Effects of Analysis Method, Bolt Pre-Stress, and Cover Plate Thickness on the Behavior of Bolted Flanges of Different Sizes by Christopher Michael Wowk An Engineering Project Submitted to the Graduate Faculty of Rensselaer Polytechnic Institute in Partial Fulfillment of the Requirements for the degree of MASTER OF ENGINEERING Major Subject: MECHANICAL ENGINEERING Approved: _________________________________________ Norberto Lemcoff, Project Adviser Rensselaer Polytechnic Institute Hartford, Connecticut May 2015 CONTENTS LIST OF TABLES ............................................................................................................................... iii LIST OF FIGURES ............................................................................................................................. iv ACKNOWLEDGMENT....................................................................................................................... v ABSTRACT....................................................................................................................................... vi NOMENCLATURE........................................................................................................................... vii 1. Introduction.............................................................................................................................. 1 1.1 Background .................................................................................................................... 1 1.2 Problem Description ...................................................................................................... 2 2. Theory/Methodology ............................................................................................................... 4 2.1 2.2 2.3 Theory ............................................................................................................................ 4 2.1.1 Joint Separation Behavior ................................................................................. 4 2.1.2 Stress in the Center of the Cover ...................................................................... 8 Methodology.................................................................................................................. 9 2.2.1 Solid Element Finite Element Model ................................................................ 9 2.2.2 Shell and Beam Element Finite Element Model ............................................. 13 Evaluation vs Analysis Type ......................................................................................... 14 3. Results and Discussion ........................................................................................................... 17 3.1 3.2 3.3 Joint Separation Behavior ............................................................................................ 17 3.1.1 Joint Separation Behavior Due to Differing Bolt Pre-Stress and Analysis Type................................................................................................... 17 3.1.2 Maximum Joint Separation Due to Cover Plate Thickness ............................. 21 3.1.3 Maximum Joint Separation Due to Normalized Bolt Pre-Stress and Nominal Pipe Size............................................................................................ 22 Location of Contact Outside Bolt Circle ....................................................................... 23 3.2.1 Location of Contact Due to Differing Bolt Pre-Stress...................................... 23 3.2.2 Normalized Location of Contact Due to Nominal Pipe Size ............................ 24 Stress in the Center of the Cover Plate ........................................................................ 25 i 3.3.1 Stress in the Center of the Cover Plate Due to Differing Bolt Preload ........... 26 3.3.2 Stress in the Center of the Cover Plate Due to Cover Plate Thickness ........... 28 4. Conclusions............................................................................................................................. 30 5. References .............................................................................................................................. 32 Appendix A ...................................................................................................................................... 1 Appendix B ...................................................................................................................................... 1 ii LIST OF TABLES Table 1 – Flange Component Geometric Parameters..................................................................... 3 Table 2 - Material Summary............................................................................................................ 3 Table 3 – Analyses Performed ...................................................................................................... 16 iii LIST OF FIGURES Figure 1 – Typical Class 3, Category 1 Appendix Y Joint (Half Section) ........................................... 2 Figure 2 – Free Body Diagram of Class 3, Category 1 Flange (Waters and Schneider 1969) .......... 5 Figure 3 – Force and Moment Diagram for Annular Ring Portion of Cover Plate & Flange ........... 6 Figure 4 – Meshed Solid Element ABAQUS Model of Class 3, Category 1 Flange Pair ................. 10 Figure 5 – Solid Element Model Boundary Conditions and Loads ................................................ 11 Figure 6 – COPEN Extraction Path................................................................................................. 12 Figure 7 – Symbol Plot of CNORMF for 80% Yield Bolt Pre-Stress Case ....................................... 12 Figure 8 – Meshed Shell/Beam Element ABAQUS Model of Class 3, Category 1 Flange Pair ...... 14 Figure 9 – Maximum Joint Separation at Sealing Element vs. Bolt Pre-Stress ............................. 18 Figure 10 – Joint Separation vs. Distance from the Outside Diameter of Flange – Radial Beam Analysis Method ......................................................................................................... 20 Figure 11 – Joint Separation vs. Distance from the Outside Diameter of Flange – Solid Element Finite Element Model ................................................................................... 20 Figure 12 – Joint Separation vs. Distance from the Outside Diameter of Flange – Shell/Beam Element Finite Element Model ................................................................................... 