Autodesk Simulation Workshop

Section 2: Linear Analysis

This section presents the theory and methods used to perform linear analyses using Autodesk Simulation. This is an important section since the bulk of the problems solved by engineers on a daily basis fall into this classification. One definition of a linear analysis is that a plot of response parameters versus input parameters is a straight line. In this section a linear analysis is one that involves linear materials. Some of the analyses may contain non-linear responses such as large displacements, stress stiffening, or buckling. However, in all cases the material will be an isotropic-linearelastic material.

Modules Contained in Section 2 1. 2. 3. 4. 5. 6. Virtual Work Equation Newton-Raphson Equations Truss Element Equations Static Analysis Natural Frequency Analysis Transient Stress (Modal Superposition) 7. Transient Stress (Direct Integration) 8. Load Factor Analysis 9. Fatigue Analysis

The first three modules present the underlying theory associated with the finite element method. Since geometric non-linear phenomena are included in subsequent modules, the theory will cover non-linear solution methods. The starting point is the differential equation of equilibrium. Equilibrium and the numerical methods used to ensure that a solution satisfies it is a central theme of the first three sections. The remaining modules, four through nine, provide additional theory and information needed to set up a specific type of problem, execute it, and interpret the results for one of the analysis types contained in Autodesk Simulation.

The intent is to present the material in the first three modules in such a way that a senior level undergraduate engineering student or a practicing engineer could use the modules as a learning resource. Standard matrix notation is used instead of the more concise indicial notation taught in graduate school. The content is presented at a depth that an interested student can work through the mathematics and obtain insight into the methods and workings of a commercial program. Each subsequent module contains just enough theory to teach a student or practicing engineer how to setup and perform each analysis type. It is possible for someone to start at Module 4 without first mastering Modules 1 through 3. In Modules 4 through 9 a cantilevered beam modeled using 3D brick elements with mid-side nodes serves as the example problem. The cantilevered beam is familiar to undergraduate students, and for some analysis types students can check results using manual calculation.

Section 2: Linear Analysis

Important Note on Archived Datasets

The datasets associated with each module in Section 2 have been Archived to facilitate downloading. An Archived dataset is a compressed file created by Autodesk Simulation Multiphysics to reduce the overall size of the file. The Archived files do not contain solution results, and it will be necessary to execute the analysis in order to obtain the results. In most cases there is a Begin and an End dataset file. The Begin file contains the problem data prior to setting the analysis parameters, and will not run properly. It provides a starting point for the module. The videos show how to change the analysis parameters in the Begin datasets to those in the End datasets. The End dataset can be executed to obtain the analysis results. Some datasets do not have a Begin and End file (i.e. Modules 1, 2, 3, and 9). Modules 1, 2, and 3 do not involve executing the software to solve a problem. Module 9 on Fatigue requires that a static analysis be executed prior to performing the fatigue analysis. The dataset for Module 9 is set up to run the static analysis when the file is retrieved. An Archived dataset can be retrieved by selecting the Autodesk Simulation Icon in the upper left corner of the screen, selecting Archive in the drop down menu, and then selecting Retrieve.

Autodesk Inventor Files

The Inventor part or assembly files are also included as part of the datasets. In some cases the Inventor files are used as in the videos or PowerPoint presentations. For example, the intent of the video may be to show how to go from Autodesk Inventor into Autodesk Simulation Multiphysics. In other cases the Inventor files are not part of the videos or PowerPoint presentations, but having them available will enable instructors or students to develop additional problems or exercises.

Section 2: Linear Analysis

Table of Contents
Click below to jump to the current Module:

1. Module 1: Virtual Work Equation ........................................................................... 4 2. Module 2: Newton-Raphson Equations ............................................................. 4 3. Module 3: Truss Element Equations ...................................................................... 6 4. Module 4: Static Analysis ...................................................................................... 7 5. Module 5: Natural Frequency Analysis ................................................................ 11 6. Module 6: Transient Stress (Modal Superposition) .............................................. 13 7. Module 7: Transient Stress (Direct Integration) ................................................... 16 8. Module 8: Load Factor Analysis .......................................................................... 18 9. Module 9: Fatigue Analysis ................................................................................. 19

Section 2: Linear Analysis

1. Module 1: Virtual Work Equation

Introduction
In this module the differential equations of equilibrium for stresses are used to develop the virtual work equation that represents an integral statement of equilibrium. The equation is first developed with respect to equilibrium in the deformed shape and is then stated in terms of conjugate stress and strain measures written with respect to a reference configuration.

Execution
1) The linear analysis types available in Autodesk Simulation can be seen by selecting Analysis Types in the browser, right clicking and selecting Linear. A list of linear analysis types is shown. These include: 1) Static Stress with Linear Material Models, 2) Natural Frequency (Modal), 3) Natural Frequency (Modal) with Load Stiffening, 4) Response Spectrum, 5) Random Vibration, 6)Frequency Response, 7) Transient Stress (Direct Integration), 8) Transient Stress (Modal Superposition), and 9) Critical Buckling Load Factor. 2) The modules in Section II that show how to perform a specific type of analysis come from this list. They are: 1) Module 4 Static Analysis, 2) Module 5 Natural Frequency Analysis, 3) Module 6 Modal Superposition, 4) Module 7 Direct Integration, 5) Module 6 - Load Factor Analysis, and 6) Module 9 - Fatigue. Module 5 addresses how to compute natural frequencies with and without load stiffening. Module 9 on fatigue requires that a stress analysis first be performed. 3) Breaking the object to be analyzed into small parts called elements is a fundamental part of the finite element method. One of the primary reasons for breaking the object into parts is that the integrals associated with the integral form of the virtual work equation can be carried out more easily. Another is that the displacement variation within the elements can be approximated by simple polynomials. 4) In addition to breaking the object into small elements, boundary conditions and loads must be defined. One end of the beam used in the modules is clamped using displacement boundary conditions that do not allow the nodes on the end surface to move. Loads are applied either on the free end or are applied as pressures on the top surface. 5) The objective of the various analysis types is usually to find the displacements and stresses within the elements that make them in equilibrium with the applied loads and boundary condition constraints.

