Professional Documents
Culture Documents
Pre Post FE
Pre Post FE
Universitat de Girona
M. Baena, C. Barris, N. Blanco, E.V. Gonzlez, J.A. Mayugo, J. Renart, D. Trias, A. Turon
September 2012
Contents
1 Introduction to Pre and Post-Process programs 1.1 Typical structure of the Finite Element programs . . . . . . . . . . . . . . . . . . . . . 1.2 Finite element generation, calculation and analysis of a typical mechanical problem 1.2.1 Generation of nite element models with the Graphical User Interface (GUI) of ANSYST M . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.2.2 Generation of ANSYST M nite element models via commands: APDL . . . . . 1.2.3 Generation of ANSYST M parametrized nite element models . . . . . . . . . . 2 Types of nite elements 2.1 Introduction . . . . . . 2.2 2D bar/truss elements 2.3 3D bar/truss elements 2.4 2D beam elements . . 2.5 Plane stress elements 2.6 Plane strain elements . 2.7 Shell elements . . . . . 2.8 Suggested problems . 1 1 2 2 6 7 9 9 10 12 14 15 17 19 20 21 21 22 34 40 40 42 44 46 47 47 49 56 56 56 59 60 61 63 65
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
3 Geometric modeling and meshing techniques 3.1 Introduction . . . . . . . . . . . . . . . . . . . 3.2 Direct generation . . . . . . . . . . . . . . . . 3.3 Solid Modeling . . . . . . . . . . . . . . . . . 3.4 Assessment of the mesh quality . . . . . . . . 3.4.1 Connectivity . . . . . . . . . . . . . . . 3.4.2 Avoid elements with high distortions, wide angles between edges. . . . . . . 3.5 Techniques to Import Models and Geometries 3.6 Suggested problems . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . warping and . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . inadmissible . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . narrow . . . . . . . . . . . . . . .
. . . . . . . . . . or . . . . . .
4 Boundary conditions 4.1 Introduction . . . . . . . . . . . . . . . . . . . . 4.2 Load application in FEM . . . . . . . . . . . . . 4.3 Constrain DOF of FE model . . . . . . . . . . . 4.3.1 Constrained displacements and rotations 4.3.2 Symmetry conditions . . . . . . . . . . . 4.3.3 Antisymmetry conditions . . . . . . . . . 4.4 Superposition . . . . . . . . . . . . . . . . . . . 4.5 Structural analysis with temperature change . 4.6 Loadcases . . . . . . . . . . . . . . . . . . . . . 4.7 Suggested problems . . . . . . . . . . . . . . . iii
. . . . . . . . . .
. . . . . . . . . .
. . . . . . . . . .
. . . . . . . . . .
. . . . . . . . . .
. . . . . . . . . .
. . . . . . . . . .
. . . . . . . . . .
. . . . . . . . . .
. . . . . . . . . .
. . . . . . . . . .
. . . . . . . . . .
. . . . . . . . . .
. . . . . . . . . .
. . . . . . . . . .
. . . . . . . . . .
. . . . . . . . . .
. . . . . . . . . .
. . . . . . . . . .
. . . . . . . . . .
. . . . . . . . . .
. . . . . . . . . .
5 Linear elastic material models 5.1 Introduction . . . . . . . . . . . . . . . 5.2 One-dimensional elastic properties . . 5.3 Bi-dimensional and Three-dimensional 5.3.1 Isotropic material . . . . . . . . 5.3.2 Orthotropic material . . . . . . 5.3.3 Anisotropic material . . . . . . . 5.4 Element coordinate system . . . . . . 5.5 Suggested problems . . . . . . . . . .
. . . . . . . . elastic . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
6 Postprocessing 6.1 Introduction . . . . . . . . . . . . . . . . . . . . 6.2 General postprocessing . . . . . . . . . . . . . 6.3 Time-history postprocessing . . . . . . . . . . . 6.4 Result analysis of combined stresses or strains 6.4.1 Combined strain . . . . . . . . . . . . . . 6.4.2 Combined stress . . . . . . . . . . . . . . 6.5 Assessment of the mesh and results quality . . 6.5.1 Energy norm to estimate the error of the 6.6 Suggested problems . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . elements . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
7 Coupling and constrain equations. Submodelling 7.1 Coupling and constrain equations. Introductory concepts 7.1.1 Coupled DOF sets . . . . . . . . . . . . . . . . . . . 7.1.2 Constraint equations of DOF . . . . . . . . . . . . 7.2 Submodelling . . . . . . . . . . . . . . . . . . . . . . . . . 7.2.1 The global model . . . . . . . . . . . . . . . . . . . 7.2.2 The submodel . . . . . . . . . . . . . . . . . . . . . 7.2.3 Submodelling procedure . . . . . . . . . . . . . . . 7.3 Suggested problems . . . . . . . . . . . . . . . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
8 Nonlinear analysis. Geometric nonlinearities 8.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.2 Geometric nonlinear behaviour due to large displacements 8.3 Buckling analysis . . . . . . . . . . . . . . . . . . . . . . . . 8.3.1 Eigenvalue buckling analysis . . . . . . . . . . . . . . 8.3.2 Nonlinear buckling analysis . . . . . . . . . . . . . . 8.4 Suggested problems . . . . . . . . . . . . . . . . . . . . . . 9 Material nonlinearities 9.1 Introduction . . . . . . . . . . . . . . . . . . 9.2 Plasticity . . . . . . . . . . . . . . . . . . . . 9.2.1 Bilinear Kinematic Plasticity . . . . . 9.2.2 Multiple-point Isotropic Plasticity . . 9.3 Two parameter Mooney-Rivlin Hyperelastic 9.4 Suggested problems . . . . . . . . . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
113 . 113 . 114 . 117 . 117 . 121 . 125 127 127 127 128 132 135 139
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
Chapter 1
Typically, the Finite Element programs are divided in three parts, the Pre-processor module, the Solution module, and the Post-processor module. In the Pre-processor, the geometry, element type, material properties, and discretisation of the problem (mesh) is achieved. Once all of the previous are dened, the analysis type, the boundary conditions, the assembly of the stiffness matrices, and the solution of the system of the equations is performed in the Solution module. Finally, once the problem has been solved, the Post-processor module helps the user to analyze the results, plotting stress, strains, displacement or reaction forces at different load levels, for example.
Figure 1.1: Structure of a FEM program. It should be mentioned here that different softwares can be used for the three different parts. For example, a specic software can be used for preprocessing the problem, and another one for solving the problem and postprocessing the results. In most of the commercial FE programs all the Pre-Process, Solution and Post-Process can be done either directly by commands or using the Graphical User Interphase (GUI). Although the latter is more user-friendly, it is important to know how to use the different commands and command scripts as it results in a much more exible way to use these programs. A very powerful tool using command scripts is the parametrization of the model, which allows the modication of the model in a very easy way. 1
1.2
The common procedures for the mechanical simulation and analysis with nite elements using ANSYST M are described in the following examples. Reminder: Units Finite element analysis are dimensionless, so before starting to create the model it is convenient to establish which dimensions are to be used and be coherent. If the material properties to be used are expressed in N/mm2, the geometric units should be in mm while forces should be expressed in N.
1.2.1
Generation of nite element models with the Graphical User Interface (GUI) of ANSYST M
Example 1.1. Use the ANSYST M graphical environment to simulate and analyse the cantilever beam shown in Figure 1.2. The beam is clamped on its left end, while on the right end a constant 1 mm displacement is imposed. The cross-section of the beam is rectangular and constant along the span. The span of the beam is 2 m, its base is 200 mm and its height is 100 mm. The material is steel, E = 210 GPa and = 0.3. In the post-process, use the graphical interface to plot the strain and stress distributions in the beam, show the reactions forces and represent the deformed shape of the beam. Obtain a list of the reaction forces, the stress variation with the vertical coordinate by dening a Path-plot and represent the temporal variation of the stress in a point with History-plot.
Figure 1.2: Cantilever beam with an imposed displacement. Solution to Example 1.1. The ANSYST M Environment for ANSYST M 13.0 contains 2 windows: the Main Window and an Output Window. Within the Main Window are 5 divisions: 1. Utility Menu. The Utility Menu contains functions that are available throughout the ANSYST M session, such as le controls, selections, graphic controls and parameters. 2. Input Line. The Input Line shows program prompt messages and allows you to type in commands directly. 3. Toolbar. The Toolbar contains push buttons that execute commonly used ANSYST M commands. More push buttons can be added if desired.
MME
4. Main Menu. The Main Menu contains the primary ANSYST M functions, organized by preprocessor, solution, general postprocessor, design optimizer. It is from this menu that the vast majority of modelling commands are issued. 5. Graphics Window. The Graphic Window is where graphics are shown and graphical picking can be made. It is here where you will graphically view the model in its various stages of construction and the ensuing results from the analysis.
Main Menu
Graphics Window
Figure 1.3: ANSYST M 13.0 interface. On the other hand, the Output Window shows text output from the program, such as listing of data etc. It is usually positioned behind the main window and can be put to the front if necessary. Before starting with the generation of the model, it is important to know how to locate and obtain help on the different commands within ANSYST M . The Help for the program can be started within the Graphic Window by different procedures. The rst option is by clicking on the Help button in the Utility Menu and selecting Help Topics. The second option is by clicking on the question mark button next to the Input Line. The last option, only valid to obtain help on a specic ANSYST M command, consists on typing help,XXXXX on the Input Line, where XXXXX is the name of the specic command. Using one of the three options, a new window appears with four different tabs. In the rst tab, Contents, all the information about the program has been arranged in accordance to different subjects. Here you can nd specic help for different commands, elements, analysis types, theory, tutorials and examples. In the second tab, Index, you can nd specic help for any command, element or analysis types arranged in alphabetical order. In the third tab, you can search all the relative information within ANSYST M (theory, command structure, analysis types, examples, etc.) related to a specic keyword. In the last tab, Favourites, you can store all the information about frequently consulted or important Help items you wish to locate easily. Once you have identied the different parts of the ANSYST M environment, start with the creation of the new model. 1. First, the model has to be identied with the simulated problem. Change the name of the model to PrePostFE_Ex101 (File>Change Jobname).
MME
2. Save the generated model. The menu command File>Save As Jobname.db saves the model as PrePostFE_Ex101.db. 3. Start with the preprocessor module (PreProcessor in the Main Menu). Dene the type of the element you are going to use throughout the analysis: Preprocessor>Element Type>Add/Edit/Delete>Add...>Solid>Brick 8node 185. ANSYST M 13.0 recommends the use of the current-technology SOLID185 element instead of SOLID45. However, for the SOLID185 element to have the same behaviour as SOLID45, the element technology should be stated to be "simplied enhanced element formulation", which stands as keyoption(2)=3. Edit the options of the element and choose this property. Spend a while looking at the different type of elements available in ANSYST M 13.0 Dene the material properties of the model: Preprocessor>Material Props>Material Models>Structural>Linear>Elastic>Isotropic. Enter the required material properties for the model, Young modulus and Poisson ratio, using the appropriate units. Again, take a while to examine the different material models available. Generate the geometry: Preprocessor>Modelling>Create>Volumes>Block>By Dimensions. Note that on the top left corner of the prompt there is the name of the associated command, [BLOCK], and a short description. Use one of the three HELP options to obtain further information on this command. Enter the dimensions of the model in order to align the span of the beam with x-axis. Use the graphical tools on the right of the Graphic Window to obtain different views of the created geometry. Mesh the model. First dene the controls to guide the mesh, for example, 40 divisions in X-direction, 4 in Y direction and 8 in Z direction. Open the mesh module Preprocessor>Meshing>MeshTool and apply the mesh controls to the model with the option Size Controls>Lines>Set and then NDIV equal to 40, 4, 8 for the corresponding lines. Dene the shape of the element to be hexahedral Hex. Then click on the Mesh button to select the block and mesh it (Mesh>Volumes>Mapped>4 to 6 sided ). 4. Continue with the solution module, Solution in the Main Menu. Dene the type of the analysis: Solution>Analysis Type>New Analysis>Static. Dene the parameters of the analysis Solution>Analysis Type>Soln Controls. Time at the end of load step =1, Number of substeps =10 and Write Items to Results le> Frequency>Write every Nth substep. With these choices the computation will be performed in 10 substeps, and the results le will be updated every Nth substep. After that, dene the boundary conditions of the model. Apply a 1mm vertical displacement at the end of the beam Solution>Dene Loads>Apply> Structural> Displacement>On Lines. Make sure the displacement is applied in the correct direction. In order to select the appropriate lines, it might result convenient to plot only the lines in the model. In the Utility Menu Plot>Lines. To display the number of the lines PlotCtrls>Numbering...>LINE Line numbers. The same procedure can be used for areas, volumes, nodes, elements, etc. Clamp the other side of the beam Solution>Dene Loads>Apply>Structural>Displacement>On Areas selecting ALL DOF. Finally, solve the model Solution>Solve>CurrentLS 5. Continue with the postprocess: General Postproc in the Main Menu. Plot the deformed shape: General Postproc>Plot Results>Deformed Shape
MME
Plot Stress or Strain quantities: General Postproc>Plot Results>Contour Plot List Reactions: General Postproc>List Results>Reaction Solu List any nodal quantity: General Postproc>List Results>Nodal Solution. Repeat the previous instructions for another substep. To do that you must move to another substep with the command General Postproc>Read Results>By Pick
Reminder: Results Viewer Alternatively, you can open the Results viewer tool General Postproc>Results Viewer to plot contour plots, list nodal or element quantities, easily change substeps and animate the results among many other utilities.
6. Create a PATH PLOT. You can dene a path through the mesh and then plot a quantity on that PATH. To do this follow the next instructions. Dene the PATH, General Postproc>Path Operations>Dene Path>By nodes, selecting a line of nodes. For example, select the nodes on the vertical edge of the clamped area and name it. Plot the PATH:General Postproc>Path Operations>Plot Paths. Create the path plot mapping the desired quantity onto the path General Postproc>Path Operations>Map onto Path>Stress>X-direction SX. Plot the path onto the geometry General Postproc>Path Operations>Plot Path Item>On Graph>SX
MME
7. HISTORY PLOT. Plot the evolution of a quantity on a node/element with step time or another quantity. Open the Time History Postprocess tool: TimeHist Postpro. Add Time-History Variable, for example Nodal Solution>Stress>X-Component of Stress. Select the desired node. Plot the Time-History variable. As an exercise plot the stress-strain evolution of a selected node.
1.2.2
In ANSYST M the generation, solving and analysis of the results can be done via commands using an specic programming language called ANSYS Parametric Design Language (APDL). Using this language the nite models can be dened as a sequence of commands in an external le, which results to be very convenient for execution and modication of the models. Example 1.2. Generate, simulate and analyse the model for the beam presented in Ex. 1.1 by means of a command le. Solution to Example 1.2. The ANSYST M command sequence for this example is listed below. You can either type these commands on the command window, or you can type them on a le, then, on the command window enter /input, le, ext or just use copy and paste.
FINISH !close all previous modules /CLEAR !clear all previous models /TITLE, Cantilever beam /PREP7 !start PRE-PROCESSOR MODULE ET,1,SOLID185 !element type #1: SOLID185 (3 DoF) KEYOPT,1,2,3 !element #1: keyoption 2=3 MP,EX,1,210000 !Youngs modulus for material #1 MP,NUXY,1,0.3 !Poisson modulus for material #1 BLOCK,0,2000,0,100,0,200 LESIZE,5,,,40 LESIZE,6,,,4 LESIZE,10,,,8 MSHAPE,0,3D VMESH,1 FINISH /SOLU ANTYPE,0 NSUBST,10,0,0 OUTRES,ALL,1 TIME,1 DL,10,,UY,-1 DA,5,ALL SOLVE FINISH !create a solid of 2mx100mmx200mm
!40 divisions in line 5 (2m) !4 division in line 6 (100mm) !8 divisions in line 10 (200mm) !hexaedral-shape: Element shape to be used for meshing !mesh volume !end of PRE-PROCESSOR MODULE !start SOLUTION MODULE !static analysis !perform the analysis in 10 sub-steps !write results for each sub-step !end time of the analysis !applied 1mm displacement in vertical direction at x=2000mm. !bottom line of the area !clamped area at x=0 !solve the current load case !end of the SOLUTION MODULE
The ANSYST M command sequence for postprocessing this example is listed below.
MME
/POST1 PLDISP,1 PLESOL,S,EQV,0,1. PLESOL,S,X,0,1. PLESOL,EPEL,X,0,1. PLESOL,EPEL,1,0,1. PRRSOL PATH,cami,2,1,1 PPATH,1,414 PPATH,2,708 /PBC,PATH,1 /REPLOT
!start Post-Processor module !plot the deformed shape !contour plot of Von Mises stress !contour plot of stress in X-direction !contour plot of strain in X-direction !contour plot of principal strain !list of reaction forces !definition of PATH "cami" !start node on PATH "cami" !end node on PATH "cami" !show path
PDEF,defcami,EPEL,X,AVG !project x-strain on path, name it defcami PLPATH,defcami !plot defined "defcami"
/POST26 !history plot module ANSOL,2,866,EPEL,X,EPELX_2 !create evolution of X-strain at node 866 PLVAR,2 !plot x-strain evolution PRVAR,2 !list x-strain evolution This le can be found at: ftp://amade.udg.edu/mms/PrePostFE/input_files/T1/PrePostFE_Ex102post.dat
1.2.3
Example 1.3. Modify the input le generated in Ex.1.2 to parametrize the geometry, loading and material properties of the model. Solution to Example 1.3. The ANSYST M command sequence for this example is listed below. You can either type these commands on the command window, or you can type them on a le, then, on the command window enter /input, le, ext. For post-processing the command sequence is the same as in the previous example. Hint 1: Parameter denition in ANSYST M Basically, there are two different ways to dene parameters in ANSYST M scripts: Using the command *SET followed by the name and value for the parameter *SET,L,2000 (denes the parameter L with a value of 2000) Directly dening the parameter and its value H=100 (denes the parameter H with a value of 100)
FINISH !close all previous modules /CLEAR !clear all previous models /TITLE, Parametrized cantilever beam /PREP7 !Parameters H=100 L=2000 B=200 Young=210000 Pois=0.3 !start PRE-PROCESSOR MODULE !beam height in mm !beam span in mm !beam base in mm !Young modulus in MPa !Poisson
MME
8
ndiv=L/H*2 !number of divisions
ET,1,SOLID185 !element type #1: SOLID185 (3 DoF) KEYOPT,1,2,3 !element #1: keyoption 2=3 MP,EX,1,Young !Youngs modulus for material #1 MP,NUXY,1,Pois !Poisson modulus for material #1 BLOCK,0,L,0,H,0,B LESIZE,5,,,ndiv LESIZE,6,,,4 LESIZE,10,,,8 MSHAPE,0,3D VMESH,1 FINISH /SOLU ANTYPE,0 NSUBST,10,0,0 OUTRES,ALL,1 TIME,1 DL,10,,UY,-1 DA,5,ALL SOLVE FINISH !create a solid block !divisions in line 5 !4 division in line 6 !8 divisions in line 10 !hexaedral-shape: Element shape to be used for meshing !mesh volume !end PRE-PROCESSOR MODULE !start SOLUTION MODULE !static analysis !perform the analysis in 10 sub-steps !write results for each sub-step !end time of the analysis !applied 1mm displacement in vertical direction at x=L. !bottom line of the area !clamped area at x=0 !solve the current load case !end SOLUTION MODULE
MME
Chapter 2
2.1
Introduction
Finite element programs usually have an element library that contains different element types. In addition, there are several different physical problems that can be modeled and simulated by means of the nite element method and every eld of application needs a different element type. For instance, the type of nite element used to simulate and obtain displacements and stresses of the beam in Fig. 1.2 cannot be used to simulate and obtain the gradients of temperature in the same beam when a heat focus is acting on the top surface. The element type determines the element formulation used, the degrees of freedom, interpolation functions, wheter the element is 2D or 3D, etc. Thus, before starting the simulation of any real problem with nite elements it is necessary to stablish the nature of the problem and determine the type of nite elements that t the simulation.
In this course, most of the analyses and simulations will be of structural nature. Table 2.1 summarises some of the ANSYST M structural elements that are more commonly used. 9
10
Element LINK1 LINK8 LINK180 BEAM3 Linear BEAM4/44 BEAM188 nodes 2 2 2 2 2 2 DOFs
ux ux ux ux ux ux
uy uy uy uy uy uy
uz uz z uz x y z uz x y z
Element description line bar/truss, 2D space line bar/truss, 3D space line bar/truss, 3D space line beam in 2D space line beam in 3D space Timoshenko beam (includes shear deformation)
Element Order linear linear linear linear linear linear(2 nodes)/ quadratic(2+1 nodes) / cubic (2+2 nodes) quadratic linear quadratic linear quadratic linear quadratic quadratic linear quadratic linear quadratic linear linear linear
BEAM189 PLANE42 2D Solid PLANE82 PLANE182 PLANE183 SOLID45 3D Solid SOLID95 SOLID92 SOLID185 SOLID186 SHELL63 SHELL93 SHELL91 SHELL99 SHELL181
3 4 8 4 8 8 20 8 8 20 4 8 4 4 4
ux uy uz x y z ux uy ux uy ux uy ux uy ux uy uz ux ux ux ux ux uy uy uy uy uy uz uz uz uz uz x y z
3D Shell
ux uy uz x y z ux uy uz x y z ux uy uz x y z ux uy uz x y z
3-node quadratic version of BEAM188 solid 4-node quadrilateral in 2D space solid 8-node quadrilateral in 2D space Similar to PLANE42 Similar to PLANE82 Basic linear brick. 8-node hexahedral 3D space 20-node version of SOLID45 10-node tetrahedral 3D space Similar to SOLID45 Similar to SOLID85 shell 4-node quadrilateral in 3D space shell 8-node quadrilateral in 3D space nonlinear layered shell 4-node quadrilateral in 3D space linear layered shell 4-node quadrilateral in 3D space Similar to SHELL63
Table 2.1: Some linear structural elements available in ANSYST M . Try to stick with the 18x series of elements when possible because they incorporate the latest technology.
Each element type has different options. For example, the planar solid element has an option to choose between plane strain and plane stress. On the other hand, some elements require the determination of real constants. These are properties that depend on the element type, such as cross-sectional properties of a beam, the ply sequence in a laminated shell element, etc. In the following sections, different structural problems are modelled using one of the suitable nite element types in order to illustrate the selection of the most appropriate nite element for each problem.
2.2
2D bar/truss elements
Example 2.1. Compute the maximum vertical displacement in the 2D overhead hoist represented in Figure 2.1, where P = 100 kN, b = 10 cm, h = 20 cm, t = 2 mm, L = 3 m, and a = 1.5 m. The material of the trusses is linear elastic with E = 210 GPa and = 0.3. Use TRUSS elements. Solution to Example 2.1. The ANSYST M command sequence for this example is listed below. You can either type these commands on the command window, or you can type them on a le, then, on the command window enter /input, le, ext.
FINISH !close all previous modules /CLEAR !clear all previous models /TITLE, Overhead hoist, 2D truss /PREP7 !start PRE-PROCESSOR MODULE
MME
11
!Parameters P=100000 b=100 h=200 t=2 Ll=3000 a=1500 S=(b*h)-(b-2*t)*(h-2*t) ET,1,LINK180 MP,EX,1,210000 MP,PRXY,1,0.3 R,1,S !Geometry K,1,0,0,0 K,2,Ll,0,0 K,3,2*Ll,0,0 K,4,Ll/2,a,0 K,5,3*Ll/2,a,0 L,1,2,1 L,2,3,1 L,3,5,1 L,4,5,1 L,1,4,1 L,2,4,1 L,2,5,1 LMESH,1,7,1 FINISH
!applied load in N !section base in mm !section height in mm !section thickness in mm !span in mm !hoist height in mm !section in mm
!element type #180: 2D/3D truss (T-C) !Youngs modulus for material #1 !Poisson modulus for material #1 !real constant #1, element section
/SOLU !start SOLUTION MODULE !Boundary and load conditions DK,1,UX !displacement x for keypoint 1 DK,1,UY DK,3,UY FK,2,FY,-P !concentrated y-load keypoint 2, value -P FK,4,FY,-P FK,5,FY,-P /PBC,all SOLVE FINISH /POST1 PLDISP,1 PRNSOL,U,Y ETABLE,STAXIL,LS,1 PLETAB,STAXIL PRETAB,STAXIL !to show BCs when solve
!start POST-PROCESSOR MODULE !deformed shape !list y-displacement !obtain axial stress !plot axial stress !list axial stress
MME
12
Observe the distribution of the axial stress in the different bars and how this, as expected, is uniform for every particular element. Try to plot the different components of the stress using the Post-Processor menu. Note that the value for all the components in all the elements is zero. The only way to obtain the axial stress is by using the ETABLE command. Use the program help for the LINK180 element and ETABLE command to understand how they work.
2.3
3D bar/truss elements
Example 2.2. The pin-articulated structure shown in Figure 2.2 is used to support a tank containing 1 tone of water. All the structural elements are hollow-circular rods with an exterior diameter D = 30 mm and thickness t = 2.5 mm. All the rods are made of steel, E = 210 GPa and = 0.3. The geometry of the structure is given by H = 5 m and B = 4 m and L = 6 m. Generate the input le to model the structure with ANSYST M . For the simulation consider an horizontal wind on the tank resulting in two horizontal forces in the upper part of the structure: FB = 5000 N and FL = 4000 N.
Figure 2.2: Pin-articulated structure supporting a tank for uids. Solution to Example 2.2. The ANSYST M command sequence for this example is listed below. You can either type these commands on the command window, or you can type them on a le, then, on the command window enter /input, le, ext.
FINISH !close all previous modules /CLEAR !clear all previous models /TITLE, Structure for fluid tank, 3D truss /PREP7 !Parameters B=4000 L=6000 H=5000 t=2.5 D=30 W=10000 FB=5000 FL=4000 !start PRE-PROCESSOR MODULE !base in mm !width in mm !height in mm !thickness in mm !outer diameter in mm !weigth in N !horitzontal force base in N !horitzontal force width in N !pi !internal diameter !rod area
MME
13
!element type #1: 2D/3D truss (T-C) !Youngs modulus for material #1 !Poisson modulus for material #1 !real constant #1, element section
!Geometry K,1,0,0,0 !keypoint, number and coordinates K,2,B,0,0 K,3,0,H,0 K,4,B,H,0 KGEN,2,ALL,,,,,L,4,1 !generate-copy all keypoints, disp. Z=L, increment number=4 L,1,3,1 L,3,4,1 L,2,4,1 L,1,4,1 L,2,3,1 L,5,7,1 L,7,8,1 L,6,8,1 L,5,8,1 L,6,7,1 L,4,8,1 L,6,4,1 L,2,8,1 L,3,7,1 L,1,7,1 L,3,5,1 LMESH,ALL FINISH !line, initial and final keypoint and divisions
/SOLU !start SOLUTION MODULE !Boundary and load conditions NSEL,S,LOC,Y,0 !select nodes located y=0 D,ALL,ALL !fix all DOF of all selected nodes NSEL,ALL !select all model nodes NSEL,S,LOC,Y,H !select nodes located y=H F,ALL,FX,FB1 !apply x-load all selected nodes F,ALL,FY,-W1 F,ALL,FZ,FL1 NSEL,ALL SOLVE FINISH /POST1 PLDISP,1 PRNSOL,U,COMP ETABLE,STAXIL,LS,1 PLETAB,STAXIL !start POST-PROCESSOR MODULE !deformed shape !list all displacements and sum !obtain axial stress !plot axial stress
Observe that in this case the model has to take into account nodes of the bars can be displaced in the three directions of the space. Thus, a 3D element is required.
MME
14
2.4
2D beam elements
Example 2.3. Compute the maximum vertical displacement in the 2D overhead hoist represented in Figure 2.3, where P = 100 kN, q = 100 N/mm, b = 10 cm, h = 20 cm, t = 2 mm, L = 3 m, and a = 1.5 m. The material of the beams is linear elastic with E = 210 GPa and = 0.3. Use BEAM elements. Plot the diagram of internal moments of each beam. Compare the results with those obtained in Ex. 2.1.
Figure 2.3: Overhead hoist with bending loads. Solution to Example 2.3. The ANSYST M command sequence for this example is listed below. You can either type these commands on the command window, or you can type them on a le, then, on the command window enter /input, le, ext.
FINISH !close all previous modules /CLEAR !clear all previous models /TITLE, Overhead hoist, 2D beam /PREP7 !Parameters P=100000 q=100 b=100 h=200 t=2 Ll=3000 a=1500 ET,1,BEAM188 KEYOPT,1,3,3 MP,EX,1,210000 MP,PRXY,1,0.3 SECTYPE,1,BEAM,HREC SECOFFSET,CENT SECDATA,h,b,t,t,t,t !Geometry K,1,0,0,0 K,2,Ll,0,0 K,3,2*Ll,0,0 K,4,Ll/2,a,0 K,5,3*Ll/2,a,0 L,1,2,10 L,2,3,10 L,3,5,10 L,4,5,10 L,1,4,10 !start PRE-PROCESSOR MODULE !applied load in N !distribuited load N/mm !section base in mm !section height in mm !section thickness in mm !span in mm !hoist height in mm !element type #1: BEAM188 (6 DoF) !element #1 keyoption 3=3 (cubic shape function) !Youngs modulus for material #1 !Poisson modulus for material #1 !Beam hollow rectangular section type for material #1 !section centered (offset=0) !section data, with base, height and four lateral thickness
MME
15
/SOLU !start SOLUTION MODULE !Boundary and load conditions DK,1,UX !displacement x for keypoint 1 DK,1,UY !displacement y for keypoint 1 DK,1,UZ !displacement z for keypoint 1 DK,1,ROTX !rotations x for keypoint 1 DK,1,ROTY !rotations y for keypoint 1 DK,3,UY !displacement y for keypoint 3 DK,3,UZ !displacement z for keypoint 3 DK,3,ROTX !rotations x for keypoint 3 DK,3,ROTY !rotations y for keypoint 3 FK,2,FY,-P !concentrated y-load keypoint 2, value -P FK,4,FY,-P FK,5,FY,-P LSEL,S,,,4 !select line 4 (top line) ESLL,S !select attached elements to line SFBEAM,ALL,2,PRES,q !apply pressure on face 2 (y-normal direction) ESEL,ALL !select all elements /PBC,all SOLVE FINISH /POST1 PLDISP,1 PRNSOL,U,Y ETABLE,AXI,SMISC,1 ETABLE,AXJ,SMISC,14 PLLS,AXI,AXJ,1,0 ETABLE,TAI,SMISC,6 ETABLE,TAJ,SMISC,19 PLLS,TAI,TAJ,1,0 ETABLE,MOI,SMISC,3 ETABLE,MOJ,SMISC,16 PLLS,MOI,MOJ,1,0 /ESHAPE,1,1 PLNSOL,S,X !to show BCs when solve
!start POST-PROCESSOR MODULE !deformed shape !list y-displacement !table of axial force FX of node i !table of axial force FX of node j !diagram of axial forces FX, scale 1, undeformed !table of section shear force SFY of node i !diagram of section shear force SFY, scale 1, undeformed !table of bending moment MZ of node i !diagram of bending moment MZ, scale 1, undeformed !display real initial shape elements
Observe that in this case the elements are subjected to bending moment, as indicated in the moment diagram. Use the program help for the BEAM188 element and ETABLE command to understand how they work.
