Artificial Radiation Frost Chamber for Frost Formation on Tea

The Yangtze River region is the main production area for famous, high-quality tea in China. Radiation frost frequently occurs in this region, especially in the early spring during calm and clear nights, and it causes substantial damage to crops, which leads to huge economic losses for tea growers. The formation of frost is difficult to experimentally control due to the complexity and variability of the agro-micrometeorological environment. The objective of this study was to evaluate an artificial radiation frost chamber based on the temperature difference between leaf and air dew point, which was designed for advanced frost-related research. Micro-meteorological data and the frost formation process were monitored in an experimental tea field during typical radiation frost nights to mimic declining temperatures that are consistent with nature. The radiation frost chamber model and main parameters were determined by theoretical calculations and computational fluid dynamics (CFD) simulation. A frost-forming experiment was conducted to evaluate the performance of the frost chamber. The observation results showed that the greatest temperature difference between leaf and air dew point (Tdiffer) was −2.3 ◦C. The simulation results showed that the desublimation cooling rate of the air vapor was greater than sublimation, and the Tdiffer should be greater than −3.2 ◦C, which could cause frost to easily form on the leaf. The performance testing results showed that leaf temperature slowly declined after a rapid decrease, which is similar to the natural condition, which results in noticeable frost formation on the leaf.


Introduction
Frost refers to the formation of ice crystals on surfaces, either by freezing of dew or a phase change from vapor to ice, which causes injurious frost formation on crops and plants [1,2]. It occurs at temperatures of less than or equal to 0 • C, as measured in a "Stevenson-screen" shelter at a height between 1.25 and 2.0 m [3]. It can also occur with an air temperature less than 0 • C, without defining the shelter type and height [4]. Snyder et al. [2] have subdivided frost into advection frost and radiation frost. Advection frosts are associated with large-scale incursions of cold air with a well-mixed, windy atmosphere and a temperature that is often subzero, even during the daytime. Radiative frosts are associated with cooling due to energy loss through radiant exchange during clear, calm nights, and with air temperature inversions. In some cases, a combination of both advection and radiation conditions will occur.
Radiation frosts commonly occur in the Yangtze River region, China, especially during early spring nights [5,6], and they are characterized by clear skies, calm winds, and air temperature inversions [7,8].

Main Structures of the Frost Chamber
The technical parts of frost chamber include the basic refrigeration unit, the cold radiation source, and the isolation unit. As shown in Figure 1, the plant samples are placed in the basic refrigeration unit and their radiation heat through the isolation unit is absorbed by the radiation refrigeration unit, which leads to a decrease of the leaf temperature. When the leaf and air temperature difference (T differ ) reduces to a certain value, the water vapor in the air will sublimate into frost on the leaf surface. The basic refrigeration unit provides a basic environment and heat transfer will occur between the samples and radiation heat source unit. The isolation unit isolates the temperature and humidity in the cold radiation source from the basic refrigeration unit, and it is also used to increase the thermal efficiency of cold source radiation.  As shown in Figure 2, the length, width, and height of the basic refrigeration unit is 800 mm, 400 mm, and 600 mm, respectively. It is composed of external and internal thermal insulation as well as a cooling plate. External thermal insulation minimizes the influence of hot air outside of the basic refrigeration unit; the cooling plate, located between the internal and external thermal insulation, is used to absorb the external heat load and it also adjusts the basic temperature to reduce the thermal load of radiation cooling.

Simulation of Tdiffer
A suitable Tdiffer is the crucial factor for frost formation on the leaf. CFD simulation was conducted to determine the technical parameters after estimating the temperature of cold radiation source unit (Tradiation) and basic refrigeration cold source unit (Tbasic), and the heating power of the external (Pext) and internal thermal insulation (Pint). The optimization indexes were adjusted to the Tdiffer, and the meshing, boundary conditions, and numerical solution were also conducted for the simulation (Table 1). The cold radiation source unit length, width, and height are 400 mm, 450 mm, and 150 mm, respectively, and it is fixed above the isolation unit and used to generate the low temperature. The isolation unit is fixed between the cold radiation source and basic refrigeration unit and it has a length, width, and height of 700 mm, 450 mm, and 600 mm, respectively; its outer surface isolation is coated with a thin film with high long wave transmittance to ensure enough radiation heat transfer between the sample and cold source. In addition, a heating wire is set up in the isolation unit to balance the cold air that is provided by the cold radiation source. Desiccant is fixed on either side of the isolation unit to prevent water vapor from condensing on the lower film.

