An Investigation into the use of FEA methods For the Prediction of Thermal Stress Ratcheting by Stephen Charles Huse A 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: _________________________________________ Ernesto Gutierrez-Miravete, Engineering Project Adviser Rensselaer Polytechnic Institute Hartford, Connecticut December, 2014 i CONTENTS LIST OF TABLES ............................................................................................................ iv LIST OF FIGURES ........................................................................................................... v LIST OF SYMBOLS ........................................................................................................ vi ACKNOWLEDGMENT.................................................................................................. vii KEYWORDS .................................................................................................................. viii ABSTRACT ...................................................................................................................... ix 1. Introduction .................................................................................................................. 1 2. Historical Review......................................................................................................... 2 2.1 Bree Diagram ..................................................................................................... 2 2.2 Linear Thermal Discontinuity ............................................................................ 4 3. Theory .......................................................................................................................... 6 3.1 Discussion .......................................................................................................... 6 3.2 Conduction in a Hollow Cylinder ...................................................................... 7 3.3 Forced Convection Inside a Hollow Cylinder .................................................... 8 3.4 ASME Requirements.......................................................................................... 9 3.5 3.4.1 Thermal Ratcheting ASME Code Requirements ................................... 9 3.4.2 Linear Regression Calculation ............................................................... 9 Numerical FEA Methods ................................................................................. 10 4. Method of Procedure.................................................................................................. 12 4.1 Discussion ........................................................................................................ 12 4.2 Thermal Analysis ABAQUS File ..................................................................... 12 4.3 4.2.1 Node Section ........................................................................................ 12 4.2.2 Elements Section .................................................................................. 13 4.2.3 Analysis Information Section ............................................................... 13 Stress Analysis ABAQUS File ......................................................................... 16 ii 4.3.1 4.4 Analysis Information Section ............................................................... 16 ABAQUS Analysis Inputs ............................................................................... 18 4.4.1 Boundary Conditions............................................................................ 18 5. Results ........................................................................................................................ 25 5.1 Calculation of Convective Heat Transfer Coefficient ...................................... 25 5.2 Thermal Analysis Results................................................................................. 28 5.3 Stress Analysis Results..................................................................................... 32 6. Discussion and Conclusions ...................................................................................... 37 7. References .................................................................................................................. 39 Appendix A, Program Files ............................................................................................. 40 iii LIST OF TABLES Table 1: Pipe Size Dimensions from Table A-6 of [8] ................................................................. 18 Table 2: Material Properties for Alloy N06600 from [1] ............................................................. 21 Table 3: Water Properties from Table A-3 of [8] ......................................................................... 22 Table 4: Thermal Transient Temperature vs Time ....................................................................... 23 Table 5: Tabular Calculation of h, Hot Flow ................................................................................ 25 Table 6: Tabular Calculation of h, Cold Flow .............................................................................. 26 Table 7: Calculation of Minimum ∆T1 ......................................................................................... 29 Table 8: Calculation of Maximum ∆T1......................................................................................... 30 Table 9: Program Files .................................................................................................................. 40 iv LIST OF FIGURES Figure 1: Bree’s Shakedown Diagram [3], [4]................................................................................ 3 Figure 2: Illustration of Temperature Gradients from Figure NB-3653.2(b)-1 of [1] .................... 4 Figure 3: Stress vs time from page 2 of [6] .................................................................................. 10 Figure 4: Valve Nozzle Model ...................................................................................................... 