Finite element analysis of residual stresses and distortion in hard faced gate valve

Transcription

1 Journal of PUNITHARANI Scientific & Industrial et al: ANALYSIS Research OF RESIDUAL STRESSES AND DISTORTION IN HARD FACED GATE VALVE Vol. 69, February 010, pp Finite element analysis of residual stresses and distortion in hard faced gate valve K Punitharani 1 *, N Murugan and S M Sivagami 1 Mechanical Engineering Department, PSG College of Technology, Coimbatore , India Mechanical Engineering Department, Coimbatore Institute of Technology, Coimbatore , India Received 0 July 009; revised 07 December 009; accepted 07 December 009 This paper studies establishing temperature distribution, distortion and residual stress field developed during plasma arc hard facing of Stellite 6 over low alloy steel (AISI 4140) gate valve by finite element analysis (FEA). In FEA, a three dimensional solid model was generated and simulated using ANSYS parametric design language (APDL) code. After simulation, residual stresses were compared and validated by stress measured by X-Ray diffraction technique. Numerical results showed good agreement with experimental test. Keywords: Distortion, Low alloy steel, Material property, Plasma transferred arc (PTA) welding, Residual stress, Stellite-6 Introduction Structural, material and welding parameters 1 strongly affect residual stresses (RSs) and distortion development in a welded component. Hardfacing (HF) is a process of depositing a filler material on surface of carbon and low alloy steel base metal. In petroleum industrial applications, gate valve made of low alloy steel substrate (AISI 4140) is used extensively to withstand high temperature and heavy load conditions. Among welding processes available for HF applications, plasma transferred arc (PTA) welding 3 has been increasingly employed due to higher deposition rate, minimum oxidation, relatively lower dilution, improved metallur gical bond between substrate and wide applicability of material 3,4. Also, PTA HF process produces high quality overlays at moderate deposition rates and at lower production costs, compared to other traditional welding processes. Stellite 6 (Co-Cr-A) is commonly used as a HF material to be deposited on gate valve surface due to its superior wear-resistance, corrosion resistance and heat resistance properties under high pressure and high temperature 5,6. During HF deposition by PTA process, material part close to weld is subject to different rates of expansion and contraction causing development of *Author for correspondence punitharani@rediffmail.com three-dimensional complex RSs 4,8. In particular, during depositing Stellite-6 over low alloy steel substrate, material properties change drastically, especially at melting point. Therefore, due to temperature dependent material properties and large deformation, material and geometrical non-linearity have to be taken into account 8. Initial expansion of material due to temperature increase has been reported to constrain by material placed away from heat source, generating compressive stress at the centre of weld pool 9. Plastic strains resulting from heating causes stress, which in turn produces internal forces that may cause buckling, bending and rotation, called distortion 4. Finite element analysis (FEA) is used in calculating welding RSs 10,11. Bonifaz 1 analyzed butt weld using FEA, while considering effects of changes with temperature in modulus of elasticity, yield stress and coefficient of linear thermal expansion. Ravichandran et al 13,14 developed a novel means of measuring axisymmetric, three-dimensional RSs, and also formulated a basic theory using FEA. Present work predicts RSs in a hardfaced gate valve using FEA, validated with measured values of RSs by X ray diffraction technique. Experimental A low alloy steel (AISI 4140) gate valve (thickness, 50.8 mm; groove, 3 mm) (Fig. 1) was preheated to 450 C for 1 h before depositing Stellite 6