2.5
Example 2.4. Generate the ANSYST M input le to model and simulate the stress concentration coupon represented in Figure 2.4. The geometry of the coupon, which is symmetric, is given by H = 40 mm, h = 20 mm, B = 100 mm and b = 40 mm. The thickness of the coupon, t = 2 mm, is constant. An horizontal displacement is applied on the right vertical edge, as shown in the gure, where u = 5102 mm. The material of the coupon is steel, E = 210 GPa and = 0.3. Solution to Example 2.4. The ANSYST M command sequence for this example is listed below.
MME
16
Figure 2.4: Stress concentration coupon. You can either type these commands on the command window, or you can type them on a le, then, on the command window enter /input, le, ext.
FINISH !close all previous modules /CLEAR !clear all previous models /TITLE, Plane stress coupon /PREP7 !Parameters B=100 B1=40 H=40 H1=20 t=2 u=0.05 ET,1,PLANE182 KEYOPT,1,1,3 KEYOPT,1,3,3 R,1,t MP,EX,1,210000 MP,PRXY,1,0.3 !start PRE-PROCESSOR MODULE !base in mm !base triangle in mm !width in mm !small widht in mm !thickness in mm !displacement in mm !element type #1: plane solid !element #1, keyopt 1=3, simplified enhanced strain formulation !element #1, keyopt 3=3, plane stress w/ thickness !real constant #1, element thickness !Youngs modulus for material #1 !Poisson modulus for material #1
!Geometry K,1,0,0 !keypoint, number and coordinates K,2,(B-B1)/2,0 K,3,B/2,(H-H1)/2 K,4,((B-B1)/2)+B1,0 K,5,B,0 K,6,0,H K,7,(B-B1)/2,H K,8,B/2,((H-H1)/2)+H1 K,9,((B-B1)/2)+B1,H K,10,B,H L,1,2,15 L,2,3,10 L,3,4,10 L,4,5,15 L,6,7,15 L,7,8,10 L,8,9,10 L,9,10,15 L,1,6,20 L,5,10,20 L,2,7,20 L,3,8,20 L,4,9,20 AL,1,11,5,9 AL,2,12,6,11 !line, initial and final keypoint and divisions
MME
17
/SOLU !start SOLUTION MODULE !Boundary and load conditions NSEL,S,LOC,X,0 !select left nodes D,ALL,UX !fix x-displacement NSEL,S,LOC,X,0 !select left bottom node NSEL,R,LOC,Y,0 D,ALL,UY !fix y-displacement NSEL,S,LOC,X,B !select right nodes D,ALL,UX,u !apply displacement NSEL,ALL !select all nodes /PBC,all !to show BCs when solve SOLVE FINISH /POST1 PLDISP,1 PLNSOL,S,EQV !start POST-PROCESSOR MODULE !deformed shape !Von Mises stress
Although the thickness of the coupon has been considered in the model (R,1,t ), this has no inuence in the results as the displacement is applied. Check the help about the element PLANE182 to see how the element can be dened without considering the thickness of the coupon. Note that in this case, if a load was applied instead of a displacement, the load should be normalized by the thickness of the coupon.
2.6
Example 2.5. Generate the ANSYST M input le to model and simulate the central cross-section of the rectangular pressure conduct represented in Figure 2.5. Consider that the material of the conduct is made of an homogeneous, linear and isotropic plastic with E = 27 GPa and = 0.25. The geometry of the element is given by H = 75 mm and B = 50 mm. The thickness of the section is t = 5 mm. The pressure in the interior of the conduct is p = 10 kg/cm2 .
MME
18
Solution to Example 2.5. The ANSYST M command sequence for this example is listed below. You can either type these commands on the command window, or you can type them on a le, then, on the command window enter /input, le, ext.
FINISH !close all previous modules /CLEAR !clear all previous models /TITLE, Plane strain square pipe /PREP7 !Parameters B=50 H=75 t=5 p=1 ET,1,PLANE182 KEYOPT,1,1,3 KEYOPT,1,3,2 MP,EX,1,27000 MP,PRXY,1,0.25 !Geometry K,1,0,0 K,2,B,0 K,3,B,H K,4,0,H K,5,t,t K,6,B-t,t K,7,B-t,H-t K,8,t,H-t L,1,2,20 L,2,3,30 L,3,4,20 L,4,1,30 L,5,6,20 L,6,7,30 L,7,8,20 L,8,5,30 L,1,5,4 L,2,6,4 L,3,7,4 L,4,8,4 AL,1,5,9,10 AL,2,6,10,11 AL,3,7,11,12 AL,4,8,9,12 AGLUE,ALL AMESH,ALL FINISH !start PRE-PROCESSOR MODULE !base in mm !height in mm !thickness in mm !pressure in N/mm2 !element !element !element !Youngs !Poisson type #1: plane solid #1, keyopt 1=3, simplified enhanced strain formulation #1, keyopt 3=3, plane strain modulus for material #1 modulus for material #1
/SOLU !start SOLUTION MODULE !Boundary and load conditions NSEL,S,LOC,X,0 !select left nodes NSEL,A,LOC,Y,0 !add bottom nodes to selection D,ALL,ALL !fix displacement LSEL,S,LINE,,5,8 !select lines 5 to 8 NSLL,S,1 !select all nodes on lines SF,ALL,PRES,p !apply pressure on all selected nodes NSEL,ALL /PBC,ALL SOLVE FINISH /POST1 PLDISP,1 PLNSOL,S,EQV !to show BCs when solve
MME
19
ftp://amade.udg.edu/mms/PrePostFE/input_files/T2/PrePostFE_Ex205.dat
Although the length of the conduct is not known, this parameter is not required for the simulation as plane strain condition is considered.
2.7
Shell elements
Example 2.6. Write the ANSYST M or ABAQUST M input le to model a simply supported rectangular plate with dimensions L = B = 1 m and thickness t = 10 mm. The material of the plate is steel and it is loaded with a uniform pressure p = 0.1 N/mm2 . Obtain the stress distribution on the x-direction at bottom, middle and top locations in the thickness direction. Solution to Example 2.6. The ANSYST M command sequence for this example is listed below. You can either type these commands on the command window, or you can type them on a le, then, on the command window enter /input, le, ext.
FINISH !close all previous modules /CLEAR !clear all previous models /TITLE, 3-D simply supported shell plate /PREP7 !Parameters B=1000 L=1000 t=10 p=0.1 ET,1,SHELL181 KEYOPT,1,3,2 KEYOPT,1,8,2 SECTYPE,1,SHELL SECDATA,t MP,EX,1,27000 MP,PRXY,1,0.25 !start PRE-PROCESSOR MODULE !base in mm !width in mm !thickness in mm !pressure in N/mm2 !element type #1: shell !element #1: keyoption 3=2 (full integration incompatible modes) !element #1, keyopt 8=2, store data top, bottom and mid surfaces !section #1 = shell !thickness !Youngs modulus for material #1 !Poisson modulus for material #1
!Geometry K,1,0,0 K,2,B,0 K,3,0,L K,4,B,L L,1,2,40 L,1,3,40 L,3,4,40 L,2,4,40 AL,ALL AMESH,ALL FINISH
/SOLU !start SOLUTION MODULE !Boundary and load conditions NSEL,S,LOC,X,B/2 !select central node NSEL,R,LOC,Y,L/2 D,ALL,UX,,,,,UY,ROTX,ROTY,ROTZ !fix DOFs NSEL,S,LOC,X,0 NSEL,R,LOC,Y,L/2 D,ALL,UY NSEL,S,LOC,X,0 D,ALL,UZ NSEL,S,LOC,X,B D,ALL,UZ NSEL,S,LOC,Y,0
MME
20
D,ALL,UZ NSEL,S,LOC,Y,L D,ALL,UZ NSEL,ALL SFE,ALL,,PRES,,p /PBC,ALL SOLVE FINISH /POST1 PLDISP,1 PLNSOL,S,X SHELL,MID PLNSOL,S,X
!start POST-PROCESSOR MODULE !deformed shape !x-stress in top and bottom surfaces !select results on mid surface (TOP,MID,BOT) !s-stress in midle surface
Observe that the value of the stress in the x-direction varies from the top to the mid and bottom surfaces. This is due to the fact that although a shell element corresponds to a planar representation, it is a 3D element and takes into account the thickness of the system.
2.8
Suggested problems
Problem 2.1. Generate the ANSYST M input le for the simulation of the beam represented in Figure 1.2 using the TRUSS, BEAM, PLANE STRESS and SHELL element types. Analyze the differences and limitations in the use of the element types listed above and the solid elements used in example 1.1. Problem 2.2. Modify the input le generated in the previous problem for shell elements and the input le in Ex. 1.3 for solid elements and simulate different ratios of H/L (height and span of the beam, respectively). Analyze for which values of the ratio, the results obtained with shell elements are similar to those obtained with solid elements. Problem 2.3. Modify the input le generated in Ex. 2.2 to simulate the structure in Figure 2.2 without the inclined bars. Consider different types of elements and boundary conditions and use the most appropriate. Problem 2.4. Modify the input le generated in Ex. 2.4 to obtain the variation of the maximum Von Mises stress versus the size of the nite element. Redene the mesh to consider 10, 20, 40, 60, 80 and 100 elements in the vertical direction and plot the obtained value of stress versus the number of divisions.
MME
Chapter 3
The goal of the meshing process is to convert a problem in the continuum real into a problem which can be solved by a computer and so, a discrete problem. This process involves two different aspects: The discretization of the geometry. The mesh must be ne enough to avoid sharp corners in curves or curved surfaces which may lead to unreal stress concentrations. The discretization of the solution. The result is obtained by the FE code only in a discrete set of locations so the mesh should also ensure that the number of solution points is enough to express the variation of the solution elds along the domain. On the other hand, a coarse mesh needs less computational resources than a ne mesh, so the user generally should establish an equilibrium position, that is, to obtain a mesh ne enough to discretize the geometry and the solution and coarse enough for the computational resources available. The meshing process creates elements and nodes. The elements are the portions (lines, surfaces or volumes) in which the continuum is decomposed. These elements are delimited by nodes which are those locations in which the solution (the variables corresponding to the degrees of freedom such as displacements, rotations and temperatures) of the system of equations is obtained. From the degrees of freedom, other relevant magnitudes which cannot be expressed in a point may be obtained and expressed in the elements. These are generally tensorial magnitudes (which need a volume face to be expressed) or uxes (which need an area to be computed and expressed). However, this magnitudes are computed by integrating the shape function of the element at the integration points of the element which guarantee a minimum error numerical integration. In this chapter some general mesh strategies and some rule-of-the-thumb recommendations are given. Meshing strategies There exist different methods to model and mesh the geometry of a structure. They can be classied in two categories: direct generation of mesh and solid modeling. 21
Generally, commercial FE codes allow the direct denition of nite element entities (nodes, elements, and boundary conditions on nodes and elements) or the denition of geometric entities (points, lines, surfaces, or volumes) which are meshed later in order to obtain the nodes and elements. Considerations Convenient for simple models Direct control of nodes and elements Program skills are needed Allows the translation of the model between different commercial codes or denition using standard codes Convenient for large models Requires larger computational resources It is easy to modify the geometry No direct control over the mesh entities
Direct generation
Solid modeling
Meshing recommendations The mesh should ensure connectivity through the nodes to guarantee a correct load distribution, that is, each element must share some nodes with its neighboring elements and all nodes, except from those nodes located at corners of the global geometry, must belong to at least two elements. Curves should be meshed with, at least, 12 elements for each circumference quarter to guarantee a smooth geometric discretization. Commercial FE codes generally provide re-meshing tools which in an iterative process rene the mesh in those places with large gradients of the solution. Elements in a mesh may be triangles (or tetrahedral) or quadrilateral (or bricks). Generally, triangular elements provide constant values for the integrated variables (stress) but can be used to mesh any geometry easily. On the other hand, quadrilateral elements offer a slightly better approximation of the distribution of the integrated variables through the element, although they may not be able to mesh some geometries accurately. Except for those cases with complicated geometries quadrilaterals are preferable. There are some element types with internal nodes which improve the discretization of the solution and are useful when large gradients are present. These are elements such as 9-node quadrilaterals, 12-node bricks, etc.
3.2
Direct generation
The process of creating elements out of nothing, without starting from any geometric entity is called direct generation or direct mesh. In this process the user imposes explicitly the location of the nodes and the size of the elements. In this process the user has a total control on the node location and element size. This approach is generally used for simple geometries.
MME
23
Direct generation consists on dening rst the node distribution in the geometry. The position of each node is dened according to a coordinate system. Some commercial codes allow to use different types of coordinate systems (cartesian, cylindrical, spherical, as so on). Hint 3: Useful ANSYS commands to directly dene nodes Command N NGEN FILL NSYM NLIST NPLOT NDELE Main Parameters x,y,z,thxy,thyz,thxz ntime,inc,n1,n2,ninc,dx,dy,dz node1,node2,nfill ncomp,inc,node1,node2,ninc node1,node2,ninc node1,node2,ninc
Dene individual node Generate nodes from a pattern of nodes Generate line nodes between two nodes Generate a reected set of nodes List nodes Display nodes Delete nodes
Once the nodes have been generated, the elements can be dened. The element is dened using the connectivity with its nodes. The set of elements dened should dene the whole geometry. Hint 4: Useful ANSYS commands to dene elements Command E EGEN ESYM EMODIF ELIST EPLOT EDELE Main Parameters i,j,k,l,m,n,o,p ntime,ninc,elem1,elem2,elinc, ,--,ninc,elem1,elem2,elinc iel,stloc,i1,i2,i3,i4,... elem1,elem2,elinc elem1,elem2,elinc
Dene individual element Generate elements from a pattern Generate elements by reection Modify element attributes List elements Display elements Delete elements
The following examples show the use of these different commands to dene nodes and elements. Example 3.1. Let us model the bi-pinned structure shown in Figure 3.1. A vertical concentrated force F = 10 kN is applied in the horizontal span, as shown. The beam has a squared cross section of 1010 mm, and is made of wood with E = 3000 GPa. Use the direct generation method in ANSYST M to dene the structure with beam elements. Solution to Example 3.1. Beam elements of length 20 mm have been used to model the structure. For this reason, node number 1 has been created at point A, node number 11 where the concentrated force is applied, point B. Thus, between these two nodes 10 elements of length 20 mm have been dened. Node number 16 has been created at point C and node 20 at the bottom end, point D. Nodes #1, #11, #16 and #20 have been dened with the N command. The distances between these nodes have been lled using the FILL command, obtaining nodes every 20 mm (see Figure 3.2). A consecutive node numeration has been obtained. Therefore, the elements can be dened easily with E and EGEN commands.
MME
24
F 8 cm
Figure 3.1: Bi-pinned structure. The ANSYST M command sequence for this example is listed below. You can either type these commands on the command window, or you can type them on a le, then, on the command window enter /input, le, ext or just use copy and paste.
FINISH !close all previous modules /CLEAR !clear all previous models /TITLE, Bi-pinned structure using BEAM188 element !Parameters L1=200 L2=300 L3=80 b=10 h=10 /PREP7 ET,1,BEAM188 KEYOPT,1,3,3 SECTYPE,1,BEAM,RECT SECOFFSET,CENT SECDATA,h,b MP,EX,1,3000e3 N,1 N,11,L1 N,16,L2 N,20,L2,-L3 FILL,1,11 FILL,11,16 FILL,16,20 E,1,2 EGEN,19,1,1 FINISH
!first horizontal distance (mm) !second horizontal distance (mm) !vertical distance (mm) !element width (mm) !element height (mm) !start PRE-PROCESSOR MODULE !element type #1: BEAM188 (6 DoF) !element #1 keyoption 3=3 (cubic shape function) !section #1 = rectangular beam !centered section (offset=0) !section width and height !material #1: E=300e3 MPa, if Poisson ratio is not defined 0.3 is assumed !define node #1 on !define node #11 on !define node #16 on !define node #20 on !fill nodes between !fill nodes between !fill nodes between coordinates coordinates coordinates coordinates #1 and #11 #11 and #16 #16 and #20 (0,0) (L1,0) (L2,0) (L2,-L3)
!define element by nodes #1, #2 !generate 18 elements incrementing 1 node like element 1 (total 19 elements) !finish PRE-PROCESSOR MODULE
Before solving the model it is necessary to dene the pins at both ends (nodes #1 and #20) and apply the vertical force in node #11. Example 3.2. Model using shell elements the 80 mm thick steel bending plate shown in Figure 3.3 . The plate has a trapezoidal shape with two right angles in B and C and is subjected to a bending load. Note that there is a simple support at 1000 mm from the clamped edge. Solution to Example 3.2. With N and FILL commands 21 rows of 13 nodes are dened to complete the whole geometry (see Figure 3.4). Between each row there is an increment in numeration
MME
25
Figure 3.2: Nodes and beam elements for the bi-pinned structure.
1500 B A 90
0.9kN 10 00 10 00
90 C
800
Figure 3.3: Schema of the bending plate (dimensions in mm). of 1, meanwhile between each column the increment in numeration is 100. With this regular distribution of the node numbers it is easy to dene all elements only using the EGEN command twice. The simple support can be dened in the central row of nodes.
Figure 3.4: Finite element model of bending plate. The ANSYST M command sequence for this example is listed below. You can either type these commands on the command window, or you can type them on a le, then, on the command window enter /input, le, ext or just use copy and paste.
FINISH !close all previous models /CLEAR !clear all previous models /TITLE,Clamped Plate under Bending /PREP7 ET,1,SHELL181 !start PRE-PROCESSOR MODULE !element type #1: SHELL181 (6 DoF)
MME
26
KEYOPT,1,3,2 SECTYPE,1,SHELL SECDATA,8 MP,EX,1,210e3 MP,PRXY,1,0.3
!element #1: keyoption 3=2 (full integration incompatible modes) !section #1 = shell !thickness !material #1: Young modulus E=210e3 MPa !material #1: Poisson coeffient=0.3
N,1 !define node #1 on coordinates (0,0) N,21,,2000 !define node #21 on coordinates (0,2000) mm N,1201,1500 !define node #1201 on coordinates (1500,0) mm N,1221,800,2000 !define node #1221 on coordinates (800,2000) mm FILL,1,21 !fill nodes between #1 and #21 FILL,1201,1221 !fill nodes between #1201 and #1221 FILL,1,1201,11,,100,21,1 !use command fill to generate rest of nodes E,1,101,102,2 EGEN,20,1,1 EGEN,12,100,1,20 FINISH !define element #1 !generate the first element column from element #1 !generate the rest of element columns !finish PRE-PROCESSOR MODULE
To generate more complex geometries with the direct method it is possible to change between different coordinate systems and also create new coordinate systems dened as "local". Hint 5: Useful ANSYS commands to manage and dene coordinate systems Command CSYS LOCAL Main Parameters ncsys: 0 - cartesian, 1 - cylindrical, 2 - spherical,... newcsys,ncsys,xc,yc,zc,thxy,...
The following example uses different coordinate systems to dene the geometry of a curved beam. Example 3.3. Model the geometry shown in Figure 3.5 using 2-D solids and the direct generation of nodes and elements. The element thickness is 4 mm and the material used is stainless steel (with a Youngs modulus of 190 GPa and a Poisson coefcient of 0.3).
MME
27
Solution to Example 3.3. The curved part of the beam is dened using a cylindrical (polar) coordinate system (where the rst coordinate corresponds to the radius R and the second to the angle ). The rectangular part of the beam is dened with a cartesian coordinate system. In this case the FILL command has been used with a tailored ratio aiming to increase the density of the mesh in the critical part (see Figure 3.6).
Figure 3.6: Nodes to dene the mesh and its numeration. The ANSYST M command sequence for this example is listed below. You can either type these commands on the command window, or you can type them on a le, then, on the command window enter /input, le, ext or just use copy and paste.
FINISH !close all previous modules /CLEAR !clear all previous models /TITLE, Curved Beam - Direct generation /PREP7 ET,1,PLANE182 KEYOPT,1,1,3 KEYOPT,1,3,3 R,1,4 MP,EX,1,190e3 CSYS,1 N,1,25,180 N,15,25,90 FILL,1,15,,,,,,3 NGEN,9,20,1,15,1,3 !start PRE-PROCESSOR MODULE !element type #1: PLANE182 (2 DoF) !element type #1: keyoption 1=3 (simplified enhanced strain formulation) !element type #1: keyoption 3=3 (plane stress w/thickness) !real constant set #1: th=4 mm !material #1: E=190000 MPa, if Poisson ratio is not defined 0.3 is assumed !activate cylindrical coordinate system !define node #1 on coordinates R=25 mm and Angle=180 !define node #15 on coordinates R=25 mm and Angle=90 !fill nodes between #1 and #15, last spacing 3 times larger than the initial one !generate nodes repeating the row: 8 new rows, increasing the radius !a distance of 3 mm in each row !activate cartesian coordinate system !define node #20 on coordinates (30,25) mm !fill nodes between #15 and #20 !generate nodes repeating the row: 8 new rows, increasing the Y position !a distance of 3 mm in each row !define element by nodes #1, #2, #22, and #21 !generates 19 elements of first row !generates rest of elements repeating the row !finish PRE-PROCESSOR MODULE
MME
28
Models can include different types of elements, real constants sets, material behaviours, etc. The ANSYS commands ET, R and MP allow to dene database tables with the different attributes of the elements used in the model. Each table uses reference numbers that associate the attribute with the element, each element keep these kind of information with the nodal connectivity. As it is shown in the Figure 3.7, the ANSYS commands TYPE, REAL and MAT dene the element type, real constant set and material, respectively. The elements generated after this selection will have the previous selected attributes. However, the change of this attributes a posteriori is also possible using the command EMODIF.
Element Types TYPE,2 ET,1,BEAM188 ET,2,SHELL181 ET,3,SOLID185 ET,4,MASS21 Real constants ET,5, ... R,1,A,Izz,h R,2,th1 R,3,th2 REAL,3 R,4,mass Materials R,5,... MP,EX,1,210e3 MP,PRXY,1,0.3 MP,EX,2,70e3 MP,PRXY,2,0.29 MAT,2 MP,DENS,2,2.7
Example 3.4. Compute the maximum vertical displacement in the 2D overhead hoist represented in Figure 3.8, where P = 100 kN, q = 100 N/mm, b = 10 cm, h = 20 cm, t = 2 mm, L = 3 m, and a = 1.5 m. The material of the beams is linear elastic with E = 210 GPa and = 0.3. Use BEAM elements on the top element and TRUSS elements for the rest. Compare the results with those obtained in 2.3.
Solution to Example 3.4. Two element types are dened (BEAM188 and LINK180). A section set is dened associated to BEAM188, to compute the section Area and Inertia. A real constant set is
MME
29
also dened associated to LINK180 dening the Area. The commands TYPE and REAL allow selecting the element type and the constant set for the elements dened next. With the ELIST command a list of dened elements are shown and the assigned type element and constant set can be checked. The ANSYST M command sequence for this example is listed below. You can either type these commands on the command window, or you can type them on a le, then, on the command window enter /input, le, ext.
FINISH !close all previous modules /CLEAR !clear all previous models /TITLE, Overhead hoist - Including beam and truss elements /PREP7 !Parameters P=100000 q=100 b=100 h=200 t=2 Ll=3000 A=1500 S=(b*h)-(b-2*t)*(h-2*t) ET,1,BEAM188 KEYOPT,1,3,3 SECTYPE,1,BEAM,HRECT SECOFFSET,CENT SECDATA,h,b,t,t,t,t ET,2,LINK180 R,2,S MP,EX,1,210000 MP,PRXY,1,0.3 N,1,0,0,0 N,2,Ll,0,0 N,3,2*Ll,0,0 N,4,Ll/2,A,0 N,14,3*Ll/2,A,0 FILL,4,14 TYPE,2 REAL,2 E,1,2 E,2,3 E,3,14 E,1,4 E,2,4 E,2,14 TYPE,1 E,4,5 EGEN,10,1,7 FINISH /SOLU ANTYPE,STATIC D,1,ALL D,3,UY,,,,,UZ D,9,UZ D,4,ROTX,,,,,ROTY D,14,ROTX,,,,,ROTY F,2,FY,-P F,4,FY,-P F,14,FY,-P !start PRE-PROCESSOR MODULE !applied load !distributed load !section base !section height !section thickness !span !hoist height !cross section area !element type #1: 3D beam (6 DoF) !element #1 keyoption 3=3 (cubic shape function) !section #1 = hollow rectangular beam !centered section (offset=0) !section width, height and thicknesses !element type #2: 3D truss (3 DoF) !geometric properties section #2: for truss elements !material #1: Youngs modulus !material #1: Poisson modulus !define node #1 on coordinates (0,0,0) !define rest of nodes
!select TRUSS element type #2 !select real constant set #2 !define TRUSS elements
!select BEAM element type #1 !define BEAM elements !finish PRE-PROCESSOR MODULE !start SOLUTION MODULE !displacement X, Y and Z for node #1 !displacement Y and Z for node #3 !displacement Z for node #9 !rotations X and Y for node #4 !rotations X and Y for node #14 !concentrated Y-load nodes #2, #4 and #14
MME
30
ESEL,S,TYPE,,1 SFBEAM,ALL,2,PRES,q ESEL,ALL /PBC,ALL SOLVE FINISH /POST1 PLDISP,1 /ESHAPE,1,1 PLNSOL,S,X !select BEAM elements (type #1) !apply distributed load on element !select all elements !show BCs when solve !solve current load state !finish SOLUTION MODULE
!start POST-PROCESSOR MODULE !plot deformed shape !plot element with real shape !contour plot of stress X distribution
Note that both BEAM188 and LINK180 are 3D elements and as we are considering a 2D analysis some displacements and rotations must be constrained to avoid displacements and deformations in the Z-direction. Example 3.5. Model the beam shown in Figure 3.9 using 2D solid elements. The beam is composed by a stainless steel bar reinforced with two aluminium plates perfectly bonded.
8 N/mm2
80 100
Figure 3.9: Two materials beam example (dimensions in mm). Solution to Example 3.5. The PLANE182 element can be used to dene a 2D solid with different thicknesses, using KEYOPTION number 3 (plain stress with thickness). Two thicknesses (16 mm thickness constant set #1 and 10 mm thickness constant set #2) and two materials (Stain Steel material #1, and Aluminium material #2) are dened. Before dening the elements with REAL and MAT commands the thickness and the material is selected. The ANSYST M command sequence for this example is listed below. You can either type these commands on the command window, or you can type them on a le, then, on the command window enter /input, le, ext.
FINISH !close all previous modules /CLEAR !clear all previous models /TITLE, Two materials beam /PREP7 ET,1,PLANE182 KEYOPT,1,1,3 KEYOPT,1,3,3 R,1,16 R,2,10 MP,EX,1,190000 MP,PRXY,1,0.29 MP,EX,2,70000 MP,PRXY,2,0.30 !start PRE-PROCESSOR MODULE !element type #1: PLANE182 (2 DoF) !element type #1: keyoption 1=3 (simplified enhanced strain formulation) !element type #1: keyoption 3=3 (plane stress w/thickness) !real constants set #1: th=16 mm, !real constants set #2: th=10 mm, !material #1: E=190000 MPa !material #1: Poisson=0.29 !material #2: E=70000 MPa !material #2: Poisson=0.30
MME
31
N,1,0,0 !define node #1 on coordinates (0,0) mm N,101,100,0 !define node #100 on coordinates (100,0) mm FILL,1,101 !fill nodes between #1 and #101 NGEN,5,200,1,101,1,,1.25 !generate 5 rows of nodes with y=+1.25 mm each row respect to pattern NGEN,3,200,811,891,1,,1 !generate 3 rows of nodes with y=+1 mm each row respect to pattern NSYM,Y,2000,ALL !generates nodes by symmetry to plane normal to Y axis REAL,1 MAT,1 E,1,2,202,201 EGEN,100,1,1 EGEN,4,200,1,100,1 REAL,2 MAT,2 E,811,812,1012,1011 EGEN,80,1,401 EGEN,2,200,401,480 ENSYM,2000,,2000,ALL NUMMRG,NODES FINISH /SOLU ANTYPE,STATIC D,1,ALL D,101,UY NSEL,S,LOC,X,20,80 NSEL,R,LOC,Y,5.75,7.25 ESLN,S,1 SFE,ALL,3,PRES,,8 ALLSEL /PBC,ALL SOLVE FINISH /POST1 PLDISP,2 PLNSOL,S,X !select constant set #1 !select material #1 !define element #1 !generate a row of elements !generete 3 more rows of elements !select constant set #2 !select material #2 !define element #301 !generate a row of elements !generate 1 more row of elements !generate elements by symmetry !merge coincident nodes !finish PRE-PROCESSOR MODULE !start SOLUTION MODULE !select Static analysis (default option) !articulation on node #1 !constrain vertically node #101 !select nodes of elements to apply surface load !select elements attached to the selected nodes !apply surface load on elements: q=8 N/mm2 !select everythhing
!solve current load state !finish SOLUTION MODULE !start POST-PROCESSOR MODULE !plot displaced model !contour plot of stress X distribution
Direct generation commands can be also used when the geometry and the mesh can be dened with mathematical functions. With the help of a programming language (Python, FORTRAN, C/C++, MatLab, etc.) the functions can be used to dene the nodes position, the element connectivity and the applied loads. After using simple commands (N, E, D, F, etc.) the whole model can be generated. The denition of a model using a MatLab code is shown in Example 3.6. Example 3.6. Model a sculpture that represents a Moebius tape (see Fig. 3.10). The mid radius R of the structure is 1325 mm. The length B of the 48 cross beams is 700 mm. The structure is made of steel (E = 210000 MPa and = 0.29), the perimetric beam is a 40 mm diameter circular rod and the cross beams are plates of 10 60 mm2 . The cross beams that support and x the sculpture to the oor have a cross-section of 25 130 mm2 . Solution to Example 3.6. The position of the nodes and the forces due to wind loads are calculated using a MATLAB code. This code generates two les: nodes.txt which contents the nodal positions, and wind.txt that contents the nodal forces due to wind.
% PrePost solution of Example 3.06 % J.A. Mayugo, 2007 % J. Renart, 2010
MME
32
perimetral beam
48 cross beams
B
Figure 3.10: Geometry of the Moebius tape sculpture.
clear all;close all;clc; number=[1:48]; angle=[0:360/48:360*(1-1/48)]; c_angle=cos(angle/180*pi); s_angle=sin(angle/180*pi); rotation=[45:180/48:225-180/48]; c_rotation=cos(rotation/180*pi); s_rotation=sin(rotation/180*pi); radius=1325; width=700; width_xy=width*c_rotation; radius_mid_X=radius.*c_angle; radius_mid_Y=radius.*s_angle;
%clear all figures, data and screen %number of cross beams %angle increment %cos(angle) %sin(angle) %rotation_ini=45 degrees %cos(rotation) %sin(rotation) %mid radius %Moebius width %width in 3D %Moebius mid radius X component %Moebius mid radius Y component
%cilindrical coordinates to define the bars of the Moebius tape radius_bot_X=radius_mid_X+0.5*width_xy.*c_angle; radius_bot_Y=radius_mid_Y+0.5*width_xy.*s_angle; radius_bot_Z=width*s_rotation; radius_top_X=radius_mid_X-0.5*width_xy.*c_angle; radius_top_Y=radius_mid_Y-0.5*width_xy.*s_angle; radius_top_Z=-width*s_rotation; hold on; plot3(radius_bot_X,radius_bot_Y,radius_bot_Z,r) plot3(radius_top_X,radius_top_Y,radius_top_Z)
%plot the external face of the Moebius tape %plot the internal face of the Moebius tape
for i=1:length(number); plot3([radius_bot_X(i) radius_top_X(i)],[radius_bot_Y(i) radius_top_Y(i)],... %Moebius tape bars [radius_bot_Z(i) radius_top_Z(i)],k); plot(radius_mid_X(i),radius_mid_Y(i),ks); %Moebius tape load points end hold off %write the results in the file nodes.txt fid0=fopen([nodes.txt],w); for inum=1:48; fprintf(fid0,n,%i,%8g,%8g,%8g \n,inum,radius_mid_X(inum),radius_mid_Y(inum),0); fprintf(fid0,n,%i,%8g,%8g,%8g \n,inum+100,radius_bot_X(inum),radius_bot_Y(inum),radius_bot_Z(inum)); fprintf(fid0,n,%i,%8g,%8g,%8g \n,inum+200,radius_top_X(inum),radius_top_Y(inum),radius_top_Z(inum)); end; fclose(fid0);
MME
33
%wind [Km/h) %wind [m/s] %pressure on the bars [kg/m^2] %pressure on the bars [N/mm^2] %pressure coefficient
%area on which de pressure is applied (plate area) %force applied on the plate
%write the results in the file wind.txt fid1=fopen([wind.txt],w); for inum=1:48; fprintf(fid1,f,%i,fz,%8g \n,inum,F_plate(inum)); end; fclose(fid1); This le can be found at: ftp://amade.udg.edu/mms/PrePostFE/input_files/T3/PrePostFE_Ex306.m
The ANSYST M command sequence for this example is listed below. You can either type these commands on the command window, or you can type them on a le, then, on the command window enter /input, le, ext. The les nodes.txt and wind.txt previously generated using the MatLabT M code should be placed in the working directory.
