Sealing Performance Analysis and Optimization of High-Pressure Hard-Sealed Trunnion-Mounted Ball Valves

2026-09-18 08:55:32

Due to the structural complexity of high-pressure hard-sealed trunnion-mounted ball valves, the forces under sealing conditions become more complex, making it difficult to use traditional analytical methods effectively. To ensure safe operation, this paper applies the finite element method to analyze the sealing performance and stress distribution of a Class 1500 NPS 14 worm-gear-operated hard-sealed trunnion -mounted ball valve in the closed state. The analysis results show that under the action of the medium load, severe stress concentration occurs around the valve stem and the root of the gland mounting hole; the macroscopic through-leakage flow rate is 5268.594 mL/min, and the specific sealing pressure is severely insufficient.

Ball valves are widely used in the petrochemical industry, especially the coal chemical industry, due to advantages such as low fluid resistance, convenient operation, rapid opening and closing, good sealing performance, and high reliability. However, because coal chemical raw materials contain corrosive media, hard particles, etc., metal hard-sealed ball valves that are high-pressure resistant, corrosion resistant, wear resistant, and suitable for a wide temperature range are required. This type of ball valve achieves sealing by pressing the ball and seat together under the dual action of preload and fluid pressure.

Ou Guofu et al. from Zhejiang Sci-Tech University studied the contact characteristics of the sealing pair of metal hard-sealed ball valves, especially the sealing performance of rough contact planes. The results show that under the condition of satisfying sealing requirements, selecting a larger pressure angle can reduce friction torque; when the radial projected width of the sealing surface is greater than 7 mm, the sealing is not tight, and when it is less than 5 mm, stress concentration occurs; the specific sealing pressure required for transverse texture characteristics is much smaller than that for longitudinal texture surfaces; the larger the sealing width, the smaller the required specific sealing pressure; under the same leakage rate, the specific sealing pressure required at high temperature is significantly greater than that at room temperature.

Yu Yong, Li Liancui, et al. from Lanzhou University of Technology also analyzed and studied the sealing performance of hard-sealed ball valves, mainly concerning the effects of materials, low pressure, and ultra-low temperature on the sealing performance of the valve seat. A review of relevant literature shows that there are no reports on research related to large-diameter trunnion-mounted ball valves with pressures above 20 MPa. Based on this, this paper conducts sealing performance analysis and optimization for the forces on a Class 1500 NPS 14 worm-gear-operated hard-sealed trunnion-mounted ball valve (hereinafter referred to as the hard-sealed trunnion-mounted ball valve) in the closed state, providing a reference for the optimal design and use of hard-sealed trunnion-mounted ball valves.

1. Sealing Principle of the Hard-Sealed Trunnion-Mounted Ball Valve

According to relevant theories of mechanical design and material strength, the main failure mode of a hard-sealed trunnion-mounted ball valve is leakage while ensuring that the valve body, ball, seat, stem, etc. do not undergo plastic deformation. Therefore, it is necessary to understand in advance the sealing principle of the sealing pair of this type of ball valve, so as to conduct an in-depth discussion. Figure 1 shows the overall half-section model of the ball valve, and Figure 2 shows the locally enlarged view within the red box in Figure 1. When the ball rotates 90° so that the valve is in the closed state, the fluid is cut off with the seal ring and O-ring as boundaries. At this time, the valve seat (inlet end) is subjected to the combined action of the spring preload F, fluid pressure P, and the force exerted by the ball on the seat. Since the resultant force of F and G to the left is greater than that to the right (the horizontal white arrows in Figure 2 indicate the regions corresponding to the forces), the rightward force needs to be offset by the force exerted by the ball on the seat. The pressure provided by this force is the specific sealing pressure. If the specific sealing pressure is greater than the required specific sealing pressure, no leakage will occur.

1. Sealing Performance Analysis and Optimization

2.1 Main Dimensions of the Model

The main dimensions involved in the model for this analysis are as follows: total length of the ball valve: 1276 mm; minimum wall thickness of the valve body: 156 mm; ball radius: 260 mm; inner diameter of the inlet, outlet, ball, and valve seat: 315 mm; minimum outer diameter of the valve seat: 332 mm; total length of the valve seat: 89 mm; upper stem diameter: 130 mm; gland outer diameter: 220 mm; minimum wall thickness of the gland: 32 mm; lower stem diameter: 100 mm.

2.2 Model Simplification Strategy

In this paper, a preliminary trial calculation was carried out by assigning the same material properties to all components (material name: ASTM A182 F304 nickel alloy, elastic modulus: 195 GPa, Poisson's ratio: 0.29, yield strength: 215 MPa). During the trial calculation, fixed constraints were applied at the flanges at both ends of the valve body, axial constraints were applied to the upper and lower stems, contact pairs connected by bolts and interference fits were set as bonded contacts, the valve seat and washer also used bonded contact, the O-ring was ignored, and other contact pairs that may produce relative sliding were set as frictional contacts, with the friction coefficient set to 0.15. According to the characteristics of stress and deformation distribution, the simplification strategy for the finite element analysis model of the ball valve was formulated as follows:

Because the valve body has very high stiffness and its deformation is negligible, the valve body is omitted.

The ball is supported by the upper and lower stems in the vertical direction, and there is no fluid action in this direction. Therefore, the supporting effect of the upper and lower stems in this direction can also be regarded as infinite; the upper stem is omitted, and an axial fixed constraint is applied.

