MagNet - Introduction
MagNet - Introduction
Introduction to
Magnetostatic 2D
Modeling
Software Version 2212
December 2022
Unpublished work. © 2022 Siemens
This material contains trade secrets or otherwise confidential information owned by Siemens Industry Software, Inc., its
subsidiaries or its affiliates (collectively, "Siemens"), or its licensors. Access to and use of this information is strictly limited as
set forth in Customer's applicable agreement with Siemens. This material may not be copied, distributed, or otherwise
disclosed outside of Customer's facilities without the express written permission of Siemens, and may not be used in any way
not expressly authorized by Siemens.
This document is for information and instruction purposes. Siemens reserves the right to make changes in specifications and
other information contained in this publication without prior notice, and the reader should, in all cases, consult Siemens to
determine whether any changes have been made. Siemens disclaims all warranties with respect to this document including,
without limitation, the implied warranties of merchantability, fitness for a particular purpose, and non-infringement of intellectual
property.
The terms and conditions governing the sale and licensing of Siemens products are set forth in written agreements between
Siemens and its customers. Siemens' End User License Agreement may be viewed at:
www.plm.automation.siemens.com/global/en/legal/online-terms/index.html.
No representation or other affirmation of fact contained in this publication shall be deemed to be a warranty or give rise to any
liability of Siemens whatsoever.
TRADEMARKS: The trademarks, logos, and service marks ("Marks") used herein are the property of Siemens or other
parties. No one is permitted to use these Marks without the prior written consent of Siemens or the owner of the Marks, as
applicable. The use herein of third party Marks is not an attempt to indicate Siemens as a source of a product, but is intended
to indicate a product from, or associated with, a particular third party. A list of Siemens' trademarks may be viewed at:
www.plm.automation.siemens.com/global/en/legal/trademarks.html. The registered trademark Linux® is used pursuant to a
sublicense from LMI, the exclusive licensee of Linus Torvalds, owner of the mark on a world-wide basis.
Introduction
Overview
The principal aim of this document is to introduce new users to the power of Simcenter™
MAGNET™ software for solving 2D static magnetic field problems. A tutorial with detailed
instructions takes the first-time user through the most important features of Simcenter
MAGNET. This is followed by a series of case studies illustrating modeling techniques and
introducing further features of the package. The document concludes with an introduction to
advanced features that make Simcenter MAGNET a uniquely powerful tool.
A feature of Simcenter MAGNET is its use of the latest methods of solving the field equations
and calculating quantities such as force and torque. To get reliable results, the user does not
need to be an expert in electromagnetic theory or numerical analysis. Nevertheless, the user
does need to be aware of the factors that govern the accuracy of the solution. One of the aims
of this document is to show how the user can obtain accurate results. In 2D, problems can be
solved very rapidly, so it is usually not necessary to consider the trade-off between speed and
accuracy. In 3D modeling, on the other hand, this is an important consideration.
For the advanced user, Simcenter MAGNET offers facilities for user-defined adjustment of the
model parameters, calculation of further results from the field solution, and control of the
operation of the package with scripts and scripting forms. Simcenter MAGNET can be linked
to other applications through the Windows ActiveX Automation interface; another application
can send commands to Simcenter MAGNET to build and solve models and retrieve solution
results.
Chapters 3 and 4 contain case studies in which Simcenter MAGNET is applied to a variety of
magnetic problems. These can be used in two ways: as reference material, and as a series of
graded exercises for developing skills after completion of the tutorial.
Chapter 5 introduces scripting in Simcenter MAGNET, including the use of Microsoft Excel to
control Simcenter MAGNET. Scripting is available in all versions of Simcenter MAGNET, and
chapter 5 indicates some of the ways of using this powerful feature.
Appendix A contains further information about the magnetic field equations and the solution
methods used in Simcenter MAGNET for 2D problems. Novice users do not require most of
this material, but advanced users may find the additional insight helpful. The discussion of
boundary conditions is relevant to all users, and includes the basis of the Kelvin
transformation technique for open-boundary problems.
Appendix B covers energy, force and inductance calculation. This includes the derivation of
some of the equations used in the case studies, and further information about the methods
used in Simcenter MAGNET.
Revised edition
This revised edition of the document is based on Simcenter MAGNET 7.5, released in May
2014. A chapter on the use of the Stack Calculator, which was included in earlier editions, has
been omitted because this feature has been removed from Simcenter MAGNET. The previous
revision introduced the Kelvin transformation technique for handling some open-boundary
problems; this is not yet part of the standard release of Simcenter MAGNET, but a manual
implementation is described in chapter 3, and a scripting example is given in chapter 5.
Translational geometry
Translational geometry means that the object has a constant cross-sectional shape generated
by translation – moving the shape in a fixed direction. The diagram below shows a C-core
formed in this way.
With translational geometry, any slice perpendicular to the axis has the same shape. Rotating
electrical machines can often be represented in this way, and so can many other devices such
as transformers and actuators. Inevitably this 2D approximation neglects fringing and leakage
fields in the third dimension, so the model must be used with caution. The shape is usually
drawn in the XY plane, with the z-axis as the axis of translation.
Rotational geometry
Rotational geometry means that the object has a shape formed by rotation about an axis, like
turning on a lathe. The diagram below shows an object formed in this way from the same
basic C shape used in the diagram above.
Objects with rotational geometry are usually described in cylindrical polar coordinates, with the
z-axis as the axis of rotation. The rotated shape is then defined in an RZ plane, which makes
an angle with the 3D X axis. This geometry differs from translational geometry in two
important respects. First, it is a true representation of a real 3D object, so highly accurate
solutions are possible. Secondly, there are different equations to be solved, and different
methods required for calculating quantities such as force and inductance. For all built-in
calculations Simcenter MAGNET handles these differences automatically, but the user needs
to be aware of the difference when interpreting flux plots for models with rotational geometry.
Magnetic concepts
Simcenter MAGNET can be used to model practical devices without knowing anything about
the differential equations of electromagnetism or the numerical methods used to solve them.
This section reviews some basic magnetic concepts that are required for making effective use
of Simcenter MAGNET; more advanced topics are covered in appendix A. The system of units
used is the SI or MKSA system, although other systems will be mentioned in the context of
magnetic materials.
Although the magnetic field is an abstract concept, the effects of B are concrete and physical.
The force in a device such as this electromagnet can be expressed in terms of B. In simplified
terms, the flux lines can be treated as elastic bands pulling the bar towards the magnet with a
tensile stress (force per unit area) given by B2 / 20. In this expression, B is the magnitude of
the vector B, and 0 = 4 10–7 is a fundamental constant. The unit of B is the tesla (T), and
the unit of 0 is the henry per meter (H/m).
f qu B (1-1)
If the moving charge is an electric current flowing in a conductor, then equation 1-1 leads to
the familiar expression f = Bli for the force on a conductor of length l carrying a current i. If the
conductor itself is moving with velocity u, then the Lorentz force causes a displacement of
charge in the conductor, leading to the expression e = Blu for the induced voltage.
Frequently it is not the flux density B that is required, but the magnetic flux and the flux
linkage . Flux is defined as = BA when the flux density B is constant and perpendicular to a
surface of area A. If the field is not constant or perpendicular to the surface then the flux is
given by an integral, but the principle is the same. Flux linkage is the sum of the fluxes for all
the turns of a coil; this is = N for a coil of N turns where each turn links a flux . The
concept of flux gains its value from Faraday’s law of electromagnetic induction, which states
that the voltage induced in a coil is e = d / dt. If the flux linkage results from current flow,
either in the same coil or in a different coil, this leads to the definition of inductance as flux
linkage per ampere. The calculation of inductance is discussed in appendix B.
Magnetic intensity H
Electric currents give rise to magnetic fields. The currents may flow in conductors or coils, or
they may take the form of electron spin currents in the atoms of a magnetic material. In either
case the problem is to define the relationship between the magnetic field described by B, and
the currents which are the source of the field. In seeking a mathematical form for this
relationship that can be used to solve practical problems, it is useful to introduce a new
magnetic quantity H, which is related both to B and to the currents that are the source of B.
For a magnetic field in free space, set up by currents flowing in conductors, H is defined
through the equation B 0 H . The relationship between H and the currents is then given by
Ampère’s circuital law:
H.dl i (1-2)
where the integral on the left is taken round a closed path, and the summation on the right is
the sum of all the currents enclosed by the path. This equation makes it easy to calculate the
field of a simple system such as a long straight conductor or a toroidal coil, and it is the basis
of the magnetic circuit concept, which is widely used for approximate calculations in
electromagnetic devices. In its differential form it leads to general methods that are applicable
to any problem; this point is expanded in appendix A.
The quantity H is known as the magnetic intensity; from equation 1-2 it has units of amperes
per meter (A/m). For magnetic fields in free space, there would be little advantage in using H;
equation 1-2 could be expressed in terms of B and 0. When magnetic materials are present,
however, the situation is completely different.
The behavior of magnetic materials can be described by modifying the relationship between B
and H. We may put:
B 0 (H M ) (1-3)
where H is the magnetic intensity given by equation 1-2, and M is an induced magnetization in
the material which depends on H. Thus H can be regarded as the cause, which is related to
currents in conductors; B is the effect, giving rise to forces and induced voltages.
From the point of view of the device designer, the magnetization M is unimportant; what
matters is the relationship between H and the resulting B. This relationship can be extremely
complex; the vectors may not be in the same direction, and the present value of B may
depend on the past history as well as the present value of H. For many practical purposes,
however, these complexities can be ignored and the properties of the material expressed by a
simple curve relating B to H. This is the B-H curve or magnetization characteristic of the
material. A typical example is the curve for transformer steel shown below. This curve has
three distinct regions: the steep initial part of the curve, where a small increase in H produces
a large increase in B; the knee of the curve; and the saturated region beyond the knee, where
a large increase in H is required for any perceptible increase in B. With this material, a flux
density of about 1.4 T marks the onset of saturation. Simcenter MAGNET provides B-H curves
for a wide range of magnetic materials, and also enables the user to create new material
curves.