2. Module 2: Newton-Raphson Equations

Introduction
In this module the Newton-Raphson method for solving non-linear equations is used to develop an incremental form of the virtual work equation presented in Module 1. This incremental form gives rise to the various components of the tangent stiffness matrix and

Section 2: Linear Analysis

residual load that are central to the types of analyses available in Autodesk Simulation. Later modules covering the change in natural frequencies due to stress stiffening and linear buckling or load factor analysis require specific components of the tangent stiffness matrix. Various equilibrium iteration methods available in Autodesk Simulation are based on the theory contained in this section.

Execution
1) The linear analysis types that use the stress-dependent contribution to the tangent stiffness matrix can be seen by selecting Analysis Types, right clicking and selecting Set Current Analysis Type, and selecting Linear. Linear analysis types that include stress stiffening are Natural Frequency (Modal) with Load Stiffening and Critical Buckling Load. The stiffness matrix used in both of these methods changes as the loads acting on the system being analyzed change. They require that loads be applied to the structure and that a static analysis be performed before the natural frequencies or buckling load factor are computed. This static analysis is automatically performed by Autodesk Simulation. 2) Non-linear problems require that equilibrium iterations be performed to obtain a solution to the non-linear equations. Module 4 presents a non-linear analysis of a cantilevered beam that experiences large displacements when subjected to a pressure load on its top surface. This simple problem is non-linear for two reasons. First, the deflections are large and the stiffness matrix changes due to the change in geometry. The final geometry is considerably different in shape than the original geometry. This is known as a geometric non-linearity. Second, the pressures remain normal to the surface and the direction of the surface normal vectors change as the beam deflects. 3) The equilibrium iteration options available for a non-linear problem can be seen by selecting Analysis Type in the browser for a Static Analysis with Non -linear Materials, right clicking and selecting Edit Analysis Parameters, and selecting the Advanced button at the bottom of the Analysis Parameters data entry box. Although the title of the analysis type includes the words Non-linear Materials, this analysis type is also applicable to linear material models. Linear materials are the default type and a non-linear material must be selected to activate the non-linear material model. A linear material model is used if no non-linear material is specified. This analysis type must be used any time a static analysis is performed that needs to account for large displacements. This is the case regardless of the type of material being used. 4) The types of equilibrium iterations available include combined Newton with and without line search, Newton with and without line search, and modified Newton with and without line search. The theoretical background for these types of equilibrium iterations is included in the Power Point slides for Module 2 and additional theory is included in Module 4. It is interesting to note that not all of these methods will converge for this relatively simple problem while others converge significantly slower than others. The reason that all of these methods are made available is that there is no single method that works best all of the time. The ability to select the best method for a particular problem requires experience, an understanding of the theory, and often some trial and error. An experienced analyst will often try different iteration methods in the course of an analysis to see which performs the best. This is particularly important when large problems are being solved that require a significant amount of time.

Section 2: Linear Analysis

3. Module 3: Truss Element Equations

Introduction
In this module the components of the tangent stiffness matrix and residual load are developed for a truss element to demonstrate how to go from the equations in Module 2 to the element level equations implemented in a production finite element program such as Simulation Mechanical. A truss element is chosen because the integrations required by the virtual work equation can be carried out by hand. More complicated elements, such as a 3D brick, use Quassian quadrature methods to evaluate the integrals. Numerical integration adds another level of sophistication that is not necessary to understand the basic steps.

Execution
1) The element types available in Autodesk Simulation can be seen by selecting an Element Type for a particular part in the browser. A list appears that contains Truss, Beam, Membrane, Brick, Plate, Tetrahedron, Spring, Rigid, Gap, Thin Composite, and Thin Composite (Sandwich). This list may change depending on the type of analysis being performed. 2) Although these element types have unique characteristics that are accounted for in their mathematical formulations, all of the element types are based on the incremental form of the equations of equilibrium. 3) Interpolation or shape functions are used to describe the variation of a parameter within an element. In the case of solid elements the interpolation functions can be used to describe the shape of an element and the variation of the displacement within the element. When the same functions are used to describe both the shape and displacement variation, the element is called an isoparametric element. The temperature variation within an element is described using interpolation functions when a heat transfer analysis is being performed. The interpolation functions for a truss element are used in the PowerPoint slides for Module 3. These same interpolation functions are used for spring and other line elements. 4) A graphical user interface makes the application of finite element methods extremely powerful. When the finite element method was first being developed, only engineers with specialized education and advanced degrees used the method. The analyses took long periods of time to perform because meshes were developed manually and the post-processing graphics tools were crude by todays standards. Now engineers at all experience and educational levels use finite element programs and most of the underlying mathematics is hidden from the user. However, it is still important to remember that all numerical methods have issues associated with accuracy, convergence, and stability. Users of the software must still ensure that their answers have converged and that the analysis being performed is sufficient to capture the underlying physics or mechanics of the problem. This more often than not requires that multiple analyses be performed with different meshes and mesh densities in high stress areas. The graphical user interfaces and speed of modern computers make this iterative process relatively quick.