FINISH !close all revious modules /CLEAR !clear all previous models /TITLE,Moebius Sculpture /PREP7 ET,1,BEAM188 KEYOPT,1,3,3 SECTYPE,1,BEAM,CSOLID SECOFFSET,CENT SECDATA,40 SECTYPE,2,BEAM,RECT SECOFFSET,CENT SECDATA,60,10 SECTYPE,3,BEAM,RECT SECOFFSET,CENT SECDATA,130,25 MP,EX,1,210000 MP,PRXY,1,0.29 MP,DENS,1,8e-6 /INPUT,nodes,txt SECNUM,1 E,101,102 EGEN,47,1,1 E,148,201 E,201,202 EGEN,47,1,49 E,248,101 SECNUM,2 E,101,1,248 E,1,201,148 E,102,2,101 E,2,202,201 EGEN,47,1,99,100 ESEL,S,,,167,172 EMODIF,ALL,SECNUM,3 !start PRE-PROCESSOR MODULE !element type #1: BEAM188 (6 DoF) !element #1 keyoption 3=3 (cubic shape function) !section #1: preimetral beams, solid cylindrical !centered section (offset=0) !section radius !section #2: internal beams, rectangular !centered section (offset=0) !section width and height !section #3: bottom internal beams, rectangular !centered section (offset=0) !section width and height !material #1: E=210000 MPa !material #1: 0.29 Poisson !material #1: density=8 T/m^3 !input nodes from file: nodes.txt !section set #1 !elements to generate the perimetral circles
MME
34
ALLSEL FINISH /SOLU D,236,ALL D,237,ALL D,238,ALL ACEL,,,10 /PBC,ALL SOLVE FINISH !select everything !finish PRE-PROCESSOR MODULE !start SOLUTION MODULE !fix sculpture to floor
!gravity load
3.3
Solid Modeling
The mesh may also be created starting from geometric entities (which can be created in the pre-processor or imported from other CAD software) and the software creates the mesh. The user, however, may impose some restrictions or conditions to the meshing process. Some of this restrictions are:
Giving some values for the element size, in terms on edge length.
Starting the mesh from lower order geometric entities and extrude or sweep the created elements to ll higher order geometries. For instance, to mesh a regular volume some of the areas may be meshed rst and then mesh the whole model following the pattern imposed in the areas.
MME
35
Hint 6: Useful ANSYS commands for solid modelling Command K L A AL V VA CIRCLE LARC AROTAT CYL4 SPH4 KBETW AADD VADD LINV BOPT VSWEEP VDRAG Main Parameters npt,x,y,z p1,p2,ndiv p1,p2,p3,...,p18 l1,l2,l3,...,l10 p1,p2,p3,..., na1,na2,na3,...,na10 pcent,rad,... p1,p2,pc,rad nl1,...,nl6, pax1,pax2,arc,nseg xc,yc,rad1,theta1,rad2, theta2,depth xc,yc,rad1,rad2 kp1,kp2,kpnew,... na1,na2,...,na9 nv1,nv2,...,nv9 nl,nv (See ANSYS Help) vnum,srca,trga,lsmo vdrag,na1,na2,...,na6, nlp1,...,nlp6
Dene individual keypoint Generate line from keypoints Generate area from keypoints Generate area from lines Generate volume from keypoints Generate volume from areas Generate circular arc lines Generate circular arc lines Generate surface by rotating lines Generate cylindrical volume Generate sphere Create keypoint between two keypoints Add areas to create single area Add volumes to create single volume Intersect volume with line Boolean operation options Mesh a volume by sweeping a meshed area Mesh a volume by sweeping a meshed area along a path
Hint 7: Useful ANSYS commands which affect the element size Command ESIZE LESIZE AESIZE Main Parameters size nline,size,ndiv, elem1,elem2,einc
Control de general size of elements Control element size on lines Control element size on areas
MME
36 Hint 8: Useful ANSYS commands to mesh entities Command LMESH AMESH VMESH LCLEAR ACLEAR VCLEAR MSHKEY
Mesh lines Mesh areas Mesh volumes Unmesh lines Unmesh areas Unmesh volumes Mesh type
EREFINE MSHAPE
Main Parameters nline1,nline2,nlinc narea1,narea2,nainc nvolu1,nvolu2,nvinc nline1,nline2,nlinc narea1,narea2,nainc nvolu1,nvolu2,nvinc key: 0 - free mesh, 1 - mapped mesh, 2 - mapped mesh if possible ne1,ne2,nicn,level,depth,... key,dimension
Example 3.7. Mesh the geometry considered in Example 3.3 using solid generation and meshing tools. Solution to Example 3.7. The ANSYST M command sequence for this example is listed below. You can either type these commands on the command window, or you can type them on a le, then, on the command window enter /input, le, ext.
FINISH !close all previous modules /CLEAR !clear all previous models /TITLE, Curved Beam - Solid Modeling /PREP7 ET,1,PLANE182 KEYOPT,1,1,3 KEYOPT,1,3,3 R,1,4 MP,EX,1,190e3 !start PRE-PROCESSOR MODULE !element type #1: PLANE182 (2 DoF) !element type #1: keyoption 1=3 (simplified enhanced strain formulation) !element type #1: keyoption 3=3 (plane stress w/thickness) !real constant set #1: th=4 mm !material #1: E=190000 MPa, if Poisson ratio is not defined 0.3 is assumed
PCIRC,25,(25+24),90,180 !define area #1: circle inner radius R=25 mm outer radius R=49 mm RECTNG,0,30,25,(25+24) !define area #2: rectangle 30 mm x 24 mm AGLUE,ALL !glue both areas LESIZE,ALL,,,8 AMESH,ALL FINISH !8 divisions per line !mesh areas !finish PRE-PROCESSOR MODULE
Example 3.8. Model a wing section using solid generation and meshing tools. Solution to Example 3.8. The ANSYST M command sequence for this example is listed below. You can either type these commands on the command window, or you can type them on a le, then, on the command window enter /input, le, ext.
FINISH !close all previous modules /CLEAR !clear all previous models /TITLE, Model extruded wing profile /PREP7 !start PRE-PROCESSOR MODULE
MME
37
ET,1,SHELL181 KEYOPT,1,3,2 SECTYPE,1,SHELL SECDATA,50 MP,EX,1,69000 MP,PRXY,1,0.3 MP,DENS,1,2.7e-3 K,1,0,0,0 K,2,2000,0,0 K,3,2300,200,0 K,4,1900,450,0 K,5,1000,250,0
!element type #1: SHELL181 (6 DoF) !element #1: keyoption 3=2 (full integration incompatible modes) !section #1 = shell !thickness !material #1: E=69000 MPa !material #1: Poisson coefficient=0.3 !material #1: density=2.7e3 kg/m3 !define keypoint #1
LSTR,1,2 !define straight line between kp #1 and #2 LSTR,5,1 !defins straight line between kp #5 and #1 BSPLIN,2,3,4,5,,,-1,0,0,-1,-0.25,0 !define spline line using point #2 to #5 AL,1,2,3 VOFFST,1,5000 ESIZE,100 AMESH,3,5,1 FINISH !define area #1 from lines #1, #2, and #3 !extrude area #1 to define volume, offset distance 5 m !element size 0.1 m !mesh areas from #3 to #5 !finish PRE-PROCESSOR MODULE
Example 3.9. Model the metallic plate part of a robot arm shown in 3.11.
Figure 3.11: Metallic plate as a component of a robot arm (dimensions in mm). Solution to Example 3.9. The ANSYST M command sequence for this example is listed below. You can either type these commands on the command window, or you can type them on a le, then, on the command window enter /input, le, ext.
FINISH !close all previous modules /CLEAR !clear all previous models /TITLE, Model robot arm part (Mesh method A) /PREP7 ET,1,SHELL181 KEYOPT,1,3,2 SECTYPE,1,SHELL SECDATA,6 MP,EX,1,75000 MP,PRXY,1,0.3 !start PRE-PROCESSOR MODULE !element type #1: SHELL181 (6 DoF) !element #1: keyoption 3=2 (full integration incompatible modes) !section #1 = shell !thickness !material #1: E=75000 MPa !material #1: Poisson coefficient=0.3
PCIRC,20,55,90,270 !generate semicircle Ri,Ro,ang1,ang2 LOCAL,11,1,320 !define new cartesian CSYS #11 WPCSYS,1,11 !put WP on CSYS #11 PCIRC,16,40,-90,90 !generate semicircle WPCSYS,1,0 !put WP on CSYS #0 PTXY,0,-55,320,-40,320,40,0,55 !define coordinate pairs x,y for polygon
MME
38
POLY PCIRC,20 WPCSYS,1,11 PCIRC,16 ASBA,3,4 ASBA,6,5 AGLUE,ALL ESIZE,6 AMESH,ALL FINISH
!define new polygon (area #3) with previous coordinates !generate circle !put WP on CSYS 11 !generate circle !substract area #4 to area #3 !substract area #5 to area #6 !glue areas (merge coincident kp and lines) !defines size of element !mesh areas !finish PRE-PROCESSOR MODULE
This other input le improves the control of the mesh for the modelled metallic plate enhancing its nal quality.
FINISH !close all previous modules /CLEAR !clear all previous models /TITLE, Model robot arm part (Mesh method B) /PREP7 ET,1,SHELL181 KEYOPT,1,3,2 SECTYPE,1,SHELL SECDATA,6 MP,EX,1,75000 MP,PRXY,1,0.3 !start PRE-PROCESSOR MODULE !element type #1: SHELL181 (6 DoF) !element #1: keyoption 3=2 (full integration incompatible modes) !section #1 = shell !thickness !material #1: E=75000 MPa !material #1: Poisson coefficient=0.3
PCIRC,20,55,90,270 !generate semicircle Ri,Ro,ang1,ang2 LOCAL,11,1,320 !define new cartesian CSYS #11 WPCSYS,1,11 !put WP on CSYS #11 PCIRC,16,40,-90,90 !generate semicircle WPCSYS,1,0 !put WP on CSYS #0 PTXY,0,0,320,0,320,40,0,55 !define coordinate pairs x,y for polygon POLY !define new polygon (area #3) with previous coordinates PTXY,0,-55,320,-40,320,0,0,0 !define coordinate pairs x,y for polygon POLY !define new polygon (area #4) with previous coordinates PCIRC,20 WPCSYS,1,11 PCIRC,16 WPCSYS,1,0 ASBA,3,5,,,KEEP ASBA,7,6,,,KEEP ASBA,4,5 ASBA,7,6 !generate circle !put WP on CSYS 11 !generate circle !substract !substract !substract !substract area area area area #5 #6 #5 #6 to to to to area area area area #3 #7 #4 #7
BLC4,55,-55,(320-95),110 !define rectangle (area #5) to define line #19 and #16 ASBL,4,19 !divide area #4 by line #19 ASBL,7,16 !divide area #7 by line #16 ASBL,3,19 !divide area #3 by line #19 ASBL,9,16 !divide area #9 by line #16 ADELE,5,,,1 AGLUE,ALL LESIZE,ALL,,,6 LESIZE,1,,,24,,1 LESIZE,3,,,24,,1 LESIZE,5,,,24,,1 LESIZE,7,,,24,,1 LESIZE,12,,,24,,1 LESIZE,13,,,24,,1 LESIZE,33,,,24,,1 LESIZE,16,,,12,,1 !delete area #5 and bellow entities !glue areas (merge coincident kp and lines) !define number of divisions in lines !modify number of division in line
MME
39
LESIZE,19,,,12,,1 LESIZE,35,,,12,,1 LESIZE,37,,,12,,1 !LCCAT,18,28 !LCCAT,14,25 !LCCAT,11,32 !LCCAT,29,36 AMESH,ALL FINISH !concatenate lines to improve mesh (uncomment it to test)
Example 3.10. Model a handle shown in the Figure 3.12 using the solid generation and meshing tools.
Solution to Example 3.10. The ANSYST M command sequence for this example is listed below. You can either type these commands on the command window, or you can type them on a le, then, on the command window enter /input, le, ext.
FINISH !close all previous modules /CLEAR !clear all previous models /TITLE, Static model for a handle /PREP7 ET,1,SOLID186 KEYOPT,1,2,1 MP,EX,1,210000 MP,PRXY,1,0.29 !start PRE-PROCESSOR MODULE !element type #1: SOLID186 (3 DoF) !element #1: keyoption 2=1 (full integration) !material #1: E=210000 MPa !material #1: Poisson=0.29
LOCAL,11,,,,,,-18,-22.5 !define local cartesian system to define central part CSYS,11 !active previous defined CSYS K,1,10,-5 K,2,10,5 K,3,-10,5 K,4,-10,-5 K,5,7.5,-5,153 K,6,7.5,5,153 K,7,-7.5,5,153 K,8,-7.5,-5,153 V,1,2,3,4,5,6,7,8 !define central part
!define volume #1 for central part !change !define !change !define WP position cylindre volume #2 WP position cylindre volume #3
MME
40
VSBV,1,2 VSBV,4,3
!substract volume #2 to volume #1 (volume #1 becomes #4) !substract volume #3 to volume #4 (volume #4 becomes #1) !change WP position !redefine cylindre volume #2 !change WP position !redefine cylindre volume #3
WPLANE,1,,,,1,,,1,,-1 CYLIND,15,30,,20 WPLANE,1,-60,,140,,,140,,,120 CYLIND,7.5,15,35,50 VGLUE,ALL ESIZE,6 MSHKEY,0 MSHAPE,1,3D VMESH,ALL FINISH
!glue volumes !define size of elements !not mapped mesh !mesh with tetrahedral elements !mesh all volumes !finish PRE-PROCESSOR MODULE
3.4
To be condent on the results of a nite element modelization of a mechanical or structural problem, it is necessary to assess the mesh quality, that is, to check if the mesh is ne enough to get good simulation results. Basically, the following aspects need to be analyzed: Assure the right connectivity between different parts of the mesh. Avoid elements with high distortions, warping and inadmissible narrow or wide angles between edges. Error estimation of the elements, typically by means of energy hypotheses (see Chapter 6). Study the sensitivity of the results on different meshes of the entire FE model or of a partial model (see Chapter 6).
3.4.1
Connectivity
The connectivity of the different parts of the mesh should be veried. The mesh edges between different parts of the geometry must assure that the generated elements share nodes. The results of a bad connectivity denition are presented in Figure 3.13, while the correct solution is presented in Figure 3.14. Example 3.11. Write the input les for the simulation of the geometry shown in Figures 3.13 and 3.14 with a bad and a good connectivity of the elements.
Figure 3.13: Example of bad connectivity of different parts of the mesh. Undeformed mesh (left) and deformed mesh (right).
MME
41
Figure 3.14: Example of right connectivity of different parts of the mesh. Undeformed mesh (left) and deformed mesh (right). Solution to Example 3.11. The ANSYST M command sequence for this example is listed below. You can either type these commands on the command window, or you can type them on a le, then, on the command window enter /input, le, ext.
FINISH !close all previous modules /CLEAR !clear all previous models /TITLE, Connectivity (Bad connectivity) /PREP7 ET,1,PLANE182 KEYOPT,1,1,3 MP,EX,1,1 MP,PRXY,1,0.2 K,1,0,0,0 K,2,5,0,0 K,3,5,1,0 K,4,0,1,0 K,5,10,0,0 K,6,10,5,0 K,7,5,5,0 A,1,2,3,4 A,2,5,6,7 AESIZE,2,.25 AESIZE,1,.5 AMESH,ALL FINISH /SOLU DK,1,ALL DK,4,ALL DK,5,UX,.1 DK,6,UX,.1 SOLVE FINISH /POST1 PLDISP,1 !start PRE-PROCESSOR MODULE !element type #1: PLANE182 (2 DoF) !element type #1: keyoption 1=3 (simplified enhanced strain formulation) !material #1: E=1 (no units) !material #1: Poisson=0.2 !define geometry: keypoints and areas
!solve current load state !finish SOLUTION MODULE !start POST-PROCESSOR MODULE
FINISH !close all previous modules /CLEAR !clear all previous models /TITLE, Connectivity (Good connectivity) /PREP7 !start PRE-PROCESSOR MODULE
MME
42
ET,1,PLANE182 KEYOPT,1,1,3 MP,EX,1,1 MP,PRXY,1,0.2 K,1,0,0,0 K,2,5,0,0 K,3,5,1,0 K,4,0,1,0 K,5,10,0,0 K,6,10,5,0 K,7,5,5,0 A,1,2,3,4 A,2,5,6,7 AGLUE,1,2 AESIZE,3,.5 AESIZE,1,.25 MSHKEY,2 AMESH,ALL FINISH /SOLU DK,1,ALL DK,4,ALL DK,5,UX,.1 DK,6,UX,.1 SOLVE FINISH /POST1 PLDISP,1 !element type !element type !material #1: !material #1:
#1: PLANE182 (2 DoF) #1: keyoption 1=3 (simplified enhanced strain formulation) E=1 (no units) Poisson=0.2
!solve current load state !finish SOLUTION MODULE !start POST-PROCESSOR MODULE
3.4.2
Avoid elements with high distortions, warping and inadmissible narrow or wide angles between edges.
Elements with high distortions, bad aspect ratio, warping and inadmissible narrow, or wide angles between the element edges can lead to numerical errors. In ANSYST M , all continuum elements (2-D and 3-D solids, 3-D shells) are tested for acceptable shape. The testing, is performed by computing shape parameters (such as Jacobian ratio) which are functions of geometry, then comparing them to element shape limits whose default values are functions of element type and settings. These limits depend on the element type and formulations, therefore it is recommended to check the guidelines given by the nite element code being used. Some example of elements with bad aspect ratio, warping and wide angles between element edges are shown in Figure 3.15. ANSYST M allows to plot the elements that have some of these element geometry checks over the limits. Main Menu>Preprocessor>Meshing>Check Mesh> Individual Elm>Select Warning/Error Elements Example 3.12. Check for elements with high distortions in the geometry and mesh shown in Figure 3.16. Solution to Example 3.12. The ANSYST M command sequence for this example is listed below. You can either type these commands on the command window, or you can type them on a le,
MME
43
Figure 3.15: Example of elements with bad aspect ratio, warping and distortion.
Figure 3.16: Example of right elements with wide corner angles. Undeformed mesh (left) and "warning" elements of the mesh (right). then, on the command window enter /input, le, ext.
FINISH !close all previous modules /CLEAR !clear all previous models /TITLE, Distortion elements /PREP7 ET,1,PLANE182 KEYOPT,1,1,3 MP,EX,1,1 MP,PRXY,1,0.2 K,1,0,0,0 K,2,5,0,0 K,3,5,1,0 K,4,0,1,0 K,5,6,0,0 K,6,6,5,0 K,7,5,1,0 A,1,2,3,4 A,2,5,6,7 AGLUE,1,2 AESIZE,3,.5 AESIZE,1,.25 MSHKEY,1 AMESH,ALL FINISH !start PRE-PROCESSOR MODULE !element type #1: PLANE182 (2 DoF) !element type #1: keyoption 1=3 (simplified enhanced strain formulation) !material #1: E=1 (no units) !material #1: Poisson=0.2 !define geometry: keypoints and areas
MME
44
3.5
In order to simulate or model different geometries or problems with nite elements it is often useful to import the geometry or the model generated by means of a CAD-system or more powerful and user-friendly Pre-process utility. In this section, some techniques to import models and geometries into a FE package are shortly described. There are different le standards that allow us to translate geometries and nite element entities between different packages. Here, in the following example, the IGES-format is used to import a geometry from a CAD software. Example 3.13. Figure 3.17 shows the schema of an arm of a grip designed to pick and carry eggs. The grip has a total of three of these arms so the eggs are clamped in between. The mechanism is activated by applying a contact load on the normal to the right-top circular area of the arm. The arm can rotate around its axis and the egg is clamped by the bottom-planar areas of the grip-arms. Model the grip-arm assuming plane-stress conditions, considering that it is made of aluminium (E = 71000 MPa and = 0.33) and the rotation axis is made of steel. It is considered that the maximum load sustained by an egg is 5 N.
14.33 10
R1 5
25
15.95
R30.32 6 5 8 40
50
25
Solution to Example 3.13. The geometry of the arm-grip has been already generated using a CAD program and can be found at ftp://amade.udg.edu/mme/PrePostFE/grip_arm.igs. The ANSYST M command sequence for this example is listed below. You can either type these commands on the command window, or you can type them on a le, then, on the command window enter /input, le, ext. Before executing the script make sure that the IGES geometry of the model is in the working directory. You can also import the geometry using the Menu commands
MME
17.79
R2
R2.5
45
and mesh the geometry avoiding the importing commands in the script.
FINISH !close all previous modules /CLEAR !clear all previous models /TITLE, Robot grip part (Model 2D plane stress) /AUX15 !import IGES file IOPTN,IGES,NODEFEAT !select options IOPTN,MERGE,YES IOPTN,SOLID,YES IOPTN,SMALL,YES IOPTN,GTOLER, DEFA IGESIN,grip_arm,igs, /PREP7 ET,1,PLANE182 KEYOPT,1,1,3 KEYOPT,1,3,3 R,1,5 ET,2,BEAM188 KEYOPT,2,3,3 SECTYPE,2,BEAM,RECT SECOFFSET,CENT SECDATA,1e9,1e9 MP,EX,1,71000 MP,PRXY,1,0.33 MP,EX,2,210000 MP,PRXY,2,0.3 !start PRE-PROCESSOR MODULE !element type #1: PLANE182 (2 DoF) !element type #1: keyoption 1=3 (simplified enhanced strain formulation) !element type #1: keyoption 3=3 (plane stress w/thickness) !real constant #1: thickness=5 mm !element type #2: BEAM188 (6 DoF) !element #2 keyoption 3=3 (cubic shape function) !section #1 = rectangular beam !centered section (offset=0) !section width and height (ultra rigid beam) !material !material !material !material #1: #1: #2: #2: E=71000 MPa Poisson=0.33 E=210000 MPa Poisson=0.3 !move lines coordinates origin
LGEN,,ALL,,,-1747.394,-416.3893,0,,,1 LSEL,S,,,1,11 LSEL,U,,,3,4 AL,ALL LSEL,S,,,3,4 AL,ALL ASBA,1,2 N,,0,0,0 LESIZE,3,,,10 LESIZE,4,,,10 AESIZE,ALL,2 MSHKEY,2 AMESH,ALL TYPE,2 MAT,2 SECNUM,2 REAL,2 E,1,178 *REPEAT,20,0,1 FINISH /SOLU D,1,ALL DDELE,1,ROTZ D,99,UX,,,,,UY F,172,FX,-4.532 F,172,FY,2.113 SOLVE FINISH /POST1 PLNSOL,S,EQV,2
!select lines !unselect circle lines !generate global area with all selected lines !select circle lines !generate circle !substract circle to global area !generate node on 0,0,0 (origin) !divisions line 3 !divisions line 4 !element size a 2 mm !try quadrangualar elements !mesh all areas !set element #2 !set material #2 !set section #2 !set real #2 !generate element between nodes #1 and #178 !repaeat previous command 19 times increment +1 second argument !finish PRE-PROCESSOR MODULE !start SOLUTION MODULE !fix articulated node !allow rotation articulated node !fix horizontal and vertical displacement contact node !apply horitzontal load node #178 !apply vertical load node #178 !solve current load state !finish SOLUTION MODULE !start POST-PROCESSOR MODULE !plot Von Mises stress + undeformed edge
MME
46
3.6
Suggested problems
Problem 3.1. Reconsider the situation in Example 3.3. Using direct generation commands, remesh the model using twice the number of divisions that were used in the example solution. Problem 3.2. Reconsider the Example 3.3. Prepare a mesh for sensitivity analysis for the following meshes:
1. The mesh in the solution of the example 2. A free mesh using triangles 3. Two different quadrilateral mapped meshes with different mesh densities. Show the obtained meshes in a report and comment on the results.
Problem 3.3. Obtain the geometry of the part presented in Example 3.9. Use a CAD program to generate this obtained geometry and its IGES le. Export the geometry to ANSYS and mesh it. Try to obtain a regular mesh. Show all the process in a report. Problem 3.4. Model the U-shaped beam shown in Figure 3.18. Using a similar procedure as in Ex. 3.6 generate the MATLAB code and the ANSYST M command sequence to dene the mesh. The codes should dene the mesh in function of the geometric parameters shown in the gure.
MME
Chapter 4
Boundary conditions
4.1 Introduction
The Finite Element Method may be applied to problems of very different nature which are affected by different physical variables. As you may know, in the structural case, after computing the stiffness matrix for each element, the nite element method solves a typical system of equations:
K =F
(4.1)
where K is the stiffness matrix, is the degree of freedom vector and F is the vector containing the external actions. Some of the components in and F will be unknown variables -and so, the solution of our problem- and some of them will be the boundary conditions of the model. A known variable can set a xed value to one nodal degree of freedom (DOF) (displacements and rotations). This type of boundary conditions are usually called Dirichlet boundary conditions or essential boundary conditions. On the other hand, a known variable can dene a nodal force or ux, now the known variable denes a natural boundary condition (also called Newman boundary condition). Focusing on the Engineering Mechanics realm, the typical problems may be divided into different physics problems as is shown in the Table 4.1.
Table 4.1: Physics Problems in FEM Problem Structural DOFs displacements and rotations temperatures Force or ux variables forces, pressures, temperatures (for thermal strain), gravity heat ow rates, convection, internal heat generation, innite surface magnetic ux, magnetic current segments, source current density, innite surface electric current, electric charges, charge densities, innite surface pressures
Thermal
Magnetic
magnetic tentials
po-
Electric
Fluid
47
48
For the moment, we will focus on structural problems. Thermal and structural-thermal analysis will be seen further during this course. The boundary conditions in structural models include externally applied forcing loads and constrained displacements. Loads are divided into different categories: DOF constraints: constrained displacements and rotations Concentrated loads: forces and moments Distributed loads: loads on edges loads on surfaces body loads inertia loads (due to accelerations or gravity) Coupled-eld loads: uid-structure interaction, thermal dilatations, etc. Temperatures which structures are expected to work at.
Hint 9: Loads on geometry vs loads on mesh The major commercial codes of FE allow to impose boundary conditions to geometric entities (points, lines, surfaces, or volumes) or to mesh entities (nodes and elements). Pros Independent of mesh. No troubles in re-meshing Generally easier to apply (less entities needed to select) Loads on affected nodes directly (more control) Cons Possible conicts local-global coordinate systems Less control on what you are doing Re-meshing involves reapplying loads Generally lots of nodes/elements involved (graphical picking may not be convenient)
Loads on Geometric Entities (points, lines, surfaces...) Loads on Mesh Entities (nodes, elements)
Load step A load step or load case is simply a conguration of loads for which a solution is obtained. Therefore, a load step in structural analysis is dened by the set of forces and moments, pressures, body and inertia loads (as gravity), and specied displacements and rotations (constrained DOF), all applied to the model. In a linear static or steady-state analysis, you can use different load steps to apply different sets of loads -wind load in the rst load step, gravity load in the second load step, both loads and a different support condition in the third load step, and so on. In a transient analysis, multiple load steps apply different segments of the load history curve. The next sections show how different kinds of loads may be applied to a structural FE model.
MME
49
Boundary conditions in two- and three-dimensional problems In two-dimensional problems, the following boundary conditions can be applied: Constrained displacements in X and Y axes, also rotations into the Z axis in exural beam elements. Concentrated forces (in X and Y directions), also moment in exural beam elements (applied into the Z axis). Distributed loads on edges and lines. Additionally, in three-dimensional problems the following boundary conditions can be applied: Imposed displacements (and rotations in shell and beam elements) to all degrees of freedom. Concentrated forces (and moments in shell and beam elements) in any direction or moments in any axis. Distributed loads on edges and surfaces. Body and inertial loads.
Hint 10: DOF notation in ANSYS UX UY UZ Displacement in X direction Displacement in Y direction Displacement in Z direction ROTX ROTY ROTZ Rotation into X axis Rotation into Y axis Rotation into Z axis
Hint 11: Useful ANSYS commands to apply boundary conditions On Mesh D on nodes F SF on nodes, SFE on elements On Geometry DK on keypoints, DL on lines FK on keypoints SFL on lines SFA on areas
4.2
Loads can be applied on nodes by means of concentrated forces or moments, as it is shown in Example 4.1. Also loads can be distributed over the elements as: surface loads, body loads, inertia loads or other coupled-eld loads (for example, thermal strains). A Surface load is a distributed load applied over a surface, for example a pressure caused by wind or any uid. A body load is a volumetric load, for example expansion of material by temperature raise in structural analysis. Inertia loads are those attributable to the inertia (mass
MME
50
matrix) of a body, such as gravitational acceleration, angular velocity, and acceleration. When a concentrated load is applied on a node in the FE model, it can be directly added to the force vector. However, if a distributed load is applied, the element shape functions are used by the nite element program to compute the equivalent forces vector. Hint 12: Avoiding rigid solid modes When modelling real life situations the boundary conditions applied to the model should guarantee that the model is not going to move without carrying any deformation. This may be done by setting a zero value to all DOF in the same or different nodes. One should do this without applying unreal boundary conditions which bring to the solution unreal stresses.
Example 4.1. Let us model a small bridge (a footbridge) intended to cross from a pier to a boat or to a small ship (see Fig. 4.1). This footbridge is 3 m long and should bear a load of a man, estimated in 150 kg, standing in its middle point. The section of the bridge is 300 mm 25 mm. The material is wood with E = 15000 MPa and = 0.3. Let us model this problem using beamtype elements. At one tip of the footbridge, the tip on the ship, all displacements and rotations are restricted and on the other one the footbridge is simply supported by the oor.
Figure 4.1: Footbridge for small ship. Solution to Example 4.1. The ANSYST M command sequence for this example is listed below. You can either type these commands on the command window, or you can type them on a le, then, on the command window enter /input, le, ext.
