See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/373015535

ABAQUS for Engineers

Book · August 2019

CITATIONS READS

0 218

1 author:

Bokwon Lee
Defense Rapid Acquisition Technology Research Institute
51 PUBLICATIONS 324 CITATIONS

SEE PROFILE

All content following this page was uploaded by Bokwon Lee on 10 August 2023.

The user has requested enhancement of the downloaded file.

Intentional Blank
ABAQUS for Engineer
A Practical Tutorial Book

By Ryan Lee
Aerospace Engineer, CAE Expert, Failure Analyst
ABAQUS for Engineers
A Practical Tutorial Book

First published 2018

Author Ryan Lee

Publisher BW Publications

CopyrightⒸ BW Publications 2019, All rights reserved

No part of this publication may be
reproduced in any material form (including
photocopying or storing in any medium by electronic
means and whether or not transiently or incidentally
to some other use of this publication) without the
written permission of the copyright holder.
Homepage https://sites.google.com/view/bw-engineering/

ISBN 9791163471264
Authors

Ryan Lee

University of Florida, Aerospace Engineering

Korea Advanced Institute of Science and Technology, Aerospace Engineering.
Currently working as an Aerospace engineer, Failure Analyst, CAE expert.
Contents
 Foreword

Foreword

Thank you for choosing this book to learn about the ABAQUS program. I think there will be a
variety of readers. The primary goal of this book is to provide practical tutorials of FE analysis
using ABAQUS program, which is a general-purpose finite element program.

The author encountered ABAQUS for the first time at the Graduate School of Aeronautical
Engineering in the United States. The supervisor assigned me a research project on the fracture
mechanics of composite materials. The first step was to define the experimental conditions
through finite element analysis before conducting any experiment. After completing the
analysis task and comparing the analysis result with the composite material tensile
experimental test, it was found that the analysis result was very similar to the test result. Here,
I started to get fascinated by the field of FEM and after graduation, I have personally taken
interest to continuously acquire information and knowledge on FEM related fields. Since then,
I have been working on computational analysis in the aerospace engineering field.
Recently, the finite element analysis process of various product designs is recognized as one
of the essential development processes to secure parts reliability and its stability through this
analysis process, or economic efficiency through optimization. Despite the increasing demand
for professional analysis in all parts of engineering schools and industries, there is a shortage
of professional analyst. There are people around the world who do not have a lot of
opportunities to undergo professional training, and also it is not easy to organize the data
scattered on the Internet, making it very difficult to learn. In order to ease this situation, I made
this book by gathering materials that I have organized little by little to share the knowledge and
know-how I had gained from my long experience to those who are interested in or want to
pursue this field.

Chapter 1 provides a basic introduction to those who have never experienced ABAQUS. In
Chapter 2, you will learn the general concept of FEM analysis using easy examples from
geometry modeling, meshing, defining the boundary and load condition, and result
interpretation. Chapter 3 examines the relationship between theoretical background and
interpretation using examples from various fields of physics analysis that can be covered by
ABAQUS. In Chapter 4, we will practice using the real analysis model that is actually applied
in the various engineering field. I hope you will learn a lot from this book and acquire what
you wanted. If you have any further questions, please feel free to contact the author.

August 2019
Part 1
What’s ABAQUS?

「 Workshop Model Folder」

Scan the QR code to download

FE models used in the workshop.

9|Part3 -W6
1.1 Starting ABAQUS

▐ Run ABAQUS / CAE

The Complete Abaqus Environment (CAE) manages the working environment for the required
analysis and also acts as a preprocessor and postprocessor for modeling and visualizing the
results. The first step in the analysis is to open the CAE for modeling and execution
environment management. There are two ways to run ABAQUS / CAE. First, as shown below,
if you enter 'abaqus cae' in the Abaqus command window, or enter 'abq6131 cae'(if your version
is 6.13) to open the CAE window

 ABAQUS Command  ABAQUS CAE icon

 How to run Abaqus 1) Using command window 2) Using the icon

When the CAE screen opens, you will see the following screen. The general CAD program is
similar in a general form. The functions of each specific element will be learned one by one
in subsequent exercises.

▲ ABAQUS/CAE screen layout

10 | P a r t 3 - W 6
 ▐Mouse usage and Part view operation

As with most CAD programs, getting used to the mouse usage has a big benefit on the
efficiency of rapid modeling. First, let's take a look at how to use a mouse for Abaqus
operations progression. Run ABAQUS / CAE to open the screen and you will see various icons
in the upper toolbar. Let's create a simple sample and try it out for yourself using the mouse. It
is recommended to use a 3-button mouse. We will call the left mouse button (M1), the right
mouse button (M2), and the center button (M3). Now let's create a random model as shown in
the following order:

 Part → Create → Continue

 Add → Line → Rectangle → Drag the mouse to create two rectangles of arbitrary size

 M3 button → Done → Extrusion Depth: enter 20 → OK

* Note that Abacus uses dimensionless units.

▲ Create two rectangles of arbitrary size

▲ Extruded to thickness 20

11 | P a r t 3 - W 6
First, let's learn how to manipulate parts in the viewer (working area). In the menu bar, there
are the following View operation buttons. We've organized the functions in order, especially
for the new users to get used to the keyboard shortcuts [Control] + [Alt] and the mouse buttons,
so you can work very quickly.

 View Manipulation tools

▪ Pan: Move parts left and right [control] + [Alt]: Mouse (2) button

▪ Rotation: Rotate the part [control] + [Alt]: Mouse (1) button

▪ Magnify: Zoom in / out the part [control] + [Alt]: Mouse (3) button

▪ Zoom: Zoom in the area

▪ Rescale: Reposition properly on the screen

▪ Cycle: Repeat previous View operation

Let's test the above functions with the parts we created earlier and learn the shortcut keys. You
will save a lot of time by following the practice models as given in the next chapter. For
example, if you hold down [Control] + [Alt] and press the mouse (M1) button and drag as
shown below, you can freely rotate without using the icon on the menu bar.

▲ Rotation Practice of Parts Using Shortcut Keys

Also, another way to do part view manipulation is by using 3D Compass in the upper right
corner. Click and drag the compass axis and vertices to move them. You can easily understand
these function while working on an upcoming tutorial.