2 130 J SCI IND RES VOL 69 FEBRUARY 010 Fig.1 Gate valve before hardfacing Fig. Gate valve after circular bead deposition alloy in order to avoid moisture, cold crack, entrapped hydrogen and to minimize induced RS. In a preheated gate valve, a circular bead (width, 14 mm; height, 5 mm) was deposited using HF technique by PTA process with 10 overlap simultaneously on both sides (Fig. ). Chemical composition of AISI 4140 and Stellite 6, respectively, are as follows (by wt): C, 0.37, 1.08; Si, 0.15, 1.09; Mn, 0.60, 1.00; Cr, 0.80, 8.75; Mo, 0.15, -; Ni, 0.50,.5; P, 0.04, -; S, 0.04, -; V, 0.10, -; W, -, 4.37; and Fe, -,.50%. Mechanical properties of AISI 4140 and Stellite 6, respectively, are as follows: tensile strength, 1075, 758 MPa; yield stress, 117, 655 MPa; density, 7750, 8380 kg/m 3 ; and Poisson ratio, 0.9, 0.5. Process parameters used during HF were as follows: welding current, 170 amp; voltage, 35.5 volts; welding speed, 150 mm/min; powder feed rate, 30 g/min; plasma gas flow rate, 3.5 lpm; oscillation frequency, 39 cycles/min; pilot arc current, 60 amp; shielding gas flow rate, 18 lpm; powder gas flow rate, 4 lpm; nozzle diam, 10 mm; nozzle to plate distance, 9 mm; and circle overlap, 10. After HF, gate valve was kept in an electric furnace, maintained at a 630 C for h, in order to release induced RSs during HF. Analysis of Residual Stresses (RSs) RSs can be introduced into engineering components during manufacture as a result of forming, bending and welding 1. RSs can also be caused by forces and thermal gradients imposed during surfacing. RSs can affect load-carrying capacity and resistance to fracture components. In order to quantify RS effect, it is necessary to know their magnitude and distribution 14 Fig.3 Residual stress measurement points. Among techniques available for measuring RS, X-ray diffraction technique is capable of measuring 3D stress 10. RSs in gate valve were estimated using X-Ray Diffraction technique and coupled FEA using ANSYS X- Ray Diffraction Technique After stress relieving HF gate valve, oxide coatings were removed using different grades of emery sheet and electrochemical polishing was applied on locations (Fig. 3) marked for RS measurements. RS was measured using X-ray stress analyzer (Rigaku strainflex MSF-M) having a back reflection type goniometer with a θ scan (range, C). Strain in surface layer was estimated by measuring peak shift due to deformation of crystal lattice using CuKα radiation. Finally, these strains were converted into stresses by assuming linear elastic distortion of crystal lattice.

3 PUNITHARANI et al: ANALYSIS OF RESIDUAL STRESSES AND DISTORTION IN HARD FACED GATE VALVE 131 Fig. 4 Meshed model of finite element Coupled Field Finite Element Analysis Finite element modelling and analysis using ANSYS 10.0 were performed to predict RSs and distortion within HF gate valve component. Model Description A cylindrical block of substrate (inner radius, 1 mm; outer radius, 44.5 mm; and thickness, 50.8 mm) was used to build 3D solid model (Fig. 4). HF material was modeled (thickness, 5 mm) above and below base metal. HF process simulated is a single pass PTA HF. FEA employed technique of element birth and death. All elements have to be created, including those that are born in later stages of analyses. In proposed method, elements were deactivated by multiplying their stiffnesses by a large reduction factor. Cylindrical block absorbs a part of the heat generated. Mesh Generation Using 8-noded brick element, solid model was meshed. Totally nodes and elements were generated in present model. Accuracy of FEA depends upon density of mesh used for analysis. Temperature around arc is generally higher than melting point of material, and it drops sharply in regions away from weld pool. Therefore, for correct temperature field in the region of high temperature gradients, it was necessary to have a more refined mesh close to weld line, while in regions located away from weld-line, a coarse mesh was generated. Sensitivity analysis, carried out in mesh generation, showed that when mesh (element) size was changed, temperature distribution and thermal stresses did not vary much. In this method, mesh size used in stress analysis was identical to that in thermal analysis (Fig. 4). Boundary Conditions and Loading Governing differential equation for heat conduction in a solid is given by δ δt k δx x δx δ δt δ δt δt + k + k + Q = ρc δy y δy δz z δz δt...(1) where k x, k y, k z are thermal conductivities in x, y and z directions, respectively. Materials were considered non-linear, therefore, parameters of k x, k y, k z, ρ, C were functions of temperatures. General solution was obtained by substituting initial and boundary conditions. Initial conditions were defined as T(x, y, z, 0) =T 0 (x, y, z) () All free surfaces convection is applied and calculated by Vinokurov s empirical relationship, which includes effect of both convection and radiation as

