What Affects Packing Temperature in Cryogenic Testing of LNG Globe Valves

Abstract: To ensure the safety and reliability of liquefied natural gas (LNG) systems, cryogenic globe valves must withstand extremely low operating temperatures of down to −196°C. Temperature field measurements were conducted on DN15 cryogenic globe valves, and temperature data were collected from the packing glands of 235 valves to characterize and verify the temperature field distribution. The results showed that the temperature distribution in the packing area approximately followed a normal distribution. The potential factors affecting the packing gland temperature during testing were then parameterized. The packing gland temperature was calculated based on three key parameters:

• the convective heat transfer coefficient
• the nitrogen layer height
• the nitrogen layer temperature

A probability distribution of the packing gland temperature was subsequently established based on these parameters.

With the rapid development of China's petrochemical, cryogenic engineering, and liquefied gas industries, the valve market has continued to expand, making cryogenic valves and related technologies an important area of research. The cryogenic globe valve investigated in this study is primarily used for LNG transportation. LNG consists mainly of methane and is produced by cooling natural gas to its liquefaction temperature, significantly reducing its volume and facilitating long-distance transportation. LNG has consequently become an increasingly important component of China's energy supply. In 2021, China imported 81.4 million tonnes of LNG, surpassing Japan to become the world's largest LNG importer.

Material selection is a key factor affecting valve strength, reliability, and service life. Cryogenic valves must be capable of long-term operation under cryogenic and ambient-temperature conditions, with a typical service life of 10 years or 3,000–5,000 operating cycles. LNG service involves cryogenic temperatures and high pressures, placing stringent demands on the design and performance of storage and transportation equipment. LNG globe valves are therefore particularly susceptible to leakage. Leakage can result in significant LNG loss, energy waste, and environmental pollution, and may even lead to serious accidents. Ensuring the reliability of cryogenic globe valves is therefore essential. Research on their reliability is critical to improving the safety of LNG systems, ensuring energy security, and protecting the environment.

Since the beginning of the 21st century, an increasing number of researchers have focused on cryogenic valves. Zhang Chaoyang summarized the classification and performance evaluation criteria for cryogenic valves, defined the conditions required to prevent icing, and provided a basis for their appropriate selection. Chen Shaorong examined the structure, operating principles, and characteristics of cryogenic emergency shut-off valves for LNG service and described the procedures for cryogenic testing. Wang Jie reviewed the application of cryogenic globe valves in LNG service. He Hang et al. investigated the cryogenic testing conditions of a DN80 cryogenic globe valve for LNG carriers, analyzed temperature variations in the packing gland and other valve components, evaluated the suitability of the valve neck length, proposed corresponding improvements, and examined the time required for the valve temperature to reach a steady state. In 2019, several Chinese valve manufacturers, including Shanghai Lianggong Valve Factory Co., Ltd., jointly drafted and issued the Technical Specifications for Cryogenic Valves, which has since been widely referenced as a technical specification for cryogenic valves in China.

Countries such as the United States, Germany, the United Kingdom, and Japan have well-established capabilities in cryogenic valve manufacturing. Lee developed a method for detecting high-pressure gas leakage in cryogenic valves, in which pressure sensors installed in a control unit monitor pressure variations to facilitate accurate and standardized testing. Pinho et al. experimentally investigated the performance of cryogenic valves using liquid nitrogen and water as working fluids. Their results showed that, under single-phase flow conditions, the flow coefficient of cryogenic globe valves was independent of the working fluid, and the numerical simulation results agreed well with the experimental data. They also simulated test conditions under subcritical and critical flow regimes and confirmed that the liquid recovery factor of cryogenic valves was independent of the working fluid. Lin et al. analyzed the application of cryogenic valves in LNG receiving terminals and the factors affecting their performance. They investigated the flow characteristics and contact stresses of hard-seated valve seats under LNG flow conditions, with particular emphasis on parameters such as pressure drop, pressure distribution, and velocity distribution.

The cryogenic globe valve investigated in this study is a DN15 valve (Figure 1) with flanged connections conforming to ASME B16.5. This study focuses on the temperature distribution in the packing area under cryogenic operating conditions, with particular emphasis on the cryogenic testing process.

