Comsol Alex Course Program Book

Finite Element
Modeling Using
COMSOL Multiphysics
Alexandria
25-27 September, 2012
PROGRAM
Elnady Engineering and Agencies
2 Elgabal Elakhdar Buildings, Nasr City, 11471 Cairo, Egypt.
Tel: +20 2 23425763
Fax: +20 2 23421793
www.elnadycompany.com
Finite Element Modeling Using COMSOL Multiphysics

Day1 Agenda
09:00 11:00
Introduction to COMSOL Multiphysics
11:00 12:30
COMSOL Demonstraion
Installation on attendees laptop
Coffee Break
12:30 1:00
01:00 03:00
Tutorial session: Getting started with COMSOL

- Parallel plate capacitor
- Thermal cube
- Stresses on a wrench
- Electrical heating and thermal stresses in a busbar
Day2 Agenda
-
09:00 12:00
Background on Finite Element Analysis

Advanced Meshing Techniques
Coffee Break
12:00 12:30
12:30 02:00
Solver Sequence and settings

Model of Day: Heat Sink
Day3 Agenda
09:00 10:30
Results and Post Processing
10:30 12:00
Advanced Structural Mechanics examples

- Elbow Bracket, different Force Analysis
- Elbow Bracket, Elasto-plastic Analysis
- Tube Connector
- Fluid Structure Interaction
Welcome Letter & Trial License Details

Dear Colleague,
It is our pleasure to welcome you to the Multiphysics High Performance Computing Workshop. During
your registration, you should receive the following items:
The Program book including step-by-step tutorials for the minicourse

DVD with the latest COMSOL Multiphysics version 4.3. For installation, see the information about
the trial license below
Your Trial License
You will find an installation DVD of COMSOL Multiphysics 4.3.

The installation Code is 9FFFBFFF0-TYSW-121009-17073896-324157832
This is a full functional trial license with all modules enabled.
Expiry date: 2012-10-09
You are entitled for technical Support. Send an email to support@comsol.com
For pricing and ordering information, please contact info@elnadycompany.com
We wish you an informative and great Workshop!

Elnady Engineering & Agencies
Introductory Models
Tutorials
Parallel plate capacitor
Thermal cube
Stresses in a wrench
Tutorial 1: Parallel Plate Capacitor

What you will learn
More experience with the interface
Sketching geometries in COMSOL
How to have a domain not active in the analysis
Tutorial 2: Thermal cube

More Practice with the User Interface
Adding a 2nd Physics
Adding Thermal Expansion
Tutorial 3: Stresses in a Wrench

What you will learn
Hands on with the interface
The basic steps in all COMSOL models
How to show deformed plots
Tutorial 4: Electrical Heating and

Thermal Stresses in a Busbar
Tutorial 1: Parallel Plate Capacitor

What you will learn
More experience with the interface
Sketching geometries in COMSOL
How to have a domain not active in the analysis
Select 2D
AC / DC > Solid Mechanics
Stationary
Checkered flag to Finish
Create Geometry
Open Model 1
Pick Geometry 1
Change Length Unit to cm
Draw Three Rectangles
Set Materials
Right Click on Materials
Choose Open Model Library
Pick Built-In Material Library
Select Air
Right Click on it, Add to Model
(This will assign it to all subdomains
but we will only use it in the air
domain: Electrodes will not be
included in the analysis)
Electrostatics Active only in Air Domain
Pick Electrostatics in Model Builder

Clear all Selections (hit X )
Only Select Air Domain
Right-click to confirm
Set Boundary Condition Ground
Right-Click Electrostatics
Choose Ground
Choose Box Selection
Select Bottom Electrode
Confirm
Set Boundary Cond Electric Potential
Right-Click Electrostatics
Choose Electric Potential
Set to 100 Volts
Choose Box Selection
Select Top Electrode
Mesh Free Triangular (Extra Fine)
Solve
Pick Study 1
Hit =
Tutorial 2: Thermal cube

More Practice with the User Interface
Adding a 2nd Physics
Adding Thermal Expansion
Choose Physics and Study
Choose File > New

Select 3D
Select Heat Transfer in Solids in Physics
Choose Stationary under Study
Draw Geometry
Right click on Geometry
Choose Block
Leave as 1 x 1 x 1 [m]
Set Materials
Right click on Materials
Choose Open Material Library
Go to Built-in Section
Right Click on Aluminum 6063
Choose Add to Model
Assign it to the block
Set Boundary Conditions
Right Click on Heat Transfer

Select Convective Cooling
Choose All Boundaries
h = 1000 Text = 500
Right Click on Heat Transfer

Select Temperature
Choose bottom surface
Right click to confirm
Set Temperature as 300
10
Mesh and Solve
Right Click on Mesh

Select Free Tetrahedral
Select Size
Choose Coarser mesh size
Right Click on Study

Select Compute
Add Structural Equations
Right Click on model 1 node

Choose Add Physics
Open Structural Mechanics
Pick Solid Mechanics
Choose the finish (checkered) flag
11
Add Thermal Expansion

Open Structural Mechanics in Model Tree
Right Click on Linear Elastic Material Model
Choose Thermal Expansion
Change Temperature from user defined
to Temperature (ht/solid1)
Leave Ref Temperature as 293 K
Constrain Bottom Surface + Solve

Right Click on Solid Mechanics
Choose Fixed Constraint
Select & Confirm bottom surface
Right Click on Study
Pick Compute
Open 3D Plot group 1
Right Click on Surface
Add Deformation
12
Tutorial 3: Stresses in a Wrench

What you will learn
Hands on with the interface
The basic steps in all COMSOL models
How to show deformed plots
1) Set Dimensions, Physics & Study-Type
Select 3D
Structural Mechanics > Solid Mechanics
Stationary
Then the checkered flag to Finish
13
2) Import Geometry
Right click on Geometry

Choose Import Then Browse
C:\COMSOL41\models\COMSOL_Multiphysics\Structural_Mechanics
Choose Wrench.mphbin
Import
3) Set Materials
In Model Builder:
Right Click on Materials
Open Model Library
Pick Built-In
Select High Strength Alloy Steel
Right Click, Add to Model
Select Steel AISI 4340
Right Click, Add to Model
Pick Bolt, right-click to add
14
4 a) Constrain Bolt
In Model Builder:
Right Click on Solid Mechanics
Choose Fixed Constraint
In Graphics Window:
Pick the Bolt Cut End
Right Click to confirm
4 b) Add Load
In Model Builder:
Right click on Solid Mechanics
Choose Face Load
Pick end of the wrench
Enter F for Fz
15
5) Mesh
In Model Tree:
Right Click on Mesh 1
Choose Add Tetrahedral
Pick Size
Change to Coarser
Right click Mesh 1
Pick Build All
6) Solve
In Model Builder:
Pick Study 1
Choose Compute
16
Display Stress on Deformed Geometry

Open Results > 3D Plot Group 1
Pick Surface 1 > Replace Expression
Plot Von Mises Stress
Right click on Surface 1

Add Deformation
Tutorial 7: Electrical Heating

and Thermal Stresses in a Busbar
17
Your first model
Exercises:
Electrical Heating in a Busbar
Bus bar made of copper.
Bolts made of titanium.
Direct current 10 mV from a
transformer to an electric device.
Heat is produced due to the resistive
losses--joule heating.
Calculate the thermal expansion.
Work though exercises, but not the
appendix
18
Your first training model: A Copper Busbar

A busbar is an electric
current conductor
typically found in highpower applications.
You will investigate:
Joule Heating
Thermal Expansion
Flow with Cooling
for a certain busbar geometry
Boundary Conditions: Electric Currents
Electric Potential 10mV
Ground
19
Boundary Conditions: Heat Transfer
Cooling with a heat transfer coefficient

of 5 W/m^2/K and surrounding air
temperature at 293.15 K
Now youre ready for your own simulations!

The only difference between the hands-on and your problem is
Perhaps a few more physics
A bit more geometry
Some more loads & b.c.s
Go to it (!!!)
20
Capture the Concept
TM
Questions
& Answers
21
An Example: Electrical Heating in a Busbar

In order to get acquainted with COMSOL Multiphysics, it is best to work
through a basic example step by step. This section will cover the model
building procedure, highlighting several features and demonstrating
common simulation tasks. The example shows how to:
Set up a multiphysics simulation
Run a geometric parametric sweep
Expand a model, taking more phenomena into account, after successfully
running a first simulation

By following the directions in this section, you will create a model that
analyzes a busbar designed to conduct a direct current from a transformer
to an electric device (see below). The current conducted in the busbar, from
bolt 1 to bolts 2a and 2b, produces heat due to the resistive losses, a
phenomenon referred to as Joule heating.
Titanium Bolt 2a
Titanium Bolt 2b Titanium Bolt 1
This presents a classic example of multiphysics, since the electric

conductivity of the conductor is temperature dependent while the heat
source in the heat transfer problem depends on the electric current density.
In fact, the phenomena are strongly coupled, meaning one affects the other
and vice versa.
6
Once you have captured the basic multiphysics phenomena, you will also
have the chance to investigate thermal expansion yielding structural stresses
and strains in the busbar and the effects of cooling by an air stream.
The Joule heating effect is described by conservation laws for electric current
and energy. Once solved for, the two conservation laws give the temperature
and electric field, respectively. All surfaces, except the bolt contact surfaces,
are cooled by natural convection in the air surrounding the busbar. You can
assume that the electric potential at the upper-right vertical bolt surface is 10
mV and that the potential at the two lower horizontal surfaces of the lower
bolts is 0 V.
The Model Wizard
Start COMSOL by double-clicking its
icon on the desktop.
When the Model Wizard opens, select
a space dimension; the default is 3D.
Click the Next button to continue
to the physics page.
In the Add Physics step, click the Heat

Transfer folder, right-click Joule
Heating and Add Selected. Click the
Next button
.
The last Model Wizard step is to select

the Study Type. Preset Studies contains
studies that have solver and equation
settings adapted to the Selected
physics, in this example, Joule Heating.
Select the Stationary study type and
click the Finish button . Any
selection from the Custom Studies list
needs manual fine-tuning.
Global Definitions
The Global Definitions in the Model
Builder stores Parameters, Variables,
and Functions with a global scope. You
can use these operations in several
Models. In this case, there is only one
Model 1 node where the parameters are
used.
Since you will run a geometric
parameter study later in the example,
define the geometry using parameters
from the start. Do this by entering
parameters for the length for the lower
part of the busbar, L, the radius of the
titanium bolts, rad_1, the thickness of
the busbar, tbb, and the width of the
device, wbb.
2xrad_1
wbb
L
tbb
Also add the parameters that will control the mesh, mh, a heat transfer
coefficient for cooling by natural convection, htc, and a value for the voltage
across the busbar, Vtot.
Right-click Global Definitions and

select Parameters to create a list of
parameters containing L, rad_1, tbb,
wbb, mh, htc, and Vtot. In the
Parameters table, click the first row
under Name and enter L. Click the first
row under Expression and enter the
value of L, 9[cm]. Note that you can
enter any unit inside the square
brackets. Continue adding the other
parameters according to the
Parameters list. It is a good idea to
type in descriptions for variables, in
case you forget or share the model
with others.
Geometry
Here you can take a shortcut by
loading the model geometry from the
Model Library, which you open by choosing Open Model Library from the
File menu. In the Model Library, select
COMSOL Multiphysics>Multiphysics>busbar_geom and then click Open.
Alternatively, you can construct the busbar geometry using the parameters
entered above with the COMSOL geometry tools. If you would like to see
this now, have a look at Appendix: Creating Geometry Sequences on
page 42.
Once you have created or imported the geometry, experimenting with
different dimensions is easy: update the values of L, tbb, or wbb, and rerun
the geometry sequence.
Under Global Definitions click Parameters. In the Settings window, select the
wbb parameters Expression column and enter 10[cm] to change the value of
the width wbb. In the Model Builder, click the Form Union node and then
the Build All button

wider busbar.
wbb=5cm
to rerun the geometry sequence. You should get a
wbb=10cm
Return to the Parameters table and change the value of wbb back to 5[cm].
Rerun the geometry sequence.
Experienced users of other CAD programs are already familiar with this
approach since all major CAD platforms include parameterized geometries.
To support this class of users and avoid redundancy, COMSOL offers the
LiveLink family of products. These products connect COMSOL
Multiphysics directly with a separate CAD program, so that all parameters
specified in CAD can be interactively linked with your simulation geometry.
The current product line includes LiveLink for SolidWorks, LiveLink for
Inventor, and LiveLink for Pro/ENGINEER.
It is also worth noting that LiveLink for Matlab is available for those who
want to incorporate a COMSOL Multiphysics model into an extended
programming environment.
Having completed the geometry for your model, it is time to define the
materials.
10
Materials
The Materials node administrates
the material properties for all
physics and all domains in a
Model node. The busbar is made
of copper, and the bolts are made
of titanium. Both these materials
are available from the Built-In
material database.
In the Model Builder, right-click
Materials and select Open
Material Browser. In the Material
Browser, expand the Built-In
materials folder, right-click
Copper and select Add Material to
Model.
Click the Material Browser window tab and repeat the above procedure for
Titanium beta-21S.
In the Model Builder, collapse the Geometry 1 node to get a better overview
of the model. Under the Materials node, click Copper. In the Settings
window, locate the Material Contents section.
The Material Contents section
provides useful feedback on the
models material property
usage. Properties that are both
required by the physics and
available from the material are
marked with a green check
mark . Properties required by
the physics but missing in the material will result in an error and are therefore
marked with a stop sign . A property that is not used in the model is
unmarked. You may note that the Thermal expansion coefficient at the
bottom of the list is not used, but you will need it later for expanding your
model with heat induced stresses and strains.
11
In the Settings window, click the Clear Selection button

selection list.
to clear the
Click in the Graphics window to

highlight the copper domain.
Right-click anywhere in the Graphics
window to add the highlighted domain
to the copper Selection list in the
Settings window. Domain number 1
should now display in the list.
In the Model Builder, click the

Titanium beta-21S node. Click the
upper titanium bolt to highlight it and
then right-click to add it to the
titanium Selection list. Repeat this
procedure for the two remaining bolts.
These should correspond to the
domains 2, 3, and 4.
Be sure to investigate the Material
section in the Settings
window. All the properties used by the
physics interfaces should have a green
check mark . Close the Material
Browser.
Contents
Physics
The domain settings for the Joule Heating physics interface are complete
now that you have set the material properties for the different domains. Next
you will set the proper boundary conditions for the heat transfer problem
and the conduction of electric current.
12
In the Model Builder, expand the Joule

Heating node to examine the default
physics interface nodes. Joule Heating
Model 1 contains the settings for heat
conduction and current conduction.
The heating effect for Joule heating is
set in Electromagnetic Heat Source 1.
Thermal Insulation 1 contains the default boundary condition for the heat
transfer problem and Electric Insulation 1 corresponds to the conservation
of electric current. Initial Values 1 contains initial guesses for the nonlinear
solver for stationary problems and initial conditions for time-dependent
problems.
Right-click the Joule Heating node. In the second section of the context
menuthe boundary sectionselect Heat Transfer >Heat Flux.
Boundary Section
Domain Section
13
In the Settings window, select All

boundaries in the Selection list.
Rotate the busbar to inspect the
back. Click one of the titanium
bolts to highlight it in green.
Right-click anywhere in the
Graphics window to remove this
boundary from the Selection list.
Repeat this for the other two bolts.
This removes boundaries 8, 12,
and 28 from the Selection list.
Inspect the list to make sure only
these boundaries are removed.
In the Settings window, click the
Inward heat flux radio button.
Enter htc in the Heat transfer
coefficient field, h. Continue by
setting the boundary conditions
for the electric current.

the Joule Heating node. In the
second section of the context
menuthe boundary section
select Electric Currents>Electric
Potential. Click the circular face of
the upper titanium bolt to
highlight it and right-click
anywhere to add it to the Selection
list. The boundary face is
numbered 28.
14
In the Settings window, enter Vtot

in the Voltage field. The last step in
the physics settings is to set the
two remaining bolt surfaces to
ground.
the Joule Heating node. In the
boundary section of the context
menu, select Electric
Currents>Ground.
In the Graphics window, click one
of the remaining bolts to highlight
it. Right-click anywhere to add it to
the Selection list. Repeat this
procedure for the last bolt.
Click the Go to Default 3D View button
on the Graphics toolbar.
Mesh
The simplest way to mesh is to
create an unstructured tetrahedral
mesh. This will do nicely for the
busbar.
the Mesh 1 node and select Free
Tetrahedral. Click the Size
node .
In the Settings window, select the
Custom button under Element
Size. Enter mh in the Maximum
element size field to set the
maximum element size to 6 mm
the value entered earlier as a global
parameter.
Click the Build All button
to create the mesh.
15
Later, in Studying Mesh Dependence and Convergence on page 29, you will create
meshes for running mesh convergence studies and to investigate the
magnitude of the change in a solution parameter after refining the mesh.
Study
To run a simulation, in the Model Builder, right-click Study 1 and select
Compute
. The Study node automatically defines a solution sequence for
the simulation based on the selected physics and the study type. The
simulation only takes a few seconds to solve.
Results
The default plot displays
the temperature in the
busbar. The temperature
difference in the device is
quite small, less than 1 K,
due to the high thermal
conductivity of copper
and titanium. The
temperature is
substantially higher than
the ambient temperature
293 K.
With this solution in hand, you can create an image that will be displayed by
COMSOL when browsing for model files. Go to the File menu and select
Save Model Image. There are two other ways to create images from this plot.
One way is to use the Image Snapshot button in the Graphics toolbar for
directly creating an image, and another is to add an Image node to the Report
16
by right-clicking the Plot Group of interest. The second option lets you reuse
the Image Settings if you update the model.
The temperature distribution is symmetric with a vertical mirror plane
running between the two lower titanium bolts and running across the middle
of the upper bolt. In this case, the model does not require much computing
power and you can model the whole geometry. For more complex models,
you should consider using symmetries in order to reduce the size of the
model.
Another plot of interest shows the current density in the device. In the Model
Builder, right-click Results and select 3D Plot Group. Right-click 3D Plot
Group 3 and select Surface. In the Settings window, in the Expression section,
. Select Electric Currents>
click the Replace Expression button
Current density norm. This is the variable for the magnitude, or absolute
value, of the current density vector.
Click the Plot button . The resulting

plot is almost uniform due to the high
current density at the contact edges
with the bolts. You can change the
Range of the color table to visualize the
current density distribution in the
Settings window. Under Range, select
the Manual color range check box.
Enter 0.6e7 in the Maximum field.
Click the Plot button . Save the
model as busbar.mph.
This plot shows how the current takes the shortest path in the 90-degree
bend in the busbar. Moreover, the edges of the busbar outside of the bolts
are hardly utilized for current conduction.
17
Rotate the device to show

the back of the busbar
where you will see the high
current density around the
contact surfaces of the
bolts. Continue the
exercise by adding your
own plots and investigating
ways of generating cross
section plots and cross
section line plots.
Now you have completed a
basic multiphysics
simulation. The following
sections are designed to increase your understanding of the steps you
implemented up to this point as well as to extend your simulation to include
other relevant effects, like thermal expansion and fluid flow.
18
Importing CAD Files