12 | P a r t 3 - W 6
  3D Compass

ABAQUS CAE has a well-organized graphical user interface (GUI), so most of the menus can
be accessed through the GUI. Some features that are not provided within the GUI can be
generated by a Keyword Editor so that almost all FE modeling procedures can be carried out
in CAE environment. When you open CAE window, the main functions are included in the
dropdown menu called Module, and the FE analysis modeling process is almost done in this
order. Inside CAE, modeling features and result visualization are tree-structured. Here most of
them can be accessed during the modeling process and module menus can be accessed directly
from the tree structure.

▲ Main Module Menu and Tree Structure

13 | P a r t 3 - W 6
Let's take a brief overview of each module.

▲ Module Menu Overview

▪ Part: Process of creating geometries or importing CAD models for parts

▪ Property: Entering material properties of components and entering properties of sections

▪ Assembly: The process of creating an assembly by moving a part to the desired location.

▪ Step: The process of defining a step for interpretation and requesting output

▪ Interaction: The process of defining the contact and fastening conditions between
components.

▪ Load: Process that defines load condition, boundary condition, initial condition, etc.

▪ Mesh: The process of creating an element for analysis

▪ Job: a process of analyzing and monitoring

▪ Visualization: The process of presenting, reviewing, and analyzing analysis results

▪ Sketch: Process of creating the basic geometry for part modeling

14 | P a r t 3 - W 6
Part 3
Intermediate Analysis

15 | P a r t 3 - W 6
16 | P a r t 3 - W 6
3.2 Post Buckling Analysis of Composite Panel | Workshop 6 |

In this exercise, we will perform a buckling analysis on the composite panel. The problem of
buckling in thin panels is very important. In recent years, the use of composite panels, which
are thin, light, strong, and fuel-efficient structural materials, is expanding. However, the skin
of such an aircraft fuselage or wing has the form of a shell or a flat plate, and these structural
elements have a relatively low buckling strength compared to the static strength of the material
itself. Since bending stiffness of thin panel is structurally weak, understanding and designing
the buckling characteristics is a very important step.

▲ The main wing structure of fighter (consists of panel and stiffener)

In this exercise, we will perform a buckling analysis of two different composite panels used in
aircraft or space projectile structures. One panel is a panel with holes inside, and the other panel
is a panel with stiffeners inside. Buckling analysis of two models will be performed to calculate
Buckling Eigenvalue. In addition, RIKS analysis will be performed to predict post-buckling
behavior

(a) (b)

▲ Composite Panel Model (a) Simple Panel (b) Stiffened Panel

17 | P a r t 3 - W 6
▐ Buckling Analysis

Buckling analysis is the process of analyzing phenomena that define buckling or collapse as
Lateral Displacement and it occurs when the structure is elastically stable and suddenly the
structure is excessively bent laterally. In general, buckling causes fatal problems with the
stability of the structure. Generally, the buckling load is calculated in the unloaded state, but
the buckling analysis can be performed even when the preload is applied. The governing
equation for buckling analysis is generally similar to the eigenvalue analysis.

𝐾𝑈 = −𝜆𝐾𝐺 𝑈

Where KU is the linear stiffness matrix, KG is the stiffness matrix of the geometry, λ is the
eigenvalue (load vector), and U is the buckling mode shape. The buckling load vector λ is a
value multiplied by the existing load value. In the practical example, the unit load of -1 is
applied for buckling analysis. The Eigenvalue calculated from the result means the load vector.
Therefore, if 100 is calculated, the buckling load is -100 N which is multiplied by these two
values.

Mode shape, which is the shape when buckling occurs, is a normalized value and does not
mean the size of actual deformation. Normalized means that the maximum value of
deformation at buckling is set to 1 and the relative value is expressed. The mode shape at this
buckling is one of the most important elements of the analysis results. This mode shape allows
designers to choose where and how to reinforce.

 Buckling of a simple bar  buckled water tank

18 | P a r t 3 - W 6
For the simple bar, Euler's equation for buckling load calculation is

π2 ∙ E ∙ lmi n
𝐹cr =
l2

Where Fcr is the critical load value at which buckling occurs, E is the modulus of elasticity,
lmin is the minimum moment of inertia of the bar cross-section, and l is the buckling length of
the structural material. ABAQUS provides Lanczos and Subspace Iteration techniques for
buckling analysis. In general, the Lanczos method is much more effective when calculating
more than 20 Eigenmodes of a model with large DOF and subspace is more effective when less
than 20 Eigenmodes.

█ Step 1: Buckling Eigenvalue

As a first step, we are going to perform analysis that calculates the Eigenvalue through buckling
analysis. In this analysis, we shall create two models, one with a stiffener and another one
without stiffener, to check the difference between the buckling load and the buckle modes for
two cases.

▐ Geometry Modelling

First, make two shell parts as shown below to make the panel. The extrusion thickness of the
panel is 0.85m and the extrusion thickness of the stiffener is 0.75mm. Create a datum plane
based on the four corners of the panel.

 Part  Create  3D: Deformable: Shell: Extrusion→sketch the panel profile→Extrusion

depth=0.85 (panel)

 Part  Create  3D: Deformable: Shell: Extrusion→ sketch the panel profile →Extrusion
depth=0.75 (Stiffener)

 Tools→Datum→Plane | 3 points→pick 3 points on corners of the panel

▲ Composite Panel Model Profile (a) Panel (b) Stiffener

19 | P a r t 3 - W 6
To make a hole in the panel, make a datum plane and cut a hole of 0.42m X 0.14m as shown
below. Then attach the stiffener in the assembly as shown below.

 Shape  Cut  Extrude  select datum plane: sketch (see image)

▲ Composite Panel Cutout profile

▐ Mesh
Double-click Mesh in the Part tree structure and set the global seed to 0.02m before creating
the mesh.

 Model tree→Double Click Mesh in Part(panel) →Seed  Part  size=0.02

 Mesh  Control  Hex-dominated: Free

 Element type  Standard | Linear | S4R

 MeshPart→Yes (repeat for stiffener)

When you set the mesh shape condition to hex-dominated, about 2000 elements are created as
follows. Compare this to a working model.

