Tribological Wear of Ball Valve Seats MANE 6960- Friction, Wear, and Lubrication of Materials Nathaniel R. Anker 5/16/2015 Table of Contents Abstract ......................................................................................................................................................... 2 Introduction ................................................................................................................................................... 2 Exxon Case Study ......................................................................................................................................... 3 Ben Jemaa Tribometer .................................................................................................................................. 4 Finite Element Model ................................................................................................................................... 5 Full Valve Model ...................................................................................................................................... 6 Geometry............................................................................................................................................... 6 Material ................................................................................................................................................. 7 Results ................................................................................................................................................... 8 Micro-Structure Model ........................................................................................................................... 10 Geometry............................................................................................................................................. 10 Results ................................................................................................................................................. 10 Conclusions ................................................................................................................................................. 12 References ................................................................................................................................................... 12 Abstract The study of tribology is crucial in the design of valves. Due to the interaction between the sealing elements of the valve, wear occurs that directly affects the life cycle of a valve in service. Specifically, in ball valves a seal is created between the valve ball, usual a metal material, and the valve seats made of a softer material. Due to the mismatch in material hardness, the valve seats exhibit wear. In this paper several different methods of examining the loading configuration and the affect it has on the wear the valve components will be studied. Introduction Valves are essential type of fitting to any system as they regulate control and regulate fluid flow within that system. There are numerous different types of valves such as ball valves, globe valves, gate valves, butterfly valves, and many more. However, most valves can be broken down into the same list of general components. The typical components found in the valve include the valve body, a plug, seats, a stem, and packing to seal the valve. Figure 1- Valve Components (Ref (a)) As a valve is designed to regulate flow it is critical that the plug properly contacts the seat surface to create a seal that is leak-proof. Seats are classified in two separate categories: hard seats and soft seats. Hard seats, made of metal and integral to the valve body, are typically found in globe and gate valves. Soft seats, made of plastics or elastomers, are typically found in ball and butterfly valves. Hard seat ball valves are not typically used as it is difficult to obtain a leak proof seal due to the metal on metal contact between the seat and ball. However, with the improvements of modern machining using CNC devices, hard seat ball valves now have applications in power plants where soft seat ball valves fail due to high temperature and pressure. Soft seat ball valves are often advantageous due to low cost as well as being small in nature they are easy to design into a system. As these are seats are made of a much softer material than the metal ball, the design life of the valve seats must be taken into consideration due to seat wear. In this paper the various methods used to study ball valve seat wear will be reviewed. Exxon Case Study ExxonMobil was experiencing an issue with valves failing in services due to heavy abrasive wear on the ball valve seats that saw a high cycle rate. A valve that had a historic service life of two years was failing in 8 months. The sealing pressure had been increased from 1000 psi to 4000 psi. To make up for this change in sealing pressure the seat material hardness was increased from 371 Bhn to 475 Bhn. Due to the failures seen in use from these changes it was recommended to increase the hardness to 613 Bhn while reducing the sealing pressure to 2000 psi. Test data did not exist to verify if this would increase the design life of the valve so the following model was created. Figure 2- Wear Model (Ref (b)) In the model, as seen in Figure 2, the wear component is the softer material (valve seat) and the wear volume is the harder material (ball). Where S is sliding distance, πis the applied pressure load, V is the sliding velocity, and πΏ is the amount of wear that occurs. The following