Temperature fluctuations during frozen storage and transportation inevitably cause quality deterioration in frozen foods, primarily through moisture loss and frost formation. While packaging is conventionally viewed as a passive barrier, its potential to actively regulate internal thermofluidic dynamics and extend shelf life remains underutilized. This study investigates the interactive effects of packaging geometry, specifically internal gas layer thickness, mean storage temperature, and temperature fluctuation dynamics on the moisture loss of packaged frozen apples. By integrating a coupled thermofluidic phase change model via computational fluid dynamics (CFD) with experimental measurements, sublimation and desublimation pathways were systematically mapped across varying enclosure configurations.The results demonstrated that dehydration rates are highly sensitive to the packaging’s physical structure. Reducing the fluctuation amplitude from ±5 °C to ±1 °C decreased the dehydration rate by 82.6%. Most notably, optimizing the packaging geometry effectively compensated for quality losses associated with elevated storage temperatures. Increasing the mean temperature from –18°C to –15°C, a strategy for energy-saving cold chain management, typically accelerates dehydration; however, simultaneously reducing the internal gas layer thickness from 1.6 cm to 0.8 cm suppressed this dehydration penalty from 23.8% to 9.5%. These findings provide quantitative guidance for the rational design of frozen food packaging, demonstrating how structural optimization can act as a geometry-enabled thermodynamic buffer to preserve food quality and support sustainable cold chain logistics.
High temperature heat pumps are an important technology for reducing industrial carbon emissions through electrification. To address challenges regarding thermodynamic efficiency degradation and elevated capital investment under large-temperature-lift conditions, a thermo-economic evaluation and multi-objective optimization model based on carbon dioxide zeotropic mixtures was developed. Binary mixtures are formed by pairing CO2 with hydrofluoroolefins, hydrocarbons, and the organic compound acetone. A multi-attribute decision-making framework is established by integrating the Non-dominated Sorting Genetic Algorithm II with the Entropy Weight Method-Technique for Order Preference by Similarity to an Ideal Solution. The influence of operating parameters, including mixture concentration, subcooling/superheating degrees, and heat source/sink boundary temperatures, on system energy efficiency and cost is investigated. Results identify CO2/Pentane with a mass fraction of 0.077/0.923 as the optimal mixture for applications. Under an 80°C temperature lift with a heat source inlet temperature of 60°C and a heat sink outlet temperature of 140°C, the optimized system achieves a COP of 3.74 and a total annual cost of 902.8 $/year. This optimal scheme reduces the initial investment by 42.0% compared to the maximum efficiency solution at a penalty of a 4.1% COP reduction. Heat transfer irreversibility is reduced by the temperature glide characteristics of the zeotropic mixtures. Lorenz efficiencies of 0.68 and 0.65 are achieved for CO2/acetone and CO2/pentane systems, respectively. Temperature matching indicators for all mixtures are optimized within the range of 2.47 to 4.25. A minimum value of 2.47 is achieved by CO2/R1233zd(E), providing a thermodynamic foundation for energy efficiency enhancement.
As a key energy-consuming device in modern buildings, the chiller unit plays an essential role, and the detection of faults in its sensors is crucial for maintaining system safety and energy efficiency. Traditional methods usually treat sensors as independent entities and mostly rely on data-driven correlation modeling, ignoring the physical coupling relationships among them. Accordingly, this study introduces a Physics-driven Temporal Graph Attention Network (PT-GAT) aimed at diagnosing sensor faults in chillers. The approach establishes a graph structure grounded in the thermodynamic cycle principles of the chiller system, adopts a 1D-CNN to extract temporal features of sensors, and utilizes a graph attention network to fuse the spatial dependencies among sensors, thereby achieving accurate sensor diagnosis. Experiments are conducted based on real operational data from a data center in Tianjin, with four typical sensor faults injected for validation. A high diagnosis accuracy of 96.8% is reported for the PT-GAT model, significantly outperforming SVM, LSTM, Transformer, and various graph neural network variants. Meanwhile, it exhibits favorable performance in terms of noise robustness, incipient fault sensitivity, and cross-condition adaptability. This study provides a physically credible and high-performance integrated diagnosis framework for sensor fault diagnosis in industrial systems with networked characteristics.
