Investigation of the mechanical behavior of a flexible solid metal seal for a cryogenic butterfly valve

Transcription

1 Journal of Mechanical Science and Technology 25 (9) (2011) 2393~ DOI /s Investigation of the mechanical behavior of a flexible solid metal seal for a cryogenic butterfly valve Jun Tae Ahn 1, Kyung Chul Lee 2, Kwon Hee Lee 1 and Seung Ho Han 1,* 1 Department of Mechanical Engineering, Dong-A University, Busan, , Korea 2 Quality Management, Dukwon Valve, Busan, , Korea (Manuscript Received December 7, 2010; Revised April 24, 2011; Accepted May 3, 2011) Abstract Seat tightness at the fully shut position should be a consideration in the development of a butterfly valve for use in a liquefied natural gas (LNG) vessel. A flexible solid metal seal offers sufficient tightness of the butterfly valve and meets the specifications for cryogenic temperature. In the present study, characteristics for a cryogenic butterfly valve, such as the flow coefficient and the pressure loss coefficient, were estimated by numerical fluid analysis carried out to simulate 3-D flow and to study performance as it was affected by the opening angles of the valve disc. A design criterion to ensure the seat tightness of the butterfly valve at the fully shut position was proposed, in which the contact pressure between the metal seal and the valve disc would be compared with the fluid pressure. Numerical structural analysis showed that the contact pressure can be calculated by simulation of the frictional contact behavior on the surface of the metal seal and the valve disc. As a result, an adequate flexibility of the metal seal and the valve disc was required in order to accomplish a contact pressure that would be high enough to satisfy the seat tightness requirement. Under cryogenic temperature, thermal shrinkage caused the metal seal to adhere closely to the valve disc periphery at both sides and raised the contact pressure to a relatively high value, though there was no contact across a small area at the center position, which is susceptible to leakage. An additional displacement of the metal seal and the valve disc appeared at an operating fluid pressure of 6.9 bar and produced sufficient contact pressure at the no-contact area. This was verified by experimental leakage tests performed at room and cryogenic temperatures. Keywords: Cryogenic butterfly valve; Metal seal; Seat tightness; Contact pressure; No-leakage Introduction This paper was recommended for publication in revised form by Associate Editor Seong Beom Lee * Corresponding author. Tel.: Fax.: address: shhan85@dau.ac.kr KSME & Springer 2011 The demand for transportation using the LNG marine system has recently increased significantly [1, 2]. Since the early part of this century, LNG and container vessels have emerged as the main shipbuilding types in the domestic shipbuilding industry. High pressure control valves operated at an extremely low temperature, i.e. cryogenic temperature, are important fluid devices that are widely used to control the storage and delivery of LNG. Many kinds of cryogenic valves, such as the glove valve, gate valve, butterfly valve, and ball valve, are mounted on LNG vessels. Among these, the butterfly valve is used most widely due to its simple configuration, in which a disc element suspended inside a tubular housing is utilized to form the desired obstruction in the fluid flow. In addition, the response is expeditious for on-off operation as well as for flow control, and the pressure loss is less than that of other types of valves [3, 4]. Despite these advantages, butterfly valves operating at cryogenic temperature often exhibit an internal leakage through the unit around the disc periphery at the fully shut position, since an actual sealing function must be accompanied by full metal-tometal contact between the disc and the metal seal [5]. The Velan Co. [6] provides a double flexible metallic O-ring seal that consists of internal and external O-rings, a retaining ring, and an Inconel spring. Each O-ring is double enveloped by an inner ring of stainless steel, an external ring of copper alloy, and an Inconel spring, in which the flexible retaining ring provides a complimentary seating pressure on the disc. To ensure full tightness in cryogenic circumferences, Amri Co. [7] has developed a 3-layer metal seal with both an external and internal sheaths and an Inconel spring to compensate for the relative displacement between the seat and the disc due either to thermal shrinkage or pressure. The degree of tightness at cryogenic temperature is sufficient, but these types of seals are susceptible to high pressure, and a high manufacturing cost is inevitable. As an alternative, a low-cost flexible solid metal seal that meets the requirements for tightness and