1.8
1.6
1.4
Flux density B (T)
1.2
1
0.8
0.6
0.4
0.2
0
0 2000 4000 6000
Magnetic intensity H (A/m)
B 0 r H (1-4)
where r is a dimensionless property of the material known as the relative permeability. A non-
magnetic material has a relative permeability of 1. For a material such as transformer steel, r
is not constant but varies with B or H; it has an initial value of several thousand, but may fall
below 100 in the saturation region. To simplify the study of some devices, and to speed up the
solution, it is useful to have a fictitious material with a constant permeability corresponding to a
linear relationship between B and H. Simcenter MAGNET provides several linear materials,
with relative permeability values ranging from 10 to 106.
Permanent magnets
Permanent magnets have the property that some magnetization remains in the material when
the external field is removed. For these materials the important part of the B-H curve is the
second quadrant, known as the demagnetization characteristic, shown below for a ceramic
ferrite.
0.4
Permanent-magnet materials have two distinctive parameters: the remanence Br and the
coercivity Hc, which are defined as follows. Br is the value of B remaining in the material when
the applied H is zero. Hc is the negative value of H that must be applied to reduce B to zero.
For the ceramic ferrite characteristic shown above, the value of Br is 0.4 T, and the value of Hc
is –270 kA/m. An additional parameter is the recoil permeability, which specifies the behavior
when the magnetic conditions change. With some materials, if the magnitude of the negative
(or demagnetizing) H is increased and then reduced, the operating point does not retrace the
original curve, but instead it follows a recoil line with a smaller slope. The recoil permeability is
the slope of this line divided by 0; it is thus a relative permeability, and its value is often close
to 1.
Some permanent-magnet materials such as samarium cobalt have a B-H characteristic which
is virtually a straight line from (0, Br) to (–Hc, 0), and the recoil line has the same slope. These
materials are treated as linear, and are specified by the values of Br and Hc instead of a B-H
curve.
It is generally best to begin with a very simple model that preserves the essential features of
the device. Shapes and dimensions can be simplified. Some parts do not need to be modeled
at all. For example, real coils will have non-magnetic insulation separating the coil from the
steel core of the device. There is no need to model such insulation; the coils can be drawn
touching the steel without any significant error. The case studies in chapters 3 and 4 give
some indication of what can be done with simple models.
For the first solution of a new model, it is desirable to get a flux plot as quickly as possible,
because the flux plot is an effective tool for revealing errors in the structure of the model. At
this stage, there is no need to use the powerful adaption feature of Simcenter MAGNET to
improve the solution accuracy, or to take other measures to control the size of the finite-
element mesh.
With any numerical method, perfect accuracy is unattainable. Even with full use of the options
for improving the accuracy, the solution generated by Simcenter MAGNET will contain errors.
In most cases these errors will be insignificant, and are likely to be smaller than the changes
caused by manufacturing tolerances or variations in the magnetic properties of the materials.
Calculated values for forces and torques are particularly sensitive to errors in the field solution,
so these values are likely to change significantly as the solution accuracy is improved. If these
are the quantities of interest in the device, then it is sensible to continue refining until the
values appear to have converged. If it is known that some torque values or force components
should be zero, then refinement should continue until the values are small in comparison with
the useful values. Similarly, where quantities are expected to be equal in magnitude, the
difference should be a small fraction of the mean magnitude.
With certain types of problem, the automatic method of refining the mesh may not yield an
accurate solution in Simcenter MAGNET. A typical example is the calculation of force or
torque in a device where the active airgap is very small in comparison with the dimensions of
the iron parts. Here the values may not converge towards a limit as the refinement level is
increased. Cases like this require the user to take control of the mesh structure, which can be
done by specifying a maximum element size in some parts of the model.
Getting help
Simcenter MAGNET is a powerful and complex package with many features that are not
covered in this introductory document. Although care has been taken to make the instructions
in the document clear and accurate, there may be occasions when the user is in difficulty. The
first point of assistance is the comprehensive help facility in Simcenter MAGNET, which gives
detailed explanations of the features and instructions for their use.
This is an example of a device that can be represented quite well by a 2D model with
translational geometry, even though it is not very long in the direction of translation. The
important leakage field between the poles is accurately modeled, and the part of the fringing
field that is neglected in a 2D model has only a minor effect on the calculation of force.
Fringing and leakage are discussed later, on pages 31 and 70. An important limitation of the
model, however, is that the calculated inductance is very inaccurate because the leakage field
in the third dimension has been neglected. If an accurate value for the inductance is required,
a 3D model must be used. Results from 2D and 3D models are compared on page 47.
Brief description
The diagram below shows the cross-section of the electromagnet with the dimensions in
millimeters. Each side of the coil is a square of side 20 mm, and the core is 10 mm thick
throughout. The armature and the core each have a depth of 40 mm, perpendicular to the
plane of the drawing. The coil has 1000 turns, and the current is initially 2.0 A.
40 5 10
20 20 10
Armature
20 10 20 40
Coil side 1 Coil side 2
Core 10
Since the electromagnet is surrounded by air, the magnetic field theoretically extends to
infinity, making it an open boundary problem. In practice the field decays rapidly with distance,
and is insignificant at a distance of about 10 times the magnet dimensions.
A technique is available in Simcenter MAGNET for exact modeling of an open boundary, and
for some types of problem this is the best approach: an example is given in chapter 3.
However, for many devices such as this electromagnet, good results can be obtained with the
simpler technique of specifying an outer boundary at a distance of 5 to 10 times the magnet
dimensions.
1. Examine the Simcenter MAGNET Main window, and identify the parts listed below.
The Project bar displays information about the model, with tabs at the top labeled
Object, Material, etc. Initially, the Object page is displayed.
The View area is the work area where the model is constructed and the results
viewed. Initially, the View 1 window is displayed.
Between the Project bar and the View area there are vertical toolbars with buttons
for viewing and selecting objects.
At the top of the Main window, there is the usual menu bar, and a row of
horizontal toolbars.
2. Move the pointer over the buttons on the toolbars, pausing on each for the “tooltip”
message that describes the action of the button.
Toolbar buttons are duplicated on the menus. For example, the View menu gives
access to the same viewing tools as the buttons on the vertical View toolbar.
Other toolbars and buttons can be added, by selecting Customize Toolbars from
the Tools menu.
Initial settings
1. On the File menu, click Save. Alternatively, click the Save button.
Select Millimeters.
5. Click OK to close the dialog.
6. On the View menu, click Set Construction Grid to display a dialog:
1. Right click in the View 1 window, and select Properties to display a dialog:
2. In the Background Style drop-down list, select System Window Solid Color.
3. If this background is preferred, Click Set as Default, and click OK.
If you go too far, roll the scroll wheel upwards to zoom in.
To restore the original display, double-click in the View 1 window. Alternatively,
on the View menu, click View All.
4. If there is no scroll wheel, proceed as follows to add a Dynamic Zoom button to the View
toolbar:
Repeat as required until the whole grid is visible. If you go too far, drag upwards
to zoom in.
To restore the original display, double-click in the View 1 window. Alternatively,
on the View menu, click View All.
In Simcenter MAGNET, all drawing takes place on a 2D construction slice. For 2D models this
is just the XY plane, but for 3D models the construction slice can be moved to other planes.
Move the pointer. Observe the coordinates displayed on the Status bar at the
bottom of the screen.
Click near the grid point at (–40, –20).
Move the pointer near (0, –20) and click.
Move the pointer near (0, –10) and click.
Continue in this way for the rest of the core.
Do not draw the lines for the coil sides or the armature at this stage. See below for the
method of deleting any lines drawn in the wrong place.
40 5 10
20 20 10
Armature
20 10 20 40
Coil side 1 Coil side 2
Core 10
P
If the last line to be drawn was in the wrong place, undo this action with
Ctrl+Z, or click the Undo button. For other lines, proceed as follows.
Double-click the last point, or press Esc, to stop line drawing.
Click the Select Construction Slice Lines/Arcs button.
Click the line that you want to delete.
The selected line should turn red.
When you type CR the display will change to the name of the required material.
Alternatively, scroll down through the list and select the material.
6. Change the Distance to 40.
7. Click OK.
A component named Core should be shown in the Object page of the Project bar.
40 5 10
20 20 10
Armature
20 10 20 40
Coil side 1 Coil side 2
Core 10
In the Object page of the Project bar, click the model name.
Air box
An outer boundary must be added to the model by creating a new component called an air
box, which encloses all the other components. The default boundary condition for the air box
is Flux Tangential (see appendix A), which means that the outer boundary is a flux line. This is
a reasonable approximation if the boundary is sufficiently far away from the device.
Since the air box is much larger than the electromagnet, it is not convenient to draw it with the
mouse. Instead, coordinates are entered with the keyboard.
If the air box is made in the same way as the other components, it will contain holes
corresponding to the shapes of those components. This is undesirable, because it will cause
problems later when the model is modified. To prevent holes being formed, the construction
slice lines for the other components will be removed, as described in step 3 below.
If the coordinates are not shown as (0, 0), edit the text.
7. Press Enter, or click the Enter button.
Nothing will change in the View 1 window, so it looks as though nothing has happened.
However, the display next to the Enter button should have changed from (x, y) to (0, 0)
mm, and the coordinate display on the Status bar should show X:0 Y:0 Xg:0 Yg:0 Zg:0.
8. Change the coordinates in the text box to (400, 0), and then press Enter, or click the
Enter button. The brackets and the comma can be omitted.
The display next to the Enter button should show (400, 0) mm, and the coordinate
display on the Status bar should show X:400 Y:0 Xg:400 Yg:0 Zg:0. A circle of radius
400 mm should be shown in the View 1 window.
9. Click the Select Construction Slice Surfaces button.
10. Click inside the circle.
11. Click the Make Component in a Line button to make the air box.
1. On the View menu, click Examine Model Dynamically, or click the Examine
Model button.
On the View menu, click Preset Views / Positive Z axis, or click the
Show XY (+Z) button.
3. To adjust the size of the displayed view, zoom in or out:
The accuracy will be higher with a fine mesh or a high-order polynomial. By default, a first-
order polynomial is used, which is fast but not very accurate.
Initial solution
1. Right-click in the View 1 window and select Initial 2D mesh. Alternatively, on the View
menu, click Initial 2D Mesh.
This should show the default mesh that Simcenter MAGNET uses to solve the field
equations.
2. On the Solve menu, click Static 2D.
The Solver Progress dialog should appear briefly. When the solution is complete, the
Results window should be displayed instead of the View 1 window.