20 Figure 5: T vs time for one cycle .................................................................................................. 23 Figure 6: T vs time for 20 cycles .................................................................................................. 24 Figure 7: h vs T for 500 gpm Hot Flow ........................................................................................ 26 Figure 8: h vs T for 500 gpm Cold Flow ...................................................................................... 27 Figure 9: Thermal Analysis Line .................................................................................................. 28 Figure 10: ∆T1 vs time .................................................................................................................. 31 Figure 11: Hoop Stress vs Hoop Strain for 1000, 2000, and 3000 psi ......................................... 33 Figure 12: Hoop Stress vs Displacement for 1000, 2000, and 3000 psi ....................................... 34 Figure 13: Plastic Hoop Strain vs time ......................................................................................... 35 Figure 14: Plastic Hoop Strain vs time, 1000 to 2000 psi in 100 psi Increments ......................... 36 Figure 15: Pressure for Onset of Thermal Ratcheting .................................................................. 37 v LIST OF SYMBOLS Symbol Description Units A Surface area in2 α Mean coefficient of thermal expansion in/in/°F cp Specific heat BTU/lb Distance from flow entry region in diameter lengths diameters D Mean diameter in di Inner diameter in Do Outer diameter in ∆T1 Linear through-wall temperature gradient °F ∆T2 Surface temperature gradient °F E Young’s Modulus psi h Convective heat transfer coefficient BTU/in2/s/°F k Thermal conductivity BTU/in/s/°F Nusselt number none Pressure psi Prandtl number none Radius in Reynold’s number none ρ Density lb/in3 T Temperature °F t Time s tw Wall thickness in σp Primary stress psi σt Thermal secondary stress psi σy Yield strength psi Kinematic viscosity ft2/s Poisson’s ratio none D/L Nu P Pr r Re vi ACKNOWLEDGMENT I would like to thank my wife, Sarah, for being supportive and helpful during the long hours spent on this project. Thanks also to my fellow workers at Electric Boat for guidance and thanks to Ernesto for being a great advisor. vii KEYWORDS ABAQUS Convection Elastic Plastic Fatigue FEA Heat Transfer Piping Ratcheting Valve viii ABSTRACT The prediction of the onset of thermal ratcheting is a necessary step in the design of nuclear piping and pressure vessels. The failure mechanism of thermal ratcheting occurs due to severe pressure and thermal stresses and is a low-cycle failure mode. This report documents a numerical FEA method for predicting the onset of ratcheting with the use of ABAQUS [2] and compares the results to the current analytical methods used in the ASME commercial code [1]. FEA calculation methodology in the computer program ABAQUS [2] was applied to 3” NPS Schedule 80 piping connected to typical valve nozzle geometry. The thermal ratcheting analysis involved the creation of two ABAQUS analyses, a heat transfer analysis and a structural elasticplastic analysis which imports the heat transfer analysis output. ix 1. Introduction Nuclear power plants, in particular, are susceptible to high thermal ratcheting strains due to rapid increases and decreases in the temperature of the water flowing through the piping and pressure vessels. When cold water from outside of the plant quickly flows through hot piping, the inside of the pipe thermally contracts while the outside circumference remains hot, causing a through wall temperature gradient resulting in tensile stress on the inside of the pipe. After the piping cools down, hot water from inside the plant can quickly flow back through the same piping resulting in the inside of the pipe thermally expanding while the outside remains cold creating a compressive thermal stress on the inside of the pipe. Related to the local through-wall temperature gradient is the gross thermal expansion and contraction of the piping arrangement due to changes in the mean temperature of the piping. Constrained expansion results in secondary moments which bend the piping and create stress. The arrangement of the pipes and support structure greatly influences this expansion moment. For this report, however, the effects of mean thermal expansion of the piping system are not included in the secondary stress. The secondary stress from through-wall temperature gradients will be focused on as the linear temperature gradient is the largest factor in thermal ratcheting as seen in the ASME requirements [1]. The previously discussed loads combined with large primary stresses due to high pressures result in plastic strain and thermal ratcheting. This report documents a method for predicting the onset of thermal ratcheting by the use of the FEA software, ABAQUS [2]. 1 2. Historical Review Thermal ratcheting failure was popularized by the work of J. Bree [3], [4]. In his articles, he proposed what is now known as the Bree diagram or shakedown diagram, as shown in Figure 1. The Bree diagram was created from analyses of thin walled tubing in nuclear fuel applications where thermal gradient stresses can be very high. The diagram predicted the stress combinations necessary for plastic strains to accumulate in piping and pressure vessels. 