2 2394 J. T. Ahn et al. / Journal of Mechanical Science and Technology 25 (9) (2011) 2393~2400 Fig. 2. Boundary conditions for the fluid analysis. The performance of the butterfly valve, as it relates to the opening angles of the disc, was numerically estimated. Here, the flow coefficient, C v, and pressure loss coefficient, K, are the performance indices. C v is the number of cubic meters per hour of water at a temperature between 5 C and 40 C that will flow through the valve given a specified pressure loss when the valve is in the open position [12]. The widely used formula for the C v with a unit of m 3 /hr/ Pa is Eq. (1): G C V = Q P 1 P 2 (1) Fig. 1. Configuration of 500A butterfly valve. strength under high pressure has been developed and marketed [8, 9]. This low-cost seal has a similar capacity to compensate for the relative displacement between the disc and metal seal and actually offers better resistance to high pressure. However, flexibility for the solid metal seal at cryogenic temperature must be assured without the use of an additional support ring. A design configuration that provides the widest available contact surface of the metal seal on the disc is also required. The purpose of the present study was to numerically estimate the characteristics of the butterfly valve due to opening angles of the valve disc. In addition, the results of the contact pressure that occurs on the contact surface of the disc and the metal seal were calculated in accordance with design criterion in development of the flexible solid metal seal to ensure seat tightness. The butterfly valve equipped to the newly developed metal seal must be certified to the satisfaction of regulations BS6755 [10] and BS6364 [11], via leakage tests performed at room and cryogenic temperatures. 2. Characteristics of a butterfly valve 2.1 Features The flanged-type cryogenic butterfly valve analyzed in the present study had a diameter of 500 mm. Fig. 1 provides an overview of a valve configuration consisting of four essential parts: body, stem, disc and metal seal. Problems due to differing thermal expansion were prevented by fabricating all parts from stainless steel - the body and disc were manufactured from SCS14A, and the stem and metal seal from STS316L. 2.2 Performance where P 1 and P 2 refer to the static pressure in a unit of Pa at both upstream and downstream of the valve disc, dimensionless G is the specific gravity, and Q is the flow rate in a unit of m 3 /hour. The dimensionless K is used to relate the pressure loss of a valve to the discharge of the valve at a given valve opening, and is defined by Eq. (2) [12]: P P K = ρv where V and ρ are the maximum velocity and density of fluid in a unit of m/s and kg f /m 3, respectively. The pressure and velocity of fluid for the opening angles of 10 to 90 in increments of 15 were calculated by fluid analysis using the commercial software ANSYS CFX Ver.11. For the fluid analysis, the water at room temperature was taken into account, since the performance indices, such as the C v and K, at cryogenic temperature differ little from those at room temperature [13]. Fig. 2 illustrates a boundary condition of the fluid analysis where the butterfly valve is located between the upstream and downstream pipes. To insure adequate flow fields, the pipe s upstream and downstream lengths were made 5 and 10 times the diameter, respectively. During simulation for the steady state of fluid flow, a uniform temperature was considered, and a standard k-ε model was chosen to predict turbulent flow calculations. For the boundary conditions, a uniform fluid velocity of 4m/s was given at the inlet boundary and an atmospheric pressure of 1atm was imposed at the outlet. Fig. 3 shows the calculated results for the pressure and velocity of the fluid as a contour plot for the disc-opening angles from 10 to 90 in increments of 15. At an opening angle of 10, the pressure loss P 1 -P 2 and the fluid velocity show maximum values of 2.2 bar and m/s, respectively. They gradually decreased as the opening angle was increased. When these results are used in Eqs. (1) and (2), the C v and K due to the change in the disc-opening angles can be shown as Fig. 4. As the valve disc rotated from a closed to a fully opened position, the C v increased from zero to 35 m 3 /hr/ Pa. On the other hand, the K decreased from 30 to zero. A similar trend for C v and K can also be found in the literature [13, 14]. (2)