4 13 J SCI IND RES VOL 69 FEBRUARY 010 Table 1 Material property of AISI 4140 and Stellite 6 Temp. K Thermal conductivity Specific heat Thermal expansion Modulus of W/mK J/kgK coefficient, K -1 elasticity GPa AISI 4140 Stellite 6 AISI 4140 Stellite 6 AISI 4140 Stellite e e e e e e e e e e-6 09 h =.41 X ε T (3) where h is convective heat transfer coefficient (W/m K), ε is emmisivity and T is temperature (K). Surface heat flux on top surface was calculated from Pavelic s circular disc model of heat source with Gaussian distribution as q r = (4) π R 3 3Q S ( x, y) e R where, Q S is irradiated heat flux on top surface, r is instantaneous radius [distance from beam axis, R, is heat source effective radius (R=r o ), r o is average keyhole radius]. Q S = η V I (5) Where η is arc efficiency, %; V is arc voltage, Volts; and I is weld current, Amps. For heat distribution along depth direction, conical heat source model represented by the equation with Gaussian distribution has been considered Q(x, y, z) = Q πr b 0 e r 1 r0 1 z H (6) where Q b is absorbed plasma power (70%), r o is initial radius (at top of keyhole), H is depth, r instantaneous radius and z is instantaneous depth. Thermal analysis was performed by using temperature dependent thermal material properties for substrate and HF layer considered; therefore, non-linear equations were solved, with all complexities related to their solutions. Among several methods of solving the system of simultaneous equations available in ANSYS program, sparse direct solver (SDS), a default solver, is used to solve equations in analysis. SDS is based on a direct elimination method, as opposed to iterative solvers, where solution is obtained through an iterative process that successively refines an initial guess to a solution that is within an acceptable tolerance of exact solution. Under thermal and structural material properties used for both base metal and HF alloy (Table 1), temperature changes encountered in heat-affected-zones are so large that change of thermal properties could not be neglected. To determine temperatures and other thermal quantities, there is a need to perform a transient thermal analysis. Therefore, implicit method of time discretisation, which allows for larger time steps, was employed. Here, time step size is actually not a problem for calculation stability but it determines accuracy of solution. Convection losses were modeled at all free surfaces of gate valve component. In thermal analysis, heat was input in 370 load steps, including 10 overlap on top side and same procedure was repeated for bottom side also. As number of load steps is fairly very high, programmic language called ANSYS Parametric Design Language, coding was employed to perform both thermal and structural analysis easily. Nodal temperature solutions obtained from thermal analysis were read as loading into stress analysis. In order to capture RSs induced due to heating and cooling cycle, temperature history needed to be read at a sufficiently large number

5 PUNITHARANI et al: ANALYSIS OF RESIDUAL STRESSES AND DISTORTION IN HARD FACED GATE VALVE 133 Residual stress, MPa stress measurement point Fig.5 Comparison of residual stress measured by XRD technique and ANSYS Table Comparison of numerical results with experimental results Measurement Residual stress Error points MPa % Measured Predicted Table 3 Statistical error analysis of experimental results with numerical results Parameter Residual stress, MPa Experimental Numerical Mean Variance Observations 5 5 Hypothesized mean difference 0 t stat t-critical one tailed of time points, especially where temperature gradient is large. However, greater the number of thermal solution steps used, greater is computational time and larger the store space required. FEA was carried out in two steps. i) A non-linear transient thermal analysis was conducted first to obtain global temperature history generated during HF process; and ii) General purpose FEA ANSYS 10.0 was used for both thermal and stress analysis performed sequentially. Structural Analysis Solid 45 elements (8 nodes and 3 degrees of freedom per node) were used for static structural analysis. With available global temperature history from thermal analysis, and with proper constraints, structural analysis was carried out to predict RSs induced and thereby distortion in HF gate valve. Results and Discussion Residual Stress Measurement by XRD Technique From measured longitudinal RSs on surface of HF gate valve using XRD technique (Fig. 5), it is found that tensile stress reaches a maximum value of 3 MPa at a distance of mm from the beginning of HF metal and then reduces. It becomes compressive RS as distance increases. Maximum compressive RS (400 MPa) is found at a distance of 1 mm from HF metal. Residual Stresses Analysis by FEA Looking into longitudinal RS distribution on HF gate valve using coupled field FEA (Fig. 5), it is clear that same trend is observed between two methods with small variations due to assumptions made in FEA. On base metal side, maximum tensile residual stress (8 MPa) is found at a distance of mm and