Figure 1. Schematic of the DN15 cryogenic globe valve

Cryogenic tests were conducted using liquid nitrogen as the cooling medium at a temperature of −196 °C (±5 °C) and an ambient temperature of 22–23 °C. Thermocouples were installed to monitor the temperature at the outer surface of the packing gland base near the bonnet and on the valve body. The valve was immersed in liquid nitrogen until the liquid level reached the top of the valve body-to-bonnet connection. The cryogenic globe valve was maintained in the liquid nitrogen for at least 1 h. After the boiling of the cooling medium subsided and the temperatures of the bonnet and valve body approached −196 °C, indicating that the valve had reached a relatively uniform temperature distribution, the temperatures of the packing gland and valve body were recorded.

Figure 2. Cryogenic globe valve after 1 h of immersion

The recorded data included the ambient temperature, cooling medium temperature, bonnet temperature, packing gland temperature, and valve body temperature. The experiments were conducted over a two-week period, with the indoor temperature maintained at 22–23 °C. The experimental data for the packing gland temperature are presented in Table 1, while the data for the internal valve temperature, valve body temperature, and cooling medium temperature are presented in Table 2.

Table 1. Experimental Data for Packing Gland Temperature

Table 2. Experimental Data for Internal Valve, Valve Body, and Cooling Medium Temperatures

Because the experimental data are discrete, the reliability of the numerical calculations can be assessed by evaluating the goodness of fit between the experimental and simulated temperature distributions. A normality test was performed on the packing temperature data. The distribution was considered to follow a normal distribution when the following criteria were satisfied:

Q–Q plot: A quantile–quantile (Q–Q) plot was generated to assess the normality of the sample data, with the theoretical quantiles of the standard normal distribution plotted on the horizontal axis and the sample quantiles on the vertical axis.

1. The normality of the sample data can be assessed by examining whether the data points lie approximately along a straight line. The closer the points are to the straight line, the more closely the sample data approximate a normal distribution. The slope of the line represents the standard deviation, while the intercept represents the mean.

4. Normality test: The significance level (p-value) of the normality test is greater than 0.05.

5. Histogram: The histogram is approximately symmetrical, with a central peak and lower frequencies toward both sides, and no obvious outliers.

The skewness and kurtosis of the packing temperature are presented in Table 3. The results show that:

both of which satisfy the above criteria.

Table 3 presents the skewness and kurtosis of the packing temperature. The results show that ∣skewness/SE(skewness)∣=0.65 and ∣kurtosis/SE(kurtosis)∣=1.19, both of which satisfy the normality criterion. Figure 3 shows the normal Q–Q plot of the packing temperature data for the cryogenic globe valve samples.

Table 3. Skewness and Kurtosis of Packing Temperature

Figure 3. Normal Q–Q Plot of Packing Temperature

Table 4 presents the results of the normality tests for the packing temperature data. The p-values obtained from both tests are greater than 0.05, indicating that the packing temperature data are consistent with a normal distribution. Figure 4 shows the distribution of the packing temperature data. The results indicate that the data approximately follow a normal distribution, supporting the use of the data in subsequent numerical calculations.

Table 4. Normality Test Results for Packing Temperature

Figure 4. Distribution of Packing Temperature

In the cryogenic tests, once the globe valve reached a steady-state temperature field, the packing gland temperature varied within a range of approximately 10 °C. This variation was mainly attributed to differences in the convective heat transfer coefficient at the heat-dissipation surface of the valve bonnet from one test batch to another. These differences were caused by factors such as liquid nitrogen boiling, manual replenishment of the cooling medium, and variations in ambient humidity.

Taking the temperature variation range at the packing gland observed in the cryogenic tests as the threshold, the corresponding variation range of the convective heat transfer coefficient was determined parametrically to obtain the best agreement between the numerical and experimental results. The factors that may affect the temperature field were then discussed, with particular attention to the valve bonnet region. The analysis focused on whether the temperature at the bottom of the packing remained above 0 °C and whether the temperature gradient was sufficiently small, as these conditions determine whether the valve stem can operate properly under cryogenic conditions.

A three-dimensional model of the DN15 cryogenic globe valve was developed using modeling software and simplified to retain only the major components, including the bonnet, valve body, valve disc, stem, handwheel, and bolts. 316L stainless steel was assigned to the major components, including the valve body, bonnet, and stem, while Stellite 6 alloy was used for the sealing surfaces. Flexible graphite was selected as the packing material, with a thermal conductivity of 151 W/(m·K) and a specific heat capacity of 709 J/(kg·K). Because of the relatively low thermal conductivity of the sealing material, a thermal conductivity of 20 W/(m·K) was assigned to the Stellite 6 alloy.