equation is derived from the model to solve for the amount of time in years, t, for a given amount of wear, πΏ . π‘= π‘π‘ 10π‘π‘π‘π‘ Where H is the Brinnell hardness of the wear component, K is the wear coefficient, D is a geometrical factor in inches, and N is the number of cycles. Based on this model is was calculated an additional design life of 1.65 years. Based on this data, the system went into a testing phase where the model was verified after running successfully for three years. Ben Jemaa Tribometer The above model is convenient to study the life cycle of a given valve configuration, however due to its simplicity it does not provide insight into the loading seen in the valve seat. Actual performance testing of the valve materials to understand how the material tribological properties is the only way to truly understand the valve’s life cycle. One way to do this is by creating a tribometer as seen in Reference (b). A tribometer is a device that models the contact configuration between the ball and seat. Reference (b) examines a case study in which a tribometer is used to examine the contact area between a brass ball between two polytetrafluorethylene (PTFE) seats. In order for the tribometer to accurately portray the wear found in a ball valve, it is designed to meet several important criteria of found in the interaction between the valve seats and the ball. The configuration of the tribometer mimics the contact of the ball and seats in an actual ball valve. The loading of the tribometer allows for the clamping loads applied to valve seats and the sliding of the ball between the seats. A seat clamping force anywhere from 100-1000 Newtons can be achieved. The sliding frequency to simulate cycling of the ball valve can operate anywhere from 0.5-2 Hertz. The tribometer measures several variables during operation including the seat clamping force and operating friction torque. Measuring this variables allows for studying several important features such as creep and relaxation of the valve materials. PTFE is a commonly used seat material in ball valves due to its many advantageous properties. PTFE has a high chemical resistivity allowing it to be used in valves for various fluids. Additionally it has a low coefficient of friction that allows the ball valve to be operated at lower torques even at greater seat clamping forces. Additionally PTFE has a higher melting temperature than other plastic materials. Despite this, a major disadvantage of PTFE is that it has a low fatigue life due to poor wear performance. As the seats wear a leak path past the ball is created rendering the valve useless. In the study a brass ball was used. Brass is a typical material used in ball valves and is much harder than PTFE, and therefore will not wear as rapidly. The ball was coated in chromium with an average roughness of 0.02 micrometers. Chromium is another material found in ball valves due to its high corrosion resistance and hardness. Finite Element Model As mentioned above, the best way to understand any tribology system is through testing. In the tribometer created in the Ben Jemaa case study certain assumptions were made to replicate the interaction in a ball valve between the ball and the seat material. The rest of this paper will describe a model using the finite element method in COMSOL used to validate some of the results found in the Ben Jemaa case study. Full Valve Model Geometry The contact between the seats and the ball within the valve can be simply described as the interaction between a cylinder and a sphere; where the seat is represented by the cylinder and the valve ball by a sphere. In order to validate the Ben Jemaa case study it is also crucial to dimension the components similarly. The following parameters were defined in COMSOL. Name Expression Description r_b 11.5e-2 Ball radius r_s 1.5e-2 Seat radius Table 1- COMSOL Geometry Parameters To model the geometry a 2D axisymmetric domain is chosen. A quarter circle with a radius of r_b is created at the center. Due to symmetry along the z and r axes only a quarter of the ball needs to be models, allowing for a much faster computing time. Another circle was created 45 degrees from the z axis with a radius of r_s to represent the valve seat. To complete the model geometry a normal density mesh was generated in COMSOL. The geometry and mesh can be seen below in Figure _. Figure 3- Geometry Material The material was also selected to recreate the Ben Jemaa Case Study. The ball was represented by the following properties per Reference (a) to represent brass. Name Value Unit Young's modulus 97e9 Pa Poisson's ratio 0.31 1 Density 8490 kg/m^3 Table 2-Brass Material Properties To represent the PTFE seats the properties from COMSOL’s material library were used. As the wear of the seats is being studied, additional properties were added to allow for plastic deformation of the PTFE seats. The yield stress of PTFE per Reference (a) is used. Assumptions were made on isotropic tangent modulus based on the study of stress-strain curves of PTFE. Name Value Unit Density 2200[kg/m^3] kg/m^3 Young's modulus 0.4e9[Pa] Pa Poisson's ratio 0.48 1 Initial yield stress 12e6 Pa Isotropic tangent modulus 0.4e4 Pa Table 3- PTFE Material Properties Loading Conditions In the model each surface must be loaded correctly to ensure the valve is constrained properly. As described in the Geometry, due to axial symmetry only a quarter of the valve ball is modeled. Both flat edges of the quarter circle were constrained by symmetry boundary conditions. Between