As a consequence of the regulatory announcement by the European Chemical Agency regarding the complete withdrawal of polyfluoroalkyl substances, which include HFE fluids, new dielectric fluids are being proposed by several companies. However, manufacturers do not provide the fluids' properties as functions of temperature. This work investigates the thermophysical properties, such as density, specific heat, kinematic viscosity, and surface tension of two novel dielectric fluids, namely Thermasolv IM1 and Thermasolv CF2. Density was determined by the buoyancy method, specific heat by the differential scanning calorimetry method, viscosity by the modified Couette method, and surface tension by the Wilhelmy plate method. Density, kinematic viscosity, and surface tension were studied in the temperature range from 273.15 to 328.15 K and in the range from 273.15 to 373.15 K for Thermasolv IM1 and Thermasolv CF2, respectively. The specific heat was investigated in the range from 263.15 to 313.15 K for Thermasolv IM1, and in the range from 273.15 to 348.15 K for Thermasolv CF2. The study proved that fluids' properties differ from the manufacturer's data. Moreover, the influence of applying the proposed property models to dimensionless numbers was investigated. The investigation showed that the error committed by using the producer data instead of temperature-dependent data is substantial. Furthermore, the study reveals that Thermasolv IM1 is a close thermophysical replacement for HFE-7100. Meanwhile, the differences between Thermasolv CF2 and HFE-7300 or HFE-7500 are too high to be considered as replacement fluids.
Refrigerated containers are transported by container vessels on deck and in cargo holds. Numbers transported are based on the vessel specification (reefer spots availability), port of loading and discharging sequence, and the overall stability of the vessel. Units loaded in underdeck spaces generate a significant amount of heat that must be continuously removed from the cargo holds. On many vessels, this is done by mechanical fans, installed in under deck spaces. As an alternative solution, a reefer fresh water cooling system can be applied, which dissipates heat from the refrigerated containers directly from the condenser unit. Inadequate removal of heat from the space occupied by the refrigerated containers, leads to increasing temperature inside the cargo hold and consequently to intensified operation of the refrigeration unit and higher power demand.This study presents a comparative analysis of the electrical energy consumption of refrigerated containers operated with both types of cooling systems. All data were collected simultaneously during a sea passage in early 2025 on a container vessel. For the most accurate results, only reefers with the same set points and same model type were compared. The analysis focuses on total energy consumption per container. Amounts of electrical energy used for the fans and the fresh water cooling system were additionally taken into consideration.Results prove, that the energy consumption of the refrigerated containers with the active reefer fresh water cooling system required less energy to maintain the required set point when compared to the containers in the cargo hold with only mechanical ventilation. Enhancing the energy efficiency of reefers transported in cargo holds with the freshwater cooling system contributed to reduced power demand, fuel consumption, exhaust gas emissions and operational costs.
R170, as a low-boiling-point environmentally friendly refrigerant, has been widely used in ultra-low temperature refrigeration systems, especially in recent years, as one component of some environmentally friendly mixed refrigerants for refrigeration and air-conditioning. In a vapor-compression refrigeration system, refrigerants should exhibit suitable solubility with refrigeration oils to ensure reliable system operation. In this study, the solubility data of R170 in three common refrigeration oils (AB100, PAG46, and POE68) which used in refrigeration and air conditioning system, was measured by means of the isochoric saturation method within a temperature interval of 283.15-323.15 K. The NRTL and PR-BM model were used to correlate the solubility data under subcritical and supercritical conditions, respectively. For subcritical conditions, the deviations of the calculated and experimental solubility data were no more than 3 %. For supercritical conditions, the deviations were within 6 % for the R170/AB100 system, 10 % for R170/PAG46 system and within 8 % for the R170/POE68 system. The calculated results exhibited excellent agreement with the experimentally obtained data. The present study provides fundamental solubility data for the application of R170 in different refrigeration systems.