When importing a design into COMSOL Multiphysics from a CAD file, the CAD
Import Module translates it into a format understood by COMSOL Multiphysics.
After importing a CAD file you continue the modeling process with meshing and
analysis in the usual fashion.
CAD Geometry for Finite Element Analysis (FEA)

A geometry created in a 3D CAD software might not automatically be suitable for
FEA. One reason is that the geometry description requirements are rather stringent.
For example, a 3D CAD geometry might contain very small gaps, edges, or surfaces
that are difficult for the user of the CAD software to control. These can make meshing
of the imported geometry impossible.
A common type of problem in CAD designs is the presence of sliver faces, which are
very long and narrow faces. While the CAD Import Module provides tools to detect
and remove such faces, in some cases the topology of the geometry is such that the
sliver face can not be removed by such tools. Such an example is illustrated in the figure
below.
4 |
CHAPTER 2: CAD IMPORT
After meshing this geometry, a very dense mesh is unfortunately created to resolve the
narrow strip of the face.
Alternatively the narrow strip can be easily avoided during the design of the geometry
in the CAD software, through making sure that the neighboring fillets have large
enough radii.
As the example above illustrates, meshing can be useful for diagnosing sliver faces and
other small features in your geometry, and you can use it as a complement to the tools
provided by the CAD Import Module. In COMSOL Multiphysics you can quickly
create a surface or volume mesh of the part, then examine it for areas of dense mesh,
which usually indicate small features.
Careful design in the CAD software is also important during the design of assemblies.
Touching parts in a CAD assembly might not be accurately positioned or might
contain slight mismatches between dimensions. Such unintentional errors also lead to
the presence of small features and sliver faces in the decomposed geometry. For this
reason it is recommended that parts of the CAD assembly are designed in the context
of the assembly.
Even geometry features on a larger scale can make meshing of imported geometries
difficult. Such features could be the subdivision of faces into smaller ones, or smaller
holes, fillets and chamfers. You can usually control these features in the CAD software.
Most modern CAD packages allow you to save different configurations of a part in the
same file. It is often beneficial to save a configuration specially for FEA, where you
IMPORTING CAD FILES
suppress features not important for the analysis. This way you can reduce the number
of mesh elements and thus the amount of memory needed to solve the problem.
3D CAD GEOMETRY OF AN ENGINE PISTON
In this example you import and mesh two versions of a 3D geometry of a diesel-engine
piston. The first version of the geometry is intended for FEA and has many features
suppressed. With the use of two symmetry planes you only need to analyze a quarter
of the original geometry. The second version is the complete 3D geometry.
1 Start COMSOL by double-clicking its icon on the desktop.
2 In the Model Wizard (
) the default space dimension is 3D. Since you do not need

any physics modeling, you can click the Finish (
)button.
)node is added to the Model Builder. The Geometry 1 node

The Geometry 1 (
contains the geometry sequence, where the geometry operations are stored. In the
Settings window to the right of the Model Builder you are able to change various
geometry settings.
3 To add an import feature to the geometry sequence right-click Geometry 1 and select
Import (
).
4 In the Settings window click the Browse button.

5 Locate the course CD on the hard disk and select the file piston_for_fem.x_b,
then click Open.
6 |
6 Click Import in the Settings window.
As soon as the import is finished the geometry appears in the Graphics window.
Continue by creating a simple unstructured tetrahedral mesh.

7 Right-click the Mesh 1 node and select Free tetrahedral.
The Mesh 1 node contains the operations that make up the mesh sequence.
IMPORTING CAD FILES
8 To build the mesh click the Build All (
) button.
The Messages window indicates that there are roughly 34,000 tetrahedral elements in
this geometry.
For the sake of comparison, import and mesh the complete 3D CAD geometry.
1 In the Model Builder select the Import 1 node.
3 Locate the course CD and select the file piston_design.x_b, then click Open.
4 Click the Mesh 1 node.
Since you have changed a setting, the Import 1 node is automatically rebuilt as you
leave the geometry sequence. This means that the newly specified file is imported
instead.
8 |
5 Click the Build All button to mesh the geometry.
According to the Messages window, this part consists of approximately 201,000

tetrahedral elements.
Observe the differences between the two geometries and how this effects the mesh.
The regions around small features such as fillets contain many small elements, which
contribute significantly to the total number of elements.
IMPORTING CAD FILES
I mport of I GES F i les

The Initial Graphics Exchange Specification, or IGES, data format was first published
in 1980, and supports data exchange for a wide range of formats including 3D solid
modeling representations. Although it is not as suitable for CAD data transfer as
modern file formats, the neutral IGES data format is still wide-spread for exchange of
geometry data.
In the IGES format, 3D data is handled by a trimmed surface representation, which,
in its basic form, does not include information about how the surfaces are connected.
Thus, the faces that make up a solid geometry have to be stitched together during the
import of the geometry, making the IGES format very sensitive to tolerances.
The latest update of the IGES format, version 5.3, addresses this issue by including
support for objects called Manifold Solid B-rep Objects (entity type 186), which can
contain surface connectivity information. Provided it is supported by your CAD
software, it is recommended that you use this option for IGES file export.
Knitting Faces after Import

In the majority of cases, IGES files do not contain any connectivity information and
the process of knitting the faces together into a solid object is sensitive to tolerances.
In this example, you can test how the knit tool included with the CAD Import Module
can help you with creating a solid object, even though there are gaps between faces
due to either design flaws or tolerance issues.
1 Click the New (
) button on the main toolbar.
2 In the Model Wizard make sure that the space dimension is 3D, then click the Finish
button.
Import.
5 Locate the course CD on the hard disk and select the file sphere.igs, then click
Open.
10 |
The geometry looks like a solid sphere, whose outer surface is split into two halves.

8 To build the mesh click the Build All button.
The mesh appears in the Graphics window, but a warning is displayed in the Settings
window that tetrahedral elements were not created since the geometry contains no
domains.
Add a Measurements node to the geometry sequence to examine the imported
geometry.
9 Right-click Geometry 1 then select Measurements.
IMPORT OF IGES FILES
11
10 Click the sphere object in the Graphics window to highlight it. Right-click anywhere
to add it to the Selection list.

The following information is displayed in the Measurements window:
The information indicates that the geometry is not a solid object since it does not
contain any domains.
By default, during import the CAD Import Module tries to knit faces and form
solids. If a solid is not created, it usually means that there were gaps, larger than the
default import tolerance of 10-5 in the geometry units, between geometry faces.
In these next steps you will create a surface mesh of the sphere to visualize the gap
between the two halves.
11 Right-click the Free Tetrahedral 1 node and select Size.
12 From the Geometric entity level list select Boundary.
13 In the Graphics window click one of the faces of the sphere to highlight it.
Right-click anywhere to add it to the Selection list.

14 From the Predefined list select Fine.
12 |
15 Click the Build All button.
Examine the meshed surfaces, note that the mesh only connects on a portion of the
common edge. On the portion where the mesh does not connect there is a narrow
gap between the faces.
Even when the geometry includes such a gap you can create a solid by using the knit
tool.
16 Right-click the Geometry 1 node and select CAD Repair, then Knit to Solid.
The Knit to Solid 1 node is inserted into the geometry sequence after the Import 1
node.
17 Select the sphere in the Graphics window, then enter 1e-4 in the Absolute repair
tolerance edit field.
18 To build the feature click the Build All button.
To check if a solid was indeed created you can use the Measurements tool.
19 Right-click the Geometry 1 node and select Measurements.
20 Add the sphere to the selection list, by selecting it in the Graphics window.
The displayed information still indicates that there are no domains in the sphere.
This means that the gap was larger than 10-4 m and the tool could not create a solid.
IMPORT OF IGES FILES
13
21 Click the Knit to Solid 1 node, then enter 1e-2 in the Absolute repair tolerance edit
field.
22 Click the Build all button to build the feature.
23 Right-click the Geometry 1 node and select Measurements, then select the sphere.
This time the object contains a domain, which indicates that a solid object was
created by the knit to solid tool.
24 Mesh the geometry by clicking the Mesh 1 node, then the Build all button.
This time a properly connected volume mesh for the solid object was created.
14 |
Modeling with Assemblies

A CAD file which contains an assembly or a multi-body part is a collection of solid
bodies. When importing such a design, each body becomes a geometry object in
COMSOL Multiphysics. In addition you can have additional geometry objects created
by various other features of the geometry sequence.
The collection of geometry objects resulting from the operations in the geometry
sequence needs to be prepared for physics modeling. This is done by the last feature
node of the geometry sequence, called the Finalize (
) feature. The Finalize feature
cannot be deleted from the sequence, and its label in the Model Builder changes
according to the finalization method. There are two methods for finalization of
geometry:
forming a union of all objects, which becomes the modeling domain
- The modeling domain, created by the union operation, consists of domains
separated by interior boundaries. COMSOL ensures continuity in the field
variables across interior boundaries.
forming an assembly object for modeling
- The geometry objects become separate domains for modeling. Identity pairs (or
contact pairs) need to be defined to ensure continuity.
- Imprints can optionally be created on touching boundaries.
Each of these methods influences the type of mesh you can create. Forming a union
imposes the largest constraints on the meshing procedure, while an assembly object
without imprints leaves you with maximum freedom for mesh generation.
In the following exercise, you import a 3D CAD assembly and test each of the above
methods for handling several geometry objects. You will also be able to test how to
prepare the same assembly for contact modeling by setting up contact pairs.
The last exercise of this section will demonstrate how gaps and dimensional
mismatches can influence the final mesh of an imported geometry.
Form a union
1 Click the New (
button.
MODELING WITH ASSEMBLIES
15
Import.
5 Locate the course CD on the hard disk and select the file piston_assembly.x_b,
then click Open.

When the import is completed, examine the geometry. Notice that there are three
geometry objects: the piston, the pin, and the rod. The pin is connected to the rod and
the piston.
pin
rod
piston
7 Click the Form Union button in the Model Builder, then click the Build Selected (
button.
The default finalization method is Form a union. After the operation is completed,
examine the resulting geometry. Note that COMSOL Multiphysics has combined the
three geometry objects into just one with three subdomains corresponding to the
piston, pin, and rod, respectively. These subdomains are delimited by inner surfaces.
8 Right-click the Geometry 1 node and select Measurements.
9 Select all three geometry objects in the Graphics window.
Note that each of the 3 objects contains one domain.
16 |
10 From the Type of geometry list select Finalized geometry.

11 From the Geometric entity level list select Entire geometry.
Note that the finalized geometry is a single object containing three domains. The
total number of faces, edges, and points does not correspond to the number of
entities before union.
Mesh two of the subdomains with different sized meshes.
12 Right-click the Mesh 1 node and select Free Tetrahedral.
13 From the Geometric entity level list select Domain.
14 Add the pin (domain 3) and the rod (domain 2) to the selection list by selecting
them in the Graphics window.

16 From the selection list select domain 2, then click the Remove from Selection button.
17 From the Predefined list select Extremely fine.
19 From the selection list select domain 3, then click the Remove from Selection button.
20 From the Predefined list select Extremely coarse.
21 Click Build All button.
17
Examine the mesh. The Wireframe Rendering (

) or Transparency (
) buttons are
useful when details of a geometry are hidden behind other parts. Notice that the
elements are connected at the surface between the subdomains.
Form an Assembly and Create Imprints

Now change the finalization method to form an assembly and create imprints.
1 Click the Form Union node in the Model Builder.
2 From the Finalization method list select Form an assembly.
3 Select the Create imprints check box.
4 Click the Build Selected button.
The operation creates imprints on the pin surface that is in contact with the piston and
the rod.
IDENTITY PAIRS
To ensure continuity between the domains you can create identity pairs between the
boundaries of the domains.
1 Right-click the Definitions node and select Identity Pair (
). This adds an identity
pair feature to the Model Builder.

To make the selection of surfaces easier first hide the rod and the piston.
2 Click the Select Domain (
) button.
3 In the Graphics window click the rod, then click the Hide Selected (
4 In the Graphics window click the piston, then click the Hide Selected (
5 Click the Select Boundaries (
) button.
) button.
) button to switch back to boundary selection.
6 Add the boundary of the pin in contact with the rod (boundary 105) to the list of
Source Boundaries.
18 |
7 To add a second identity pair, between the pin and the piston, right-click the
Definitions node and select Identity Pair.
8 Add the boundary of the pin in contact with the piston (boundary 104) to the list
of Source Boundaries.
Before selecting the destination boundaries, you can make the rod and the piston
visible again and hide the pin instead.
9 Click first the Reset Hiding (
) button, then click the Select Domain (
10 In the Graphics window click the pin, then click the Hide Selected (
11 In the Settings window of Identity Pair 2 click Activate Selection (
) button.
) button.
) next to the
Destination Boundaries section.

12 Add the boundary of the piston in contact with the pin (boundary 42) to the list of
Destination Boundaries.
13 Click the Identity Pair 1 node.

14 Click Activate Selection (
) next to the Destination Boundaries section.
19
15 Add the boundary of the rod in contact with the pin (boundary 53) to the list of
Destination Boundaries.
16 To show all domains again click the Reset Hiding (
) button.
MESHING
The three domains of the assembly object all have their own boundary surfaces. This
option gives you greater freedom in meshing the different parts than the union
method.
17 Right-click the Mesh 1 node and select Build All.
20 |
As you can see, you can now freely mesh the two objects with different sized meshes.
Note that this mesh is for illustrative purposes and is not appropriate for FEA, due to
the quality of the mesh.
Form an Assembly
Now test the third, and last, method of finalizing the geometry by creating an assembly
object without any modification of the geometry.
1 Click the Form Assembly node in the Model Builder.
2 De-select the Create imprints check box.
The operation forms an assembly object for modeling, but the geometry objects are
not modified.
MESHING
This procedure gives you even greater freedom in meshing the objects. Because there
are no imprints on the boundary surfaces, you can create a swept mesh of the pin.
1 Click the Free Tetrahedral 1 node.
2 In the Selection list select domain 3 and click the Remove from Selection (
) button.
3 Right-click the Size 1 node and select Delete.

4 Right-click Mesh 1 and select Swept.
6 Select the pin (domain 3).
7 Right-click the Swept 1 node and select Size.
8 From the Predefined list select Extremely Fine.
21
9 Click the Build All (
) button.
In this example you can see that after importing CAD assemblies, the method that
gives you the greatest freedom in meshing is to form an assembly object without
creating imprints. This method is also suitable for complex CAD geometries where a
geometry analysis by the other two methods fails.
Gaps in Assemblies
The CAD model in this exercise contains some gaps on the assembly level, which can
prevent the creation of a good mesh. One of the parts also contains holes we would
like to remove.
You can start by importing the CAD file.
1 Click the New (
button.
Import.
22 |
5 Locate the course CD on the hard disk and select the file assembly_gaps.x_b,
then click Open.

plate 2
plate 1
The assembly consists of two parts, plate 1 and plate 2. Plate 2 has four screw holes,
and two feet that fit into corresponding holes on plate 1. Besides the imported parts,
the surrounding air volume is also important for the analysis. Therefore, you can
continue by creating a block around the objects.
2 Right-click the Geometry 1 node and select Block.
3 Use data from the following table to draw a block:
POSITION: BASE: CORNER
SIZE
-0.01
Width
0.06
-0.03
Depth
0.06
-0.06
Height
0.12
) button.
5 Right-click the Mesh 1 node and select Free Tetrahedral.
The finalize operation is performed automatically as you click the Mesh 1 node.
During the finalize operation, COMSOL Multiphysics determines the geometric
23
difference of the objects and creates one object with three subdomains delimited by
single surfaces.
) button.
After the meshing is completed, note that there are about to 130,000 tetrahedral
elements in this seemingly simple geometry.
7 Click the Transparency (
) button to visualize the inner domains.
Dense mesh regions
You can see that there is a very fine mesh in several boundary regions between the
two parts of the assembly. Usually this occurs if small, dimensional discrepancies
exist between parts. They might be present in the design to provide important
clearance for assembling the final product. In this case the clearance between the feet
24 |
of plate 1 and the corresponding slots on plate 2 causes a very thin air gap. However,
such fine details do not affect the overall physics phenomena.
narrow gaps
You cannot change the dimension of imported objects with the CAD Import
Module. Yet, you can, easily remove both the feet and the slots. To this you need to
use the defeaturing tools.
8 Right-click the Geometry 1 node, then select CAD Defeaturing>Delete Faces.
In the Tools window that appears you can select faces of the geometry to be
removed. When a delete operation is completed a feature node will be added to the
geometry sequence.
To make the selection of faces easier first hide the block and plate 1.
9 Click the Select Objects (
) button.
10 In the Graphics window click the block, then click the Hide Selected (
11 In the Graphics window click plate 1, then click the Hide Selected (
) button.
) button.
25
13 Add the faces highlighted in the figure below, boundaries 1, 2, 5, 6, 7, 8 on object
imp1(1), to the list of Faces to delete.
The selected faces delimit the features to be removed.

14 Click the Delete Selected button to delete the feet.
The default heal method, Patch, heals the wound that results from removing the
faces, by shrinking and growing neighboring faces to cover the hole.
When the operation completes a Delete Faces 1 feature node is added to the Model
Builder, right after the current feature in the geometry sequence, which was the
Block 1 feature.
With the feet removed you can continue with removing the corresponding slots on
plate 1.
Hide some objects again to make face selection easier, this time the block and plate
2.
15 Click the Select Objects (
) button.
16 In the Graphics window click the block, then click the Hide Selected (
17 In the Graphics window click plate 2, then click the Hide Selected (
26 |
) button.
) button.
19 Add the faces highlighted in the figure below, boundaries 5-8, 9-12 on object
imp1(2), to the list of Faces to delete.
20 Select the Heal as through hole check box.
27
21 Click the Delete Selected button to delete the holes.
Before continuing with meshing, you can also delete the four mounting holes,
shown in the figure below, since they are assumed to not influence the analysis.
You can easily select the faces delimiting the holes by using the Select Box tool.
22 Click the Select Box (
) button, then in the Graphics window draw a rectangle

around the faces delimiting one of the holes. Confirm the selection by
right-clicking.
23 Repeat the previous step for the other three holes.