6 134 J SCI IND RES VOL 69 FEBRUARY 010 compressive stress (340 MPa) is found at a distance of 1 mm and comparing this with allowable limits (117 MPa) of yield stress of base metal, it is much low, due to stress relieving of gate valve after HF. It is observed that, tensile stresses occur at HF surface and below that compressive stresses were noticed. Maximum stress could be noticed at sharp edges and wherever cross section varies. A comparison of RSs obtained by both methods (Fig. 5), shows a maximum deviation of 19.50%. By using student s t-test in MS Excel solver, a comparison is made for numerical result with experimental result (Fig. 5 and Tables & 3) and it was observed that t-stat is less than t-critical one tailed. So, a good agreement of experiment result with numerical result, indicates simulated model is reliable in predicting RS. Errors can be minimized by proper selection of precision machine for measuring RS if all material properties at different temperatures are known for Stellite-6. Conclusions FEA was used to evaluate RSs in 50.8 mm OFE gate valve. Predicted RSs were compared and validated with X ray diffraction technique. A generalized plane strain assumption yielded reasonable stress distributions in both longitudinal and transverse direction, as was verified by XRD experimental data. Temperature near circular bead and HAZ decreased rapidly with distance from center of heat source. A very large tensile residual stress occurs on the surface of circular bead, and a compressive stress appears below circular bead. Maximum distortion (0.467 mm) was obtained from FEA. Maximum deviation between measured and predicted RSs is found to be 19.50%, due to assumptions made in FEA. Acknowledgements Authors thank M/s BHEL, Trichy for providing necessary support for successful experiment conduct. Authors also thank managements of PSG College of Technology and Coimbatore Institute of Technology, Coimbatore for providing necessary facilities in completion of the work. References 1 Mahapatra M M, Datta G L & Pradhan B, Three-dimensional finite element analysis to predict the effects of shielded metal arc welding process parameters on temperature distributions and weldment zones in butt and one-sided fillet welds, IMechE, 0 (006) Yang Q X, Yao M & Park J K, Numerical simulations and measurements of temperature and stress field in medium-high carbon steel specimen after hard-face-welding, Comput Mater Sci, 9 (004) Garg A K, Plasma Arc Welding and Surfacing Process (Welding Research Institute, Department of Science and Technology, SERC School on Advances in Welding Technology) 00, Michaleris P & DeBiccari, Prediction of welding distortion, Weld J, 76 (1997) Davis J R, Metals Handbook, 10 th edn (ASM International) 1998, Wu A P, Ren J L, Peng Z S, Murakawa H & Ueda Y, Numerical simulation for the residual stresses of Stellite hardfacing on carbon steel, J Mater Process Technol, 101 (000) Crook P, ASM Handbook vol : Nonferrous Alloys and Special Purpose Materials, 10 th edn (ASM International) 1998, Canas J, Picon R, Paris F & Marin J, Experimental and numerical analysis of residual stress in welded AL-5083-O Aluminum plates, w:weld Int, 8 (1994) Maeder G, Leburn J L & Sprauel J M, Present possibilities for the X-ray diffraction method of stress measurements, NDT Int, 14 (1981) Prasad S & Narayanan S, Finite element analysis of temperature distribution during arc welding using adaptive grid technique, Weld J, 75 (1996) Murugan S, Kumar P V, Gill T P S, Raj B & Bose M S C, 1999, Numerical modeling and experimental determination of temperature distribution during manual metal arc welding, J Sci Technol Weld & Joining, 4 (1999) Bonifaz E A, Finite element analysis of heat flow in single-pass arc welds, Weld Res Suppl, 79 (000) Ravichandran G, Raghupathy P, Ganesan N & Krishnakumar R, Prediction of axis shift distortion during circumferential welding of thin pipes using the finite element method, Weld J, 76 (1997) Peng-Hsiang Chang & Tso-Liang Teng, Numerical and experimental investigations on the residual stresses of the butt-welded joints, Comput Mater Sci, 9 (004) Rajamanickam N, Balusamy V, Thyla P R & Hari vignesh G, Numerical simulation of thermal history and residual stresses in friction stir welding of AL 014-T6, J Sci Ind Res, 68 (009)