3. Return to the View 1 window by clicking the View 1 tab at the bottom of the View area.
The Polynomial Order is shown as Default. For this model, the default is order 1.
Right-click in the View 1 window and select Initial 2D mesh. Alternatively, in the
View menu, click Initial 2D Mesh.
This should show the original mesh.
Each of the available fields can be viewed as a smoothed field. Smoothing applies averaging
to remove the discontinuities that are present in the computed field. A smoothed field is
continuous across mesh element boundaries where the materials are the same. Smoothed
fields are not continuous across element boundaries where the materials differ (true
discontinuity exists at these boundaries).
Move the mouse pointer anywhere in the model region, but do not click.
The Status bar should show two values at the left-hand side: the flux function, and
|B| smoothed.
Ignore the flux function. The other number is the flux density magnitude at the position of
the pointer.
Move the pointer without clicking, and observe the change on the Status bar.
3. Hold the mouse still, with the pointer anywhere in the model region, and click once.
A window should open at the bottom of the Main window, called the Text Output
bar.
This displays the coordinates of the point, and values of the flux function and |B|
smoothed.
Every click in the model region displays a new set of values.
4. Close the Text Output bar by clicking the Close box on the left-hand side of the bar, or
clicking Text Output Bar on the Tools menu.
Notice two features of the magnetic field in the electromagnet. First, the magnetic field in the
airgap between the poles and the armature is not confined to the pole region, but spreads into
the surrounding air; this is termed fringing. Secondly, some flux takes a short cut across the
space between the poles, instead of crossing the airgap and passing through the armature;
this is termed leakage.
Field Extractor
It is instructive to display graphs of the flux density components Bx and By along a line passing
through the airgap between the armature and the poles, with end-point coordinates (2.5, –30)
and (2.5, 30). This will show the variation of the flux density in each airgap, and the rapid
decay beyond the edges of the poles.
5.
For X, entering only a Start value automatically sets the number of sampling points to 1.
Since the Y coordinate range is 60 mm, setting the number of sampling points to 61 will
sample the field at increments of 1 mm.
The display precision can be adjusted with a control at the top of the Results window; the
default precision is 3. For the results given below, the precision was set to 4.
Energy
By default, the Results window should display the Energy page. If this is not visible, click the
Energy tab. The displayed values should be similar to the following:
Coil#1 0.3209 Wb
The self-inductance of the coil may be calculated from the flux linkage:
λ 0.3209
L= = = 0.1605 H = 160.5 mH
i 2.0
Alternatively, the self-inductance may be calculated from the stored magnetic energy:
2W 2 × 0.3202
L= = = 0.1601 H = 160.1 mH
i2 (2.0)2
See appendix B for details of these methods of calculating inductance.
With some models, the self-inductance can be calculated automatically with one of the
standard extensions to Simcenter MAGNET: the RLC Matrix Calculator (see page 74).
However, this method is valid only for models constructed from linear magnetic materials, so
that the inductance values are independent of current. It is not applicable to the C-core
electromagnet, where the inductance falls with increasing current.
The force components and magnitudes, in newtons, should be similar to the following:
Body X Y Z Magnitud
e
Parameterization
1. In the Object page of the Project bar, right click the Coil#1 object, and select Properties
to display a dialog:
3. Edit the Expression field for the Current parameter so that it contains the following list of
values:
2, 4, 6, 8, 10
Observe that five problems have been created, each with a different coil current.
If Simcenter MAGNET is licensed for parameterization, all of the problems will
be marked for solution, otherwise only the first problem will be marked.
8. On the Solve menu, click Static 2D.
Post-processing – 1
The instructions below are applicable when Simcenter MAGNET is licensed for
parameterization. For the Trial Edition, only one problem ID will be visible in any of the lists,
and it will not be possible to display animations or graphs of results.
Click the Magnetic Permeability tab and resize the dialog box to display the B-H
characteristic:
For the Armature, click in the text box for the Force Magnitude.
The Graph Selection… button should be enabled.
Note that the current values are not displayed in this graph, because Simcenter
MAGNET uses this method to show the variation of force with any parameter. The
current values corresponding to the problem numbers 1 to 5 are 2, 4, 6, 8 and 10 A
respectively.
6. The force values might be expected to increase as the square of the current. Observe
that the magnitude of the force does not increase in this way, because the core is
saturating at high current values.
7. Close the graph window.
This change of material has made a small improvement. For significant improvement, the
shape of the core must be changed.
5 10
20
20 20 10
20 20 20 40
10
20
To make this change, the original model will be modified with the shift and distort facilities in
Simcenter MAGNET. Before doing this, the original drawing lines will be deleted as described
below.
In the Shift Vector text box, change the text to (–10, 0, 0).
Click OK to apply this shift and close the dialog.
The left-hand coil side should have moved 10 mm to the left.
On the Model menu, click Distort Vertices, or click the Distort Vertices button.
The model vertices are marked in the View 1 window:
V1
V2
Move the pointer to the required position, displaced from the original position by
10 mm in both x and y. Do not press the mouse button while doing this.
“Rubber band” lines follow the pointer.
V1
V2
On the View menu, click Preset Views / Negative Z axis, or click the Show XY (–
Z) button.
This shows the back view of the model (see the next page):
V1
V3
V4
V2
In the absence of saturation, the force would increase by 9 times when the current increases
from 2 A to 6 A, and by 25 times when the current increases from 2 A to 10 A. With the
modified model, the force increase is 8.8 times at 6 A, and 17.6 times at 10 A. Thus, the
change has almost eliminated saturation for currents of up to 6A. There is considerable
improvement in the performance at 10 A, but saturation is still significant at this current.
In the Model Properties dialog, there is a Parameters page where the user can create a new
named parameter and give it a range of values. The new parameter can be used to modify
other properties of model components, so that they depend on one parameter. In this case, a
user-defined parameter will be used to vary the Y coordinate of a shift vector that determines
the displacement of the armature.
It is necessary to remove the list of values from the coil current parameter, as described in the
instructions on the next page; otherwise Simcenter MAGNET will generate a new problem for
each combination of values for the current parameter and the shift parameter.
2. Select the Parameters page and scroll down to the end of the list of parameters:
Observe that five problems have been created, with different shift vectors.
Shift (mm) 0 1 2 3 4 5
–Fx (N) 25.26 24.95 24.53 23.78 22.82 21.73
–Fy (N) 0.00 0.91 1.70 2.50 3.18 3.83
2D 3D
These results show that the 2D model predicts the force of attraction with good accuracy, but
it seriously under-estimates the flux linkage and hence the inductance of the coil. These
results may be explained by considering the 3D model shown below.
In the middle of the airgap, the magnitude of the flux density is approximately 0.249 T. If the
field were confined to the area of the pole face, with a uniform value of 0.249 T, equation B-13
of Appendix B gives the force of attraction as
B2 A
fn 19.74 N
20
The 2D result of 24.15 N is 22.3% higher than the idealized value. This increase can be
attributed to the contribution of the fringing flux in the region of the long edges of each pole
face, visible in the flux plot on page 29, which increases the effective area of the pole face. In
the 3D model, there is a further small contribution from the fringing flux in the region of the
short edges, giving a force value that is 1.6% higher than the 2D result.
The flux linkage with the coil, on the other hand, is increased significantly by two effects. The
leakage field between the poles will extend well beyond the device in the third dimension; this
will increase the flux in the yoke, giving a corresponding increase in the flux linkage with the
coil. Additionally, the curved ends of the coil will generate a magnetic field which is ignored in
the 2D solution, and this will further increase the flux linkage. Consequently, the 3D solution
gives an inductance value that is 53% higher than the 2D result. There is a similar increase in
the stored magnetic energy.
Case Studies:
Translational Geometry
Introduction
The case studies in this chapter cover a range of modeling problems for devices with
translational geometry. Devices with rotational geometry are discussed in chapter 4. These
case studies are arranged in order of increasing difficulty, progressively introducing further
features of Simcenter MAGNET, so it is advisable to work through them in sequence. The
detailed descriptions of basic Simcenter MAGNET operations given in chapter 2 will not be
repeated, but any new operations will be fully explained.
In some of the case studies, a 2D model does not always give accurate results because the
device is not very long in the translational direction. Since the user needs to be aware of the
limitations of 2D modeling, these case studies also include 3D results for comparison.
For all of the case studies in chapter 3 and chapter 4, the instructions assume that a new
model is being started, as described in the tutorial in chapter 2. To avoid tedious repetition,
this instruction is given in abbreviated form in the case studies.
Some of the case studies require shapes to be drawn from arcs and straight lines. Drawing an
arc requires the user to specify the coordinates of the center, the start point and the end point.
Arcs are normally drawn counter-clockwise from the start point to the end point, but the
direction can be changed to clockwise by holding down the Ctrl key. If the arc drawing tool is
selected from the Draw toolbar, the order of the points is center, start, end.
The Draw menu has other arc tools with the points in different orders.
E-core electromagnet
The diagram below shows an E-core electromagnet. It is similar in principle to the C-core
electromagnet of chapter 2, but it is a better magnetic design because the coil is nearer to the
airgap, and both sides of the coil are active. The objectives are to determine the self-
inductance of the coil and the force on the armature, and to explore the magnetic field
distribution in the device.
6x
x
g x x 2x x x
3x 3x
4x
The dimensions in this diagram are chosen so that the armature and the core can be made
from laminations punched from sheet steel without any waste, as shown in the diagram below.
The basic dimension x is 10 mm, the airgap length g is 5 mm, and the depth of the
electromagnet is 60 mm. The coil has 1000 turns, carrying a current of 2.0 A.
60
10
5 10 10 20 10 10
P
30 30
40
As with the C-core electromagnet, an air box must be drawn around the model with a radius of
about 10 times the model dimensions.
1. Set the options for solving, using the Solve menu, with the following changes from the
default values:
Solver Options:
Polynomial order 2.
Adaption Options:
Use h-adaption,
Tolerance 0.2%.
2. Solve as Static 2D.
3. Inspect the contour plot of the flux function and the shaded plot of the |B| values.
4. Inspect the computed global quantities, and calculate self-inductance values as follows
(see appendix B for details):
From the flux linkage: L = / i, where is the flux linkage for the coil, and i is
the coil current.
From the stored energy: L = 2W / i2, where W is the stored magnetic energy.