21 Figure 13 – Maximum Joint Separation at Sealing Element vs. Bolt Pre-Stress (Different Cover Plate Thicknesses)....................................................................................................... 22 Figure 14 – Maximum Joint Separation at Sealing Element vs. Bolt Pre-Stress (4 and 16 NPS Sizes) ........................................................................................................................... 23 Figure 15 – Normalized Contact Distance from Bolt Circle vs Bolt Pre-Stress.............................. 24 Figure 16 – Normalized Contact Distance from Bolt Circle vs Bolt Pre-Stress (16 NPS Flange Pair)............................................................................................................................. 25 Figure 17 – Radial Stress (S11) in Cover Plate vs Bolt Pre-Stress .................................................. 27 Figure 18 – Tangential Stress (S22) in Cover Plate vs Bolt Pre-Stress........................................... 27 Figure 19 – Radial Stress (S11) in Cover Plate vs Cover Plate Thickness ...................................... 29 Figure 20 – Tangential Stress (S22) in Cover Plate vs Cover Plate Thickness ............................... 29 iv ACKNOWLEDGMENT I would like to thank my project advisor, Professor Norberto Lemcoff, for his support and understanding during the process of completing this project. His advice and patience were invaluable to this project. I would also like to thank the many professors I have had over the course of my education at Rensselaer Polytechnic Institute. The knowledge and insight in the field of mechanical engineering they have transferred to me during my time here were also invaluable to the completion of this project. Finally, I would like to thank my wife, Lara, and my parents for all of their love and support, without which completion of this project and degree program would not have been possible. v ABSTRACT Standards shapes and sizes exist for flanges with contact outside the bolting circle; however, if a custom design is needed, it is the responsibility of the designer to ensure a leak-free joint is maintained at the design pressure, and that the components of the assembly do not fail in service. Guidelines for sizing the flange geometry and bolting requirements are provided in Non-Mandatory Appendix Y of the American Society of Mechanical Engineers Boiler and Pressure Vessel Code, however the Appendix Y method does not specify a method for determining the separation of the flange components when subjected to internal pressure loading, an important parameter when selecting a sealing element. This paper investigates the influence of bolt preload, cover plate thickness, nominal pipe size, and analysis method on the separation behavior of the flanged joint, the predicted location of contact outside the bolt circle, and stress in the cover plate for Class 3, Category 1 Appendix Y flange pairs. The radial beam theory method that was used to develop the Appendix Y method, finite element analysis using solid elements, and finite element analysis using shell and beam elements were used to evaluate the flanged joint. This investigation identified similarities between the results of the different analysis methods and also potential shortcomings and limitations for the analysis methods. vi NOMENCLATURE a Width Of The Outermost Region Of The Radial Beam b Distance From Bolt Circle To Location Of Contact bmax Distance From Bolt Circle To Outside Diameter Of Flange B1 Diameter of the Sealing Element C Width of the Innermost Region of the Radial Beam CNORMF Normal Force Due To Contact Acting On Each Node COORD1 Radial Location of Each Node COPEN Distance of Each Node to Contact Surface f Bolt Hole Flexibility Constant f’ Bolt Hole Flexibility Constant f” Bolt Hole Flexibility Constant F Axial Force Acting On Cover Plate/Flange Due To Pressure I Moment Of Inertia β Distance from the Bolt Circle to the Seal L Length of the Flange from Point of Contact to Seal M Bending Moment Acting On Flange/Cover Plate M1 Total Bending Moment Acting On Cover Plate MSII Total Bending Moment Acting On Cover Plate Nbolts Number of Bolts in Bolt Circle P Internal Pressure of the Joint Rm Radius of the Sealing Element S11 Radial Stress Predicted By ABAQUS S22 Tangential Stress Predicted By ABAQUS SRIIBC Radial Stress in the Center of the Cover Plate Predicted By BPVC vii SRIIWS Radial Stress in the Center of the Cover Plate Predicted By RBT STIIBC Tangential Stress in the Center of the Cover Plate Predicted By BPVC STIIWS Radial Stress in the Center of the Cover Plate Predicted By RBT t Thickness of the Cover Plate or Flange tII Thickness of the Cover Plate v(x) Separation of the Joint as a Function of x x Distance from Outside Diameter Greek Letters θ Rotation of the Radial Beam θi Rotation of the Radial Beam at Outside Diameter θreq_sol Required Angular Section for Solid Element FEA θreq_shell Required Angular Section for Shell/Beam Element FEA ν Poisson’s ratio Subscripts A Referring to the outermost portion of the radial beam B Referring to the innermost portion of the radial beam viii ACRONYMS ASME American Society of Mechanical Engineers BPVC Boiler and Pressure Vessel Code FEA Finite Element Analysis RBT Radial Beam Theory ix 1. Introduction 1.1 Background Bolted pressure vessel flanges are typical for fluid power applications when disassembly of the joint is required for maintenance or access to the internals of the system. Bolted flanges can be broken into two general categories: flanges with no contact beyond the bolting circle, and flanges with contact beyond the bolting circle. Flanges without contact beyond the bolting circle are stressed during bolt up. Flanges with contact between the mating flanges outside the bolting circle are not stressed until pressurization. The behavior of these flanges are dependent on the prestress carried in the bolts and the interaction between the flanges in contact. This project will focus on the behavior of the latter type of flanges. Standards shapes and sizes exist for flanges with contact outside the bolting circle; however, if a custom design is needed, it is the responsibility of the designer to ensure a leak-free joint is maintained at the design pressure, and that the components of the assembly do not fail in service. Guidelines for sizing the flange geometry and bolting requirements are provided in Non-Mandatory Appendix Y of the American Society of Mechanical Engineers (ASME) Boiler and Pressure Vessel Code (BPVC) (ASME 2013). The basis of the analysis method utilized by Appendix Y involves considering each flange as a collection of discrete, radial beams whose behavior can be described through beam theory (Schneider 1968). The analytical procedures presented in Appendix Y cover many different configurations of flanged joints with contact outside the bolt circle. Past work has been performed on the agreement between the behavior of symmetric flange pairs predicted by the radial beam theory, classical plate theory, and finite element analysis (Galai and Bouzid 2010). 