Section 2: Linear Analysis

5) Students or inexperienced users always have questions about what time step or element size to use. There is no universal answer to this question. The more that is understood about the underlying theory, the better a user will be able to judge what a good starting value might be. However, even the educated and experienced user will still perform multiple analyses using different parameters to determine the sensitivity of the results to selected input data.

4. Module 4: Static Analysis

Introduction
A static analysis involves loads that are applied so slowly that they do not create a dynamic response, or steady state loads that exist after the dynamic transients have been damped out. A static analysis is probably the most common type of analysis performed to evaluate the strength of a mechanical component or system. The failure mechanism most usually associated with a static analysis is yielding of the material and stress results are depicted using the von Mises stress that can be compared directly to the yield strength. A factor of safety can be computed that provides a measure of how close the stresses at a point are to causing permanent deformation due to exceeding the yield strength. The problem analyzed in this module is a cantilevered beam that has a pressure loading that follows the surface of the beam as it deforms. The beam experiences large displacements and equilibrium iterations are required to obtain accurate results. Although a relatively simple problem, the large deformations and associated pressure load that follows the deformation causes some iteration methods to fail or converge slowly. The videos associated with this module are divided into four parts. The first part addresses meshing the model and applying loads. The second part shows how to set up the problem by editing the analysis parameters. The third part explains the information that is created and displayed on the computer screen during the analysis. The fourth part addresses tools and methods available for analyzing the results. The Execution section that follows is also divided into these four parts.

Execution
Note that the Analysis Type displayed in the FEA Editor is Static Stress with Nonlinear Material Models. Although Nonlinear Material Models are available in this analysis type, linear elastic materials are also available. In this case we are using material properties for the aluminum alloy 6061-T6.

Video 4A: Meshing the model 1) With the selection tab set on surfaces, select the surface on the end of the beam that
is to be fixed.

2) Right click and select Add. Select Surface Boundary Condition in the pop-up
menu. Click the Tx, Ty, and Tz boxes. Click Ok.

3) Select the top surface that is to have the pressure load applied.

Section 2: Linear Analysis

4) Right click and select Add. Select Surface Pressure/Traction in the pop-up menu.
Select the Pressure button and then add 2 in the Magnitude field. Check the box Follows Displacement. Click Ok. Checking the box Follows Displacement is important for problems that experience large displacements. If the box is not checked, the pressure will always act normal to the original surface orientation. If the box is checked, the pressure will act normal to the surface of the last computed equilibrium configuration. You are encouraged to run the problem with and without the box being checked to see the difference in the computed displacements and stresses.

5) Select Element Definition in the FEA Editor browser. Right click and select Edit
Element Definition. At the bottom of the pop-up menu beside Midside Nodes, select the arrow to obtain the Included option. Click Ok. This is an important option for thin sections that experience bending. Standard 8-noded brick elements do not have sufficient flexibility to model bending. Two options exist: 1) add a large number of elements through the thickness, or 2) increase the flexibility of the brick elements by adding mid-sized nodes. The two elements through the thickness are insufficient to accurately model the bending through the thin cross section, so mid-sized nodes are used. The mid-side nodes permit a quadratic displacement variation within the element. A quadratic displacement variation will result in a linear strain variation. The linear strain variation will result in a linear stress distribution.

6) Select 3D Mesh Settings in the ribbon menu across the top of the screen. Select
Options. Click on Model. Uncheck the top box, Use automatic geometry-based mesh size function.

7) Select the Surface icon. Click on the down arrow head beside percent of automatic
to obtain the Absolute mesh size option. Enter 0.0625 in the Size field. Click Ok.

8) Select Generate 3D Mesh in the ribbon menu across the top of the screen to
generate the mesh on the outside surfaces of the beam. Note that the pressure and displacement constraints are automatically applied to the nodes of elements associated with the designated surfaces.

Video 4B: Problem Setup

1) Select Analysis Type in the FEA Editor browser, then select Edit Analysis Parameters. 2) Set the Event Duration to 1 second. Time is used with a static analysis because some materials have time dependent properties. Examples include viscoelastic models for polymers and creep models for high temperature metal applications. In our problem we can use any time value. 3) Set the Capture rate to 1. This indicates that the load will be applied in 1 increment and that the stress and displacement data will be saved for later analysis at each increment. If a value of 2 were used, then the load would be applied in 2 increments (50% and 100%) and the stress and displacement data would be saved for each load increment. 4) Set the first row of the Time column for the Load Curve table to 0 and the first row of the Multiplier column to 0. Set the second row of the Time column to 1 and the second row of the Multiplier column to 1. Note that the time value of 1 entered in the second row agrees with the Event Duration entered in step 2.

Section 2: Linear Analysis

5) Select View Plot to see a plot of the load curve. It appears as a ramp which means that the load will be increased linearly from a value of 0% at time 0 to 100% at time 1. Click Ok. 6) Select the Gravity/Acceleration tab. Set the X-multiplier to 0, the Y-multiplier to -1, and the Z-multiplier to 0. This will cause the gravitational force to be applied in the negative-y direction. Set the acceleration due to body force to 386.4. If a problem is modeled that has a constant acceleration or deceleration, this same location can be used to enter the information. Click Apply. 7) Select the Output tab. Uncheck the box Nodal forces due to contact. Since the problem being solved does not have contact, unchecking this box will not have any effect. It is being unchecked to make the user familiar with the option. 8) Click on the Advanced button at the lower right hand corner of the Output menu. This provides access to several addition options that can have a significant effect on the solution of a non-linear problem. 9) Select the down-arrow associated with the Nonlinear iterative solution method. There are six equilibrium iteration methods available. Which of these methods is best for a particular problem is not always clear and a trial-and-error process may be required. A Full Newton w/o line search works best for this problem. The modified Newton options will not converge and the Full Newton w/line search converges slowly. Select Full Newton w/o line search. The user is encouraged to select the other iteration methods to see what happens.