Transient thermal analysis requires both the thermal conductivity (k) and specific heat capacity (cp) of the material to be considered. As shown in Table 5, the thermal conductivity of the alloy materials remains relatively stable over the cryogenic temperature range; therefore, a constant thermal conductivity was assumed for the sealing alloy. In contrast, the specific heat capacity of stainless steel varies with temperature and was therefore treated as a temperature-dependent property.

Table 5. Thermophysical Properties of 316L Stainless Steel at Different Temperatures

To ensure the accuracy of the numerical results, the mesh was locally refined in the nonlinear contact regions while maintaining a consistent mesh density for identical contact areas. Local mesh refinement was also applied to the sealing ring. Although the minimum element quality was 0.15, this value occurred in a non-critical region, whereas the mesh quality in the main calculation regions exceeded 0.80 (Figure 5). The model contained a total of 2,188,068 nodes and 611,665 elements. The thermal–structural coupled analysis confirmed that the mesh quality was sufficient for the subsequent calculations.

Figure 5. Mesh Model of the Cryogenic Globe Valve

A first-kind (Dirichlet) boundary condition with a temperature of −196.3 °C was applied to the portions of the valve body in contact with the cooling medium. Because the section of the bonnet below the middle flange is immersed in the cooling medium, and cryogenic tests showed that the temperature inside the valve cavity stabilized at approximately −196 °C, the region at and below the middle flange of the bonnet was treated as an adiabatic boundary, consistent with engineering practice. The section of the bonnet above the long neck and the portion of the stem above the packing were treated as extended heat-dissipation surfaces, with a third-kind (convective) boundary condition applied to these surfaces.

Owing to the vaporization and boiling of liquid nitrogen, a relatively unstable region of cold nitrogen vapor forms above the bonnet flange face. The height of this nitrogen vapor region, measured upward from the flange face, was denoted by x, and the ambient temperature within this region was parameterized with a lower bound of −150 °C. Above the vapor region, the ambient temperature increases along the y-axis according to a power-law relationship with the distance from the heat source.

where y₁is the fluctuating height of the liquid nitrogen, y₂ is the height of the handwheel, and a and b are constants.

Substituting the values into the equation yields:

For the transient thermal analysis, the total simulation time was set to 3,600 s. Time integration was enabled, with an initial time step of 0.5 s, a minimum time step of 0.5 s, and a maximum time step of 72 s. A direct solver was used, while all other parameters were kept at their default settings. The boundary conditions used in the analysis are shown in Figure 6.

Figure 6. Boundary Condition Settings

The packing temperature results are shown in Figure 7. The simulated packing gland temperature is 8.5 °C, which is close to the mean of the normal distribution obtained from the cryogenic test data, thereby supporting the reliability of the numerical simulation.

Cryogenic experiments revealed an irregular layer of nitrogen vapor above the liquid nitrogen surface, resulting from the evaporation of liquid nitrogen and the condensation of water vapor from the surrounding air. In addition, factors such as manual replenishment of liquid nitrogen and ambient humidity affect the convective heat transfer coefficient at the bonnet surface. Therefore, the following parameters were considered in the parametric analysis:

• the convective heat transfer coefficient
• the height of the nitrogen vapor layer
• the temperature of the nitrogen vapor layer

Figure 7. Packing Temperature (°C)

Under steady-state conditions, the convective heat transfer coefficient h between the valve surface and the surrounding air typically ranges from 5 to 10 W/(m²·K). Accordingly, the mean and standard deviation of hh can be determined. The mean values and standard deviations of the nitrogen vapor layer height and temperature obtained from the experimental observations are also presented in Table 6.

Table 6. Parameterization of Influencing Factors

A central composite design (CCD) was selected for the design of experiments (DOE), with the design type set to “AutoDefined.” This option automatically determines the appropriate design type based on the number of input variables, allowing the design space to be adequately sampled and the response surface to be effectively fitted. The design was also configured to allow individual variables to be controlled when selecting design points. For example, the effects of other factors on temperature could be evaluated while keeping the convective heat transfer coefficient constant. Figure 8 illustrates the workflow for the parameterized temperature-field simulation and validation.