Simulation of T differ
A suitable T differ is the crucial factor for frost formation on the leaf. CFD simulation was conducted to determine the technical parameters after estimating the temperature of cold radiation source unit (T radiation ) and basic refrigeration cold source unit (T basic ), and the heating power of the external (P ext ) and internal thermal insulation (P int ). The optimization indexes were adjusted to the T differ , and the meshing, boundary conditions, and numerical solution were also conducted for the simulation (Table 1). Three-dimensional (3D) drawing software (UG 11.0) was used to construct the 3D model of the frost chamber Figure 3a and it was subsequently imported to the grid software (ICEM 15.0). The hexahedral mesh division method was applied to the model to attain the advantages of high mesh quality and easy convergence, while the O-type meshing method was used around heating wire. The mesh boundary was encrypted with a total of 684,976 mesh cells and 646,017 nodes. The grid quality inspection showed a minimum determination coefficient of 0.5 and minimum aspect ratio of 16, which met the requirements of the numerical calculation, so the mesh was converted into an unstructured grid and then saved as a mesh file Figure 3b. Three-dimensional (3D) drawing software (UG 11.0) was used to construct the 3D model of the frost chamber Figure 3(a) and it was subsequently imported to the grid software (ICEM 15.0). The hexahedral mesh division method was applied to the model to attain the advantages of high mesh quality and easy convergence, while the O-type meshing method was used around heating wire. The mesh boundary was encrypted with a total of 684,976 mesh cells and 646,017 nodes. The grid quality inspection showed a minimum determination coefficient of 0.5 and minimum aspect ratio of 16, which met the requirements of the numerical calculation, so the mesh was converted into an unstructured grid and then saved as a mesh file Figure 3  The detailed numerical simulation steps were showing as follows: (1) import mesh file; conduct mesh check and smoothing; set proportional size; gravity acceleration was set to −9.81 m /s 2 ; and, pressure-based steady-state solution was selected.
(2) Energy equation and radiation heat transfer equation were added; the radiation model and DO model were chosen to calculate the semi−transparent medium and the flow equation with standard k-ϵ model.
(3) The gas was set as an ideal incompressible gas, and the absorption coefficient of the film was set as the following: where, Io is the output energy, W; Ii is the input energy, W; x is the characteristic length, m; and, α is the absorption factor, m −1 . The transmittance of PE film is 80% and the thickness is 0.08 mm, he α is 2790 m −1 .
(4) The boundary condition setting is shown in table 1. Heating power is expressed by heat flow density in Fluent, and the relationship between heat flow density and power is (1) import mesh file; conduct mesh check and smoothing; set proportional size; gravity acceleration was set to −9.81 m /s 2 ; and, pressure-based steady-state solution was selected.
(2) Energy equation and radiation heat transfer equation were added; the radiation model and DO model were chosen to calculate the semi−transparent medium and the flow equation with standard k-model.
(3) The gas was set as an ideal incompressible gas, and the absorption coefficient of the film was set as the following: where, I o is the output energy, W; I i is the input energy, W; x is the characteristic length, m; and, α is the absorption factor, m −1 . The transmittance of PE film is 80% and the thickness is 0.08 mm, he α is 2790 m −1 .
(4) The boundary condition setting is shown in Table 1. Heating power is expressed by heat flow density in Fluent, and the relationship between heat flow density and power is where, ϕ is the heat flow density, W·m −2 ; P is the heating power, W; d is the diameter of hot heating wire, m; l is the axial length of heating wire, m. P, d and l is set to 40 W, 0.001 m and 0.07 m, respectively. So the ϕ is 1000 W·m −2 .
(5) Monitoring value was set and the default residual convergence standard was chosen; leaf temperature, lower film temperature and upper film temperature were selected as monitoring items.
(6) Simple algorithm and standard pressure option and first−order upwind format were selected. (7) After initializing the flow field, the number of iterations was set to 5000 before start the iteration. After the calculation of convergence, the second−order upwind format was used to continue the iteration until convergence.