Video 4C: Running the Simulation

1) Click on the traffic light icon, Run Simulation, to start the analysis. 2) The first thing that occurs during an analysis is the generation of the nodes and elements within the interior volume of the part. For large systems this can take considerable time. 3) As the analysis proceeds, load step status information will be displayed in tabular form. The columns are Time, Percent, DT, L, Iter, and Residual. a. Time gives the instance in time for which ASM is seeking a solution. b. Percent is the percent load being applied. c. d. L is the parameter used to decrease the size of the load increment if convergence is not being obtained fast enough. DT and L are related through the formula

DT

1 1 ( L1) CaptureRat e 2

e. Iter gives the current number of the equilibrium iteration being performed (incremented sequentially by ASM). f. Residual is the convergence parameter. Convergence is achieved when this number becomes less than the value specified in the analysis parameters. 4) The program will iterate until the specified convergence tolerance is achieved. Convergence is indicated by placing a small c immediately following the value of the Residual. If convergence is proceeding too slowly, the L parameter will be increased and a new time step will be computed.

Section 2: Linear Analysis

Video 4D: Static Analysis Results

1) ASM will automatically display a contour plot of the von Mises stress. The von Mises stress can be compared directly to the yield stress of ductile materials. It is also very easy to see the neutral axis of beams in bending using the von Mises stress. Since the stress at the neutral axis is zero, the von Mises stress will appear as a blue line in the contour plot. 2) The pressure and boundary conditions can be turned off to improve the visibility of the contours. Select the Results Options tab and then click on Loads and Constraints. 3) A safety factor defined as the yield stress divided by the von Mises stress can also be displayed as a contour plot. The safety factor defined by the following formula provides a factor that indicates how much the load can be increased before yielding occurs:

S.F .
where

ys eff

, is the von Mises effective

ys is the yield strength of the material and eff

stress. If the Safety Factor is less than or equal to one, the material will yield. The material will not yield if the Safety Factor is greater than one. The Safety Factor contour plot is obtained by first loading the yield strength of the material. Select the Results Contours tab and then click on the down arrow beside the Safety Factor icon. Select Set Allowable Stress Values and click on Load Yield Stress. Next click on the down arrow beside the Safety Factor icon and select View. A contour plot of the safety factor will be displayed. 4) In many cases an engineer has a good idea how a part will deform when it is subjected to loads and boundary conditions. Verifying that the deformations are consistent with expectations is an important way to check that loads and boundary conditions have been applied correctly. The displacements often have to be scaled to enable the part deformation to be seen on the screen. This is accomplished by going to Results Contours and selecting the Show Displaced on the top ribbon menu. Select Displaced Options to obtain a set of controls. These controls enable the displacements to be scaled as a percentage of the model size or as a multiple of the actual value. The cantilevered beam has large displacements that can be seen using a scale factor of one times the actual displacements. Select As an Absolute Value and enter a scale factor of 1 and close the box. The displacements can be turned off using this set of controls if no deformation is wanted in the plot. 5) Principal stress components and the six tensor components of stress can also be displayed as a contour plot. Tensor components are related to the Cartesian coordinate system used to model the part. The icons for selection of the Principal stress and the tensor components of stress are located on the ribbon menu of Results Contours. The

zz component of stress can be used to view the bending

stress through the thickness of the beam. Select Results Contours and then select the down arrow beside the Tensor icon. Select the ZZ Component. Note that a tensile stress occurs on the top surface while a compressive stress occurs on the bottom surface. The two values are essentially the same, which is consistent with beam theory.

Section 2: Linear Analysis

5. Module 5: Natural Frequency Analysis

Introduction
The natural frequencies and mode shapes of a system are important parameters that affect the dynamic response of the system. The natural frequency of a system is sometimes computed simply to compare it to the frequency of excitation to see if a resonance situation will exist. In other cases, the natural frequencies and mode shapes must be computed prior to performing a transient analysis. This is the case for a Modal Superposition analysis covered in Module 6. The time step used to perform a dynamic response analysis (Modules 6 and 7) is also based on the period of the highest natural frequency of interest. In addition, practicing engineers often compute the natural frequencies and mode shapes of a system to make sure that a system is constrained and responds as intended. Considerable insight can be gained about a system by studying its mode shapes. Accurate mode shapes are important to obtaining accurate dynamic stresses. Natural frequencies are less sensitive to modeling technique than mode shapes since they are based on the square root of the ratio of stiffness divided by mass. The stiffness can be off by a factor of two and the natural frequency will only have a 40% error. However, strains are based on the derivative of displacements and displacements are directly related to the mode shapes. Errors in mode shapes lead directly to errors in strains and associated stresses. The problem analyzed in this module is a cantilevered beam. The natural frequencies and mode shapes are first calculated without any static loads (video for Module 5A). The natural frequencies and mode shapes are then calculated when the beam is subjected to a tension force acting along its length (video for Module 5B). The shift in frequencies due to stress stiffening is very apparent.