>>Total number of nodes: 2026

>>Total number of elements: 1917

1899 linear quadrilateral elements of type S4R

18 linear triangular elements of type S3

20 | P a r t 3 - W 6
▐ Material Property
In this step, the material properties of composite materials are applied to panels which are made
of carbon composite. Since composite materials are already widely used in various fields,
engineers who understand FEM may understand some of its physical characteristics. In order
to model the composite material, classification needs to be looked. It can be classified into
Lamina, Engineering Constants, Orthotropic, and Anisotropic according to its shape and type.
In this analysis, Lamina is selected and modeled because it is a 3D Shell model. Enter the
physical properties as follows: Both panels and stiffeners will be modeled using the same
material.

 Density=1450kg/m3,

 Elastic Property: E1=139GPa, E2=35GPa, Nu12=0.15, G12=G13=50GPa, G23=42GPa

▲ Composite Material Property

 Material → Create | Material-1 → General | Density→ Mechanical | Elastic | Lamina (see

image)

The next step is to set up the stacked configuration of a composite panel. One of the most
important things in modeling composites is to create lamination and directional properties. The
figure below shows the dialog box that defines the stacking type of panel. Discrete was chosen
in this model to set the layup orientation. If you select Discrete, Normal Axis selects the surface
of the panel, and the Primary Axis selects the linear axis in the longitudinal direction as shown
in the figure. The thickness is 0.0015m, that is, 4 sheets of 1.5mm thickness per lamination
surface, and the lamination angle is [45/0 / -45 / 90] s. In the dialog box • Make calculated
sections symmetric means that the composite is made up of layers stacked symmetrically.
Finally, select the Offset tab to select the bottom surface.

21 | P a r t 3 - W 6
  Composite→Element type=Conventional Shell, Initial ply=4→Layup Orientation:
Discrete  Normal Axis: Surface (select Panel surface)  Primary Axis: Edge (select one
edge of Panel)

 Double click Region | select panel→double click Material | select Material-1→ double
click thickness | 0.0015→ input rotation angle (see image)

▲ Composite Layup setup

If this setting is made, the figure with the coordinate axis is displayed as shown below. N-3 is
defined as 3 axes in the normal direction of the shell, and P -1 is defined as 1 axis in the
direction of the primary axis.

▲ Layup Orientation display

In this way, the stiffener is also modeled based on the below figure. The lamination thickness
is 1.5 mm and the lamination angle is [45/0/45] s. Apply the same material and thickness.

22 | P a r t 3 - W 6
 ▲ Stiffener Layup Orientation

▐ Assembly
Copy the model you created so far from the model tree to create one more model. In the
Assembly module, model A loads only panel, and another model B loads panels and stiffener.

 [model A] Instance→Create→select panel

 [model B] Instance→Create→select panel and stiffener

▐ Step

In this analysis, we will be identifying the Eigenvalue in the Buckle analysis. Create a step as
follows and keep the default values for the other variables.

Step  Create  Buckle: Eigensolver | Lanczos: Number of eigenvalue= 5

▲ Step setup

23 | P a r t 3 - W 6
▐ Interaction

In the interaction phase, the relationship between parts is defined. In the model, the relationship
between load point and part is defined and the contact state between stiffener and panel is
defined. First, create RP (Reference Point) at the position (0, 0, 0) where the load is applied as
shown below. Connect the upper part of the panel with the Constraint as shown below.

 Constraint  Create  Coupling | Control Point (select RP), Surface (select top edge of
the panel)  Coupling Type: Kinematic  select all DOF

Load point
(0, 0, 0)

Top edge
▲ Constraint setup

Stiffener and panel are connected by applying a tie condition. Create a tie condition by selecting
the bottom surface of the stiffener and the inner surface of the panel.

 Constraint  Create  Tie  Surface(select inner surface of the panel)  Surface (select
bottom surface of Stiffener)

 Repeat it for other stiffeners

▲ Tie Constraint Setup

24 | P a r t 3 - W 6
Post-Buckling Analysis – Riks

Post buckling analysis after buckling is generally unstable and tends to diverge rather than
convergence. ABAQUS applies the RIKS method in this post-buckling analysis. Riks analysis
is generally used for very insecure structures, geometrically nonlinear buckling analysis, etc.
This technique is used when the stiffness less than zero occurs in the load-displacement as
shown below. Riks method is an analysis method that finds load and displacement at the same
time with load magnitude as an unknown variable.

▲ Typical Unstable Behaviors

In Riks analysis, the load increases proportionally. To understand the load applied to the riks,
it is necessary to understand the LPF (Load Proportional Factor). The relationship between the
applied Load and LPF is as follows.
LPF X Applied Load = Current Load

LPF inc X Applied Load = Load Increment

Ptotal = P0 + λ(Pref − 𝑃0 )

Ptotal is the current load, P0 is the dead load and the initial load at the start of the step. In
ABAQUS, LPF value is recorded every increment, so it is possible to check whether the
analysis proceeds normally by monitoring this value during job execution.

25 | P a r t 3 - W 6
█ Step-2: Buckling RIKS
In this exercise, we will use the results from Step-1 to perform a post-buckling analysis that
analyzes how the panel behaves after buckling. Copy the model created previously and enter a
new model name for Riks analysis.

 Model tree→Copy model (created previously)→enter a new name

▐ Step
Delete the existing step and set the step for Riks analysis as shown below.

 Step→Create→Static, Riks→Arc length | initial=0.01, min=0.0001, max=1 | max

number=100 | estimated total length=10

▲ Riks analysis step setup

The Total Arc Length entered in the Step is a factor that determines the initial LPF.

Δ𝑙𝑖𝑛
Δλin =
𝑙𝑝𝑒𝑟𝑖𝑜𝑑

The initial LPF increase is calculated at the rate of Art Length increase. In most cases, entering
1 as the total arc length will solve most problems. However, if you look at the status file while
26 | P a r t 3 - W 6
monitoring the analysis, the LPF will continue to print. If the increase in LPF is very small,
you can increase this total arc length. The maximum value is 1e + 36. In this analysis, 10 was
applied to obtain an appropriate value. LPF can be seen as the main variable to check progress
in all static analysis.