2.1 Bree Diagram Bree analyzed a condition in which pressure builds up in nuclear fuel cans due to off gassing of fission materials. Combined with the pressure was a thermal gradient that was present when the reactor was operating, but not present when the reactor was cold. This cyclic thermal load causes yielding of the material, maintaining stress at the yield strength [3]. When the plant cools down, the residual stress may cause further plastic strains. Therefore, both cooldown and heatup can result in plastic deformation that accumulates until fatigue failure occurs. The prevention of this fatigue failure is the purpose for thermal ratcheting requirements in the ASME commercial code. 2 Figure 1: Bree’s Shakedown Diagram [3], [4] Figure 1 is Bree’s shakedown diagram, Figure 3 of [3], for non-work hardening material and constant yield strength σy with respect to temperature. The different regions of Figure 1 are as follows: E is the purely elastic region where no plastic strain occurs, S1 and S2 are the plastic shakedown regions where initially, plastic strain accumulates but then tapers off as the pipe settles into a purely elastic response, P is the plastic stability region where plastic strain will cycle between the maximum and minimum stresses, but will not continue to accumulate to 3 failure, and lastly, R1 and R2 are the ratcheting regions where the combination of primary and secondary stresses are sufficient to result in eventual failure of the structure. The X axis of Figure 1 is equal to the primary stress over the yield strength. For primary stress due to internal pressure in a cylinder, the stress can be calculated with a thin-walled approximation resulting in p PD PD which leads to X _ axis where the material 2t w y 2t w yield strength is taken at the average bulk fluid temperature of the thermal transient. The Y axis of Figure 1 is equal to the maximum secondary stress range due to a linear thermal gradient over the yield strength. The stress resulting from a linear through wall temperature gradient is t ET1 ET1 Y _ axis [1], [3] which leads to . 21 v y 21 v 2.2 Linear Thermal Discontinuity The thermal discontinuity that Bree considered was a linearized temperature gradient through the wall of the piping. Temperature gradients, as illustrated in Figure 2, are the sum of the mean temperature, T, the linearized temperature gradient, V (also known as ∆T1), and the surface temperature gradient, ∆T2. Figure 2: Illustration of Temperature Gradients from Figure NB-3653.2(b)-1 of [1] 4 Changes in mean temperature do not cause local stresses to occur, but do cause thermal expansion moments in a constrained run of piping. The linearized or average temperature difference creates thermal stresses that lead to ratcheting failure. The surface temperature gradient creates surface stresses which results in crack initiation and fatigue crack failure. 5 3. Theory 3.1 Discussion Thermal ratcheting is a low cycle fatigue mechanism that accumulates plastic strain with each stress cycle [5]. Structures such as nuclear piping systems are subjected to the type of low cycle, high stress conditions that result in plastic strain and thermal ratcheting. Current ASME analysis requirements in Section III NB-3653.7 are designed to prevent ratcheting from starting [1]. Primary and secondary stresses are limited such that the structure does not enter the ratcheting regime. Primary stresses are loads such as deadweight and pressure that do not reduce when strain occurs, but will continue until ductile failure occurs. Secondary stresses are loads such as thermal expansion moments and thermal gradient stress that will reduce when strain occurs. In the design of piping systems, it is important to give special attention to locations prone to stress concentrations such as welds or geometry discontinuities [5]. Accurate modeling of accumulated plastic strain due to ratcheting is hindered by many complex and hard to model factors. Material hardening and cyclic stress history are two of the major factors that are difficult to accurately model. Kinematic hardening, the increase in strength after yielding, occurs in many materials and continues as loading increases until the ultimate tensile strength is reached at which point the material experiences ductile failure. An isotropic linear kinematic hardening model will tend to under predict thermal ratcheting accumulated strains while a nonlinear kinematic hardening model will tend to either over predict ratcheting strains or predict elastic shakedown [6]. For this report, an elastic perfectly plastic material model is assumed. The stress history is not always well known and can affect the analysis. The earlier that larger stress cycles are applied the earlier that failure of the material will occur. However, because cyclic history is usually unknown, the worst case loading history is assumed for design analyses. Thermal ratcheting strain is calculated using the current requirements of the ASME Boiler and pressure vessel code [1] Section III, Division 1 – NB-3653.7. As input, the code requires that the 6 linear through-wall gradient of temperature, ∆T1, be known. The following sections will describe the calculation of ∆T1. 3.2 Conduction in a Hollow Cylinder The general heat transfer partial differential equation for a hollow cylinder is 1 T T kr c p r r r t [1] where temperature, T, is time and location dependent and material properties are for the cylinder. For steady-state conditions, the right hand side of Equation [1] goes to zero and simplifies to 1 T kr 0 . Multiplying by r, dividing by k (independent of r for isotropic materials) and r r r integrating gives r T A , where A is the first integration constant. Dividing by r gives r T A , which integrates to T r A ln r B . Boundary conditions are then used to solve for r r A and B. For non steady state conditions, such as when temperature varies with time, the easiest way to solve Equation [1] for ∆T1 is by numerical methods. Also, a common and conservative analysis assumption is that the outside of the pipe is perfectly insulated, having convective heat loss of zero resulting in a slightly higher ∆T1. This simplifying assumption is reasonable based on the heat transfer rate for free convection between metal and air versus the rate for forced convection between water and metal, and the rate of thermal conduction in metals. The result of this comparison is that heat transfer for metal conduction and forced convection is much faster than metal to air heat transfer in free convection. Additionally, much of the hot piping in proximity to manned areas is insulated for safety, further reducing heat loss to the environment. 