Execution Video 5A: Natural Frequency (Modal)

The cantilevered beam used in Module 4 will also be used in this module. Once the beam is meshed, the Analysis Parameters must be entered. Right click on Analysis Type and select Edit Analysis Parameters. 1) First set the Number of frequencies/modes to calculate to 5. The number of modes to be calculated depends on what their use is going to be. In our case, we will use the natural frequencies and mode shapes to compute the transient response of the cantilever in Module 6. Several runs were made with each using a different number of natural frequencies and mode shapes. Five modes were determined to provide a sufficiently accurate result. 2) Next set the Lower cut-off frequency to zero and the Upper cut-off frequency to 1000. This establishes a range within which the algorithm will search for five frequencies. If it doesnt find five natural frequencies, the algorithm will only return the number found within this range. Zero is a common value for the Lower cut -off

Section 2: Linear Analysis

frequency. Some trial and error may be required to find a suitable Upper cut-off frequency. If the value you enter doesnt result in the desired number of modes, you will need to re-run the analysis using a higher number. Make sure that the Rigid body modes are expected box is not checked. The numerical algorithm will use a frequency shift more suited to extracting natural frequencies of zero Hz if this box is selected. Close the Analysis Parameters box. The natural frequencies and mode shapes are computed by selecting the Run Simulation icon under the Analysis tab located at the top of the screen. After the analysis is complete the first mode will be displayed in the results section. The frequency associated with the first mode is displayed along with the mode number in the lower left hand corner of the screen. The scale factor used to display the mode shape can be changed to enhance the visibility of the mode shape. Select Show Displaced on the Results Contours menu. Select Displaced Options. The scale factor can be specified as a percentage of the model size or as an absolute value. As an example, set the scale factor to 10% of the model size, press enter, and close the box. Notice the increased amplitude of the mode shape. Other modes can be displayed by selecting them using the Load Case options. Each mode shape is treated as a load case and the arrows allow you to move up or down to obtain a specific mode shape. Note that the natural frequency for each mode shape is displayed in the lower left hand portion of the screen. Each of the mode shapes can be animated to aid in understanding the motion. The animation start, pause, and stop buttons are located at the top of the Captures area on the top menu.

3)

4) 5)

6)

7)

8)

Video 5B: Natural Frequency (Modal) with Load Stiffening

The major difference between this type of analysis and a standard natural frequency and mode shape analysis discussed in Video 5A is that a static analysis is first performed by ASM before computing the natural frequencies and mode shapes. The stresses computed in the static analysis are used to form the stress-dependent contribution to the tangent stiffness matrix. Tensile stresses will stiffen the beam and raise the natural frequencies. Compressive stresses will decrease the stiffness of the beam and lower the natural frequencies. The same cantilevered beam used in Video 5A will be used in this demonstration. The process used to mesh the beam is shown in Video 4A. 1) The analysis type for this analysis should be Natural Frequency (Modal) with Load Stiffening. If it is not, select Analysis Type, right click and select Set Current Analysis Type, select Linear then Natural Frequency (Modal) with Load Stiffening. 2) With the meshed model in the FEA Editor, select the end surface on the free end of the beam, then right click and select Add Surface Force. Enter a magnitude of 100 and select the z-direction button to apply a 100 lb load acting in the z-direction. This creates a tensile zz stress of 1,600 psi in the beam. 3) Select the Analysis Type, right click and select Edit Analysis Parameters. The same parameters will be used as in Video 5A. Set the Number of frequencies/modes to calculate to 5, Lower cut-off frequency to 0, and Upper cutoff frequency to 1,000 Hz. 4) We are not going to use the Linear Modal results as a starting point for the current eigenvalue extraction so the Begin with Linear Modal results box is not checked.

Section 2: Linear Analysis

Using this option could decrease the analysis time and is recommended for large problems. Our problem is small, so it is not necessary. Select the Gravity/Acceleration tab and set the X multiplier, Y multiplier, and Z multiplier to zero. Gravity would normally be set to an appropriate direction. However, in this problem we want to see what effect the axial load by itself will have, so gravity is ignored. Close the Analysis Parameters box. Select the Analysis tab and then select the Run Simulation icon. The tools available for analyzing the results are the same as those shown in Video 5A. A comparison of the natural frequencies obtained without and with load stiffening shows that even a relatively small stress can have a big effect on some of the modes. The results are compared in the following table.

5)

6) 7) 8)

Mode

1 2 3 4 5

Natural Frequency (Linear) [Hz] 28.1 175.7 222.0 491.9 615.6

Natural Frequency (Load Stiffening) [Hz] 54.8 220.4 228.0 533.9 616.7

6. Module 6: Transient Stress (Modal Superposition)

Introduction
Modal Superposition is the first of two methods available in ASM for determining the dynamic response of a system to time-dependent loads. Modal Superposition is limited to linear systems having constant natural frequencies and mode shapes. Problems involving contact or other mechanisms that result in changes in the natural frequencies must use direct integration methods covered in Module 7. Modal superposition uses natural frequencies and mode shapes as input data and a natural frequency analysis must be performed prior to performing a modal superposition analysis. Computing the dynamic response of a system involves solving the equations of motion using numerical methods. The solutions are more time consuming and require more computer resources (disk space for storing results) than static analyses. The modal superposition method offers a relatively quick method for computing the dynamic response of systems in which the response is controlled by a few low frequency modes. In these cases the computation time can be significantly reduced. A cantilevered beam subjected to a step input at its free end is used as an example problem to demonstrate how to setup and perform a modal superposition analysis. The same problem is used in Module 7. The results for both methods are the same.