1 1.2 Problem Description This project will focus on the behavior of a typical flat faced flange with a flat cover plate, as predicted by three different analysis methods: a) radial beam theory (RBT), b) finite element analysis (FEA) using solid elements, and c) finite element analysis using shell elements for the components of the flange and beam elements for the bolts, for differing cover plate thicknesses, bolt preload, and flange sizes. In Appendix Y nomenclature, this joint configuration is designated as a Class 3, Category 1 Appendix Y flange pair. The flange pair is made up of a cover plate, flange/hub, and fastener hardware. A typical joint of this type is shown in Figure 1. Figure 1 – Typical Class 3, Category 1 Appendix Y Joint (Half Section) Joint separation behavior is an important characteristic in determining the leak behavior of the joint, as it quantifies separation of the flanges at the location of the sealing element. In addition to the investigation into the behavior of the joint under pressure loading, additional analysis will be performed to characterize the stress predictions for different analysis methods in the center of the cover plate. 2 The geometry of the flanged joints evaluated by this project are based on standard sizes given in ASME B16.5, Pipe Flanges and Flanged Fittings for Class 1500 flanges, and are shown in Table 1. The ASME B16.5 class designation dictates the operating pressure rating for the flange components, with Class 1500 being rated for a 1,500 psi operating pressure. Materials allowed by ASME B16.5 were used for all joint components. The cover and flange/hub are made from ASTM A515 Grade 60 carbon steel, and the bolts are made of ASTM A193 Grade B7 stainless steel. Material properties are shown in Table 2. Table 1 – Flange Component Geometric Parameters NPS Size 4 16 Outside Diameter of Flange and Cover (in) 12.25 32.50 Flange and Cover Plate Thickness (in) 2.12 5.75 Bolt Circle Diameter (in) 9.50 27.75 Bolt Hole Diameter (in) 1.38 2.63 8 16 Nominal Bolt Diameter (in) 1.25 2.50 Bolt Tensile Diameter (in) 1.11 2.26 Bolt Tensile Area (in) 0.969 4.00 Pipe Bore (in) 4.60 16.19 Hub Outside Diameter (in) 6.38 21.75 Flange/Hub Length (in) 3.56 10.25 Center of Seal Groove Diameter (in) 5.49 18.97 Number of Bolts Table 2 - Material Summary Flange Component Material Specification Young’s Modulus Yield Strength Poisson’s Ratio Cover Plate & Flange/Hub ASTM A515 Grade 60 30,000 ksi 32 ksi 0.3 Bolts ASTM A193 Grade B7 30,000 ksi 105 ksi 0.3 3 2. Theory/Methodology 2.1 Theory This project will investigate the joint separation behavior and cover plate stresses for Class 3, Category 1 Appendix Y flanges with varying cover plate thickness, bolt preload, and nominal pipe size. Bolted joint behavior will be determined using the RBT developed by Waters and Schneider (1969), equations given in Appendix Y, finite element models consisting of solid elements, and finite element models consisting of more computationally efficient shell and beam elements. 2.1.1 Joint Separation Behavior Appendix Y of the BPVC provides a method for sizing Class 3, Category 1 flange pairs. Included in this method are equations to predict the rotation of the cover plate and flange at the location of the sealing element, stresses in the flange components, and the pre-stress and operating stress of the bolts to ensure contact outside the bolt circle at a location specified by the designer. Although useful in confirming a design will satisfy the stress requirements of the BPVC, the equations in Appendix Y do not readily support determination of the joint separation behavior, or allow for analysis of cases of low bolt pre-stress where the relative rotations of the cover and flange at the outside diameter are non-zero. In order to evaluate cases with low bolt pre-stress, additional investigation and manipulation of the method used to create the Appendix Y method is required. This method was first proposed by Schneider (1968), and considers the bending behavior of the flange components as a series of radial beams. Schneider (1968) initially applied this method to identical flange pairs where symmetry could be exploited to simplify the analysis, but later modified the identical flange pair method for use in analysis with non-identical flange pairs (Waters and Schneider 1969). Waters and Schneider’s (1969) method for evaluating non-identical flange pairs is presented below. 4 A free body diagram of a Class 3, Category 1 Appendix Y flange configuration is shown in Figure 2. Fluid pressure acts on the area of the cover and flange within the diameter of the seal, which is assumed to be located at the mid-thickness of the hub, Rm. The internal pressure not only acts to separate the joint between the cover and flange, but also creates a large overturning moment which creates a counter-clockwise rotation of the outer portion of the flange. Figure 2 – Free Body Diagram of Class 3, Category 1 Flange (Waters and Schneider 1969) Contact is assumed to occur as a uniform line load outside the bolt circle between the flange and cover, at a distance b, to react the prying loads acting on the bolts. The distance between the bolt circle and the location of contact, b, is determined by equating the moments acting on the flange and cover, respectively, about the location of contact, R. The true location of R is dependent on the preload of the bolts and the separation of the flanges. In the method presented by Waters and Schneider (1969), it is more computationally efficient to select a location of R and then solve for the required bolt preload to create contact at the selected R, than to specify the preload and solve for R. 5 For non-identical flanges, Waters and Schneider’s method assumes that the two sets of