The development of novel refrigerant blends requires reliable vapor-liquid equilibrium (VLE) data. In this work, a machine-learning framework for VLE prediction of binary refrigerant blends was established. The framework integrates the construction of a dataset from experimental data for 157 binary refrigerant blends, feature extraction using basic thermophysical properties and COSMO-based sigma-profile descriptors, model development with artificial neural network (ANN) and eXtreme Gradient Boosting (XGBoost), and feature interpretation through Shapley additive explanation (SHAP). Results showed that both models reproduced the VLE behavior of binary refrigerant blends with satisfactory accuracy. The AARD values of the ANN model were 3.50% for the training set and 3.45% for the test set, both lower than those of the XGBoost model (5.67% for the training set and 5.94% for the test set). For azeotrope identification, XGBoost achieved a higher prediction accuracy (70.70%) than ANN (63.69%). SHAP analysis indicated that among the descriptors based on sigma-profile, near-neutral and weakly-negative nonpolar regions of high-boiling components made the most significant contributions to the prediction results. The proposed framework provides an efficient and practical approach for rapid VLE prediction and preliminary azeotrope screening of binary systems, offering useful guidance for the design and selection of novel refrigerant blends.
To achieve decarbonization of the heating sector, heat-pump-based steam generation is crucial for the transition of industrial heating. Since the output temperature for steam generation is higher than conventional domestic heating, the temperature lift is also larger, leading to reduced COP. In this case, transporting waste heat or other heat sources through a heat network is important for improving system performance. However, temperature mismatch exists in multiple sections of such heat-network-source heat pump for steam generation. To address this issue, an enhanced heat-network-source heat pump system for efficient steam generation is proposed, with double-section heat pump and feed water preheating. Thermodynamic analysis shows that the system can generate saturated steam at 100–130°C with a maximum COP of 3.34, under 65°C heat network inlet temperature. Compared to the basic heat pump steam generation system, the COP can be improved by 34.2%. Moreover, the proposed system is more advantageous as the heat network temperature difference increases. The exergy analysis further reveals that the performance improvement is associated with reduced exergy loss in the evaporators, owing to improved matching between the heat network and the heat pump system. Besides, exergy loss in condenser is reduced by the preheating process, due to reduced temperature difference in steam generation. With a higher-quality heat source from heat network and improved COP, the proposed system provides a practical configuration for improving the efficiency of steam production.
Achieving ultra-low temperatures (ULT) of -80 °C efficiently presents a significant technical challenge due to intrinsic difficulties associated with high temperature lifts. Furthermore, few refrigerants are capable of operating in this range. Consequently, ULT systems traditionally exhibit low efficiency and rely on R23, a high-GWP fluid. This article experimentally analyzes a two-stage cascade compression cycle, featuring internal heat exchangers in both stages, utilizing natural refrigerants R290 and R170 to reach evaporation temperatures of −80 °C and −70 °C. This proposed pair is benchmarked against conventional R404A/R23 at three condensation levels, simulating various laboratory ambient conditions, from regular use to high-load scenarios. Results demonstrate a Coefficient of Performance improvement of up to 40% using very low-GWP refrigerants. Moreover, the natural refrigerant pair delivers higher cooling capacity despite a lower mass flow rate. An environmental impact analysis using the Total Equivalent Warming Impact (TEWI) methodology reveals a 37% reduction in equivalent CO₂ emissions for the natural refrigerant system. This study also identifies high-risk operational points in ULT cycles. The low-temperature stage expansion valve inlet is identified as a critical safety risk, particularly with flammable natural refrigerants, due to extremely low material temperatures. To mitigate this, an interstage heat exchanger is proposed, connecting the high-temperature stage liquid line to the low-temperature stage vapor line. This configuration ensures an adequate compressor suction temperature for proper lubrication. This interstage solution maintains comparable performance, with a maximum COP reduction of 6.5%, while maintaining all high-risk points of the cycle within safe operating temperatures.
The global demand for space cooling is growing rapidly, making effective management of latent loads increasingly critical. Liquid desiccant dehumidification (LDD) systems enable energy efficient air dehumidification at temperatures above the dew point. Ionic liquids (IL) have recently emerged as promising desiccants due to their low melting temperatures and reduced corrosivity towards metals. Despite their potential, comparative experimental studies investigating different ionic liquids in this context remain scarce. However, meaningful quantitative comparison is often limited due to differences in operating conditions.This study experimentally investigates the dehumidification performance of [Emim]DEP in a LDD system featuring diabatic contactors. Experiments were conducted for hot water temperatures 55 - 105°C and ambient humidity ratios of 10.3 - 22.3 g·kg-1. The absorber outlet humidity ratio ranged from 5.3 to 14.3 g·kg-1, while the latent COP ranges from 0.47–0.76 at a driving temperature of 55°C and 0.32–0.52 at 105°C, increasing with larger ambient humidity ratios but showing a clear leveling-off at higher driving temperatures, where the gain in desorption capacity becomes smaller relative to the thermal input due to increasing sensible heat losses. A qualitative comparison with [Emim]Ac is presented.