Section 2: Linear Analysis

Execution
The results from Module 5A are used as a starting point for this analysis. The natural frequency analysis is contained in Design Scenario 1. 1) Select Design Scenario 1 in the browser, right click and select copy. This creates a new design scenario with the same information that is in Design Scenario 1. 2) Select Analysis Type in the browser, right click and select Set Current Analysis Type, select Linear, and then select Transient Stress (Modal Superposition). 3) We must determine the identification numbers of the nodes where the loads will be applied. The loads will be applied along the top edge of the free end of the beam. With the selection mode set to vertices (select the Selection tab and select Vertices), select each node where a load will be applied, right click and sele ct Inquire. The node number and coordinates of the node will be displayed. Write the node numbers down, they will be needed in the next step. Select only the corner nodes. Loads are typically not applied to the mid-side nodes. 4) We are now ready to setup the analysis parameters required to perform a modal superposition analysis. 5) Select Analysis Type in the browser, then right click on Edit Analysis Parameters. A box appears that contains information required to perform the analysis. 6) Load information is entered in two places: 1) the Dynamic Load Data table and 2) in the Load Curves. The Load Curves contain multipliers that are a function of time. The multipliers are used to scale the vector components of the load that are entered into the Dynamic Load Data table. The vector components are entered in the X scale, Y scale, and Z scale columns. We will first define the applied loads by filling in the Dynamic Load Data table. 7) A row of the Dynamic Load Data table must be filled in for each nod e along the top edge of the free end of the beam. The node numbers found in Step 3 were determined for this purpose. The Add Row button is used to add rows to the table. Note that the Index in the first column is automatically incremented. The node numbers are entered in the second column. It does not matter what order is used to enter the node numbers. The Load Curve associated with each node will be entered in the next column. In this problem we will only use one load curve, so a 1 should be placed in each row of the Load Curve column. The Type column is used to tell ASM whether the vector components are for a force or a moment. Type 0 is a force and Type 1 is a moment. We will only be entering forces, so each row of the table should contain a 0 in the Type column. The X Scale, Y Scale, and Z Scale columns contain the vector components of the force to be applied to each node. In our problem a 5 lb force will be applied in the negative y-direction. This force will be divided equally among the 17 nodes located at the corner of each element. This gives a Y Scale factor of -0.294 lb in each row of the table. The X Scale and Z Scale columns should contain zeros. The last column contains an Activation Time. The activation time can be used to specify a time at which the vector components become active. They are zero before becoming active. In this problem we will use the activation time to simulate a step function that is zero until 0.05 seconds. At all times greater than 0.05 seconds, the vector components will be as entered in the X Scale, Y Scale, and Z Scale columns. The 0.05 seconds is an

Section 2: Linear Analysis

arbitrary number used in this example. The step function could also be specified using the Load Curves. This will be done in Module 7. 8) Select the Load Curves button to open the Load Curve data box. 9) Click the Add Rows button three times to add three rows to the table. 10) Add the following data into the load curve table:

Time(sec) 0.0 0.05 0.051 2.0

Factor 0 0 1 1

Note that the plot of the load curve that has a step at 0.05 seconds is automatically updated. Click Ok to exit the Load Curves data box. 11) The time step will be based on the highest natural frequency for a weak axis bending mode computed in Module 5A. The third weak axis bending mode has a frequency of 491.9 Hz. The period of this frequency is 0.00203 seconds. Common practice is to use a time step of 1/8 -1/10 of this period. This works out to a time step of 0.000203 seconds. 12) The lowest weak axis bending mode has a frequency of 28.1 Hz or a period of 0.0356 seconds. If we capture three complete cycles of this mode it will take 0.1068 seconds (3 * 0.0356 sec). We must also wait for the step function to start at 0.05 seconds. Adding 0.05 to 0.1068 seconds gives a total analysis time of 0.1568 seconds. A time step of 0.000203 seconds will require 772 time steps. Based on these calculations well set the number of time steps to 1000. This will give us slightly more than three cycles of the lowest weak axis bending mode. 13) The data stored when a transient analysis is performed can be quite large. We will set the Output Interval to 2, which will cause the results associated with every other time step to be saved. Setting this value to 1 would cause the results associated with every time step to be saved. 14) The Damping Factor is set to 0.005 or 0.5% of critical. This damping value is typical of that found in a structure not having any connections. 15) All of the analysis parameters have been entered and we are ready to submit the analysis. Select the Analysis tab in the top menu, and press the Run Simulation icon. 16) After the analysis is complete, the program will switch to the Results mode. 17) All of the tools available for static stress analysis are also available for each load case of dynamic analysis. Each time step is considered to be a load case and different load cases can be selected using the Load Case arrows located at the top of the screen. 18) The ability to plot stress or displacement data as a function of time is a very important tool for the analysis of data from a transient analysis. 19) Click on a node in the model mesh. The default selection mode in Results is Vertex. If you cannot select a node, select Selection at the top of the screen and then select Vertex. 20) Right click and select Edit New Graph. A time series graph of the von Mises stress at the node you selected will appear. You can change the parameter being plotted by selecting a stress tensor component or displacement component at the top of the screen.

Section 2: Linear Analysis

7. Module 7: Transient Stress (Direct Integration)

Introduction
The second method available in Autodesk Simulation to determine the response of a system to time-dependent loads is Direct Integration. This method gains its name from the fact that it is a direct approach that does not attempt to reduce the size of the problem through a transformation process, as is done in a modal superposition analysis. Instead it uses numerical methods that integrate the equations of motion containing the full size and form of the stiffness and mass matrices. Although this method will be applied to linear problems in this module, it can handle a wide range of non-linear phenomena and forms the basis for the methods used for non-linear problems. The direct integration method presented in this module is an implicit method that requires the formation of the stiffness and mass matrices. Explicit integration methods (not covered in this Section) are used for problems requiring very small time steps to capture high frequency content. The same cantilevered beam used in Module 7 is used for this module so that the results from both methods can be compared.