FINISH !close all previous modules /CLEAR !clear all previous models /TITLE, Footbridge 2D with beam elements /PREP7 !start PRE-PROCESSOR MODULE !define geometric parameters (mm) length=3000 width=300 height=25
!element definition ET,1,BEAM188 KEYOPT,1,3,3 SECTYPE,1,BEAM,RECT SECOFFSET,CENT SECDATA,height,width MP,EX,1,15000 MP,NUXY,1,0.3 !material !define keypoints
!element type #1: BEAM188 (6 DoF) !element #1 keyoption 3=3 (cubic shape function) !section #1 = rectangular beam !centered section (offset=0) !section width and height !material property #1, Youngs modulus property #1, Poisson modulus
MME
51
!keypoint,#,x,y
K,1,0,0 K,2,length/2 K,3,length !define lines L,1,2,6 L,2,3,6 LMESH,all FINISH /SOLU !apply constraints DK,1,ALL,0 DK,3,UY,0 !apply loads FK,2,FY,-1500 SOLVE FINISH /POST1 PRRSOL,F PLDISP,2 PLNSOL,U,SUM,0,1
!start POST-PROCESSOR MODULE !list reaction forces !plot deformed shape !contour plot of deflection
ETABLE,TAI,SMISC,6 !Shear diagram at node i ETABLE,TAJ,SMISC,19 !Shear diagram at node j PLLS,TAI,TAJ,1,0 !Plot the shear diagram PRRSOL,M !list reaction moments ETABLE,MOI,SMISC,3 !Bending moment at node i ETABLE,MOJ,SMISC,16 !Bending moment at node j PLLS,MOI,MOJ,1,0 !Plot the bending moment diagram /ESHAPE,1,1 PLNSOL,S,X This le can be found at: ftp://amade.udg.edu/mms/PrePostFE/input_files/T4/PrePostFE_Ex401.dat
Observe that the left end of the deformed footbridge remains horizontal due to the applied constraints while the right end rotates upwards. Example 4.2. We want to dimension a roof. The loads to consider are the weight of the roof and the load due to snow. The roof is 0.9m 2.1m with a 10 mm thickness and it is settled horizontally and supported at its four vertex. The snow load over the roof is 400 N/m2 . Model the roof and obtain the deformed shape plot. Consider carefully which are the boundary conditions to apply. The roof is from a reinforced polymer with a Youngs modulus of 15 GPa, a Poissons coefcient of 0.3, and a mass density of 40 kg/m3 . Solution to Example 4.2. The ANSYST M command sequence for this example is listed below. You can either type these commands on the command window, or you can type them on a le, then, on the command window enter /input, le, ext.
MME
52
length=900 width=2100 thickness=10
ET,1,SHELL181 KEYOPT,1,3,2 R,1,thickness MP,EX,1,15000 MP,PRXY,1,0.3 MP,DENS,1,40e-9 !define keypoints K,1,0,0 K,2,width,0 K,3,width,length K,4,0,length !define area A,1,2,3,4 !element definition AESIZE,ALL,100 AMESH,ALL FINISH
!element type #1: SHELL181 (6 DoF) !element #1: keyoption 3=2 (full integration incompatible modes) !real constant #1: th=10 mm !Youngs modulus for material #1 in N/mm2 !Poisson ratio for material #1 !material property #1: Density in kg/mm^3
!keypoint, #, x, y
!specifies the element size to be meshed onto areas !mesh all areas
/SOLU !start SOLUTION MODULE !apply boundary conditions DK,1,UZ,0,,,UX,UY !define a DOF constraint at a keypoint (ux=uy=uz=0) DK,2,UZ,0,,,UY !define a DOF constraint at a keypoint (uy=uz=0) DK,3,UZ,0 DK,4,UZ,0 !apply loads ACEL,,,9.81 LSWRITE,1 SFA,1,1,PRES,-0.4e-3 LSWRITE,2 LSSOLVE,1,2 FINISH /POST1 SET,1 PLNSOL,U,SUM,0,1 SET,2 PLNSOL,U,SUM,0,1 !start POST-PROCESSOR MODULE !first load case !contour plot of deflection !second load case !contour plot of deflection
!acceleration in m/s^2, kg*m/s^2 = N !write the first load case !define snow load over surface; 0.4 kN/m^2 = 0.4e-3 N/mm^2 !write the second load case !solve the resulting system of equations, of load cases #1 and #2
Note that in this case the maximum displacement of the roof caused by its own weight is very small. However, when the weight of the snow is added, the displacement of the roof is increased considerably. Example 4.3. Use a commercial FE code to model the effect of the distributed force q = 5 N/mm2 on the two types of short cantilever beams shown in Fig.4.2. Consider that the two beams are made of steel with a thickness of 5 mm and take L1 = 80 mm, L2 = 40 mm, B = 150 mm and Q = 160 mm. Use eight-noded (quadratic) quadrilateral plane elements. Solution to Example 4.3. The ANSYST M command sequence for this example is listed below. You can either type these commands on the command window, or you can type them on a le, then, on the command window enter /input, le, ext. Next we present two different solutions using different approaches to implement the clamped side.
MME
53
q B Q Q (a) (b)
Figure 4.2: Short cantilever beams
FINISH !close all previous modules /CLEAR !clear all previous models /TITLE, Distributed load on short cantilever: direct clamped /PREP7 !start PRE-PROCESSOR MODULE !Geometric parameters (mm) L1=80 !clamped length L2=40 !short side length B=150 !distributed load span th=5 !thickness ET,1,PLANE183 KEYOPT,1,3,3 MP,EX,1,210000 MP,PRXY,1,0.3 R,1,th K,1,0,0 K,2,0,-L1 K,3,B,0 K,4,B,-L2 L,1,2,10 L,1,3,15 L,2,4,15 L,3,4,10 AL,1,2,3,4 AMESH,ALL FINISH /SOLU DL,1,,ALL SFL,2,PRES,5 SOLVE FINISH /POST1 PLDISP,1 PLNSOL,S,EQV PLVECT,S !element type !plane stress with thickness !Youngs modulus for material #1 !Poisson ratio for material #1 !real constant #1, thickness for element #1 !geometry: keypoints
q B
!start SOLUTION MODULE !campled line #1 !load on line #2 !solve the current load state
!start POST-PROCESSOR MODULE !deformed shape !contour plot Von Mises equivalent stress !vector plot principal stress
FINISH !close all previous modules /CLEAR !clear all previous models /TITLE, Distributed load on short cantilever: not direct clamped /PREP7 !start PRE-PROCESSOR MODULE !Geometric parameters (mm) L1=80 !clamped length L2=40 !short side length B=150 !distributed load span Q=160 !square side th=5 !thickness
MME
54
ET,1,PLANE183 KEYOPT,1,3,3 MP,EX,1,210000 MP,PRXY,1,0.3 R,1,th
!element type !plane stress with thickness !Youngs modulus for material #1 !Poisson ratio for material #1 !real constant #1, thickness for element #1
K,1,0,0 !geometry: keypoints K,2,0,-L1 K,3,B,0 K,4,B,-L2 K,5,0,((Q-L1)/2) K,6,0,-(L1+((Q-L1)/2)) K,7,-Q,((Q-L1)/2) K,8,-Q,-(L1+((Q-L1)/2)) K,9,-Q,0 K,10,-Q,-L1 L,1,2,10 L,1,3,15 L,3,4,10 L,2,4,15 L,1,5,5 L,5,7,16 L,7,9,5 L,9,1,16 L,9,10,10 L,10,8,5 L,10,2,16 L,8,6,16 L,2,6,5
!line clamped with 5 divisions !line clamped with 10 divisions !line clamped with 5 divisions
!areas definition from lines AL,1,2,3,4 !area 1 AL,5,6,7,8 AL,8,9,11,1 AL,11,10,12,13 AMESH,ALL FINISH /SOLU DL,7,,ALL DL,9,,ALL DL,10,,ALL SFL,2,PRES,5 SOLVE FINISH /POST1 PLDISP,1 PLNSOL,S,EQV PLVECT,S !mesh generation
!start POST-PROCESSOR MODULE !deformed shape !contour plot Von Mises equivalent stress !vector plot principal stress
After analysing the results of both short cantilever beam congurations, it can be concluded that the inclusion of more material does not signicatively affect the maximum values of the achieved stress. Example 4.4. Use plane strain conditions to simulate the cross-section of an hydraulic dam shown in Fig. 4.3. The dam is made of reinforced concrete, E = 27000 MPa and = 0.25, and its left side is full of water up to a height H = 30 m. The other dimensions of the dam are h = 20 m, B = 4 m and b = 2 m. The density of the uid contained by the dam is = 104 N/m3 . Solution to Example 4.4. The ANSYST M command sequence for this example is listed below. You can either type these commands on the command window, or you can type them on a le, then, on the command window enter /input, le, ext.
MME
55
FINISH !close all previous modules /CLEAR !clear all previous models /TITLE, 2-D plane strain dam section /PREP7 !start PRE-PROCESSOR MODULE !Parameters (mm) H=30000 !left height H1=20000 !right height B=4000 !bottom base B1=2000 !top base t=1000 !thickness D=0.00001 !fluid density in N/mm3 p=D*H !max pressure in N/mm2 ET,1,PLANE182 KEYOPT,1,3,2 MP,EX,1,27000 MP,NUXY,1,0.25 K,1,0,0 K,2,B,0 K,3,0,H1 K,4,B1,H1 K,5,0,H K,6,B1,H L,1,2,8 L,1,3,20 L,2,4,20 L,3,4,8 L,3,5,10 L,4,6,10 L,5,6,8 AL,1,2,3,4 AL,4,5,6,7 AMESH,ALL FINISH !element type !keyopt3=2, plane strain !thickness is not used in plane strain !Youngs modulus !Poisson !geometry: keypoints
!areas by lines
!mesh
/SOLU !start SOLUTION MODULE!Boundary conditions DL,1,,ALL !clamped nodes on base !Apply pressure SFGRAD,PRES,0,Y,0,-(p/H) !Gradient on surface load, type pressure, cartesian coordinate system, ! along Y direction, with slope -p/H
MME
56
In this case, the cross-section of the dam acts as a cantilever beam under the effect of a distributed load following a triangular distribution.
4.3
4.3.1
In a general FEA, a node can have more than one DOF. For example, if the FE model uses beam elements in 2-D space we have three DOF: the horizontal displacement, the vertical displacement, and the rotation around perpendicular axis to plane. When we constraint different DOF we dene different kinds of actual boundary conditions. In the 2-D beam element case if we constraint only the horizontal and vertical displacements we get a simple support, if we constraint all the DOF we obtain a clamped condition. Hint 13: ANSYS command to constrain DOF on nodes D, nnode, dir where nnode is the node number, dir is the degree of freedom to impose, that is uy, ux or rotz.
4.3.2
Symmetry conditions
Symmetry conditions can be applied to reduce the size of the model without loss of accuracy. However, to use such conditions the four types of symmetry must be present: geometry, boundary conditions, material and loads. Under this condition the solution will be also symmetric. For example, symmetry with respect to the y-z plane involves that the nodes on the symmetry plane have the following conditions
(4.2)
Symmetric boundary conditions on nodes in the symmetry plane involve the restriction of the out-of-plane DOF translation and the restriction of the in-plane DOF rotations. Example 4.5. Use a commercial FE code to nd the theoretical stress concentration factor in a rectangular notched bar made of aluminium, E = 190 GPa and = 0.3. The dimensions and the load state is dened in Fig. 4.4. Use eight-noded (quadratic) quadrilateral plane elements. Solution to Example 4.5. The ANSYST M command sequence for this example is listed below. You can either type these commands on the command window, or you can type them on a le, then, on the command window enter /input, le, ext.
MME
57
!start SOLUTION MODULE !symmetry conditions in line 1 !symmetry conditions in line 9 !apply pressure on line 2 (tension)
!start POST-PROCESSOR MODULE !contour plot Von Mises equivalent stress, undeformed edge display, scale=1
o =
10 37 4 P = = 14.8 MPa A 25 4
The maximum horizontal stress close to the notch obtained from FE model is 28 MPa. Therefore, the concentration factor is
k=
max = 1.89 o
Axysymmetric conditions can be also used when revolution geometries are simulated. In the following example inertial loads are applied in an axisymmetric model of a ywheel. Example 4.6. Consider the steel ywheel schematized in Figure 4.5. Its inner diameter is 50 mm and its outer diameter is 600 mm. Assume that it is rotating at 2000 rad/s. Write an input le for ANSYST M which models this situation and computes the total inertia of the ywheel.
MME
58
Rcen2
rotation axis
B1/2
Figure 4.5: Quarter section of a ywheel. Solution to Example 4.6. The ANSYST M command sequence for this example is listed below. You can either type these commands on the command window, or you can type them on a le, then, on the command window enter /input, le, ext.
FINISH /CLEAR /TITLE, Flywheel !close all previous modules !clear all previous models
/PREP7 !start PRE-PROCESSOR MODULE !Parameters, this is a dynamic analysis and the international !system units are used to avoid problems Rin=25 !mm Rcen1=Rin*4 Rout=300 Rcen2=250 B1=300 B2=90 B3=25 B4=200 Rr=Rcen2-(25) ET,1,PLANE182,,,1 MP,EX,1,210e3 MP,PRXY,1,0.3 MP,DENS,1,7800e-9 K,1,Rin,0 K,2,Rin,B1/2 K,3,Rcen1,0 K,4,Rcen1,B2/2 K,5,Rcen1,B1/2 K,6,Rr,0 K,7,Rcen2,0 K,8,Rcen2,B3/2 K,9,Rcen2,B4/2 K,10,Rout,0 K,11,Rout,B4/2 L,1,2,10 L,1,3,8 L,2,5,8 L,3,4,6 L,4,5,4 L,4,6,12 L,5,8,16 L,6,7,4 !element type with axysymmetric condition, KEYOPTION3=1 !Young modulus, N/mm^2 !Poisson !Density, kg/mm^3 !geometry: keypoints
B2/2
MME
B3/2
B4/2
59
L,7,8,4 L,8,9,6 L,7,10,8 L,9,11,8 L,10,11,10 AL,1,2,3,4,5 AL,5,6,8,9,7 AL,10,9,11,13,12 LCCAT,4,5 LCCAT,6,8 LCCAT,9,10 AMESH,ALL FINISH /SOLU DK,1,UY !areas by lines
!concatenate lines in one to improve meshing !it helps to obtain continuity in the mesh !and avoid irregularities
OMEGA,,2000 !rotational speed in rad/s SOLVE FINISH /POST1 !start POST-PROCESSOR MODULE PLDISP,1 ETABLE,evol,VOLU !compute the volume of each node, save it with "evol" label ETABLE,ecent,CENT,X !compute the centroidal of each node, save it with "ecent" label SMULT,mass,evol,,7800e-9 !compute the total mass in label "mass" by multiplying 7800*evol SMULT,mx,mass,ecent,1,1 !compute the first moment of inertia in label "mx" by multiplying ! mx=1*mass+1*ecent SMULT,einer,mx,ecent,1,1 !compute element inertia in label "einer" by multiplying ! einer=1*mx+1*ecent SSUM !calculates the summ of every element table item *GET,totalinertia,SSUM,,ITEM,einer !compute total inertia (kg mm^2) *STATUS,totalinertia !List of the parameter "totalinertia" This le can be found at: ftp://amade.udg.edu/mms/PrePostFE/input_files/T4/PrePostFE_Ex406.dat
Observe that although only a quarter section of the ywheel is modelled, the simulation takes into account the whole ywheel and the total inertia is calculated.
4.3.3
Antisymmetry conditions
Antisymmetric conditions are similar to the symmetry conditions. They can be applied when it exists symmetry of geometry, symmetry of boundary conditions, symmetry of material and antisymmetry of loads. Antisymmetric boundary conditions involve the restriction of the in-plane DOF translations of the nodes in the antisymmetry plane and the restriction of the out-of-plane DOF rotations. Hint 14: Symmetry conditions vs Antisymmetry conditions The following table shows which are the DOF to restrict in function of the symmetry plane normals. Symmetry 2D UX, ROTZ UY, ROTZ Antisymmetry 2D UY UX
Normal X Y Z
MME
60
4.4
Superposition
The superposition principle states that linear problems can be decomposed as the addition of several sub-problems. In large models involving symmetric geometry but un-symmetric loads the superposition can be applied by decomposing the problem as the addition of a symmetric problem and an antisymmetric one. Example 4.7. Use a commercial FE code to nd analyze the beam structure of Figure 4.6 made of thin-walled circular tubes with an outer diameter of 20 mm and 4 mm thick. Use two-node beam elements. Decompose the problem in a symmetric part and an antisymmetric part and obtain the total displacements.
Figure 4.6: Geometrically symmetric unsymmetrically loaded beam structure Solution to Example 4.7. The ANSYST M command sequence for this example is listed below. You can either type these commands on the command window, or you can type them on a le, then, on the command window enter /input, le, ext.
FINISH !close all previous modules /CLEAR !clear all previous models /TITLE, Superposition method of thin walled beam structure /PREP7 !start PRE-PROCESSOR MODULE !define parameters (mm) length=3000 height=1000 Ri=6 !Inner radius of the tube Ro=10 !Outer radius of the tube
ET,1,BEAM188 !element type #1: 3D beam (6 DoF) KEYOPT,1,3,3 !element #1 keyoption 3=3 (cubic shape function) SECTYPE,1,BEAM,CTUBE !section #1 = circular tube SECOFFSET,CENT !centered section (offset=0) SECDATA,Ri,Ro !inner radius, outer radius MP,EX,1,210000 !Youngs modulus for material #1 MP,PRXY,1,0.25 !Poisson ratio for material #1 K,1,0,0 K,2,length/4,0 K,3,length/2,0 K,7,length/2,height K,8,length/4,height/2 L,1,2,20 L,2,3,20 L,3,7,20 L,3,8,20 L,2,8,20 L,1,8,20 !geometry: keypoints
MME
61
L,8,7,20 LMESH,ALL FINISH /SOLU !Symmetric case DK,1,ALL,0 DK,3,UX,0,,,ROTZ DK,7,UX,0,,,ROTZ !DSYM,SYMM,X FK,8,FY,-75 FK,7,FY,-100 LSWRITE,1 !Antisymmetric case DKDELE,ALL,ALL DK,1,ALL,0 DK,3,UY,0 DK,7,UY,0 FKDELE,ALL,ALL FK,8,FY,-25 LSWRITE,2 LSSOLVE,1,2 FINISH /POST1 SET,1 PLDISP,2 PRRSOL,F
!start SOLUTION MODULE !define a DOF constraint at a keypoint !constrain displacement normal to symmetric plane !to apply symmetry conditions on selected nodes !apply load to a keypoint !write first load step
!delete previous constrains in keypoints !constrain displacent in antisymmetic plane !delete previous forces in keypoints !apply load to a keypoint !write second load step !solve load steps 1 and 2
!start POST-PROCESSOR MODULE !activate first load step solution !deformed shape !list reaction forces
SET,2 !activate second load step solution PLDISP,2 !deformed shape PRRSOL,F !list reaction forces This le can be found at: ftp://amade.udg.edu/mms/PrePostFE/input_files/T4/PrePostFE_Ex407.dat
Note that the addition of the vertical reaction force in node 1 for the two load steps results in the total reaction force this node would have if the whole structure was simulated with the specied loads.
4.5
Although we are focusing on structural problems we may also be interested in having some prediction about our structure working at a different temperature. From the nite element software point of view, we may understand this problem as something like:
K T
=F
(4.3)
where, again, K is the stiffness matrix, is the degree of freedom vector and F is the vector containing the external actions, and T is the variation on the degree of freedom caused by thermal dilatation:
T = T
(4.4)
where is the coefcient of thermal expansion (CTE) vector and T the temperature change. The ANSYS commands useful for these situations are summarised next.
MME
62
Hint 15: ANSYS commands for temperature change in structural analysis ANSYS command MP,CTEX,NMAT,VALUE MP,CTEY,NMAT,VALUE MP,CTEZ,NMAT,VALUE TREF,TEMP TUNIF,TEMP Use CTE through X direction CTE through Y direction CTE through Z direction reference temperature (at which it is measured) working temperature
Example 4.8. Compute the thermal strain and the total strain of the steel plate with a hole shown in Figure 4.7. The dimensions of the plate are measured at a temperature of 25 C while it is expected to work at 80 C. Note: The coefcient of thermal expansion for steel is 12 106 C1 .
Solution to Example 4.8. The ANSYST M command sequence for this example is listed below. You can either type these commands on the command window, or you can type them on a le, then, on the command window enter /input, le, ext.
FINISH !close all previous modules /CLEAR !clear all previous models /TITLE, Temperature change in structural analysis /PREP7 !Parameters R=10 L=100 T=5 p=1 !start PRE-PROCESSOR MODULE !radius in mm !height in mm !thickness in mm !pressure in N/mm2
MME
63
ET,1,PLANE182 !element type #1: plane solid KEYOPT,1,1,3 !simplified enhanced strain formulation KEYOPT,1,3,3 !element #1, keyopt 3=3, plane stress w/thickness R,1,T !element thickness MP,EX,1,207000 !Youngs modulus for material #1 MP,PRXY,1,0.25 !Poisson ratio for material #1 MP,CTEX,1,12e-6 !coefficient of thermal expansion for material #1 K,1,0,0 K,2,L,0 K,3,L,L K,4,0,L L,1,2,20 L,2,3,20 L,3,4,20 L,4,1,20 AL,1,2,3,4 CYL4,2*R,3*R,R ASBA,1,2 AMESH,ALL FINISH /SOLU NSEL,S,LOC,X,0 D,ALL,ALL LSEL,S,LINE,,3 NSLL,S,1 SF,ALL,PRES,p NSEL,ALL TREF,25 TUNIF,80 SOLVE FINISH /POST1 PLDISP,1 PLNSOL,EPTH,X PLNSOL,EPEL,X PLNSOL,EPTT,X !geometry: keypoints
!define areas by lines !define circular hole !substract area 2 from area 1 !mesh area
!start SOLUTION MODULE !select left nodes !fix displacement !select line 3 !select all nodes on lines !apply pressure on all selected nodes
!start POST-PROCESSOR MODULE !deformed shape !thermal strain in direction X !elastic strain in direction X !thermal+elastic strain in direction X
Observe that the value of the strain in the X direction is similar for the thermal expansion and the elastic strain due to the mechanical load.
4.6
Loadcases
In many occasions the same structure is analyzed under different sets of boundary conditions usually called loadcases. FE commercial codes allow to dene these loadcases and to obtain the solution separately for each of them. Hint 16: ANSYS commands for load step denition ANSYS command LSWRITE, LSNUM LSDELE, LSMIN, LSMAX, LSINC LSSOLVE, LSMIN, LSMAX, LSINC SET, LSNUM Usage Write load steps with created loads Delete load steps Read and solve multiple load steps Denes load step to be post-processed
MME
64
Example 4.9. Reconsider the moebius tape sculpture of example 3.6 and introduce different load cases considering: people walking on the structure, gravity load (density equal to 8 T/m3 ), wind load. Solution to Example 4.9. The ANSYST M command sequence for this example is listed below. You can either type these commands on the command window, or you can type them on a le, then, on the command window enter /input, le, ext.
FINISH !close all revious modules /CLEAR !clear all previous models /TITLE,Moebius Sculpture /PREP7 ET,1,BEAM188 KEYOPT,1,3,3 SECTYPE,1,BEAM,CSOLID SECOFFSET,CENT SECDATA,40 SECTYPE,2,BEAM,RECT SECOFFSET,CENT SECDATA,10,60 SECTYPE,3,BEAM,RECT SECOFFSET,CENT SECDATA,25,130 MP,EX,1,210000 MP,PRXY,1,0.29 MP,DENS,1,8e-6 /INPUT,nodes,txt SECNUM,1 E,101,102 EGEN,47,1,1 E,148,201 E,201,202 EGEN,47,1,49 E,248,101 SECNUM,2 E,101,1,248 E,1,201,148 E,102,2,101 E,2,202,201 EGEN,47,1,99,100 ESEL,S,,,167,172 EMODIF,ALL,SECNUM,3 ALLSEL FINISH /SOLU D,236,ALL D,237,ALL D,238,ALL ACEL,,,10 LSWRITE,1 F,12,FY,-500 F,13,FY,-500 F,14,FY,-500 LSWRITE,2 FDEL,12,ALL FDEL,13,ALL FDEL,14,ALL !start PRE-PROCESSOR MODULE !element type #1: BEAM188 (6 DoF) !element #1 keyoption 3=3 (cubic shape function) !section #1: preimetral beams, solid cylindrical !centered section (offset=0) !section radius !section #2: internal beams, rectangular !centered section (offset=0) !section width and height !section #3: bottom internal beams, rectangular !centered section (offset=0) !section width and height !material #1: E=210000 MPa !material #1: 0.29 Poisson !material #1: density=8 T/m^3 !input nodes from file: nodes.txt !section set #1 !elements to generate the perimetral circles
!redefine section bottom elements !select everything !finish PRE-PROCESSOR MODULE !start SOLUTION MODULE !fix sculpture to floor
!gravity load !first load step: only gravity load !additional loads: somebody transit over sculture
MME
65
!wind load !third load step: gravity load + wind !solve load states !finish SOLUTION MODULE !start POST-PROCESSOR MODULE
/INPUT,wind,txt LSWRITE,3 LSSOLVE,1,3 FINISH /POST1 SET,1 PLDISP,1 SET,2 PLDISP,1 SET,3 PLDISP,1
4.7
Suggested problems
Problem 4.1. Reconsider the situation of Example 4.1. Solve the problem again but considering that displacements and rotations are null at both tips of the footbridge. Compare both solutions with the analytical solution you might know. Problem 4.2. Reconsider the situation of Example 4.2 in a different constructive solution. Consider that now it is an inclined roof and that, in addition to being supported by its four corners, it is supported by a perfectly rigid beam located under one of its shorter sides. The inclination is such that the beam supported edge is in a height of 1 m with respect to the opposite edge. Notice that now the normal to the roof surface is not coincident with the gravity direction. Consider how could you apply the snow load.
Problem 4.3. Model the U-shaped beam shown in Figure 4.9 with ANSYST M . The inner radius of the beam is ri = 10 mm, the outer radius is ro = 30 mm, the thickness is t = 5 mm and the length of the straight part is L = 30 mm. The material of the beam is steel, E = 210 GPa and = 0.3, and the applied force is F = 100 N. Evaluate the value and location of the maximum Von Mises stress. Parametrize the geometry of the model, the applied force and the material properties. Show all work in a report.
MME
66
(b) Repeat the problem now applying symmetry conditions. boundary conditions to avoid unreal stresses.
(c) Compare the solutions of (a) and (b) to verify you applied the boundary conditions correctly. (d) Assume the reference temperature is 10 and obtain the thermal strain and the total strain when the temperature of the structure is 60 . The coefcient of thermal expansion for steel is 12106 .
MME
Chapter 5
Material behaviors are very varied depending on different causes: the own physical constitution of the material (metallic materials, geomaterials, polymers, etc), the range of applied strain, or other physic parameters such as temperature, velocity of load application, large time behavior, etc. The main categorization between static non-time dependent models is between linear material models (linear elastic models) and non-linear material models (materials with hyperelasticity, plasticity, etc). An analysis with these non-linear models requires an iterative solution. This chapter is focussed on linear models. Linear material models for structural analysis are basically destined to study the linear elastic response of materials. Hint 17: ANSYS command to dene linear elastic properties Command MP Main Parameters lab,mat,C0,C1,C2,C3,C4
Material property label (lab) for static structural analysis EX - Elastic moduli (also EY, EZ) PRXY - Major Poissons ratios (also PRYZ, PRXZ) NUXY - Minor Poissons ratios (also NUYZ, NUXZ) GXY - Shear moduli (also GYZ, GXZ) DENS - Mass density These properties (which may be functions of temperature) are called linear properties because typical non-thermal analyses with these properties require only a single iteration. X, Y, and Z refer to the element coordinate system. In general, if a material is isotropic, only the EX, the PRXY and possibly the DENS terms are used.
The models should be dened in the different element types according to their formulation: uniaxial models for one-dimensional elements (truss, pipes, etc.), plane models for bi-dimensional elements (plane solid elements, beams, etc.), or three-axial approach for three-dimensional elements (shells, solids). 67
68
5.2
x = Ex
(5.1)
The model only needs a single property, the Youngs modulus (E ) (in ANSYS denoted by label EX). If inertial loads are employed in any loadcase, the density should be also introduced (in ANSYS denoted by label DENS). Examples 2.1, 2.2, 4.1, and so on use this material model.
5.3
ij = Cijkl kl
(5.2)
5.3.1
Isotropic material
When the material is isotropic the stiffness tensor in plane linear elasticity is
1 E 1 C= 1 2 0 0
0 0 1 2
(5.3)
For a plane-stress case, E = E and = , while for a plane-strain or axisymmetrical model = E/(1 2 ) and = /(1 ). E
The stiffness tensor in isotropic solid linear elasticity is
C= 0 0
where G = E/2(1 + ).
1 E E E 0
E 1 E E 0 0 0
E E 1 E 0 0 0
0 0 0 1 G 0 0
0 0 0 0 1 G 0
0 0 0 0 0 1
(5.4)
Only two properties are necessary to dene this model: Youngs modulus (E ) and the Poissons ratio ( ).
MME
69
Hint 18: ANSYS properties for isotropic linear elastic materials In ANSYS, Youngs modulus must be input using the label EX. Poissons ratio (PRXY or NUXY) should not be equal to or greater than 0.5, the value assumed by default is 0.3, if a zero value is desired input PRXY or NUXY with a zero or blank value. The shear modulus (GXY) defaults to EX/(2(1+NUXY)) (ANSYS internally converts PRXY to NUXY when solving the analysis). If GXY is input, it must match EX/(2(1+NUXY)). Hence, the only reason for inputting the shear modulus is to ensure consistency with the other two properties. Isotropic Elastic Example: High Carbon Steel MP,EX,1,210e3 !MPa MP,PRXY,1,.29 !no units MP,DENS,1,7850e-9 !kg/mm3
Examples 2.4, 2.6, 4.5 and so on use isotropic plane stress materials model. Examples 2.5, 2.2 and 4.4 use isotropic plane strain model. Example 4.6 uses isotropic plane strain with axysymmetry. Example 3.10 uses solid isotropic material model.
5.3.2
Orthotropic material
C=
1 Ex xy Ex xz Ex 0 0 0
yx Ey 1 Ey yz Ey 0 0 0
zx Ez zy Ez 1 Ez 0 0 0
0 0 0 1 Gyz 0 0
0 0 0 0 1 Gxz 0
0 0 0 0 0 1 Gxy
(5.5)
where (yx /Ey ) = (xy /Ex ), (zx /Ez ) = (xz /Ex ), and (zy /Ez ) = (yz /Ey ). Depending on Ex , Ey , and Ez values, the pairs related Poissons ratios become major o minor coefcients.