moments act on the flanges. A system of balanced forces and moments acts equally on the flange and cover. This system of forces and moments produces the same flange separation behavior as the identical flange pair method (Schneider 1968). The remaining unbalanced system of forces and moments causes rigid body rotation of the flanges as a pair, and does not act to further separate the joint. The joint must be considered as four separate pieces as shown in Figure 2. The outer rings of the cover plate and flange must be analyzed as systems of discrete, radial beams in order to use beam theory to determine displacements. The free body and moment diagram for the annular portion of the cover plate and flange considered as a beam is shown in Figure 3. Figure 3 – Force and Moment Diagram for Annular Ring Portion of Cover Plate & Flange 6 The moment along the length of the radial beam can be expressed as: ππ΄ = π" π+π"πΉβ π₯ π ππ΅ = (ππ + π ′ πΉβ) − π ′ πΉ(π₯ − π) 0≤x<b (1) b≤x≤L (2) Note that in the above equations, factors f, f’, and f” are included to account for the increased flexibility allowed by the bolt holes. The curvature of the beam can be determined by integrating the expression M/EI. To account for the decrease in width of the beam as x increases (moves towards the center), the beam will be considered as made up of two regions, the outer region with width a, and the inner with width c. To maintain continuity in the circumferential direction, the term (1-ν2) is applied to the moment of inertia equations. ππ‘ 3 πΌπ΄ = 12∗(1−π2 ) 0≤x<b (3) b≤x≤L (4) π ππ΄ ππ₯ πΈπΌπ΄ 0≤x<b (5) πΏ ππ΅ ππ₯ πΈπΌπ΅ b≤x≤L (6) 0≤π₯<π π≤π₯≤πΏ (7) ππ‘ 3 πΌπ΅ = 12∗(1−π2 ) The curvature of the beam can be determined as: ππ΄ (π₯) = ∫0 ππ΅ (π₯) = ∫π π(π₯) = { ππ΄ (π₯) ππ΅ (π₯) πππ with boundary conditions θA(0) = θi and θA(b)= θB(b). When b<bmax, θi is equal to zero, since the flange and cover are in contact over the length from b to bmax. When contact occurs at the outside diameter, b=bmax, a non-zero rotation can occur at the point of contact. The deflection of the beam can then be determined as: π π£π΄ (π₯) = ∫0 ππ΄ (π₯) ππ₯ 0≤x<b 7 (8) πΏ π£π΅ (π₯) = ∫π ππ΅ (π₯) ππ₯ π£ (π₯) π£(π₯) = { π΄ π£π΅ (π₯) πππ b≤x≤L (9) 0≤π₯<π π≤π₯≤πΏ (10) with boundary conditions vA(0) = 0 and vA(b)= vB(b). The expression for v(x), too cumbersome to be presented in this report, is dependent on the reaction forces and moments that must be determined by solving the system of equations for the entire flange assembly. The equations necessary to determine the forces and moments were presented by Waters and Schneider (1969), and are shown in the Appendix A Maple worksheet. To solve for the forces, a value of b and/or θi must be assumed in order to calculate the resulting bolt pre-stress and joint separation. Trial and error was used to determine the values of b and θi for the bolt preloads investigated in this project. The joint separation between the cover plate and flange was then calculated based on the resulting values of b and θi. In addition to the values of b determined by RBT, for higher bolt prestresses, equations in Appendix Y were used to predict the location of contact as function of bolt pre-stress (ASME 2013). 2.1.2 Stress in the Center of the Cover Stress in the center of the cover was determined using equations presented in Appendix Y of the BPVC (ASME 2103) and by Waters and Schneider (1969). Appendix Y of the BPVC predicts the radial and tangential stress at the center of the cover plate to be equal. The stresses can be calculated using equation 38 of Appendix Y: ππ πΌπΌπ΅πΆ = πππΌπΌπ΅πΆ = 0.3094∗π∗π΅12 2 π‘πΌπΌ 6∗πππΌπΌ 2 1 ∗π‘πΌπΌ − π∗π΅ (11) where P is the operating pressure, B1 is the diameter of the sealing element, tII is the thickness of the cover, and MSII is the total flange moment for the cover plate at the sealing element diameter, which can be calculated using the equations in Appendix Y of BPVC. 8 Waters and Schneider (1969) offer an equation to predict the radial and tangential stress at the center of the cover also. As in Appendix Y, both the radial and tangential stresses at the center of the cover plate are predicted to be equal. The magnitude of the stresses are given by equation 48 of Waters and Schneider (1969): 3 π∗π π2 ∗(3+π£) − 8 ππ πΌπΌπππ = πππΌπΌππ = π‘ 2 ∗ [ πΌπΌ 2 ∗ π1 ] (12) where P is the operating pressure, Rm is the radius of the sealing element, tII is the thickness of the cover plate, ν is the Poisson’s ratio of the cover material, and M1 is the total flange moment for the cover plate at the location of the sealing element, which can be calculated as described in Section 2.1.1. 2.2 Methodology Maple worksheets, provided in Appendix A, were created to perform the RBT analyses. Microsoft Excel spreadsheets, provided in Appendix B, were developed to perform the Appendix Y calculations. In addition to the analytical modeling performed by the Maple worksheets and Microsoft Excel spreadsheets, joint behavior was investigated using ABAQUS finite element models. Models using solid elements and more computationally efficient shell and beam elements were created and used for evaluating the behavior of the bolted joints. 2.2.1 Solid Element Finite Element Model Cylindrical symmetry of the flanged joint was exploited to reduce the number of elements needed to properly characterize the contact between the cover plate, flange, and bolts. An angular segment of the flange pair from the center of one bolt to the midpoint of the cover/flange between the initial bolt hole and subsequent bolt hole is all that is required to fully define the behavior of the joint. The angular segment is dependent on the number of bolts in the bolt circle and is determined by: 9 180 ππππ−π ππππ = π (13) ππππ‘π So, for a flange with eight bolts in the bolt pattern, a 22.5° segment is required, and for a flange with 16 bolts in the bolt pattern, an 11.25° segment is required. Solid models of the joint segment were created using ABAQUS CAE with the dimensions shown in Table 1. The parts were partitioned to force nodes of the contact surfaces between the bolts, cover plate, and flange to align to allow for improved contact recognition. All parts were meshed using C3D8 elements (8 noded linear brick elements). The meshed assembly is shown in Figure 4. Figure 4 – Meshed Solid Element ABAQUS Model of Class 3, Category 1 Flange Pair To account for the effects of symmetry, a cylindrical coordinate system was created and symmetrical