7 The initial temperature of the pipe and the temperature versus time of the inside radius are needed for solving Equation [1]. The temperature of the inside of the cylinder depends on the energy transferred due to forced convection from the fluid flowing inside of the cylinder. 3.3 Forced Convection Inside a Hollow Cylinder The convective heat transfer coefficient, h, for turbulent flow inside a cylinder is calculated with the Dittus-Boelter equation which is given in Equation (3.2.99) of [7]. Nu 0.023 Re 0.8 Pr n where Nu [2] vdi hd i , Re , n is 0.4 for the fluid cooling the pipe and 0.3 for the fluid heating k the pipe, k is for the fluid, and v in the numerator of the equation for the Reynold’s number is velocity. All properties are at bulk fluid temperature. The qualifications for Equation [2] is that 0.7 ≤ Pr ≤ 160, Re > 10000, and D/L>10. By inspection, the water properties from Table 3 satisfy the requirement for Pr. Re is satisfied based on the problem parameters. D/L is the measure of lengths in diameters from the entry region. It is assumed that the location of analysis is more than 10 diameters from the entry region. Knowing the fluid temperature versus time and fluid flow rate versus time, the convective heat transfer coefficient, h, can be calculated. The convective heat transfer coefficient is then used to calculate the heat transferred through convection to the piping, Q hAT where A is the area of heat transfer and ∆T is the temperature difference between the bulk fluid temperature and the inside surface of the cylinder. Heat transferred by convection is based on the surface area, the difference in temperature between the bulk fluid and inside surface of the pipe, and the convective heat transfer coefficient, h. 8 3.4 ASME Requirements 3.4.1 Thermal Ratcheting ASME Code Requirements ASME Section III, Division 1 – NB-3653.7 [1] requirements for thermal ratcheting is that the range of ∆T1 between any two transients is T1 y ' y 0.7 E C4 [3] where C4 is an equation constant (equal to 1.0 for NiCrFe material), E and α are taken at the ambient temperature of 70 °F, σy is at the average fluid temperature of the transients, and y’=1/X for 0 < X < 0.5 and y’=4*(1-X) for 0.5 < X < 1.0 from ASME NB-3222.5, where X is the Bree diagram axis, X _ axis PD . 2t w y 3.4.2 Linear Regression Calculation ΔT1 at each time increment is found by calculating the linear regression of the temperatures through the wall. Once the linear regression fit is known, ∆T1 is the temperature difference from the inside of the pipe to the outside. The line equation is in the form T A Bx where x is the x coordinate, A is A T B x , B is x x T T x x n B i 1 i i 2 n i 1 i [4] , and a horizontal bar over a variable denotes the average of the variable through the wall. The temperature difference from the inside of the pipe to the outside is then B times the wall thickness or T1 Bt w . 9 3.5 Numerical FEA Methods ABAQUS accepts the convective heat transfer coefficient and bulk fluid temperature as input to calculate the heat transferred between the fluid and the piping. ABAQUS also calculates the temperatures necessary to calculate ∆T1 through the numerical analysis of Equation [1]. To model cyclic thermal cycles, the analysis temperatures are increased and decreased repeatedly. The stress analysis ABAQUS file then imports the varying temperatures at each node and applies a constant pressure. The pressure is applied to the inside of the pipe at the nominal value and at the ends of the pipe due to end effects. The end effect pressure is equal to the nominal pressure times Pend the ratio of cross sectional area of the fluid over the metal. PnomDi2 Pnom Di2 2 . 4 Do2 / 4 Di2 / 4 Do Di2 The constant pressure and varying thermal cycles result in a stress load set similar to Figure 3 where the first curve is primary stress versus time and the second curve is secondary stress versus time. Figure 3: Stress vs time from page 2 of [6] 10 The geometry, material properties, and pressure films for the analysis files were created in the ABAQUS pre-processor software, HYPERMESH. Load conditions are added by direct editing of the .inp file as described in Section 4. The ABAQUS stress analysis can use nonlinear FEA methods for calculating large plastic strains; however this restricts the output of the total strain. 11 4. Method of Procedure 4.1 Discussion This section describes the analysis files and the inputs to the thermal and stress analyses which are provided in the Appendices. The student version of ABAQUS limits the user to 1000 nodes per model. In order to conserve the number of nodes, modeling is done axisymmetrically. ABAQUS axisymmetric analysis, by default, defines the Y axis as the axis of symmetry equating R,Z,θ with X,Y,Z respectively. Bending moments were not calculated for this analysis since a three-dimensional model would be needed, requiring the full version of ABAQUS. The slight disadvantage to three-dimensional modeling is the increased computational times whereas an axisymmetric model may take seconds, a complex three-dimensional model could take minutes or hours to complete. Section 4.2 details the thermal analysis model. Section 4.3 details the changes from the thermal model for the stress analysis. Section 4.4 details the inputs to ABAQUS. 4.2 Thermal Analysis ABAQUS File The ABAQUS file is separated into three main sections which are nodes, elements, and analysis information. The majority of manual editing is done in the analysis information section of the ABAQUS input file. ** is a delimiter in the files that tells ABAQUS to ignore the line, which is useful for commenting or blank space. 4.2.1 Node Section The first section defines node locations. *NODE, NSET=ALL denotes the start of the node section. *NODE tells ABAQUS that the following lines will have a node number then node coordinates based on analysis type. Since the analysis is 2D axisymmetric, two coordinates are given: radial (X) and longitudinal (Y). NSET=ALL creates a set of node numbers. Appending the *NODE card with NSET=ALL places all nodes into the set ALL which is then used for assigning the initial temperature of all the nodes. 