Experimental Investigation and Data-Driven Modeling for Variable Refrigerant Systems
by Po-Ching Hsu
Variable refrigerant flow (VRF) systems have gained widespread adoption in commercial and residential buildings due to their high part-load efficiency, zoning flexibility, and superior thermal comfort performance. As buildings account for a substantial portion of global energy consumption and carbon emissions, improving the modeling, control, and operational efficiency of VRF systems has become increasingly important. However, the growing structural and operational complexity of modern VRF systems, characterized by variable-speed compressors, multiple indoor units, electronic expansion valves, and heat recovery configurations, poses significant challenges for accurate system modeling and advanced control implementation. Traditional curve-based models, while computationally efficient and physically interpretable, often lack sufficient control-oriented inputs and struggle to capture transient dynamics. Conversely, purely data-driven models demonstrate strong predictive performance but may suffer from limited physical consistency, robustness, and generalizability. These limitations highlight the need for a hybrid modeling framework that combines the strengths of both approaches.
This dissertation presents a comprehensive experimental and modeling investigation of VRF systems based on long-term field test data. A multi-functional VRF (MFVRF) system installed in a campus office building was instrumented to capture detailed operational data across diverse seasonal and operating conditions. The resulting database enabled exploratory analysis of system behavior, part-load performance, and the impact of control strategies on energy consumption and thermal comfort.
Building on this dataset, multiple data-driven models were developed to capture VRF system behavior, including decision trees (DTs), artificial neural networks (ANNs), and long short-term memory (LSTMs). A systematic evaluation framework was established to assess predictive accuracy, physical consistency, computational efficiency, and model compactness. Results demonstrate that deep learning models, particularly LSTM-based architectures, effectively capture transient and nonlinear system behavior, achieving higher predictive accuracy than shallow machine learning (ML) models. To enhance robustness and physical interpretability, a hybrid modeling framework was proposed by integrating a modified VRF-SysCurve physics-based model with ML models such as ANN. The hybrid structure preserves key thermodynamic relationships while incorporating control-oriented variables, enabling improved predictive accuracy and greater flexibility for control applications. Transfer learning strategies were implemented to fine-tune sub-models for underrepresented indoor units using limited additional data, thereby improving generalizability and data efficiency. Compared with standalone curve-based and purely data-driven models, the proposed hybrid model demonstrates superior physical consistency and reduced performance degradation under extrapolated conditions.
The developed hybrid model was further embedded within a model predictive control (MPC) framework to evaluate advanced control strategies. The MPC implementation effectively regulates control inputs to reduce power consumption while maintaining thermal comfort. Pareto-front analyses reveal the trade-offs between energy savings and comfort objectives across cooling and heating seasons.
Overall, this research establishes an end-to-end digital-twin framework for VRF systems that integrates field experimentation, data-driven modeling, hybrid physics-informed modeling, and optimized control. The proposed framework advances the accuracy, robustness, and practical deployability of VRF modeling for real-time control and smart building applications.
Future work will enhance data representativeness for underutilized indoor units (IDUs) through targeted experiments to improve model generalization in rarely observed operating regions. The scalability of the proposed framework will be evaluated across VRF systems with different configurations. Multi-domain co-simulation integrating building thermal and distribution-network models will be explored to assess demand-side flexibility and grid-interactive performance. Finally, field implementation of the proposed MPC strategy will be conducted to validate energy savings, comfort impact, robustness, and operational reliability under practical conditions.
Doctoral Dissertation
Modeling and Performance Evaluation of Cold Climate Heat Pump Systems in Commercial and Residential Buildings
by Dhahyun Kang
Heat pumps are increasingly recognized as a critical technology for building decarbonization due to their ability to provide space heating and cooling with efficiencies well above those of conventional fossil-fuel systems. However, their adoption for electrified heating in cold climates remains limited in both residential and commercial applications, owing to capacity degradation and efficiency losses at low ambient temperatures. While substantial research exists on the component-level performance of cold climate heat pumps (CCHPs), there is a lack of building-level analysis evaluating their operation under real-world cold climate conditions. Additionally, few studies have incorporated the performance of low-GWP (Global Warming Potential) refrigerant alternatives directly into whole-building energy models or examined the feasibility of integrated cascaded heat pump systems for simultaneous space heating and domestic hot water production. This thesis addresses these gaps through two complementary studies.
The first evaluates the quasi-steady-state heating performance of a commercially available air-source rooftop heat pump in a modeled outpatient healthcare facility located in ASHRAE Climate Zone 5B, using EnergyPlus™ as the simulation engine. R-410A serves as the baseline refrigerant, with R-454B, R-32, R-290, and R-1234yf assessed as low-GWP alternatives using literature-based performance correction factors applied under consistent system boundary conditions. R-32 emerges as the most promising alternative, offering improved SCOP over R- 410A with comparable capacity and near-term market feasibility as an A2L refrigerant. Although R-290 achieved the highest SCOP and greatest cost savings, its capacity loss at low ambient temperatures and charge limitation constraints reduce its suitability for large-scale commercial adoption. R-454B performed comparably to R-410A with minimal efficiency penalty, making it a viable near-term drop-in replacement. R-1234yf showed higher winter electricity consumption but exhibited greater efficiency during warmer months, suggesting it is better suited to mild- climate applications. A Life Cycle Climate Performance (LCCP) analysis confirmed that indirect emissions from energy use dominate total climate impact, making system efficiency a more critical factor than refrigerant GWP alone.