boundary conditions were applied to the appropriate faces of the parts. Axial displacement of the assembly was restrained by fixing the axial displacement of the free face of the hub to zero. Contact was defined throughout the model with the use of the “General Contact” interaction. The analysis was performed in two steps. In the first step, the pre-stress was applied to the bolt. This was done using a “Bolt Preload” load applied to the middle of the bolt. Since only half of the 10 bolt appeared in the model, only half of the preload was required to be applied in order to generate the required bolt pre-stress. In the second step, a pressure load of 1,500 psi was applied to the cover from the center up to the location of the hypothetical sealing element at Rm, the bore of flange/hub, and also from the face of the flange up to the location of the hypothetical sealing element. Boundary conditions and loads applied to the assembly are shown in Figure 5. Figure 5 – Solid Element Model Boundary Conditions and Loads To determine the joint separation behavior of the solid element model, the Contact Opening (COPEN) field output was extracted from the model after the pressure load was applied along a path at the edge of the flange between bolt circles. This field output tracks the distance between two faces in close proximity as part of ABAQUS’s contact algorithm. The extraction path for COPEN is shown in Figure 6. 11 Figure 6 – COPEN Extraction Path Since the solid element FEA model allowed for contact between the flange and cover at multiple radii outside the bolt circle, the location of contact predicted by the solid element FEA model was approximated by calculating the total moment acting on the inside face of the cover due to contact with the flange, and dividing it by the total contact force acting on the inside face of the cover. To accomplish this, the Contact Normal (CNORMF) and radial node location (COORD1) field outputs were extracted for every node on the inside surface of the cover plate. The location of contact and distance from the bolt circle could then be calculated as shown in Equations (14) through (17). Figure 7 – Symbol Plot of CNORMF for 80% Yield Bolt Pre-Stress Case 12 πππ‘ππ ππππππ‘ = ∑ππ=0 πΆπππ π·1π ∗ πΆπππ ππΉπ (14) πππ‘ππ πΉππππ = ∑ππ=0 πΆπππ ππΉπ (15) πΏππππ‘πππ ππ πΆπππ‘πππ‘ = π= π·π΅πππ‘πΆπππππ 2 πππ‘ππ ππππππ‘ πππ‘ππ πΉππππ (16) − πΏππππ‘πππ ππ πΆπππ‘πππ‘ (17) Radial and tangential stresses in the center of the cover were directly extracted from the FEA model. The analysis results were transformed into cylindrical coordinates and the radial and tangential stress were extracted as S11 and S22, respectively. 2.2.2 Shell and Beam Element Finite Element Model Cylindrical symmetry of the flanged joint was once again exploited to reduce the number of elements needed to properly characterize the contact between the cover plate, flange, and bolts. An angular segment of the flange pair from subsequent midpoints between bolt holes was required to fully define the behavior of the joint, since it is not feasible to model half the bolt with a beam element. The angular segment required was determined by the number of bolts in the bolt circle and was determined by: 360 ππππ−π βπππ = π (18) ππππ‘π So for a flange with eight bolts in the bolt pattern, a 45° segment is required, and for a flange with 16 bolts in the bolt pattern, a 22.5° segment is required. Shell models of the flange segment were created using ABAQUS CAE with the dimensions shown in Table 1. The bolt was represented in the analysis as a B33 element (2 node cubic beam in space) with a circular cross section equal to the bolt’s stress area. Each end of the beam was attached to the cover and flange, with a kinematic coupling constraint to represent the bolt head and nut. The influence radius for the kinematic couplings were set equal to the diameter of the head of the 13 bolt. The parts were partitioned to force nodes of the contact surfaces between the cover plate and flange to align to allow for improved contact recognition. The cover plate and flange were meshed using S4 elements (4 noded doubly curved general-purpose shell, finite membrane strain elements). Shell offsets were set so that the inner faces of the cover plate and flange/hub had zero separation. The meshed assembly is shown in Figure 8. Figure 8 – Meshed Shell/Beam Element ABAQUS Model of Class 3, Category 1 Flange Pair Boundary conditions, load steps, and contact interactions were applied to the shell model in the same manner as they were applied to the solid element model with the following exception: the full bolt preload was applied since the full cross-section of the bolt was included in the analysis. The joint separation behavior, radial location of contact between the cover plate and flange and stress in the center of the cover plate were extracted from the shell/beam model in the same manner as they were extracted from the solid element model, as described in Section 2.2.1. 2.3 Evaluation vs Analysis Type In order to reduce the number of analyses required to compare the effects of the multiple independent variables investigated by this project, not all combinations of nominal pipe size, bolt 14 preload, and cover plate thickness were evaluated with each analysis method. Instead, the influences of each independent variable were evaluated using some of the analysis methods presented in Sections 2.1 and 2.2 in order to allow for general statements regarding their effect to be made. The analysis type performed for each perturbation of independent variable is shown in Table 3. 15 Table 3 – Analyses Performed 4" NPS Size Analysis Method Cover Thickness Bolt Pre-Stress Joint Separation Along Length t1 0 Pressure Load 0.5*t1 25% Yield 80% Yield 0 Pressure Load Radial Beam Theory Solid Element FEA Shell Element FEA Radial Beam Theory Radial Location of Contact Appendix Y Equations Solid Element FEA Shell Element FEA Radial Beam Theory Stress In Center of Cover Appendix Y Equations Solid Element FEA Shell Element FEA Maximum Joint Separation 16" Radial Beam Theory Solid Element FEA Shell Element FEA 16 25% Yield 2*t1 80% Yield 0 Pressure Load t2 25% Yield 80% Yield 0 Pressure Load 25% Yield 80% Yield 3. Results and Discussion 3.1 Joint Separation Behavior The separation of the joint between the cover plate and flange with differing analysis types, bolt preload, cover plate thickness, and nominal pipe size, under a 1,500 psi internal pressure load, was investigated by this project. The separation of the joint from the outside diameter of the flanges inward to the location of the sealing element was evaluated for a standard sized 4 NPS Class 3, Category 1 Appendix Y flange pair with differing bolt preloads using different analysis methods. The maximum separation of the joint at the sealing element was also determined for the same joint with different cover plate thicknesses using RBT, solid element finite element models, and shell/beam element finite element models. RBT and solid element finite element models were also used to compare the joint separation of standard sized 4 NPS and 16 NPS Class 3, Category 1 Appendix Y flange pairs with differing preloads. 