The ball is supported by the gland and the lower stem in the inlet and outlet directions, respectively. Since their stiffness is sufficiently high that the deformation after fluid action is negligible, the supporting effect of the gland and lower stem in this direction is also regarded as infinite; the gland and lower stem are omitted, and axial fixed constraints in the inlet and outlet directions are applied.

The O-ring and washer have no influence on the contact between the ball and the valve seat. Therefore, when modeling, the grooves for the O-ring and washer are filled flat, and the O-ring and washer are omitted.

Because it is a trunnion-mounted ball valve, symmetry can be used. That is, the final model consists of 1/8 of the ball and 1/4 of the valve seat.

2.3 Boundary Conditions

The application of boundary conditions is shown in Figure 4. In the figure, A (hatched area) represents the restraining effect of the gland on the ball, which is set as a fixed constraint along the Z-axis; B (area with "#" lines) represents the restraining effect of the upper stem on the ball, which is set as a fixed constraint along the X-axis; C (darkened area) is where the valve body provides radial restraint on the valve seat, which is set as a radial fixed constraint; symmetry constraints are applied to each cut plane.

The contact pair between the valve seat and the ball is set as a stainless steel–stainless steel contact pair. To effectively simulate the actual contact condition, it is set as frictional contact under asymmetric conditions. Because the sealing surface has high machining accuracy and low surface roughness, the friction coefficient is taken as 0.15.

According to the conversion relationship between American standards and Chinese standards, the fluid-applied pressure is P = (1500 × 2 + 1500 / 2) × 0.00689 = 26 MPa, and its application positions are shown in Figures 4b and 4c (darkened areas). Assume the spring pre-compression is 6 mm, the spring stiffness is 63 N/mm, and the number of springs is 40; then the spring preload acting on the 1/4 valve seat is F = 40 × 63 × 6 / 4 N = 3740 N. Its application position is shown in Figure 4c; that is, the circled position in Figure 4c is subjected to both F and P. The fluid pressure P and the spring preload F are applied in two time steps. In the first time step, F is applied; in the second time step, P is applied while F remains unchanged.

2.4 Mesh Independence Verification

In finite element analysis, the mesh density has a certain influence on the solution accuracy. Generally, the denser the mesh, the more accurate the solution results, but the computational workload is large, the computer configuration requirements are higher, and the solution time is lengthy. Therefore, under the premise of ensuring computational accuracy, it is necessary to seek the optimal mesh size to shorten the computation time.

Settings are made according to the material properties and boundary conditions in Section 2.2. Taking the minimum mesh size as the baseline and a relative deformation error of 0.1% for the valve seat as the solution target, adaptive meshing is performed for the ball and contact surface by successively reducing the mesh sizes of 16 mm and 4 mm by 1/2^n (n = 0, 1, 2, 3, 4). The results are shown in Table 1. It can be seen from Table 1 that the error of the third group of data is already less than 0.1%. Therefore, considering the computation time, the ball element size is finally taken as 4 mm, and the contact surface element size is taken as 1 mm. Since the valve seat has a relatively small volume and has little influence on computer configuration and computation time, mesh independence verification is not performed for it, and 1 mm is taken. The meshing result is a hexahedron-dominated structured mesh, with 3,394,370 nodes, 899,007 elements, and a mesh quality of 0.827. The meshing result is shown in Figure 5.

The contact pair between the valve seat and the ball is set as a stainless steel–stainless steel contact pair. To effectively simulate the actual contact condition, it is set as frictional contact under asymmetric conditions. Because the sealing surface has high machining accuracy and low surface roughness, the friction coefficient is taken as 0.15.

According to the conversion relationship between American standards and Chinese standards, the fluid-applied pressure is P = (1500 × 2 + 1500 / 2) × 0.00689 = 26 MPa, and its application positions are shown in Figures 4b and 4c (darkened areas). Assume the spring pre-compression is 6 mm, the spring stiffness is 63 N/mm, and the number of springs is 40; then the spring preload acting on the 1/4 valve seat is F = 40 × 63 × 6 / 4 N = 3740 N. Its application position is shown in Figure 4c; that is, the circled position in Figure 4c is subjected to both F and P. The fluid pressure P and the spring preload F are applied in two time steps. In the first time step, F is applied; in the second time step, P is applied while F remains unchanged.

2.4 Mesh Independence Verification

In finite element analysis, the mesh density has a certain influence on the solution accuracy. Generally, the denser the mesh, the more accurate the solution results, but the computational workload is large, the computer configuration requirements are higher, and the solution time is lengthy. Therefore, under the premise of ensuring computational accuracy, it is necessary to seek the optimal mesh size to shorten the computation time.

Settings are made according to the material properties and boundary conditions in Section 2.2. Taking the minimum mesh size as the baseline and a relative deformation error of 0.1% for the valve seat as the solution target, adaptive meshing is performed for the ball and contact surface by successively reducing the mesh sizes of 16 mm and 4 mm by 1/2^n (n = 0, 1, 2, 3, 4). The results are shown in Table 1. It can be seen from Table 1 that the error of the third group of data is already less than 0.1%. Therefore, considering the computation time, the ball element size is finally taken as 4 mm, and the contact surface element size is taken as 1 mm. Since the valve seat has a relatively small volume and has little influence on computer configuration and computation time, mesh independence verification is not performed for it, and 1 mm is taken. The meshing result is a hexahedron-dominated structured mesh, with 3,394,370 nodes, 899,007 elements, and a mesh quality of 0.827. The meshing result is shown in Figure 5.