The second study develops and validates a field-calibrated BEopt / EnergyPlus™ model for a 382.4 m 2 (4,116 ft² ) single-family residence in Ellicott City, MD (ASHRAE Climate Zone 4A), and uses it to evaluate the energy, economic, and environmental performance of a gas-to- electric retrofit. The retrofit replaces a natural gas furnace and gas-powered water heater with a cascaded R-32 variable refrigerant flow ( VRF) system and R-134a heat pump water heater (HPWH). The model was validated against field measurements collected during the 2024–2025 heating season, with simulation errors within the ±15% tolerance specified by ASHRAE Guideline 14. Field data were used to calculate the COP of the cascaded system across three water heater setpoint temperatures, with the system COP ranging from 3.13 at 55°C to 2.22 at 65°C, confirming that setpoint temperature is a meaningful driver of water heating efficiency. Utility bill analysis showed source energy savings of up to 50% in the peak winter months, though overall utility costs increased due to the higher per-unit electricity costs in the Maryland market. Performance projections across five U.S. climate regions showed annual source energy savings of 11.1% to 37.0% and CO2 emissions reductions of 26% to 64%, with the strongest results in colder climates. Life-cycle cost analysis over a 30-year horizon indicates that the retrofit carries a modest cost premium, the smallest in high-HDD climates where energy savings are greatest.
Together, these results demonstrate that both low-GWP refrigerant substitution and cascaded heat pump retrofits represent technically viable, energy-efficient pathways toward building decarbonization, and highlight the importance of building-level analysis over component-level evaluation alone.
Master's Thesis
A Silicon and Silicon Carbide Bonding Methodology for Cooling High Heat Flux Electronics
by Kyle Martin
This thesis involves the development of a novel bonding procedure for Silicon and Silicon Carbide substrates used in microfluidic cooling devices for high-heat-flux electronics. As power densities in computer chips increase, cooling devices with small, complex structures need to be developed. Silicon and Silicon Carbide are two of the most popular material choices for these devices. Many cooling devices made from these materials require effective bonding between their components to prevent leakage of the process fluids used. While extremely important, leak-proof bonding at elevated pressures has proven difficult to achieve. To address this challenge, a novel bonding procedure for a Silicon manifold-microchannel heat sink was developed.
Since the components of the manifold-microchannel heat sink were made from Silicon wafers with microscale etched geometries, solder-assisted wafer bonding was chosen as the bonding method. A unique multi-layer solder comprising three different metals was developed as the bonding material. After depositing the material onto the manifold and microchannels, a novel process using a Rapid Thermal Annealer (RTA) was developed to bond the components.
After the manifold and microchannel components were bonded to form the complete cooling device, pressure testing was performed to assess the bond strength. These tests revealed that the device was leak-proof up to 60 psi and could likely support pressures up to higher pressures if needed. While the pressure-tested device used microchannels and manifolds, both made of Silicon, recent tests indicate that the developed bonding procedure is also effective for both Si-to-SiC and SiC-to-SiC substrates.
Master's Thesis
Development and Optimization of Low-GWP Heat Pump Systems for Industrial Wood Drying
by Tamoy Seabourne
Wood drying is a quality-critical and energy-intensive process in lumber manufacturing, and it remains an important target for industrial electrification and decarbonization. Dehumidification kilns using vapor-compression heat pumps can remove moisture by condensing water at an evaporator and reheating the dried air at a condenser. Still, kiln schedules impose a wide operating envelope that challenges conventional low-temperature heat pump designs, especially under large temperature lift and evolving latent loads. This thesis develops a schedule-aware, component-based modelling framework for heat-pump-driven wood drying and couples it with steady-state experimental evaluation to quantify performance, operating constraints, and improvement pathways. A system-level quasi–steady-state model is constructed to link kiln schedule targets to psychrometric states, coil duties, electrical input, and the specific moisture extraction rate (SMER), using air-side bypass and airflow assumptions consistent with industrial operation and a 10-coefficient heat pump performance representation for computationally efficient cycle simulation. The model is validated against industrial drying data, showing good agreement in moisture content evolution and capturing characteristic declines in drying rate, dryer effectiveness, and SMER over time. An experimental campaign using a closed-loop facility with a virtual load establishes baseline performance of a commercial dehumidification unit (R-134a). It evaluates targeted upgrades, including low-GWP refrigerant substitution, resized heat exchangers, an expansion valve, and high-temperature-capable air handling. A direct comparison at 40 °C and 50% RH shows an increased heating capacity of 6.3 kW and an improved SMER of 1.8, with a comparable COP to the baseline that has a heating capacity of 5 kW and a SMER of 1.65. The upgraded configuration demonstrates stable operation over a broader operating range, including conditions above the conventional low-temperature envelope and up to conditions of 63 oC and 28% RH, demonstrating a SMER of 1.5 with a capacity of 7.1 kW.
Finally, steady-state test results are integrated into the schedule-driven simulation framework to assess full-drying implications, indicating that the upgraded system can maintain schedule-consistent moisture trajectories while reducing total power by reducing auxiliary heating demand and increasing SMER during key portions of the drying process. Overall, this thesis provides an experimentally anchored modelling workflow for evaluating and improving heat pump wood drying systems under realistic schedule variation and highlights design levers such as evaporator approach temperature, pressure drop, airflow/bypass control, and cycle architecture that govern energy use and moisture removal efficiency.
Master's Thesis
Top