This proposed study has elaborated a tool for identifying optimal specifications of adsorbent-adsorbate pairs for heat pump and refrigeration applications based on a figure of merit such as refrigerant uptake swing, COP, Specific Heating (SH) or Specific Cooling (SC). The tool is designed to establish optimum performance map and to outline key manufacturing specifications of adsorbent namely Specific Volume (vs), Specific Surface Area (S) and Micro-Pore Size Distribution (MPSD) as well as both median micropore diameter (ϕo) and standard deviation (Δ). SDA Gamma model and BET method were used to establish MPSD and Specific Surface Area (S) respectively. For demonstration purposes with Ammonia refrigerant (R717), the specifications of carbon-based absorbents are provided for adsorption heat pumps (TE = 5°C, TC = 40°C and TD = 175°C), adsorption air conditioning systems (TE = 10°C, TC = 35°C and TD = 150°C) and adsorption air ice making machines (TE = -5°C, TC = 35°C and TD = 150°C). The absorbents CHP-0.4-35-2.9 (COPopt = 1.74, SH = 887 kJ/kg, ϕo = 0.69 nm, Δ=0.91 nm, SBET=734 m2/g), CAC-0.4-30-2.4 (COPopt = 0.84, SC = 379 kJ/kg, ϕo = 0.91 nm, Δ = 0.80 nm, SBET=743 m2/g) and CIM-0.4-30-2.4 (COPopt = 0.67, SC=294 kJ/kg, ϕo = 0.91 nm, Δ=0.80 nm, SBET =744 m2/g) are among the best candidates for adsorption heat pump, air conditioning and ice making respectively. With those three candidates, each system will still operate effectively within ±20°C around the designed operating driving temperature (TD) with marginal change of optimum COP value.
Regenerative thermal oxidisers (RTOs) are widely used for volatile organic compounds (VOCs) abatement in automotive paint shops. They produce significant heat with variable exhaust temperatures, complicating the conversion of this energy source into a stable and usable service. This study assesses whether using liquid desiccant solutions to recover RTO stack heat, store it thermochemically through changes in solution concentration and supply temperature and humidity control is techno-economically viable. An annual performance model coupled with a techno-economic assessment was developed for two utilisation strategies in a paint shop: on-site for flash-off drying and off-site for climate control in a nearby gym via pipeline transport. The analysis considers aqueous lithium chloride (LiCl) and calcium chloride (CaCl2), outdoor air conditions, RTO exhaust temperature variability and operational adjustments, including regenerator exhaust air recirculation and auxiliary heat, to balance dehumidification and regeneration year-round. Results show that coupling RTO heat recovery with liquid desiccant regeneration is technically feasible and that on-site flash-off drying offers the strongest economic case, with payback periods of 5.3 years (NPV £134.4k, IRR 18.1%) and 10.9 years (NPV £22.9k, IRR 6.6%) for aqueous CaCl2 and LiCl, respectively. The off-site gym application is less favourable for the investigated distance and end-user demand. Sensitivity analysis confirms that viability is strongly dependent on energy prices, pipeline distance and working fluid. The results indicate that on-site utilisation should always be prioritised, while off-site applications require larger end-user demand and/or shorter distances, together with more favourable operating conditions.