3.1.1 Joint Separation Behavior Due to Differing Bolt Pre-Stress and Analysis Type The maximum joint separation at the sealing element as a function of the pre-stress in the bolts is shown in Figure 9. It can be seen that the three analysis methods show a similar relationship, the joint separation decreases as the bolt pre-stress increases. This relationship is important, as larger separation of the cover plate and flanges indicates a potential for the joint to leak in service, because the sealing element can only expand a finite amount. Typically, bolts in pressure containing applications are torqued to a pre-stress equal to at least the load they will see in service, in order to reduce cyclic stresses caused by pressurization and depressurization cycles. The results of this analysis indicate that a preload greater than the operating load may be required for Class 3, Category 1 Appendix Y flange pairs in order to minimize joint separation at the sealing element. The magnitude of the preload is dependent on the capabilities of the sealing element, however 17 the results of this project indicate that a preload at least 6x the operating load can significantly reduce joint separation at the seal. Additionally, at higher pre-stresses for each analysis method, the total joint separation appears to approach a non-zero limit. It is suspected that these limits are the result of flexure of the cover plate and flange/hub. This non-zero lower limit at the sealing element indicates that the capabilities of the sealing element must be matched to the expected joint separation, even when high bolt pre-stresses are utilized in order to ensure a leak-free joint. Figure 9 – Maximum Joint Separation at Sealing Element vs. Bolt Pre-Stress The total joint separation as a function of the distance from the outside diameter of the flange using RBT is shown in Figure 10, using solid element FEA models in Figure 11, and using shell/beam element FEA models in Figure 12. The general magnitudes of separation from RBT and solid element FEA are in good agreement for lower bolt pre-stresses. The shell/beam element FEA models predicts much smaller joint separations for all preloads. Upon further evaluation of the deflection curves, the separation along the length of the joint appears to be linear for the RBT and solid 18 element FEA results, whereas the separation appears to be exponential for the shell/beam element FEA results. This suggests that the elongation of the bolts has a larger influence in the joint separation when RBT and solid element FEA are used, and the flexibility of the cover plate and flange has a larger influence when the shell/beam element FEA analysis is used. This difference in results is suspected to be caused by the artificial stiffening of the bolts caused by the kinematic coupling constraint used to model the bolt heads in the shell/beam element FEA models. For the RBT analysis, there seems to be a limit where there is much smaller variation in the behavior of the joint at higher bolt pre-stresses. This appears to occur between the 25% and 80% Yield curves. This limit is not seen when the joint is analyzed using either FEA methods. Both FEA methods predict complete separation of the cover plate and flange at the outside diameter when the bolts are not preloaded. This complete separation was not observed in the RBT analyses. It must be noted that complete separation of the flanges is not supported by the RBT analysis method, as contact is assumed to occur at least at the outside diameter. This limitation in the RBT analysis indicates a potential shortcoming in the RBT analysis at lower bolt pre-stresses. 19 Figure 10 – Joint Separation vs. Distance from the Outside Diameter of Flange – Radial Beam Analysis Method Figure 11 – Joint Separation vs. Distance from the Outside Diameter of Flange – Solid Element Finite Element Model 20 Figure 12 – Joint Separation vs. Distance from the Outside Diameter of Flange – Shell/Beam Element Finite Element Model 3.1.2 Maximum Joint Separation Due to Cover Plate Thickness The maximum joint separation at the sealing element as a function of the pre-stress in the bolts and cover plate thickness is shown in Figure 13. The maximum joint separation was calculated using RBT and solid element FEA models. It can be seen that both analysis methods show similar behavior; joint separation decreases as bolt pre-stress and cover plate thickness increase. For the thinner cover plate models, the results of the RBT and solid element FEA diverge as bolt preload increases. For the standard size and double thickness models, both analysis methods showed good agreement, however smaller joint separations were predicted at higher preloads by the solid element FEA models. The results from this analysis indicate that joint separation can be decreased by increasing the bolt pre-stress or cover plate thickness. 