3 J. T. Ahn et al. / Journal of Mechanical Science and Technology 25 (9) (2011) 2393~ Fig. 5. Schematic illustration of the seat tightness criterion. (a) Pressure contour (b) Velocity contour Fig. 3. Pressure and velocity contours for disc-opening angles. K Disc-Opening Angles ( o ) Fig. 4. Cv and K due to disc-opening angles. C V : Flow Coefficient K : Pressure Loss Coefficient C V (a) Overview Fig. 6. Finite element model for structural analysis. (b) Contact area Since the P contact is strongly influenced by the amount of the relative displacement, the mechanical behavior in the vicinity of the contact surface between the disc and the seal must be taken into account in the design of the flexible metal seal. 3. Criterion for seat tightness Research on the seat tightness of metal seals has been performed mainly to estimate the contact pressure that occurs on the surface between the disc and the seal under fluid pressure [15, 16]. However, there continues to be a lack of research on mechanical behavior, which includes contact pressure for the flexible solid metal seals of butterfly valves in controlling bidirectional flows. In the present study, a design criterion that would ensure seat tightness in a bi-directional flow was proposed as shown in Eq. (3), in which a contact pressure P contact occurred on the contact surface of the valve disc and the metal seal was compared with the fluid pressure P fluid. P contact > P fluid (3) Fig. 5 illustrates the criterion of the seat tightness, where the subscripts n and r refer to the normal and reverse directions of the bi-directional flows. In the fully shut position, the P fluid caused not only a force pushing the seal upward to the disc, but also a relative displacement between the disc and the seal. The friction caused by this force and relative displacement created the P contact. 4. Mechanical behavior of the flexible metal seal 4.1 Modeling for structural analysis The mechanical behavior of the metal seal was investigated numerically by structural analysis using ANSYS Workbench Ver.12 [17]. Fig. 6(a) shows an overview of the finite element model generated by pre-processing for numerical analysis. The contact surface between the disc and the metal seal was implemented using a contact tool in the pre-processing option, for which the surface of the disc represents the contact surface and the surface of the metal seal is the target surface. The CONTA 174 element [17] was applied to the finite element model, as shown in Fig. 6(b), which simulated the contact and sliding between the 3-D target surfaces and the deformable surface. The pure penalty method, as a contact algorithm, and the isotropic friction model were used as a constant coefficient of friction based on the assumption of uniform stick-slip behavior in all directions [17]. The loading conditions applied a maximal pressure P fluid of 6.9 bar bi-directionally: normal and reverse pressure. The boundary conditions were given as follows: the stem and seat were fixed, the seat and seal were bonded, and the coefficient of friction was given as 0.3 for the contact conditions between the disc and the metal seal.

4 2396 J. T. Ahn et al. / Journal of Mechanical Science and Technology 25 (9) (2011) 2393~2400 (a) Normal pressure (a) Overview (b) Reverse pressure Fig. 8. Relative displacement of the contact area at room temperature. (b) Loading conditions at the inlet and outlet pipes Fig. 7. Loading and boundary conditions of the TSI analysis. To investigate the effect of cryogenic temperature on the mechanical behavior of the metal seal, thermo-structure interaction (TSI) analysis was carried out. The loading and boundary conditions were imposed with a pressure and temperature for the inlet and the outlet of the pipe at -196 C and 6.9 bar, respectively, and with convection conditions applied to the other parts of the butterfly valve. Fig. 7 is a schematic illustration of the loading and boundary conditions used in the TSI analysis. The thermal conductivity, coefficient of thermal expansion and elastic modulus in relation to the cryogenic temperature, which are required for TSI analysis, were taken from the literature [2, 18]. (a) Normal pressure 4.2 Numerical results (b) Reverse pressure Fig. 9. Shape and location of the contact zone at room temperature. The contact surfaces at the one-quarter and center positions were selected for the present study, as shown in Fig. 6(a), because this is where both the maximum and minimum numerical results were found. Fig. 8 shows the relative displacement for the contact areas at the one-quarter and center positions for a bi-directional P fluid of 6.9 bar at room temperature. For better understanding, the displacement was magnified using a scale of 200 provided by a post-processing tool from ANSYS Workbench. The displacement of the disc at the center position was significantly small under both normal and reverse pressure due to its geometric shape where the stem and supported body provided high stiffness for the disc (see Fig. 6(a)). On the other hand, the disc at the one-quarter position was deformed significantly. In the case of the metal seal, the displacement at the onequarter position was also shown to be much larger than that at the center position. This can be caused by an interaction of the relative displacement, which takes place at the contact surface of the disc and the metal seal. Therefore, the displacement of the metal seal increased with an increase in the displacement of the disc. When the direction of the P fluid was changed, there was a different manner of the displacement behavior. This caused a change of the shape and location of the contact zone of the disc and the metal seal, in which the P contact took place.

