For the purpose of proper refrigerant distribution in heat exchangers for air conditioners,we prototyped a refrigerant distributor with a new structure that distributes gas-liquid separated refrigerant using a slit pipe.Observation of the flow state by the visualization model and evaluation of the variation of the liquid-phase branch ratio as a distribution characteristic revealed the following. The height of the refrigerant liquid level formed in the gas-liquid separation space of the prototype distributor (slit type) decreases with increasing dryness. Comparing the distribution characteristics of the current collision type four-branch distributor and the prototype distributor (slit type), the conditions without inclination and the high flow rate condition (Gr=20 kg/h), the current distributor is superior,but the under inclination conditions and low flow conditions (Gr=10 kg/h), the prototype distributor has an advantage over the conventional four-branch distributor.

Polysaccharides are widely used to preserve the texture of frozen desserts like ice cream by inhibiting ice recrystallization. Despite recent significant advances in research, a clear comprehension of the ice recrystallization inhibition mechanism of polysaccharides remains elusive. Some previous studies hypothesized that the ice recrystallization inhibition originates from the interaction between polysaccharides and ice crystal surface and tried to correlate it with the ice crystal shape. In this work, we have developed a novel method for analyzing the shape change of ice crystals during storage by curvature calculation. This method depicts the ice crystal shapes as the distribution of curvatures well expressing the changes of the ice crystal shapes. The newly developed method has been applied to evaluate the shape changes of ice crystals during the storage of sucrose solutions with and without kappa-carrageenan. The changes in curvature distributions revealed that ice crystals were rectangular with flat edges in the pure sucrose solution and were comparatively round in the sucrose/kappa-carrageenan solutions at longer storage time, which was not characterized in the roundness changes, showing the potentiality of using the curvature distribution to evaluate the shape change of ice crystals appropriately during storage.

The performance of a 2 kW class heat pump cycle was calculated to evaluate COP, volumetric capacity, and irreversible losses for the binary mixture refrigerants containing HFO-1123 with a very low global warming potential, HFO-1123/R32, HFO-1123/R290, and HFO-1123/R152a. In the calculations, the minimum temperature difference at the outlet of the cross flow heat exchanger was fixed at 5 K assuming an air heat source, and the required heat transfer area was varied for comparison. At an HFO-1123 mass fraction of 0.6, HFO-1123/R32 exhibited the highest cooling COP of 6.1, meanwhile HFO-1123/R290 and HFO-1123/R152a exhibited COPs 1.1 and 1.4 lower than that. The large temperature glide increased the irreversible losses in the heat exchangers and significantly influenced the COP. Therefore, calculations were performed using a counter-flow heat exchanger model. The COP of HFO-1123/R290 was increased by 1.2 compared to that of the cross flow heat exchange, which was comparable to that of HFO-1123/R32.

In this study, we analyzed the impact of feature selection on the predictive performance of common machine learning models for estimating the coe-cient of performance (COP) of a heat pump system. The dataset consisted of heating operation test data from a heat pump cycle using refrigerant R32. Various feature groups were created from control variables and measurement data, and the relationship between these features and the machine learning models was examined. The results indicated that stepwise linear regression provided the highest predictive accuracy when only control variables were used. While, linear regression and Gaussian process regression provided high accuracy models when measured data were included as features. Moreover, kernel-based machine learning models, such as Gaussian process regression and support vector regression, demonstrated high predictive accuracy even with a limited number of features or when key explanatory variables, such as the COP-defining variable, were excluded. Additionally, the study showed that models with high accuracy could be built using a small number of features, particularly those with a strong correlation to the objective variable.

In air conditioning systems where the transition to environmentally friendly refrigerants is progressing, the challenge is to use refrigerants with a temperature gradient. Refrigerants with a temperature gradient have different compositions in the liquid and gas phases, so the bias in the gas-liquid distribution at the pipe branch has a significant impact on controllability. In this study, we conducted a gas-liquid two-phase flow analysis using Large eddy simulation (LES) for a shape that branches from a horizontal pipe in the vertical upward direction, and clarified the effect of the length of the outlet pipe on the gas-liquid distribution characteristics. In addition, by expressing the outlet dryness obtained by LES with a one-dimensional model based on the length of the outlet pipe, we developed a simple distribution prediction formula that can be applied to the analysis of the entire air conditioning system.