21 Figure 13 – Maximum Joint Separation at Sealing Element vs. Bolt Pre-Stress (Different Cover Plate Thicknesses) 3.1.3 Maximum Joint Separation Due to Normalized Bolt Pre-Stress and Nominal Pipe Size The maximum joint separation at the sealing element as a function of the pre-stress in the bolts for standard sized 4 and 16 NPS flange pairs is shown in Figure 14. The maximum joint separation for the 16 NPS flange pair was calculated using RBT and solid element FEA models, whereas the results of the RBT analysis are only shown for the 4 NPS flange pair. The RBT and solid element FEA results are in good agreement for maximum separation of the flanges at the sealing element. It can be seen that greater joint separation is predicted for the larger size flange pair when the bolt pre-stress is low. As bolt pre-stress increases, the joint separation of the larger flange pair converges to the predicted separation for the smaller flange pair. 22 Figure 14 – Maximum Joint Separation at Sealing Element vs. Bolt Pre-Stress (4 and 16 NPS Sizes) 3.2 Location of Contact Outside Bolt Circle The location of contact outside the bolt circle under a 1,500 psi internal pressure with differing analysis types, bolt pre-stress, and nominal pipe sizes was investigated by this project. The location of contact was calculated using RBT, Appendix Y equations, solid element FEA models, and shell/beam FEA models for a 4 NPS Class 3, Category 1 Appendix Y flange pair with varying bolt pre-stress. The radial location of contact for a 16 NPS flange pair was also calculated using RBT, Appendix Y equations, and solid element FEA models, and the results were normalized and compared to the results for the 4 NPS flange pair. 3.2.1 Location of Contact Due to Differing Bolt Pre-Stress The normalized contact distance from the bolt circle due to bolt pre-stress predicted by the different analysis methods is shown in Figure 15. All of the analysis methods are in good agreement at bolt pre-stresses greater than 20% of the bolt yield strength. Agreement between the analysis 23 methods is expected because the location of contact is the result of a force and moment balance about the bolt circle. However, slight variations do exist between the analytical and FEA methods at lower preloads. These variations are caused by the lack of capability of the analytical methods to predict complete separation of the joint at low bolt pre-stress. For the RBT results, contact is assumed to occur at the outside diameter of the joint, bmax, until the bolt pre-stress is great enough to force the slope of the cover plate and flange/hub to be zero at the outside diameter of the joint. A similar assumption is carried in the Appendix Y equations, where the minimum bolt pre-stress that is able to be calculated corresponds to contact at the outside diameter and a slope between the flanges of zero. Figure 15 – Normalized Contact Distance from Bolt Circle vs Bolt Pre-Stress 3.2.2 Normalized Location of Contact Due to Nominal Pipe Size The normalized contact distance from the bolt circle due to bolt pre-stress predicted by the different analysis methods for the 16 NPS flange pair is shown in Figure 16. The contact distance was calculated using RBT, Appendix Y equations, and solid element FEA models. All three analysis 24 methods show the expected relationship: the distance of contact from the bolt circle decreases as bolt pre-stress increases. There appears to be more variation in the contact location results between analyses for the 16 NPS flange pair than for the 4 NPS flange pair. The cause of the observed variation is unclear, as the location of contact is the result of a force and moment balance around the bolt circle. Figure 16 – Normalized Contact Distance from Bolt Circle vs Bolt Pre-Stress (16 NPS Flange Pair) 3.3 Stress in the Center of the Cover Plate The radial and tangential stresses in the center of the cover plate, under the internal pressure load, were investigated using differing analysis types, bolt pre-stresses, and cover plate thicknesses. The stresses were calculated using RBT, Appendix Y equations, solid element FEA models, and shell/beam FEA models for a 4 NPS Class 3, Category 1 Appendix Y flange pair with varying bolt pre-stress. The effect of cover plate thickness on these stresses was also investigated. 25 3.3.1 Stress in the Center of the Cover Plate Due to Differing Bolt Preload Radial (S11) and tangential (S22) stress in the center of the cover plate for a standard sized 4 NPS Class 3, Category 1 Appendix Y flange pair as a function of bolt pre-stress are shown in Figure 17 and Figure 18, respectively. The calculated stresses vary greatly between the four analyses types for both radial and tangential stress. RBT calculated high values for stress that appear to approach a limit as the bolt pre-stress increases. Appendix Y results calculated a fairly constant stress over the entire bolt pre-stress range studied in the present analysis. Both FEA methods calculated stresses that decreased as the pre-stress approached 25% of the bolt yield strength, but increased as the bolt pre-stress was increased further. Additional mesh refinement work, that is outside the purview of this project, is required in order to determine if the cover plate stress increase with increasing bolt pre-stress is due to element distortion near the center of the cover plate. Although it is unclear what is the cause for the differences in stress results for the four analysis types, it is suspected that the shell/beam element FEA model yields incorrect stress results because of the large differences between them and the results of the other analysis types, and the previously mentioned increased rigidity the kinematic coupling representing the bolt head adds to the model. 26 Figure 17 – Radial Stress (S11) in Cover Plate vs Bolt Pre-Stress Figure 18 – Tangential Stress (S22) in Cover Plate vs Bolt Pre-Stress 27 3.3.2 Stress in the Center of the Cover Plate Due to Cover Plate Thickness Radial (S11) and tangential (S22) stress in the center of the cover plate for a 4 NPS Class 3, Category 1 Appendix Y flange pair with different cover plate thicknesses are shown in Figure 19 and Figure 20, respectively. The thickness of the cover plate is normalized by the standard thickness specified in ASME B16.5. The stresses were calculated using RBT, Appendix Y equations, and solid element FEA models. Due to the large stress variations noted in Section 3.3.1 regarding the shell/beam element FEA results, the shell/beam FEA analysis method was not included when investigating the effect of cover plate thickness on stress. Figure 19 and Figure 20 show that the stress in the cover plate decreases as cover plate thickness increases. This relationship is expected. Since the pressure area is the same regardless of cover plate thickness, the forces and moments acting on the cover plate are expected to be independent of cover plate thickness. Although the forces and moments are independent of thickness, the stress in a plate is inversely proportional to its thickness squared. For the thinner cover plate, there were differences in stresses calculated by each analysis method. As the cover plate thickness increases, the results from the different analysis methods converge. The results of the different analysis methods were shown to be in good agreement for the standard and double-thickness cover plates. 