5 J. T. Ahn et al. / Journal of Mechanical Science and Technology 25 (9) (2011) 2393~ Table 1. Numerical results at a room temperature of 20 C. Positions Displacement (10-3 mm) Disc Seal Contact pressure (MPa) Normal pressure 1/ Center Reverse pressure 1/ Center (a) B/Cs: inside inlet pipe (b) B/Cs: inside outlet pipe Table 2. Numerical results at a cryogenic temperature of -196 C. Fig. 10. Thermal shrinkage under cryogenic temperature imposed at the inlet and outlet pipe. B/Cs* Positions Thermal shrinkage (mm) Disc Seal Contact pressure (MPa) Inside inlet pipe 1/4 Center Inside outlet pipe 1/4 Center (a) Normal pressure * : A temperature of -196 C was imposed on the inside of both the inlet and outlet pipes, which are the boundary conditions for TSI analysis. : Displacement and contact pressure was caused by the P fluid following thermal shrinkage at -196 C. Fig. 9 shows the contact zone on the metal seal under bidirectional fluid pressure. The shape of the contact zone at the center and one-quarter positions differs in some respects. The contact zone at the center position was scattered on the surface of the metal seal, but the contact zone at the one-quarter position had the shape of a line. The shape of the contact zone is dependent on the behavior of the relative displacement of the disc and the metal seal. At the center position showing a small displacement, the contact occurred slightly on the spotted contact surface, and the P contact was small. Meanwhile, the P contact at the one-quarter position was high enough where a strong interaction of the relative displacement of the disc and the metal seal took place. In addition, this interaction under a bi-directional fluid pressure influenced the location of the contact zone. The contact zone under normal pressure occurred at the rear of the metal seal, while under reverse pressure it occurred at the front. The relative displacement of the disc and the metal seal under bi-directional fluid pressure in the butterfly valve affected not only the shape and the location of the contact zone on the metal seal, but also the magnitude of the contact pressure. The details of the numerical results are listed in Table 1. This shows that the P contact increases significantly when the displacement of the disc and the metal seal increases. Consequently, an adequate flexibility of both the disc and the seal is (b) Reverse pressure Fig. 11. Relative displacement of the contact area caused by the thermal shrinkage under cryogenic temperature imposed at the inlet and outlet pipes and bi-directional fluid pressure P fluid. required in order to acquire a contact pressure that is high enough to meet the requirements of seat tightness. At room temperature, the butterfly valve investigated in the present study showed that the P contact was greater than the P fluid of 6.9 bar (0.69 MPa), which satisfies the criterion for seat tightness, according to Eq. (1). TSI analysis was carried out to investigate the effect of cryogenic temperature on the mechanical behavior of the disc and the metal seal. Contour plots of the thermal shrinkage for the valve body under cryogenic temperature imposed at the inlet and outlet pipes are shown in Fig. 10. The displacement presented as thermal shrinkage that occurred when the temperature changed from 20 C to -196 C was significantly large in comparison to that at room temperature. The numerical results under these loading conditions imposed by a cryogenic temperature are listed in Table 2, wherein the values marked

6 2398 J. T. Ahn et al. / Journal of Mechanical Science and Technology 25 (9) (2011) 2393~2400 (a) Normal pressure Fig. 13. Schematic illustration of the experimental set-up for the leakage test at cryogenic temperature. (b) Reverse pressure Fig. 12. Shape and location of the contact zone at cryogenic temperature. (a) At room temperature (b) At cryogenic temperature Fig. 14. Photos of the experimental set-up for the leakage test. by mean the additional displacement and contact pressure at P fluid of 6.9 bar after thermal shrinkage was completed at C. The imposition of a cryogenic temperature to the inside of an inlet pipe produces a displacement of 1.94~2.08 mm, but another case shows smaller values of 1.03~1.22 mm. This can be caused by a larger surface area on the inside of the inlet pipe being imposed as the boundary conditions of temperature. Fig. 11 shows the relative displacement for the contact areas of the valve disc and metal seal, where the displacement was magnified by using a scale factor