Main Circuit Design for the Frost Chamber
The main control circuit design consists of a circuit breaker (QF), thermal relay (KM1, KM2, and KM3), fuse, time relay, radiative refrigeration compressor (RRC), basic refrigeration compressor (BRC), decompression valve (DV), 24 V DC battery, and heating wire. The control circuit includes an automation temperature controller (ATC1, ATC2, and ATC3), the time relay coil, and the switch relay coil; the automation temperature controller received signals that were collected by the temperature sensor for controlling the main circuit ( Figure 4).
(5) Monitoring value was set and the default residual convergence standard was chosen; leaf temperature, lower film temperature and upper film temperature were selected as monitoring items.
(6) Simple algorithm and standard pressure option and first−order upwind format were selected. (7) After initializing the flow field, the number of iterations was set to 5000 before start the iteration. After the calculation of convergence, the second−order upwind format was used to continue the iteration until convergence.

Main Circuit Design for the Frost Chamber
The main control circuit design consists of a circuit breaker (QF), thermal relay (KM1, KM2, and KM3), fuse, time relay, radiative refrigeration compressor (RRC), basic refrigeration compressor (BRC), decompression valve (DV), 24 V DC battery, and heating wire. The control circuit includes an automation temperature controller (ATC1, ATC2, and ATC3), the time relay coil, and the switch relay coil; the automation temperature controller received signals that were collected by the temperature sensor for controlling the main circuit ( Figure 4).

The Cooling Ability of the Cold Radiation Source Unit and Basic Refrigeration Unit
The sampled tea cultivar is Maolu, which was well maintained in the greenhouse that was located in the experimental tea field. The tea plants that were used for the experiment had healthy leaves and they were sampled on March 28, 2016, when the ambient temperature was 18.0 °C. Performance testing was conducted on the following day, March 29, 2016, which had the same ambient temperature as the day before. Table 2 shows the temperature setting of the cold radiation source unit (Tradiation) and basic refrigeration cold source unit (Tbasic). In each treatment, only turned on the basic refrigeration unit was chosen as the control group.

The Cooling Ability of the Cold Radiation Source Unit and Basic Refrigeration Unit
The sampled tea cultivar is Maolu, which was well maintained in the greenhouse that was located in the experimental tea field. The tea plants that were used for the experiment had healthy leaves and they were sampled on March 28, 2016, when the ambient temperature was 18.0 • C. Performance testing was conducted on the following day, March 29, 2016, which had the same ambient temperature as the day before. Table 2 shows the temperature setting of the cold radiation source unit (T radiation ) and basic refrigeration cold source unit (T basic ). In each treatment, only turned on the basic refrigeration unit was chosen as the control group.  In this study, T radiation and T basic were continuously maintained to realize the continuous decrease of T l . The cold radiation source unit system was opened first to decrease the T l , followed by opening the base refrigeration unit to absorb the heat load outside the box and reduce the heat load of the radiant refrigeration. As the heat load decreased, the cold radiation source could further reduce the T l .
To determine the temperature control parameters, the change in T l under different T radiation was tested. The initial T l was set to 10.0 • C, T radiation was set to 7.5 • C, and T radiation was set to 0, 10.0, −20.0, −30.0, and −40.0 • C.

Observation Results
During the three tea field observation nights, T differ showed an obvious decrease from 16:00 to 18:00 ( Figure 5), as the cooling rates (V cool ) were both 3.5 • C·h −1 on January 26 and February 4, and that was 2.9 • C·h −1 for March 1. There was no frost on the night of January 26 to 27, while only a minute ice crystal formed on the tea leaf at 19:00 on March 1. While the ice crystals gradually disappeared with the increase of u, subtle frost formations were observed on the leaf from 1:00 to 4:00 on March 2. At 18:00 of February 4, some tiny ice crystals formed on the tea leaf, which entirely covered the leaf by 20:00. The lowest leaf−air temperature difference (T differ-min ) was −1.0 • C on the night of January 26 to 27, as it altered between 0 • C and 1.0 • C due to the cloudy weather condition. The desublimation rate of the water vapor in the air (V de ) was lower than the sublimation rate of frost (V sub ), which did not result in any frost formations on the leaf. On the clear and calm night of February 4 to 5, T differ-min was −2.3 • C and it was kept stable at around −1.1 • C. When V de was higher than the V sub , it resulted in frost formation on the leaf. T differ-min was −1.0 • C on the night of March 1 to 2, whereby the V de was similar to the V sub , which resulted in no noticeable frost on the leaf.
Tl. The cold radiation source unit system was opened first to decrease the Tl, followed by opening the base refrigeration unit to absorb the heat load outside the box and reduce the heat load of the radiant refrigeration. As the heat load decreased, the cold radiation source could further reduce the Tl.
To determine the temperature control parameters, the change in Tl under different Tradiation was tested. The initial Tl was set to 10.0 °C, Tradiation was set to 7.5 °C, and Tradiation was set to 0, 10.0, −20.0, −30.0, and −40.0 °C.