2D 3D
Discussion
2D model
Since the airgap length and the coil ampere-turns are the same as for the C-core
electromagnet, the airgap flux density values are expected to be similar. The total pole-face
area is three times as great, so the force of attraction and the self-inductance are also
expected to be about three times as great, which is the case. If the coil ends are neglected,
the coil resistance will be twice that of the C-core electromagnet, because the coil cross-
sectional area is halved. Thus, the E-core electromagnet has a lower power loss per unit
force, and a larger value of time constant L / R, than the C-core electromagnet.
3D model
As with the C-core electromagnet, the 3D solution gives a significantly higher value for the
self-inductance: in this case 19% higher than the 2D result. Similarly, the 3D solution gives a
higher value for the force: 5.2% higher than the 2D result for the force on the armature. For
comparison, with the C-core electromagnet the 3D solution gives an inductance value that is
53% higher than the 2D result, and a force value that is 1.6% higher.
40 5 10
10
20 P
20
20
10
As with the C-core electromagnet, an air box must be drawn around the model with a radius of
about 10 times the model dimensions.
1. Set the options for solving, using the Solve menu, with the following changes from the
default values:
Solver Options:
Polynomial order 2,
Newton Tolerance 0.5%.
Adaption Options:
Use h-adaption,
Tolerance 0.1%.
2. Solve as Static 2D.
3. Inspect the contour plot of the flux function and the shaded plot of the |B| values.
4. Inspect the computed global quantities.
2D 3D
Discussion
Energy
The stored magnetic energy values are negative in this model because B and H are in
opposite directions in the permanent-magnet material, so the energy density calculated from
equation B-1 (Appendix B) is negative in this part of the model. When the total energy is
calculated by integrating the energy density over the volume of the model, the negative energy
in the permanent-magnet material exceeds the positive energy in the rest of the model. See
the Simcenter MAGNET help for further information.
2D model
The airgap flux density is somewhat higher than in the C-core electromagnet, giving an
increased force of attraction, and the maximum flux density in the steel is also higher.
Because the permanent magnet has a recoil relative permeability close to 1, there is much
more leakage flux at the back of the magnet than with the coil in the C-core electromagnet.
3D model
In this 3D model the force on the armature is 16% lower than the value from the 2D model. In
contrast, with the C-core electromagnet, the force from a 3D model is 1.6% larger. The
reduction in force can be explained as follows. There is significant flux leakage between the
poles at the ends, which is ignored in the 2D model. Thus, more of the magnet flux is diverted
into leakage paths, so that less flux crosses the airgap, giving a reduction in the force of
attraction. This effect is greater if the width of the permanent magnet is reduced.
There is a simple analytical expression for the force per unit length between two infinitely long
busbars in free space:
2 10 7 ki1i2
f [N/m] (3-1)
d
where i1 and i2 are the currents in the bars, k is a constant that depends on the shape of the
conductors and their separation, and d is the distance between centers of the busbars. This is
based on the formula given by Bewley [1] and Steele [2]. The value of k in this case study can
be calculated from the analytical formula to be 0.94980 when d = 0.3 m, so the theoretical
force is 6.3320 kN/m; this serves as a check on the results from Simcenter MAGNET.
0.2 0.2
Bar#1 Bar#2
The open boundary representation has two parts: (a) a circular boundary surrounding the
model, enclosing an air space; (b) a second small circular boundary, enclosing another air
space, outside the first. Corresponding points on the two boundaries are linked so that the
field values will be identical, by applying an even periodic boundary condition.
Number of turns: 1.
Current: 1.0e5 A.
6. Select and delete all the construction-slice lines.
Click Scale Factor and enter the value 0.1 in the text box.
Click Shift Vector and edit the text box: (0.4, 0.4, 0).
Click Apply the Transformation to a Copy of the Selection.
Click OK.
A new air box should appear in the View 1 window, as shown below, and there should
be a new component in the Object page.
4. In the Object page, change the name of the new component to Exterior.
5. Display the tree directory for the Airspace component.
6. Hold down the Ctrl key, and select Face#3 and Face#4.
These are the half-cylinders which form the curved surface of the Airspace component,
as can be seen by rotating the view of the model.
Click Scale Factor and enter the value 0.1 in the text box.
Click Shift Vector and edit the text box: (0.4, 0.4, 0).
These settings must be identical to those used in the Transform Components dialog to
create the Exterior component.
Solver Options:
Polynomial order 2,
CG Tolerance 0.001%.
Adaption Options:
Use h-adaption,
Tolerance 0.01%.
2. Solve as Static 2D.
3. Inspect the resulting solution mesh as well as the force values.
4. Compare the force with the theoretical value of 6.332 kN.
BoundaryCondition#1 (EP),
Exterior.
3. Right-click Coil#1 and select Properties to display the Coil#1 Properties dialog:
4. Scroll down and click Reverse Direction for CoilSide#2. Click OK.
5. Display the model Properties dialog. In the Parameters page, define a new numeric
parameter BoxRadius with values:
0.4, 1, 2, 4
6. On the vertical View toolbar, click the Examine Model button.
7. In the View 1 window, drag the pointer to rotate the model so that the front and
back faces of the air box are visible:
10. In the Vertex page, note the sign in the first Local Position value, e.g. (–0.4, 0)
11. Open the Parameters page.
12. In the Expression field for the Position parameter, enter an expression such as:
[–%BoxRadius, 0] with the same sign as in the first Local Position value.
13. Click Apply.
14. Click the next vertex and repeat the procedure. Do this for all four vertices.
Solver Options:
Polynomial order 2,
CG Tolerance 0.001%.
Adaption Options:
Use h-adaption,
Tolerance 0.01%.
Sample results
The results below were obtained with Simcenter MAGNET version 7.5.
0.4 m 6803.2
N
1m 6344.0
N
2m 6334.4
N
4m 6332.3
N
Discussion
The analytical value for the force is 6332.0 N. With currents in opposite directions, the open
boundary method gives a result very close to this value. With currents in the same direction,
where a closed boundary is required, the force is over-estimated by 7.4% when the air box
radius is 0.4 m. An air box radius of 2 m, which is about 10 times the radius of the model, is
sufficient to give high accuracy in the force calculation. To achieve this accuracy, the default
CG tolerance has been reduced to 0.001%, and the adaption tolerance has been set to a very
low value of 0.01%. Nevertheless, this 2D model solves rapidly because there are no
magnetic materials present.
If a cylindrical hole is bored in the conductor from one end to the other, there will be no
magnetic field in this hole, provided the hole is coaxial with the conductor. In other words,
there is no field inside a hollow cylindrical conductor.
A remarkable result, which is not widely known, is the nature of the field in the hole when it is
bored off-center, so that its axis is displaced from the axis of the cylinder, as shown in the
diagram below. This field is perfectly uniform; the flux lines are parallel straight lines in a
direction normal to the plane containing the two axes [3].
Number of turns: 1.
Current: 69000 A.
6. Construct an air box from a circle centered at the origin, radius 500 mm:
Solver Options:
Polynomial order 2,
CG Tolerance 0.0001%.
Adaption Options:
Use h-adaption,
Tolerance 0.001%.
2. Solve as Static 2D.
3. Inspect the contour plot of the flux function and the shaded plot of |B| smoothed.
Observe the nature of the flux lines in the hole. Use the Probe Field Values tool to get
values of the flux density at several points in the hole.
4. Use the Field Extractor on the Extensions menu to obtain values for By over a grid of
nine equally spaced points in the hole as follows.
x (mm) 20 40 60
y (mm)
This problem makes an interesting case study for Simcenter MAGNET, firstly because the
results can be compared with an analytical solution if the magnetic material is linear [3], and
secondly because it illustrates a technique for producing a uniform magnetic field. For this
case study, the inner radius of the cylinder is 24 mm, the outer radius is 30 mm, the relative
permeability of the iron is 1000, and the magnitude of the applied field is 0.1 T.
If this coil is surrounded by air, the field takes the form shown below.
As this flux plot shows, the field in the middle of the coil is approximately uniform. It can be
made exactly uniform by a simple modification: the coil is enclosed in a close-fitting air box
with a Field Normal boundary condition on four surfaces. The diagram below shows the
resulting field pattern.
Boundary conditions are discussed in appendix A. This use of the Field Normal condition
effectively embeds the device in a material of infinite permeability. Flux lines have to enter the
top and bottom bounding surface at right angles, and they find return paths round the sides
(not visible in Simcenter MAGNET) of zero reluctance. The result is a perfectly uniform field of
finite extent, which is a useful approximation to the theoretical uniform field of infinite extent.
When an iron cylinder is placed in this field, the result is shown in the flux plot on page 64.
0i
B0 [T] (3-2)
l
where B0 is the flux density in the absence of an iron cylinder, i is the coil current, and l is the
length of each coil side in the direction of the field. The result does not depend on the
separation between the coil sides, or on their thickness.
For the screening problem, we need to generate a uniform field over a region that is
significantly larger than the cylinder. A region 240 mm square should be satisfactory for a
cylinder with an outer diameter of 60 mm. The required current is therefore:
Select the left-hand coil side by clicking CoilSide#1 in the Object page.
On the Model menu, click Shift Components.
Click the check box for making a copy.
Enter the following value for the shift vector:
(234, 0, 0)
Click OK.
Rename the new component CoilSide#2.
7. Make a coil from the two coil sides:
Number of turns: 1.
Current: 19100 A.
Solver Options:
Polynomial order 2,
CG Tolerance 0.01%.
Adaption Options:
Use h-adaption,
Tolerance 0.01%.
2. Solve as Static 2D.
3. Inspect the flux plot and the shaded plot of |B|.
Use the Probe Field Values tool to explore the flux density values inside and
outside the cylinder, and in the material of the cylinder.
Confirm that the value inside the cylindrical cavity is about 1% of the value far
away from the cylinder.
4. Use the Field Extractor to explore Bx and By over a square grid of points, ranging from –
15 mm to +15 mm in both X and Y directions, with a spacing of 5 mm. See page 63 for
the procedure.
D B
A
Creating the model
1. Start a new model and save it as Cylindrical screen 2.
2. Set the model length units to millimeters.
3. The coordinate origin is in the center of the hollow cylinder. Set the display to show a
corresponding range of x and y values with a construction grid. Use a grid spacing of 6
mm to simplify drawing the component outlines.