LPF X Applied Load = Current Load

LPF inc X Applied Load = Load Increment

▐ Keyword Edit

For RIKS analysis, the ‘* imperfection’ keyword should be inserted using the keyword editor.
The imperfection keyword is used to read the nodal displacement field calculated by the
eigenvalue analysis of buckling and to continue the post-buckling analysis. Job-2(could be
different for your case) inserted in Keyword refers to the analysis result of an existing model
(job-2.fil).

 Model Edit Keywords

▲ Select edit Keyword in the model tree

 Open Keyword edit→find **STEP→insert keyword(*imperfection, see image)→OK

▲ Riks model Keyword Editing

27 | P a r t 3 - W 6
* The imperfection keyword requires nodal displacement data in the buckling analysis and
should be requested in the buckling analysis. In order to generate * .fil(nodal displacement)
result when analysis Step-1 model, the following keyword should be inserted between **
FIELD OUTPUT and * End Step in Keyword of the model.

 Open Keyword edit (in previous buckling analysis model)→find *Output→insert

keyword(*Node file, see image)→OK

*NODE FILE, FREQUENCY=1

U

▲ Previous Step-1 model Keyword Editing (insert *node file)

If you revise the keyword of Step-1 model and re-analyze it, then you will see that job-2.dat
and job-2.fil files are created as shown below. The RIKS analysis is a post-buckling analysis
using the existing analysis results. Therefore, the following files should be included in the
working folder.

If you have followed well so far, open the * .dat file of the Riks model result later, and you
can see that the scale factor for each mode is properly applied in the job.dat file as shown
below.

*IMPERFECTION, STEP=1, FILE= __your_modal_analysis__

1, 0.002
2, 0.001
3, 0.0005

28 | P a r t 3 - W 6
 ▲ Riks model result file (Data file)

▐ Load and Boundary Condition

In Load applied to Riks model, input the first buckling load (-4.78MN) obtained from the
previous analysis result. For panel equipped with stiffeners [Model B], the buckling load (-
14.34 MN) is applied.

 Load→Create→step-1 | concentrated force | CF3=-4.78e6 (for Model A)

 Load→Create→step-1 | concentrated force | CF3=-14.34e6 (for Model B)

▲ Loading condition setup

When this is done, run the Job to analyze and check the analysis results.

▐ Result Visualization
To analyze the post-buckling behavior of the Riks model, plot the LPF history output data using
Abaqus's XY plot function as shown below.

29 | P a r t 3 - W 6
 ▲ History output and LPF XY plot setup

▲ LPF Analysis Results of Without Stiffener

Remaining are removed

30 | P a r t 3 - W 6
Part 4
Experienced Analysis

31 | P a r t 4 - W 1 2
4.4 Collision Analysis of Crash Box in Auto-body | Workshop 12 |

Crash Box or crush cans is one of the vehicle structures, located at both front and rear side
absorbs the energy when the vehicle crashes and delivers the minimum impact energy to the
body frame. Recently, as vehicles have become lighter, the vehicle safety issue has to be
considered as a priority. As shown in the figure below, the crash box is attached to the inside
of the bumper located in the front and rear of the vehicle, effectively distributing the collision
energy generated at low-speed collision and transmitting it to the rail. Minimized energy is
delivered to the passengers, enhancing the safety of the injured. In order to improve the energy
absorption rate, various types of crash box designs have been developed. Recently, a thin tube
type filled with a shock-absorbing foam material inside the crash box has been widely used,
and various shapes have been studied.

In this example, collision / Impact analysis will be performed in the octagonal tube shape to
improve the plastic deformation energy generated during the Buckle by including the groove
on the surface.

Linkage

▲ Crash Box Absorption Concept

▲ The various Crash Box design

32 | P a r t 4 - W 1 2
The first step of collision analysis is to do a Buckle analysis and then the * imperfection
keyword is used to reflect the axial buckling characteristics of the collision.

 Step 1: Buckling Analysis (Buckle)

 Step 2: Crash Analysis (Explicit, dynamic)

█ Step 1: Buckling Analysis

▐ Geometry Modelling
The crash box consists of two rigid plates and two crash box shells (Type 1). The crash box
part was created by CAD program and imported into CAE as * .igs or * .stp to create the part.
To compare the relationship between shape change and collision energy absorption, we shall
model and compare a simple crash box with no surface groove (Type 2).

 [Type 1] Crash box with side groove, [Type 2] Crash box without groove

 File→Import→Select crash box(*.stp)

▲ Crash Box (Type 1) Assembly

▲ Crash Box (Type 2) Assembly

33 | P a r t 4 - W 1 2
For the first stage of buckling analysis, model the ground support plate and loading plate to
apply the load acting on the upper part.

 Part → Create → 3D | Discrete | Rigid | Shell | Planar → Create 0.8 X 0.8

 Tools→Reference point→select center at the center of the plate

▲ Plate (Discrete rigid)

▐ Material Property
Crash Box is made of steel. Recently, lightweight functional structural materials such as
aluminum alloy and composite materials have been used a lot. In this example, a 3mm-thick
shell box made of steel was analyzed. Material properties can be created by entering Density,
Elastic, and Plastic property values referring to the picture below.

 Material→Create→ input Steel properties (see image)

 Section→Create→section-1 | Sell, Homogeneous | thick=0.003→OK

 Assign→Section→Select a part→set section-1→OK

▲ Elastic-plastic material properties for crash box shell

34 | P a r t 4 - W 1 2
▐ Assembly
In the Assembly module, the two created plates with two crash boxes are placed together to
form a total assemble model as shown below. There are many ways to do this, so please try to
use the assembly function of Assembly. Basically, you can assemble them by using only
translation and rotation, but another simplest method is too easily assemble is by using position
constraint function.

 Instance→Create→Part | select part → OK (repeat for two plates and two boxes)

 Assembled parts (see image)

▲ Assembled model and aligning tools

▐ Step
The first step of the analysis creates a bucket step because the crash box component is used to
check the buckling phenomenon under the axial compressive load. Request the analysis to write
up to the fifth Buckle mode, define the maximum number of interactions to five, and set the
maximum number of iterations to 100.