Observation Results
During the three tea field observation nights, Tdiffer showed an obvious decrease from 16:00 to 18:00 ( Figure 5), as the cooling rates (Vcool) were both 3.5 °C·h −1 on January 26 and February 4, and that was 2.9 °C·h −1 for March 1. There was no frost on the night of January 26 to 27, while only a minute ice crystal formed on the tea leaf at 19:00 on March 1. While the ice crystals gradually disappeared with the increase of u, subtle frost formations were observed on the leaf from 1:00 to 4:00 on March 2. At 18:00 of February 4, some tiny ice crystals formed on the tea leaf, which entirely covered the leaf by 20:00. The lowest leaf−air temperature difference (Tdiffer-min) was −1.0 °C on the night of January 26 to 27, as it altered between 0 °C and 1.0 °C due to the cloudy weather condition. The desublimation rate of the water vapor in the air (Vde) was lower than the sublimation rate of frost (Vsub), which did not result in any frost formations on the leaf. On the clear and calm night of February 4 to 5, Tdiffer-min was −2.3 °C and it was kept stable at around −1.1 °C. When Vde was higher than the Vsub, it resulted in frost formation on the leaf. Tdiffer-min was −1.0 °C on the night of March 1 to 2, whereby the Vde was similar to the Vsub, which resulted in no noticeable frost on the leaf.

Simulation Results
The simulation results showed that, when Tl was −2.5 °C and cabinet air temperature was 0 °C, Tdiffer was −2.5 °C, which did not meet the design requirement. Thus, further reduction of the temperature of the cold radiation source was needed. The refrigerating fluid used in the chamber was Freon and its lowest boiling point was −50.0 °C under normal temperature and pressure. While taking the load of the refrigeration compressor into consideration, Tradition was set to −40.0 °C. The power of the heating wire was set to 40.0 W based on the theoretical calculation, and it showed a condensation degree

Simulation Results
The simulation results showed that, when T l was −2.5 • C and cabinet air temperature was 0 • C, T differ was −2.5 • C, which did not meet the design requirement. Thus, further reduction of the temperature of the cold radiation source was needed. The refrigerating fluid used in the chamber was Freon and its lowest boiling point was −50.0 • C under normal temperature and pressure. While taking the load of the refrigeration compressor into consideration, T radition was set to −40.0 • C. The power of the heating wire was set to 40.0 W based on the theoretical calculation, and it showed a condensation degree of 0 on the upper and bottom films of the isolation layer. As shown in Figures 6 and 7, T differ then reached −3.2 • C, meeting the design requirement.

Tl Variation
As Tbasic was set to −5.0 °C, Tradiation was set to −40.0, −20.0, and 0 °C, the result showed that the decrease rate of Tl was 14.0 °C·h −1 at the first 0.5 h, which was same as the control group. This means that the cooling rate of Tl was primarily affected by the basic refrigeration unit. After the combined action conducted over 3.0 h, Vcool became higher than that of the control group, whereas Tl was −3.8, −2.9, and −0.4 °C, respective to Tradiation set at −40, −20, and 0 °C. Therefore, the cold radiation source further reduced Tl when the influence from basic refrigeration unit tended to be stable (Figure 8a).
Tbasic was set to 0 °C and at the first 0.