24 The Faces to delete list should now contain faces 7-14 and 17-24 of object dfa1.
25 Click the Delete Selected button to delete the holes.
The last step is to create a new mesh of the geometry.
28 |
26 Right-click the Mesh 1 node in the Model Builder, then select Build All.
Dense mesh regions
The mesh now consists of about 31,000 tetrahedral elements. As you can see in the
figure, there are still two regions of high mesh density left. The reason for this is a slight
difference in width between the two objects.
Objects are not of the same width
If this dimensional mismatch is unintentional and assumed to not be important for the
analysis, it can easily fixed in a CAD software. With the CAD Import Module, you can
export defeatured parts to a Parasolid file, which you can open and edit with a CAD
program.
29
27 Right-click the Geometry 1 node and select Export to File.

28 The Input list shows all three objects selected (dfa2, blk1, and dfa3). To accept and
export the objects click Export.

29 From the Save as type list select either Parasolid binary file or Parasolid text file.
30 Select a directory, enter a file name, then click Save.
The resulting Parasolid file can be imported into most CAD software.
You can also use the CAD Import Module to fix the dimensional discrepancy, by for
example using the repair functionality of the Finalize feature.
31 Click the Form Union feature node.
32 In the Relative repair tolerance edit field enter 1e-3.
33 Right-click the Mesh 1 node in the Model Builder, then select Build All.
As you can see the resulting mesh contains only about 5000 elements.
30 |
The Meshing Sequence

COMSOL Multiphysics provides an interactive meshing environment where, with a
few mouse clicks, you can easily mesh individual faces or domains. Each meshing
operation is added to the meshing sequence. The final mesh is the result of building
all the operations in the meshing sequence.
This example demonstrates how to use the meshing sequence. On a realistic geometry
you create a mesh consisting of different element types. You learn how to mesh certain
parts of a geometry, modify this mesh, change the parameters to your liking, and
rebuild the mesh.
Meshing with Default Settings

To start, import a 3D CAD geometry and mesh it with the default settings using the
free mesher to generate a mesh with tetrahedral elements.
1 Click the New (
button.
Import.
5 Locate the course CD on the hard disk and select the file solder_joints.x_b,
then click Open.

Note that for many of the imported objects a warning is displayed that the default
import tolerance, 10-5 m, might be too large for the geometry, which includes parts
on the order of a tenth of a millimeter in size. Features smaller than the import
tolerance are automatically removed during import. Therefore, if you know that
your geometry contains small features you can change the import tolerance in the
32 |
C H A P T E R 3 : M E S H I N G TE C H N I Q U E S
Settings window of the Import feature. There you also have the option of turning off
the repair feature during import. In this particular case, this is not necessary.
The geometry represents a small part of a circuit board with an electronic

component mounted by means of several solder ball joints.
Electronic component
Solder joints
Circuit board

7 Click the Mesh 1 node.
As you click the Mesh 1 node the Finalize feature is automatically built, which means
that the various objects are combined into the one object with inner boundaries and
subdomains.
8 Click the small plus sign (or triangle) in front of the Mesh 1 node to expand the
meshing sequence.
Initially the meshing sequence contains only the Size feature node. This feature is
called a global attribute feature, since it influences all subsequent operation
features in the meshing sequence. This first Size feature node cannot be deleted
from the meshing sequence. You can also add attribute features under an operation
feature node, in which case it is called a local attribute feature.
The Free Tetrahedral 1 feature node is added after the Size feature node. COMSOL
Multiphysics always inserts new nodes in the meshing sequence after the current
feature node. To indicate the current feature node, it appears with a quadratic frame
around its icon. As soon as it is inserted, the Free Tetrahedral 1 node becomes the
current feature.
THE MESHING SEQUENCE
33
According to information displayed in the Messages window this mesh consists of

approximately 35,000 elements. While the geometry is resolved quite well by this
mesh, you may want to reduce the number of elements to reduce the memory required
for solving the problem.
Adding Local Attribute Features

Assume that you are investigating the solder joints and would therefore like to keep
the detailed mesh in the spherical subdomains, but create a mesh with fewer elements
in the remaining objects.
1 Right-click the Free Tetrahedral 1 feature node and select Size.
34 |
3 Use the Graphics window to select the domains 1-3, which represent the circuit
board and the electronic component.
4 From the Predefined list select Coarser.

5 Click the Build All button to build the new mesh.
According to the Messages window the mesh now consists of approximately 24,000
elements.
The Swept Mesher

For even fewer elements in the circuit board and electronic component, you can create
a swept mesh. This technique sweeps a boundary mesh through the domains to create
a structured mesh in the sweep direction.
The swept mesher operates on a 3D subdomain by first meshing a source face, and then
sweeping the resulting face mesh through the subdomain to an opposite target face.
For straight and circular sweep paths, you can use several connected faces as source
35
faces. All faces that encompass a domain are classified as either source faces, a target
face, or boundary faces. The boundary faces are the faces connecting the source and
target faces.
MESHING A SET OF BOUNDARIES
First you can modify the free tetrahedral mesh operation to operate only on the solder
joints.
1 Click the Free Tetrahedral 1 feature node.
3 From the Selection list select domains 1,2, and 3, then click the Remove from
Selection (
) button.
You can also remove the Size 1 feature node.

4 Right-click the Size 1 node, then select Delete.
You can now create a coarse mesh on the source boundaries for the sweep mesh
operation. Build the mesh before continuing.
) button.
6 Right-click the Mesh 1 node then select More Operations then Free Triangular.
7 From the Graphics window select the two faces (boundaries 2 and 14) highlighted
in the figure below.
To make selection easier click the Wireframe Rendering (

to turn of wireframe rendering.
8 Right-click the Free Triangular 1 node and select Size.
9 From the Predefined list select Coarser.
36 |
) button. Click it again
) button to build the mesh.
The mesh is built, but a Problems node appears under the Free Triangular 1 feature
node.
11 Expand the Problems 1 node, then click the Warning 1 node to read the message.
COMSOL Multiphysics generates the warning because the Coarser predefined mesh
size you have applied to the two surfaces has been overridden by the relatively finer
Normal mesh settings specified in the first Size feature node of the meshing sequence.
The reason is to avoid bad quality meshes, that may result if the boundary of a
domain is meshed with a coarser mesh size then the volume.
To avoid this situation a good practice is to set the first global Size setting to the
coarsest mesh that you plan to have in the geometry, then specify local size feature
nodes for the mesh operations that need finer mesh.
12 Click the Size node, then select Coarser from the Predefined list.
13 Right-click the Free Tetrahedral 1 node, and select Size.
14 From the Predefined list select Normal.
Since the global setting is Coarser, the size attribute to the Free Triangular 1 node is
no longer necessary.
15 Right-click the Size 1 node under the Free Triangular 1 node, then select Delete.
37
16 Right-click the Mesh 1 node then select Build All (
) to build the mesh.
Notice that the triangular mesh is coarser this time.

CREATING A SWEPT MESH
Now that the source faces are meshed, you can sweep this mesh through the
subdomains.
1 Right-click the Mesh 1 node and select Swept.
The Geometric entity level list is set to Remaining by default for new mesh operations.
In this case the remaining domains correspond to the domains we would like to
sweep mesh.
38 |
) button to build the mesh.
The mesh now consists of approximately 15,000 elements. The swept mesher used
the Coarser predefined mesh size to determine the number of elements along the
sweep direction. By specifying a distribution for the swept mesher you have the
possibility to manually control the number and distribution of elements along the
sweep direction.
3 Right-click the Swept 1 feature node and select Distribution.
4 In the Number of elements edit field enter 2.
5 Click the Build All button to build the mesh.
39
This latest mesh consists of approximately 18,000 elements, keeping the higher
resolution for the subdomains which are important for the analysis, while providing a
less dense structured mesh for the remaining subdomains.
40 |
Free Mesh Parameters

The free meshing algorithm using tetrahedral elements is the most general meshing
technique, and does not pose any constraints on the structure of the geometry. Hence
it can be used to mesh any arbitrary object.
This exercise shows how to use the various parameters of the free mesh algorithm.
There are a number of predefined parameter sets, and each parameter can be tuned
individually.
Predefined Parameter Settings

There are nine predefined parameter sets ranging from extremely fine to extremely
coarse. These settings result in a good mesh for most geometries and simulation
problems.
Follow the next steps to import a 3D CAD model and to test two of these settings.
1 Click the New (
button.
Import.
5 Locate the course CD on the hard disk and select the file piston_for_fem.x_b,
then click Open.

) button.
The Messages window indicates that there are roughly 34,000 tetrahedral elements
in this geometry.
Use the mouse to rotate the object and look at the geometry more closely. You can
see that it contains some small details such as fillets and chamfers. Assume that the
existing mesh does not resolve these details sufficiently for your simulation needs
and so a finer parameter setting is required.
8 Right-click the Free Tetrahedral 1 node, then select Size.
FREE MESH PARAMETERS
41
9 From the Predefined list select Fine.
This changes the predefined parameter set from Normal to Fine.

11 In the Messages window, note that this mesh consists of approximately 91,000
elements.
If you examine the mesh you can see that, even with the Fine settings, some narrow
areas are not resolved sufficiently. You could decrease the mesh size once more but, to
achieve a fine mesh only on certain types of boundaries, you also have the possibility
to tune individual mesh parameters. This is demonstrated in the following sections.
Mesh Curvature
Assume that you want a better resolution on regions with a high curvature, such as
fillets. To find the right setting you can test it first on just one of the fillets in the
geometry.
1 Click the Size 1 node.
2 From the Predefined list select Normal.
3 From the Geometric entity level list select Boundary.
4 From the Graphics window select the boundary highlighted (boundary 12) in the
figure below.
5 In the Element Size section click the Custom radio button.

6 Select the Minimum element size check box.
7 Change the value of Minimum element size to 5e-4.
With the Minimum element size parameter, you can prevent the generation of a fine
mesh around small curved regions of the geometry. When the radius of curvature is
smaller than this value, the mesh generator considers the radius of curvature as being
this value.
42 |
) button.
9 Select the Resolution of curvature check box.

10 In the Resolution of curvature enter 0.5.
The Resolution of curvature parameter determines the size of boundary elements

compared to the curvature of the geometric boundary. The curvature radius
multiplied by the resolution of curvature gives the maximum allowed element size
along the boundary. A lower value gives a finer mesh along curved boundaries.
) button.
The selected fillet is resolved even finer with this mesh.
43
Resolution of Narrow Regions

Assume that you also want better resolution on narrow regions with no curvature such
as chamfers.
12 From the Graphics window add the boundary highlighted (boundary 44) in the
figure below to the Selection list of the Size 1 feature node.
13 Select the Resolution of narrow regions check box.

14 In the Resolution of narrow regions edit field enter 2.1.
The mesh Resolution of narrow regions mesh parameter approximately controls the
number of layers of elements that are created in narrow regions (approximately). If
the value of this parameter is less than one, the mesh generator might create
elements that are anisotropic in size in narrow regions.
44 |
) button.
Element Growth Rate

Assume that you are happy with the new parameter settings for curved and narrow
regions. Now use these to mesh the entire geometry.
1 Click the Zoom Extents (
) button to see the entire geometry.
2 From the geometric entity level list in the Size 1 feature node select Entire geometry.
) button.
The fine details of the geometry are resolved satisfactorily. However, the mesh
consists of approximately 308,000 elements, which is too many. Assume that the
inner part of the volume does not need such a high resolution. You can change it by
specifying the rate of growth from the small elements on the surface towards the
inner part of the geometry.
4 Select the Maximum element growth rate check box.
5 In the Maximum element growth rate edit field enter 2.3.
The Maximum element growth rate determines the maximum rate at which the
element size can grow from a region with small elements to a region with larger
elements.
) button.
7 In the Messages window you can see that this mesh consists of approximately
112,000 elements.
45
Mesh Quality
Changing various mesh parameters can influence the quality of the mesh. In general,
the default settings result in a mesh with rather good quality. However, this quality can
decrease drastically when changing individual parameters.
First take a look at the mesh statistics for the current mesh on the piston.
1 Right-click the Mesh 1 node and select Statistics.
The Statistics window appears, which contains details about the mesh, including the
number and type of elements, and a histogram of element quality.
The element quality is a number between 0 and 1, where 1 describes a perfectly

isotropic element, and 0 describes a flat element. For 3D meshes in general a minimum
quality of about 0.1 means a satisfactory mesh. However, this depends also on the type
of geometry and physics application. Note that the quality number is calculated based
on the linear elements.
46 |
In the following, you learn how to visualize the elements with the worst mesh quality.
Assume that you want to achieve the opposite goal as described in the previous
exampleyou want much to resolve the narrow regions with less mesh.
1 Click the Size 1 feature node.
2 Select the Predefined radio button, then select the Custom radio button.
This way the custom mesh parameters are reset to values corresponding to the
Normal predefined mesh size.
3 Select the Resolution of narrow regions check box.
4 Change the value of Resolution of narrow regions to 0.1.
) button.
6 Right-click the Mesh 1 node and select Statistics.
Now your mesh consists of only 28,000 elements. However, in the area of the
narrow regions the quality becomes worse. The minimum element quality has
decreased, and according to the histogram there is a relatively small number of
elements that have the lowest quality.
Visualize this situation with the following steps.
7 Right-click the Mesh 1 node and select Plot.
The Mesh 1 plot is added to the 3D Plot Group 1 under the Results section of the Model
Builder. The mesh plot that appears in the Graphics window contains the volume
elements colored according to quality.
8 Expand the Element Filter section, and select Enable filter.
9 From the Criterion list select Worst quality.
10 In the Fraction edit field enter 0.005.
47
11 Click the Plot button.
You can now see those 0.5% of elements with the worst quality, where almost all of
them are located around the narrow faces. As anticipated, decreasing the mesh
resolution in narrow regions detrimentally influences the quality of the mesh.
48 |
Mapped and Swept Meshing

COMSOL Multiphysics provides the ability to create structured meshes, specifically,
mapped meshes on surfaces and swept meshes in subdomains. Each of these techniques
can be used interactively and has its own set of parameters.
This exercise demonstrates how to use the mapped and swept meshing techniques. It
shows how to create these meshes interactively so as to achieve a desired distribution
of elements.
Swept Mesh with Default Settings

First create a swept mesh with the default settings.
1 Click the New (
button.
Import.
5 Locate the course CD on the hard disk and select the file thin_layer.mphbin,
then click Open.

6 Right-click the Mesh 1 node and select Swept.
) button.
MAPPED AND SWEPT MESHING
49
This procedure creates a very coarse mesh consisting of hexahedral elements. Assume
that you want dense layers at the bottom because you expect high gradients in the
simulation. In addition, you want a better resolution in the curved part of the
geometry. The following section shows how to achieve these goals.
Mapped Mesh Parameters

In order to force a certain element distribution in a cross-section of the geometry, you
must create a mesh on a source face. This mesh is later swept through the subdomains
of the geometry.
1 Click the Size node in the Model Builder.
2 Click the Build Selected (
) button.
By building the Size feature, it becomes the current feature node in the mesh
sequence after which the mapped mesh operation can be inserted.
3 Right-click the Mesh 1 node and select More Operations>Mapped.
4 Add boundary 1, which is highlighted in the figure below, to the Selection list.
5 Right-click the Mapped 1 node and select Distribution.

6 Add Edges 1 and 6 to the Selection list of the Distribution 1 feature node.
7 From the Distribution properties list select Predefined distribution type.

8 Type 8 in the Number of elements edit field.
50 |
9 In the Element ratio edit field type 5.
This parameter controls the ratio in size between the last element and the first
element along the edge.
10 In the Distribution method list select Geometric sequence.
11 Right-click the Mapped 1 node and select Distribution.
12 Add Edges 2 and 4 to the Selection list of the Distribution 2 feature node.
13 Type 2 in the Number of elements field.

) button.
15 Use the zoom and rotation features to magnify the modified mesh.
You have achieved the desired element distribution in the cross section. The mesh
looks to be adequate on the straight parts. However, the curved parts are not
resolved satisfactorily.
Swept Mesh Parameters

Change the swept mesh parameters in order to achieve the desired element
distribution along the subdomains.
MAPPED AND SWEPT MESHING
51
1 Right-click on the Swept 1 node and select Distribution.

2 In the Selection list select domains 2, 3 and 4, then click the Remove from Selection
) button.

6 From the Distribution method list select Geometric sequence.
7 Select the Reverse direction check box.
This defines the distribution in the opposite direction for the selected subdomain.
2 In the Selection list select domains 1, 3 and 4, then click the Remove from Selection
) button.

6 From the Distribution method list select Geometric sequence.
The last step is to define the distribution for the curved parts.
8 In the Selection list select domains 1 and 2, then click the Remove from Selection
) button.

52 |
) button.
Solved with COMSOL Multiphysics 4.1
Heat Sink
Introduction
This model is intended as a first introduction to simulations of fluid flow and conjugate
heat transfer. It is also a very useful model for exploring the modeling procedure in
COMSOL Multiphysics. More specifically, it demonstrates the use of three important
concepts:
The concept of sequencing. Each operation that you can use to set up a simulation
is displayed as a node in the Model Tree. As you proceed with your model, the
Model Tree forms a sequence of operations. You can make changes to any of the
nodes in the tree and then re-run the sequence of operations to update your model
according to the changes in the node settings.
The selection of materials in the Materials node. In this node, you can specify and
investigate materials and material properties for all the domains and for all physics
interfaces in your model.
The use of selections. In the Selections node, you can define selections of domains,
boundaries, edges, and points, which you can later use in other steps in the
modeling procedure.
Model Definition
The modeled system consists of an aluminum heat sink for cooling of components in
electronic circuits mounted inside a channel of rectangular cross section; see Figure 1.
Such a set-up is used in order to measure the cooling capacity of heat sinks. Air enters
the channel at the inlet and exits the channel at the outlet. The base surface of the heat
sink is kept at a constant temperature through an external heat source. All other
external faces are thermally insulated.
HEAT SINK
inlet
outlet
base surface
Figure 1: The model set-up including channel and heat sink.