MME
70
Hint 19: ANSYS properties for orthotropic linear elastic material In ANSYS, the properties required for an orthotropic material are: Youngs moduli, EX, EY, EZ, Poissons ratios, may be input in either major (PRXY, PRYZ, PRXZ) or minor (NUXY, NUYZ, NUXZ) form, but not both for a particular material, and shear moludi GXY, GYZ, and GXZ. Note that, for example, if only EX and EY are input (with different values) to a plane stress element, an error will result indicating that the material is orthotropic and that GXY and NUXY are also needed. For axisymmetric analyses, the X, Y, and Z labels refer to the radial (R), axial (Z), and hoop ( ) directions, respectively. Orthotropic Elastic Example: Aluminum Oxide MP,EX,1,307e3 MP,EY,1,358.1e3 MP,EZ,1,358.1e3 MP,GXY,126.9e3 MP,GXZ,126.9e3 MP,GYZ,126.9e3 MP,NUXY,1,.20 MP,NUXZ,1,.20 MP,NUYZ,1,.20 MP,DENS,1,3750e-9 !MPa !MPa !MPa !MPa !MPa !MPa !no units !no units !no units !kg/mm3
Example 5.1. Use ANSYS to model a rectangular plate with dimensions ax =4000 mm, ay =2000 mm, thickness t =10 mm subjected to a uniform tensile load q0 = 1200 N/mm applied in the 2000 mm edge. The material is an unidirectional laminate AS4D/9310 which elastic properties are summarised in Table 5.1, with the bers oriented in the x-axis. Obtain the elongation of the plate under the applied load and compare it with the analytical solution obtained using Hookes Law. Table 5.1: Material properties of unidirectional AS4D/9310 carbon/epoxy composite. Property AS4D/9310 133.86 GPa 7.71 GPa 4.36 GPa 2.76 GPa 0.301 0.396
Solution to Example 5.1. As the problem can be considered as a 2D situation, one of the coordinates, in the thickness direction, can be eliminated from the governing equations so that the 3D problem simplies to 2D. In the process, the thickness becomes a parameter, which is known and supplied to the modeling software. Typically, we differentiate material properties and
MME
71
parameters even though both are supplied as known input data. Most software packages make this difference as well. For example, the solid thickness is supplied to ANSYS as a real constant set (R command), while material properties are entered separately (MP command). In the FE model we use PLANE183, a 8-node bi-dimensional solid element. Symmetry with respect to the x z and y z planes is used to model 1/4 of the plate. The ANSYST M command sequence for this example is listed below. You can either type these commands on the command window, or you can type them on a le, then, on the command window enter /input, le, ext.
FINISH !close all previous modules /CLEAR !clear all previous models /TITLE,Rectangular plate under distributed load /PREP7 !start PRE-PROCESSOR MODULE t=10 !thickness ET,1,PLANE183,,,3 !element type #1, 2D solid, plane stress w/thickness R,1,t !real constant #1, thickness !material properties FOR AS4D/9110 orthotropic laminate MP,EX,1,133.86e3 !MPa MP,EY,1,7.706e3 MP,EZ,1,7.706e3 MP,GXY,1,4.36e3 MP,GYZ,1,2.76e3 MP,GXZ,1,4.36e3 MP,PRXY,1,0.301 MP,PRYZ,1,0.396 MP,PRXZ,1,0.301 RECTNG,0,2000,0,1000 ESIZE,250 AMESH,all FINISH /SOLU ANTYPE,STATIC DL,1,1,SYMM DL,4,1,SYMM SFL,2,PRES,-1200/t SOLVE FINISH /POST1 PLDISP,1 PLESOL,S,X !geometry: square area !element size 250 mm !mesh the area
!start SOLUTION MODULE !static analysis !symmetry conditions in line 1 !apply uniform pressure N/mm in line #2
!start POST-PROCESSOR MODULE !plot deformed shape !contour plot of x direction stress
The obtained maximum displacement of the model is 1.813 mm, due to used symmetry conditions, the total elongation of the plate is the double, L = 3.626 mm. Applying the Hookes Law the total elongation is
L = F
ax L = q0 = 3.586mm EA Ex t
MME
72
5.3.3
Anisotropic material
The stiffness tensor for an general anisotropic material with linear elasticity is
C=
which is dened by 21 constants.
C11
C12 C22
(5.6)
Hint 20: ANSYS properties for anisotropic linear elastic material In ANSYS the order of the vector is expected as {x, y, z, xy, yz, xz}, whereas in the Voigt notation the order is given as {x, y, z, yz, xz, xy}. The anisotropic values are dened in ANSYS using TB,ANEL,1 and one TBDATA command for each value (see example next). The sequence number in TBDATA command is obtained following the rows showed in the above stiffness tensor (C11 , C12 , C13 , C14 , C15 , C16 , C22 , C23 ...). Anisotropic Elastic Example: Cadmium MP,DENS,1,3400e-9 TB,ANEL,1 TBDATA,1,121e3 TBDATA,2,48.1e3 TBDATA,3,44.2e3 TBDATA,7,121e3 TBDATA,8,44.2e3 TBDATA,12,51.3e3 TBDATA,16,18.5 TBDATA,19,18.5 TBDATA,21,24.2 !kg/mm3 !C11 !C12 !C13 !C22 !C23 !C33 !C44 !C55 !C66 (MPa) (MPa) (MPa) (MPa) (MPa) (MPa) (Pa) (Pa) (Pa)
Only some elements in ANSYS have the anisotropic elasticity capability (SOLID5, PLANE13, SOLID64, SOLID98, SHELL181, PLANE182, PLANE183, SOLID185, SOLID186, SOLID187, among others).
5.4
During the denition of the model when we use orthotropic or anisotropic material, we need to be sure about that the material properties are applied in the correct orientation. In most of FEA commercial codes, the material coordinate system is associated to the element coordinate system, one right-handed orthogonal system associated to each element. Also, the element coordinate system can be used to obtain the derived results (strains and stress) in these material directions.
MME
73
The orientation of the element coordinate system is associated with the element typology. For bar or beam elements the orientation of the x-axis is generally along the element line dened by two end nodes. For solid elements (in two and three dimension), generally the orientation is dened parallel to the global or geometric coordinate system. However, for shell elements this is not useful as the x and y directions must be dened on the element surface and the z -axis always normal to it. The default orientation of x and y axes depends of the commercial code and the element type. There are various ways to dene the default orientation of x and y in shell elements. Two of them are shown in Fig. 5.1. In Fig. 5.1(a) the x-axis is aligned with the edge dened by the rst and second nodes of each element, the z -axis normal to the shell surface (with the outward direction determined by the right-hand rule), and the y -axis perpendicular to the x- and z -axis (ANSYS uses this rule as default). Other packages, such as MSC.MARC, calculate the orientation of x-axis from the lines dened by the middles points of the edges as shown in Fig. 5.1(b).
Figure 5.1: Default orientations of element (or material) coordinate systems in shells elements: (a) ANSYS, (b) MSC.MARC. In the example 5.1, we have dened only a rectangular plate with rectangular elements and all of them have the rst and the second nodes aligned with the global x-axis. The material axes have been chosen parallel as the global axis. But this is not the general case. Most commercial codes have utilities to control or change the element coordinate system. In Example 5.2, it is shown how it can be done in a shell with curvature, and in Example 5.3, how different orientations in different locations of the structure can be dened. Example 5.2. Dene a 3D curved shell in ANSYS and mesh it aligning the coordinate system of the elements with the global y -axis. Solution to Example 5.2. For shells dened in 3D, the ESYS orientation uses the projection of the local system on the shell surface. The element x-axis is determined from the projection of the local x-axis on the shell surface. The z -axis is determined normal to the shell surface (with the outward direction determined by the right-hand rule), and the y -axis perpendicular to the x- an z -axis. For elements without midside nodes (linear shape functions), the projection is evaluated at the element centroid and it is assumed constant in direction throughout the element. For elements with midside nodes (quadratic shape functions), the projection is evaluated at each integration point and may vary in direction throughout the element. In ANSYS a local coordinate system, which can be cartesian, cylindrical or spherical, must dened using the LOCAL command. Then, each element is associated to a previously dened local coordinate system using the element property ESYS. The result is the orientation of the x-axis element coordinate system parallel to x-axis local coordinate system. Also, it is possible to dene element coordinate system orientations by user written subroutines. The ANSYST M command sequence to align the element x-axis with the global y -axis for this example is listed below. You can either type these commands on the command window, or you can type them on a le, then, on the command window enter /input, le, ext.
MME
74
FINISH !close all previous modules /CLEAR !clear all previous models /TITLE, Element orientation in a curved shell /PREP7 ET,1,SHELL281 SECTYPE,1,SHELL SECDATA,5 K,1,30,0,2 K,2,0,5,5 K,3,-30,0,2 K,4,20,35,0 K,5,0,35,2 K,6,-20,35,0 L,1,4 L,3,6 BSPLIN,1,2,3 BSPLIN,4,5,6 AL,ALL LOCAL,11,0,8,,,90,0,0
!start PRE-PROCESSOR MODULE !element type #1: SHELL281 (6 DOF) !section #1 = shell !thickness !define geometry by keypoints and lines
!create spline between nodes 1, 2 and 3 !create spline between nodes 4, 5 and 6 !create a area with all lines !define local coord. system, with label 11,according to cartesian !coordinate system, with origin in point (8,0,0) !and rotation about local Z,X,Y equal to (90,0,0) !set coord. system for elements meshed !define element size !mesh the area #1 !go back to default coord. system !set on display laminate orientation !not hidden surfaces !change arrow scaling !display elements
The results of the orientation of the elements coordinate systems can be observed in Figure 5.3. The element coordinate systems may be displayed as a triad with the /PSYMB command with the label ESYS.
Figure 5.2: Element coordinate system orientation, not using "ESYS" command (left) and using "ESYS" command (right).
Example 5.3. Model in ANSYS a ange pipe with axial and radial material orientation. In the cylindrical part the reference axis (x-axis) will be in longitudinal direction. In the ange the reference axis will be radial (see Fig. 5.3). Solution to Example 5.3. In the model different orientation systems in different model locations must be dened. Two local reference axes are dened and activated using the ESYS command. Then, the elements on the cylindrical part are aligned to the axial direction and the elements on
MME
75
Figure 5.3: Reference x-axis in a pipe ange. the ange are aligned according to the radial direction. The ANSYST M command sequence for this example is listed below. You can either type these commands on the command window, or you can type them on a le, then, on the command window enter /input, le, ext.
FINISH !close all previous modules /CLEAR !clear all previous models /TITLE, Flange pipe with axial and radial material orientations /PREP7 ET,1,SHELL281 SECTYPE,1,SHELL SECDATA,5 CYL4,0,0,350,,,,300 CYL4,0,0,350,,550 AGLUE,3,4,5 LOCAL,11,0,,,,0,0,90 LOCAL,12,1,,,,0,0,0 ESIZE,50 ESYS,11 AMESH,3,4 ESYS,12 AMESH,6 CSYS,0 /PSYMB,ESYS,1 /TYPE,1,0 EPLOT !start PRE-PROCESSOR MODULE !element type #1 SHELL281 (6 DOF) !section #1 = shell !thickness !create geometry by solid modeling !glue areas, area 5 becomes area 6 !define rotation=90 deg around Y (cylinder) !define polar coordinate system (flange) !define element size !set coord. system for elements meshed !mesh the cylindrical areas (areas 3 and 4) !set coord. system for elements meshed !mesh the flange area (area 6) !go back to default coord. system !set on display local orientation !not hidden surfaces !display elements
Observe that independently of the geometry of the element, all the element coordinate systems in the cylindrical part are oriented in the longitudinal direction and those of the ange part are oriented in the radial direction.
5.5
Suggested problems
Problem 5.1. Reconsider the rectangular plate made of an orthotropic material described in Ex.5.1 but introducing the same material properties by using the anisotropic behavior option command instead of the orthotropic one. Check that the obtained results are identical to those of Ex. 5.1. Problem 5.2. Using PLANE182 element, generate a mesh from the spring geometry shown in Figures 5.4, where L =100 mm, ri =10 mm ,ro =30 mm and thickness equal to 10mm. The spring is made of the same orthotropic material as in Ex. 5.1. The dashed line on middle of the strip shows the y-axis of the material orientation. Report graphically the obtained mesh and show the material axis orientation.
MME
76
MME
Chapter 6
Postprocessing
6.1 Introduction
The gathering, interpretation and understanding of the calculated results in a Finite Element problem is as important as a good denition of the boundary conditions, type of element, etc. In this chapter, different useful methods to obtain and analyze the calculated results are presented. All Finite Element programs have the possibility to analyze the obtained results by using two different methods: general postprocessing and time-history postprocessing. The rst method is intended to review the results for the whole model at specic steps and sub-steps (specic timeincrements of frequencies). With this method, for instance, the stress distribution in the model can be plotted for a structural analysis. The second method allows to obtain the variation of a certain variable or result item at particular locations in the model with respect to time, frequency or other result variables. In a nonlinear structural analysis, for example, the deection of a certain node can be plot versus time or the applied load.
Hint 21: ANSYST M methods for post-processing The general postprocessing in ANSYST M is carried out by using the General Postproc option in the Main Menu or invoking the /POST1 module in a command script. The time-history postprocessing in ANSYST M is carried out by using the TimeHist Postpro option in the Main Menu or invoking the /POST26 module in a command script.
6.2
General postprocessing
The general postprocessor is used to review the results for the entire model or parts of the model for a specic time, frequency or step. In this module, simple graphic representations of the results can be obtained as well as listing the results in a table and more complex data manipulation. By default, different results are directly obtained after a simulation. These results depend on the type of analysis and the type of element used. 77
The rst step to analyze the obtained results is to handle the results for the specic step, sub-step or time increment desired. By default, most of the commercial Finite Element programs read the results data for the last calculated increment. Hint 22: ANSYST M methods for reading results data SET reads all the results data from the results le for the whole model at a particular time. SUBSET reads all the results data from the results le for specic parts of the model at a particular time. Other commands such as LAYER, TOP, MID, BOTTOM can be used to specify particular parts of the model where results are to be considered. APPEND reads all the results data from the results le for the whole model at a particular time and adds them to the previously stored results data.
Reading selected results data Most of the results data, such as stresses, strains, reaction forces and displacements in a structural analysis, are available during postprocessing in commercial Finite Element codes. In some, these data results need to be specied beforehand (this is the case of ABAQUST M , for instance). In others, not all the data results are directly available in the general postprocessing but have to be obtained afterwards (this is the case of ANSYST M ). Hint 23: ANSYST M methods for reading selected results data In ANSYST M some results (depending on the analysis and element type) must be accessed by creating an Element Table, ETABLE. ETABLE allows accessing, storing and manipulating data results for afterwards representation. The lling of the element table and the use of the command ETABLE must be specically done in accordance to the available results for the considered element type and analysis type. It is strongly recommended to check which results are available for the element and analysis type and identify the variable names or numbers that correspond to the desired results.
Coordinate system for the results data Commercial Finite Elements codes allow reviewing the results data in different coordinate systems. Usually, results are calculated and stored in the nodal or the element coordinate systems and rotated into the active coordinate system, which, by default, is the global cartesian system.
MME
Chapter 6. Postprocessing
79
These results, can be afterwards rotated into any previously dened coordinate system for display, list and storage. Hint 24: ANSYST M coordinate systems for results data Use RSYS,0 to obtain the results in the global cartesian coordinate system (default). Use RSYS,1 to obtain the results in the global cylindrical coordinate system. Use RSYS,2 to obtain the results in the global spherical coordinate system. Use RSYS,N to obtain the results in the local coordinate system N . Use RSYS,SOLU to obtain the results in the nodal or element coordinate systems.
Manipulation of the results data Most commercial Finite Elements programs allow the manipulation of the data results, including arithmetic operations, comparisons, failure criteria denition, etc. Hint 25: ANSYST M manipulation of results data Use SADD to add two different specied stored results. Use SMULT to multiply two different specied stored results. SEXP can be used to exponentiate and multiply two different specied stored results. SMIN and SMAX allow to compare and obtain the minimum and maximum, respectively, of two different specied stored results. SALLOW can be used to dene the allowable stress for safety factor calculations. SFCALC calculates safety factors for stored results.
Reviewing results data All the commercial Finite Elements codes allow reviewing the results data both in graphical plots and listings. Graphical plots include different options such as contour displays, deformed shapes, vector plots, path plots, etc. Contour plots display the variation of the results variable over the model. Deformed shapes represent the nal shape of the system in equilibrium after taking into account all the boundary conditions and loads. Vector plots show the variation and direction of a vector quantity result using arrows. Path plots are useful to observe the variation of a results variable along a predened path.
MME
80
Hint 26: ANSYST M methods for graphically reviewing results data PLNSOL displays continuous contour lines over the entire model. Use PLESOL to display contour lines that are discontinuous across the elements. PLETAB displays contour lines of stored data in element tables. Use PLDISP to display the deformed shape of the model. PLVECT generates vector displays using arrows. PLPATH plots the results data on the predened path.
Listings are an efcient tools to report resulting nodal, element solution and reaction data. Hint 27: ANSYST M methods for listing results data Use PRNSOL to list specied nodal solution data. Use PRESOL to list specied results for the selected elements. PRRSOL lists the node reactions for the selected nodes. PRNLD lists the summed element nodal loads for the selected nodes.
Example 6.1. Consider the clamped beam simulated in Chapter 1 and the input le generated in Ex. 1.2. After simulating the beam, postprocess the results and plot the deformed shape of the beam, the contour plots for the distribution of longidtudinal stress, equivalent (Von Mises) stress and principal elastic strain and list the reaction forces for 5th and the last substeps. Solution to Example 6.1. As it was already explained in Chapter 1, the requested plots can be obtained by means of the GUI menus, the Results Viewer module or commands. In this case, an input le with all the required commands is used. The ANSYST M command sequence for this example is listed below.
/POST1 SET,1,5 PLDISP,1 PLESOL,S,X,0,1. PLESOL,S,EQV,0,1. PLESOL,EPEL,1,0,1. PRRSOL SET,1,LAST PLDISP,1 PLESOL,S,X,0,1. PLESOL,S,EQV,0,1. PLESOL,EPEL,1,0,1. PRRSOL !start Post-Processor module !set step 5 !plot the deformed shape !contour plot of stress in X-direction !contour plot of Von Mises stress !contour plot of principal strain !list of reaction forces !set last step (default) !plot the deformed shape !contour plot of stress in X-direction !contour plot of Von Mises stress !contour plot of principal strain !list of reaction forces
MME
Chapter 6. Postprocessing
81
Observe that as the analysis is linear elastic, the results for the 5th substep are half of those for the last. Example 6.2. Consider the clamped beam simulated in Chapter 1 and the input le generated in Ex. 1.2. After simulating the beam, postprocess the results and obtain the variation of the equivalent stress along the vertical line dened by the nodes on the centre of the clamped section. Solution to Example 6.2. The ANSYST M commands to dene the required pathplot, obtain the results along this pathplot and represent it are summarised in the following input le.
/POST1 PATH,vertax,2 PPATH,1,,0,0,100 PPATH,2,,0,100,100 /PBC,PATH,1 /REPLOT PDEF,Svertax,S,X,AVG PLPATH,Svertax !start Post-Processor module !definition of PATH "vertax" by 2 points !start node located on X,Y,Z !end node located on X,Y,Z !show path
Observe that as expected for a linear elastic analysis, the variation of the equivalent stress along the path is linear. Example 6.3. A clamped plate 100 mm long, 25 mm wide and 5 mm thick is made of AS4D/9310 unidirectional composite material (see Ex. 5.1 for material properties). The material is oriented at 45 with respect to the longitudinal axis of the plate. The plate is bended 10 mm at the free end. Simulate the plate using shell elements and postprocess the results to obtain the contour plot of the resulting stresses in the material. As a simple failure criterion, it is assumed that the material fails when the absolute value of the stress in the bre direction is equal or greater than 500 MPa. Compute the failure criterion and obtain its contour plot. Solution to Example 6.3. The ANSYST M commands to generate, simulate and postprocess the previous model are listed in the summarised input le.
FINISH !close all previous modules /CLEAR !clear all previous models /TITLE, Clamped plate L=100 B=25 h=5 fi=45 FA=500 !plate length !plate width !plate thickness !orientation angle !allowable fibre stress
/PREP7 !start PRE-PROCESSOR MODULE ET,1,SHELL181 !element type #1: SHELL181 (6 DoF) KEYOPT,1,3,2 !element #1: keyoption 3=2 (full integration incompatible modes) SECTYPE,1,SHELL !shell section type for material#1 SECDATA,h !section data, with thickness h=5mm MP,EX,1,133.86E3 !material is UD AS4D/3100 Carbon/Epoxy MP,EY,1,7.706E3 MP,EZ,1,7.706E3 MP,GXY,1,4.36E3 MP,GYZ,1,2.76E3 MP,GXZ,1,4.36E3
MME
82
MP,PRXY,1,0.301 MP,PRYZ,1,0.396 MP,PRXZ,1,0.301 RECTNG,0,L,0,B ESIZE,5 LOCAL,11,0,,,,fi ESYS,11 AMESH,ALL CSYS,0 FINISH /SOLU NSEL,S,LOC,X,0 D,ALL,ALL NSEL,S,LOC,X,L D,ALL,UZ,-10 NSEL,ALL SOLVE FINISH /POST1 /VIEW,1,1,1,1 PLNSOL,S,X,2,1 RSYS,11 PLNSOL,S,X,2,1 ETABLE,S1,S,X SABS,1 SMULT,FF,S1,,1/FA PLETAB,FF,AVG !creates a rectangle with x=L and y=B !element size 5 mm !define material orientation !use material orientation !mesh the area !return to global system
!select encastred nodes !fix encastred nodes !select end nodes !apply displacement end nodes !select all !solve the current load case
!iso-view !stress in direction 1, global !results in material orientation !stress in direction 1, local !store fibre stress as S1 !use absolute values !store fibre failure as FF !plot fibre failure, averaged
Observe that the stress contour plots depend on the coordinate system employed. If the global coordinate system is used, the x-stress is in the longitudinal direction of the plate while when the element coordinate system is used, the x-stress corresponds to the bre stress. The contour plot of the longitudinal stress in the material direction shows a maximum of 789.67 MPa, which is higher than the material allowable stress in the bre direction. However, the contour plot for the assumed failure criterion shows a maximum of 0.75, indicating that there is no material failure. This incoherence is due to the fact that the failure criterion employed has been calculated using the ETABLE command and element results. Actually, ETABLE assumes a unique interpoled value for the whole element from the nodal values. In this way, part of the information is lost and some differences appear. These differences can be reduced by reducing the size of the element or performing a mesh quality assessment (see Section 6.5). Try to recalculate the same example using smaller elements and compare the results. Another possibility to obtain more accurate results, especially for failure prediction, is to use the nodal results. In ANSYST M the FC command can be used for this purpose. The following input le shows an example of the use of the maximum stress failure criterion in the previous example.
/POST1 !Failure criterion definition and calculation FC,1,S,XTEN,500 !mat1,stress,max stress 11t FC,1,S,XCMP,-500 !mat1,stress,max stress 11c FC,1,S,YTEN,1e6 !mat1,stress,max stress 22t (large value, dont compute) !max stress 22c=-22t FC,1,S,ZTEN,1e6 !mat1,stress,max stress 33t (large value, dont compute)
MME
Chapter 6. Postprocessing
83
!max stress 33c=-33t !mat1,stress,max stress 12 (positive, large value, dont compute) !mat1,stress,max stress 23 (positive, large value, dont compute) !mat1,stress,max stress 13 (positive, large value, dont compute)
PRNSOL,S,FAIL !print table with FAIL index PLNSOL,FAIL,SMAX !maximum stress failure criterion This le can be found at: ftp://amade.udg.edu/mms/PrePostFE/input_files/T6/PrePostFE_Ex603_fail.dat
Observe that in this case the predicted failure index almost doubles that of the initial calculation.
6.3
Time-history postprocessing
The time-history postprocessor is used to review the results for specic parts of the model as a function of time or frequency. As in the case of the general postprocessor, it is also possible to obtain graphic plots and listings as well as perform data manipulation. General uses of the time-history postprocessing include the variation plot of a certain result as a function of time in a transient analysis or obtain the force-deection plot in a nonlinear structural analysis. Obtaining results data The rst step to analyse the obtained results is to handle the results for the specic part of the model and assign them to a results variable. Hint 28: ANSYST M methods for reading time-history results data In the time-history postprocessor of ANSYST M different data results for specic locations are retreived from the results le by using different commands and assigning them a numerical label. Use NSOL to specify the nodal data from the results le to be stored. ESOL species the element data from the results le to be stored. Use ANSOL to specify the averaged nodal data from the results le to be stored in the solution coordinate system. RFORCE species the total reaction force to be stored.
Manipulating results data Once the time-history result variables are dened, they can be manipulated and combined to obtain new variables. Hint 29: ANSYST M methods for manipulating time-history results data In the time-history postprocessor of ANSYST M there are different commands to manipulate and combine result variables in order to obtain new ones. Some of the more useful are: ABS, ADD, ATAN, EXP, PROD, SQRT. Check their use in the HELP module.
MME
84
Reviewing results data As in the case of the general postprocessing, time-history variables can be both graphically displayed and listed. Hint 30: ANSYST M methods for reviewing time-history results data Use PLVAR to graph the result variables. Up to ten variables can be plot. XVAR species the result variable to be used for the X-axis in the graphs. By default, is time in static and transient analyses and frequency in harmonic analyses. PRVAR lists up to six specied result variables. Times or frequencies for which results are to be listed can be controlled with NPRINT and PRTIME.
Example 6.4. Consider the clamped beam simulated in Chapter 1 and the input le generated in Ex. 1.2. After simulating the beam, postprocess the results and obtain the evolution of the longitudinal stress and strain at the top node of the vertical line on the centre of the clamped section. Plot the evolution of both variables in a combined plot versus time taking into account a scale factor if necessary. List the results and plot the variation of the longitudinal stress versus the longitudinal strain. Solution to Example 6.4. The ANSYST M command sequence for obtaining the time-history postprocessing of this example is listed below.
/POST26 !history plot module of x-strain at node 708, name it as EPELX reference number 2 of x-stress at node 708, name it as SX reference number 3
ANSOL,2,708,EPEL,X,EPELX !create evolution !and save it with ANSOL,3,708,S,X,SX !create evolution !and save it with PROD,4,2,,,EPELX5,,,1e5
!create a new variable, with 4 as number reference and EPELX5 !as name, that takes the variable with reference number 2 and applies !a scaling factor of 1e5 !plot x-strain (scaled) and stress evolution !list x-strain and stress evolution !set the variable with reference number 2 to be on the x-axis !plot the variable SX versus EPELX (x-stress versus x-strain)
Observe that the longitudinal stress is proportional to the longitudinal strain, as expected for a linear elastic model. A scale factor of 1 105 has been considered to plot in the same graph the evolution of the longitudinal stress and strain.
6.4
When performing a structural simulation, the comparison between the resulting strains or stresses with an allowable value in a uniaxial model is straightforward. However, when there is more than
MME
Chapter 6. Postprocessing
85
one component of strain or stress, these components are normally combined into one resulting strain or stress and then compared with the allowable value. Especic techniques for strain and stress combination are presented next.
6.4.1
Combined strain
Principal strains can be calculated from the strain components by the cubic equation:
x 0 xy /2 xz /2 xy /2 y 0 yz /2 xz /2 yz /2 z 0
=0
(6.1)
where 0 are the three solutions and principal strains. The three principal strains are labeled 1 , 2 , and 3 (in ANSYST M output as 1, 2 and 3 with strain items such as EPEL). The principal strains are ordered so that 1 is the most positive and 3 is the most negative. The strain intensity I (output as INT with strain items such as EPEL) is the largest of the absolute values of 1 - 2 , 2 - 3 , or 3 - 1 . That is:
I = max (|1 2 |, |2 3 |, |3 1 |)
(6.2)
The von Mises or equivalent strain e (in ANSYST M output as EQV with strain items such as EPEL) is computed as
1 e = 1+
6.4.2 Combined stress
1 (1 2 )2 + (2 3 )2 + (3 1 )2 2
1 2
(6.3)
Principal stresses are calculated from the stress components by the cubic equation:
x 0 xy xz xy y 0 yz xz yz z 0
=0
(6.4)
where 0 are the three solutions and principal stresses. The three principal stresses are labeled 1 , 2 and 3 (in ANSYST M output as S1, S2 and S3). The principal stresses are ordered so that 1 is the most positive and 3 is the most negative. The stress intensity I (in ANSYST M output as SINT) is the largest of the absolute values of 1 - 2 , 2 - 3 , or 3 - 1 . That is:
I = max (|1 2 |, |2 3 |, |3 1 |)
The von Mises or equivalent stress e (in ANSYST M output as SEQV) is computed as
(6.5)
e =
1 (1 2 )2 + (2 3 )2 + (3 1 )2 2
1 2
(6.6)
Example 6.5. Graph the contour plots of the principal and equivalent (Von Mises) strains and stresses for the cantilever beam simulated in Ex. 1.2. Also graph the contour plot of the stress and strain intensity and a vector plot of the principal stress and strain.
MME
86
Solution to Example 6.5. The following ANSYST M input le can be used to obtain the above strains and stresses contour plots and vector plots.
/POST1 PLNSOL,S,1 PLNSOL,S,2 PLNSOL,S,3 PLVECT,S PLNSOL, S,INT, PLNSOL, S,EQV, PLNSOL,EPEL,1 PLNSOL,EPEL,2 PLNSOL,EPEL,3 PLVECT,EPEL PLNSOL,EPEL,INT PLNSOL,EPEL,EQV !start Post-Processor module !contour plot of 1st principal stress !contour plot of 2nd principal stress !contour plot of 3th principal stress !vector plot of principal stress !contour plot of stress intensity !contour plot of Von Mises stress !contour plot of 1st principal strain !contour plot of 2nd principal strain !contour plot of 3th principal strain !vector plot of principal strain !contour plot of strain intensity !contour plot of Von Mises strain
Observe that in the vector plot of the principal strains and stresses, the maximum principal stress and strain are located on the top part of the beam, while the minimum principal stress and strain are located on the bottom part of the beam and represented by an inverted arrow (negative). In both cases, the direction of the arrows is longitudinal with the beam, indicating the direction of the principal stress and strain directions. Also observe that the size of the arrow is proportional to the value of the represented stress or strain. The equivalent (Von Mises) stress and strain can be also calculated using the ETABLE command. However, as in the case of Ex. 6.3, there is a certain difference between the values calculated in this way and the values directly obtained with ANSYST M . The following input le can be used to calculate the Von Mises stress for the current example.
/POST1 ETABLE,Sp1,S,1 ETABLE,Sp2,S,2 ETABLE,Sp3,S,3 SADD,Si1,Sp1,Sp2,,-1 SADD,Si2,Sp2,Sp3,,-1 SADD,Si3,Sp3,Sp1,,-1 SEXP,Si11,Si1,,2 SEXP,Si22,Si2,,2 SEXP,Si33,Si3,,2 SADD,Si12,Si11,Si22 SADD,Si123,Si12,Si33 SMULT,S123,Si123,,0.5 SEXP,Seq,S123,,0.5 PLETAB,Seq,avg !start Post-Processor module !store max principal stress in Sp1
!store Si12 as (Sp1-Sp2)^2+(Sp2-Sp3)^2 !store Si123 as (Sp1-Sp2)^2+(Sp2-Sp3)^2+(Sp3-Sp1)^2 !store S123 as Si123/2 !store Seq1 as S123^0.5
MME
Chapter 6. Postprocessing
87
6.5
As mentioned in Section 3.4, it is necessary to assess the mesh quality to be condent on the results of a mechanical or structural simulation. That is, to check if the mesh is ne enough to get good simulations results. Basically, the following aspects need to be analysed: Assure the right connectivity between different parts of the mesh. Avoid elements with high distortions, warping and inadmissible narrow or wide angles of the element edges. Error estimation of the elements, typically by means of energy hypotheses. Study the sensitivity of the results on different meshes of the entire FE model or of a partial model. From the above, the error estimation of the elements by means of energy hypotheses is carried out during the postprocessing. The mesh quality assessment in this way is explained next.