4. Construct the quarter cylinder from two concentric arcs, of radii 24 mm and 30 mm, as
follows:
Number of turns: 1.
Current: 9550 A (this is half of 19100 A for the full model).
Solver Options:
Polynomial order 2,
CG Tolerance 0.01%.
Adaption Options:
Use h-adaption,
Tolerance 0.01%.
2. Solve as Static 2D.
3. Inspect the flux plot and the shaded plot of |B|.
4. Check that the flux density values are similar to those obtained with the full model of the
cylinder and coil.
Discussion
The theoretical value [3] for the flux density inside the cylindrical cavity is 1.101 mT for the
cylinder used in this case study, when it is placed in a uniform field of 100 mT, of infinite
extent. When averaged over a square grid of 49 points in the hole, the value in the full model
is 1.165 mT. The difference of 5.6% can be attributed to the approximate representation of an
infinite field. In practice this error is unimportant, because the screening effect is strongly
dependent on the relative permeability of the screen, which is likely to be highly variable in
practice. To see this effect, try replacing the linear material MU3 with a non-linear material
such as CR10.
The full equivalent circuit for the transformer is shown below, where x1 and x2 are the primary
and secondary leakage reactances respectively, and Xm is the magnetizing reactance.
jx1 jx2
R1 R2
jXm Rc
These reactances are related to the magnetic field through the inductances:
x1 l1 , x 2 l 2 , X m M (3-4)
where is the angular frequency. Although this appears to be a time-harmonic problem, a
static solution will be sufficient for determining these inductances because transformers are
normally designed to minimize eddy-currents in the core.
When the windings have equal numbers of turns, the leakage inductances l1 and l2 are defined
as follows:
l1 L1 M , l 2 L2 M (3-5)
where M is the mutual inductance, and L1 and L2 are the self-inductances of the two windings.
These leakage inductances represent flux produced by one winding which fails to link with the
second winding, but “leaks” into the surrounding air (or other non-magnetic material) instead.
From appendix B, the mutual inductance is given by
21 12
M (3-6)
i1 i2
where 21 is the flux linkage with winding 2 when winding 1 carries a current i1, and 12 is the
flux linkage with winding 1 when winding 2 carries a current i2.
Experimentally, the reactance parameters are determined from open-circuit and short-circuit
tests. In the open-circuit test, the secondary is open-circuited so that i2 = 0, and the normal
voltage is applied to the primary winding. Measurements at the primary terminals give the
value of the total primary reactance x1 + Xm.
In the short-circuit test, the secondary terminals are short-circuited and a low voltage is
applied to the primary, sufficient to circulate the normal full-load current. The secondary
current is then close to its normal full-load value. Under these conditions the current flowing in
the magnetizing reactance Xm is negligible, so measurements at the primary terminals give the
total leakage reactance x = x1 + x2. It is not possible to determine x1 and x2 separately by this
method, so it is usually assumed that each is equal to half of the total.
To simulate the conditions of a short-circuit test in Simcenter MAGNET, the windings are
supplied with equal and opposite currents. From appendix B, the sum of the leakage
inductances is given by
2Ws
l1 l2 , (3-7)
is2
where is is the current in each winding and Ws is the stored energy.
Each coil has 1000 turns. Under no-load conditions, when only one coil carries current, the
current is 2.0 A. For the simulated short-circuit condition where the coils carry equal and
opposite currents, the current is 200 A, giving a current density of 5 MA/m2 (or 5 A/mm2).
Because there is virtually no external field, the air box can be quite close to the core of the
transformer.
Solver Options:
Polynomial order 2.
Adaption Options:
Use h-adaption,
Tolerance 0.05%.
2. Solve as Static 2D.
This solution is for the no-load condition with the primary energized.
3. Inspect the contour plot of the flux function and the shaded plot of the |B| values.
4. Inspect the computed global quantities and calculate inductances as follows (see
appendix B for details):
Self-inductance: L1 = / i1, where is the flux linkage with the primary, and i1
is the coil current.
Mutual inductance: M = / i1, where is the flux linkage with the secondary.
5. Change the primary current to 0 A and the secondary current to 2.0 A. Solve again and
calculate the inductances:
Self-inductance: L2 = / i2, where is the flux linkage with the secondary, and
i2 is the coil current.
Mutual inductance: M = / i2, where is the flux linkage with the primary.
6. Change the primary current to 200 A and the secondary current to 200 A. Solve again
and calculate the total leakage inductance from equation 3-7
1. On the Extensions menu, click RLC Matrix Calculator to display the Calculator dialog:
The results given by the Calculator should be identical to those found from the
current and flux linkage values.
i1 = 2.0 A, i2 = 0 2D 3D
i1 = 0, i2 = 2.0 A
i1 = 200 A, i2 = 200 A
2D results
The first two solutions give very similar values for self-inductance, mutual inductance and
leakage inductance. At first sight this is surprising because the primary winding occupies the
space between the secondary and the center limb of the core, so the two are not obviously
equivalent. But in a 2D model, if the core were infinitely permeable the current and flux
patterns would have a symmetry that results in equal inductance values. Evidently a
permeability of 1000 is sufficient to give very similar results.
The result from the energy calculation is very close to the sum of the two leakage reactances
computed from the separate flux linkages. This is useful confirmation of the accuracy of the
flux calculation, and shows that it is sometimes possible to obtain a good value for the leakage
reactance by subtraction.
3D results
These results are significantly different from the 2D solution. Both of the leakage inductances
are larger than their 2D counterparts because of the end-winding field, and as expected, the
secondary value is considerably larger than the primary value.
The leakage inductance from the energy calculation is slightly smaller than the sum of the two
leakage reactances computed from the separate flux linkages, which is an indication of
numerical errors in the 3D results.
The motor has a 3-phase stator, with two coils in each phase shown by the colors in the 2D
model. Here, the red phase is energized, pulling one pair of rotor poles into alignment with the
corresponding stator poles. Successively energizing the red, yellow and blue coils will make
the rotor move through successive rotational steps of 30º.
13. Select the stator surface and make the component in a line:
1. Draw the two sides of the top stator coil, as rectangles of 20 mm by 35 mm.
2. Make the two components named CoilSide#1a and CoilSide#1b:
Material: AIR
Sweep distance: 100 mm.
There is negligible flux leakage outside the stator, so there is no need to leave a gap
between the air box and the stator.
Solver Options:
Polynomial order 2.
Adaption Options:
Use h-adaption,
Tolerance 0.1%.
2. Solve as Static 2D.
3. Inspect the contour plot of the flux function and the shaded plot of the |B| values.
The flux plot should resemble the diagram below on the left.
The diagram on the right shows the effect of connecting a coil the wrong way
round.
If the coil connections are wrong, open the Coil page, right-click on a coil side,
and click Reverse Coil Side Direction. Repeat for the other coil side.
4. Examine the effect of switching the coil currents by setting the current in Coil#1 to 0 and
the current in Coil#2 to 2.0 A.
Use the Keyboard Input bar, or set the Construction Grid spacing to 2.5 mm.
3. Select the upper half of the circle.
1. In the Parameters page of the model Properties, define a new numeric parameter Angle
as a number with the list of values: 0, 5, 10, 15, 20, 25, 30.
2. For the Rotor component, display the Parameters page of the Properties dialog and
enter the following expression for the RotationAngle parameter:
%Angle%deg
The rotor will be rotated in the anticlockwise direction for successive values of the
RotorAngle parameter in degrees.
3. Solve as Static 2D.
With the Trial Edition of Simcenter MAGNET, solutions of successive problems can be
obtained by the method described in chapter 2 (page 37).
4. In the Results window, open the Force page and examine the magnitude of the torque
on the rotor for each problem.
Sample results
The values below were obtained for the magnitude of the torque on the rotor, with Simcenter
MAGNET version 7.5. The rotor angle represents the displacement from a position of
alignment with the stator poles.
0 0.012
5 2.112
10 3.142
15 3.621
20 3.706
25 2.845
30 1.714
Discussion
The dimensions of this simple stepper motor have been chosen to make the model easy to
build, with no attempt at an optimal design. In particular, the radial depth of the core is
excessive, with a correspondingly low flux density. The stator and rotor poles are narrow, so
the restoring torque decreases rapidly when the rotor displacement angle exceeds 20º.
Consequently, the torque developed at each switching transition is quite low.
Primary
The primary has a full-pitched 3-phase winding in a laminated steel core, with 6 slots per pole,
and the secondary comprises blocks of permanent-magnet material partially embedded in a
steel backing. Alternating currents in the primary winding will produce a traveling magnetic
field, and at one instant of time the flux plot is shown below for a load angle of 90º.
This model represents two pole pitches of an infinitely long machine. Since the field pattern
repeats every two pole pitches, the field values will be identical at corresponding points on the
two ends. To represent this condition in Simcenter MAGNET, a periodic boundary condition is
specified for the two ends. Because of the symmetry of this device, it is possible to model just
one pole pitch, since the field values repeat with opposite signs at intervals of a pole pitch.
If a field has identical values at corresponding points, an even periodic boundary condition is
used; if it has equal and opposite values, an odd periodic boundary condition is used.
The case study will consider two ways of simulating the machine operation:
For a fixed position of the secondary, the phase angle of the currents will be
varied. This will show how the tractive force varies with load angle, for one
position of the secondary.
Both the phase of the currents and the position of the secondary will be varied, so
as to hold the load angle constant. This will show the cogging effect that will be
present during normal operation.
In this machine, getting accurate results requires control of the mesh density in the airgap.
This is best accomplished by defining two air regions for the airgap, and specifying the
maximum size for elements in these regions. Automatic mesh refinement by adaption is not
used in this model.
20 200 15 20
20 10 20
10
(1) (2) (3) (4) (5) (6)
40
60
10 20 20 20 10
240
To speed up the solution, linear materials will be used for the model: MU3 for the primary and
secondary cores, and PM04 for the permanent magnet. These are reasonably representative
of ordinary steels and ceramic ferrite, respectively.
For the coils, new materials will be defined with the same properties as copper, but with
distinctive red, yellow and blue colors to identify the phases. The color sequence for the slots
will be red, blue, yellow, where the blue slots will carry reversed current from the blue phase.
This gives the normal 60º phase progression between coil groups for an AC machine winding.