12 4.2.2 Elements Section The second section is initiated with the card *ELEMENT, TYPE=DCAX8, ELSET=Pipe. *ELEMENT tells ABAQUS that the following lines will have an element number then nodes defining the element. These are automatically created by HYPERMESH in the correct order. TYPE=DCAX8 defines the element type as D for diffusive heat transfer, C for non-twisting, AX for axisymmetric, and 8 for 8-noded quadratic second order element. ELSET=Pipe creates a set of elements under the name “Pipe”. Appending the *ELEMENT card with ELSET places all elements defined in the card into the set which is then used for assigning material properties to the elements. 4.2.3 Analysis Information Section The third section is where most editing of ABAQUS input files occurs. While it is laborious to manually enter node and element information, the analysis section is much faster to manually edit rather than navigating through a user interface that was designed to run every type of analysis that ABAQUS is capable of. The following is one of the many ways to order and build the analysis section. 4.2.3.1 Material Definitions *MATERIAL, NAME=N06600 tells ABAQUS that the following material property cards apply to the material named N06600. *CONDUCTIVITY, TYPE=ISO tells ABAQUS that the following lines will have thermal conductivity in BTU/s/in/°F then the temperature in °F at which each applies. ISO denotes the property applies equally in all directions. *SPECIFIC HEAT tells ABAQUS that the following lines will have specific heat in BTU/lb then the temperature in °F at which each applies. 13 *DENSITY tells ABAQUS that the following line will have density in lb/in3 at 70 °F. For material property cards with only one line, the property is applied to all temperatures. *ELASTIC, TYPE = ISOTROPIC tells ABAQUS that the following lines contain Young’s modulus in psi then Poisson’s ratio then the temperature in °F at which each applies. ISOTROPIC denotes the property applies equally in all directions. *EXPANSION, ZERO = 70.0, TYPE = ISO tells ABAQUS that the following lines contain the mean coefficient of thermal expansion in in/in/°F then the temperature in °F at which each applies. ZERO defines the ambient temperature at which no thermal expansion occurs. ISO denotes similar properties in all directions. *PLASTIC tells ABAQUS that the following lines will have stress in psi then plastic strain then the temperature in °F at which each applies. A plastic strain of 0.0 denotes the yield strength at which plastic deformation begins. Entering plastic strain of 0.0 at each temperature creates an elastic perfectly plastic material definition. *SOLID SECTION, ELSET=Pipe, MATERIAL=N06600 places the material properties labeled N06600 onto the named set of elements. The line following this card is the attribute line, for which 1.0 is for default attributes. 4.2.3.2 Transient Information *ELSET, ELSET=P2 creates a set of elements from the following lines and labels the set as P2. This is used to define a set of elements that border the inside edge and have the second edge of the element at the inside of the piping. An easy way to find this set of elements is by defining a pressure on the inside of the model in HYPERMESH. *INITIAL CONDITIONS, TYPE=TEMPERATURE tells ABAQUS the initial temperature of the nodes. In the following lines is the node set, ALL, then the initial temperature, 70 °F. 14 *AMPLITUDE, NAME=TEMPAMP1, VALUE=ABSOLUTE tells ABAQUS that the following lines have time in seconds then the temperature in °F, repeating up to 4 times per line. This inputs the temperature versus time curve for use in the calculation of energy transferred in convection. Multiple curves were used to define the full transient in order to minimize run time of the stress analysis. *AMPLITUDE, NAME=FILMAMP1, VALUE=ABSOLUTE is the same card type as for the temperature curves but is instead inputting the convective heat transfer coefficient versus time. Similar to the temperature curve, the film curve is divided into multiple curves. *INCLUDE,INPUT=5cycles.th.inp tells ABAQUS to insert the lines found in the 5cycle.th file. This card is used to reduce the repetition of lines in the main file by running 5 thermal cycles with one line of code. 4.2.3.3 Step Definition in 5cycles.th In order to reduce the repetition of multiple lines in the main ABAQUS thermal analysis file, lines were added in a separate file. After properties and thermal inputs are defined in the main file, the analysis steps are imported from this file. *STEP, INC=5000 initiates a step with up to 5000 discrete analysis increments. The cards between this and the following *END STEP card will define a step of the analysis. Multiple steps are entered to reduce run times of the analysis. *HEAT TRANSFER, DELTMX=20.0 tells ABAQUS that the following line defines the initial time increment, the length of time to run the step for, the minimum time step size, the maximum time step size, and steady state option where 0.0 denotes no steady state analysis. DELTMX defines the maximum difference in temperature allowed between adjacent nodes. The ABAQUS program will use the DELTMX control to automatically increase or decrease the time of each increment. 15 *FILM, AMPLITUDE=TEMPAMP1, FILM AMPLITUDE=FILMAMP1 tells ABAQUS that the following lines apply the time versus temperature and time versus heat transfer coefficient curves to the elements by element set, edge of element, temperature (dummy value since AMPLITUDE=TEMPAMP is appending the card), and film coefficient (dummy value since FILM AMPLITUDE=FILMAMP is appending the card). The lines *NODE FILE, FREQUENCY=1 | NT | *EL FILE | COORD, TEMP | *EL FILE,POSITION=NODES, FREQUENCY=1 | TEMP create a binary data file of temperatures at each time step which are then imported into the stress analysis later. *END STEP defines the completion of the analysis step. The lines from *STEP to *END STEP are then repeated to define the full transient and to create five thermal cycles. 