the two surfaces a contact pair is created to ensure there is adequate interferences between the two bodies. In addition to the contact, COMSOL allows for friction to be accounted for at the contact pair. The contact pair can be seen below in Figure 4. Figure 4 -Contact Pair Finally, a boundary load is applied to top half of the seat to simulate the clamping load applied to the seat by the valve body. Various boundary loads were used to compare the results found in the Ben Jemaa Case Study. Results A surface fringe plot of the Von Mises stress can be found below. Under a boundary load of 1000 N, the peak stress was approximately 1.1 MPa, well below the PFTE yield stress of 12 MPa. As such the seats did not undergo plastic deformation in this load condition. Figure 5- Surface Von Mises Stress As the loading was within the elastic region, the seat did not see stresses beyond yield as such there was no plastic strain present. As expected the peak strains can be found in the area of contact between the seat and the ball. A fringe plot of the strain can be found below. Figure 6- Surface Strain As the values for stress and strain were found to be low in the nominal case several other finite element runs were performed for various boundary loads. As expected, with a larger clamping load imparted on the seat, a greater stress is found. For typically used clamping loads no plastic deformation occurs. In order to find tangible plastic strains an unrealistic clamping load of 20,000 N was used. While a clamping load this high may create a strong seal between the seats and ball it is highly disadvantageous as the valve operating torque becomes unreasonably high as well. Data for the different boundary loads can be found in Table 4. Load 100 1000 2000 20000 Strain Stress Strain Displacement (plastic) 1.24E+05 2.03E-04 1.04E-05 N/A 1.23E+06 0.00204 1.05E-04 N/A 2.46E+06 0.0041 2.12E-04 N/A 1.34E+07 0.04781 0.00218 0.02441 Table 4- Boundary Load vs Stress/Strain/Displacement Micro-Structure Model The above macro model does not describe the entire story in regards to the wear seen in a ball valve. As designed the loading is well below the yield strength of both materials allowing for a low risk of failure. However, even a failure on the microscopic level can create a leak path past valve seats. To verify this a second finite element model was created at a much smaller scale. Geometry As mentioned in the Ben Jemaa study, the ball has an average roughness of 0.02 micrometers. The model below created to equilateral triangles with a height of the average roughness value. Figure 7-Microstructure Geometry The upper material is the PTFE seat while the lower material represents the brass ball. The lower edge of the brass was fully contstrained. The upper edge of the PTFE seat was given a downward displacement into the sphere. Results Seen below is the surface Von Mises stress in the contact area between the two asperities. It is noted the peak stresses are found in the brass. However the peak stress, approximately 71 MPa, is below the yield stresses found in brass, 124 MPa. Figure 8 - Microstructure Asperity Von Mises Stress It is observed that the stresses found in the area of contact in the PTFE are approximately equal to the material yield strength. However, this is due to nature of PTFE in plastic deformation, as the tangential elastic modulus is much lower compared to the Young’s modulus. Figure 9- Plastic Strain In the areas above the PTFE yield stress deformation occurs by way of plastic strain. This is due to the fact that the PTFE is a weaker and softer material than brass. The data discussed here is for a single static displacement. As the valve cycles potentially thousands of times in a life cycle contact of this nature will occur, creating larger values of strain. Eventually, fractures will occur creating a leak path through the lost material. Additionally wear issues could occur such as erosion due to the debris or metal asperities ploughing through the seat material. Conclusions Based on the results of the finite element analysis, it can be shown the models used to investigate valve seat wear adequately predict stresses and strains found in operation. The theoretical and finite element models accurately predict preliminary stresses and strains for different material combinations for seat and ball materials. It is also advantageous to use this methods during preliminary design as they can often be cheaper than testing. However, testing offers the most accurate results as certain assumptions will be made in any model. Additionally, the finite element model used in this study provided a reasonable baseline for the material combination in this study. However several improvements can be made to the model. While the model used a non-linear solver due to contact and plastic material, including additional, viscoelastic properties for the seat material could be used. While this will help the accuracy of the results, it comes at the cost of much higher computing power needed to run the model. In conclusion, the finite element model used in this study quantified the results to validate the materials chosen to proceed with cycle testing the valve. References a) Jemaa, M. C. Ben, R. Mnif, K. Fehri, and R. Elleuch. "Design of a New Tribometer for Tribological and Viscoelasticity Studies of PTFE Valve Seats." Tribol Lett Tribology Letters 45.1 (2011): 177-84. b) T. Sofronas, "Case 11: Excessive ball valve seat wear," Hydrocarbon Processing, 2002. c) http://www.ezlok.com/TechnicalInfo/MPBrass.html