Execution
1) The first step is to apply a five lb force acting in the negative y-direction to the top edge of the free end of the beam. This is accomplished by changing the selection mode to Edge, selecting the top edge of the free end, right clicking and selecting Add edge load. After the data entry box appears, enter a Magnitude of -5, and set the direction cosine for the y-direction to 1 by selecting the y button. Closeout the box by selecting Ok. 2) The next step involves setting up the analysis controls by editing the analysis parameters. Select Analysis Type in the browser, then right click and select Edit Analysis Parameters. 3) The Analysis Parameters data entry box contains areas for defining the Event, Damping, Load Curve, and Dynamic Load Data. The Dynamic Load Data table could be used to define the loads acting on the structure. This table was used in conjunction with the Modal Superposition Analysis in Module 6. Since we entered the 5 lb load using the FEA Editor, we dont need to use this table. 4) Next we will define the Event information. This includes the number of time steps, time step size and output interval. Determining the time step can be a trial and error process. In our case we know the natural frequencies of the beam from Module 5. The time step will be based on the highest natural frequency of a weak axis bending mode. The third weak axis bending mode has a frequency of 491.9 Hz. The period of this frequency is 0.00203 seconds. Common practice is to use a time step of 1/8 1/10 of this period. This works out to a time step of 0.000203 seconds. 5) The lowest weak axis bending mode has a frequency of 28.1 Hz or a period of 0.0356 seconds. If we capture three complete cycles of this mode, it will take 0.1068 seconds (3 * 0.0356 sec). We must also wait for the step function to start at 0.05

Section 2: Linear Analysis

seconds. Adding 0.05 to 0.1068 seconds gives a total analysis time of 0.1568 seconds. A time step of 0.000203 seconds will require 772 time steps. Based on these calculations well set the number of time steps to 1000. This will give us slightly more than three cycles of the lowest weak axis bending mode. The data stored when a transient analysis is performed can be quite large. We will set the Output Interval to 2, which will cause the results associated with every other time step to be saved. Setting this value to 1 would cause the results associated with every time step to be saved. Next well define the load multiplier that specifies how the 5 lb load will vary with time. The magnitude of the load is 5 lb and the load multiplier ranges from 0 to 1. In our case we will define a step function that goes from 0 to 1 at 0.05 seconds. The load multiplier information is entered in the table and is automatically plotted in the graph window. We will need to add three rows to the table. Press the Insert Row button three times. Next enter the following information into the table: (0,0), (0,0.05), (0.0501,1), and (2,1). Note the step function plotted in the graph. It should be noted that the step function could be started at time t=0 by defining a constant load curve of 1. The 0.05 second delay is for demonstration purposes only. Close out the box by selecting Ok. Next, the Rayleigh Damping coefficients Alpha and Beta will be specified. These coefficients must be estimated based on the modal damping expected for each mode. We have used a 0.5% of critical modal damping value for the first three weak axis bending modes. Using the procedure outlined in the PowerPoint slides, we have determined that the values of Alpha and Beta are 4.768 and 0 respectively. These values are entered. Close the Analysis Parameters box. Next we will submit the analysis by selecting the Analysis tab and then picking the Run Simulation icon. After the analysis has completed, the results can be reviewed using the tools available in the Results section. These tools are discussed and illustrated in Video 6B.

6)

7)

8)

9)

Section 2: Linear Analysis

8. Module 8: Load Factor Analysis

Introduction
Buckling is an important failure mechanism for compressively loaded components. Buckling loads can be estimated using a Load Factor Analysis. Some systems will buckle after the loads have caused yielding to occur. A Load Factor Analysis is restricted to systems that are in the elastic region when buckling occurs. A load factor analysis uses the small-deformation and stress-dependent components of the tangent stiffness matrix to set up an eigenvalue problem. The eigenvalue problem returns a load factor that indicates how much the applied load must be changed to reach the buckling load. The stress-dependent component of the tangent stiffness matrix is a function of the stress in the system and a static analysis must be performed prior to setting up the eigenvalue problem. The static analysis is automatically performed by ASM as part of the analysis. A cantilevered beam subjected to a compressive axial load is used as an example problem for this module. The numerical results can be compared directly to the finite element results.

Execution
1) A 5 lb load acting in the negative z-direction is added to the end of the cantilevered beam using the FE Editor. 2) Next we edit the analysis parameters prior to executing the analysis. Select Analysis Type in the browser, then right click and select Edit Analysis Parameters. 3) The Load Multipliers can be used to change the loads or combinations of loads acting on the structure without having to change them in the FE Editor. This allows the analysis to be done multiple times using baseline loads. Our demonstration will not use these since we have only one load. 4) Next we will set gravity to zero so that the results can be compared directly to calculations made using hand calculations. The formulas used in hand calculations do not generally include gravitational effects. Select the Gravity/Acceleration tab. Set the Acceleration due to body force to zero. Close out the Analysis Parameters box. 5) The analysis is executed by selecting the Analysis tab followed by picking the Run Simulation icon. 6) Once the analysis is complete, Simulation automatically switches to the Results section and displays a deformed shape of the lowest buckling mode. The Buckling Load Multiplier is displayed in the lower-left-hand corner of the graphics window. The load multiplier of 5.6 times the applied load of 5 lb gives a buckling load of 28 lbs. This agrees with a 28.1 lb load computed using the buckling equation for fixedfree beam. 7) The various buckled shapes and load factors can be seen by toggling through the five modes computed by using the Load Case Options arrows located at the top of the screen. These higher order buckling modes are generally of little practical use since the design is controlled by the lowest load factor.