4.3 Stress Analysis ABAQUS File The stress analysis file has the same geometry and material properties as the thermal file, but the analysis information and element type are different. The element type is CAX8 for structural analysis instead of DCAX8. 4.3.1 Analysis Information Section Other than the material property cards, the analysis information section for the stress analysis is different from the thermal analysis section as detailed below. *BOUNDARY tells ABAQUS that the following lines will have a node then degree of freedom (2 is Y) then prescribed displacement where 0.0 is no deflection, essentially anchoring the node in the selected degree of freedom. *EQUATION tells ABAQUS that the following lines will have the number of variables for an equation for node displacements multiplied by a factor and equal to zero, alternating with the next line which inputs the variable information. The variable information is given as the first 16 node, degree of freedom, multiplication factor, second node, degree of freedom, and multiplication factor. This card is used to tell ABAQUS that the nodes on the free end of the pipe can move in the Y direction but must all have the same Y displacements. *AMPLITUDE, NAME=PRESS,VALUE=ABSOLUTE defines the time versus pressure curve in the following lines. This value controls the pressure on the model and is iterated to induce ratcheting. PRESS is defining the pressure on the inside of the pipe. PRESE is defining the longitudinal pressure due to end effects on the pipe. *ELSET, ELSET=P1E creates a set of elements from the following lines and labels the set as P1E. This is used to define a set of elements that border the top edge of the pipe and has the first edge of the element at the end. This set will have the PRESE amplitude pressure applied. *STEP initiates the load set. When INC is not included, the default number of analysis increments allowed is up to 100. *STATIC, DIRECT tells ABAQUS to discretize the stress analysis by the input in the following line which gives the time of each increment and the total time. *TEMPERATURE, FILE=valve.th, BSTEP=1, BINC=1,ESTEP=2,EINC=1 tells ABAQUS to import temperatures from the thermal file from step 1, increment 1 to step 2, increment 1 up to the amount of time requested, 10 seconds. Therefore the next analysis step does not duplicate an analysis time. Modifying the thermal file usually requires modifying this card as well. *DLOAD, AMPLITUDE=PRESS tells ABAQUS that the following lines have the following information: element, edge of element, and dummy value for pressure as the appended amplitude card for PRESS overwrites these values. *DLOAD, AMPLITUDE=PRESE is the same card except that it applies the end pressure effects. 17 The analysis steps are repeated until all thermal analysis steps are used. The use of many time steps allows for the varying of time increments to speed up the run time of the total analysis. 4.4 ABAQUS Analysis Inputs This section presents the information that is entered into the ABAQUS input files. Table 1 details the geometry of the piping which is connected to the valve nozzle. Table 1: Pipe Size Dimensions from Table A-6 of [8] Description Value Units Geometry Outer Diameter, Do Thickness, tw Inner Diameter, di Mean Diameter, D Length 3 NPS, Schedule 80 3.5 0.3 2.9 3.2 10.0 in in in in in 4.4.1 Boundary Conditions The geometry for the valve nozzle is detailed in Figure 4. The valve end is anchored axially while the pipe end is allowed to thermally grow, simulating a flexible piping system. The pipe end is constrained to axially displace equally at all points along the radius, simulating the attaching pipe. If a piping system is arranged as a straight run from anchor to anchor, then the boundary conditions would be modeled as axially constrained at both ends. However, this would produce enormous compressive stress, and so is avoided in practice. Common practice is to introduce flexibility into the arrangement with bends and stress loops in order to allow the piping to thermally grow. 18 Remains parallel to x axis, simulating attaching pipe Line of Symmetry Pipe Length = 10.0” Tangent Length = 0.5” Length = 2.0” Length = 4.0” Restrained Axially 19 Figure 4: Valve Nozzle Model Table 2 details the material properties entered into ABAQUS for the piping and the valve nozzle. The material is assumed to be a Nickel Chromium Iron composition, commonly known as Inconel. The specific material properties taken are for NiCrFe, Alloy N06600 seamless pipe and tube, Spec SB-167 for sizes ≤ 5 inches from Reference [1], Section II, Part D, Material Properties, Tables Y-1, TE-4, TCD, TM-4, and PRD. Conductivity was converted from units of BTU/hr/ft/°F by dividing by (3600*12). Also, specific heat was calculated from the equation cp=k/TD/ρ where TD is thermal diffusivity from Table TCD, and ρ is converted to units of lb/ft3 = 0.3*123=518.4 20 Table 2: Material Properties for Alloy N06600 from [1] Temperature T (°F) 70 100 150 200 250 300 350 400 450 500 550 600 650 700 750 800 850 900 Conductivity k (10-3 BTU/s/in/°F) 0.199 0.201 0.206 0.211 0.215 0.222 0.227 0.234 0.238 0.245 0.250 0.257 0.262 0.269 0.273 0.280 0.287 0.292 Specific Heat cp (BTU/lb) 0.108 0.109 0.111 0.113 0.114 0.116 0.116 0.118 0.118 0.120 0.121 0.122 0.123 0.125 0.126 0.128 0.130 0.131 Density ρ (lb/in.3) Young’s Modulus E (106 psi) Poisson’s Ratio v Mean Coefficient of Thermal Expansion α (10-6 in./in./°F) 0.31 6.8 6.9 7.0 7.1 7.2 7.3 7.4 7.5 7.6 7.6 7.7 7.8 7.9 7.9 8.0 8.0 8.1 8.2 31.0 30.3 29.9 29.4 0.30 29.0 28.6 28.1 27.6 27.1 21 Yield Stress σy (ksi) 30.0 30.0 29.2 28.6 28.0 27.4 26.8 26.2 25.7 25.2 24.7 24.3 23.9 23.5 23.2 22.9 22.6 22.3 Table 3 details the water properties used to calculate the convective heat transfer coefficient that is input into ABAQUS. The results of this calculation are provided in Section 5. Table 3: Water Properties from Table A-3 of [8] Temperature T (°F) 32 40 50 60 70 80 90 100 150 200 250 300 350 400 450 500 550 600 Conductivity Kinetic Viscosity Density Prandtl K (BTU/hr/ft/°F) -5 ρ (lb/ft ) Number 0.319 0.325 0.332 0.34 0.347 0.353 0.359 0.364 