 Step→Create→Buckle | Eigensolver=subspace | number of eigenvalue=5 | max number

=100 → OK

▲ Step Setting (Buckle)

35 | P a r t 4 - W 1 2
▐ Interaction

The lower ground plate and the crash box is fixed by using a tie constraint, and the upper plate
and box where the axial compressive load is applied are set the contact condition. Then, connect
the left and right flange surfaces of the box with * Tie to connect half of the crash box part.
In practical world, the joining of metal plates is mainly performed by spot welding or fastener.
An analysis to see the macroscopic behavior of a structure may simply use * Tie. Modeling of
the fastening type of the metal plate may use connector elements, spring elements, and cohesive
elements in ABAQUS. Later in this example, we will cover spot welding modeling techniques
using connector elements.

 Interaction → Create → Contact | Name=No Fric | default → Ok

 Interaction → Create → Surface to Surface(Standard) | Initial Step → Property | No Fric

→ Select Top Plate and top flange of the Box (see image)

 Constraint → Create → Tie → [1] Ground plate-Box Lower Flange

 Constraint → Create → Tie → [2] Crash box Left flange set (See image)

 Constraint → Create → Tie → [3] Crash box Right flange set (See image)

▲ Interaction Setting

36 | P a r t 4 - W 1 2
▐ Loading and Boundary Condition

The axial compressive load applied during the buckling analysis is applied as -100N to the RP
of the upper plate. The boundary condition is constrained to all other DOF = 0 except for U1,
the direction of loading on the upper RP, and all DOF = 0 on the RP of the lower plate.

 Load → Create → step-1, Mechanical | Concentrated force→Select RP of top plate →

CF1=-100

 BC → Create → Initial, Mechanical | Displacement→ Select RP of top plate → constrain

all except for U1

 BC → Create → initial, Mechanical | Displacement→ Select RP of ground plate →

constrain all DOF

▲ Boundary and Loading Condition

▐ Mesh

Mesh can be created similar to the figure below by using the auto mesh option by double-
clicking the Mesh item in the Part tree. Set the Seed size to 0.02 and keep all the element types
in Quad. Since the shape is relatively simple, there will be no difficulty in creating the mesh.
S4R elements are used for both standard and explicit, so there is no need to set the element
type separately.

 Seed→Part→global size=0.02→OK

 Mesh→Control→Select part→Quad→OK

 Mesh→Part→Yes

37 | P a r t 4 - W 1 2
 ▲ Generated Mesh Model

To extract the deformation information for the buckling shape, right-click on the model item
in the model tree, enter Edit Keyword, and enter * Node file to extract the deformation
information of the node. The extracted Buckling Mode Shape data will be used in the later
post-buckling analysis. If the following keyword is entered, the file 'jobname'.fil is created and
the U (displacement) information of the node is generated in ASCII code in the file.

 Right Click on Model tree→Edit Keyword→ enter keyword after output request

*Node File

U,

▲ *Node file in Keyword Editor

So far, we have completed the buckling analysis model using the type 1 crash box. Apply the
same procedure to create a model for a Type 2 crash box.

38 | P a r t 4 - W 1 2
▐ Results Visualization
After completing the bucket analysis, check the buckling load and buckling mode. As shown
in the figure below, it can be seen that the buckling mode shape occurs in the local area without
grooves. Critical load (24.1kPa) of Mode 1 and Mode 2 is almost similar and the shape is
similar, so it should be judged as Buckle phenomenon of the same mode. In Mode 5, it can be
seen that buckling occurs globally on the side surface.

▲ Mode 1 Buckling shape(24.1k) ▲ Mode 2 Buckling shape(24.1k)

▲ Mode 3 Buckling shape(24.9k) ▲ Mode 5 Buckling shape (26.6k)

▲ Buckling Mode Shape (Type 1)

▐ Model Modification (Collision analysis)

So far, we have performed buckling analysis on the model including the spot weld, and then
copy the model and convert it into a collision analysis. This can be done by repeating the Step
2A/B collision analysis (including *imperfection) When applying imperfection, the job file
name must specify the result file name of the spot weld model.

39 | P a r t 4 - W 1 2
 ▲ Tie constraint model ▲ Spot weld model

▲ Comparison of deformed shape after a collision

Comparing the deformation state of the final frame after the analysis, the spot-welded model
will show that the welded part of the side is opened. If you check the view option to see the
connector element to observe the load applied to the spot weld, you can check the position and
local coordinate system of the connector element for the weld as shown below. If the primary
variable is selected as the CTF, the total force acting on the spot weld area can be displayed as
shown below.

 View → ODB display option → Entity display | show connectors

▲ von Mises stress distribution ▲CTF distribution

Extract CTF3 using XY data from ODB field output function. Here CTF3 is the normal force
of the spot weld, which can be seen in the chart below which shows several connector elements
that rapidly decrease after an increase of more than 30k. This graph shows the broken spot
welds.

40 | P a r t 4 - W 1 2
 ▲ CTF distribution over time

In this workshop, the crash box collision analysis of automobiles has been introduced,
including the concept of buckling analysis, geometry imperfection, spot weld modeling with
connector elements and an overview of the crash energy absorption mechanism of the crash
box assembly part.

41 | P a r t 4 - W 1 2
Remaining are removed

42 | P a r t 4 - W 1 2
4.7 Bolt Thread Failure Analysis using XFEM | Workshop 15 |

In this exercise, we will study the bolt damage analysis using XFEM. Bolts and bolted joints
have been used for a long time for fastening various mechanical parts. From simple bolts
commonly found in homes to aircraft, it will be difficult to find mechanical parts without a
fastener. Mechanics on bolts have already been well studied, but damage from bolt breakdowns
can still be found easily, from simple toy damage to reactor damage. Research on the damage
mechanism of various bolts using FEM is still ongoing. ABAQUS has a feature called XFEM.
ABAQUS provides considerable functions in the field of fracture mechanics and XFEM
(Extended Finite Element Method) can simulate crack growth irrelevant to the mesh by using
contour integral of fracture mechanics to predict crack size and growth direction. to be.

◄ XFEM mesh independent crack

Through this exercise, we will examine the crack propagation phenomenon of the thread
through XFEM for the modeling method of the thread and the damage mechanism, and learn
the analysis method. The main contents to learn through this exercise are as follows. Since this
is an intermediate level, the basic modeling procedure has been simplified. Please check the
beginner's level in Part 2 and Part 3 for further explanation.