The cooling capacity of the heat sink can be determined by measuring the power
required to keep the base surface of the heat sink at a constant temperature.
The model solves a thermal balance for the heat sink and the air flowing in the
rectangular channel. Thermal energy is transported through conduction in the
aluminum heat sink and through conduction and convection in the cooling air. The
temperature field is continuous across the internal surfaces between the heat sink and
the air in the channel. The temperature is set at the inlet of the channel and at the base
of the heat sink. Alternatively, you can also simulate the presence of a layer of adhesive
material between the heat sink and the heating device used to keep a constant
temperature. In such case, you have to define a heat transfer coefficient for the adhesive
layer and then set the temperature at the heater side of the layer. The transport of
thermal energy at the outlet is dominated by convection.
The flow field is obtained by solving one momentum balance for each space coordinate
(x, y, and z) and a mass balance. The inlet velocity is defined by a parabolic velocity
profile for fully developed laminar flow. At the outlet, a constant pressure is combined
the assumption that there are no viscous stresses in the direction perpendicular to the
outlet. At all solid surfaces, the velocity is set to zero in all three spatial directions.
2 |
HEAT SINK
The thermal conductivity of aluminum, the thermal conductivity of air, the heat
capacity of air, and the air density are all temperature-dependent material properties.
You can find all of the settings mentioned above in the physics interface for Conjugate
Heat Transfer in COMSOL Multiphysics. You also find the material properties,
including their temperature dependence, in the Material Browser
Results
In Figure 2, the hot wake behind the heat sink visible in the plot is a sign of the
convective cooling effects.
Figure 2: The surface plot visualizes the temperature field on the channel walls and the
heat sink surface, while the arrow plot shows the flow velocity field around the heat sink.
Model Library path: Heat_Transfer_Module/Tutorial_Models/heat_sink
HEAT SINK
Modeling Instructions
MODEL WIZARD
1 Go to the Model Wizard window.

2 Click Next.
3 In the Add Physics tree, select Heat Transfer>Conjugate Heat Transfer>Laminar Flow
(nitf).
4 Click Next.
5 In the Studies tree, select Preset Studies>Stationary.
6 Click Finish.
GLOBAL DEFINITIONS
Parameters
1 In the Model Builder window, right-click Global Definitions and choose Parameters.
Define some parameters that you can use when specifying the channel dimensions.
2 Go to the Settings window for Parameters.
3 Locate the Parameters section. In the Parameters table, enter the following settings:
NAME
EXPRESSION
DESCRIPTION
L_channel
7[cm]
Channel length
W_channel
3[cm]
Channel width
H_channel
1.5[cm]
Channel height
U0
5[cm/s]
Mean inlet velocity
GEOMETRY 1
Import 1
1 In the Model Builder window, right-click Model 1>Geometry 1 and choose Import.
2 Go to the Settings window for Import.
3 Locate the Import section. Click the Browse button.
4 Browse to the models Model Library folder and double-click the file
heat_sink_n19.mphbin.

6 In the Model Builder window, right-click Geometry 1 and choose Work Plane.
4 |
HEAT SINK
Rectangle 1
1 In the Model Builder window, right-click Geometry and choose Rectangle.
2 Go to the Settings window for Rectangle.
3 Locate the Size section. In the Width edit field, type L_channel.
4 In the Height edit field, type W_channel.
5 Locate the Position section. In the x edit field, type -4.5e-2.
6 In the y edit field, type -W_channel/2.
Extrude 1
1 In the Model Builder window, right-click Work Plane 1 and choose Extrude.
2 Click the Go to Default 3D View button on the Graphics toolbar.
3 Go to the Settings window for Extrude.
4 Locate the Distances from Work Plane section. In the table, enter the following
settings:
DISTANCES (M)
H_channel

MATERIALS
1 In the Model Builder window, right-click Model 1>Materials and choose Open Material
Browser.
2 Go to the Material Browser window.
3 Locate the Materials section. In the Materials tree, select Built-In>Air.
4 Right-click and choose Add Material to Model from the menu.
Air
By default, the first material you add applies to all domains. Typically, you can leave
this setting and add other materials that override the default material where applicable.
In this example, specify aluminum for Domain 2:
1 In the Model Builder window, right-click Materials and choose Open Material Browser.
3 Locate the Materials section. In the Materials tree, select Built-In>Aluminum
3003-H18.
HEAT SINK
Aluminum 3003-H18
1 In the Model Builder window, click Aluminum 3003-H18.
2 Select Domain 2 only.
C O N J U G A T E H E A T TR A N S F E R
Fluid 1
1 In the Model Builder window, right-click Model 1>Conjugate Heat Transfer and choose
Fluid.
Inlet 1
1 In the Model Builder window, right-click Conjugate Heat Transfer and choose the
boundary condition Laminar Flow>Inlet.

2 Select Boundary 115 only.
3 Go to the Settings window for Inlet.
4 Locate the Boundary Condition section. From the Boundary condition list, select
Laminar inflow.
5 In the Uav edit field, type U0.
Outlet 1
boundary condition Laminar Flow>Outlet.

Temperature 1
boundary condition Heat Transfer>Temperature.

The default temperature, corresponding to 20 degrees Celsius or 68 degrees

Fahrenheit, applies at the inlet.
Temperature 2
boundary condition Heat Transfer>Temperature.

3 Go to the Settings window for Temperature.
6 |
HEAT SINK
4 Locate the Temperature section. In the T0 edit field, type 393.15[K].
This value corresponds to 120 degrees Celsius or 248 degrees Fahrenheit.
Outflow 1
boundary condition Heat Transfer>Outflow.

MESH 1
Free Tetrahedral 1
In the Model Builder window, right-click Model 1>Mesh 1 and choose Free Tetrahedral.
Size 1
1 In the Model Builder window, right-click Free Tetrahedral 1 and choose Size.
2 Go to the Settings window for Size.
3 Locate the Geometric Scope section. From the Geometric entity level list, select
Domain.
5 Locate the Element Size section. From the Predefined list, select Finer.
DEFINITIONS
In the Model Builder window, right-click Study 1 and choose Compute.
Selection 1
1 In the Model Builder window, right-click Model 1>Definitions and choose Selection.
2 Go to the Settings window for Selection.
Boundary.
4 Right-click Selection 1 and choose Rename.
5 Go to the Rename Selection dialog box and type walls in the New name edit field.
6 Click OK.
7 Right-click Selection 1 and choose Select Box.
8 Select Boundaries 3 and 5114 only.
HEAT SINK
RESULTS
Data Sets
1 In the Model Builder window, right-click Results>Data Sets>Solution 1 and choose
Add Selection.
Boundary.
4 From the Selection list, select walls.
3D Plot Group 1
1 In the Model Builder window, click Surface 1.
2 Go to the Settings window for Surface.
3 Locate the Coloring and Style section. From the Color table list, select Thermal.
4 In the Model Builder window, right-click 3D Plot Group 1 and choose Arrow Volume.
5 Go to the Settings window for Arrow Volume.
6 In the upper-right corner of the Expression section, click Replace Expression.
7 From the menu, choose Conjugate Heat Transfer (Laminar Flow)>Velocity field (u, v,
w).
8 Locate the Arrow Positioning section. Find the x grid points subsection. In the Points
edit field, type 40.

9 Find the y grid points subsection. In the Points edit field, type 20.
10 Find the z grid points subsection. From the Entry method list, select Coordinates.
11 In the Coordinates edit field, type 5e-3.
12 Right-click Arrow Volume 1 and choose Color Expression.
13 Go to the Settings window for Color Expression.
15 From the menu, choose Conjugate Heat Transfer (Laminar Flow)>Velocity magnitude
(nitf.U).
8 |
HEAT SINK
Solved with COMSOL Multiphysics 4.2a
Various Analyses of an Elbow Bracket

Introduction
The component depicted in Figure 1 is part of a support mechanism and is subjected
to both mechanical loads and thermal loads. This tutorial model takes you through the
steps to carry out a detailed analysis of the part using the Structural Mechanics
Module.
Figure 1: Geometry of the elbow bracket.

In the various parts of the model you are introduced to using the available basic
analysis types, together with numerous postprocessing possibilities. These analysis
types are:
Static analysis
Eigenfrequency analysis
Damped eigenfrequency analysis
Transient analysis
Modal based transient analysis
Frequency response analysis
Modal based frequency response analysis
Parametric analysis
Linear buckling analysis
This tutorial model consists of a single model, with nine studies, corresponding to
these analysis types, which are described in the section Available Study Types in the
Structural Mechanics Module Users Guide. The chapter Structural Mechanics
Modeling in the same manual provides further assistance.
2011 COMSOL
1 |
VA R I O U S A N A LY S E S O F A N E L B OW B R A C KE T
Note: If you have already built the short model version described in Static and
Eigenfrequency Analyses of an Elbow Bracket, you can proceed directly to the section
Time-Dependent Analysis.
Model Definition
The geometry for this part, see Figure 1, has been created with a CAD software, and
it is available for you to import into COMSOL Multiphysics.
Material
Structural steel, as taken from the material library, with Youngs modulus of 200 GPa,
Poissons ratio of 0.33, and thermal expansion coefficient 12.310-6 K1.
Damping
The Structural Mechanics Module supports Rayleigh damping and loss factor
damping. You can also use no damping, which is the default option.
In some of the analyses here you will use Rayleigh damping, where you specify
damping parameters that are proportional to the mass (dM) and stiffness (dK) in the
following way:
C = dM M + dK K
where C is the damping matrix, M is the mass matrix, and K is the stiffness matrix. The
damping is specified locally in each domain; this means that you can specify different
damping parameters in different parts of the model.
To find the values for the Rayleigh damping, you can use the relations between the
critical damping ratio and the Rayleigh damping parameters. It is often easier to
interpret the critical damping ratios, which are given by
dM
--------- i + dK i
i = -------------------------------------------2
where i is the critical damping ratio at a specific angular frequency i. Knowing two
pairs of corresponding i and i results in a system of equations
2 |
VA R I O U S A N A LY S E S O F A N E L B O W B R A C KE T
2011 COMSOL
1 -
-----------------------1
( 2 1 ) 2 dM
dK
1
------------------- ------2
( 2 2 ) 2
1
2
with the damping parameters as the unknown variables.

Assume that the structure has a constant damping ratio of 0.1. Select two frequencies
near the excitation frequency, 400 Hz and 600 Hz, to calculate the damping
parameters. You can do this in MATLAB with the following commands:
b = [0.1;0.1];
A = [1/(2*400*2*pi) 2*pi*400/2; 1/(2*600*2*pi) 2*pi*600/2];
% A*damp = b
damp = A\b;
alphadM = damp(1)
betadK = damp(2)
The result is dM = 300 and dK = 3.2105.

For more information see the section Modeling Damping and Losses in the Structural
Mechanics Module Users Guide.
If modal based dynamic response studies are performed it is usually easier to give the
critical damping ratios directly. This will also give more detailed control over the
damping properties over a large frequency range.
2011 COMSOL
3 |
Loads and Constraints

The displacement are fixed in all directions on the face shown in Figure 2. The load is
described under each study, but in all cases it is distributed over the face shown in this
figure.
Fixed face
Loaded face
Figure 2: Constraint and loading of the bracket.

The Model Library note immediately below appears in the discussion of every model.
The path indicates the location of the model file in the Model Library root directory.
The most convenient way to open it is from the Model Library window in the
COMSOL Desktop, which you can open from the File menu.
Model Library path: Structural_Mechanics_Module/Tutorial_Models/

elbow_bracket
MODEL WIZARD

2 Click Next.
3 In the Add physics tree, select Structural Mechanics>Solid Mechanics (solid).
4 Click Next.
5 Find the Studies subsection. In the tree, select Preset Studies>Stationary.
4 |
2011 COMSOL
6 Click Finish.
GEOMETRY 1
Import 1
3 Locate the Import section. From the Geometry import list, choose COMSOL
Multiphysics file.
4 Click the Browse button.
elbow_bracket.mphbin.
6 Click the Import button.

7 Click the Wireframe Rendering button on the Graphics toolbar.
The view in the Graphics window should look like that in the image below.
8 Click the Wireframe Rendering button on the Graphics toolbar to return to the
default surface rendering.
2011 COMSOL
5 |
MESH 1
Size
1 In the Model Builder window, right-click Model 1>Mesh 1 and choose Free
Tetrahedral.
The default free mesher has nine predefined combinations of mesh parameter
settings. They range from Extremely fine to Extremely coarse, with Normal as the
default setting. Unless any other mesh parameters are set, this is the setting that is
used if you use Build All or Build Selected to generate the mesh.
3 Locate the Element Size section. From the Predefined list, choose Coarse.
MATERIALS
Next, specify the material properties. You can do this either by explicitly typing them
in or by selecting a library material in the Material Browser. For this model, use a
library material.
Browser.
3 Locate the Materials section. In the Materials tree, select Built-In>Structural steel.
Static Analysis
A static analysis has no explicit or implicit time dependencies. This situation
corresponds to the steady state with constant (in time) boundary conditions and
material properties.
The purpose of such analysis can be to find the maximum stress level and compare it
with the materials yield strength, as well as to check that the deformation of the
component is within the limits of the design criteria.
Results and Discussion

The analysis shows that the von Mises effective stress has a maximum value of
187 MPa, which, compared with the materials yield strength of 350 MPa, results in a
utilization factor of 53%.
6 |
2011 COMSOL
The analysis also gives the maximum static displacements as 1.14 mm

Three different representations of the stress state are shown in Figure 3 through
Figure 5.
Figure 3: Effective stresses on the boundary of the volume.
2011 COMSOL
7 |
Figure 4: Isosurface plot of the effective stress.
Figure 5: Arrow plot of the principal stresses.
8 |
2011 COMSOL
SOLID MECHANICS
Fixed Constraint 1
1 In the Model Builder window, right-click Model 1>Solid Mechanics and choose Fixed
Constraint.
Boundary Load 1
1 In the Model Builder window, right-click Solid Mechanics and choose Boundary Load.
3 Go to the Settings window for Boundary Load.
4 Locate the Force section. Specify the FA vector as
3[MPa]
3[MPa]
STUDY 1
In this model, where there are many different studies, it is a good idea to assign manual
names to some features.
1 In the Model Builder window, right-click Study 1 and choose Rename.
2 Go to the Rename Study dialog box and type Study 1 (Static) in the New name
edit field.
3 Click OK.
The default settings in the generated solver sequence are OK for this model.
STUDY 1 (STATIC)
1 In the Model Builder window, right-click Study 1 (Static) and choose Compute.
Before moving on to analyzing the solution, rename the solver sequence.

2 In the Model Builder window, expand the Study 1 (Static) node.
Solver 1
1 In the Model Builder window, expand the Study 1 (Static)>Solver Configurations node.
2 Right-click Study 1 (Static)>Solver Configurations>Solver 1 and choose Rename.
2011 COMSOL
9 |
3 Go to the Rename Solver dialog box and type Static Sequence in the New name
edit field.
4 Click OK.
RESULTS
Similarly, rename the solution data set.
Data Sets
Rename.
2 Go to the Rename Solution dialog box and type Static Solution in the New name
edit field.
3 Click OK.
In the Results branch, you can create various plot types, evaluate expressions, or
animate the results. The result features can visualize any expression containing, for
example, the solution variables, their derivatives, and the space coordinates. Many
frequently used expressions are predefined as postprocessing variables, and they are
directly available in the Expression section menus for the various plot types.
When the solver finishes, two default plots appear. One shows total displacement as a
surface plot and the other shows the same quantity as a slice plot. Re-use these plot
groups, beginning by creating a volume plot of the von Mises stress with markers for
maximum and minimum values together with the deformed shape of the component.
For this purpose, modify the first plot group.
Stress (solid)
1 In the Model Builder window, right-click Results>Stress (solid) and choose Rename.
2 Go to the Rename 3D Plot Group dialog box and type Static Stress Contour in
the New name edit field.

3 Click OK.
4 Click the Zoom Extents button on the Graphics toolbar.
To evaluate the maximum displacement, use a Maximum coupling operator.

DEFINITIONS
Maximum 1
1 In the Model Builder window, right-click Model 1>Definitions and choose Maximum.
2 Go to the Settings window for Maximum.
10 |
VAR IOUS ANALYSES O F AN ELBOW BR ACKET
2011 COMSOL
3 Locate the Source Selection section. From the Geometric entity level list, choose
Boundary.
4 From the Selection list, choose All boundaries.
Variables 1a
1 In the Model Builder window, right-click Definitions and choose Variables.
2 Go to the Settings window for Variables.
3 Locate the Variables section. In the Variables table, enter the following settings:
NAME
EXPRESSION
DESCRIPTION
U_max
maxop1(solid.disp)
Maximum deflection
STUDY 1 (STATIC)
Static Sequence
1 In the Model Builder window, right-click Study 1 (Static)>Solver Configurations>Static
Sequence and choose Solution>Update.
This step is necessary in order to access variables that were created after the solution
was performed.
RESULTS
Derived Values
1 In the Model Builder window, right-click Results>Derived Values and choose Global
Evaluation.
2 Go to the Settings window for Global Evaluation.
4 From the menu, choose Definitions>Maximum deflection (U_max).
5 Locate the Expression section. From the Unit list, choose mm.
6 Click the Evaluate button.
The result, approximately 1.1 mm appears in the Results window.

Next, replace the second default plot group by an isosurface plot. The resulting plot
should resemble that in Figure 4.
3D Plot Group 2
1 In the Model Builder window, right-click Results and choose 3D Plot Group.
2 Right-click Results>3D Plot Group 2 and choose Isosurface.
2011 COMSOL
11 |
3 Go to the Settings window for Isosurface.

5 From the menu, choose Solid Mechanics>Stress>von Mises stress (solid.mises).
6 Right-click Isosurface 1 and choose Deformation.
7 Right-click 3D Plot Group 2 and choose Rename.
8 Go to the Rename 3D Plot Group dialog box and type Static Stress Isosurface
in the New name edit field.

9 Click OK.
With the following steps you can reproduce the principal stress arrow plot shown in
Figure 5:
3D Plot Group 3
1 Right-click Results and choose 3D Plot Group.
2 In the Model Builder window, right-click Results>3D Plot Group 3 and choose More
Plots>Principal Stress Volume.
3 Go to the Settings window for Principal Stress Volume.
4 Locate the Positioning section. Find the X grid points subsection. In the Points edit
field, type 10.