In recent years, zeotropic mixed refrigerants have become a strong candidate for next-generation refrigerants due to restrictions on conventional refrigerants to reduce global warming. The refrigeration oil used in the compressor is selected for its compatibility with refrigerants to return to the compressor more easily. However, it is also known that the dissolution of refrigerants into refrigeration oil cause a change in the physical properties of the mixture, which affects the performance of the refrigerant compressor. Therefore, to realize early conversion to next-generation refrigerants, it is necessary to find out how refrigerants dissolve in refrigeration oil to ensure that compressors work well and are reliable. Especially for zeotropic mixed refrigerant, the dissolved composition may differ from the original composition. This study aimed to reveal the dissolution behavior of zeotropic mixed refrigerants into the refrigeration oil. The experiment was set to measure the dissolved composition at saturation condition and compare the dissolved composition at the initial dissolution state and at saturation condition. R454C, a mixture of R32 and R1234yf, was used as the zeotropic mixed refrigerant, and POE oil was used as the refrigeration oil. The measurement results showed that the refrigerant dissolved in the refrigeration oil had almost the same composition as the charged before mixing. However, when the dissolved composition was measured in the pressure range below the dew point at a liquid temperature of 80℃, the amount of R1234yf was approximately 4% higher than that of the charged composition. Furthermore, at a liquid temperature of 80℃, the liquid phase separation with different refrigerant mixture ratios was observed in the solubility range from 67% to 82%.

A study was conducted to quantify the amount of refrigerant dissolved in refrigeration oil for the purpose of estimating the composition change of mixed refrigerants that are non-azeotropic in an air-conditioner. In the case of a binary non-azeotropic mixture of R 1132(E) [trans-1,2-difluoroethylene] and R 1234yf, R 1234yf preferentially dissolves in the refrigeration oil, and the amount of R 1234yf dissolved in the refrigeration oil was estimated as the temperature and pressure increased. It was found that the ratio of R 1132(E) in the mixed refrigerant composition tended to increase relative to the inclusion composition as the temperature decreased and the pressure increased.

R 1132(E), an HFO refrigerant with extremely low GWP and high efficiency, is known to occur self-decomposition reaction by external energy source. In this study, we evaluated the conditions for occurrence of self-decomposition and the discharge energy in compressor for using of R 1132(E) mixed refrigerants. It was found that self-decomposition was less likely to occur as the R 1132(E) concentration decreased, and refrigerants containing R 1132(E) at < 35mass% was stable for excessive discharge energies. The quantification of the discharge energy showed that the energy in 2.2kW compressor was less than 1J. From these results, the composition ratio of R 1132(E) mixed refrigerants which can be used stably as refrigerant were clarified. .It was clarified that these results indicated clarifying the conditions for stable use of R 1132(E) mixed refrigerant.

Considering the importance of low GWP refrigerant blends and their constituents, this study aimed to clarify the flammability characteristics of CO2/propane (R290) blends at all blend compositions up to non-flammable. First, the flammability limits were evaluated using the evaluation method newly adopted in the High Pressure Gas Safety Act (HPGSA), and the results were compared with the ASHRAE method. The burning velocity was evaluated by microgravity experiments in the immediate vicinity of the non-flammable composition. Furthermore, to contribute to practical safety evaluation, the quenching distance was evaluated near the boundary of the mildly flammable (2L) class with a maximum burning velocity of 10 cms^-1. Second, by formulating these properties as a function of the mass or mole fraction of CO2, it became possible to predict all the flammability characteristics values at all blend compositions up to non-flammable and determine the flammability class according to ISO 817 and HPGSA.

Under the worldwide situation of the shift to the next-generation refrigerants with low global warming potential (GWP), there is a demand to evaluate the fundamental thermophysical properties of low-GWP refrigerants, which is essential for their practical application. In this study, we developed a speed of sound measurement apparatus in the liquid phase using a commercially available ultrasonic-pulse sound velocity sensor. The apparatus was calibrated using pure water and R1336mzz(Z) (cis-1,1,1,4,4,4-hexafluoro-2-butene) as the reference standards for speed of sound, and R1234yf (2,3,3,3- tetrafluoropropene) was measured to confirm the soundness of the speed of sound measurement. As a result, it was confirmed that the speed of sound for R1234yf in the liquid phase agreed with the existing values within the claimed uncertainty. The relative standard uncertainty of the speed of sound measurement in the liquid phase was estimated to be 0.084 %. Using the developed liquid -phase speed of sound measurement apparatus, R1336mzz(E) (trans-1,1,1,4,4,4-hexafluoro-2-butene) and R13I1 (trifluoroiodomethane) were measured as the low-GWP refrigerants since their liquid-phase speed of sound have not been reported yet. The speed of sound data for R1336mzz(E) were obtained at 28 points in the temperature range from 283 K to 343 K and pressure range from 1 MPa to 7 MPa, and those for R13I1 were obtained at 28 points in the temperature range from 273 K to 333 K and pressure range from 1 MPa to 7 MPa. The obtained speed of sound data were compared with the calculated values from the existing equations of state for R1336mzz(E) and R13I1, and it was found that the present measurement data for R1336zz(E) and R13I1 were smaller by down to 9 % and 2.4 %, respectively.