 Understanding bolt threads and Axisymmetric modeling methods

 Plastic property input method for bolt material

 Understanding Damage Initiation and Damage Evolution for XFEM Modeling

43 | P a r t 4 - W 1 5
  Understand the phenomenon of thread stripping

▐ What’s Thread Stripping

The term thread stripping can be described as shear failure of threads. As shown in the figure
below, it means that the whole thread is pulled out in the shear direction without enduring the
fastening force or working load acted by the bolt or screw thread and the thread falls off like a
coil. This is mainly due to design problems or damage due to abnormal overload.

▲ Damage due to thread stripping

The figure below shows the results of the model analyzed in this exercise. The stress
concentration in the thread and the crack growth pattern in the individual thread due to thread
stripping can be confirmed.

▲ Bolt analysis model and XFEM analysis result

The bolts we used range from cheap bolts to very expensive titanium high precision and high-
performance bolts. Most machine parts use standard products and the shape standard is largely
divided into Inch bolt and Metric bolt. The inch bolts to be modeled in this process are 7 / 16-
20UNF-3A bolts. Bolt nomenclature can be understood as follows. If you are a mechanical
design engineer, understanding this nomenclature will help you work. UNF bolts of the
following specifications are mainly used for the aerospace industry.

44 | P a r t 4 - W 1 5
 NOTE

▲ 7 / 16-20UNF-3A / B Thread Shape and Key Dimensions

For bolt analysis, model two parts of the bolt thread and female thread. The shape of the thread
is Helicoil, but the parts are constructed by assuming Axisymmetric about the circular axis.
Note that in Axisymm modeling, the axis that is symmetric must be aligned with the origin Y-
axis.

▲ Internal Thread, External thread geometry

45 | P a r t 4 - W 1 5
▐ Material Property

Bolt mechanisms seem simple, but they are not really simple mechanisms as many assume and
thread damage to bolts can take many different forms. All metals that make up the bolt and nut
have plastic properties, and the self-locking bolts also have fastening force in the plastic zone.
Therefore, for accurate analysis, the physical properties must be entered into the Elastic-Plastic
properties. The bolts and female threads used in this exercise will model the mounting of the
AMS4340 Steel bolts to the Aluminum 2025 site with female threads, respectively. First, enter
the basic properties of the material. Please refer to the picture below for plasticity.

 Aluminum Part: Density 2580, Elastic Modulus 72GPa, Poisson's ratio 0.33

 Steel Bolt: Density 7850, Elastic Modulus 210GPa, Poisson's ratio 0.29

(a) (b)

▲ Aluminum Plasticity (a) and Steel Plasticity (b)

The next step is to enter properties for Damage Initiation and Evolution to model crack growth.
To understand this clearly, it would be easy to understand the prior knowledge of Fracture
Mechanics. Although it is good to learn while following the modeling procedure, it is difficult
to perform accurate analysis without understanding the underlying background knowledge, and
it is recommended to fully understand the background knowledge and the meaning of each
parameter because there is a possibility of applying meaningless results to the actual design.
For XFEM, input the following properties.

 MAXS damage

 Damage Evolution

 Damage Stabilization Cohesive

46 | P a r t 4 - W 1 5
 ▲ XFEM Properties

Enter the following values in the MAXS Damage and Damage Evolution input fields and enter
0.012 for Damage stabilization cohesive.

▲ Maxs Damage and Damage Evolution Input Value

NOTE
▐ Damage modeling
ABAQUS provides a variety of models to define the damage of materials. Basically, it is
applicable to both 2D and 3D models. For metal materials, it is divided into a cohesive method
using the transaction-separation principle and LEFM (Linear Fracture Mechanics) method
using the virtual crack closure technique. The method to define the damage is to model crack
initiation and propagation. The model for the initiation of a crack is divided as follows:

 Cohesive: MAXS, MAXE, QUADS, QUADE, MAXPS or user subroutine(UDMGINI)

 LEFM: VCCT

The MAXS (Maximum nominal stress) model applied in this exercise is a model that defines
the damage based on the normal stress of the stress region and the stresses of the 1-axis and 2-
axis of the model.

The initiation and growth of the cracks in the XFEM are calculated using two parameters. Φ
Phi, PHILSM which means the crack surface and Ψ Psi, PSILSM which means crack front.
47 | P a r t 4 - W 1 5
Another variable, STATUSXFEM, refers to the state of an enriched region defined by crack
growth and has a size between 0 and 1, and 1 means a region where cracks have occurred
completely. There are two methods for modeling damage evolution using the Fracture Energy
and the Fracture Displacement criterion. In this exercise, Fracture energy calculated from the
KIC (critical stress intensity factor) was used.

▐ Create Assembly
The assembly module assembles the bolt part and the internal thread part as shown below. In
order to analyze the effect of stripping on the threaded area, two models are utilized, one is the
full model and the other 50% engagement models. Creating two models in a CAE file is easy
when performing an analysis. It is recommended to model the partial fastening model as a
procedure of copying and modifying the completed model after completing the fastening model.
Since the height of one thread is about 0.68mm, this shape is generated by translating about
0.34mm. When assembling threads, you can use the edge to edge constraints to easily align the
configuration parts as desired. In order to define the enriched zone in the XFEM modeling in
the future, the individual screw parts are partitioned as shown below. The reason for doing this
is to create an environment in which the cracks all form on individual threads.

 create constraint > edge to edge

(a) (b)

▲ Assembly (a) Full engagement (b) 50% engagement model

▐ Create Step

This analysis is a quasi-static analysis in which the working load increases linearly with the
step time. The analysis step should include two steps. The first step is to apply pre-tension load

48 | P a r t 4 - W 1 5
acting on bolt and second step is to apply pull-out force acting on the bolt. Create an analysis
step as shown below and check that Nlgeom is On. Enter step time as automatic, initial step as
0.01 and max increment as 1000. Also, as shown in Figure (c), set to generate 50 frames per
step to check the process of crack generation and growth.

(a)

(b) (c)

▲ Creating Step

As with this analysis, the XFEM analysis, which includes contact and plastic and failure
properties, may have problems with convergence due to significant nonlinearity. Therefore, to
compensate for this, the solution control factors are modified as follows. Check the
discontinuous analysis as shown below and increase the IA parameter to 20 and enter the other
elements together.