Tl Variation
As Tbasic was set to −5.0 °C, Tradiation was set to −40.0, −20.0, and 0 °C, the result showed that the decrease rate of Tl was 14.0 °C·h −1 at the first 0.5 h, which was same as the control group. This means that the cooling rate of Tl was primarily affected by the basic refrigeration unit. After the combined action conducted over 3.0 h, Vcool became higher than that of the control group, whereas Tl was −3.8, −2.9, and −0.4 °C, respective to Tradiation set at −40, −20, and 0 °C. Therefore, the cold radiation source further reduced Tl when the influence from basic refrigeration unit tended to be stable (Figure 8a).
Tbasic was set to 0 °C and at the first 0.5 h, Vcool−leaf was 12.0, 12.0, and 10.0 °C·h −1 , respective to Tradiation set at −40.0 °C, −20.0 °C, and 0 °C. Vcool−leaf of the control group was 7.0 °C·h −1 . After the combined action conducted over 3.0 h, Tl was −0.9, 2.7, 2.4, and 4.3 °C. This showed that there was an enhanced effect of the cold radiation source unit on Tl (Figure 8b).
Tbasic was set to 5 °C and at the first 0.5 h, Vcool−leaf was 9.8, 7.8, and 4.0 °C·h −1 , respective to Tradiation at −40.0, −20.0, and 0 °C. Vcool−leaf of the control group was 2.6 °C·h −1 . After the combined action conducted over nearly 3.0 h, the Tl was 2.3, 3.8, 7.0, and 7.8 °C, respectively. This showed a further enhanced effect of the cold radiation source unit on Tl strengthened (Figure 8c).

T l Variation
As T basic was set to −5.0 • C, T radiation was set to −40.0, −20.0, and 0 • C, the result showed that the decrease rate of T l was 14.0 • C·h −1 at the first 0.5 h, which was same as the control group. This means that the cooling rate of T l was primarily affected by the basic refrigeration unit. After the combined action conducted over 3.0 h, V cool became higher than that of the control group, whereas T l was −3.8, −2.9, and −0.4 • C, respective to T radiation set at −40, −20, and 0 • C. Therefore, the cold radiation source further reduced T l when the influence from basic refrigeration unit tended to be stable (Figure 8a).
T basic was set to 0 • C and at the first 0.5 h, V cool−leaf was 12.0, 12.0, and 10.0 • C·h −1 , respective to T radiation set at −40.0 • C, −20.0 • C, and 0 • C. V cool−leaf of the control group was 7.0 • C·h −1 . After the combined action conducted over 3.0 h, T l was −0.9, 2.7, 2.4, and 4.3 • C. This showed that there was an enhanced effect of the cold radiation source unit on T l (Figure 8b).
T basic was set to 5 • C and at the first 0.5 h, V cool−leaf was 9.8, 7.8, and 4.0 • C·h −1 , respective to T radiation at −40.0, −20.0, and 0 • C. V cool−leaf of the control group was 2.6 • C·h −1 . After the combined action conducted over nearly 3.0 h, the T l was 2.3, 3.8, 7.0, and 7.8 • C, respectively. This showed a further enhanced effect of the cold radiation source unit on T l strengthened (Figure 8c).

Tdiffer Variation
Tl was influenced by the combined action of the cold radiation source and basic refrigeration units (Figure 9). Tdiffer was used as an index to evaluate the cooling capacity of the cold radiation source. Tradiation was set to −40.0 °C, and Tdiffer was −4.4, −5.2, and −5.5 °C, relative to Tbasic at −5.0, 0, and 5 °C. Furthermore, Tradiation was −20.0 °C, Tdiffer was −3.5, −3.6, and −3.7 °C, respectively. For Tradiation at 0 °C, Tdiffer was −1.0, −1.1, and −0.8 °C, respectively. These results showed that different Tbasic had little influence on the cooling ability of the cold radiation source. Vcool−leaf was similar to that of the natural environment under the individual effect of the cold radiation source within the first one hour. Following those results, Tl slowly decreased.

T differ Variation
T l was influenced by the combined action of the cold radiation source and basic refrigeration units ( Figure 9). T differ was used as an index to evaluate the cooling capacity of the cold radiation source. T radiation was set to −40.0 • C, and T differ was −4.4, −5.2, and −5.5 • C, relative to T basic at −5.0, 0, and 5 • C. Furthermore, T radiation was −20.0 • C, T differ was −3.5, −3.6, and −3.7 • C, respectively. For T radiation at 0 • C, T differ was −1.0, −1.1, and −0.8 • C, respectively. These results showed that different T basic had little influence on the cooling ability of the cold radiation source. V cool−leaf was similar to that of the natural environment under the individual effect of the cold radiation source within the first one hour. Following those results, T l slowly decreased.

