Heat Exchanger Optimization for Next-generation Air-Source Heat Pump Split Systems with Low-GWP High-Glide Zeotropic Mixtures

Air source heat pumps (ASHPs) are a viable and efficient technology for building space conditioning. Due to the ongoing transition to next-generation low-GWP refrigerant alternatives (e.g., HFC/HFO mixtures, hydrocarbons, etc.), research efforts are needed to optimize ASHPs. These mixtures can mitigate flammability concerns while meeting GWP requirements, but can yield lower performance in comparison to legacy HFC refrigerants without careful system optimization. Among these mixtures, zeotropic refrigerant mixtures (ZRMs) with temperature glide show potential for efficient heat exchange processes, making heat exchanger design crucial in maximizing the benefits of high-glide ZRMs in ASHPs.

However, there has been very little experimental study of refrigerant and heat exchanger impacts for next-generation low-GWP high-glide ZRMs, such as R454C, in residential ASHPs. Previous studies using a drop-in approach led to unfair comparisons among alternative refrigerants with different volumetric capacities, leaving the potential improvements of heat exchanger performance unclear. To address this gap, experimental evaluations were conducted in this study using an off-the-shelf R410A system modified for R454C with a compatible compressor. Experimental results showed that although the R454C compressor provided a higher mass flow rate, the system's capacity and efficiency were lower compared to the R410A system. Additionally, significant differences in refrigerant charge between cooling and heating modes indicate the need for heat exchanger optimization to make R454C a viable alternative refrigerant for this heat pump.

To further investigate and optimize heat exchangers, a detailed steady-state heat pump system model was developed and validated using experimental data. The model includes discretized heat exchanger models with up-to-date correlations for high-glide ZRMs, a dimensionless compressor model, an electrical expansion device model, and line-set models. Parametric studies showed that refrigerant-to-air flow pattern, heat exchanger structural and circuitry designs, as well as airflow rate, collectively impact system performance. Implementing a cross-counterflow pattern and appropriate sizing led to a 16.1% increase in cooling capacity and a 6.7% increase in heating capacity compared to the baseline system. Additionally, the issue of refrigerant charge imbalance was resolved.

To advance research on heat exchanger optimization for high-glide ZRMs, a machine learning-assisted multi-objective optimization framework has been proposed for optimizing FTHXs in reversible ASHPs using high-glide ZRMs. Structural, circuitry, and airflow parameters are simultaneously optimized to maximize system efficiencies (COP, SEER2, and HSPF2) while minimizing HX material costs and maintaining capacity targets. Data-driven surrogate metamodels using ANNs significantly reduce computational costs, achieving a 95% reduction while maintaining high prediction accuracy. Case studies optimizing the baseline AHU and ODU for R454C show that the optimized heat pump system increases HSPF2 by 15% over the R410A baseline and 21% over the R454C drop-in system at the baseline HX cost, while SEER2 improves by 12% and 32%, respectively, when enforcing a 3 RT capacity target and no charge difference. Relaxing capacity and charge constraints further enhances performance but results in overcapacity and minor charge imbalances. Optimized FTHXs for R454C suggest that increasing the row number, applying appropriate circuitry configurations, optimizing tube spacing, and controlling fan speed can collectively enhance performance by improving temperature glide matching and avoiding excessive fan power due to increased air pressure drop.

The framework was validated experimentally by testing an optimized AHU selected based on manufacturer availability, with results aligning well with numerical simulations. The selected AHU significantly improved heat pump performance and reduced the optimal charge difference between cooling and heating modes to zero. The rated SEER2 and HSPF2 showed increases of 23.8% and 4.6%, respectively, over the R454C drop-in results. These experimental results validate the capability of the developed optimization framework. Finally, experimental findings highlighted that appropriate design of the header and distributor in conventional FTHXs is necessary in future work to enable refrigerant flow reversal and maintain efficient cross-counterflow operation.