 Other > General solution control > Step1:Edit > Specify

▲ General solution control editor

49 | P a r t 4 - W 1 5
▐ Create Boundary Condition

The loads and boundary conditions acting on this model are as shown below. The pre-tension
load is 20kN and pull-out force 27kN is applied which is due to the torque acting on the bolt.
The bolt modeled in this analysis is a pull-out force acting as a bolt for fastening components
mounted on an aircraft propeller. To calculate this pull-out force, a separate model was created
as shown in the following figure to calculate the centrifugal force at the maximum rpm to
calculate the load on the bolt. The load acting on the propeller's component fastening was
calculated to be approximately 55.8 kN and the two bolts were fastened to about 27 kN for the
individual bolts. There is also an off-axis force other than the fastening shaft, but it is not
considered because the relative size is not large.

In Axisymmetric modeling, the significant point in the load input is that complete load should
be applied, unlike the overall shell model. In other words, although we only model 1/2 of the
model based on the axis of rotation, ABAQUS assumes that the entire part is generated in the
Axisymm model so that the entire load should be entered. Entering only half the load or
dividing the cross-section area is one of the common mistakes in Axisymm load input.

▲ Working load component and fastening bolt load calculation model

To create the load point, create a reference point on the shaft as shown in the figure below (a)
and connect the top of the bolt and RP with a kinematic constraint as shown in figure (b).

(a) (b)

▲ Generate load point RP and Kinematic constraint

50 | P a r t 4 - W 1 5
The load acts as a concentrated force on the RP for both pre-tension and pull-out forces, and
the magnitude is 1. Then, the magnitude of the load is entered using the Amplitude function so
that the load increases with the total time as shown in the figure below, and the generated
amplitude is selected in the load input window. Enter the boundary conditions as shown below

 Load > create > concentrated force > pick RP:CF1=0, CF2=1, amplitude= …

▲ Create Load Amplitude

U1=UR3=0
U1=U2=UR3=0
U2=0

▲ Boundary conditions

▐ Create Interaction

To model threaded contacts, you create a simple contact property. Contact Property>
Tangential behavior: Penalty: Define Fric Coefficient 0.2. Generally, the coefficient of
friction of the thread should be 0.2 ~ 0.3, and the lubricated thread or the cadmium (Cd) plated
thread should be 0.17 ~ 0.18. For the surface where contact is made, define the entire surface
of thread as a surface, and then set these two surfaces as Surface to Surface contact as shown
in the below image. The next step is to define the areas where cracks occur and grow, one of
the important procedures in XFEM. In the previous part or assembly phase, all individual
thread areas were partitioned. In order to create and grow cracks in the partitioned individual
51 | P a r t 4 - W 1 5
areas, select Special> Crack> Create and set the crack creation areas individually as shown
below. Initial cracks can be set in the area where cracks are generated, or damage model can
be modeled to create cracks in the area that meets the condition by defining a damage model
that is damaged initiated like this model.

▲ Define Contact surface

▲ Define Crack enriched region

▐ Create Mesh

To create a mesh, double-click Mesh option in the Model Tree. This allows you to modify the
mesh in an individual part, regardless of the mesh you created in the assembly. The mesh must
be created with the edge seed tightly through the edge seed as shown in the figure below so
that the cracks can be seen and developed in detail. Refer to the figure below to create a mesh
that looks similar. Element is set to CAX4R (Asymmetric stress) as Hex element.

(a) (b)

▲ Creating Mesh
52 | P a r t 4 - W 1 5
53 | P a r t 4 - W 1 5
Remaining are removed

54 | P a r t 4 - W 1 5
Appendix
Useful Tips and Information

55 | A p p e n d i x
A3. Meshing Tips
The core element of the FEM is the finite element created, as we have seen so far. There is no
more important factor that determines the reliability of analysis than creating a good quality
element. In general, the generation of brick elements is known to give the best FEM analysis
results. The figure below (left) is a model that meshes with Tetrahedron through automesh, and
the right side shows a hexagonal Brick element through a proper partition.

▲ Tet mesh and Hex mesh model according to the shape of the 3D element

▐ why do you put so much effort to create Hex mesh?

Comparing the analysis results based on a simple static bending test, a 4X4 Tet mesh created
is sufficient to obtain similar results as similar to the 2X2 Hex mesh, which is close to the
theoretical solution. Hex mesh is the preferred method for shorter calculation time because
most models can yield quite good results with fewer nodes and elements.

▲ Comparison of Hex Mesh and Tet Mesh

56 | A p p e n d i x
However, the hex mesh can be done only when the model is fairly structured, and it can be
seen that the hex mesh is possible only by partitioning effectively in ABAQUS. In most cases,
the model should be divided into squares so that the hex mesh can be achieved through
automesh. Therefore, Hex mesh is easy to make by simplifying a small hole or gap, step, fillet,
and chamber as much as possible. The Hypermesh program has a very good algorithm for such
a Hex mesh, so it may apply a procedure that separates the mesh using a dedicated program
such as Hypermesh. There was a time when many research papers were published only on
Mesh Generation techniques when the program's mesh generation function was not good as it
is these days. Especially in the case of CFD analysis, mesh quality is known to have more
influence on analysis results. The paper “Mesh generation is Art or Science?” Shows how much
research has been done on its importance.

▐ Meshing Technique

The model looks like this: This model includes Round, Chamber, Hole, Lug and so on,
ABAQUS / CAE cannot auto-mesh Hex mesh. However, high-quality Hex Mesh can be
achieved through proper partitioning.

▲ Complex geometry ▲ Automesh-results (Tet Mesh 18k)

▲ Hex mesh (20k) with Partition

57 | A p p e n d i x
CAE provides a variety of partitioning tools, but for some of the models, it is sometimes
difficult to determine which tool should be used for which partition. Here is a very effective
way to use it. First of all, what shape do you need to partition to make the hex mesh effectively?
It's important to cut the Rule of Thumb into the simplest shape right through the partition. In
other words, if the cube can be shaped in a partition, almost Hex Mesh is possible.

Import the Part and select Create Shell from the left menu of the Part module. Select the top
surface as shown below and sketch the profile so that the part can be cut into squares.