28 Figure 19 – Radial Stress (S11) in Cover Plate vs Cover Plate Thickness Figure 20 – Tangential Stress (S22) in Cover Plate vs Cover Plate Thickness 29 4. Conclusions The joint separation behavior and cover plate stresses for Class 3, Category 1 Appendix Y flanges with varying cover plate thickness, bolt preload, and nominal pipe size were investigated in this project. Bolted joint behavior was determined using the RBT analysis method developed by Waters and Schneider (1969), equations given in Appendix Y of ASME BPVC, finite element models consisting of solid elements, and finite element models consisting of more computationally efficient shell and beam elements. The following conclusions were able to be drawn from the analyses performed by this project: Differences exist in the joint behaviors predicted by RBT, solid element FEA models, and shell/beam FEA models. Shell/beam FEA models predicted joint separations much less than the other analysis types. The suspected cause of this difference is the increased rigidity added to the model by the kinematic coupling constraint that represented the interaction between the heads of the fastener and cover plate and flange. The magnitude of joint separation predicted by RBT and solid element FEA models were generally in good agreement, however it was determined that RBT lacks the capability to predict complete separation of the cover plate and flange faces at low bolt pre-stresses. Additionally, a limit was found to exist for the RBT analysis, where joint separation does not decrease any longer as the bolt pre-stress increases. The thickness of the cover plate and the overall size of the flange pair was also found to have an influence on joint separation. Joint separation decreased as thickness of the cover plate increased. For the 16 NPS flange pair, predicted joint separation for lower bolt pre-stresses were greater than those predicted for the 4 NPS flange pair, however the separations appeared to converge as the bolt pre-stress was increased. 30 Predictions for the radial location of contact between the cover plate and flange for the four analysis types were found to be in good agreement for the 4 NPS flange pair. Comparison of the normalized location of contact between flange pair sizes were found not to show any direct relationship. There appeared to be differences in predicted radial and tangential stresses for the four different analysis types, however all showed that the stresses decrease as cover plate thickness increases. The shell/beam element FEA results were generally much lower than the results of the other analysis methods. It is possible that the method of modeling bolt head contact artificially stiffened the joint, leading to lower cover plate deflections and stresses. Additionally, further mesh refinement analysis must be performed to determine whether the increase in stress between bolt prestresses of 25% and 80% yield strength of the bolt were caused by excessive element distortion at the center of the cover plate for the solid element FEA model. 31 5. References ASME Boiler and Pressure Vessel Code, Section VIII – Rules for Construction of Pressure Vessels, 2013 Edition, Appendix Y. Galai, H and Bouzid, A.H, “Analytical Modeling of Flat Face Flanges with Metal-to-Metal Contact Beyond the Bolt Circle,” Journal of Pressure Vessel Technology, Vol 132, 2010. Schneider, R.W, “Flat Face Flanges with Metal-to-Metal Contact Beyond the Bolt Circle,” Transactions of the ASME, Vol 90, 1968, pgs 82-88. Waters, E.O and Schneider, R.W, “Axisymmetric, Nonidentical, Flat Face Flanges with Metalto-Metal Contact Beyond the Bolt Circle,” Journal of Engineering for Industry, Vol 91, 1969, pgs 615-622. 32 Appendix A Sample MAPLE Worksheet to Determine Joint Behavior using Radial Beam Theory Included in this appendix is the MAPLE worksheet used to determine the flange behavior using RBT for the standard sized 4 NPS flange with no bolt pre-stress. The behavior of different flange sizes, cover plate thicknesses, and bolt preload can be determined by changing the necessary values in the “Case Constants” and “Calculations” sections. A1 A2 A3 A4 A5 A6 Appendix B Sample EXCEL Worksheet to Determine Bolt Pre-Stress with Varying Locations of Contact Using Appendix Y Formulas Included in this appendix is the Excel worksheet used to determine the required bolt pre-stress to engage contact at differing distances using Appendix Y equations for the standard sized 4 NPS flange. The behavior of different flange sizes and cover plate thicknesses can be determined by changing the necessary values in the “Flange Properties” section. Material Properties E 30000000 nu 0.3 psi P Flange Properties t1 1500 psi 2.12 in t2 Bolt Hole Diameter Flange Outside Diameter 2.12 in 1.38 in 12.25 in Pipe Bore 4.6 in Diameter of Seal Bolt Circle Diameter Pipe Outside Diameter 5.49 in 9.5 in 6.38 in g0 0.89 g1 0.89 re1 Bolt Properties Tensile Diameter Stress Area Number of Bolts Bolt Length 1 1.11 in 0.969 in^2 8 5.35 in B1 Calculated Values Estar1 Estar2 X AR rB a β F V Fprime H HD HT h0 hc or "b" JS JP MP C1 C2 C3 C4 θrb1 θrb2 Ms1 Ms2 Mu1 Mu2 Mb1 Mb2 θB1 θB2 HC Wm1 Sigb Si % YS Fbi SR2_c ST2_c 285843840 285843840 0.5 0.369909594 0.039464491 1.980874317 1.365209472 0.9082 0.550103 5.685849602 35507.96878 24928.53771 10579.43107 2.023363536 0.103 0.668475969 0.400964295 73547.40081 -0.204855533 -7474.244217 -0.12541143 -12579.73298 -1.52528E-05 1.52528E-05 -12032.94804 -8367.399809 -1,833 1,833 -10,200 -10,200 0.00012 0.00015 615,022 650,530 83,918 83,916 80% 81,315 3,760 3,760 hD hT 0.25 0.681993 0.414482 73547.4 -0.202947 -8089.674 -0.122282 -12929.27 -1.45E-05 1.45E-05 -12421.71 -8932.051 -1,745 1,745 -10,677 -10,677 0.00013 0.00016 251,482 286,990 37,021 37,013 35% 35,866 3,804 3,804 B2 0.36 0.692108 0.424596 73547.4 -0.201543 -8542.753 -0.119964 -13188.21 -1.4E-05 1.4E-05 -12708.72 -9348.314 -1,680 1,680 -11,029 -11,029 0.00013 0.00016 173,664 209,172 26,983 26,966 26% 26,130 3,836 3,836 2.005 2.2275 1 0.750959 0.483447 73547.4 -0.19374 -11059.3 -0.10686 -14652 -1.1E-05 1.1E-05 -14315.61 -11669.12 -1,323 1,323 -12,992 -12,992 0.00015 0.00017 60,555 96,063 12,392 12,265 12% 11,885 4,016 4,016 1.375 0.7854415 0.5179298 73547.401 -0.189443 -12445.25 -0.099474 -15477.14 -9.39E-06 9.389E-06 -15210.16 -12953.7 -1,128 1,128 -14,082 -14,082 0.00015 0.00017 43,248 78,756 10,159 9,924 9% 9,617 4,115 4,115