0.384 0.394 0.396 0.395 0.391 0.381 0.367 0.349 0.325 0.292 2 v x 10 (ft /s) 1.93 1.67 1.4 1.22 1.06 0.93 0.825 0.74 0.477 0.341 0.269 0.22 0.189 0.17 0.155 0.145 0.139 0.137 3 62.4 62.4 62.4 62.3 62.3 62.2 62.1 62 61.2 60.1 58.8 57.3 55.6 53.6 51.6 49 45.9 42.4 22 13.7 11.6 9.55 8.03 6.82 5.89 5.13 4.52 2.74 1.88 1.45 1.18 1.02 0.927 0.876 0.87 0.93 1.09 Table 4 provides the assumed temperature versus time data used for the thermal transient. This transient is then repeated twenty times in order to calculate if ratcheting is occurring as seen in Figure 6. Figure 5 graphs the information entered in Table 4. Table 4: Thermal Transient Temperature vs Time t (s) 0 5 50 55 100 T (°F) 70 600 600 70 70 T vs time 700 Temperature (°F) 600 500 400 300 T (°F) 200 100 0 -10 0 10 20 30 40 time (s) Figure 5: T vs time for one cycle 23 50 60 T vs time 700 Temperature (°F) 600 500 400 300 T (°F) 200 100 0 0 500 1000 1500 time (s) Figure 6: T vs time for 20 cycles 24 2000 5. Results Section 5 details the results of the thermal and stress analysis as well as the calculation of the convective heat transfer coefficient. 5.1 Calculation of Convective Heat Transfer Coefficient Table 5 and Table 6 provide the calculated values for the convective heat transfer coefficient with an assumed flow rate of 500 gallons per minute, gpm. Flow rate was converted from gpm to in/s using the conversions 231 in3 = 1 gallon, 60 sec = 1 min, and by dividing by the cross-sectional area, πdi2/4=6.605 in2. Table 5 is graphed in Figure 7 and Table 6 is graphed in Figure 8. Table 5: Tabular Calculation of h, Hot Flow T (°F) 70 100 150 200 250 300 350 400 450 500 550 600 Flow (gpm) 500 Flow (in/s) Re 291.44 553700 793137.8 1230444 1721179 2181866 2667827 3105407 3452482 3786593 4047738 4222460 4284102 Pr 6.82 4.52 2.74 1.88 1.45 1.18 1.02 0.927 0.876 0.87 0.93 1.09 25 Nu 1609 1896 2318 2708 3028 3344 3614 3823 4046 4259 4495 4769 h (BTU/in2/s/°F) 0.00446 0.00551 0.0071 0.00852 0.00957 0.01054 0.01128 0.01163 0.01185 0.01187 0.01166 0.01112 Table 6: Tabular Calculation of h, Cold Flow T (°F) Flow (gpm) 600 550 500 450 400 350 300 250 200 150 100 70 500 Flow (in/s) Re 291.44 4284102 4222460 4047738 3786593 3452482 3105407 2667827 2181866 1721179 1230444 793137.8 553700 Pr 1.09 0.93 0.87 0.876 0.927 1.02 1.18 1.45 1.88 2.74 4.52 6.82 Nu h (BTU/in2/s/°F) 4810 4462 4201 3993 3794 3621 3399 3143 2884 2564 2204 1949 0.011212 0.011576 0.011702 0.011698 0.011537 0.011302 0.010718 0.009935 0.009071 0.007858 0.006404 0.005399 It is seen in Figure 7 and Figure 8 that the coefficient is not well represented by only the start and end points; therefore, each data point is entered into ABAQUS for the amplitude card containing the curve of film coefficient versus time. h vs T for 500 gpm Hot Flow 0.012 0.011 h (BTU/in^2/s/°F) 0.01 0.009 0.008 0.007 0.006 h (BTU/in^2/s/°F) 0.005 0.004 0 100 200 300 T (°F) 400 Figure 7: h vs T for 500 gpm Hot Flow 26 500 600 h vs T for 500 gpm Cold Flow 0.012 0.01 0.009 0.008 0.007 0.006 h (BTU/in^2/s/°F) 0.005 0.004 600 500 400 300 200 100 T (°F) Figure 8: h vs T for 500 gpm Cold Flow 27 0 h (BTU/in^2/s/°F) 0.011 5.2 Thermal Analysis Results The results of the thermal analysis file was exported into MS Excel for an analysis slice in order to calculate the linear temperature gradient. The line of analysis that ∆T1 is calculated from, is at the transition from the pipe outer diameter to the 30° slope as shown in Figure 9. Line of ∆T1 analysis Node 145 Figure 9: Thermal Analysis Line 28 Table 7 provides the data taken from the ABAQUS thermal file for the first time of minimum ∆T1 (-383 °F). Node, time, and temperature are from the ABAQUS output and the remaining cells are calculated in accordance with Section 3.4.2. Table 7: Calculation of Minimum ∆T1 time (s) 5.31 Average Sum B*tw node 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 T x (°F) (in) 544.94 499.58 455.87 414.51 376.14 340.82 308.79 280.15 254.77 232.65 213.78 198.01 185.22 175.28 168.17 163.86 162.38 292.64 (Ti-Tm)(xi-xm) 1.45 1.47 1.49 1.51 1.53 1.54 1.56 1.58 1.60 1.62 1.64 1.66 1.68 1.69 1.71 1.73 1.75 1.60 (xi-xm)2 -37.85 -27.16 -18.36 -11.43 -6.26 -2.71 -0.61 0.23 0.00 -1.12 -2.96 -5.32 -8.06 -11.00 -14.00 -16.90 -19.54 0.0225 0.0172 0.0127 0.0088 0.0056 0.0032 0.0014 0.0004 0.0000 0.0004 0.0014 0.0032 0.0056 0.0088 0.0127 0.0172 0.0225 -183.05 0.14 -183.05/0.14*0.3= -382.84 29 Table 8 provides the data taken from the ABAQUS thermal file for the first time of maximum ∆T1 (316 °F). Node, time, and temperature are from the ABAQUS output and the remaining cells are calculated in accordance with Section 3.4.2. Table 8: Calculation of Maximum ∆T1 time (s) 55.35 Average Sum B*tw node 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 T x (°F) (in) 162.38 205.924 246.177 282.868 315.847 345.366 371.56 394.543 414.549 431.814 446.487 458.7 468.542 476.129 481.525 484.768 485.863 380.77 1.45 1.47 1.49 1.51 1.53 1.54 1.56 1.58 1.60 1.62 1.64 1.66 1.68 1.69 1.71 1.73 1.75 1.60 30 (Ti-Tm)(xi-xm) (xi-xm)2 19.54 11.38 5.23 0.92 -1.74 -2.97 -2.96 -1.91 0.00 2.61 5.77 9.34 13.19 17.20 21.25 25.22 28.98 0.0225 0.0172 0.0127 0.0088 0.0056 0.0032 0.0014 0.0004 0.0000 0.0004 0.0014 0.0032 0.0056 0.0088 0.0127 0.0172 0.0225 151.05 151.05/0.14*0.3= 0.14 315.92 Figure 10 plots the result of the linear regression of temperature through the wall of the valve nozzle transition versus time for the first thermal cycle. Negative ∆T1 indicates a decrease in temperature from the inside edge to the outside edge. ∆T1 vs time 400 55.4, 316 300 Temperature (°F) 200 100 0 -100 0 10 20 30 40 50 -200 ∆T1 ( F) -300 5.3, -383 -400 -500 time (s) Figure 10: ∆T1 vs time 31 60 70 5.3 Stress Analysis Results For the assumed thermal transient, the structural analysis was iterated to predict the onset of ratcheting. Ratcheting was analyzed at Node 145, as shown in Figure 9, which is at the outside of the thermal analysis slice shown in Figure 9. Figure 11 plots the hoop stress (psi) versus hoop strain for 1000 psi (blue), 2000 psi (green), and 3000 psi (yellow). Ratcheting is easily seen in the 3000 psi iteration, and is slightly seen in the 2000 psi iteration. In the 1000 psi iteration, it is difficult to judge whether ratcheting is occurring. 