Workflow for Parameterized Temperature-Field Simulation and Validation:

Cryogenic Testing → DOE Model Selection → Design Point Selection → Response Surface Generation → Factor Sensitivity Analysis → Experimental Data Collection & Probability Distribution Generation → Simulated Probability Distribution Generation → Similarity Test for Simulation Validation

To facilitate the experimental analysis, the packing gland temperature was used for approximate calculations. Therefore, to ensure the accuracy of the numerical simulation, the packing gland temperature was also set as the output parameter. Based on the mean values and standard deviations of the three input parameters, CCD sampling was used to select a total of 15 design points (see Table 7), and the corresponding packing gland temperatures were calculated.

Table 7 shows the following trends:

1. Effect of the convective heat transfer coefficient: With the nitrogen vapor layer height and temperature held constant, an increase in the convective heat transfer coefficient leads to an increase in the packing gland temperature. Increased ventilation can enhance the convective heat transfer coefficient and thereby increase the packing temperature.
2. Effect of the nitrogen vapor layer height: With the convective heat transfer coefficient and nitrogen vapor layer temperature held constant, a decrease in the nitrogen vapor layer height leads to an increase in the packing gland temperature. Therefore, when measuring the packing gland temperature, the measurement should be taken only after the cooling medium has stopped boiling and the temperatures of the valve body and bonnet have approached −196 °C.
3. Effect of the nitrogen vapor layer temperature: With the nitrogen vapor layer height and convective heat transfer coefficient held constant, an increase in the nitrogen vapor layer temperature leads to an increase in the packing gland temperature. Therefore, higher ambient temperatures are preferable for the experiments to help maintain the packing temperature above 0 °C.

Table 7. The 15 Design Points and Corresponding Packing Gland Temperatures

Once the calculations for all design points were completed, a response surface was constructed. A Kriging model was selected to represent the response surface. The model provides a global representation of the design space while accounting for local deviations. The DOE points were interpolated using the Kriging model, and the resulting response surface was used to calculate the sensitivity contributions of the three parameters (Figure 8). The results indicate that the convective heat transfer coefficient has the greatest influence on the local temperature field.

Figure 8. Sensitivity Contributions of the Three Parameters to the Local Temperature Field

Finally, the probability distribution of the packing gland temperature was generated (Figure 9). Latin hypercube sampling (LHS) was employed. Compared with conventional Monte Carlo sampling, LHS requires 20%–40% fewer simulation runs while providing comparable accuracy. The sample size was set to 235, corresponding to the number of cryogenic globe valves tested experimentally.

Figure 9. Probability Density Distribution of Packing Gland Temperature

When the sample size was increased to 10,000, the calculated minimum packing temperature remained above 0 °C, indicating that the globe valve is unlikely to experience temperature-related failure under room-temperature conditions. Because globe valves may operate under varying environmental conditions, including adverse weather, the effect of ambient temperature was further investigated. Since 0 °C was defined as the critical threshold in this study, the ambient temperature was varied from 0 to 22 °C to determine the temperature at which the packing temperature reached 0 °C. As shown in Figure 10, the packing temperature remained above 0 °C when the ambient temperature exceeded 11 °C. Therefore, an ambient temperature of at least 11 °C is required to maintain the packing temperature above 0 °C during valve operation.

Figure 10. Effect of Ambient Temperature on Packing Gland Temperature

Temperature field experiments were conducted on DN15 cryogenic globe valves, and packing gland temperature data were collected from 235 valves. Statistical analysis showed that the packing gland temperature data followed a normal distribution, providing a reliable probability distribution model and an experimental basis for the subsequent parametric analysis. A 3D model of the cryogenic globe valve was then established and simplified to retain the main components, including the bonnet, valve body, disc, stem, handwheel, and bolts. After mesh generation and the application of boundary conditions, a transient thermal analysis was performed. The simulated packing gland temperature was 8.5 °C, which was close to the mean of the normal distribution obtained from the cryogenic test data, indicating good agreement between the simulation and experimental results.

A central composite design (CCD) was employed for the design of experiments (DOE), and three factors affecting the packing gland temperature were parameterized: the convective heat transfer coefficient, the nitrogen vapor layer height, and the nitrogen vapor layer temperature. The packing gland temperature was calculated for different combinations of these parameters, after which a probability distribution of the packing gland temperature was generated and the sensitivity contribution of each parameter was determined. The results showed that the convective heat transfer coefficient had the greatest influence on the packing gland temperature.