Coastal industries (e.g., food processing, chemical distillation), commercial facilities, and residential buildings demand high-temperature heat supply (90–140°C). Conventional air-source heat pumps were limited to below 80°C, while transcritical CO₂ seawater-source heat pumps offered a viable pathway by leveraging abundant seawater. However, existing systems suffered from suboptimal expansion work recovery and severe exergy destruction when the water outlet temperature exceeded 120°C. This paper proposed a novel two-stage expander system (TSES) that closely matched the entire expansion process of the transcritical CO2 cycle, enabling efficient cascade energy recovery. Consequently, TSES overcame the operating range limitations of single-stage turbine system (STES) and single-stage volumetric expander systems (SVES). Thermodynamic modeling and performance comparisons demonstrated that TSES delivered clear advantages across typical seawater-source heating conditions. For instance, for gas-cooler inlet water temperatures between 35°C and 60°C, the heating coefficient of performance (COP) and exergy efficiency of the TSES consistently outperformed those of the single-stage expander systems by roughly 2%. Its relative improvement grew further with rising inlet water temperature, maintaining a lead of more than 8% over the reference systems at 60°C. Furthermore, in the high-temperature heating region with gas-cooler outlet water temperature of 120–140°C, the COP and exergy efficiency advantages of the TSES over the STES gradually widened from 2.71% to 6.27%, while its advantage over the SVES remained consistently above 8%. These gains stemmed from structural improvements: the TSES reduced the expansion exergy destruction share by over 30% and increased expansion work recovery rate by over 60% relative to both single-stage systems.
Cryo-compressed hydrogen storage technology offers high volumetric density and relatively low energy consumption, making it a promising solution for hydrogen storage and transportation. Direct cooling of high-pressure gaseous hydrogen provides a lower intrinsic energy demand. Nevertheless, time-dependent compression heat during charging is not properly addressed, causing a 30-40 K temperature rise and a consequent reduction in the achievable hydrogen storage density. A dual-cycle cooling-filling process is proposed that combines a mixed-refrigerant Joule-Thomson cycle with a closed nitrogen cycle. The closed nitrogen cycle enables thermal load decoupling: gaseous nitrogen matches distributed sensible heat, while liquid nitrogen removes compression heat under near-isothermal conditions. Cold energy storage further allows dynamic adaptation to compression heat loads. Thermodynamic simulation shows that at 50 MPa and 85 K, the proposed system achieves a specific power consumption of 6.74 kWh/kg, approximately 39% lower than that of the conventional LH2 pressurisation-vaporisation method for CcH2 production.
In this paper, a swing compressor with two discharge ports for discharge regulation is introduced. The prototype design and discharge regulation process of different operation conditions have been introduced in detail. A SCDR prototype with stroke volume of 9.88cm3 has been displayed and an experiment using R32 as the working fluid has been conducted to comprehensively assess the thermal performance of this prototype. The evaluation has been carried out from the perspectives of volumetric efficiency, cooling capacity, input power, COP, and discharge temperature, under four operating conditions that varied in evaporation temperature, condensation temperature, suction temperature and operating frequency. The experimental cooling capacities of SCDR for four operation conditions are 2628.9 W, 3505.3 W, 3899.4 W and 4311.9 W respectively, with the corresponding input power of 350.8 W, 851.5 W, 1240.7 W and 1847.4 W. The volumetric efficiencies are 91.53%, 88.41%, 84.7% and 86.58% respectively, with the corresponding COP of 7.49, 4.12, 3.14 and 2.33. The experimental discharge temperatures of SCDR are 68.3 °C, 92.4 °C, 103.5 °C, and 115.6 °C, which remain within an acceptable range. The experimental results demonstrate the thermal performance of SCDR, which is suitable to be an alternative rotary compressor applied in the household air conditioner.
Due to the extremely high ratio of sensible heat to latent heat for helium (>70), a sufficient precooling process is critical to reducing the consumption of limited cooling capacity at 4 K by helium’s sensible heat and improving the helium liquefaction rate. However, because of the difficulties in cryogenic measurement, practical helium precooling process lacks sufficient experimental date to guide its optimization. In this study, an analysis method for precooling-related losses is proposed and an experimental platform is established based on a two-stage 4 K GM cryocooler. The extraction characteristics of excess cooling and its impact on the cooling capacity at 4 K are measured, and the distribution of losses during the precooling process is quantitatively analyzed. Results indicate that the cooling capacity loss caused by the superheat of helium is the main loss affecting the liquefaction rate. This loss is governed by the minimum extractable temperature of excess cooling, which is determined by cylinder-end temperature of the cryocooler. Effective use of the excess cooling at the end of cylinder for precooling helium can reduce the helium temperature from above 10 K to 8.2 K and the corresponding liquefaction rate loss from 63% to 47.8%, representing an effective approach to improve precooling efficiency and increase the helium liquefaction rate.