5 Find the Y grid points subsection. In the Points edit field, type 15.
6 Find the Z grid points subsection. In the Points edit field, type 10.
8 In the Model Builder window, right-click 3D Plot Group 3 and choose Rename.
9 Go to the Rename 3D Plot Group dialog box and type Static Principal Stress
Arrow Plot in the New name edit field.
10 Click OK.
Eigenfrequency Analysis
An eigenfrequency analysis finds the eigenfrequencies and modes of deformation of a
component. The eigenfrequencies f in the structural mechanics field are related to the
eigenvalues returned by the solvers through
12 |
2011 COMSOL
f = --------2i
(1)
In COMSOL Multiphysics you can choose between working with eigenfrequencies

and working with eigenvalues according to your preferences. Eigenfrequencies are the
default option for all physics interfaces in the Structural Mechanics Module.
If no damping is included in the material, the undamped natural frequencies are
computed.
The purpose of the following eigenfrequency analysis is to find the six lowest
eigenfrequencies and corresponding mode shapes.

The first six eigenfrequencies are:
EIGENFREQUENCY
FREQUENCY
f1
417 Hz
f2
571 Hz
f3
1930 Hz
f4
2450 Hz
f5
3110 Hz
f6
3930 Hz
The mode shapes corresponding to the two lowest eigenfrequencies are shown in
Figure 6. The deformed plot indicates an oscillation in the xy-plane for the lowest
eigenfrequency, while the second lowest eigenmode shows an oscillation in the
yz-plane.
Figure 6: Eigenmodes of the two lowest eigenfrequencies.
2011 COMSOL
13 |
Notes About the COMSOL Implementation

Any loads present on the model, such as the load from the static load case above, are
ignored in the default eigenfrequency analysis. It is also possible to include effects from
prestress. You can find an example of such an analysis in the model Vibrating String.
Add a new study to your model.
MODEL WIZARD
1 In the Model Builder window, right-click the root node and choose Add Study.
3 Find the Studies subsection. In the tree, select Preset Studies>Eigenfrequency.
4 Click Finish.
STUDY 2

2 Go to the Rename Study dialog box and type Study 2 (Eigenfrequency) in the
New name edit field.
3 Click OK.
STUDY 2 (EIGENFREQUENCY)
Solver 2
1 In the Model Builder window, right-click Study 2 (Eigenfrequency) and choose
Compute.
2 Expand the Study 2 (Eigenfrequency)>Solver Configurations node.
3 Right-click Study 2 (Eigenfrequency)>Solver Configurations>Solver 2 and choose
Rename.
4 Go to the Rename Solver dialog box and type Eigenfrequency Sequence in the
5 Click OK.
14 |
2011 COMSOL
RESULTS
Data Sets
Rename.
2 Go to the Rename Solution dialog box and type Eigenfrequency Solution in the
3 Click OK.
Mode Shape (solid)

As a default, the first eigenmode is shown. Follow these steps to reproduce the plot in
the left panel of Figure 6:
1 In the Model Builder window, expand the Mode Shape (solid) node, then click Surface
1.
3 Locate the Coloring and Style section. Clear the Color legend check box.
The displacement values do not have any real significance for an eigenmode plot such
as this one.
Take a look at the second mode as well.
4 In the Model Builder window, click Mode Shape (solid).
5 Go to the Settings window for 3D Plot Group.
6 Locate the Data section. From the Eigenfrequency list, choose 570.790707.
Compare the resulting plot to that to the right in Figure 6.

You can give the plot a more descriptive name:
8 Right-click Mode Shape (solid) and choose Rename.
9 Go to the Rename 3D Plot Group dialog box and type Mode Shapes in the New name
edit field.
10 Click OK.
11 Right-click Mode Shape (solid) and choose Player to create a movie showing how the
elbow bracket would deform if subjected to a harmonic load with a frequency near
the selected eigenfrequency, in this case 570.6 Hz.
To play the movie again, click the Play button on the Graphics toolbar.
2011 COMSOL
15 |
Export
The default sweep type is dynamic data extension. If you set the Sweep type to Stored
solutions and then click the Generate Frame button, you get an animation where each
frame corresponds to an eigenmode in the Eigenfrequency list. By using the Frame
number slider in the Frames section you can browse the eigenmodes.
1 In the Model Builder window, right-click Results>Export>Player 1 and choose
Rename.
2 Go to the Rename Player dialog box and type Mode Shapes in the New name edit
field.
3 Click OK.
Damped Eigenfrequency Analysis

If the material has damping, the eigenvalue solver will automatically switch to
computation of the damped eigenfrequencies. The damped eigenfrequencies and
eigenmodes are complex. The real part of the eigenfrequency corresponds to the
frequency and the imaginary part is the damping.

The first six eigenfrequencies are given below, and can be compared with the results
from the undamped model.:
EIGENFREQUENCY
FREQUENCY
UNDAMPED
FREQUENCY
f1
415+41.3i Hz
417 Hz
f2
568+56.6i Hz
571 Hz
f3
1885+397i Hz
1930 Hz
f4
2372+629i Hz
2450 Hz
f5
2946+996i Hz
3110 Hz
f6
3600+1577i Hz
3930 Hz
The relative damping of a certain mode is the ratio between the imaginary and the real
part. It can be seen that the relative damping increases rapidly as the natural frequency
increases. This is an effect of the Rayleigh damping model.
16 |
2011 COMSOL

As the eigenvalues exist as complex conjugate pairs, the first six damped
eigenfrequencies correspond to the first three undamped eigenfrequencies. For this
reason, twelve computed frequencies are requested.
MODEL WIZARD
3 Find the Studies subsection. In the tree, select Preset Studies>Eigenfrequency.
4 Click Finish.
SOLID MECHANICS
Add damping and specify the mass and stiffness parameters.
Damping 1
1 In the Model Builder window, right-click Linear Elastic Material Model 1 and choose
Damping.
2 Go to the Settings window for Damping.
3 Locate the Damping Settings section. In the dM edit field, type 300.
4 In the dK edit field, type 3.2e-5.
STUDY 3

2 Go to the Rename Study dialog box and type Study 3 (Damped Eigenfrequency)

3 Click OK.
STUDY 3 (DAMPED EIGENFREQUENCY)
Solver 3
1 In the Model Builder window, right-click Study 3 (Damped Eigenfrequency) and choose
Compute.
2 Expand the Study 3 (Damped Eigenfrequency)>Solver Configurations node.
2011 COMSOL
17 |
3 Right-click Study 3 (Damped Eigenfrequency)>Solver Configurations>Solver 3 and
choose Rename.
4 Go to the Rename Solver dialog box and type Damped Eigenfrequency Sequence

5 Click OK.
RESULTS
Data Sets
Rename.
2 Go to the Rename Solution dialog box and type Damped Eigenfrequency
Solution in the New name edit field.
3 Click OK.
Mode Shape (solid)

1 In the Model Builder window, right-click Results>Mode Shape (solid) and choose
Rename.
2 Go to the Rename 3D Plot Group dialog box and type Damped Mode Shapes in the
3 Click OK.
The mode shape is very similar to the one obtained when solving the undamped
problem.
Damped Mode Shapes

1 In the Model Builder window, expand the Results>Damped Mode Shapes node, then
click Surface 1.
18 |
2011 COMSOL
Time-Dependent Analysis
This analysis solves for the transient solution of the displacements and velocities as
functions of time. The material properties, forces, and boundary conditions can vary
in time.
The purpose of this analysis is to find the transient response from a harmonic load
during the first five periods. The excitation frequency is 500 Hz, which is between the
first and second eigenfrequencies found in the eigenfrequency analysis.
This load is applied on the face indicated in Figure 2. The expression for the load can
be written as
F x = 1.5 [ 1 + sin ( 2 500 t 2 ) ] MPa
(2)
where t denotes the time in seconds.

Because the loading is harmonic, the expected solution will consist of an initial
transient, and after long time the response will be a stationary harmonic solution with
its amplitude controlled by the damping of the system.
2011 COMSOL
19 |
The following plot shows the x-displacement at a point on the loaded face:
Figure 7: x-displacement at a point on the loaded face.

The figure below shows the von Mises stress in the bracket at 0.0036 s. The maximum
value is 252 MPa.
Figure 8: von Mises stress at t = 3.6 ms.
20 |
2011 COMSOL

In order to get a good accuracy, you should set the absolute tolerance value to a
reasonable value smaller than the actual expected displacements. The solution is always
performed in the selected base unit system, so tolerances should in this case be
expressed in m.
When a harmonic load is used, the time step can sometimes oscillate in an inefficient
manner, causing longer solution times. This can be avoided by using the more
restrictive time stepping obtained by selecting the check box for Time step increase
delay.
For more information on the settings for the time-dependent solver, see Time
Dependent in the COMSOL Multiphysics Reference Guide.
If you are working from the beginning of this example, ignore the next four
instructions. If you are starting from the short model version described in Static and
Eigenfrequency Analyses of an Elbow Bracket, load that model as described here:
1 From the View menu, choose Model Library.
2 Go to the Model Library window.
3 In the Model Library tree, select Structural Mechanics Module>Tutorial Models>elbow
bracket brief.
4 Click Open.
MODEL WIZARD
1 In the Model Builder window, right-click elbow_bracket_brief.mph and choose Add

Study.
3 Find the Studies subsection. In the tree, select Preset Studies>Time Dependent.
4 Click Finish.
STUDY 4
Solving for five periods with an excitation frequency of 500 Hz means solving for 10
ms. Save the solution every 0.2 ms.
Step 1: Time Dependent

1 In the Model Builder window, click Study 4>Step 1: Time Dependent.
2011 COMSOL
21 |
2 Go to the Settings window for Time Dependent.

3 Locate the Study Settings section. In the Times edit field, type
range(0,2e-4,10e-3).

5 Go to the Rename Study dialog box and type Study 4 (Time-Dependent) in the
6 Click OK.
7 Right-click Study 4 and choose Show Default Solver.
STUDY 4 (TIME-DEPENDENT)
Solver 4
1 In the Model Builder window, expand the Study 4 (Time-Dependent)>Solver
Configurations node.
2 Right-click Study 4 (Time-Dependent)>Solver Configurations>Solver 4 and choose
Rename.
3 Go to the Rename Solver dialog box and type Time-Dependent Sequence in the
4 Click OK.
Time-Dependent Sequence
1 In the Model Builder window, expand the Study 4 (Time-Dependent)>Solver
Configurations>Time-Dependent Sequence node, then click Time-Dependent Solver 1.
2 Go to the Settings window for Time-Dependent Solver.
3 Click to expand the Absolute Tolerance section.
4 In the Tolerance edit field, type 1e-5.
5 Click to expand the Time Stepping section.
6 Select the Time step increase delay check box. Keep the default value of 15.
This setting instructs the solver not to increase the time step until 15 consecutive
steps have been successful.
7 In the Amplification for high frequency edit field, type 0.95.
By raising this value from its default value of 0.75 you reduce the damping of high
frequencies.
You can reduce the file size significantly by not storing time derivatives when not
needed.
22 |
2011 COMSOL
8 Click to expand the Output section.

9 Clear the Store time-derivatives check box.
RESULTS
Before computing the solution, prepare a plot for displaying the results during the
solution process.
Data Sets
Rename.
2 Go to the Rename Solution dialog box and type Time-Dependent Solution in the
3 Click OK.
1D Plot Group 6
3 Locate the Data section. From the Data set list, choose Time-Dependent Solution.
4 Right-click Results>1D Plot Group 6 and choose Point Graph.
5 Select Point 30 only.
6 Go to the Settings window for Point Graph.
7 In the upper-right corner of the y-Axis Data section, click Replace Expression.
8 From the menu, choose Solid Mechanics>Displacement>Displacement field
(Material)>Displacement field, X component (u).
9 Locate the y-Axis Data section. From the Unit list, choose mm.
10 Select the Description check box.
11 In the associated edit field, type X displacement on loaded face.
13 Go to the Rename 1D Plot Group dialog box and type Time-Dependent
Displacement Graphs in the New name edit field.
14 Click OK.

1 In the Model Builder window, click Study 4 (Time-Dependent)>Step 1: Time Dependent.
2011 COMSOL
23 |

3 Click to expand the Results While Solving section.
4 Select the Plot check box.
5 From the Plot group list, choose Time-Dependent Displacement Graphs.
SOLID MECHANICS
A new load is needed for this study. You could just change the existing one, but when
you have multiple studies it is better to have individual load features and disable the
ones not currently used.
Boundary Load 1
1 In the Model Builder window, right-click Boundary Load 1 and choose Rename.
2 Go to the Rename Boundary Load dialog box and type Static Load in the New name
edit field.
3 Click OK.
4 Right-click Boundary Load 1 and choose Disable.
Boundary Load 2
1 Right-click Solid Mechanics and choose Boundary Load.
3 Go to the Rename Boundary Load dialog box and type Time-Dependent Load in the
4 Click OK.
1.5[MPa]*(1+sin(2*pi*500[Hz]*t-pi/2))
In the Model Builder window, right-click Study 4 (Time-Dependent) and choose Compute.
24 |
2011 COMSOL
RESULTS
Time-Dependent Displacement Graphs

Compare the result for the x-displacement with the graph shown in Figure 7.
3D Plot Group 7
2 Right-click Results>3D Plot Group 7 and choose Surface.
5 From the menu, choose Solid Mechanics>Stress>von Mises stress (solid.mises).
6 Right-click Surface 1 and choose Deformation.
7 In the Model Builder window, click 3D Plot Group 7.
9 Locate the Data section. From the Data set list, choose Time-Dependent Solution.
10 From the Time list, choose 0.0036.
Before generating a movie of the solution, you need to fix the scale factor for the
deformation; otherwise this factor is automatically adjusted for each movie frame in
such a way that the elbow does not move in the Graphics window.
12 In the Model Builder window, click Surface 1>Deformation 1.
13 Go to the Settings window for Deformation.
14 Locate the Scale section. Select the Scale factor check box.
16 Go to the Rename 3D Plot Group dialog box and type Time-Dependent Stress
Contour in the New name edit field.
17 Click OK.
18 Right-click 3D Plot Group 7 and choose Player.
Time-Dependent Modal Analysis

In a modal-based analysis, the problem is reduced by representing the dynamics of the
structure as a combination of small number of its most significant eigenmodes. This is
very efficient when the frequency content of the loads applied to the structure is
limited, so that only a small number of modes will be excited.
2011 COMSOL
25 |
You solve the same problem as in the previous section. Just as for the direct solution,
it is important that you set the absolute tolerance small enough.

The plot in Figure 9 below shows the same x-displacement as in the previous section
but with results from both the full and the modal based time-dependent analysis. The
correspondence between the solutions is good even though only the six first
eigenmodes are used.
Figure 9: x-displacement at a point on the loaded surface for full and modal analyses.
Figure 10 below shows the von Mises stress in the bracket at 0.0036 s. The maximum
value is 237 MPa, which can be compared with the 252 MPa computed using the
direct solution above. In general, more modes than what is needed to compute
accurate displacements are required to obtain good stress solutions.
26 |
2011 COMSOL
Figure 10: von Mises stress for modal solution.

When you create a new study, it is possible to directly select Time-Dependent
Modal. Such a study, however, generates a complete solver sequence including the
eigenvalue computation step. Because the eigenvalues are already available, you create
an Empty Study and add the study steps manually.
The undamped eigenmodes are used as the base, and the damping is provided by the
material.
In the modal procedure, all loads must have the same variation in time, specified in the
study step. This means that you should not enter any time-dependent loads (that is,
loads with an explicit dependency on the time using the time variable t).
MODEL WIZARD
2011 COMSOL
27 |
3 Find the Studies subsection. In the tree, select Custom Studies>Empty Study.
4 Click Finish.
SOLID MECHANICS
Time-Dependent Load
In the Model Builder window, right-click Model 1>Solid Mechanics>Time-Dependent Load
and choose Disable.
Boundary Load 3
3 Go to the Rename Boundary Load dialog box and type Modal Time-Dependent
Load in the New name edit field.
4 Click OK.
1.5[MPa]
STUDY 5

2 Go to the Rename Study dialog box and type Study 5 (Modal Time-Dependent)

3 Click OK.
4 Right-click Study 5 and choose Time-Dependent Modal.
STUDY 5 (MODAL TIME-DEPENDENT)
Step 1: Time-Dependent Modal

1 Go to the Settings window for Time-Dependent Modal.
2 Locate the Study Settings section. In the Times edit field, type
range(0,2e-4,10e-3).
28 |
2011 COMSOL
Solver 5
1 In the Model Builder window, right-click Study 5 (Modal Time-Dependent) and choose
Show Default Solver.
2 Expand the Study 5 (Modal Time-Dependent)>Solver Configurations node.
3 Right-click Study 5 (Modal Time-Dependent)>Solver Configurations>Solver 5 and
choose Rename.
4 Go to the Rename Solver dialog box and type Modal Time-Dependent Sequence

5 Click OK.
Modal Time-Dependent Sequence

1 In the Model Builder window, expand the Study 5 (Modal Time-Dependent)>Solver
Configurations>Modal Time-Dependent Sequence node, then click Modal Solver 1.
2 Go to the Settings window for Modal Solver.
3 Locate the General section. Find the Tolerance subsection. In the Absolute global
tolerance edit field, type 1e-5.
4 Locate the Eigenpairs section. From the Solution list, choose Eigenfrequency Sequence.
5 Click to expand the Advanced section.
6 In the Load factor edit field, type 1+sin(2*pi*500*t-pi/2).
7 In the Model Builder window, right-click Study 5 (Modal Time-Dependent) and choose
Compute.
RESULTS
Data Sets
Rename.
2 Go to the Rename Solution dialog box and type Modal Time-Dependent Solution

3 Click OK.
Stress (solid)
Reproduce the plot in Figure 10 by following these instructions:
1 In the Model Builder window, click Stress (solid).
3 Locate the Data section. From the Time list, choose 0.0036.
2011 COMSOL
29 |

5 In the Model Builder window, right-click Stress (solid) and choose Rename.
6 Go to the Rename 3D Plot Group dialog box and type Modal Time-Dependent
Stress Contour in the New name edit field.
7 Click OK.
Add the results from this study to the graph of the direct time dependent results, so
that the methods can be compared.
Time-Dependent Displacement Graphs

1 In the Model Builder window, right-click Results>Time-Dependent Displacement
Graphs and choose Point Graph.
5 From the menu, choose Displacement field, X component (u).
6 Locate the Data section. From the Data set list, choose Modal Time-Dependent
Solution.
7 Locate the y-Axis Data section. From the Unit list, choose mm.
8 Click to expand the Coloring and Style section.
9 Find the Line style subsection. From the Color list, choose Green.
Frequency Response Analysis

A frequency response analysis solves for the linearized steady-state response from
harmonic excitation loads. The loads can have amplitudes and phase shifts that may
depend on the excitation frequency, f:
F freq = F ( f ) cos 2f t + F Ph ( f ) ----------
180
(3)
where F(f) is the amplitude and FPh(f) is the phase shift of the load.
The result of a frequency response analysis is a complex time-dependent displacement
field, which can be interpreted as an amplitude, uamp, and a phase angle, uphase. The
actual displacement at any point in time is the real part of the solution
30 |
2011 COMSOL
u = u amp cos ( 2f t + u phase )
(4)
The software can visualize the amplitudes and phases as well as the solution at a specific
i
angle (time). When plotting, COMSOL Multiphysics multiplies the solution by e ,
where is the angle in radians that corresponds to the angle specified in degrees in
the Solution at angle (phase) edit field in the solution feature under the Data sets. The
plot then shows the real part of the evaluated expression
u = u amp cos ( + u phase )
(5)
The angle is available as the variable phase (radians) and can be used in plot
expressions.
The purpose of this analysis is to find the response from a harmonic load with an
excitation frequency in the range 350650 Hz, which includes the first two
eigenfrequencies found in the eigenfrequency analysis. The loads are
F x = 3 10
F z = 3 10
N/m
N/m
2
2
and there is no time shift between the load components.
2011 COMSOL
31 |

The amplitudes of the x-, y-, and z-displacements as functions of excitation frequency,
at a point on the face where the load is applied, appear in the following figure:
Figure 11: Displacement amplitudes vs. excitation frequency.

