Designing a clean and efficient air conditioner with AI intervention to optimize energy-exergy interplay

Abstract

Conventional refrigerants in air conditioners (A/Cs) although deliver localized comfort within dwellings, their overall detrimental impact on the environment is an alarming issue. This study investigates eco-friendly alternatives to the refrigerants R410A and R22 (which contribute to global warming and ozone depletion) through a comprehensive analysis of salient parameters such as coefficient of performance (COP), volumetric cooling capacity, and exergetic efficiency (ηex) featuring a sustainably retrofitted vapor compression based A/C. Diverse thermophysical properties of alternate refrigerants yield multiple options for contriving a sustainable A/C. Performance enhancement of the retrofitted system is then realized through a multifaceted genetic algorithm (GA) coupled with an artificial neural network (ANN). Subsequently, the incongruence of optimality between maximum system COP and maximum ηex is dealt by a dual ANN powered non dominated sorting genetic algorithm-2 (NSGA-2) optimizer which provides balanced output in terms of COP (4.37 for L20a and 4.238 for ARM71a refrigerant respectively) and ηex (26.208% for L20a and 25.413% for ARM71a refrigerant respectively). The artificial intelligence (AI) based approach helps comprehend the trade-off between different system performance indices (having different units/ranges of variation) during optimum design selection. Furthermore, the data-driven surrogate model reveals the dominating effect of energy performance over exergy performance of the system, urging for the higher priority of resource allocation for COP upgrade than ηex upgrade. Finally, the multi-objective optimization yields a broader set of Pareto optimal points which offer flexibility to the stakeholders to thrust the sustainable system towards higher COP or higher ηex mode of operation.