, commonly used in air-conditioning applications was assessed for the optimized circuitries of a 36-tube evaporator while converging to a given cooling capacity, degree of superheating, and heat source boundary conditions. At a given output capacity and air outlet temperature, larger coefficient-of-performance improvements (up to 9.99% with reference to a common serpentine configuration) were realized for zeotropic refrigerant mixtures, such as R454C, where appropriate matching of the temperature glide with the temperature variation of the air yielded the possibility of further reducing the required compression ratio under the corresponding operating conditions. Hence, it was demonstrated that low-GWP zeotropic mixtures with temperature glide can realize a performance comparable to that of R32 and higher than that of R410A by approaching the Lorenz cycle operation.