of 10. At the one-quarter position, the thermal shrinkage allowed the metal seal to adhere closely to the disc with a contact pressure of 42.7 MPa. At the center position, however, the gap between the disc and the metal seal became wider and the criterion for the seat tightness could not be satisfied. This phenomenon took place distinctly under the imposition of -196 C on the inside of the inlet pipe, because the massive volume around the center position, such as the stem and extended bonnet structures, can be a cause for the disagreement in thermal shrinkage. The normal P fluid of 6.9 bar after thermal shrinkage for the imposition of a cryogenic temperature of -196 C at the inside of the inlet pipe caused an additional displacement of the valve disc and the metal seal. This additional displacement created a contact pressure of 0.9 MPa at the center position, which satisfies the criterion for seat tightness. Although the additional displacement that occurred from normal pressure after the thermal shrinkage was small, it affected the interaction of the relative displacement of the disc and the metal seal even at this low temperature. Fig. 12 shows the numerical results for the shape and location of the contact zone on the metal seal at cryogenic temperature under bi-directional fluid pressure. Distinct lineshaped contact zones at the one-quarter position are shown due to their high contact pressure. Meanwhile, an irregular contact zone appears at the center position. In the case of normal pressure, no contact area occurred at the center position, i.e. near the periphery at the connection of the stem and disc, although the average values of the contact pressure were calculated to be 0.9 MPa higher than a fluid pressure of 6.9 bar (0.69 MPa). Since this could exhibit an internal leakage under normal pressure through the no-contact area at the fully shut position, an experiment for leakage testing was required to verify whether the proposed flexible solid metal seal could ensure the required seat tightness according to international regulations related to this case. 5. Experiment setup for the leakage test A leakage test was carried out at room and cryogenic temperature according to regulations BS6755 [10] and BS6364 [11], respectively. At room temperature, the butterfly valve was tested using helium gas as an inert gas at 6.9 bar with 6 steps, and at the maximum permitted leakage rate for a valve with a metal-to-metal seal, 0.3 mm 3 /s x DN where DN is designated to the diameter of the valve. The leakage test at cryogenic temperature was performed using a low-temperature type test rig as shown in Fig. 13. The valve was set into an insulated tank and was soaked at -196 C for at least 1 hour until the temperature had stabilized. The valve body was pressurized up to 6.9 bar in a bi-directional flow direction with close/off on the outlet side, and this was maintained for a period of 15 minutes. The applied pressure was increased stepwise by 6 steps of equal height up to 6.9 bar, and then the leakage rate was measured using a flowmeter. The requirement of the leakage rate was not to exceed 100 mm 3 /s x DN according to BS6364 [11]. Fig. 14 shows photos of the ex-

7 J. T. Ahn et al. / Journal of Mechanical Science and Technology 25 (9) (2011) 2393~ Leakage Rate (10 3 mm 3 /sec) perimental set-up for the leakage test at room and cryogenic temperature. Results of the measured leakage rates are plotted in Fig. 15. No-leakage was observed during the leakage test performed at room temperature. However, the leakage rate at cryogenic temperature increased incrementally to 3,300 mm 3 /s with a gas pressure increase of up to 6.9 bar. The leakage was observed to increased as the gas pressure increased. This can be explained by the no-contact area at the center position near the periphery at the connection of the stem and disc, as shown in Fig. 11(a). Although the leakage rate reached as high as 3,300 mm 3 /s, taking into account the DN value of 500 mm at the maximum gas pressure of 6.9 bar, this was much smaller than that of the maximum permitted leakage rate of 50,000 mm 3 /s recommended in BS6364 [11]. Consequently, via a leakage test performed at room and cryogenic temperature, the butterfly valve equipped to the newly developed metal seal was certified to the satisfaction of requirements BS6755 and BS Conclusions At Room Teprature At Cryogenic Temperature Pressure (Bar) Fig. 15. Test results of the leakage rate due to the pressure of helium gas at room and cryogenic temperature. The