The model for this machine will use separate air boxes for the primary and the secondary, with
a common surface in the middle of the airgap. A separate air box enables the secondary to be
moved relative to the primary.
RedPhase: %Magnitude*cos(%Phase%deg).
YellowPhase: %Magnitude*cos((%Phase–120)%deg).
BluePhase: %Magnitude*cos((%Phase+120)%deg).
%deg converts degrees to radians.
4. Construct an air box for the primary from the lower rectangle:
Name: PrimaryAirBox
Material: AIR.
Sweep distance: 100 mm.
5. Similarly, construct an air box for the secondary from the upper rectangle.
6. Draw two more lines in the airgap, across the whole width of the model, touching the
primary and secondary surfaces as shown below.
7. Construct air regions from the resulting narrow rectangles, named PrimaryAirgap and
SecondaryAirgap.
9. Click Maximum Element Size, and in the text box enter 2.5.
10. Similarly, set the maximum element size for the SecondaryAirgap to 2.5 mm.
11. Apply an Odd Periodic boundary condition to the model as follows.
Solver Options:
Polynomial order 2.
Adaption Options:
not used.
2. Solve as Static 2D.
With the Trial Edition of Simcenter MAGNET, solutions of successive problems can be
obtained by the method described in chapter 2 (page 37).
3. Inspect the flux plot and the shaded plot of |B| for each problem.
4. In the Results window, open the Force page and examine the force components on the
two bodies for each problem.
Sample results
The following values were obtained for the X component of force on the secondary, with
Simcenter MAGNET version 7.5:
–30 0 0.0
–15 15 17.96
0 30 34.99
15 45 50.02
30 60 62.18
45 75 70.49
60 90 74.27
75 105 73.01
90 120 66.48
Boundary conditions
1. Open the model LSM 1 and save it as LSM 2.
2. Delete the periodic boundary condition.
3. Display the tree directories for the PrimaryAirBox and the SecondaryAirBox.
4. Hold down the Ctrl key and click the following faces:
PrimaryAirBox Face#5
PrimaryAirBox Face#6
SecondaryAirBox Face#6.
The model view should resemble the following:
0 62.18
5 58.61
10 67.05
15 77.64
20 83.13
25 86.23
30 84.45
35 73.54
40 61.96
45 58.67
50 66.94
55 78.13
60 84.23
65 87.50
70 85.16
75 73.49
80 61.85
85 58.23
90 66.36
Discussion
For the simple linear synchronous motor used in this case study, the technique of shifting the
secondary shows clearly the cogging effect of the primary slot openings on the force
characteristic.
To simplify the modeling, a full-pitched winding has been used, so that each slot carries
current from only one phase. However, it is a straightforward matter to define upper and lower
coil sides in each slot, and it is then possible to model two-layer chorded windings.
The methods of using periodic boundary conditions can be applied to other devices such as
rotating electrical machines, where it is only necessary to model one pole pitch of the physical
device. In spite of their shape, rotating machines must also be modeled using translational
geometry.
Constructing a model with rotational geometry is different from the procedures used in
chapters 2 and 3. The model is constructed as part of a solid of revolution by rotating shapes
about an axis. This axis must be the Y axis of the normal XY drawing plane. Components are
formed by sweeping in an arc instead of sweeping in a line. The subtended angle of the arc is
unimportant for a 2D model, so the Simcenter MAGNET default angle of 90º will be used for
all of the case studies. This construction technique is described in the first case study:
Inductance of a Brooks coil.
Flux plots are not as easy to interpret with rotational geometry as they are with translational
geometry. Consider the case of an infinitely long straight solenoid. The magnetic field is
confined to the interior of the solenoid, where it is perfectly uniform and parallel to the axis.
This device can be modeled in Simcenter MAGNET by setting boundary conditions – see the
case study on a cylindrical screen in a uniform field on page 64. The diagram below shows the
flux plot generated by Simcenter MAGNET, with the axis of the solenoid on the left. This
diagram represents the flux in the right-hand half of a cross-section of the device.
The spacing between the flux lines decreases with increasing distance from the axis. For
users who are familiar with ordinary 2D flux plots for translational geometry in the XY plane,
the plot gives the impression that the flux density increases with distance from the axis. This is
an illusion. If r is the radial distance between two successive flux lines, and B is the axial flux
density, then the increment of flux is:
Brooks coils have a simple shape, and are close to the optimum of a coil that has the largest
inductance for wire of a given length and cross-sectional area. The inductance can be
calculated analytically, so this is another useful test problem for checking the accuracy of the
results produced by Simcenter MAGNET. From the formula in Grover [5] for solenoidal coils,
the inductance is:
A value for the coil current must be specified before the problem can be solved in Simcenter
MAGNET, although a value is not required in the description of the problem. There are no
magnetic materials present in this device, so it is a linear problem in which the flux is directly
proportional to the current. Since the inductance is the flux linkage per unit current, it must be
independent of the current, so any reasonable value of current may be used; a value of 1 A is
chosen arbitrarily.
0.1 0.1
P 0.1
Material: AIR.
The suggested solver settings should give an accurate 2D solution without excessive
computing time. The user is invited to try the effect of different settings.
1. Set the options for solving, using the Solve menu, with the following changes from the
default values:
Adaption Options:
Use h-adaption,
Tolerance 0.01%.
2. Solve as Static 2D.
3. Inspect the contour plot of the flux function and the shaded plot of the |B| values.
4. Inspect the computed global quantities, and calculate self-inductance values as follows
(see appendix B for details):
From the flux linkage: L = / i, where is the flux linkage for the coil, and i is
the coil current.
From the stored energy: L = 2W / i2, where W is the stored magnetic energy.
5. As an alternative, use the RLC Matrix Calculator (see page 74) to determine the self-
inductance of the coil.
Sample results
The results below were obtained with Simcenter MAGNET version 7.5:
Discussion
The computed value of the self-inductance is only 0.1% lower than the theoretical value, which
indicates that the surfaces of the air box have been taken sufficiently far away from the coil. In
this case the radius of the box is 10 times the outer radius of the coil, and the axial length is 20
times the axial length of the coil. The contour plot of the flux function shows no apparent
“squashing” of the flux pattern near the box surfaces. These ratios of air box dimensions to
coil dimensions may be taken as guides for other problems where the inductance of an air-
cored coil is to be computed but there is no known analytical solution.
0.1 0.1
0.1 coil 2
0.1 P
0.1 0.1
0.1 coil 1
The coils have the same dimensions as for the single Brooks coil (page 94), and the number
of turns is N = 100. The spacing is 0.1 m, which is equal to the height of one coil. From Grover
[4], the mutual inductance of this configuration is
There are two methods of determining the mutual inductance with Simcenter MAGNET. The
first method is to use the relationships based on the flux linkage with one coil when current
flows in the other coil. For this linear system, appendix B gives:
21 12
M (4-4)
i1 i2
where 21 is the flux linkage with coil 2 when coil 1 carries a current i1, and 12 is the flux
linkage with coil 1 when coil 2 carries a current i2.
The second method of determining mutual inductance is based on the stored energy.
Consider two cases: (a) equal currents, so that i1 = i2 = i; (b) equal and opposite currents, so
that
i1 = –i2 = i. From appendix B, the mutual inductance is given by
Wa Wb
M , (4-5)
2i 2
where Wa and Wb are the stored energy values in the two cases.
Material: AIR.
Solving and post-processing
1. Set the options for solving, using the Solve menu, with the following changes from the
default values:
Adaption Options:
Use h-adaption,
Tolerance 0.01%.
2. Solve as Static 2D.
3. Inspect the contour plot of the flux function and the shaded plot of the |B| values.
4. Inspect the computed global quantities and calculate inductances as follows (see
appendix B for details):
Self-inductance: L1 = / i1, where is the flux linkage with coil 1, and i1 is the
coil current.
Mutual inductance: M = / i1, where is the flux linkage with coil 2.
5. Change the current in coil 1 to 0 A and the current in coil 2 to 1.0 A. Solve again and
calculate the inductances:
Self-inductance: L2 = / i2, where is the flux linkage with coil 2, and I2 is the
coil current.
Mutual inductance: M = / i2, where is the flux linkage with coil 1.
Sample results
The results below were obtained with Simcenter MAGNET version 7.5:
i1 = 1.0 A, i2 = 0
Stored magnetic energy: 0.001273 J
Flux linkage : 0.002546 Wb
Flux linkage : 0.0004895 Wb
Self-inductance L1: 2.546 mH
Mutual inductance M: 0.4895 H
i1 = 0, i2 = 1.0 A
Stored magnetic energy: 0.001273 J
Flux linkage : 0.002546 Wb
Flux linkage : 0.0004895 Wb
Self-inductance L2: 2.546 mH
Mutual inductance M: 0.4895 mH
i1 = 1.0 A, i2 = 1.0 A
Stored magnetic energy: 0.00303554 J
i1 = 1.0 A, i2 = –1.0 A
Stored magnetic energy: 0.00205666 J
Mutual inductance M: 0.4894 mH
Discussion
The computed value of the self-inductance is the same as for a single Brooks coil (see
page 94), which is about 0.1% lower than the theoretical value. All three methods of
calculation give the similar values for the mutual inductance, which is about 0.3% lower than
the theoretical value. These results indicate that the air box is large enough to give an
accurate result, and that the solver options are correctly chosen.
When the slug is centrally positioned, as shown above, the voltages induced in the sensing
coils are equal, so the output from the transducer is zero. If the slug is displaced from this
position, the voltage in one sensing coil increases while the voltage in the other decreases;
the differential voltage will vary from zero in an almost linear manner. The designer has the
problem of ensuring that the voltage/displacement characteristic is a linear one.
Although the device uses alternating current, it can be modeled as a static problem because it
does not depend on eddy currents in the slug. The voltage depends on the flux linkage with
the sensing coils, which can be determined from a static field solution.
2
3
P 7 4
The slug has a radius of 2 mm and a length of 7 mm; the dimensions of the coils are shown in
millimeters in the diagram, where the inner radius is 2 mm. The central coil has 100 turns and
a current of 0.1 A; the outer coils each have 200 turns, with no current.
Material: AIR.
Parameterization
1. In the Parameters page of the model Properties, define a new numeric parameter Shift
with the following values –1.5, –1.0, –0.5, 0, 0.5, 1.0, 1.5.