6.5.1
The nite element solution is an approximation to the true solution of a mathematical problem. For the analyst is important to know the magnitude of error involved in the solution. The ANSYST M program offers a method for a posteriori estimation of the solution error due to mesh discretisation. The method involves calculating the energy error within each nite element and expressing this error in terms of a global error energy norm. The primary unknown in conventional nite element analysis is the displacement at every node in the model. The nite elements used in the model have shape functions associated with them that characterize the resulting displacement eld of the physical model. The principal of minimization of total potential energy is used to obtain the equation in the form of:
[K] {x} = {F }
(6.7)
where [K] is the global stiffness matrix, {x} is the displacement vector, and {F } is the force vector. This equation is solved for the unknown displacement vector {x}. The stresses and strains are then obtained as the rst derivative of these displacements. So while the displacement eld in the nite element model is continuous, the stress eld in the model is dicontinuous. The stress at a node, as printed from ANSYST M , is the average of the stresses from all the elements attached to that node. This introduces an error in the magnitude of stress at a node and is referred to as mesh discretization error. The phenomenon of mesh discretization error is graphically illustrated in Figure 6.1. The coarser the mesh, the greater the potential for this error. Similarly, for a very ne mesh, the contribution to stress at a node from all elements attached to it will be the same, exact value. However, rarely do we have the luxury of repeated mesh renement. This way, the nodal stress error vector is dened as:
i a i {n } = {n } {n }
(6.8)
a i where {n } is the stress error vector of node n of element i, {n } is the average stress vector i } is the stress vector of node n of element i. As an example, the nodal stress at node n, and {n 6 error vector of node 13 of element 6, {13 }, in Figure 6.1 can be computed as:
MME
88
1 2 3
1 6 7
2 8
3 9
4 10
5 11 12
8 15
9 16 17
12 20
13
14
15
16
21
22
23
24
25
x13 =
y13 = 37.5;
xy13 = 25
(6.9)
(6.10)
(6.11)
(6.12)
Once all the nodal stress error vectors are computed, the energy error for each element can be dened as:
Ei =
1 2
(6.13)
where Ei is the energy error of element i and Dkl is the constitutive tensor. By adding all element energy errors Ei , the global energy error in the model, E , can be determined:
Ne
E=
i=1
Ei
(6.14)
The energy error can be normalized against the strain energy U and expressed as a percent error in energy norm, E :
ERR =
E U +E
1 2
(6.15)
The percent error in energy norm E is a good overall global estimate of the discretization or mesh accuracy. According to this, in ANSYST M , the structural energy error (labeled SERR) is a measure of the discontinuity of the stress eld from element to element, and the thermal energy error (TERR) is a measure of the discontinuity of the heat ux from element to element. With these energy errors (SERR and TERR) ANSYST M program calculates a percent error in energy
MME
Chapter 6. Postprocessing
89
norm (SEPC for structural percent error and TEPC for thermal percent error). PRERR command
lists SEPC and TEPC for all selected elements, and PLESOL displays the contours of SERR or TERR to nd out where to rene the mesh. Example 6.6. Evaluate the accuracy of the mesh and the solution of the FE model used to solve problem 4.5. Plot nodal stresses, element stresses, and the percentage of energy error. Evaluate two different meshes. Solution to Example 6.6. Two different element types and two different mesh renements are considered in order to establish a comparison. The rst element type considered is a 8-noded 2D solid element, PLANE183, while the second is the 4-noded version of the same, PLANE182. The rst mesh corresponds to an element size of about 5.5 mm, while in the second the element size is of 1.5 mm. Before getting the energy error quantities the PowerGraphics option of ANSYST M has to be disabled, /GRAPHICS,FULL. To plot the percentage of structural energy error of each element type, PLESOL,SERR,,0,1.0. To list the total pencentage of energy error of each element type PRERR. The rst model to be analyzed corresponds to a mesh with 8-noded 2D solid elements and an element size of 5.5mm. The mesh obtained and the energy error of each element are shown in Figure 6.2. The total percentage error in energy norm returned by ANSYST M is STRUCTURAL PERCENTAGE ERROR IN ENERGY NORM (SEP C) = 5.0734. This difference is appreciated if the nodal and the element quantities for the stress in the direction perpendicular to the load axis are compared. These stresses are shown in Figure 6.3.
Figure 6.2: Mesh and energy error with elements of nominal size of 5.5 mm and 8-noded plane elements. The results obtained when 4-noded plane elements of nominal size 5.5 mm are used are shown in Figure 6.4. The total percentage error in energy norm returned by ANSYS in this case is STRUCTURAL PERCENTAGE ERROR IN ENERGY NORM (SEP C) = 11.389. The difference between the nodal and the element quantities for the stress in the direction perpendicular to the load axis is higher than when 8-noded elements are used, as it is shown in Figure 6.5.
MME
90
Figure 6.3: Nodal (left) and element (right) stress in the direction perpendicular to the load axis with elements of nominal size of 5.5 mm and 8-noded plane elements.
Figure 6.4: Mesh and energy error with elements of nominal size of 5.5 mm and 4-noded plane elements.
Figure 6.5: Nodal (left) and element (right) stress in the direction perpendicular to the load axis with elements of nominal size of 5.5 mm and 4-noded plane elements.
Using plane stress elements with 8-noded (PLANE183) and a mesh with elements with a nom-
MME
Chapter 6. Postprocessing
91
inal size of 1.5mm, the energy error is smaller as it is shown in Figure 6.6. Now, the total percentage error in energy norm returned by ANSYS is STRUCTURAL PERCENTAGE ERROR IN ENERGY NORM (SEP C) = 0.28475. The difference between the nodal and the element quantities for the stress in the direction perpendicular to the load axis is smaller than in previous analysis as it is shown in Figure 6.7.
Figure 6.6: Mesh and energy error with elements of nominal size of 1.5 mm and 8-noded plane elements.
Figure 6.7: Nodal (left) and element (right) stress in the direction perpendicular to the load axis with elements of nominal size of 1.5 mm and 8-noded plane elements.
For comparison purpouses, the last simulation considers plane stress elements with 4-noded (PLANE182) and a mesh with elements with a nominal size of 1.5mm (see Figures 6.8 and 6.9). In this last case, the total percentage error in energy norm returned by ANSYS is STRUCTURAL PERCENTAGE ERROR IN ENERGY NORM (SEP C) = 0.4.9002. This energy error is larger than the one of the 8-noded and 1.5mm size case, but lower than the one of the 8-noded and 5.5mm size case (see summary in Table 6.1).
MME
92
Figure 6.8: Mesh and energy error with elements of nominal size of 1.5 mm and 4-noded plane elements.
Figure 6.9: Nodal (left) and element (right) stress in the direction perpendicular to the load axis with elements of nominal size of 1.5 mm and 4-noded plane elements.
Element size
1.5 mm 5.5 mm
Table 6.1: Variaton on the percent error energy norm (SEP C) as a function of the type of element and the element size.
MME
Chapter 6. Postprocessing
93
Reminder: Reporting results data Appart from reviewing the results and assess their quality, reporting the obtained results is also an important task. This can be done in a simple way by capturing an image of the display or capturing a list. In ANSYST M this can be done in two ways: 1. PlotCtrls>Write Metale and select Standard Color or Invert White/Black. 2. PlotCtrls>Capture Image and print the captured image, either to a le or in paper. 3. PlotCtrls>Hard Copy and either send the captured image to a printer or save it with a image le extension (i.e. BMP Postscript, TIFF, JPEG, PNG). , Moreover, in the case of ANSYST M , there is also the option to elaborate a report including images, tables, etc. (File>Report Generator). For the specic case of ANSYST M , but also for other commercial Finite Element codes, the reviewing of the different results data can be more convenient by using the Graphical User Interface (GUI) than using commands. For instance, you can use the Results Viewer tool General Postproc>Results Viewer to plot contour plots, list nodal or element quantities, easily change substeps and animate the results among many other utilities. Try to repeat the previous examples using the GUI or Results Viewer and decide which method is more convenient for you.
6.6
Suggested problems
Problem 6.1. Reconsider the problem in Ex. 2.4. Recalculate it applying symmetry conditions on the meshes listed below. Evaluate the energy error norm obtained for the three different meshes used. Analyse the difference on the nodal stresses and element stresses.
1. A mesh using 10 quadrilateral elements in the vertical direction. 2. A mesh using 80 quadrilateral elements in the vertical direction. 3. A free mesh using 10 triangle elements in the vertical direction.
Problem 6.2. Reconsider the problem in Ex. 2.4. Calculate the principal stresses using ETABLE commands and considering plane stress, see equation 6.16. Compare these results whith the principal stresses obtained directly from the program. Calculate the relative error between them.
1 =
x + y + 2
x y 2
2 2 + xy
(6.16)
2 =
x + y 2
x y 2
2 2 + xy
MME
Chapter 7
In the Finite Element method the relationships between the different degrees of freedom (DOF) of the nodes in the mesh are dened by the elements. This is represented by the stiffness matrix as expressed in the next equation:
F =K
(7.1)
where K is the stiffness matrix, is the DOF vector and F is the vector containing the equivalent external loads. However, some degree-of-freedom related features such as rigid regions, structural joints, sliding symmetry boundaries, or periodic conditions, cannot be described by means of elements. A general option which most FE commercial codes offer is the possibility of dening special associations, or internodal connections between nodal DOF in ways that elements are not able. One possibility is to use equalities between a DOF of a set of nodes (coupling or linking DOF). Another option is to dene equations between different DOF (constraint equations). Both techniques add equations to equation (7.1).
7.1.1
Coupling a degree of freedom of a set of nodes imposes that the DOF results calculated for one member of the set must be the same for all the members in the set for this particular coupled DOF. Coupling can be used to model various joint and hinge effects. A more general form of coupling can be done with constraint equations (see section 7.1.2). For structural analyses, the DOF available for a node depend upon the DOF associated to the element type. For example, degrees of freedom available with two-dimensional beam elements are UX, UY, and ROTZ only. A set of coupled nodes in a nodal coordinate direction are forced to take the same displacement (or rotation) in the specied direction. Coupling operates in the nodal coordinate system of each node coupled. Therefore, you should usually keep the nodal coordinate systems consistent. In commercial codes, one of the DOF of the coupled set is the reference (or prime) DOF (in ANSYS is the rst DOF on the coupled set). All other degrees of freedom in the coupled sets are 95
96
eliminated from the solution matrices by their relationship to the reference DOF. For this reason, DOF with specied constraints (grounded DOF) should not be included in a coupled set and the coupled DOF must not appear in more than one coupled set.
Hint 31: Useful ANSYS commands to dene or modify Coupled DOF sets Command CP CPNGEN CPLIST CPDELE Main Parameters nset,label,nd1,nd2,nd3,... nset,label,n1,n2,ninc cp1,cp2,cpinc cp1,cp2,cpinc
Denes (or adds to) a set of coupled DOF Denes (or adds to) a set of coupled DOF List coupled set DOF Delete coupled set DOF
Example 7.1. Model a structure that consists of two pipes of 10 mm in diameter and 2 mm thick and a steel bar of cross-sectional area 20 mm2 . One of the beams has a distributed load of 5 N/mm2 on its edge. The material is steel with E = 210 GPa and = 0.3. The two beams are joined by a pin. Model this problem using beam type elements for beams and link bar elements for the bar.
200
Solution to Example 7.1. The joint between the two beam-pipes can be solved using a common node because the beam element have rotational z-axis DOF. For this reason, the coupling DOF of UX and UY is used there, dening a cylindrical joint in this point. The ANSYST M command sequence for this example is listed below. You can either type these commands on the command window, or you can type them on a le, then, on the command window enter /input, le, ext.
FINISH /CLEAR /TITLE, Pinned structure /PREP7 ET,1,BEAM188 KEYOPT,1,3,3 ET,2,LINK180 SECTYPE,1,BEAM,CTUBE SECOFFSET,CENT SECDATA,3,5 R,2,20 MP,EX,1,210e3 MP,PRXY,1,0.3 !start PRE-PROCESSOR MODULE !element type #1: BEAM188 (6 DoF) !element #1 keyoption 3=3 (cubic shape function) !element type #2: 3D truss (3 DoF) !section #1=hollow tube !centered section (offset=0) !section internal and external radius !geometric properties bar: cross-section !material #1: Youngs modulus !material #1: Poisson modulus
MME
97
N,1,-100 N,11,0,50 N,12,0,50 N,22,100 FILL,1,11 FILL,12,22 CP,1,UX,11,12 CP,2,UY,11,12 TYPE,1 REAL,1 E,1,2 EGEN,10,1,1 E,12,13 EGEN,10,1,11 TYPE,2 REAL,2 E,6,17 FINISH
!define node #1 on coordinates !pinned node of left steel pipe !pinned node of right steel pipe !define nodes between #1 and #11 !define coupling DOF #1: UX direction node set #11 and #12 !define coupling DOF #2: UY direction node set #11 and #12 !select BEAM element type (#1) and real constant set #1 !define BEAM elements
!select TRUSS element type (#2) and real constant set #2 !define TRUSS elements for rope
/SOLU !start SOLUTION MODULE ANTYPE,STATIC D,1,UX,,,,,UY,UZ,ROTX,ROTY !displacement X and Y for node #1 D,22,UX,,,,,UY,UZ,ROTX,ROTY !displacement X and Y for node #22 SFBEAM,1,2,PRES,5 !apply distributed load on element #2 !repeats 9 times more the previous command incrementing +1 number element *REPEAT,10,1 /PBC,ALL !show BCs when solve SOLVE !solve current load state FINISH /POST1 PLDISP,1 /ESHAPE,1,1 PLNSOL,S,X !start POST-PROCESSOR MODULE !plot deformed shape !contour plot of stress X distribution
Observe that the structure deforms in such a way that the pinned-node remains part of the two pipes, imposing common displacements at this point, but allowing them to rotate independently from each other.
Forces applied to coupled nodes (in the coupled DOF direction) will be summed up and applied to the reference DOF. Output forces are also summed at the reference DOF. A set of coupled nodes which are not coincident, or which are not along the line of the coupled displacement direction, may produce an applied moment which will not appear in the reaction forces. If the structure rotates, a moment may be produced in the coupled set in the form of a force couple. This moment is in addition to the real reaction forces and may make it appear that moment equilibrium is not satised by just the applied forces and the reaction forces.
Example 7.2. A cantilever T-beam is loaded with a vertical 1 kN force distributed on the end section (see Fig. 7.2). The T-beam is made of aluminum (E =69 GPa and = 0.3) with plates 2.2 mm thick. Obtain the deformed shape and the stress in the Z direction. Solution to Example 7.2. A coupling of vertical DOF set is applied on the nodes of the end section. In this way, applying the total vertical force on one of the nodes of the section is equivalent to apply a distributed force. The ANSYST M command sequence for this example is listed below. You can either type these commands on the command window, or you can type them on a le, then, on the command window enter /input, le, ext.
MME
98
FINISH /CLEAR /TITLE, Cantilever T-beam /PREP7 ET,1,SHELL181 KEYOPT,1,3,2 MP,EX,1,69e3 MP,PRXY,1,0.3 R,1,2.2 K,1,0,0,0 K,2,0,20,0 K,3,-15,0,0 K,4,15,0,0 KGEN,3,1,4,1,,,80,4 K,10,0,5,160 A,1,2,6,5 A,1,3,7,5 A,4,1,5,8 A,5,6,10,9 A,5,7,11,9 A,8,5,9,12 ESIZE,5 MSHKEY,1 AMESH,ALL NSEL,S,LOC,Z,160 CP,1,UY,ALL FINISH /SOLU ANTYPE,STATIC NSEL,S,LOC,Z,0 D,ALL,ALL NSEL,ALL NODE1=NODE(0,0,160) F,NODE1,FY,-1e3 SOLVE FINISH /POST1 PLDISP,1 /ESHAPE,1,1 PLNSOL,S,Z !start PRE-PROCESSOR MODULE !element type #1: SHELL181 (6 DoF) !element #1: keyoption 3=2 (full integration incompatible modes) !material #1: Youngs modulus !material #1: Poisson modulus !real constant set #1: 2.2 mm thickness !defines coordinates keypoint #1
!generates keypoint #5 to #12 !redefines keypoint #10 !defines area #1 !defines area #2 !defines area #3 !defines area #4 !defines area #5 !defines area #6 !set 5 mm elements size !mesh all areas !select nodes on section z=160 !coupling UY on all selected nodes
!set reference node NODE1 on coordinates 0,0,160 !apply total force of 1 kN on z=160 !solve current load state
!start POST-PROCESSOR MODULE !plot deformed shape !contour plot of stress Z distribution
MME
99
7.1.2
Linear constraint equations provide a more general way of relating degree of freedom values than it is possible with simple coupling. Constraint equations must have the form:
A=
i=1
i Ui
(7.2)
where A is a constant, Ui is the degree of freedom of term i, i is the coefcient of term i, and n is the number of terms in the equation.
The rst unique degree of freedom in the equation is eliminated in terms of all other degrees of freedom in the equation. A unique degree of freedom is one which is not specied in any other constraint equation, coupled node set, specied displacement set, or master degree of freedom set. It is recommended that the rst term of the equation be the degree of freedom to be eliminated. The rst term of the equation cannot contain a master degree of freedom, and no term can contain coupled degrees of freedom. The same degree of freedom may be specied in more than one equation but care must be taken to avoid over-specication (over-constraint).
The degrees of freedom specied in the equation (i.e., UX, UY, ROTZ, etc.) must also be included in the model (as determined from the element types [ET]). Also, each node in the equation must be dened on an element (any element type containing that degree of freedom will do).
Hint 32: Useful ANSYS commands to dene constraint equations Command CE CELIST CEDELE Main Parameters neq,A,nd1,lab1,coeff1,... ce1,ce2,ceinc ce1,ce2,ceinc
Denes (or adds to) a set of coupled DOF List constraint equation Delete constraint equation
Example 7.3. A composite material has two different components distributed in a regular array (see Figure 7.3). The matrix material has a Youngs modulus of 5 GPa, while for the rectangular particles is 150 GPa. Model one RVE (Representative Volume Element) of this geometry with appropriate periodic conditions. Note, take a = b = 5 mm.
MME
100
Figure 7.3: RVE example. Hint 33: Two-dimensional periodic conditions for a rectangular RVE In an rectangular RVE, such as shown in Figure 7.3, four faces (x1 = a and x2 = b) and the four edges dened by these faces need to be considered to dene the constraint equations to apply the periodic conditions. On the periodic pair of faces x1 = a and x2 = b, the constrain equations are
u1 (a, x2 ) u1 (a, x2 ) 2a0 = 0 11 u2 (a, x2 ) u2 (a, x2 ) 2b0 = 0 21 u1 (x1 , b) u1 (x1 , b) 2a0 = 0 12 u2 (x1 , b) u2 (x1 , b) 2b0 = 0 22
x2 (b, +b)
(7.3)
x1 (a, +a)
(7.4)
The intersection of planes x1 = a and x2 = b dene two pairs of periodic edges, for which the following equations apply,
u1 (+a, +b) u1 (a, b) 2a0 2b0 = 0 12 11 u2 (+a, +b) u2 (a, b) 2a0 2b0 = 0 22 21 u1 (+a, b) u1 (a, +b) 2a0 + 2b0 = 0 21 12 u2 (+a, b) u2 (a, +b) 2a0 + 2b0 = 0 22 21
(7.5)
(7.6)
Solution to Example 7.3. The ANSYST M command sequence for this example is listed below. You can either type these commands on the command window, or you can type them on a le, then, on the command window enter /input, le, ext.
FINISH /CLEAR /TITLE, Square RVE with two materials /PREP7 a=5 b=5 !start PRE-PROCESSOR MODULE !geometry of the cell
MME
101
a1=3*(a/5) b1=2*(b/5) a2=1*(a/5) b2=1*(b/5) epxx=0.001 epyy=0.000 epxy=0.001 ET,1,PLANE182 KEYOPT,1,1,3 MP,EX,1,5000 MP,PRXY,1,0.3 MP,EX,2,150000 MP,PRXY,2,0.3 BLC5,0,0,(2*a),(2*b) LESIZE,ALL,,,10 TYPE,1 MAT,1 AMESH,1
!horizontal strain !vertical strain !shear strain !element type !element type !material #1: !material #1: !material #2: !material #2: #1: PLANE182 (2 DoF) #1: keyoption 1=3 (simplified enhanced strain formulation) Youngs modulus Poisson modulus Youngs modulus Poisson modulus
!mesh area #1 !select nodes material #2 !select elements material #2 !set material #2 !reselect all items !plot elements
NSEL,S,LOC,X,(-(a1+(a/100))+a2),((a1+(a/100))+a2) NSEL,R,LOC,Y,(-(b1+(b/100))+b2),((b1+(b/100))+b2) ESLN,S,1 EMODIF,ALL,MAT,2 ALLSEL,ALL /PNUM,MAT,1 EPLOT !periodic conditions of vertex (2 couple of vertex) CE,1,(epxx*(a*2))+(epxy*(b*2)),1,UX,-1,12,UX,1 CE,2,(epxx*(a*2))-(epxy*(b*2)),22,UX,-1,2,UX,1 CE,3,(epyy*(b*2))+(epxy*(a*2)),1,UY,-1,12,UY,1 CE,4,(epyy*(b*2))-(epxy*(a*2)),2,UY,-1,22,UY,1
!periodic conditions of vertical sides (9 couple of nodes) CE,11,(epxx*(a*2)),32,UX,-1,21,UX,1 !uniaxial strain XX CE,12,(epxx*(a*2)),33,UX,-1,20,UX,1 CE,13,(epxx*(a*2)),34,UX,-1,19,UX,1 CE,14,(epxx*(a*2)),35,UX,-1,18,UX,1 CE,15,(epxx*(a*2)),36,UX,-1,17,UX,1 CE,16,(epxx*(a*2)),37,UX,-1,16,UX,1 CE,17,(epxx*(a*2)),38,UX,-1,15,UX,1 CE,18,(epxx*(a*2)),39,UX,-1,14,UX,1 CE,19,(epxx*(a*2)),40,UX,-1,13,UX,1 CE,21,(epxy*(a*2)),32,UY,-1,21,UY,1 CE,22,(epxy*(a*2)),33,UY,-1,20,UY,1 CE,23,(epxy*(a*2)),34,UY,-1,19,UY,1 CE,24,(epxy*(a*2)),35,UY,-1,18,UY,1 CE,25,(epxy*(a*2)),36,UY,-1,17,UY,1 CE,26,(epxy*(a*2)),37,UY,-1,16,UY,1 CE,27,(epxy*(a*2)),38,UY,-1,15,UY,1 CE,28,(epxy*(a*2)),39,UY,-1,14,UY,1 CE,29,(epxy*(a*2)),40,UY,-1,13,UY,1 !shear strain XY
!periodic conditions of horizontal sides (9 couple of nodes) CE,32,(epyy*(b*2)),3,UY,-1,31,UY,1 !uniaxial strain yy CE,33,(epyy*(b*2)),4,UY,-1,30,UY,1 CE,34,(epyy*(b*2)),5,UY,-1,29,UY,1 CE,35,(epyy*(b*2)),6,UY,-1,28,UY,1 CE,36,(epyy*(b*2)),7,UY,-1,27,UY,1 CE,37,(epyy*(b*2)),8,UY,-1,26,UY,1 CE,38,(epyy*(b*2)),9,UY,-1,25,UY,1 CE,39,(epyy*(b*2)),10,UY,-1,24,UY,1 CE,40,(epyy*(b*2)),11,UY,-1,23,UY,1 CE,43,(epxy*(b*2)),3,UX,-1,31,UX,1 CE,44,(epxy*(b*2)),4,UX,-1,30,UX,1 CE,45,(epxy*(b*2)),5,UX,-1,29,UX,1 !shear strain XY
MME
102
CE,46,(epxy*(b*2)),6,UX,-1,28,UX,1 CE,47,(epxy*(b*2)),7,UX,-1,27,UX,1 CE,48,(epxy*(b*2)),8,UX,-1,26,UX,1 CE,49,(epxy*(b*2)),9,UX,-1,25,UX,1 CE,50,(epxy*(b*2)),10,UX,-1,24,UX,1 CE,51,(epxy*(b*2)),11,UX,-1,23,UX,1 FINISH /SOLU NSEL,S,LOC,X,0 NSEL,R,LOC,Y,0 D,ALL,ALL ALLSEL SOLVE FINISH /POST1 PLDISP,1 PLNSOL,U,SUM PRNSOL,S !start SOLUTION MODULE !select central node !central node constrained !solve current load state
!start POST-PROCESSOR MODULE !plot deformed shape !total displacement contour !list stress state of nodes
See how the obtained solution is symmetric in different directions for different parameters. To simplify the denition of the periodic conditions, they can be dened by a loop, see the following command sequence.
FINISH /CLEAR /TITLE, Square RVE with two materials /PREP7 a=5 b=5 a1=3*(a/5) b1=2*(b/5) a2=1*(a/5) b2=1*(b/5) epxx=0.001 epyy=0.000 epxy=0.001 ET,1,PLANE182 KEYOPT,1,1,3 MP,EX,1,5000 MP,PRXY,1,0.3 MP,EX,2,150000 MP,PRXY,2,0.3 BLC5,0,0,(2*a),(2*b) LESIZE,ALL,,,10 TYPE,1 MAT,1 AMESH,1 !start PRE-PROCESSOR MODULE !geometry of the cell !geometry of the material #2 !position of center of the material
!horizontal strain !vertical strain !shear strain !element type !element type !material #1: !material #1: !material #2: !material #2: #1: PLANE182 (2 DoF) #1: keyoption 1=3 (simplified enhanced strain formulation) Youngs modulus Poisson modulus Youngs modulus Poisson modulus
!mesh area #1 !select nodes material #2 !select elements material #2 !set material #2 !reselect all items !plot elements
NSEL,S,LOC,X,(-(a1+(a/100))+a2),((a1+(a/100))+a2) NSEL,R,LOC,Y,(-(b1+(b/100))+b2),((b1+(b/100))+b2) ESLN,S,1 EMODIF,ALL,MAT,2 ALLSEL,ALL /PNUM,MAT,1 EPLOT !periodic conditions of vertex (2 couple of vertex)
MME
103
CE,1,(epxx*(a*2))+(epxy*(b*2)),1,UX,-1,12,UX,1 CE,2,(epxx*(a*2))-(epxy*(b*2)),22,UX,-1,2,UX,1 CE,3,(epyy*(b*2))+(epxy*(a*2)),1,UY,-1,12,UY,1 CE,4,(epyy*(b*2))-(epxy*(a*2)),2,UY,-1,22,UY,1 !periodic conditions j=0 *DO,i,1,9,1 CE,i+10,(epxx*(a*2)),i+31,UX,-1,21-j,UX,1 CE,i+20,(epxy*(a*2)),i+31,UY,-1,21-j,UY,1 CE,i+31,(epyy*(b*2)),i+2,UY,-1,31-j,UY,1 CE,i+42,(epxy*(b*2)),i+2,UX,-1,31-j,UX,1 j=j+1 *ENDDO FINISH /SOLU NSEL,S,LOC,X,0 NSEL,R,LOC,Y,0 D,ALL,ALL ALLSEL SOLVE FINISH /POST1 PLDISP,1 PLNSOL,U,SUM PRNSOL,S !start SOLUTION MODULE !select central node !central node constrained !solve current load state
!start POST-PROCESSOR MODULE !plot deformed shape !total displacement contour !list stress state of nodes
7.2
Submodelling
As was introduced in Chapter 3, nite-element software users generally establish in the meshing process some equilibrium between the available computational resources (where coarser meshes are preferred) and the needs of the solved problem (where ner meshes may be needed). Moreover, some structural components may exhibit nearly constant stress through its dominium but a small area showing large stress concentration. Submodelling is a useful tool to analyze parts of a model which exhibit stress concentrations or which need a detailed geometrical modelling. Submodelling is also known as the cut boundary displacement method or the specied boundary displacement method. The cut boundary is the boundary of the submodel which represents a cut through the global model. Displacements calculated on the cut boundary of the coarse model are specied as boundary conditions for the submodel. Submodelling is based on St. Venants principle, which states that if an actual distribution of forces is replaced by a statically equivalent system, the distribution of stress and strain is altered only near the regions of load application. The principle implies that stress concentration effects are localized around the concentration; therefore, if the boundaries of the submodel are far enough away from the stress concentration, reasonably accurate results can be calculated in the submodel. Aside from the obvious benet of giving more accurate results in a region of your model, the submodelling technique has other advantages: It reduces, or even eliminates, the need for complicated transition regions in solid nite element models.
MME
104
It enables you to experiment with different designs for the region of interest (different llet radii, for example). Some restrictions for the use of submodelling are: It is valid only for solid elements and shell elements. The principle behind submodelling assumes that the cut boundaries are enough far away from the stress concentration region. You must verify that this assumption is adequately satised. In the following sections you can nd a summary of Chapter 9 of ANSYS 13 Help, regarding submodelling.
7.2.1
In many cases, the coarse model does not need to include local geometric details. However, the nite element mesh must be ne enough to produce a reasonably accurate degree of freedom solution. This is important because the results of the submodel are almost entirely based on interpolated degree of freedom results at the cut boundary.
7.2.2
The submodel
In the submodel the same element type (solid or shell) that was used in the global model should be employed1 . Also, the same element real constants (such as shell thickness) and material properties should be introduced. Specify appropriate node rotations. Node rotation angles on cut boundary nodes should not be changed after they have been written to the node le in interpolation step 1. Command: NROTAT 2 . The presence or absence of node rotation angles in the coarse model has no effect upon the submodel.
7.2.3
Submodelling procedure
The basic steps for any submodelling procedure are: 1. Perform Cut Boundary Interpolation. This is the key step in submodelling. You identify the nodes along the cut boundaries, and the nite element program calculates the DOF values (displacements) at those nodes by interpolating results from the global model. For each node of the submodel along the cut boundary, the program uses the appropriate element from the coarse mesh to determine the DOF values. These values are then interpolated onto the cut boundary nodes using the element shape functions.
Shell-to-solid may also be used but will not be covered in this chapter. See ANSYS help. Be aware that node rotation angles might be changed by application of nodal constraints [DSYM], by transfer of line constraints [SFL], or by transfer of area constraints [SFA], as well as by more obvious methods [NROTAT and NMODIF]
2 1
MME
105
Hint 34: Cut boundary interpolation in ANSYS. Global model Identify and write the cut boundary nodes of the submodel to a le (Jobname.NODE by default). You can do this in PREP7 by selecting nodes along the cut boundaries and then using one of these methods to write the nodes to a le with command NWRITE (you perform temperature interpolation see ANSYS help). To restore the full set of nodes, use command ALLSEL. To write the database to Jobname.DB, use command SAVE. To do the cut boundary interpolation (and the temperature interpolation), the database must contain the geometry for the coarse model. Therefore, you must resume the database use command RESUME, making sure to identify the name of the coarse model database le. Enter POST1 with command /POST1. Interpolation can only be performed in POST1. Point to the coarse results le with command FILE. Read in the desired set of data from the results le with command SET. Initiate cut boundary interpolation with command CBDOF. By default, the CBDOF command assumes that the cut boundary nodes are on le Jobname.NODE. The ANSYS program will then calculate the cut boundary DOF values and write them in the form of D commands to the le Jobname.CBDO. All interpolation work is now done, so leave POST1 (FINISH) and restore the submodel database (RESUME).
2. Analyze the Submodel. In this step, you dene the analysis type and analysis options, apply the interpolated DOF values (and temperatures), dene other loads and boundary conditions, specify load step options, and obtain the submodel solution. It is important that on the submodel any other loads and boundary conditions that existed on the coarse model are introduced as well. Examples are symmetry boundary conditions, surface loads, inertia forces (such as gravity), concentrated force loads, etc.