32 Figure 11: Hoop Stress vs Hoop Strain for 1000, 2000, and 3000 psi Figure 12 plots the hoop stress (psi) versus displacement (in) for iterations of 1000 psi (blue), 2000 psi (green), and 3000 psi (yellow). All three iterations show shakedown occurring. Ratcheting is easily seen in the 3000 psi iteration. The 1000 and 2000 psi iterations are difficult to judge whether ratcheting is occurring from Figure 12. 33 Figure 12: Hoop Stress vs Displacement for 1000, 2000, and 3000 psi Figure 13 plots the plastic hoop strain versus time for iterations of 1000 psi, 2000 psi, and 3000 psi. This metric easily shows iterations with accumulating plastic strain. From Figure 13 it is seen that the pressure for the onset of thermal ratcheting is between 1000 and 2000 psi. 34 Plastic Strain vs time 0.012 0.01 Strain 0.008 1000 psi 0.006 2000 psi 0.004 3000 psi 0.002 0 0 500 1000 1500 2000 time (s) Figure 13: Plastic Hoop Strain vs time 35 2500 Figure 14 plots the plastic hoop strain versus time in increments of 100 psi between 1000 and 2000 psi. Up to 1600 psi, the plastic strain levels off, with each steady state strain slightly different due to the increase in pressure. From 1700 psi and up, the plastic strain is seen to accumulate. Plastic Hoop Strain vs time 0.0025 0.002 1000 psi 1100 psi 1200 psi 0.0015 Strain 1300 psi 1400 psi 1500 psi 0.001 1600 psi 1700 psi 1800 psi 1900 psi 0.0005 2000 psi 0 0 500 1000 1500 2000 time (s) Figure 14: Plastic Hoop Strain vs time, 1000 to 2000 psi in 100 psi Increments 36 6. Discussion and Conclusions The prediction of the onset of thermal ratcheting with the use of ABAQUS is possible for complex geometry in order to facilitate the design of piping and pressure vessels. The thermal and structural analysis models successfully calculated a pressure limit at which plastic strain begins to accumulate. Maintaining design pressures below the calculated pressure results will prevent the failure mechanism of thermal ratcheting from occurring. The thermal models also facilitated the calculation of thermal through wall temperature gradients for comparison to the ASME code limits. Figure 15 plots the difference in strain at the end time for each increment from the previous, lower increment, as seen on the right of Figure 14. From Figure 15, it is seen that ABAQUS predicts a pressure for the onset of accumulated plastic strain of just over 1500 psi. Difference in Plastic Hoop Strain from Previous Pressure Iteration after 20 cycles vs Pressure 4.50E-04 4.00E-04 Difference in Strain 3.50E-04 3.00E-04 2.50E-04 2.00E-04 Difference in Plastic Hoop Strain 1.50E-04 1.00E-04 1600 5.00E-05 0.00E+00 1000 1500 2000 Pressure (psi) Figure 15: Pressure for Onset of Thermal Ratcheting 37 From the equations in Section 3.4.1 the pressure when ratcheting begins based on ASME code can be solved for. First, the equation for y’ is selected. The pressure for onset of yield is likely less than 2000 psi and the yield strength is 26980 psi, linearly interpolated at the average transient temperature of 335 °F, X= PD 2000 * 3.2 0.40. 2t w y 2 * .3 * 26980 For X<0.5, y’=1/X. Substituting for X and y’ in equation [3] gives T1 Solving for P gives P 2t w y2 0.7T1 DE 2t w y2 0.7 PDE . 2 * 0.3 * 26980 2 =1323 psi, less than 0.7 * 699 * 3.2 * 31.0 * 6.8 the 1500 psi result from ABAQUS. Using the yield strength at 70 °F of 30000 psi gives a higher pressure, 1631 psi, which is slightly over the ABAQUS results. The results from ABAQUS provide realistic results which are, as expected, slightly higher than the result given for the ASME code. 38 7. References [1] 2010 ASME boiler & pressure vessel code an international code. (2010). New York, NY: American Society of Mechanical Engineers. [2] ABAQUS (Version 6.13) [Software]. (2013). Providence, RI: Dassault Systèmes Simulia Corp. [3] Bree, J. (1967). Elastic-plastic behaviour of thin tubes subject to internal pressure and intermittent high-heat fluxes with application to fast nuclear reactor fuel elements. Journal of Strain Analysis, (2), 226-38. [4] Bree, J. (1989). Plastic deformation of a closed tube due to interaction of pressure stresses and cyclic thermal stresses. International Journal of Mechanical Sciences, 865-892. [5] Bari, S. (2001). Constitutive Modeling for Cyclic Plasticity and Ratcheting. [6] Cailletaud, G. (2003). UTMIS Course 2003 – Stress Calculations for Fatigue - 6. Ratcheting. Ecole des Mines de Paris: Centre des Materiaux. [7] Kreith, F. (2000). The CRC handbook of thermal engineering. Boca Raton, Fla.: CRC Press. [8] Kreith, F. (1965). Principles of heat transfer. Second edition. Scranton, Pa.: International Textbook. 39 Appendix A, Program Files Table 9 lists the program files which were used in the creation of this report. Table 9: Program Files File Description Valve.th.inp ABAQUS Standard heat transfer analysis 5cycles.th.inp 5 thermal cycles imported into valve.th.inp Valve10.st.inp ABAQUS Standard structural analysis, 1000 psi iteration Valve11.st.inp ABAQUS Standard structural analysis, 1100 psi iteration Valve12.st.inp ABAQUS Standard structural analysis, 1200 psi iteration Valve13.st.inp ABAQUS Standard structural analysis, 1300 psi iteration Valve14.st.inp ABAQUS Standard structural analysis, 1400 psi iteration Valve15.st.inp ABAQUS Standard structural analysis, 1500 psi iteration Valve16.st.inp ABAQUS Standard structural analysis, 1600 psi iteration Valve17.st.inp ABAQUS Standard structural analysis, 1700 psi iteration Valve18.st.inp ABAQUS Standard structural analysis, 1800 psi iteration Valve19.st.inp ABAQUS Standard structural analysis, 1900 psi iteration Valve20.st.inp ABAQUS Standard structural analysis, 2000 psi iteration Valve30.st.inp Report Calculations.xls ABAQUS Standard structural analysis, 3000 psi iteration MS Excel Workbook for calculating h, ∆T1, and plotting results Python program for sequentially running ABAQUS files using the command “abaqus python 1st.py” 1st.py 40