2. In the slug Component Parameters page, specify the shift vector as [0, %Shift%mm, 0].
Adaption Options:
Use h-adaption,
Tolerance 0.05%.
2. Solve as Static 2D.
With the Trial Edition of Simcenter MAGNET, solutions of successive problems can be
obtained by the method described in chapter 2 (page 37).
3. Inspect the contour plot of the flux function and the shaded plot of the |B| values.
4. Inspect the computed global quantities.
If Simcenter MAGNET is licensed for parameterization, display a graph of the differential
flux linkage against displacement. The differential flux linkage is the flux linkage with the
complete sensing coil, which is the difference between the flux linkages of the individual
coil components.
Sample results
The results below were obtained with Simcenter MAGNET version 7.5:
–1.5 –6.497
–1.0 –4.731
–0.5 –2.482
0.0 0.0
+0.5 +2.482
+1.0 +4.730
+1.5 +6.498
Discussion
From the symmetry of the device, the differential flux linkage should be an odd function of the
displacement, which is confirmed by the results. There is noticeable non-linearity as the
displacement approaches the extremes of ±1.5 mm.
In this case study, the sphere has a diameter of 20 mm; the magnet has a diameter of 20 mm
and a length of 20 mm. Initially, the model will have the sphere in contact with the magnet. A
controlled gap will be introduced by defining a set of values for a displacement parameter and
using this parameter in a shift vector for the sphere.
Material: AIR.
Parameterization
1. In the Parameters page of the model Properties, define a new numeric parameter Shift
with the following values: 0.1, 0.2, 0.3, 0.4, 0.5.
2. In the sphere Component Parameters page, specify the shift vector as
[0, –%Shift%mm, 0].
Note the negative sign for the y shift. If this is omitted, the sphere will intersect the
magnet, and the solution will fail.
Solver Options:
Polynomial order 4.
Adaption Options:
Use h-adaption,
Tolerance 0.01%.
2. Solve as Static 2D.
With the Trial Edition of Simcenter MAGNET, solutions of successive problems can be
obtained by the method described in chapter 2 (page 37).
3. Inspect the contour plot of the flux function and the shaded plot of the |B| values.
4. Inspect the computed global quantities. Take the force of attraction to be the mean of the
magnitudes of the Y components of force on the sphere and the magnet.
5. Repeat the study with polynomial orders 2 and 3 for the same model, and with
polynomial order 4 for new models with air box radii of 80 mm and 320 mm.
Sample results
The results below were obtained with Simcenter MAGNET version 7.5 for an air box radius of
160 mm and solver polynomial order 4:
Discussion
The reference model, with an air box radius of 160 mm and polynomial order 4, gives almost
identical results for the magnitudes of axial forces on the magnet and the sphere. At a
displacement of 0.3 mm, reducing the boundary radius to 80 mm changes the force by 1%,
but increasing the radius to 320 mm changes the force by only 0.14%. The polynomial order
has very little effect on the results when adaption is used with a small tolerance of 0.01%. It
appears that the computed results for a radius of 320 mm are close to convergence on the
true values, and the numerical error is unlikely to exceed 1%. Since the force depends on the
square of the flux density, a change of only 0.5% in the remanence of the magnetic material
would give a force change of 1%. The accuracy of the computed force value is therefore
limited by the variability of material properties rather than numerical errors.
Center pole
Top plate
Moving coil
Permanen
t magnet
Bottom plate
It is required to find the coil inductance and explore the linearity of the force/current
relationship under the following conditions:
2
7 26
P
5
20 7
15
10 10 25
5
40
i 0
L (4-6)
i
The force on the coil will be calculated for the following conditions:
The coil current is varied in steps of 0.2 A with the coil in its mid position.
The coil position is varied in steps of 1 mm with the current at its rated value.
Material: AIR.
Parameterization – current
In the Coil Parameters page, enter the following list of values for the Coil Current
parameter:
–1.0, –0.8, –0.6, –0.4, –0.2, 0, 0.2, 0.4, 0.6, 0.8, 1.0
Adaption Options:
Use h-adaption,
Tolerance 0.5%.
2. Solve as Static 2D.
With the Trial Edition of Simcenter MAGNET, solutions of successive problems can be
obtained by the method described in chapter 2 (page 37).
3. Inspect the contour plot of the flux function and the shaded plot of the |B| values.
4. Inspect the computed global quantities for the different current values.
If Simcenter MAGNET is licensed for parameterization, display a graph of the Y
component of force on the coil.
5. Determine the coil inductance for a current of –1 A from equation 4-5 on page 108.
Note that the RLC Matrix Calculator extension to Simcenter MAGNET (see page
74) will not give the correct value for the inductance in this case.
Parameterization – displacement
1. In the Coil Parameters page, change the current parameter values from a list to the
single number: –1.0
2. In the Parameters page of the model Properties, define a new numeric parameter Shift
with the following values: –5, –4, –3, –2, –1, 0, 1, 2, 3, 4, 5
3. In the Coil Component Parameters page, specify the shift vector as [0, %Shift%mm, 0]
Discussion
There is noticeable non-linearity in the force/current characteristic, which is not symmetrical
for positive and negative current values. Its origin is the flux generated by the coil current
itself, which aids or opposes the permanent-magnet flux in the airgap. The origin of the effect
can be demonstrated by substituting a non-magnetic material such as aluminum for the
ceramic ferrite material, since the recoil permeability of ceramic ferrite is close to 1. A current
magnitude of 1.0 A in the coil then gives an axial force of –0.389 N, independent of the
direction of the current. This force is in the negative Y direction, drawing the coil in to the
magnet, and it is proportional to the square of the current.
The inductance calculated from equation 4-6 on page 108 is a static value, which ignores
eddy currents in the magnetic circuit, particularly the center pole, when the coil carries
alternating current.
Scripting
Introduction
Up to this point, Simcenter MAGNET has been used interactively, with the mouse and the
keyboard, to build models and analyze the results. Simcenter MAGNET can also be controlled
by scripts and scripting forms.
Scripts are text files containing commands that control Simcenter MAGNET. A script can be
recorded during a Simcenter MAGNET session. When this script is run, all the operations that
were carried out during the recording session will be repeated automatically. Scripts created in
this way can be edited to change the operations, and scripts that are more powerful can be
created with the VBScript programming language.
Recording a script is often an effective way of finding out how to use the Simcenter MAGNET
scripting commands, in preparation for writing special-purpose user scripts.
Scripting forms take scripting a stage further by providing a graphical user interface for the
user to interact with the script. A form can have text boxes for entering values, buttons for
starting actions, and areas for displaying results. Scripting forms are not covered in this
document, but sources of information are given below.
The most advanced kind of scripting uses another application to communicate with Simcenter
MAGNET through the Microsoft ActiveX Automation Interface. Microsoft Excel, for example,
can be used in this way. A script in the form of an Excel macro can command Simcenter
MAGNET to build and solve a model, using data entered on the spreadsheet, and then get
results back from Simcenter MAGNET to display on the spreadsheet. An example of this form
of scripting is given in the section “Automation with Excel” on page 117.
Further information
The Simcenter MAGNET help gives full particulars of the Simcenter MAGNET scripting
commands. Guidance on writing scripts is available from the Support area of the Siemens
website.
https://support.sw.siemens.com/
A B
P
The cross-section of each block is a square of side 20 mm, the gap between the blocks is
20 mm, and the blocks are 100 mm deep. The material is ceramic ferrite.
This is an open boundary problem, which is best handled by the open boundary technique
described in the busbar forces case study on page 55. However, for simplicity, a simple
circular boundary will be used in this example.
Material: AIR.
Sweep distance: 100 mm.
Solver Options:
Polynomial order 2.
Adaption Options:
Use h-adaption,
Tolerance 0.05%.
2. Solve as Static 2D.
3. Inspect the contour plot of the flux function.
4. Inspect the forces on the blocks.
Sample results
The results below were obtained with Simcenter MAGNET version 7.5.
In the message box “Do you want to edit the user script file … immediately?”,
click No.
The core of the Excel implementation is a set of macros or subroutines, written in Visual
Basic, which are similar in principle to the scripts discussed above. There are four buttons on
the worksheet, each linked to a macro, which do the following:
Start Simcenter MAGNET: starts the Simcenter MAGNET application and set the length units.
Close Simcenter MAGNET: closes the model file, close the application and release resources.
Simcenter MAGNET Visibility: makes Magnet visible or invisible.
Run Model: gets data values from cells D5 to D11, sends commands to Simcenter MAGNET to
build and solve the model, gets the x-components of force on the blocks from Simcenter
MAGNET, and displays the results in cells D13 and D14.
The macro activated by the Run Model button is the core of the implementation. For clarity,
this macro calls other subroutines that carry out specific tasks such as creating one magnet
block.
The instructions are based on Microsoft Excel 2007; other versions may behave differently.
Excel worksheet – 1
1. Start Excel.
2. Ensure that there is access to Visual Basic as follows:
Click the Microsoft Office button, and then click Excel Options.
Click Trust Center, and then click Trust Center Settings.
Click Macro Settings.
In Macro Settings, click Disable all macros with notification.
In Developer Macro Settings, set the check box for Trust access to the VBA
project object model.
Click OK to close the Trust Center window.
Click Popular, and set the check box for Show Developer tab in the Ribbon.
Click OK to close the Excel Options window.
3. Save the current blank workbook as PM Blocks.xls
4. Enter the text and numerical values exactly as shown on the previous page, except for
cells D13 and D14.
1. Click the Developer tab of the Ribbon, and in the Code group click Visual Basic.
This should display the Visual Basic Editor in a new blank window.
2. On the Insert menu, click Module
This should open a new Code window named Module1, with an insertion point for text
entry.
3. Type the text listed on the next two pages, taking care to copy it accurately. See the
section “Comments on the Code” for an explanation of the content.
' Permanent magnet blocks with Simcenter MAGNET.
Option Explicit
Sub Visibility()
' Subroutine called by the Simcenter MAGNET Visibility button.
' Toggles the Simcenter MAGNET visibility flag.