The peaks in the displacement amplitude curves are associated with the two lowest
eigenfrequencies of the bracket. Note that the lowest eigenfrequency, around 410 Hz,
corresponds to the peak on the x-displacement amplitude curve, while the next
eigenfrequency, around 570 Hz, corresponds to the peak on the z-displacement
amplitude curve. This is just as indicated by the eigenmode shapes obtained from the
eigenfrequency analysis.

The loads are given without explicit time dependencies, since the harmonic variation
is an underlying assumption in this analysis type. The loads can have different phase
angles. This can either obtained by adding a Phase feature to the load, or by writing
the load in complex form.
Usually when performing a frequency response analysis you want to sweep over a
frequency range. This can be done using the parametric solver, using the frequency as
parameter.
32 |
2011 COMSOL
MODEL WIZARD
3 Find the Studies subsection. In the tree, select Preset Studies>Frequency Domain.
4 Click Finish.
The load is the same as in the stationary study, so you can reuse the first load feature.
SOLID MECHANICS
Static Load
1 In the Model Builder window, right-click Model 1>Solid Mechanics>Modal
Time-Dependent Load and choose Disable.
2 Right-click Static Load and choose Enable.
STUDY 6

2 Go to the Rename Study dialog box and type Study 6 (Frequency Domain) in the
3 Click OK.
STUDY 6 (FREQUENCY DOMAIN)
Step 1: Frequency Domain

1 In the Model Builder window, click Study 6 (Frequency Domain)>Step 1: Frequency
Domain.
2 Go to the Settings window for Frequency Domain.
3 Locate the Study Settings section. In the Frequencies edit field, type
range(350,10,650).
Solver 6
1 In the Model Builder window, right-click Study 6 (Frequency Domain) and choose
Compute.
2 Expand the Study 6 (Frequency Domain)>Solver Configurations node.
3 Right-click Study 6 (Frequency Domain)>Solver Configurations>Solver 6 and choose
Rename.
2011 COMSOL
33 |
4 Go to the Rename Solver dialog box and type Frequency Domain Sequence in the
5 Click OK.
RESULTS
Data Sets
Rename.
2 Go to the Rename Solution dialog box and type Frequency Domain Solution in

3 Click OK.
Stress (solid)
1 In the Model Builder window, expand the Stress (solid) node, then click Surface 1.
3 Locate the Expression section. From the Unit list, choose MPa.
6 Go to the Rename 3D Plot Group dialog box and type Frequency-Response Stress
Contour in the New name edit field.
34 |
2011 COMSOL
7 Click OK.
Add a 1D plot group and reproduce the displacement amplitude graphs in Figure 11.
1D Plot Group 10
3 Locate the Data section. From the Data set list, choose Frequency Domain Solution.
4 Locate the Title section. From the Title type list, choose Manual.
5 In the Title text area, type Displacement amplitudes.
6 Right-click Results>1D Plot Group 10 and choose Rename.
7 Go to the Rename 1D Plot Group dialog box and type Frequency Response
Displacement Graphs in the New name edit field.
8 Click OK.
Frequency Response Displacement Graphs

1 In the Model Builder window, right-click Results>Frequency Response Displacement
5 From the menu, choose Solid Mechanics>Displacement>Displacement amplitude
(Material)>Displacement amplitude, X component (solid.uAmpX).
6 Click to expand the Legends section.
7 Select the Show legends check box.
8 From the Legends list, choose Manual.
9 In the table, enter the following settings:
LEGENDS
10 In the Model Builder window, right-click Frequency Response Displacement Graphs and
choose Point Graph.

2011 COMSOL
35 |

(Material)>Displacement amplitude, Y component (solid.uAmpY).
15 Locate the Coloring and Style section. Find the Line style subsection. From the Color
list, choose Green.

16 Locate the Legends section. Select the Show legends check box.
LEGENDS
choose Point Graph.

(Material)>Displacement amplitude, Z component (solid.uAmpZ).
list, choose Red.

LEGENDS
Click the Plot button.Compare the resulting plot with that in Figure 11.
Frequency Response Modal Analysis

You can also solve the same frequency response problem using the modal method. The
same remarks as for the Time-dependent Modal analysis above are relevant.
36 |
2011 COMSOL

In Figure 12 below, the results from the modal frequency response analysis are overlaid
on the results from the previous direct frequency response analysis (see Figure 11).
The curves are almost indistinguishable. The response is to a large degree controlled
by the lowest eigenmodes, which are used in the modal analysis. The modal method is
however more effective in terms of computer resources.
Figure 12: Displacement amplitudes vs. excitation frequency for direct and modal
frequency-response analyses.
SOLID MECHANICS
Static Load
In the Model Builder window, right-click Model 1>Solid Mechanics>Static Load and
choose Disable.
Boundary Load 4
3 Go to the Rename Boundary Load dialog box and type Modal Frequency Load in
2011 COMSOL
37 |
4 Click OK.
linper(3)[MPa]
linper(3)[MPa]
MODEL WIZARD
3 In the Studies tree, select Custom Studies>Empty Study.
4 Click Finish.
STUDY 7

2 Go to the Rename Study dialog box and type Study 7 (Modal Frequency
Response) in the New name edit field.
3 Click OK.
STUDY 7 (MODAL FREQUENCY RESPONSE)
Step 1: Frequency-Domain Modal

1 In the Model Builder window, right-click Study 7 (Modal Frequency Response) and
choose Study Steps>Frequency-Domain Modal.

2 Go to the Settings window for Frequency-Domain Modal.
3 Locate the Study Settings section. In the Frequencies edit field, type
range(350,10,650).
Solver 7
choose Show Default Solver.

2 Expand the Study 7 (Modal Frequency Response)>Solver Configurations node.
3 Right-click Study 7 (Modal Frequency Response)>Solver Configurations>Solver 7 and
choose Rename.
38 |
2011 COMSOL
4 Go to the Rename Solver dialog box and type Modal Frequency-Domain Sequence

5 Click OK.
Modal Frequency-Domain Sequence

1 In the Model Builder window, expand the Study 7 (Modal Frequency Response)>Solver
Configurations>Modal Frequency-Domain Sequence node, then click Modal Solver 1.
2 Go to the Settings window for Modal Solver.
3 Locate the Eigenpairs section. From the Solution list, select Eigenfrequency Sequence.
choose Compute.
RESULTS
Data Sets
Rename.
2 Go to the Rename Solution dialog box and type Modal Frequency-Domain
3 Click OK.
Stress (solid)
1 In the Model Builder window, expand the Stress (solid) node, then click Surface 1.
3 Locate the Expression section. From the Unit list, select MPa.
2011 COMSOL
39 |
6 Go to the Rename 3D Plot Group dialog box and type Modal Frequency-Response
Stress Contour in the New name edit field.
7 Click OK.
Frequency Response Displacement Graphs

Reproduce the plot in Figure 12 as follows:
1 In the Model Builder window, right-click Results>Frequency Response Displacement
3 Locate the Data section. From the Data set list, choose Modal Frequency-Domain
Solution.
(Material)>Displacement amplitude, X component (solid.uAmpX).
list, choose Blue.

8 From the Line list, choose Dashed.
40 |
2011 COMSOL
9 Find the Line markers subsection. From the Marker list, choose Asterisk.
LEGENDS
X (modal)
choose Point Graph.

Solution.
(Material)>Displacement amplitude, Y component (solid.uAmpY).
list, choose Green.

LEGENDS
Y(modal)
choose Point Graph.

Solution.
2011 COMSOL
41 |

(Material)>Displacement amplitude, Z component (solid.uAmpZ).
list, choose Red.

LEGENDS
Z(modal)
Parametric Analysis
A parametric analysis solves for the response as a function of a parameter. You can freely
define the parameter name and what it affects; it can be a material property, a load
parameter, or some other expression.
The purpose of this example is to find the response to static loading of the bracket as
a function of the direction of the load parameterized by the angle .
Apply the load on the face shown in Figure 2. To control the direction of the load,
introduce a parameter in the load expressions:
F x = 3 cos ( 180 ) MPa
F z = 3 sin ( 180 ) MPa
(6)
where is the angle of the load direction in the xz-plane. Let 45 45.
42 |
2011 COMSOL

The following plot shows the x-, y-, and z-displacements as functions of the direction
of the load, , at a point on the surface where the load is applied:
Figure 13: Displacement amplitudes versus load direction.

When you perform a parametric study, you will select the underlying study type (here:
Stationary) first. The you add a Parametric feature, and define the parameter values.
MODEL WIZARD
3 Find the Studies subsection. In the tree, select Preset Studies>Stationary.
4 Click Finish.
STUDY 8
2011 COMSOL
43 |
2 Go to the Rename Study dialog box and type Study 8 (Parametric Static) in

3 Click OK.
4 Right-click Study 8 and choose Parametric Sweep.
It is necessary to define the parameters to be used. The values given here do not
influence the parametric solver, but can be used for tests outside it.
GLOBAL DEFINITIONS
Parameters
NAME
EXPRESSION
DESCRIPTION
alpha
-45
Solver parameter
Add a load with a parameter dependency.

SOLID MECHANICS
Modal Frequency Load

In the Model Builder window, right-click Modal Frequency Load and choose Disable.
Boundary Load 5
3 Go to the Rename Boundary Load dialog box and type Parametric Load in the New
name edit field.
4 Click OK.
44 |
3[MPa]*cos(alpha*pi/180)
3[MPa]*sin(alpha*pi/180)
2011 COMSOL
STUDY 8 (PARAMETRIC STATIC)
Parametric Sweep
1 In the Model Builder window, click Study 8 (Parametric Static)>Parametric Sweep.
2 Go to the Settings window for Parametric Sweep.
3 Locate the Study Settings section. Under Parameter names, click Add.
4 Go to the Add dialog box.
5 In the Parameter names list, select alpha (Solver parameter).
6 Click the OK button.
7 Go to the Settings window for Parametric Sweep.
8 Locate the Study Settings section. In the Parameter values edit field, type
range(-45,5,45).
Solver 8
1 In the Model Builder window, right-click Study 8 (Parametric Static) and choose
Compute.
2 Expand the Study 8 (Parametric Static)>Solver Configurations node.
3 Right-click Study 8 (Parametric Static)>Solver Configurations>Solver 8 and choose
Rename.
4 Go to the Rename Solver dialog box and type Parametric Static Sequence in the
5 Click OK.
RESULTS
Data Sets
Rename.
2 Go to the Rename Solution dialog box and type Parametric Static Solution in

3 Click OK.
Stress (solid)
In the Model Builder window, right-click Results>Stress (solid) and choose Delete.
1D Plot Group 12
2011 COMSOL
45 |

3 Locate the Data section. From the Data set list, choose Parametric Static Solution.
4 Click to expand the Title section.
5 From the Title type list, choose None.
6 Locate the Plot Settings section. Select the x-axis label check box.
7 In the associated edit field, type Force direction (degrees).
8 Select the y-axis label check box.
9 In the associated edit field, type Displacement (mm).
13 Locate the y-Axis Data section. In the Expression edit field, type u.
LEGENDS
17 Right-click Point Graph 1 and choose Duplicate.

19 Locate the y-Axis Data section. In the Expression edit field, type v.
list, choose Green.

21 Locate the Legends section. In the table, enter the following settings:
LEGENDS

24 Locate the y-Axis Data section. In the Expression edit field, type w.
list, choose Red.
46 |
2011 COMSOL
LEGENDS

29 Go to the Rename 1D Plot Group dialog box and type Parametric Response
Graphs in the New name edit field.
30 Click OK.
Linear Buckling Analysis

A structure under compression can sometimes become unstable due to buckling. The
critical buckling load can be estimated using a Linear Buckling Analysis.
To perform this analysis, you first run a stress analysis with an arbitrary load level. In a
second study step, the buckling load is computed as a scale factor with respect to the
load used in the first analysis.
2011 COMSOL
47 |

The computed eigenvalue is 103.1. Because the load applied in the stationary step was
1 kN, the estimated buckling load is 103.1 kN. The shape of the buckling mode is
shown below.
Figure 14: Buckling mode shape.

It can be noted that the stresses caused by the pre-load are so large that for in this
structure a plastic collapse would occur long before the buckling load was reached.

When a Linear Buckling study is selected, both study steps are automatically prepared.
It is only necessary to define the reference load.
MODEL WIZARD
3 Find the Studies subsection. In the tree, select Preset Studies>Linear Buckling.
4 Click Finish.
48 |
2011 COMSOL
SOLID MECHANICS
Next, set up the buckling pre-load, beginning by disabling the parametric load.
Parametric Load
In the Model Builder window, right-click Parametric Load and choose Disable.
Boundary Load 6
3 Go to the Rename Boundary Load dialog box and type Buckling Pre-Load in the
4 Click OK.
7 Locate the Force section. From the Load type list, choose Total force.
8 Specify the Ftot vector as
0
1[kN]
STUDY 9

2 Go to the Rename Study dialog box and type Study 9 (Linear Buckling) in the
3 Click OK.
STUDY 9 (LINEAR BUCKLING)
Solver 9
1 In the Model Builder window, right-click Study 9 (Linear Buckling) and choose
Compute.
2 Expand the Study 9 (Linear Buckling)>Solver Configurations node.
3 Right-click Study 9 (Linear Buckling)>Solver Configurations>Solver 9 and choose
Rename.
4 Go to the Rename Solver dialog box and type Linear Buckling Sequence in the
2011 COMSOL
49 |
5 Click OK.
RESULTS
Data Sets
Rename.
2 Go to the Rename Solution dialog box and type Linear Buckling Solution in the
3 Click OK.
4 Right-click Solution 10 and choose Rename.
5 Go to the Rename Solution dialog box and type Linear Buckling Pre-Load
6 Click OK.
Mode Shape (solid)

With the following steps you can reproduce the plot in Figure 14:
1 In the Model Builder window, expand the Mode Shape (solid) node, then click Surface
1.
5 In the Model Builder window, right-click Mode Shape (solid) and choose Rename.
6 Go to the Rename 3D Plot Group dialog box and type Buckling Shape Plot in the
7 Click OK.
Export
1 In the Model Builder window, right-click Results>Export>Player 2 and choose
Rename.
2 Go to the Rename Player dialog box and type Time-Dependent Stress Contour

3 Click OK.
50 |
2011 COMSOL
Solved with COMSOL Multiphysics 4.0a. COPYRIGHT 2010 COMSOL AB.
Tube Connection
Introduction
A tube connection consisting of a flange with four prestressed bolts is subjected to
tensile forces. A sketch of the connection is shown below.
Model Definition
The tube is made of steel and has an outer diameter of 220 millimeters and an inner
diameter of 200 millimeters. The flange has a diameter of 360 millimeters and it is
30 millimeters thick. The connection consists of four prestressed M24 bolts. The bolts
are prestressed to 80% of the yield strength. The tensile force in the tube varies from 0
to 500 kN.
To compute the influence of the tensile force on the stress level in the bolt, the model
includes a parametric analysis. Because of symmetry in both load and geometry, you
only need to analyze one eighth of one of the flanges. The geometry has been created
in the CAD software SolidWorks and is available as an IGES file.
Two contact regions are modeled. One contact pair acts between the bottom surface
of the flange and the top surface of an additional fixed solid which supplies the
symmetry condition with respect to contact. The other contact pair acts between the
washer under the bolt head and the flange.
TU B E C O N N E C T I O N
Note: This model requires both the Structural Mechanics Module and the CAD
Import Module.

After the pretension step there is a tensile stress in the bolt, and compressive stress in
the flange under the bolt. This is illustrated in Figure 1.
Figure 1: The axial stress after the pretension step.

The general stress state at maximum external load is shown in Figure 2. In addition to
the stress that has developed in the fillet between tube and flange, there are some
interesting features. The stress in the bolt has increased significantly and is no longer
2 |
constant over the cross section. Furthermore a stress of the order 300 MPa has
developed at the outer edge of the flange.
Figure 2: Effective stress at maximum external load.

The applied external load in this example severely overloads the bolted joint. Up to
about half the full load it works fairly well, but then the contact between the two
mating flanges start to open up, so that the external force is transmitted directly by the
bolts. This is displayed in Figure 3. Actually, the conditions are even worse than the
average force indicates. The bolt is subjected to bending with a non-uniform stress
distribution over the cross section. The maximum stress is well above the yield limit,
and is approaching the ultimate stress. The development of the axial stress in two
points on opposite sides of the bolt is displayed in Figure 4. The points are located in
the x-z symmetry plane. One point is as close to the tube centerline as possible and the
other as far out as possible.
Figure 3: The bolt force as a function of the tensile force.
Figure 4: The development of the bolt stress at two different positions in the cross section.
4 |
The plots of the contact pressure between the mating flanges shown in Figure 5 and
Figure 6 are informative. In the first plot the contact pressure after bolt pretensioning
is shown. It can be seen that there are too few bolts, since almost no contact pressure
is established over a large part of the flange. When the full external load is applied, the
flanges are separating even directly under the bolts and the contact region has moved
outwards. This explains the high stress seen at the outer edge of the flange.
Figure 5: Contact pressure between flanges after pretensioning the bolts.
Figure 6: Contact pressure between the flanges at full external load.