MME
106 Hint 35: ANSYS commands for the Submodel The rst step is to enter SOLUTION (/SOLU).
Dene the appropriate analysis type (usually static) and analysis options. To apply the cut boundary DOF constraints, simply read in the le of D commands (created by CBDOF) with command /INPUT,,CBDO. Specify what DOF constraint values and nodal body force loads are to be accumulated with commands DCUM,ADD and BFCUM,,ADD. Be sure to reset the DCUM and BFCUM commands to their default status before proceeding. Then specify load step options (such as output controls) and initiate solution calculations using Command(s): SOLVE. After the solution is obtained, leave SOLUTION (FINISH).
3. Verify the Distance Between the Cut Boundaries and the Stress Concentration. The nal step is to verify that the cut boundaries of the submodel are far enough away from the concentration. You can do this by comparing results (stresses, magnetic ux density, etc.) along the cut boundaries with those along the corresponding locations of the coarse model. If the results are in good agreement, it indicates that proper cut boundaries have been chosen. Otherwise, you will need to recreate and reanalyze the submodel with different cut boundaries further away from the region of interest.
Example 7.4. Model a variation of the hydraulic dam considered in Ex. 2.5 using the submodelling procedure according to what is shown in Figure 7.4. Start with a simple model with geometric parameters h1 = 10 m, h2 = 5 m, L = 2 m and W = 1 m and consider that the dam is subjected to a uniform 10 kg/cm2 pressure at its deepest end. Solution to Example 7.4. The ANSYST M command sequence for this example is listed below. You can either type these commands on the command window, or you can type them on a le, then, on the command window enter /input, le, ext.
FINISH /CLEAR /TITLE, Dam submodeling - Global Model /FILNAME,Dam_Global /PREP7 h1=10000 h2=5000 L=8000 W=2000 p=1 ET,1,PLANE182 KEYOPT,1,1,3 KEYOPT,1,3,2 MP,EX,1,27000 MP,PRXY,1,0.25 K,1,0,0 !start PRE-PROCESSOR MODULE !total wall height !height !length !width !pressure in N/mm2 !element type #1: PLANE182 (2 DoF) !element type #1: keyoption 1=3 (simplified enhanced strain formulation) !element type #1: keyoption 3=2 (plane strain) !Young Modulus !Poissons ratio !define geometry by keypoints
MME
107
W Submodel h2
h1
Cut Boundary
K,2,L,0 K,3,L,(h1-h2)/2 K,4,L,h1-2*h2/3 K,5,L,h1 K,6,L-W,h1 K,7,L-W,h1-2*h2/3 K,8,L-W,h1-h2 K,9,(L-W)/2,(h1-h2)/2 K,10,L-W/2,(h1-h2)/2 K,11,L-W/2,h1-2*h2/3 L,1,2,40 L,2,3,20 L,3,4,20 L,4,5,20 L,5,6,10 L,6,7,20 L,7,8,10 L,8,9,20 L,9,1,20 L,9,10,15 L,10,11,20 L,11,7,5 !define lines by keypoints
AL,1,2,3,4,5,6,12,11,10,9 !create areas from lines AL,10,11,12,7,8 MSHKEY,2 !use mapped meshing if possible AMESH,ALL !mesh all areas SAVE !save the file GLOBAL.db FINISH /SOLU DL,1,1,ALL LSEL,S,LINE,,2,4,1 SFL,ALL,PRES,p ALLSEL SOLVE FINISH /POST1 PLDISP,1 PLNSOL,S,EQV SAVE,ALL This le can be found at: !start SOLUTION MODULE !fix the line 1 of area 1 !select lines 2,4,1 !apply the pressure !select all
!start POST-PROCESSOR MODULE !plot deformed shape !show Von Mises stress !save solution
MME
108
ftp://amade.udg.edu/mms/PrePostFE/input_files/T7/PrePostFE_Ex704.dat
Observe that a stress concentration is produced in the corner where the straight part from the dam and its basis unite. Example 7.5. Introduce a round corner, that is, a more detailed geometry, in the cross-section of the previous hydraulic dam and analyze the results. Set the llet radius to 2 m. Solution to Example 7.5. The ANSYST M command sequence for this example is listed below. You can either type these commands on the command window, or you can type them on a le, then, on the command window enter /input, le, ext.
FINISH /CLEAR /TITLE, Dam submodeling - Submodel /FILNAME,Dam_Submodel /PREP7 h1=10000 h2=5000 L=8000 W=2000 divl=100 RAD=2000 ET,1,PLANE182 KEYOPT,1,1,3 KEYOPT,1,3,2 MP,EX,1,27000 MP,PRXY,1,0.25 K,7,L-W,h1-2*h2/3 K,8,L-W,h1-h2 K,9,(L-W)/2,(h1-h2)/2 K,10,L-W/2,(h1-h2)/2 K,11,L-W/2,h1-2*h2/3 !start PRE-PROCESSOR MODULE !total wall height !height !length !width !element size !fillet radius !element type #1: PLANE182 (2 DoF) !element type #1: keyoption 1=3 (simplified enhanced strain formulation) !element type #1: keyoption 3=2 (plane strain) !Young Modulus !Poissons ratio !define geometry by keypoints
L,7,8 !define lines by keypoints L,8,9 L,9,10 L,10,11 L,11,7 LFILLT,1,2,RAD !define fillet between intersection lines #1 and #2 LESIZE,ALL,RAD*2*3.14159/96 !the divisions of the fillet radius AL,1,2,3,4,5,6 MSHKEY,2 AMESH,ALL NROTAT,ALL LSEL,S,LINE,,3,5,1 NSLL,S NWRITE ALLSEL SAVE RESUME,Dam_Global,DB /POST1 FILE,Dam_Global SET CBDOF FINISH RESUME,Dam_Submodel,DB /SOLU DCUM,ADD /INPUT,,CBDO !define area by lines !use mapped meshing if possible !mesh all areas !rotate nodal coordinate sys into local !select cut boundary lines !select nodes on lines !write node list and coordinates into file SUBMODEL.node !select all !save the file SUBMODEL.db !resume Global Model previously saved !start POST-PROCESSOR MODULE !point to the global model results file !defines the data set to be read from the results file GLOBAL.db !perform cut boundary interpolation (file SUBMODEL.cbdo)
!resume the file SUBMODEL.db !start SOLUTION MODULE !cumulative loads !input BC from global model (file SUBMODEL.cbdo)
MME
109
!start POST-PROCESSOR MODULE !plot deformed shape !show Von Mises stress
Note that the submodel allows for a more detailed analysis of the corner or llet area. In fact, the maximum Von Mises stress in the submodel is about 23 MPa, while in the global model it is approximated to about 31 MPa. Example 7.6. Use ANSYST M to model the steel trapezoidal cantilever beam with holes shown in Figure 7.5. The beam is subjected to a uniform pressure of 20 MPa applied on its top edge. The diameter of the holes is 7 mm. Use the submodel technique to obtain the Von Mises stress in the area of the holes. Consider a plane strain model.
20 MPa 50 350
510
55
80 750
Solution to Example 7.6. First, only the general beam, without the small holes, is analyzed to obtain the global model. The ANSYST M command sequence for this example is listed below. You can either type these commands on the command window, or you can type them on a le, then, on the command window enter /input, le, ext.
FINISH /CLEAR /TITLE, Cantilever beam submodeling - Global Model /FILNAME,CANTILEVER_GLOBAL /PREP7 h1=350 h2=50 L=750 d=7 top_off=10 p=20 holes1x=510 holes2x=590 sep1=55 sep2=35 xmin=490 !start PRE-PROCESSOR MODULE !cantilevered tip height !free tip height !length !hole diameter !offset to top holes !pressure !x coord left holes !x coord right holes !vertical separation left holes !vertical separation right holes !x coordinates of submodel
MME
35 35
10
55
110
xmax=610 ET,1,PLANE182 KEYOPT,1,1,3 KEYOPT,1,3,2 MP,EX,1,210000 MP,PRXY,1,0.3 K,1,0,0 K,2,L,0 K,3,L,-h2 K,4,0,-h1 L,1,2,40 L,2,3,20 L,3,4,40 L,4,1,20 AL,1,2,3,4 MSHKEY,2 AMESH,ALL SAVE FINISH /SOLU DL,4,1,ALL LSEL,S,LINE,,1 SFL,ALL,PRES,p ALLSEL SOLVE FINISH /POST1 PLDISP,1 PLNSOL,S,EQV SAVE,ALL
!element type #1: PLANE182 (2 DoF) !element type #1: keyoption 1=3 (simplified enhanced strain formulation) !element type #1: keyoption 3=2 (plane strain) !Young Modulus !Poissons ratio !define geometry by keypoints
!create areas from lines !use mapped meshing if possible !mesh all areas !save the file as CANTILEVER_GLOBAL.db
!start SOLUTION MODULE !cantilever boundary conditions !select line 1 !apply pressure to line 1 !select all
!start POST-PROCESSOR MODULE !plot deformed shape !show Von Mises stress !save the solution
Afterwards, the hole region is analysed by using the submodelling technique incorporating the results of the global model. The ANSYST M command sequence for this example is listed below. You can either type these commands on the command window, or you can type them on a le, then, on the command window enter /input, le, ext.
FINISH /CLEAR /TITLE, Cantilever beam submodeling - Submodel /FILNAME,CANTILEVER_SUBMODEL /PREP7 h1=350 h2=50 L=750 d=7 top_off=10 p=20 holes1x=510 holes2x=590 sep1=55 sep2=35 xmin=490 xmax=610 ET,1,PLANE182 KEYOPT,1,1,3 KEYOPT,1,3,2 !start PRE-PROCESSOR MODULE !cantilevered tip height !free tip height !length !hole diameter !offset to top holes !pressure !x coord left holes !x coord right holes !vertical separation left holes !vertical separation right holes !x coordinates of submodel
!element type #1: PLANE182 (2 DoF) !element type #1: keyoption 1=3 (simplified enhanced strain formulation) !element type #1: keyoption 3=2 (plane strain)
MME
111
MP,EX,1,210000 MP,PRXY,1,0.3
K,1,xmin,0 !define geometry by keypoints K,2,xmax,0 K,3,xmax,-h1+xmax*(h1-h2)/L K,4,xmin,-h1+xmin*(h1-h2)/L L,1,2 L,2,3 L,3,4 L,4,1 !define lines by keypoints
AL,1,2,3,4 CYL4,holes1x,-top_off,d/2 CYL4,holes1x,-(top_off+sep1),d/2 CYL4,holes1x,-(top_off+2*sep1),d/2 CYL4,holes2x,-top_off,d/2 CYL4,holes2x,-(top_off+sep2),d/2 CYL4,holes2x,-(top_off+2*sep2),d/2 ASEL,S,AREA,,1,7 ASBA,1,ALL LESIZE,ALL,d/2 MSHKEY,2 AMESH,ALL NROTAT,ALL LSEL,S,LINE,,2,4,2 NSLL,S NWRITE ALLSEL SAVE
!define the element size !use mapped meshing if possible !mesh all areas !rotate nodal coordinate sys into local !select cut boundary lines !select nodes on lines !write node list into file !select all
!start POST-PROCESSOR MODULE !point to the global model results file !data set to be read from the results file !perform cut boundary interpolation
RESUME,CANTILEVER_SUBMODEL,DB /SOLU LSEL,S,LINE,,1 SFL,ALL,PRES,p DCUM,ADD /INPUT,,CBDO DCUM,REPL SOLVE FINISH /POST1 PLDISP,1 PLNSOL,S,EQV
!resume Submodel
!start POST-PROCESSOR MODULE !plot deformed shape !show Von Mises stress
Observe that although the general tendency of the Von Mises stress is the same for the results of the global and submodel, in the latter stress concentrations can be observed around the holes, which cannot be captured by the global model.
MME
112
7.3
Suggested problems
Problem 7.1. Using the same geometry and model of Example 3.9, apply a force of 100 N in one of the holes producing a bending moment on the plate while the other hole is clamped. Use coupled DOF set to apply the force. Problem 7.2. Using similar code of Ex. 7.3 model a RVE representative of a masonry wall (see Figure 7.6). The bricks are 240 mm long and 45 mm thick. The vertical gap between bricks is 30 mm while the horizontal one is 60 mm. Use the results of the model to compute the homogenized Youngs Modulus of the masonry wall.
Problem 7.3. Reconsider the situation of Ex. 7.4. Solve the problem again with h1 = 30 m, h2 = 18 m, L = 6 m and W = 2.5 m. Notice that a new stress concentration appears. Construct a new submodel for this area and try to introduce a llet radius to avoid stress concentrations. Comment the obtained results for different values of the llet radius and the validity of the analysis. Problem 7.4. Model the simple structure of Figure 7.7. The material is steel and the parameters A = 11 m, B = 2.5 m and C = 3.5 m. The applied load is F = 10 N/m. Construct submodel for the area showing stress concentrations and try to introduce a llet radius to avoid large stress concentration.
A F B C
C
Figure 7.7: Submodel for a T-structure.
MME
Chapter 8
Many structural problems present nonlinear behaviours that have to be taken into account for a correct simulation. For example, in a cantilever beam subjected to an end load (see Figure 8.1), as the beams deects the distance between the load application point and the encastred area is reduced, consequently reducing the corresponding bending moment. Moreover, as the beam deects and rotates the effect of the load on the beam can change due to the relative orientation between beam and load. For instance, the bending behaviour of the system varies if the load has a predened direction, e.g. gravitational load, or the applied force follows the surface of the beam, e.g. pressure. Obviously, this effect is only important when large displacements are involved. Another example of structural nonlinear behaviour is the case of a wood shelf loaded with books. Even if the load is kept constant, as time passes, the deection of the shelf increases. The interaction of different parts of a model also corresponds to a nonlinear structural problem.
Figure 8.1: Geometric nonlinear behaviour in a cantilever beam. Actually, the three previous examples correspond to the three different types or categories of nonlinear structural behaviour: Geometric nonlinearities Material nonlinearities 113
In the rst case, the nonlinear structural behaviour is due to a change in the geometric conguration of the structure. This change is usually characterised by large displacement and/or rotations. Material nonlinearities are caused by nonlinear stress-strain relationships. These can be inuenced by different factors such as plasticity, temperature or creep. Finally, structures involving change of state can also results in a nonlinear structural behaviour: a cable can only transmit load under tension, a roller support is either in contact or not, etc. In all the cases, though, the load-deection curve shows a variation in structural stiffness, which is the basic characteristic of the nonlinear structural behaviour. Figure 8.2 shows characteristic load-deection curves for the three categories of nonlinear structural behaviour. In general, this nonlinear behaviour implies more convergence problems complicating the analysis and requiring more iterations and different methods and strategies to ensure a converged solution.
Figure 8.2: Characteristic nonlinear load-deections curves for: (a) geometric nonlinear behaviour, (b) material nonlinear behaviour and (c) change of state. In this chapter, the nonlinear structural behaviour caused by geometric nonlinearities are considered. Two different types of geometric nonlinearities are taken into account: Geometric nonlinearities due to large displacements and/or rotations Geometric nonlinearities due to buckling
8.2
Most of the structural analyses involve small displacements, rotations and deections resulting in null or small stiffness variations. However, some problems involving large displacements, rotations and deections result in stiffness variations. In these cases, a nonlinear analysis has to be carried out activating large displacements or geometric nonlinearities. In order to facilitate the convergence of the solution during the analysis, it is convenient to divide the application of load or displacement in different steps. The smaller the steps, the better the options for a converged solution. However, many load or displacements increments might result in excessive calculation times. Hint 36: Activating the geometric nonlinear effects In the majority of the commercial nite element codes, the geometric nonlinearities are taken into account by activating a nonlinear geometric effects option. In the case of ANSYST M this option is activated by using NLGEOM,ON.
MME
115
Example 8.1. Consider the cantilever beam shown in Figure 8.3. The end-load P = 100 kN is distributed along 50 mm in the longitudinal direction. The beam is made of steel (E = 207 GPa and = 0.3), is 2 m long, its base is 20 mm and its thickness is 100 mm. Compare the results obtained considering a linear behaviour of the system, a nonlinear behaviour with a constant direction of the applied load (gravity) and a nonlinear behaviour with the load always applied perpendicular to the surface of the beam.
Solution to Example 8.1. The cantilever beam in Figure 8.3 has been simulated with and without considering the geometric nonlinearities. The two different cases for the nonlinear system have been taken into account. In the rst case, the direction of the load remains the same during the whole simulation, gravity load. In the second case, the direction of the load varies accordingly to the deection of the beam in order to remain perpendicular to the load application area on the beam. Hint 37: Dening a Follower Force In ANSYST M an applied force always perpendicular to the surface of the element in a nonlinear problem has to be simulated dening the force as pressure. In other commercial nite element codes, such as ABAQUST M and MSC-MarcT M there is an option to indicate that the direction of the force varies according to the deection of the element, FOLLOWER FORCE.
As expected, the results of the simulation show a linear relationship between the applied load and the deection of the beam for the linear case. However, when the nonlinear geometric effects such as large displacements are taken into account, the resulting relationship between applied load and deection is not linear. Figure 8.4 summarises the load-deection curves for the three cases considered. Note that the applied load is the same for the three cases but the resulting deection varies depending on the assumptions. If the axial stress is considered, the linear model predicts an axial stress of 758.5 MPa, while the nonlinear model with gravity load 685 MPa and the nonlinear model with pressure 742 MPa. Note that in the linear case the results are exactly the same if the load is applied as a gravity load or as pressure. The ANSYST M command sequence for this example is listed below. You can either type these commands on the command window, or you can type them on a le, then, on the command window enter /input, le, ext.
FINISH !close all previous modules /CLEAR !clear all previous models /TITLE, Nonlinear cantilever beam /PREP7 !Parameters H=100 L=2000 !start PRE-PROCESSOR MODULE !beam height in mm !beam span in mm
MME
116
1.E+05 1 10 8.E+04 8 10
Load (N)
0 0.E+00
0 200 400 600 800
B=20 p=1e5
!beam base in mm !end-load !element type #1: BEAM188 (6 DoF) !element #1 keyoption 3=3 (cubic shape function) !section #1 = rectangular beam !centered section (offset=0) !section width and height
ET,1,BEAM188 KEYOPT,1,3,3 SECTYPE,1,BEAM,RECT SECOFFSET,CENT SECDATA,H,B MP,EX,1,207000 MP,PRXY,1,0.3 K,1,0,0 K,2,L,0 L,1,2 LESIZE,1,50 LMESH,1 FINISH
!Youngs modulus for material #1 !Poisson modulus for material #1 !generate key-points !define line !define element size
/SOLU !start SOLUTION MODULE ANTYPE,0 !static analysis NLGEOM,ON !activate geometric nonlinearities NSUBST,10,0,0 !perform the analysis in 10 sub-steps OUTRES,ALL,1 !write results for each sub-step TIME,1 !end time of the analysis NSEL,S,LOC,X,0 !encastred node D,ALL,ALL NSEL,S,LOC,X,L-50,L !apply end-load nodes ESLN,S,1 !select end elements !SFBEAM,ALL,1,PRES,p/50 !distributed end-load on beam element F,ALL,FY,-p/2 !concentrated end-load in 2 nodes (gravity load) ALLSEL !select all the entities SOLVE !solve the current load case FINISH /POST1 PLNSOL,U,Y /POST26 NSOL,2,41,U,Y,uy ABS,3,2,,,U_Y XVAR,3 RFORCE,4,1,F,Y,F_Y PLVAR,4 PRVAR,3,4 !start POST-PROCESSOR MODULE !plot vertical displacement !start HISTORY-PLOT MODULE !evolution of Y-displacement at node 41 !absolute value !use as X-axis !evolution of reaction force !plot reaction force vs. displacement !list displacement and reaction force
2.E+04 2 10
4 10 4.E+04
6.E+04 6 10
Deflection (mm)
MME
117
8.3
Buckling analysis
Buckling and geometric instabilities in structural parts are typical problems that exhibit geometric nonlinear effects. Actually, a structure subjected to compressive loads can present no lateral displacements until a critical load is reached, Pcr . Beyond this critical load, or bifurcation point, the structure deforms in such a way that lateral displacements occur. Depending on the conguration of the structure, the occurrence lateral displacements can imply the global instability of the structure. In other cases, the occurrence of these lateral displacements only imply a change of behaviour and the structure can still sustain load under the post-buckled conguration. Figure 8.5 schematically shows the possible theoretical load versus lateral displacement curves for a structure under buckling loads. The curve on the left corresponds to a stable structure that can sustain load after reaching the critical load as the structure tends to stiffen. On the contrary, the curve on the right is for a unstable structure where after reaching Pcr the structure can deform in such a way that the lateral displacement increases without increasing the applied load. Although in this case this behaviour occurs in one direction for the lateral displacement, it is not possible to ensure that the structure will deform on the other way.
Figure 8.5: Theoretical load versus lateral displacement curves for a structure under buckling loads. In general, two different techniques are usually available in all FEM codes for predicting the buckling load and buckling mode shape of a structure: nonlinear buckling analysis and eigenvalue (or linear) buckling analysis. Because the two methods can yield different results, it is necessary to understand the differences between them.
8.3.1
A real structure sustaining load can suffer buckling or become unstable in many ways and adopting different deformed shapes that depend on the applied load. For every buckling mode or deformed shape there is a critical load or bifurcation point. Eigenvalue buckling analysis can be used to predict the theoretical buckling load (or critical load or bifurcation point) of an ideal linear elastic structure. For instance, an eigenvalue buckling analysis of a column will match the classical Euler solution. However, real structures usually contain imperfections and nonlinearities that might cause the instability or buckling at a force lower than the theoretical critical load. Thus, eigenvalue buckling analysis often yields unconservative results and must be used carefully and just as a tentative analysis. The following equation summarises the calculation required to determine the eigenvalues of a structure:
MME
118
([K] + i [S]) i = 0
(8.1)
where [K] is the stiffness matrix, [S] is the stress stiffness matrix, i is the i-th eigenvalue and i is the i-th eigenvector of displacements. As the stress stiffness matrix is required for the calculation of the eigenvalues, this matrix has to be previously obtained. In the case of ANSYST M , this can be achieved by performing rst a static analysis activating the PSTRES option. The resulting critical loads are obtained multiplying the applied load per the resulting eigenvalues. It is a common practice to simulate the problem applying unit loads so the resulting eigenvalues directly correspond to the critical loads. The eigenvectors are normalized so that the largest component is 1.0. Thus, the stresses (when output) may only be interpreted as a relative distribution of stresses. Hint 38: Negative eigenvalues. In general, negative eigenvalues indicate that the applied load should be applied in the opposite direction to obtain the represented eigenvalue. However, in ANSYST M , if the rst eigenvalue closest to the shift point is negative (indicating that the loads applied in a reverse sense will cause buckling), the program will terminate.
Example 8.2. Compute the rst ten bifurcation loads of the prismatic bi-supported column shown in Figure 8.6 using the eigenvalue buckling analysis. The column is made of steel (E = 207 GPa and =0.3), its base is b = 200 mm, its thickness is h = 100 mm and the length is L = 5 m.
Solution to Example 8.2. Although the geometry of the column is prismatic with a thickness half of the base and the rst buckling modes will be in the same deformation plane (X-Y), it is possible that one or more of the ten rst eigenvalues correspond to a deformed shape in the other deformation plane (X-Z). Thus, a plane analysis could not take into account some buckling modes and a 3D analysis is required. The ANSYST M command sequence for this example is listed below. You can either type these commands on the command window, or you can type them on a le, then, on the command window enter /input, le, ext.
FINISH !close all previous modules /CLEAR !clear all previous models /TITLE, Prismatic column /PREP7 !Parameters H=100 L=5000 B=200 !start PRE-PROCESSOR MODULE !column thickness in mm !column span in mm !column base in mm
MME
119
p=1
!applied load !element type #1: BEAM188 (6 DoF) !element #1 keyoption 3=3 (cubic shape function) !section #1 = rectangular beam !centered section (offset=0) !section width and height
ET,1,BEAM188 KEYOPT,1,3,3 SECTYPE,1,BEAM,RECT SECOFFSET,CENT SECDATA,H,B MP,EX,1,207000 MP,PRXY,1,0.3 K,1,0,0 K,2,L,0 L,1,2,100 LMESH,1 FINISH /SOLU ANTYPE,0 PSTRES,ON D,1,UX,,,,,UY,UZ D,2,UY,,,,,UZ F,2,FX,-p SOLVE FINISH /SOLU ANTYPE,BUCKLE BUCOPT,SUBSP,10 MXPAND SOLVE /POST1 SET,LIST SET,FIRST PLDISP,1 SET,NEXT PLDISP,1
!Youngs modulus for material #1 !Poisson modulus for material #1 !generate key-points !define line and divisions !mesh the line
!start SOLUTION MODULE !static analysis !calculate stress stiffness matrix !fixed displacements node 1 !fixed displacements node 2 !apply load node 2
!start SOLUTION MODULE !buckling analysis !buckling options: subspace iteration method, 10 eigenvalues !expand all buckling modes
!start POST-PROCESSOR MODULE !list eigenvalues !set first eigenvalue !plot buckling mode !set next eigenvalue !plot buckling mode
As expected, the buckling mode that corresponds to a deformed shape in the X-Z plane is one of the rst modes which would not be considered in a 2-D analysis. Actually, this buckling mode is the third found, although the critical load coincides with that of the second buckling mode in the X-Y plane. Table 8.1 summarises the critical loads for the rst ten calculated eigenvalues. Table 8.1: Resulting eigenvalues for the bi-supported prismatic column. Eigenvalue Critical load (106 N) 1 1.36 2 5.43 3 5.43 4 12.1 5 21.4 6 21.4 7 33.2 8 47.3 9 47.3 10 63.6
Example 8.3. The cantilever right angle frame shown in Figure 8.7 is subjected to an in-plane end load F . Determine the rst ve bifurcation loads and deformed shapes for the structure. The frame is made of steel with the following geometry: L = 250 mm, b = 30 mm and t = 1 mm. Solution to Example 8.3. As in the previous example, the deformed shape of the structure can produce out-of-plane displacements. Thus, a 3-D analysis has to be carried out. In this case, shell elements will be used to this end.
MME
120
L t b
b F
Figure 8.7: Cantilever structural frame. The ANSYST M command sequence for this example is listed below. You can either type these commands on the command window, or you can type them on a le, then, on the command window enter /input, le, ext.
FINISH !close all previous modules /CLEAR !clear all previous models /TITLE, L-Frame: eigenvalues /PREP7 !Parameters t=1 L=250 b=30 p=1 ET,1,SHELL181 KEYOPT,1,3,2 R,1,t MP,EX,1,207000 MP,PRXY,1,0.3 K,1,0,0 K,2,0,-b K,3,L-b,-b K,4,L,-b K,5,L,0 K,6,L-b,-L K,7,L,-L L,1,2,6 L,2,3,44 L,3,4,6 L,4,5,6 L,5,1,50 L,3,6,44 L,6,7,6 L,7,4,44 AL,1,2,3,4,5 AL,8,7,6,3 AGLUE,1,2 AMESH,ALL FINISH /SOLU ANTYPE,0 !start PRE-PROCESSOR MODULE !thickness in mm !span in mm !column base in mm !applied load !element type #1: SHELL181 (6DoF) !element #1: keyoption 3=2 (full integration imcompatible modes) !real constant #1, thickness !Youngs modulus for material #1 !Poisson modulus for material #1 !generate key-points
MME
121
PSTRES,ON NSEL,S,LOC,X,0 D,ALL,ALL NSEL,S,LOC,X,L NSEL,R,LOC,Y,-L F,ALL,FX,p NSEL,ALL SOLVE FINISH /SOLU ANTYPE,BUCKLE BUCOPT,SUBSP,5 MXPAND SOLVE FINISH /POST1 SET,LIST SET,FIRST PLNSOL,U,Z,2,10 SET,NEXT PLNSOL,U,Z,2,10
!apply load
!start SOLUTION MODULE !buckling analysis !buckling options: subspace iteration method, 5 eigenvalues !expand all buckling modes
!start POST-PROCESSOR MODULE !list eigenvalues !set first eigenvalue !plot buckling mode !set next eigenvalue !plot buckling mode
8.3.2
Nonlinear buckling analysis is usually the more accurate approach and the recommended for design or evaluation of real structures. This technique uses a nonlinear static analysis with gradually increasing loads to determine the load level at which the structure becomes unstable. Using the nonlinear technique, the model can include features such as initial imperfections, plastic behaviour, gaps, and large-deection response. In addition, using deection-controlled loading, the post-buckled performance of the structure that buckles into a stable conguration can be evaluated. Example 8.4. Consider the structural frame in Ex. 8.3 and use a nonlinear buckling analysis to determine the critical load and buckling curve for the structure. Use an out-of-plane perturbation load applied at the free end of the frame to drive to the structure to buckling. This perturbation is removed close to the buckling load. Solution to Example 8.4. The out-of-plane perturbation load is only used to cause a small deformation and force the structure to buckle as a consequence of an assumed imperfection. In this case, this assumed imperfection causes a small out-of-plane deection. This perturbation is set to Fz = 1 102 N. As the eigenvalues of the structure have been evaluated in the previous example and it has been determined that the rst buckling mode occurs when F is 17.66 N, during the simulation a load F = 24 N will be considered in order to obtain the post-buckling behaviour of the structure. As expected, the critical load or bifurcation point predicted by using the nonlinear buckling method is close to the rst eigenvalue found in Ex. 8.3 although slightly lower. Figure 8.8 shows the resulting buckling curve where it can be seen that the bifurcation point is about 16 N while the structure starts to behave in a nonlinear way at about 10 N. It can be also seen in the gure that the behaviour of the structure is stable after the bifurcation point as the structure tends to stiffen and larger displacements requires larger loads.
MME
122
1
POST26 F_X
25 22.5 20 17.5 15 12.5 10 7.5 5 2.5 0 0 8 16 24 32 40 48
VALU
Load (N)
64 56 72
80
Figure 8.8: Buckling curve for the cantilever structural frame - load perturbation. The ANSYST M command sequence for this example is listed below. You can either type these commands on the command window, or you can type them on a le, then, on the command window enter /input, le, ext.