Modification of Initial Vcool
Vcool in the chamber was dissimilar to the natural environment after the system had initially functioned for 1.0 h ( Figure 10). Thus, it was necessary to provide a new kind of temperature control scheme to make sure that it was consistent with the natural conditions in order to make Tl decrease 10.0 °C in the first 2.0 h.  (Figure 10). Therefore, these cooling trends can be used to control Tl and Table 3 shows the adjustment scheme. Firstly, Tbasic was set to 7.5 °C, and it rapidly decreased the Tl to 10.0 °C. Meanwhile, Tradiation was set to −10 °C and kept at that temperature for 0.5h, and then Tbasic was set to 5.4 °C. After that, Tradiation was set to −20.0 °C and kept at that temperature for 0.5 h, and then Tbasic was set at 3.6 °C. Subsequently, Tradiation was set at −30.0 °C and kept for 0.5 h, after which When T radiation was set to 0, 10.0, 20.0, 30.0, and 40.0 • C, T l decreased 1.4, 2.1, 2.8, 3.4, and 3.8 • C, respectively, within the first 0.5 h ( Figure 10). Therefore, these cooling trends can be used to control T l and Table 3 shows the adjustment scheme. Firstly, T basic was set to 7.5 • C, and it rapidly decreased the T l to 10.0 • C. Meanwhile, T radiation was set to −10 • C and kept at that temperature for 0.5h, and then T basic was set to 5.4 • C. After that, T radiation was set to −20.0 • C and kept at that temperature for 0.5 h, and then T basic was set at 3.6 • C. Subsequently, T radiation was set at −30.0 • C and kept for 0.5 h, after which the T basic was set at −1.2 • C. Finally, T radiation was set at −40 • C and it was kept there for 0.5 h. As shown in Figure 11, the change in actual T l was consistent with that of an ideal one in which the natural cooling process for a frost night was realized. When Tradiation was set to 0, 10.0, 20.0, 30.0, and 40.0 °C, Tl decreased 1.4, 2.1, 2.8, 3.4, and 3.8 °C, respectively, within the first 0.5 h ( Figure 10). Therefore, these cooling trends can be used to control Tl and Table 3 shows the adjustment scheme. Firstly, Tbasic was set to 7.5 °C, and it rapidly decreased the Tl to 10.0 °C. Meanwhile, Tradiation was set to −10 °C and kept at that temperature for 0.5h, and then Tbasic was set to 5.4 °C. After that, Tradiation was set to −20.0 °C and kept at that temperature for 0.5 h, and then Tbasic was set at 3.6 °C. Subsequently, Tradiation was set at −30.0 °C and kept for 0.5 h, after which the Tbasic was set at −1.2 °C. Finally, Tradiation was set at −40 °C and it was kept there for 0.5 h. As shown in Figure 11, the change in actual Tl was consistent with that of an ideal one in which the natural cooling process for a frost night was realized.

Conclusions
In a typical radiation frost night, the desublimation rate of the water vapor in the air must be greater than the sublimation rate of frost, which could cause frost formations on the leaf. Observation results showed that greatest T differ was −2.3 • C in the natural frost night. T differ should be greater than −3.2 • C to meet the design requirement. The performance testing results showed that the leaf temperature slowly declined after a rapid decrease. Additionally, the processes of frost formation on the surfaces of tea leaf samples in this chamber is similar to the naturally frost night. The change in actual T l was consistent with that of the natural conditions in frost night.
Used routinely for research aimed at understanding the physiology and mechanisms of frost injury to tea plants, the chamber was running for over 10 months and proved to be reliable and easy to use at normal temperature (around −10 to 25 • C). If the chamber works in the summer (ambient temperature higher than 25 • C), the frost formation is not obvious. We think it is caused by the bad airtightness between radiation cold source, thermal-protective coating, and basic refrigeration unit. We will keep on improving the chamber to make it potentially used in all kinds of ambient temperature. It could be applied in plant physiology research where dew formation is required, or where energy exchange is to be studied. The chamber should be further studied to improve the control over the amount of frost.