The analysis is performed in two steps, represented by two studies. In the first step, the
effects of pretensioning the bolt are computed, and in the second step the external load
on the tube is applied as a parametric sweep.
The pretensioning of the bolt is modelled using an auxiliary variable, representing a
prestrain on a piece of the bolt. The level of this prestrain is determined by an extra
equation, stating that the prescribed pretension force must equal the integral of the
axial stress over a section through the bolt. In the following step when the external
load is applied, the prestrain is kept constant, while the axial force is free to adapt to
the changes in external load.
Model Library path: Structural_Mechanics_Module/Contact_and_Friction/

tube_connection
6 |
MODEL WIZARD

2 Click Next.
3 In the Add Physics tree, select Structural Mechanics>Solid Mechanics (solid).
4 Click Add Selected.
5 In the Add Physics tree, select Mathematics>ODEs and DAEs (ge).
7 Click Next.
8 In the Studies tree, select Preset Studies for Selected Physics>Stationary.
9 Click Finish.
GLOBAL DEFINITIONS
Parameters
NAME
EXPRESSION
DESCRIPTION
A_tube
pi/
4*(0.22^2-0.2^2)[m^
2]
Tube area
S_pre
800[MPa]*0.8*0.8
Prestress in the bolt (class

8.8)
Di_wash
25[mm]
Washer inner diameter
Do_wash
45[mm]
Washer outer diameter
T_wash
4[mm]
Washer thickness
Do_bolthead
38[mm]
Bolthead diameter
H_bolthead
15[mm]
Bolthead height
As_bolt
353[mm^2]
Stress area of bolt
Ds_bolt
sqrt(4/pi*As_bolt)
Effective bolt diameter
F_pre
S_pre*As_bolt
Pretension force
R_bc
150[mm]
Bolt circle radius
NAME
EXPRESSION
F_tube
500[kN]
DESCRIPTION
Axial force in tube
loadpar
Tube load multiplier
GEOMETRY 1
Import 1
3 Locate the Import section. Click the Browse button.
tube_connection.igs.
5 Click the Import button.

Cut out 1/8 of the structure to make use of the repetitive symmetry.
Work Plane 1
Rectangle 1
1 In the Model Builder window, right-click Work Plane 1>Geometry and choose
Rectangle.
3 Locate the Position section. In the y edit field, type -0.1.
4 Locate the Size section. In the Width edit field, type 0.2.
5 In the Height edit field, type 0.3.
Revolve 1
1 In the Model Builder window, right-click Work Plane 1 and choose Revolve.
2 Go to the Settings window for Revolve.
3 Locate the Revolution Angles section. In the End angle edit field, type 45.
8 |
Intersection 1
1 In the Model Builder window, right-click Geometry 1 and choose Boolean
Operations>Intersection.
2 Select the objects imp1 and rev1 only.
Next, create the bolt.
Work Plane 2
Bzier Polygon 1
1 In the Model Builder window, right-click Work Plane 1>Geometry and choose Bzier
Polygon.
2 Go to the Settings window for Bzier Polygon.
3 Locate the Polygon Segments section. Click the Add Linear button.
4 Find the Control points subsection. In row 1, set x to R_bc.
5 In row 1, set y to -0.03.
6 In row 2, set x to R_bc+Ds_bolt/2 and y to -0.03.
7 Click the Add Linear button.
8 In row 2, set y to T_wash.
10 In row 2, set x to R_bc+Di_wash/2.
12 In row 2, set y to 0.
14 In row 2, set x to R_bc+Do_wash/2.
16 In row 2, set y to T_wash.
18 In row 2, set x to R_bc+Do_bolthead/2.
20 In row 2, set y to T_wash+H_bolthead.

22 In row 2, set x to R_bc.
24 Click the Close Curve button.
Add a region for prestrain application. The size is not important, but it is reasonable
to make it of same order as the bolt diameter, in order not to impose any unnecessary
constraints on the mesh size.
Rectangle 1
Rectangle.
3 Locate the Position section. In the x edit field, type R_bc.
4 In the y edit field, type -0.03.
5 Locate the Size section. In the Width edit field, type Ds_bolt/2.
6 In the Height edit field, type 10[mm].
Copy 1
Transforms>Copy.
2 Select the object r1 only.
3 Go to the Settings window for Copy.
4 Locate the Displacement section. In the y edit field, type 0.01.
Union 1
1 In the Model Builder window, right-click Work Plane 2>Geometry and choose Boolean
Operations>Union.
2 Select the objects r1, copy1, and b1 only.
Revolve 2
1 In the Model Builder window, right-click 1>Work Plane 2 and choose Revolve.
2 Go to the Settings window for Revolve.
10 |
3 Locate the Revolution Angles section. In the End angle edit field, type 180.
4 Locate the Point on the Revolution Axis section. In the x edit field, type R_bc.
Form Union
1 In the Model Builder window, click Form Union.
2 Go to the Settings window for Finalize.
3 Locate the Finalize section. From the Finalization method list, select Form an assembly.
Block 1
1 In the Model Builder window, right-click Geometry 1 and choose Block.
2 Go to the Settings window for Block.
3 Locate the Size and Shape section. In the Width edit field, type 0.15.
4 In the Depth edit field, type 0.01.
6 Locate the Position section. In the x edit field, type 0.05.
7 In the y edit field, type -0.04.
8 In the z edit field, type -0.14.
10 Click in the Graphics window, Press Ctrl+A, and then right-click to select all objects.
11 Click the Zoom Selected button on the Graphics toolbar.
11
This completes the geometry modeling stage. The geometry should now look like that
in the figure below.
DEFINITIONS
1 In the Model Builder window, right-click Model 1>Definitions and choose Contact Pair.
3 Go to the Settings window for Contact Pair.
4 In the upper-right corner of the Destination Boundaries section, click Activate
Selection.
6 In the Model Builder window, right-click Definitions and choose Contact Pair.
8 Go to the Settings window for Contact Pair.
9 In the upper-right corner of the Destination Boundaries section, click Activate
Selection.
Set up an integration through the bolt, so that the axial force can be computed from
the stress.
12 |
Integration 1
1 In the Model Builder window, right-click Definitions and choose Model
Couplings>Integration.
2 Go to the Settings window for Integration.
3 Locate the Operator Name section. In the Operator name edit field, type IntSec.
4 Locate the Source Selection section. From the Geometric entity level list, select
Boundary.
Create a variable for the current bolt force, taking the symmetry into account.
Variables 1
1 In the Model Builder window, right-click Definitions and choose Variables.
NAME
EXPRESSION
DESCRIPTION
F_bolt
2*IntSec(solid.sy)
Bolt force
MATERIALS
Browser.
3 Locate the Materials section. In the Materials tree, select Built-In>Structural steel.
SOLID MECHANICS
Fixed Constraint 1
1 In the Model Builder window, right-click Model 1>Solid Mechanics and choose
More>Fixed Constraint.
Symmetry 1
1 In the Model Builder window, right-click Solid Mechanics and choose Symmetry.
13
2 Select Boundaries 8, 14, 17, 18, 27, 28, 30, 34, 36, 39, and 42 only.
You can do this by first copying the text 8, 14, 17, 18, 27, 28, 30, 34, 36, 39, and
42 and then clicking the Paste Selection button next to the Selection box or clicking
in the box and pressing Ctrl+V.
Contact 1
1 In the Model Builder window, right-click Solid Mechanics and choose Pairs>Contact.
2 Go to the Settings window for Contact.
3 Locate the Pair Selection section. In the Pairs list, select Contact Pair 1.
4 Locate the Normal Contact section. In the pn edit field, type
min(1e-3*5^segiter,1)*solid.E/h*50.
5 Locate the Initial Values section. In the Tn edit field, type S_pre/30.
Contact 2
1 In the Model Builder window, right-click Solid Mechanics and choose Pairs>Contact.
3 Locate the Pair Selection section. In the Pairs list, select Contact Pair 2.
4 Locate the Normal Contact section. In the pn edit field, type
min(1e-3*5^segiter,1)*solid.E/h*50.
5 Locate the Initial Values section. In the Tn edit field, type S_pre/3.
Boundary Load 1
1 In the Model Builder window, right-click Solid Mechanics and choose Boundary Load.
4 Locate the Force section. Specify the F vector as
0
loadpar*F_tube/A_tube
ODES AND DAES
1 In the Model Builder window, click Model 1>ODEs and DAEs.

2 Go to the Settings window for ODEs and DAEs.
3 Locate the Interface Identifier section. In the Identifier edit field, type prestrain.
14 |
Global Equations 1
1 In the Model Builder window, expand the ODEs and DAEs node, then click Global
Equations 1.
2 Go to the Settings window for Global Equations.
3 Locate the Global Equations section. In the table, enter the following settings:
NAME
EQUATION F(U,UT,UTT,T)
INITIAL VALUE (U)
eps_p
F_pre-F_bolt
-S_pre/2E11[Pa]*0.75*7
A reasonable initial guess for the prestrain is based on that 75% of the deformation
occurs in the bolt, and that the region you are using for prestrain here is roughly 1/
7 of the total bolt length.
SOLID MECHANICS
Initial Stress and Strain 1

1 In the Model Builder window, right-click Linear Elastic Material Model 1 and choose
Initial Stress and Strain.
2 Go to the Settings window for Initial Stress and Strain.
3 Locate the Domains section. Click Clear Selection.
5 Locate the Initial Stress and Strain section. In the 0 table, enter the following
settings:
0
eps_p
MESH 1
Mapped 1
1 In the Model Builder window, right-click Model 1>Mesh 1 and choose More
Operations>Mapped.
Size 1
1 Right-click Mapped 1 and choose Size.
15
3 Locate the Element Size section. From the Predefined list, select Coarse.
Swept 1
1 In the Model Builder window, right-click Mesh 1 and choose Swept.
2 Go to the Settings window for Swept.
3 Locate the Domains section. From the Geometric entity level list, select Domain.
Distribution 1
1 Right-click Swept 1 and choose Distribution.
2 Go to the Settings window for Distribution.
3 Locate the Distribution section. In the Number of elements edit field, type 1.
Free Tetrahedral 1
1 In the Model Builder window, right-click Mesh 1 and choose Free Tetrahedral.
2 Go to the Settings window for Free Tetrahedral.
3 Locate the Domains section. From the Geometric entity level list, select Domain.
Edge 1
1 In the Model Builder window, right-click Mesh 1 and choose More Operations>Edge.
2 Select Edge 98 only.
Distribution 1
1 Right-click Edge 1 and choose Distribution.
2 Select Edge 98 only.
Swept 2
1 In the Model Builder window, right-click Mesh 1 and choose Swept.
2 Go to the Settings window for Swept.
3 In the upper-right corner of the Source Faces section, click Activate Selection.
4 Click to expand the Source Faces section.
16 |
5 Select Boundaries 36, 39, and 42 only.

6 Click to expand the Destination Faces section.
7 In the upper-right corner of the Destination Faces section, click Activate Selection.
8 Select Boundaries 18, 27, and 30 only.
Size 1
1 Right-click Swept 2 and choose Size.
3 Locate the Element Size section. From the Predefined list, select Extremely fine.
Distribution 1
1 In the Model Builder window, right-click Swept 2 and choose Distribution.
STUDY 1
1 In the Model Builder window, right-click Study 1 and choose Show Default Solver.
2 Expand the Study 1>Solver Configurations node.
Solver 1
1 In the Model Builder window, expand the Solver Configurations>Solver 1 node, then
click Dependent Variables 1.

2 Go to the Settings window for Dependent Variables.
3 Locate the Scaling section. From the Method list, select Manual.
4 In the Scale edit field, type 1e8.
5 In the Model Builder window, expand the Dependent Variables 1 node, then click
mod1_u.
6 Go to the Settings window for Field.
8 In the Scale edit field, type 1e-5.
9 In the Model Builder window, click Dependent Variables 1>mod1_ODE1.
10 Go to the Settings window for State.
17
The default solver sequence segregates the ODE variable from the displacement
variables, which is not what you want to do. Modify the segregated groups as
follows:
13 In the Model Builder window, expand the Stationary Solver 1>Segregated 1 node.
14 Right-click Stationary Solver 1>Segregated 1>Segregated Step 1 and choose Delete.
15 In the Confirm Delete dialog box, click Yes.
16 Go to the Settings window for Segregated Step.
17 Locate the General section. Under Variables, click Add.
19 In the Variables list, select mod1_ODE1.
21 In the Model Builder window, right-click Stationary Solver 1>Segregated 1>Lumped
Step 1 and choose Delete.
23 In the Model Builder window, right-click Stationary Solver 1>Segregated 1 and choose
Segregated Step.
27 In the Variables list, select mod1_solid_Tn_p1 and mod1_solid_Tn_p2.
29 In the Model Builder window, right-click Study 1 and choose Compute.
RESULTS
Reproduce the plot in Figure 1 with the following steps:
3D Plot Group 1
1 In the Model Builder window, expand the Results>3D Plot Group 1 node, then click
Surface 1.
4 From the menu, choose Solid Mechanics>Stress tensor>Stress tensor y (solid.sy).
5 Click to expand the Range section.
18 |
6 Select the Manual color range check box.

7 In the Minimum edit field, type -3e8.
8 In the Maximum edit field, type 5e8.
MODEL WIZARD
3 In the Studies tree, select Preset Studies for Selected Physics>Stationary.
4 Click Finish.
STUDY 2
In the Model Builder window, right-click Study 2 and choose Show Default Solver.
Solver 2
1 In the Model Builder window, expand the Study 2>Solver Configurations>Solver 2
node, then click Dependent Variables 1.

2 Go to the Settings window for Dependent Variables.
3 Locate the Initial Values section. From the Method list, select Solution.
4 From the Solution list, select Solver 1.
6 In the Scale edit field, type 1e8.
7 Locate the Variables Not Solved For section. From the Method list, select Solution.
8 From the Solution list, select Solver 1.
9 In the Model Builder window, expand the Dependent Variables 1 node, then click
mod1_u.
The pre-tension strain is to be kept constant during the application of the external
load, so it should not be solved for.
13 In the Model Builder window, click Dependent Variables 1>mod1_ODE1.
14 Go to the Settings window for State.
15 Locate the General section. Clear the Solve for this state check box.
19
16 In the Model Builder window, expand the Stationary Solver 1>Segregated 1 node.
17 Right-click Stationary Solver 1>Segregated 1>Segregated Step 1 and choose Disable.
18 Right-click Stationary Solver 1>Segregated 1>Lumped Step 1 and choose Delete.
20 In the Model Builder window, right-click Stationary Solver 1>Segregated 1 and choose
Segregated Step.
24 In the Variables list, select mod1_solid_Tn_p1 and mod1_solid_Tn_p2.
26 In the Model Builder window, right-click Stationary Solver 1 and choose Parametric.
27 Go to the Settings window for Parametric.
28 Locate the General section. In the Parameter names edit field, type loadpar.
29 In the Parameter values edit field, type range(0,0.1,1).
Because the contact is well established, the contact iterations do not need the
smooth ramping of the penalty factor.
SOLID MECHANICS
Contact 1
1 In the Model Builder window, click Contact 1.
3 Locate the Normal Contact section. In the pn edit field, type solid.E/
solid.hmin_dst*50.
Contact 2
1 In the Model Builder window, click Contact 2.
3 Locate the Normal Contact section. In the pn edit field, type solid.E/
solid.hmin_dst*50.
STUDY 2
In the Model Builder window, right-click Study 2 and choose Compute.
20 |
RESULTS
The following steps reproduce the plot in Figure 2:
3D Plot Group 3
1 In the Model Builder window, expand the Results>3D Plot Group 3 node, then click
Surface 1.
4 From the menu, choose Solid Mechanics>von Mises stress (solid.mises).
6 Locate the Range section. Select the Manual color range check box.
Proceed to plot the bolt force as a function of the tensile force as in Figure 3.
1D Plot Group 5
3 Locate the Data section. From the Data set list, select Solution 2.
4 Right-click Results>1D Plot Group 5 and choose Global.
5 Go to the Settings window for Global.
6 Locate the Expressions section. In the Expressions table, enter the following settings:
EXPRESSION
F_bolt/1000

9 Clear the Show legends check box.
11 Find the Line style subsection. In the Width edit field, type 2.
12 Locate the X-Axis Data section. From the Parameter list, select Expression.
13 In the Expression edit field, type 500*loadpar.
21
16 Locate the Plot Settings section. In the x-axis label edit field, type Axial force on
tube (kN).
17 In the y-axis label edit field, type Axial force in bolt (kN).
The following steps create the plot in Figure 4:
1D Plot Group 6
5 Select Vertices 33 and 51 only.
8 From the menu, choose Solid Mechanics>Stress tensor>Stress tensor y (solid.sy).
9 Locate the Expression section. From the Unit list, select MPa.
10 Locate the X-Axis Data section. From the Parameter list, select Expression.
11 In the Expression edit field, type 500*loadpar.
13 Find the Line style subsection. From the Line list, select Cycle.
14 In the Width edit field, type 2.
16 Select the Show legends check box.
17 From the Legends list, select Manual.
LEGENDS
Outside
Inside

21 Locate the Plot Settings section. In the x-axis label edit field, type Axial force on
tube (kN).
22 |
22 In the y-axis label edit field, type Bolt stress (MPa).