FINISH !close all previous modules /CLEAR !clear all previous models /TITLE, L-Frame: nonlinear buckling - load perturbation /PREP7 !Parameters t=1 L=250 b=30 ET,1,SHELL181 KEYOPT,1,3,2 R,1,t MP,EX,1,207000 MP,PRXY,1,0.3 K,1,0,0 K,2,0,-b K,3,L-b,-b K,4,L,-b K,5,L,0 K,6,L-b,-L K,7,L,-L L,1,2,6 L,2,3,44 L,3,4,6 L,4,5,6 L,5,1,50 L,3,6,44 L,6,7,6 L,7,4,44 AL,1,2,3,4,5 AL,8,7,6,3 AGLUE,1,2 AMESH,ALL !start PRE-PROCESSOR MODULE !thickness in mm !span in mm !column base in mm !element type #1: SHELL181 (6DoF) !element #1: keyoption 3=2 (full integration imcompatible modes) !real constant #1, thickness !Youngs modulus for material #1 !Poisson modulus for material #1 !generate key-points
MME
123
nNODE=node(L-b/2,-L,0) FINISH /SOLU ANTYPE,0 NLGEOM,ON ARCLEN,ON NSUBST,2 NSEL,S,LOC,X,0 D,ALL,ALL ALLSEL F,nNODE,FZ,1e-2 SOLVE FINISH /SOLU OUTRES,ALL,ALL NSUBST,40 F,nNODE,FX,24 SOLVE FINISH /POST1 SET,LAST PLNSOL,U,Z
!start SOLUTION MODULE !static analysis !activate geometric nonlinearities !use Arc Length method (for improved convergence) !use 2 substeps !fixed encastred nodes !apply perturbation load
!start SOLUTION MODULE !write results for every substep !use 40 substeps !apply load
!start POST-PROCESSOR MODULE !set last eigenvalue !plot buckling mode !start HISTORY-PLOT MODULE !evolution of Z-displacement at nNODE !absolute value !use as X-axis !evolution of applied force (element 389, nNODE) !evolution of applied force (element 433, nNODE) !evolution of applied force (change sign) !plot reaction force vs. displacement !list displacement and reaction force
Note that in this case the calculation requires more time as it is a nonlinear calculation. Thus, it is important to adjust correctly all the control parameters in order to reduce the calculation time and ensure the convergence of the solution. Example 8.5. Consider the structural frame in Ex. 8.3 and 8.4 and use a nonlinear buckling analysis to determine the critical load and buckling curve for the structure. In this case, use the deformed shape of the rst buckling mode multiplied by 1/10 of the thickness of the frame as the initial perturbation. Solution to Example 8.5. In this case, the eigenvalue analysis must be run rst in order to upload the deformed geometry of the structure for the rst buckling mode as the initial perturbation in the structure. As in the previous case, the eigenvalue analysis of the structure results in the rst buckling mode taking place when F is 17.66 N. Thus, during the simulation a load F = 24 N will be considered in order to obtain the post-buckling behaviour of the structure. Similarly to the previous example, the critical load or bifurcation point predicted is close to the rst eigenvalue but slightly lower and the structure is stable after the bifurcation point. As it can be seen in Figure 8.9, the bifurcation point in the resulting buckling curve is about 16.5 N, while the nonlinear behaviour of the structure starts at about 10 N. Observe that the resulting load-displacement curve shown in Figure 8.9 is very similar to the one obtained in Ex. 8.4. The ANSYST M command sequence for this example is listed below. You can either type these commands on the command window, or you can type them on a le, then, on the command window enter /input, le, ext.
MME
124
1
POST26 F_X
25 22.5 20 17.5 15
VALU
Load (N)
Figure 8.9: Buckling curve for the cantilever structural frame - deformed shape perturbation.
FINISH !close all previous modules /CLEAR !clear all previous models /TITLE, L-Frame: nonlinear buckling - deformed shape perturbation /PREP7 !Parameters t=1 L=250 b=30 p=1 delta=t/10 ET,1,SHELL181 KEYOPT,1,3,2 R,1,t MP,EX,1,207000 MP,PRXY,1,0.3 !start PRE-PROCESSOR MODULE !thickness in mm !span in mm !column base in mm !applied load !shape factor multiplier (t/10) !element type #1: SHELL181 (6DoF) !element #1: keyoption 3=2 (full integration imcompatible modes) !real constant #1, thickness !Youngs modulus for material #1 !Poisson modulus for material #1
K,1,0,0 !generate key-points K,2,0,-b K,3,L-b,-b K,4,L,-b K,5,L,0 K,6,L-b,-L K,7,L,-L L,1,2,6 !define line and divisions L,2,3,44 L,3,4,6 L,4,5,6 L,5,1,50 L,3,6,44 L,6,7,6 L,7,4,44 AL,1,2,3,4,5 !define areas through lines AL,8,7,6,3 AGLUE,1,2 !bond areas AMESH,ALL nNODE=node(L-b/2,-L,0) !store node number FINISH /SOLU ANTYPE,0 !start SOLUTION MODULE !static analysis
MME
125
PSTRES,ON NSEL,S,LOC,X,0 D,ALL,ALL ALLSEL F,nNODE,FX,p SOLVE FINISH /SOLU ANTYPE,BUCKLE BUCOPT,SUBSP,5 MXPAND SOLVE FINISH
!start SOLUTION MODULE !buckling analysis !buckling options: subspace iteration method, 5 eigenvalues !expand all buckling modes
!Add displacement from previous analysis !multiplier for added displacements, !load step, substep (equivalent to mode=1), !results file containing displacements and extension /PREP7 !start PRE-PROCESSOR MODULE UPGEOM,delta,1,1,file,rst FINISH /SOLU ANTYPE,STATIC NLGEOM,ON ARCLEN,ON OUTRES,ALL,ALL NSUBST,40 F,nNODE,FX,24 SOLVE FINISH /POST1 SET,LAST PLNSOL,U,Z !start SOLUTION MODULE - continuation loads !static analysis !activate geometric nonlinearities !use Arc Length method (for improved convergence) !write results of every substep !use 40 substeps !apply load
!start POST-PROCESSOR MODULE !set last eigenvalue !plot buckling mode !start HISTORY-PLOT MODULE !evolution of Z-displacement at nNODE !absolute value !use as X-axis !evolution of applied force (element 389, nNODE) !evolution of applied force (element 433, nNODE) !evolution of applied force (change sign) !plot reaction force vs. displacement !list displacement and reaction force
8.4
Suggested problems
Problem 8.1. The bi-supported column in Example 8.2 can be considered as a symmetric structure and therefore could be simulated using symmetry boundary conditions (as schematically shown in Fig. 8.10). Determine if the resulting buckling eigenvalues are the same for the complete structure and the symmetric half column or not. Justify the results. Problem 8.2. Compute the bifurcation load and the post-buckling behaviour of the column in Example 8.2 by means of a nonlinear analysis. Problem 8.3. Extract the rst ve bifurcation loads and deformed shapes of the structure presented in Figure 8.11. The thickness of the beam is 1.5 mm and it is made of steel.
MME
126
S Sy
z x
L/2 P
MME
Chapter 9
Material nonlinearities
9.1 Introduction
As has been introduced in Chapter 8, a nonlinear structural behaviour can be induced by a geometric nonlinearity, a contact or boundary nonlinearity, or by a material nonlinearity. This chapter deals with the nonlinear structural behaviour casued by material nonlinearities. It must be taken into account that an analysis including nonlinear models requires an iterative solution that might compromise its convergence and can be very expensive in terms of computational time.
Nonlinear material models arise from the presence of a time-independent behaviour, such as hyperelasticity, plasticity or damage, or to take into account other time-dependent behaviours, such as creep or viscoplasticity.
A number of material-related factors can cause changes in the stiffness of the structure during the course of an analysis. Nonlinear stress-strain relationships of plastic, multilinear elastic, and hyperelastic materials will cause a stiffness change in the structure at different load levels (and, typically, at different temperatures). Creep, viscoplasticity, and viscoelasticity will give rise to nonlinearities that can be time-, rate-, temperature-, or stress-related. Swelling will induce strains that can be a function of temperature, time, neutron ux level (or some analogous quantity), and stress. Some nonlinear material models are presented here as an example involving different physical responses.
9.2
Plasticity
Most common engineering materials exhibit a linear stress-strain relationship up to a stress level known as the proportional or elastic limit. Beyond this limit, the stress-strain relationship becomes nonlinear, although not necessarily inelastic. Plastic behaviour, characterized by nonrecoverable strain, begins when the stress level exceeds the material yield point. Because there is usually little difference between the yield point and the proportional limit, the most common plasticity approach assumes that these two points are coincident in plasticity analyses (see Figure 9.1). Plasticity is a nonconservative, path-dependent phenomenon. In other words, the sequence in which loads are applied and in which plastic responses occur affects the nal solution results. If plastic response is anticipated for the analysis, loads should be applied as a series of small incremental load steps or time steps. 127
128
9.2.1
As shown in Figure 9.1, bilinear plasticity assumes an elastic linear behaviour up to the yield point, and another linear behaviour after this point. The unload path is parallel to the elastic response. This model is known as the bilinear kinematic hardening model and is recommended for small-strain analyses involving materials that obey von Mises yield criteria (most metals).
Hint 39: ANSYS properties for Bilinear Kinematic Hardening model The bilinear kinematic hardening model is activated in ANSYS with the command TB,BKIN (see the following example). The sequence number in TBDATA command is the yield stress in the rst position and the tangent modulus d. Bilinear Kinematic Plasticity Example: Nickel Alloy FINISH /CLEAR /PREP7 MP,EX,1,180e9 MP,PRXY,1,.31 MP,DENS,1,8490 TB,BKIN,1,1 TBDATA,1,900e6,445e6 TBPLOT,BKIN,1
!Pa !no units !kg/m3 !bilinear kinematic hardening, material #1, 1 temperature !Yield stress (Pa) and Tangent modulus (Pa) !display the data table
In the case of an isotropic hardening, the hardening of the material is independent of the direction and it is the same for tension and compression. The Bilinear Isotropic Hardening model uses the von Mises yield criteria coupled with an isotropic work hardening assumption. This model is recommended for large strain analyses and it is activated in ANSYS with the command TB,BISO. The denition parameters are the same as for the case of the bilinear kinematic model.
Example 9.1. Use a bilinear plasticity model to simulate in ANSYS an hexagonal Allen wrench 10 mm wide across ats made of high carbon steel (HCS) with a Youngs modulus of 210 103 MPa, a Poissons ratio of 0.3, a Yield stress of 500 MPa, and a Tangent modulus of 20 103 MPa.
Solution to Example 9.1. Small strains are expected for this analysis and the Bilinear Kinematic Hardening model is used. Two different load cases are considered to show the effect of
MME
129
the bilinear model and the generated plastic strain. Observe that once the rst applied load is removed there is no remaining deformation or plastic strain. However, when the second applied load is removed, there is a remaining deformation and plastic strain. The ANSYS command list is presented next.
FINISH /CLEAR /TITLE, Allen wrench with plastic strain /UNITS,MPA !units are in mm, MPa, and Newtons /PREP7 !start PRE-PROCESSOR MODULE W_HEX=10 !distance across flats (10mm) !specifies units for angular functions in parameter expressions *AFUN,DEG W_FLAT=W_HEX/2/TAN(60)*2 !wrench edge length L_SHANK=60 !shank length L_HANDLE=112.5 !handle length BENRAD=10 !bending radius between the shank and the handle L_ELEM=7.5 !element length NO_D_HEX=2 !number of element divisions in the hexagon ET,1,SOLID185 KEYOPT,1,2,3 ET,2,PLANE182 KEYOPT,2,1,3 MP,EX,1,2.1E5 MP,PRXY,1,0.3 TB,BKIN,1,1 TBDATA,1,500,0.2E5 TBLIST,BKIN,1 /XRANGE,0,0.01 TBPLOT,BKIN,1 RPOLY,6,W_FLAT K,7 K,8,,,-L_SHANK K,9,,L_HANDLE,-L_SHANK L,4,1 L,7,8 L,8,9 LFILLT,8,9,BENRAD ASBL,1,7 CM,BOTAREA,AREA LESIZE,2,,,NO_D_HEX LESIZE,5,,,NO_D_HEX LESIZE,1,,,1 LESIZE,3,,,1 LESIZE,4,,,1 LESIZE,6,,,1 TYPE,2 ESHAPE,2 AMESH,ALL TYPE,1 ESIZE,L_ELEM VDRAG,2,3,,,,,8,10,9 CMSEL,,BOTAREA ACLEAR,ALL ASEL,ALL /VIEW,1,1,1,1 EPLOT FINISH /SOLU ANTYPE,STATIC AUTOTS,ON NSUBST,3,5,1 CNVTOL,F,,,,1E-30 CMSEL,,BOTAREA LSEL,,EXT NSLL,,1 D,ALL,ALL !element type #1: SOLID186 (3 DoF) !element #1: keyoption 2=3 (simplified enhanced strain integration) !element type #2: PLANE182 (2 DoF) !element type #2: keyoption 1=3 (simplified enhanced strain formulation) !define material, plastic steel, Youngs modulus !Poisson coeficient !bilinear isotropic hardening, material #1, 1 temperature !yield stress, tangent modulus !list plastic model properties !x-axis range on TBPLOT !plot plastic strain-stress model !define the geometry of the Allen wrench section (hexagon) !define the shank and handle lengths
!generates a fillet line between intersecting lines 8 and 9 !subtracts lines from areas !groups geometry entities in the variable BOTAREA !define control mesh and meshing of the hexagonal section
!change the element type to mesh the volume !define control mesh !volume by dragging an area along a path, area, line path !select the group BOTAREA and assemblies !clear the area and associate nodes !select all areas !plot elements !end PRE-PROCESSOR MODULE !start SOLUTION MODULE !automatic stepping on !number of substeps, max, min !set convergence tolerance on force, minimum !apply boundary conditions: clamp bot area !select the external lines of the BOTAREA !select nodes associated with selected lines !clamp
MME
130
NSEL,S,LOC,X,W_FLAT/2-0.1,W_FLAT+0.1 !select nodes to apply load NSEL,R,LOC,Y,L_HANDLE-L_ELEM-0.1,L_HANDLE+0.1 CM,NLOAD,NODE !define group of nodes ALLSEL T=60 !apply a load of 60 N to nodes F,NLOAD,FX,-T SOLVE !solve 1st load set FDELE,ALL,ALL T=1e-10 F,NLOAD,FX,-T SOLVE T=120 F,NLOAD,FX,-T SOLVE FDELE,ALL,ALL T=1e-10 F,NLOAD,FX,-T SOLVE FINISH /POST1 /DSCALE,,1 SET,1 PLNSOL,EPEL,EQV PLNSOL,EPPL,EQV SET,2 PLNSOL,EPEL,EQV PLNSOL,EPPL,EQV SET,3 PLNSOL,EPEL,EQV PLNSOL,EPPL,EQV SET,4 PLNSOL,EPEL,EQV PLNSOL,EPPL,EQV !delete the loads !apply a load of 0 N to nodes !solve 2nd load set: no loads !apply a load of 120 N to nodes !solve 3rd load set !delete the loads !apply a load of 0 N to nodes !solve 4th load set: only remain plastic strain !end SOLUTION MODULE !start POST-PROCESSOR MODULE !do not scale displacements !contour plot of von Mises equivalent elastic strains !contour plot of von Mises equivalent plastic strains !contour plot of von Mises equivalent elastic strains !contour plot of von Mises equivalent plastic strains !contour plot of von Mises equivalent elastic strains !contour plot of von Mises equivalent plastic strains !contour plot of von Mises equivalent elastic strains !contour plot of von Mises equivalent plastic strains
The same problem can be simulated by dening an initial deformed state for a nal load step corresponding to the deformation-state resulting from an initial load step. Once this nal load step is solved with zero loads applied, the remaining deformations are the plastic deformations of the initial step. In ANSYS this can be achieved by dening this initial deformed state by using the INISTATE command. The ANSYS command list for the generation of the initial load step and writing of the initial state le for the considered the example is presented next.
FINISH /CLEAR /TITLE, Allen wrench with plastic strain /UNITS,MPA !units are in mm, MPa, and Newtons /PREP7 !start PRE-PROCESSOR MODULE W_HEX=10 !distance across flats (10mm) !specifies units for angular functions in parameter expressions *AFUN,DEG W_FLAT=W_HEX/2/TAN(60)*2 !wrench edge length L_SHANK=60 !shank length L_HANDLE=112.5 !handle length BENRAD=10 !bending radius between the shank and the handle L_ELEM=7.5 !element length NO_D_HEX=2 !number of element divisions in the hexagon ET,1,SOLID185 KEYOPT,1,2,3 ET,2,PLANE182 KEYOPT,2,1,3 !element !element !element !element type #1: SOLID186 (3 DoF) #1: keyoption 2=3 (simplified enhanced strain integration) type #2: PLANE182 (2 DoF) type #2: keyoption 1=3 (simplified enhanced strain formulation)
MME
131
!define material, plastic steel, Youngs modulus !Poisson coeficient !bilinear isotropic hardening, material #1, 1 temperature !yield stress, tangent modulus !list plastic model properties !x-axis range on TBPLOT !plot plastic strain-stress model !define the geometry of the Allen wrench section (hexagon) !define the shank and handle lengths
MP,EX,1,2.1E5 MP,PRXY,1,0.3 TB,BKIN,1,1 TBDATA,1,500,0.2E5 TBLIST,BKIN,1 /XRANGE,0,0.01 TBPLOT,BKIN,1 RPOLY,6,W_FLAT K,7 K,8,,,-L_SHANK K,9,,L_HANDLE,-L_SHANK L,4,1 L,7,8 L,8,9 LFILLT,8,9,BENRAD ASBL,1,7 CM,BOTAREA,AREA LESIZE,2,,,NO_D_HEX LESIZE,5,,,NO_D_HEX LESIZE,1,,,1 LESIZE,3,,,1 LESIZE,4,,,1 LESIZE,6,,,1 TYPE,2 ESHAPE,2 AMESH,ALL TYPE,1 ESIZE,L_ELEM VDRAG,2,3,,,,,8,10,9 CMSEL,,BOTAREA ACLEAR,ALL ASEL,ALL /VIEW,1,1,1,1 EPLOT FINISH /SOLU ANTYPE,STATIC AUTOTS,ON NSUBST,3,5,1 CNVTOL,F,,,,1E-30 CMSEL,,BOTAREA LSEL,,EXT NSLL,,1 D,ALL,ALL
!generates a fillet line between intersecting lines 8 and 9 !subtracts lines from areas !groups geometry entities in the variable BOTAREA !define control mesh and meshing of the hexagonal section
!change the element type to mesh the volume !define control mesh !volume by dragging an area along a path, area, line path !select the group BOTAREA and assemblies !clear the area and associate nodes !select all areas !plot elements !end PRE-PROCESSOR MODULE !start SOLUTION MODULE !automatic stepping on !number of substeps, max, min !set convergence tolerance on force, minimum !apply boundary conditions: clamp bot area !select the external lines of the BOTAREA !select nodes associated with selected lines !clamp
NSEL,S,LOC,X,W_FLAT/2-0.1,W_FLAT+0.1 !select nodes to apply load NSEL,R,LOC,Y,L_HANDLE-L_ELEM-0.1,L_HANDLE+0.1 CM,NLOAD,NODE !define group of nodes ALLSEL T=60 !apply a load of 60 N to nodes !T=120 !uncomment to apply load of 120 N to nodes F,NLOAD,FX,-T INISTATE,WRITE,1,,,,0,EPPL !write plastic deformation state in file SOLVE !solve initial load set FINISH !end SOLUTION MODULE /POST1 /DSCALE,,1 PLNSOL,EPEL,EQV PLNSOL,EPPL,EQV !start POST-PROCESSOR MODULE !do not scale displacements !contour plot of von Mises equivalent elastic strains !contour plot of von Mises equivalent plastic strains
The ANSYS command list for the generation of nal load step and reading of the initial state le for the considered example is presented next.
MME
132
FINISH /SOLU INISTATE,READ,file.ist FDELE,ALL,ALL SOLVE FINISH /POST1 /DSCALE,,1 PLNSOL,EPEL,EQV PLNSOL,EPPL,EQV
!start SOLUTION MODULE !read plastic deformation state in file !delete all loads !solve final load set: no loads !end SOLUTION MODULE !start POST-PROCESSOR MODULE !do not scale displacements !contour plot of von Mises equivalent elastic strains !contour plot of von Mises equivalent plastic strains
Observe that additionally to the expected plastic deformation there is a remaining elastic deformation and stress. What is the cause of this remaining elastic deformation and stress?
9.2.2
Multiple-point plasticity models can be used to dene multilinear plasticity behaviours. Hint 40: ANSYS properties for Multiple-point Isotropic Plasticity model The multiple-point isotropic plasticity model is activated in ANSYS with the command TB,MISO (see example next), the sequence number in TBPT command are the different stress-strain points that dene the material response. Multiple-point Isotropic Plasticity Example: Aluminium FINISH /CLEAR /PREP7 MP,EX,1,69000 !material #1: Elastic modulus MP,PRXY,1,0.3 !material #1: Poisson coeficient TB,MISO,1,1,11 !material #1, Multi-point isotropic plasticity, n- points TBPT,,0.003,207 !nonlinear curve point,, x coordinate, y coordinate TBPT,,0.004,230.7885 TBPT,,0.005,242.7903 TBPT,,0.006,252.0527 TBPT,,0.007,259.8827 TBPT,,0.008,266.7906 TBPT,,0.009,273.0415 TBPT,,0.01,278.7934 TBPT,,0.024,335.648 TBPT,,0.04,379.7314 TBPT,,0.063,428.6213 TBPLOT,MISO,1
Example 9.2. Considerer a thin-walled aluminum sphere with a radius r = 250 mm and a thickness t = 5 mm. The sphere is subjected to an internal pressure p = 12 MPa. The elastic modulus
MME
133
and Poisson ratio of the material are E = 69 GPa and = 0.3. The plastic behaviour of aluminium is governed by
Solution to Example 9.2. The following MATLAB le is used to obtain the stress vs. total strain curve shown in Figure 9.2 and write the command sequence to dene the material behaviour in the ANSYS input le.
%% PrePostFE. Solution of Example 9.2. % J.A. Mayugo, 2008 clear all;close all E=69000 nu=0.3 sigma_eff_0=207 sigma_eff=930 % MPa % MPa % MPa
%% Multi-point plasticity model epsilon_MISO=[[0.003:0.001:0.01],0.024,0.040,0.063]; sigma_MISO=interp1(epsilon,sigma_eff,epsilon_MISO,linear); [m,n_MISO]=size(sigma_MISO) figure1=figure hold on plot(epsilon,sigma_eff,LineWidth,1.8); plot(epsilon_elastic,sigma_elastic,r--,LineWidth,1.8); plot(epsilon_MISO,sigma_MISO,kd,LineWidth,1.8,MarkerFaceColor,w,MarkerSize,6); hold off xlabel(Total strain,Fontsize,16);ylabel(Stress,Fontsize,16); legend(Plastic response,Elastic response,Discrete response to MISO,Location,Best);legend(boxoff); set(gca,Fontsize,16); saveas(figure1,[PrePostFE_Ex902],eps); n_file = PrePostFE_Ex902; tline{1} = [MP,EX,1,,num2str(E), !material #1: Elastic modulus ]; tline{2} = [MP,NUXY,1,,num2str(nu), !material #1: Poisson coeficient]; tline{3} = [TB,MISO,1,1,,num2str(n_MISO), !material #1, Multi-point isotropic plasticity, n- points]; for ii=1:n_MISO tline{3+ii} = [TBPT,,,num2str(epsilon_MISO(ii)),,,num2str(sigma_MISO(ii))]; end fid0 = fopen([n_file,.txt],w); fprintf(fid0,%10s\n\n,!INPUT FILE TO ANSYS, Multi-point isotropic plasticity); for i=1:n_MISO+3 fprintf(fid0,%10s\n,tline{i}); end fclose(fid0); This le can be found at: ftp://amade.udg.edu/mms/PrePostFE/input_files/T9/PrePostFE_Ex902.m
MME
134
450 400 350 300 Stress 250 200 150 100 50 0 0 0.01 0.02 0.03 0.04 0.05 Total strain
0.06
0.07
0.08
!create keypoints
!create arc !specify 20 elements on line !mesh !exit PRE-PROCESSOR MODULE !start SOLUTION MODULE !write results for each sub-step !number of substeps, max, min !select nodes at Y=0 !fix the node displacement in Y direction !apply internal pressure 12 MPa !select all !solve the first load-step
MME
135
!delete all pressure loads !solve a second load-step !exit SOLUTION MODULE !start POST-PROCESSOR MODULE !set the first load-step !set the second load-step !exit POST-PROCESSOR MODULE
9.3
A material can be considered as hyperelastic if there is an elastic potential function (or strain energy density function), which is a scalar function of one of the strain or deformation tensors, whose derivative with respect to a strain component determines the corresponding stress component. Hyperelasticity can be used to analyse elastomers or rubber-like materials subjected to large strains and displacements with small volume changes (nearly incompressible materials). Therefore, it is necessary to activate the large strain theory during the analysis. One of the hyperelasticity models that can be used in ANSYS is the Mooney-Rivlin model. In the two parameter Mooney-Rivlin model the form of the strain energy potential is:
(9.1)
where u is the strain energy potential, I1 the rst deviatoric strain invariant, I2 the second deviatoric strain invariant, c01 and c10 are material constants characterising the deviatoric deformation of the material, J determinant of the elastic deformation gradient and d is the material incompressibility parameter.
The initial shear modulus is dened as:
= G = 2 (c10 + c01 )
and the initial bulk modulus is dened as:
(9.2)
K=
where if d is not dened explicity but as
2 d
(9.3)
d=
1 + 2 c10 + c01
MSTER EN MECNICA DE MATERIALS I ESTRUCTURES
(9.4)
MME
Mooney-Rivlin model for elements SHELL181, PLANE182, PLANE183, SOLID185, SOLID186, SOLID187, SOLSH190, SHELL208, and SHELL209. The Mooney-Rivlin model is activated in ANSYS with the command TB,HYPER,,,,MOONEY or TB,MOONEY (see example next), the sequence number in TBDATA command is c10 , c01 , and d. Mooney-Rivlin Example: Rubber MP,DENS,1,.0018 MP,PRXY,1,.499 TB,MOONEY,1 TBDATA,1,80 TBDATA,2,20 !lb/in3 !no units !C10 (psi) !C01 (psi)
Example 9.3. Use ANSYS to model a circular plate made of rubber with applied pressure in one surface. Use an hyperelastic model.
N,1,0.,0. !geometry and mesh N,2,0.17143,0. N,3,0.47143,0. N,4,0.90000,0. N,5,1.4571,0. N,6,2.1429,0. N,7,2.9571,0. N,8,3.9000,0. N,9,4.9714,0. N,10,6.1714,0. N,11,7.5000,0. N,101,0.,0. N,102,0.16996,2.23759E-02 N,103,0.46740,6.15338E-02 N,104,0.89230,0.11747 N,105,1.4447,0.19020 N,106,2.1245,0.27970 N,107,2.9318,0.38598 N,108,3.8666,0.50905 N,109,4.9289,0.64890 N,110,6.1186,0.80553 N,111,7.4358,0.97895 E,1,2,102,102 E,2,3,103,102
MME
137
E,3,4,104,103 E,4,5,105,104 E,5,6,106,105 E,6,7,107,106 E,7,8,108,107 E,8,9,109,108 E,9,10,110,109 E,10,11,111,110 LOCAL,11,0,0.0,0.0,0.0,7.5,0.0,0.0 !new local coordinate system NROTAT,102,111,1 !rotates nodes: 102 to 111 FINISH !finish PRE-PROCESSOR MODULE /SOLU ANTYPE,STATIC AUTOTS,ON NSUBST,400,1200,25 NLGEOM,ON NROPT,FULL,,OFF OUTRES,ALL,ALL NEQIT,20 !start SOLUTION MODULE !set static analysis !use automatic time stepping !specify the number of substeps, max, min !non-linear analysis !option of convergence: Newton-Raphson !solution data to be written !max number of equilibrium iterations
D,1,UY,0.0,,11,1,ROTX,ROTZ !boundary conditions D,102,UY,0.0,,111,1,ROTX,ROTZ D,11,UX,0.0,,111,100,UY,UZ D,1,UX,0.0,,,,UY,ROTX,ROTY,ROTZ SF,ALL,PRES,50.0 SOLVE FINISH /POST1 /NOPR /VIEW,1,,-1 !/ANG,1 !/USER /FOCUS,1,4,,8,0 /DIST,,12 /TRIAD,OFF SET,FIRST PLDISP,0 /NOERASE SET,,10 PLDISP,0 SET,,20 PLDISP,0 SET,,25 PLDISP,0 SET,LAST PLDISP,1 /ERASE /TRIAD,ON /GOPR /ESHAPE,0 FINISH !apply uniform pressure in psi !solve current load state !finish SOLUTION MODULE !start POST-PROCESSOR MODULE !suppress graphing data !change the view to plane x-z
!setup center of graphics screen for disp. plot !set distance to zoom out !hide the coordinate system !set displacement data for first substep !plot displacement data !set display to overlay plot
The following le allows to obtain the thickness evolution an the displacement evolution (in this case with an experimental comparison).
/POST26 !start TIME-HISTORY MODULE /XRANGE,0,3.0 !x-axis scale range /YRANGE,0,1 !y-axis scale range /AXLAB,X,UZ OF CENTER/R-INITIAL !x-axis label /AXLAB,Y,THICKNESS/ORIGINAL THICKNESS !y-axis label NSOL,2,1,U,Z,UZ_1 !nodal data to be stored
MME
138
ESOL,3,1,,SMIS,17,TH_1 ADD,4,2,,,UZRATIO,,,0.13333333,0,0, ADD,5,3,,,SH.181,,,2,0,0, /COLOR,CURVE,MRED XVAR,4 PLVAR,5 /ERASE /NOPR *DIM,X,TABLE,20,1 *DIM,Y,TABLE,20,1 X(1,1)= 1.25 Y(1,1)= 1.25 Y(2,1)= 2.5 X(3,1)= 2.25 Y(3,1)= 4.0 X(4,1)= 2.6 Y(4,1)= 5.9 X(5,1)= 2.9 Y(5,1)= 7.8 X(6,1)= 3.2 Y(6,1)= 9.8 X(7,1)= 3.5 Y(7,1)= 11.6 X(8,1)= 3.62 Y(8,1)= 12.6 X(9,1)= 4.1 Y(9,1)= 15.3 X(10,1)= 4.9 Y(10,1)= 18.8 X(11,1)= 5.7 Y(11,1)= 22.1 X(12,1)= 6.2 Y(12,1)= 24.0 X(13,1)= 7.2 Y(13,1)= 27.9 X(14,1)= 8.3 Y(14,1)= 31.2 X(15,1)= 8.9 Y(15,1)= 32.9 X(16,1)= 9.9 Y(16,1)= 35.8 X(17,1)= 10.9 Y(17,1)= 38.0 X(18,1)= 13.1 Y(18,1)= 42.9 X(19,1)= 14.4 Y(19,1)= 45 X(20,1)= 15.2 Y(20,1)= 46 /GOPR
!element data bo be stored !adds variables !curve color !x-variable to be displayed !y-variable to be displayed
!supress the input data !add a table: experimental data !table x data point !table y data pointX(2,1)= 1.8
!reactivates the supressed printout !x-axis scale range !y-axis scale range !x-axis label !y-axis label !curve color !plot a curve from an array !nodal data to be stored !multiply solution by 50 !curve color !specify x variable to be displayed !display solution in graph file !finish TIME-HISTORY MODULE
/XRANGE,0,20 /YRANGE,0,60 /AXLAB,X,UZ OF CENTER (IN) /AXLAB,Y,PRESSURE (LB/SQ IN) /COLOR,CURVE,YGRE *VPLOT,X(1,1),Y(1,1) /NOERASE NSOL,2,1,U,Z,UZ_1 PROD,7,1,,,SH.181,,,50,0,0, /COLOR,CURVE,MRED XVAR,2 PLVAR,7 /ERASE FINISH
MME
139
9.4
Suggested problems
Problem 9.1. Reconsider the situation in Example 3.3 and introduce a bilinear plasticity model. The yield stress is 270 MPa and ET = E/180. Increase the load until some plastic deformation appears. Obtain the Von Misses contour plot of the plastic strain and comment the results. Show your results in a report. Problem 9.2. Generate an ANSYS model to simulate the behavior of a 5 mm thick rectangular plate (see gure 9.3) with material nonlinearities dened by a bilinear kinematic plasticity law. Use PLANE182 elements and consider the following material properties to dene the non-linear behaviour: an elastic modulus of 210 GPa, Poissons coefcient = 0.3, and a yield stress of 500 MPa. Dene an appropriate value for the applied pressure p and obtain the stress-strain curve of the bilinear plasticity approach when: a) the tangent modulus is 1% of the elastic modulus and b) the tangent modulus is 0.1% of the elastic modulus.
Use the same applied pressure as before to simulate the same problem when the tangent modulus is 0.01% of the initial modulus. Comment the obtained results.
MME