performance of the butterfly valve, e.g., flow coefficient C v and pressure loss coefficient K, was numerically estimated. As the valve disc rotated from closed to fully opened positions, the C v increased up to 35 m 3 /hr/ Pa, and the K decreased from 30 to zero. A design criterion for the seat tightness was proposed as an inequality formula of the P contact and P fluid, where a higher value of P contact is required to ensure seat tightness. The P contact was strongly affected by the frictional displacement of the disc and the metal seal. At room temperature, the P contact increased significantly when the displacement of the disc and the metal seal increased. Consequently, an adequate flexibility of disc and seal is required in order to accomplish a contact pressure that is high enough to satisfy the seat tightness requirement. At cryogenic temperature, the thermal shrinkage caused the metal seal to adhere closely to the disc and raised the contact pressure to a relatively high value. However, the gap between the disc and the metal seal at the center position widened, at which point the criterion for seat tightness could not be satisfied. When the P fluid was applied, an additional displacement of the disc and the metal seal took place and produced a contact pressure that met the criterion for seat tightness. Via a leakage test performed at room and cryogenic temperature, the butterfly valve equipped to the newly developed metal seal was certified to the satisfaction of requirements BS6755 and BS6364. Acknowledgement The authors are grateful for the financial support provided by research funds from Dong-A University. Reference [1] Y. J. Ahn, B. J. Kim and B. R. Shin, Numerical analysis of 3-D Flow through LNG marine control valves for their advanced design, Journal of Mechanical Science and Technology, 22 (2008) [2] D. K. Kim and J. H. Kim, A study on structure analysis of glove valve for LNG carrier, Journal of the Korean Society of Marine Engineering, 31 (8) (2007) [3] X. G. Song, L. Wang, S. H. Baek and Y. C. Park, Multidisciplinary optimization of a butterfly valve, ISA Transaction, 48 (2009) [4] T. Kimura, T. Tanaka, K. Fujimoto and K. Ogawa, Hydrodynamic characteristics of a butterfly valve-prediction of pressure loss characteristics, ISA Transaction, 34 (1995) [5] Feature butterfly valves, new developments in triple-offset butterfly valves, World Pumps (September 2004) [6] Cryogenic metal seated butterfly valves, www. velan.com. [7] Valves all along the LNG chain, [8] BFGoodrich, Metal C-Rings, [9] E-Seals, JE/JH Series, [10] BS6755, Testing of valves - Part 1 : Specification production pressure testing requirements, British Standard Institution (1986). [11] BS6364, Specification for valve for cryogenic service, British Standard Institution (1984). [12] P. L. Skousen, Valve handbook - 2 nd Edition, McGraw-Hill, USA. (2006). [13] S. W. Kim, Y. D. Choi, B. S. Kim and Y. H. Lee, Flow characteristics of cryogenic butterfly valve for LNG carrier (Part 2 : Flow characteristics under cryogenic condition), Journal of KFMA, 11 (2) (2008) [14] L. Wang, X. G. Song and Y. C. Park, Dynamic analysis of fluid flow passing a large butterfly valve, 3 rd global conference on PCO, Gold Coast, Australia (2-4 Feb. 2010). [15] S. M. Lee, S. X. Quan, J. H. Kang and Y. C. Park, Analysis of butterfly valve s metal seat using F.E.M, KSMPE Spring Conference, Jeju, Korea (2008) [16] R. D. Villepoix, M. Lefrancois, J. Montucard and C. Rouaud, Latest development on valve seat-seal assembly, Fusion Engineering and Design, 29 (1995) [17] ANSYS Workbench - Simulation Introduction, Release 12.0 (2009),

8 2400 J. T. Ahn et al. / Journal of Mechanical Science and Technology 25 (9) (2011) 2393~2400 [18] J. K. Kim, Application of LNG to complex power generation system (internal report), Kumho Engineering Co. (1999) Jun Tae Ahn received his B.S degree in Mechanical Engineering from Donga University, Korea, in He is currently an M.S. student at the Department of Mechanical Engineering of the same university. His research interests include machine elements design, fatigue life design and FSI-analysis. Seung Ho Han received his B.S. and M.S. degrees in Mechanical Engineering from Hanyang University, Korea, in 1989 and 1991, respectively. He then received his Doctor degree from RWTH Aachen, Germany, in Dr. Han is currently a Professor at the Department of Mechanical Engineering of Dong-A University in Busan, Korea. Dr. Han s research interests include reliability based design and fatigue life extension technique.