Finally, reproduce the contact pressure plots shown in Figure 5 and Figure 6:
3D Plot Group 7
4 In the upper-right corner of the Expression section, click Contact pressure, contact
pair p1 (solid.Tn_p1).
5 Locate the Range section. Select the Manual color range check box.
23
24 |
Tutorial: 6
A L E Flui d- S t r uc t u re In t erac t i on
Introduction
The following example demonstrates techniques for modeling fluid-structure
interactions in COMSOL Multiphysics. It illustrates how fluid flow can deform
structures and how to solve for the flow in a continuously deforming geometry using
the arbitrary Lagrangian-Eulerian (ALE) technique.
The model geometry consists of a horizontal flow channel in the middle of which is an
obstacle, a narrow vertical structure (Figure 1). The fluid flows from left to right,
except where the obstacle forces it into a narrow path in the upper part of the channel,
and it imposes a force on the structures walls resulting from the viscous drag and fluid
pressure. The structure, being made of a deformable material, bends under the applied
load. Consequently, the fluid flow also follows a new path, so solving the flow in the
original geometry would generate incorrect results.
The ALE method handles the dynamics of the deforming geometry and the moving
boundaries with a moving grid. COMSOL Multiphysics computes new mesh
coordinates on the channel area based on the movement of the structures boundaries
and mesh smoothing. The Navier-Stokes equations that solve the flow are formulated
for these moving coordinates.
The structural mechanics portion of the model does not require the ALE method, and
COMSOL Multiphysics solves it in a fixed coordinate system as usual. However, the
strains the model computes in this way are the only source for computing the
deformed coordinates with ALE.
Figure 1: Fluid flows into this horizontal flow channel from the left, and it enters with a
parabolic velocity profile. A narrow vertical structure in the channel (the straight vertical
structure) forces the flow into a narrow path. Due to fluid pressure and viscous drag, the
originally vertical structure bends. This simulation models the fluid flow in a deformed,
moving mesh that follows the movement of the bending structure.
Model Definition
In this example the flow channel is 100 m high and 300 m long. The vertical
structure5 m wide, 50 m high, and with a semicircular topsits 100 m away
from the channels left boundary. Assume that the structure is long in the direction
perpendicular to the image.
The fluid is a water-like substance with a density = 1000 kg/m3 and dynamic
viscosity = 0.001 Pas. To demonstrate the desired techniques, assume the structure
consists of a flexible material with a density = 7850 kg/m3 and Youngs modulus
E = 200 kPa.
FLUID FLOW
The fluid flow in the channel is described by the incompressible Navier-Stokes

equations for the velocity field, u = (u, v), and the pressure, p, in the spatial
(deformed) moving coordinate system:
u
T
[ p I + ( u + ( u ) ) ] + ( ( u um ) )u = F
t
(1)
u = 0
In these equations, I denotes the unit diagonal matrix and F is the volume force
affecting the fluid. Assume that no gravitation or other volume forces affect the fluid,
so that F = 0. The coordinate system velocity is um = (um, vm).
At the channel entrance on the left, the flow has fully developed laminar characteristics
with a parabolic velocity profile but its amplitude changes with time. At first flow
increases rapidly, reaching its peak value at 0.215 s; thereafter it gradually decreases to
a steady-state value of 5 cm/s. The centerline velocity in the x direction, uin (see
Figure 4), with the steady-state amplitude U comes from the equation
2
Ut
u in = ----------------------------------------------------------2 2
2
( 0.04 t ) + ( 0.1t )
(2)
where t must be expressed in seconds.

At the outflow (right-hand boundary), the condition is p = 0. On the solid
(nondeforming) walls, no-slip conditions are imposed, u = 0, v = 0, while on the
deforming interface the velocities equal the deformation rate, u0 = ut and v0 = vt (the
default condition; note that u and v on the right-hand sides refer to the displacement
components).
STRUCTURAL MECHANICS
The structural deformations are solved for using an elastic formulation and a nonlinear
geometry formulation to allow large deformations.
The obstacle is fixed to the bottom of the fluid channel. All other object boundaries
experience a load from the fluid, given by
T
F T = n ( p I + ( u + ( u ) ) )
(3)
where n is the normal vector to the boundary. This load represents a sum of pressure
and viscous forces.
MOVING MESH
The Navier-Stokes equations are solved on a freely moving deformed mesh, which
constitutes the fluid domain. The deformation of this mesh relative to the initial shape
of the domain is computed using Winslow smoothing. This is the default smoothing
when using the Fluid-Structure Interaction interface. For more information, please
refer to the chapter The Fluid-Structure Interaction Interface in the MEMS Module
Users Guide. Inside the obstacle, the moving mesh follows the deformations of the
obstacle. At the exterior boundaries of the flow domain, the deformation is set to zero
in all directions.

Figure 2 shows the geometry deformation and flow at t = 4 s when the system is close
to its steady state. Due to the channels small dimensions, the Reynolds number of the
flow is small (Re << 100), and the flow stays laminar in most of the area. The swirls are
restricted to a small area behind the structure. The amount of deformation as well as
the size and location of the swirls depend on the magnitude of the inflow velocity.
Figure 2: Flow velocity and geometry deformation at t = 4 s. The streamlines indicate the
flow direction and the color indicates flow-velocity magnitude.
Figure 3 shows the mesh velocity at t = 0.15 s. The boundaries of the narrow structure
are the only moving boundaries of the flow channel. Therefore the mesh velocity also
has its largest values near the structure. Depending on the current state of the
deformationwhether it is increasing, decreasing or stationarythe mesh velocity can
have a very different distribution. Figure 4 further illustrates this point; it compares the
average inflow velocity to the horizontal mesh velocity and the horizontal mesh
displacement just beside the top of the structure. Most of the time the deformation
follows the inflow velocity quite closely. Whenever the inflow velocity starts to
decrease, the deformation also decreases, which you can observe as the negative values
on the horizontal mesh velocity. Toward the end of the simulation, when inflow and
structure deformation approach their steady-state values, the mesh velocity also
decreases to zero.
Figure 3: Mesh velocity (arrows) and mesh and geometry deformation at t = 0.15 s.
Figure 4: Inflow velocity, horizontal mesh velocity, and mesh deformation. The curve with
triangles shows the average x direction velocity at the inflow boundary (m/s); the curve
with circles shows 104 mesh displacement in the x direction (dx_ale; m) at the geometry
point (1.0510-4, 0.510-4); and the curve with squares shows 103 mesh velocity in the
x direction (xt; m/s), also at the point (1.0510-4, 0.510-4).
Figure 5 compares the meshes at different times. The first image shows the initial
mesh, which you generate prior to solving the model. This mesh is equally distributed
around the top of the structure. The second image shows the mesh in its deformed
form. Because the structure deforms more in the horizontal direction, the mesh also
changes more in this direction: On the left, the mesh elements are stretched; on the
right, they are compressed in the x direction.
Figure 5: Geometry and mesh near the top of the structure at t = 0 s and 2 s.

This example implements the model using Fluid-Structure Interaction interface. By
default the Fluid-Structure Interaction interface treats all domains as fluid. Activate
solid material model node in the area of the narrow structure. To get a more accurate
computation of the large strains, large deformation analysis is the default setting. The
interface automatically identifies the fluid-solid interaction boundaries and assigns the
boundary condition to those boundaries.
Model Library path: MEMS_Module/Microfluidics_Models/

fluid_structure_interaction
MODEL WIZARD

2 Click the 2D button.
3 Click Next.
4 In the Add Physics tree, select Fluid Flow>Fluid-Structure Interaction (fsi).
6 Click Next.
7 In the Studies tree, select Preset Studies>Time Dependent.

8 Click Finish.
GLOBAL DEFINITIONS
Parameters
NAME
EXPRESSION
DESCRIPTION
3.33[cm/s]
Inlet mean velocity at steady state
Variables 1
1 In the Model Builder window, right-click Global Definitions and choose Variables.
NAME
EXPRESSION
DESCRIPTION
u_in
U*t^2/
sqrt(t^4-0.07[s^2]*t^2+
0.0016[s^4])
Inlet mean velocity
GEOMETRY 1
1 In the Model Builder window, click Model 1>Geometry 1.

2 Go to the Settings window for Geometry.
3 Locate the Geometry Settings section. Find the Units subsection. From the Length
unit list, select m.
Rectangle 1
1 Right-click Model 1>Geometry 1 and choose Rectangle.
3 Locate the Size section. In the Width edit field, type 300.
4 In the Height edit field, type 100.
Rectangle 2
1 In the Model Builder window, right-click Geometry 1 and choose Rectangle.

3 Locate the Size section. In the Width edit field, type 5.
5 Locate the Position section. In the x edit field, type 100.
Fillet 1
1 In the Model Builder window, right-click Geometry 1 and choose Fillet.
2 On the object r2, select Vertices 3 and 4 only.
3 Go to the Settings window for Fillet.
4 Locate the Radius section. In the Radius edit field, type 2.5.
5 Click the Build All button. The geometry should look like that in the figure below.
FLUID-STRUCTURE INTERACTION
By default the Fluid-Structure Interaction interface treats all domains as fluid. Add the
appropriate domain selection to the default node for the solid domain and proceed to
the material specification.
Linear Elastic Material Model 1

1 In the Model Builder window, expand the Model 1>Fluid-Structure Interaction node,
then click Linear Elastic Material Model 1.

MATERIALS
Material 1
1 In the Model Builder window, right-click Model 1>Materials and choose Material.
3 Go to the Settings window for Material.
4 Locate the Material Contents section. In the Material Contents table, enter the
following settings:
PROPERTY
NAME
VALUE
Density
rho
1e3
Dynamic viscosity
mu
1e-3
Material 2
1 In the Model Builder window, right-click Materials and choose Material.
3 Go to the Settings window for Material.
4 Locate the Material Contents section. In the Material Contents table, enter the
following settings:
PROPERTY
NAME
VALUE
Young's modulus
2e5
Poisson's ratio
nu
0.33
Density
rho
7850
FLUID-STRUCTURE INTERACTION
1 In the Model Builder window, click Model 1>Fluid-Structure Interaction.

2 Go to the Settings window for Fluid-Structure Interaction.
3 Locate the Physical Model section. From the Compressibility list, select Incompressible
flow.
Inlet 1
1 In the Model Builder window, right-click Fluid-Structure Interaction and choose the
boundary condition Laminar Flow>Inlet.

3 Go to the Settings window for Inlet.

4 Locate the Boundary Condition section. From the Boundary condition list, select
Laminar inflow.
5 In the Uav edit field, type u_in.
Outlet 1
boundary condition Laminar Flow>Outlet.

Fixed Constraint 1
boundary condition Solid Mechanics>Fixed Constraint.

MESH 1
Free Triangular 1
1 In the Model Builder window, right-click Model 1>Mesh 1 and choose Free Triangular.
2 Go to the Settings window for Free Triangular.
3 Locate the Domains section. From the Geometric entity level list, select Entire
geometry.
Size
1 In the Model Builder window, click Size.
3 Locate the Element Size section. From the Calibrate for list, select Fluid dynamics.
STUDY 1

1 In the Model Builder window, click Study 1>Step 1: Time Dependent.
3 Locate the Study Settings section. In the Times edit field, type range(0,0.005,1.5)
range(1.7,0.02,4).
4 Select the Relative tolerance check box.

5 In the associated edit field, type 0.001.
6 In the Model Builder window, right-click Study 1 and choose Show Default Solver.
Solver 1
You can improve the displacement scaling by changing the scale to 105.
1 In the Model Builder window, expand the Study 1>Solver Configurations>Solver
1>Dependent Variables 1 node, then click Dependent Variables 1>mod1_u_solid.
4 In the Scale edit field, type 1.0e-5.
5 In the Model Builder window, right-click Study 1 and choose Compute. You can
ignore the non-ideal constraints related warning shown in the log.

RESULTS
2D Plot Group 1
The default plot shows the pressure distribution in the fluid domain. To reproduce
Figure 1, plot the velocity field.
2 Locate the Data section. From the Time list, select 2.
3 In the Model Builder window, expand the 2D Plot Group 1 node, then click Surface 1.
6 From the menu, choose Fluid-Structure Interaction (Laminar Flow)>Velocity
magnitude (fsi.U).
7 Locate the Coloring and Style section. Select the Wireframe check box.
8 Clear the Color legend check box.
10 In the Model Builder window, right-click 2D Plot Group 1 and choose Arrow Surface.
11 Go to the Settings window for Arrow Surface.
13 From the menu, choose Fluid-Structure Interaction (Laminar Flow)>Velocity field
(u_fluid, v_fluid).
14 Locate the Coloring and Style section. In the Scale factor edit field, type 1.5.
2D Plot Group 2
Proceed to reproduce Figure 2, which shows flow velocity and streamlines.
3 Locate the Plot Settings section. From the Frame list, select Spatial (x, y, z).
7 From the menu, choose Fluid-Structure Interaction (Laminar Flow)>Velocity
magnitude (fsi.U).
9 In the Model Builder window, right-click 2D Plot Group 2 and choose Streamline.
10 Go to the Settings window for Streamline.
12 From the menu, choose Fluid-Structure Interaction (Laminar Flow)>Velocity field
(u_fluid, v_fluid).
13 Locate the Streamline Positioning section. From the Entry method list, select
Coordinates.
14 In the x edit field, type 0^(range(1,15)) 125*1^(range(1,2)).
15 In the y edit field, type range(0,100/14,100) 20 5.
16 Locate the Coloring and Style section. From the Color list, select Red.
Report
To animate flow around the structure, do the following:
1 In the Model Builder window, right-click Results>Report and choose Player.
2 Go to the Settings window for Player.
3 Locate the Scene section. From the Subject list, select 2D Plot Group 2.
4 Locate the Parameter Sweep section. From the Select via list, select Interpolated
times.
5 In the Times edit field, type range(0.025,0.025,0.5).
6 Right-click Player 1 and choose Play.
To inspect the deformed geometry and deformed mesh near the top of the structure
(Figure 5), proceed with the following steps.
Data Sets
1 In the Model Builder window, right-click Results>Data Sets and choose Solution.
2 Right-click Solution 2 and choose Add Selection.
Domain.
2D Plot Group 3
3 Locate the Plot Settings section. From the Frame list, select Spatial (x, y, z).
7 Locate the Expression section. In the Expression edit field, type 1.
8 Locate the Coloring and Style section. From the Coloring list, select Uniform.
9 From the Color list, select Blue.
10 Select the Wireframe check box.
11 Click to expand the Quality section.
13 In the Model Builder window, right-click 2D Plot Group 3 and choose Surface.
16 From the Time list, select 0.
17 Locate the Expression section. In the Expression edit field, type 1.
18 Locate the Coloring and Style section. From the Coloring list, select Uniform.
19 Click the Zoom Box button on the Graphics toolbar and then use the mouse to zoom
in on the obstacle.

Add the arrow plot, to reproduce Figure 3.

3 Locate the Data section. From the Time list, select 0.15.
6 Locate the Data section. From the Time list, select 0.15.
7 In the Model Builder window, right-click 2D Plot Group 3 and choose Arrow Surface.
8 Go to the Settings window for Arrow Surface.
9 Locate the Expression section. In the x component edit field, type xt.
10 In the y component edit field, type yt.
11 Locate the Coloring and Style section. In the Scale factor edit field, type 1.5.
RESULTS
Finally, plot the horizontal mesh velocity, the mesh deformation at the point beside the
top of the structure, and inflow velocity (Figure 4).
1D Plot Group 4
2 Right-click Results>1D Plot Group 4 and choose Global.
3 Go to the Settings window for Global.
4 In the upper-right corner of the Expressions section, click Replace Expression.
5 From the menu, choose Definitions>Inlet mean velocity (u_in).
Data Sets
1 In the Model Builder window, right-click Results>Data Sets and choose Cut Point 2D.
2 Go to the Settings window for Cut Point 2D.
3 Locate the Point Data section. In the x edit field, type 105.
4 In the y edit field, type 50.
1D Plot Group 4
1 In the Model Builder window, right-click Results>1D Plot Group 4 and choose Point
Graph.
3 Locate the Data section. From the Data set list, select Cut Point 2D 1.
4 Locate the Expression section. In the Expression edit field, type xt.
5 From the Unit list, select mm/s.
7 From the Legends list, select Manual.
LEGENDS
mesh velocity in the x direction (mm/s)

11 Locate the Expression section. In the Expression edit field, type x-X.
12 From the Unit list, select mm.
LEGENDS
mesh displacement in the x direction (mm)

Comsol Alex Course Program Book

Diunggah oleh

Informasi Dokumen

Hak Cipta

Format Tersedia

Bagikan dokumen Ini

Bagikan atau Tanam Dokumen

Opsi Berbagi

Apakah menurut Anda dokumen ini bermanfaat?

Apakah konten ini tidak pantas?

Hak Cipta:

Format Tersedia

Comsol Alex Course Program Book

Diunggah oleh

Hak Cipta:

Format Tersedia

Finite Element

Finite Element Modeling Using COMSOL Multiphysics

Introduction to COMSOL Multiphysics

Tutorial session: Getting started with COMSOL

Background on Finite Element Analysis

Solver Sequence and settings

Results and Post Processing

Advanced Structural Mechanics examples

Welcome Letter & Trial License Details

The Program book including step-by-step tutorials for the minicourse

Your Trial License

You will find an installation DVD of COMSOL Multiphysics 4.3.

We wish you an informative and great Workshop!

Tutorial 1: Parallel Plate Capacitor

Tutorial 2: Thermal cube

Tutorial 3: Stresses in a Wrench

Tutorial 4: Electrical Heating and

Tutorial 1: Parallel Plate Capacitor

Electrostatics Active only in Air Domain

Pick Electrostatics in Model Builder

Set Boundary Condition Ground

Set Boundary Cond Electric Potential

Mesh Free Triangular (Extra Fine)

Tutorial 2: Thermal cube

Choose Physics and Study

Choose File > New

Set Boundary Conditions

Right Click on Heat Transfer

Right Click on Heat Transfer

Mesh and Solve

Right Click on Mesh

Right Click on Study

Add Structural Equations

Right Click on model 1 node

Add Thermal Expansion

Constrain Bottom Surface + Solve

Tutorial 3: Stresses in a Wrench

1) Set Dimensions, Physics & Study-Type

Right click on Geometry

Display Stress on Deformed Geometry

Right click on Surface 1

Tutorial 7: Electrical Heating

Your first model

Your first training model: A Copper Busbar

Boundary Conditions: Electric Currents

Electric Potential 10mV

Boundary Conditions: Heat Transfer

Cooling with a heat transfer coefficient

Now youre ready for your own simulations!

Capture the Concept

An Example: Electrical Heating in a Busbar

An Example: Electrical Heating in a Busbar

running a first simulation

Titanium Bolt 2b Titanium Bolt 1

This presents a classic example of multiphysics, since the electric

An Example: Electrical Heating in a Busbar

In the Add Physics step, click the Heat

An Example: Electrical Heating in a Busbar

The last Model Wizard step is to select

An Example: Electrical Heating in a Busbar

Right-click Global Definitions and

An Example: Electrical Heating in a Busbar

the Build All button

to rerun the geometry sequence. You should get a