Section 2: Linear Analysis

9. Module 9: Fatigue Analysis

Introduction
Fatigue is a failure mechanism that results when cyclic loads cause internal dislocations in the microstructure to move to the surface. The molecular-level dislocations lead to surface cracks and subsequent failure. There are two methods used to determine the fatigue life of a component: 1) stress-life, and 2) strain-life. The stress-life approach is used for high-cycle fatigue situations that have stresses significantly below the yield strength of the material. The strain-life approach is applicable to both low and high-cycle fatigue and is recommended when stresses are near or exceed the yield strength. The stress-life approach uses an S-N curve, mean stress interaction curve, and a cumulative damage law to provide a framework for evaluating whether a part is susceptible to fatigue crack initiation and failure. The strain-life approach uses cyclic-strain data, mean stress correlations, and a cumulative damage law to estimate the fatigue life. Both methods require a technique for counting the number of cycles or strain reversals when variable amplitude loading is considered. The Fatigue Wizard in Autodesk Simulation can estimate the life of a component using either the stress-life or strain-life approach. The cumulative damage rule used in each case is Miners rule. The rain flow counting method is used to determine the stress/strain range and associated mean values for variable amplitude loading. In this module a cantilevered beam subjected to cyclic loads applied to its free end is used to demonstrate how to set up and perform a fatigue analysis.

Execution
A fatigue analysis is performed after a stress analysis has been performed. The Fatigue Wizard is accessed in the Results Section using the Analysis tab located at the top of the screen. 1) After entering the Results Section, select the Analysis tab then select the Fatigue Analysis icon. 2) The Fatigue Wizard guides you through the seven steps needed to perform a fatigue analysis. The first step is to select the type of fatigue analysis that will be used. The two options are Strain Based or Stressed Based. Note the warning on the screen that the Stress Based approach is only suitable for high cycle fatigue calculations. In this problem we will perform a Strain-Life analysis, so the radial button beside Strain Based (EN) is selected. We are using the Strain-Life method because we expect to be in the low cycle fatigue regime based on how close the peak stress is to the yield stress, and by how much it exceeds the endurance limit of the 6061-T6 aluminum used in the cantilevered beam. 3) In the next step we have the opportunity to modify the material properties used in the fatigue analysis. We will use the default values for the elastic modulus and tensile strength contained in the Simulations material library. We must enter Poissons ratio in the yellow field. A value of 0.3 is entered. The strain-life method requires material

Section 2: Linear Analysis

properties that define the slope and y-intercepts of cyclic stress-range and strainrange data plotted against reversals or (2N) using log-log scales. This information can be entered by selecting the Advanced button. It is possible to estimate these quantities from information such as the fracture strain, ultimate strength, etc. In this problem we will use the estimated values by selecting the Approximate button. We must also uncheck the My material is steel button since our material is aluminum. The next allows a stress concentration factor to be added. If the stresses in the area that controls the fatigue life are modeled accurately enough to capture a stress concentration, then we can enter a value of 1. If the model is too coarse to accurately estimate the stress, then a stress concentration factor can be used to raise the stress or strain to a more realistic value. In a linear analysis where the stress exceeds the yield strength, the program will estimate the actual stress and strain using the Neuber method. We can also select a surface finish life reduction factor based on values contained in the Specimen Data Modifiers area. Click the Surface Finish box and select the Machined option. Other life reduction factors can be specified by selecting the Miscellaneous box. In this case a number is entered to account for other factors. In the next step we define a load multiplier curve that specifies how the load varies during a cycle. In our problem we will define a curve that starts at zero, ramps linearly to a value of 1, and then ramps down to a value of zero. A more complicated load multiplier can be specified to accommodate variable amplitude loads. The Rain Flow Counting method is used to determine the amplitudes and mean values. Click on the Load Curve box. A box containing a load curve table and graph area is displayed. Additional rows can be added to the table by right clicking on the Multiplier box and selecting Insert row. We will need a total of three rows to enter our load multiplier data. Enter the following data pairs (0 sec, 0), (0.5 sec, 1), and (1.0 Sec, 0). Close the box by selecting Ok. In the next step we specify the type of information that we want from the fatigue analysis. We can select either Safety Factor or Cycles to Failure. Since we can easily calculate a safety factor if we know the life, well select Cycles to Failure. We must also enter a Desired Life. This information will be used for comparison purposes only. If the life is computed to be less than this desired life, the Fatigue Wizard will tell us that the design is unsafe. If it is greater, then Fatigue Wizard will tell us that it is safe. The value entered does not affect that computed fatigue life. In the next step we submit the analysis by selecting the stop light icon. In the final step we see the fatigue analysis results. Based on our desired life of 100,000 cycles, our design is considered to be unsafe because we have a computed life of 71,200 cycles. The columns in the table contain Log(N) at different nodes using various mean stress corrections. The second column has no mean stress correction; the third column is based on the Morrow correction; and the fourth column is based on the Smith-Watson-Topper correction. The Morrow correction is generally considered to be the most correct for compressive mean stresses while the SmithWatson-Topper correction is considered to be the most correct when the mean stresses are in tension. In our problem the top surface has a tensile alternating stress and a tension mean stress. Therefore the last column data provides the most appropriate estimate.

4)

5)

6)

7) 8)