◀Create shell

▲ Create shell – Profile sketch

After setting the extrusion thickness so that the sketched shell can penetrate the whole part,
make sure to check ‘Keep internal boundary’ before extrusion. If you check in the Mesh module,
the parts will be displayed in a yellow (green) and green (structured) where Hex Mesh is
possible. If you still have an orange (Tet), that means you still need more partitions.

▲ Create shell – Profile sketch

58 | A p p e n d i x
If you make a shell to distinguish parts like this, if you delete the cell after clicking the Remove
selected icon among the display group management functions, only the remaining shell is
displayed as shown below. If you delete this shell from Geometry Edit, only the partitioned
parts will remain as shown below. If you mesh this, a Hex mesh will be created.

◀Removed selected

 Part Module → Tools → Geometry Edit | Face | Remove → select remaining Shell → OK

▲ Geometry of remaining shell element ▲ Part partitioned after deleting shell

Applying the process described so far, a high-quality Hex mesh is created as shown below.
What does quality mean here? The quality of a mesh is how well it maintains the shape of a
formal element. It means the aspect ratio of the element (the ratio of height to width), the plane
of the element, and the angle of the plane.

▲ A model that creates Hex mesh through the partition

59 | A p p e n d i x
▐ Some misconceptions about good meshes
So far we have looked at some tips for making a Hex mesh, but it is true that the Hex mesh
calculates better results than the model having a same number of Tet elements. But how
meaningful is this difference because today's computing power is so good on personal PCs?
Nowadays, few graduate schools have classes on meshing itself. Not surprisingly, today's
programs have good algorithms and are replacing many. So, today's generation of CAE
engineers will not have much understanding of the meshing algorithm. In the 80s and 90s, there
have been quite a few cases where degrees are earned by grid generation alone. So, what are
the misperceptions about meshes that are indispensable in FEM?

 Is Hexadron Mesh always better than Tetrahedron mesh?

As we have seen before, the Hex mesh gives a better result for the same number of elements.
Therefore, many textbooks have emphasized the creation of Hex (square) meshes. Also,
elements that simulate most of the physics accurately are often only available in Hex. However,
since the solver technology of the recent FEM program is mature enough, it is difficult to find
elements that do not support Tet mesh, and the speed of computation is so fast that the influence
of the number of elements is becoming less important. In addition, if the shape is a little
complicated, various pretreatments such as partitions for hex mesh should be performed. If we
compare the time required for this pretreatment and the gain of computation time with the hex
mesh, would it make sense to make a Hex mesh with a lot of effort?

 The smaller the mesh, the better?

One of the biggest misconceptions is the size of the mesh. Of course, it is true that the size of
the mesh greatly affects the analysis results. It is also true that the smaller the mesh, the more
it converges to a specific value. What we need to think about here is that a good mesh is not a
small mesh but how well a mesh can represent physics. In fact, it is much more important to
apply boundary conditions, load conditions, material models, etc., similar to actual physics,
rather than the mesh alone. Meshes are good enough to represent this kind of physics, but
should not always be small. In other words, a good mesh is not necessarily made small but
depends on how well you understand the physics you are trying to analyze.

 A good mesh should accurately represent your CAD model?

Most CAE designers are familiar with CAD and have a good level of use. So they produce
quite detailed and sophisticated models. Is a mesh that reflects all these details accurately in
the FE model? In most cases, this is not the case. A good mesh is a representation of the physics
experienced by the object structure, not an exact copy of the CAD model. The purpose of our
FE analysis is to calculate physical factors such as stress, strain, pressure, displacement and

60 | A p p e n d i x
heat of the system. Therefore, the FE model is a model that is simplified as much as possible
through engineering judgment to determine whether the precise and detailed parts of the CAD
model affect physics.

▲ geometry Simplification (Example)

 A good mesh can only be done through a separate professional mesh work.

Of course, it is true that it is good to do the mesh work as a separate program for creating the
mesh. If you are a power user who deals with the daily analysis program, you can understand
the physical phenomena of the system and the optimal conditions for the shape model. However,
if you are not an experienced user, you can create a mesh that is sufficiently good with the
automesh function provided by your program. Today's programs can calculate and control the
conditions for good mesh creation (calculation of shape, shape variables of mesh, hard edge
control, small gap control, etc.) numerically so that a good enough mesh can be generated by
automesh itself. If you are not an experienced user, try using the automesh provided by the
program rather than manually meshing.

 Are complex 3D meshes always better than 2d and 1d?

One of the biggest misconceptions is that this complex 3D model approach is the best. In this
industry, 3D is also known as a model for beginners. Experts who fully understand the physics
of the structural system to be analyzed create accurate shell models or beam models that can
sufficiently represent the physics and extract the desired results within the project period.

Although a dense mesh might look good and seem to have put a lot of effort into a realistic 3D
model, creating a shell model actually takes more time and effort. The FE model, which is
made of a huge mesh of 3D models, makes the problem bigger and more complicated and may
be far from a solution for problem-solving. Effective modeling, that is, a model that properly
reflects physics, can be a critical factor in the success of a project. FEA engineers should not
be misled by the complex and colorful results.

61 | A p p e n d i x
 Epilogue

A6. Which CAD format is best for you?

▌ Which one is the best for CAE?

In FEM work, modeling in ABAQUS / CAE is often insufficient. In fact, since the data sent
and received from suppliers is provided in a wide variety of formats, it is often difficult to work
with the FEM analysis model with the format shared. So which format is the most suitable for
CAE and can be handled without problems? Let's take a quick look at the CAD format.

There are many different CAD programs used by different companies or engineering sectors.
Someone who works with a company-provided CAD file will often be faced with: In some
cases, the imported file is missing parts, in some cases very badly distorted, or in some cases
no error at all. This often results in very small gaps that are inconspicuous, or where the faces
are separated.

In this case, data is sent and received in a standardized format. Understanding what a
standardized format is and how it was created, even if it is standardized, can help in
troubleshooting.

▐ Why doesn't the CAD file open?

Every company has its own style and form of program. Here, converting the generated model
into a standardized model and bringing it back to ABAQUS and restoring it to its original shape
often produces an uncorrectable error. In this section, we looked at the various CAD formats
and what was going on with Abacus.

62 | E p i l o g u e

View publication stats