If Not Running Then
Call MsgBox("Simcenter MAGNET is not running.", vbOKOnly)
Exit Sub
End If
If Visible Then
Visible = (MsgBox("Simcenter MAGNET is visible. Change to invisible?", vbYesNo) =
vbNo)
If Not Visible Then
Mag.Visible = False
End If
Else
Visible = (MsgBox("Simcenter MAGNET is invisible. Change to visible?", vbYesNo) =
vbYes)
If Visible Then
Mag.Visible = True
End If
End If
End Sub
Excel worksheet – 2
1. Return to the Excel window by clicking the Excel icon on a left-hand toolbar at the top of
the Visual Basic Editor window.
2. In the Controls group, click Insert.
In the Form Controls area, click the Button icon.
3. Click on worksheet cell F1 to insert a button.
A button and a dialog box should appear.
4. In the Macro Name drop-down list, select StartSimcenter MAGNET.
5. Click OK.
6. Click inside the new button, and change the text to Start Simcenter MAGNET.
7. Hold down the Alt key, and adjust the size and position of the button by dragging the
border.
The button will snap to the worksheet cells.
8. Click outside the button, so that the button is no longer marked with a border.
After this, avoid clicking the button. Two more buttons are required before testing
can begin.
9. Save the file.
10. In a similar way, insert two more buttons:
A button named Close Simcenter MAGNET, linked to the CloseSimcenter
MAGNET subroutine.
A button named Simcenter MAGNET Visibility, linked to the Visibility
subroutine.
Simcenter MAGNET should close, and the icon should disappear from the task
bar.
If there are errors, correct them as before.
5. Check that all three buttons work correctly.
A message box should be displayed if you try to start Simcenter MAGNET again
when it is already running, or use either of the other two buttons when Simcenter
MAGNET is not running.
6. Ensure that Simcenter MAGNET is closed before continuing with the development.
If you start a second instance of Simcenter MAGNET, it may not work correctly.
If this happens, press Ctrl+Alt+Delete and use the Windows Task Manager to
close all instances of Simcenter MAGNET.
Dim ArrayOfValues(0)
Press F8.
The subroutine header line is marked in yellow.
Press F8 again.
The subroutine NewModel is entered.
If you start a second instance of Simcenter MAGNET, it may not work correctly.
If this happens, press Ctrl+Alt+Delete and use the Windows Task Manager to
close all instances of Simcenter MAGNET.
11. When all the errors have been corrected, single-stepping will reach the end of the
RunModel() subroutine, and the result will be the same as if the Run Model button had
been clicked on the Excel worksheet.
12. Check that the force values are similar to those displayed in the view of the Excel
worksheet on page 117. The values will be slightly different, because the model uses a
Kelvin open boundary.
This text must be entered accurately, or the model will fail because the material is
unknown.
The solution will be very much faster, because no Newton steps are required.
3. Try changing the dimensions of the blocks and the gap between the blocks.
Subroutine CloseMagnet
This subroutine closes the current model if one has been built, then closes the Simcenter
MAGNET application and resets variables.
Subroutine Visibility
This subroutine determines whether the Simcenter MAGNET window should be visible, by
testing the value of the global variable Visible. The value (True or False) is set by the user’s
response to a question in a dialog box.
Subroutine RunModel
This subroutine calls other subroutines described below to get data from the worksheet and
build the model. It solves the model, and displays results for force on the worksheet.
Subroutine NewModel
This subroutine closes the current model and starts a new model. It also resets some object
variables so that they refer to the new model, and sets the Simcenter MAGNET solving
parameters.
Subroutine GetData
This subroutine gets data from designated cells on the first worksheet, which is assumed to
have the Excel default name Sheet1. The subroutine does no data checking, and simply
assumes that valid data have been entered in the worksheet. A production version of the
subroutine should include checks for valid data.
Subroutine MakeBlock
This subroutine makes a block of permanent-magnet material, using data supplied in the
parameter list. It is called twice by the subroutine SolveModel to make the two blocks. The
Simcenter MAGNET commands are similar to those found in a User Script Log file, apart from
the string manipulation to use the name of the material from the worksheet.
Subroutine OpenBoundary
This subroutine implements the Kelvin transformation to represent an open boundary. It
creates an interior air space component surrounding the model, and a small exterior air
component. The components are linked with an even periodic boundary condition.
Subroutine MakeCylinder
This subroutine makes a cylindrical air region, using data supplied in the parameter list. It is
called twice by the subroutine OpenBoundary to make the air space and exterior regions.
The static magnetic field is described by the magnetic flux density B and the magnetic
intensity H, which satisfy the equations
curl H J , (A-1)
div B 0 , (A-2)
B H 0 r H , (A-3)
where J is the current density, which is the source of the magnetic field. Equation A-1 is the
differential form of Ampère’s circuital law – equation 1-2 of Chapter 1 – and equation A-2
embodies the fact that there are no free magnetic poles, so magnetic flux lines are closed
curves.
To solve these equations we use equation A-2 to express the flux density B in terms of
another vector A through the equation
B curl A , (A-4)
where A is known as the magnetic vector potential. Equation A-1 then becomes
1
curl curlA J , (A-5)
which is solved numerically by the finite-element method to determine the magnetic field in the
device (see the section “Numerical Solution” on page 130).
In two dimensions, the magnetic vector potential has some very useful properties. For the XY
plane of translational geometry, the current and therefore the vector potential is in the
z direction. The components of B are then given by
A A
Bx , By (A-6)
y x
where A = Az is the magnitude of the vector potential. The scalar quantity A will be termed the
magnetic potential. From equation A-6 it may be shown that equipotentials – lines of constant
A – are flux lines. Moreover, the magnetic flux between any two points in the XY plane, for a
depth d in the z direction, is given by
d ( A1 A2 ) , (A-7)
With the RZ plane of rotational geometry, a modified potential is required for numerical
stability and accuracy [2]; Simcenter MAGNET uses the quantity
U rA . (A-8)
It may be shown that lines of constant U are flux lines in rotational geometry, and that the flux
between two points is given by
2 (U 1 U 2 ) , (A-9)
where U1 and U2 are the values of U at those points. Thus U is the flux per radian. The flux
plots for rotational geometry in Simcenter MAGNET are contours of constant U.
The components of B are not given by equation A-6 in rotational geometry with r and z
substituted for x and y. Instead, in cylindrical polar coordinates, we have:
A 1U
Br , Bz
1
rA 1 U (A-10)
z r z r r r r
In 2D, an artificial boundary takes the form of a closed curve along which a property of the
field is specified. The field property is the boundary condition. Two kinds of boundary condition
are relatively easy to implement when solving the field equations:
The Dirichlet, or flux tangential, boundary condition. The flux function (A or U) is constant over
any portion of the boundary with this condition, so the portion becomes part of a flux line.
The Neumann, or field normal, boundary condition. The direction of B is at right angles to any
portion of the boundary with this condition, so the flux lines enter the portion at right angles.
The default boundary condition in Simcenter MAGNET is flux tangential, so the entire outer
boundary will become a flux line unless the user specifies otherwise. For most problems this is
the best choice. It is equivalent to putting the model in a cavity of a material with zero
permeability, so that no flux can escape from the model. If the boundary is taken sufficiently
far away from the components of the model, it is a good approximation to an open boundary.
A radius of about 10 times the model dimensions will be sufficient in many applications.
It is a simple way of simulating the effect of a magnetic screen around the model.
It can be used for creating artificial field patterns, such as the uniform field of finite extent in the
case study on the cylindrical screen (page 64).
It can be used to reduce the size of a model by exploiting symmetry (see below).
For any given artificial boundary shape, it may be shown that the true field in the model lies
between two extremes: the result with a flux tangential boundary condition, and the result with
a field normal boundary condition. A method of checking the size of the boundary is therefore
to repeat the solution using the other boundary condition and compare the results.
Devices such as rotating electrical machines have a periodic structure, where the magnetic
field conditions in one part of the device are similar to the conditions in another part. It is
possible to represent the device by a model of a small part, for example one or two pole
pitches in a rotating machine. The periodic nature of the device is represented by a constraint
between the field values at the two ends of the representative portion. This constraint is
termed a periodic boundary condition. If the field values at the two ends are equal in
magnitude and sign, the constraint is an even periodic boundary condition. A model of two
complete pole pitches of a machine would require this constraint. If the field values are equal
in magnitude but opposite in sign, the constraint is an odd periodic boundary condition. A
model of one pole pitch of a machine would require this constraint, and an example is given in
the case study of linear synchronous motors on page 84.
Open boundaries
Consider a 2D static problem with translational symmetry that has an open boundary. Let the
2D model be enclosed in a circle of radius R, so that all the material parts of the model are
inside the circle. The field region comprises two parts: the finite interior of the circle, and the
infinite space outside the circle. Since the infinite space region is current-free, so that J = 0, it
follows from equation A-1 that the magnetic intensity H can be represented by the gradient of
a scalar potential:
H grad (A-11)
There is a further restriction, however: for equation A-11 to hold, we must also have:
H dl 0 (A-12)
H dl i (A-13)
where the right-hand side is the sum of all currents passing through the contour of integration.
Equation A-12 therefore implies that the model cannot contain any unbalanced currents.
Provided equation A-11 holds, the Kelvin transformation [7] gives the following useful result:
the infinite space outside the circle of radius R is equivalent to the interior of another circle of
any finite radius kR, if the potentials of all corresponding points on the two circles are identical.
To implement this in Simcenter MAGNET, it is necessary to construct two air boxes: a normal
circular air box surrounding the model, and a second circular air box to represent the infinite
external space. An even periodic boundary condition is imposed on corresponding curved
surfaces of the two air boxes, forcing the potentials to have the same values at corresponding
points. An example of a manual implementation is given in the case study on busbar forces
(page 55), and a scripting example is given on page 117.
Other techniques can be used to simulate open boundaries in 2D models with rotational
symmetry, and in 3D models, but they will not be considered in this document.
Numerical solution
Introduction
The core of Simcenter MAGNET is a powerful technique for solving the electromagnetic field
equations numerically. Most of this process is automatic and virtually transparent to the user,
but it is necessary to control the process by setting the solver and adaption options. As with
the fundamental equations, it is helpful if the user has some understanding of the method.
Simcenter MAGNET employs the finite-element method [8] to solve the 2D form of equation
A-5 for the magnetic potential. With this method, the region of the problem is divided into a
mesh of triangular elements, and the potential in each element is approximated by a simple
function of the x and y (or r and z) coordinates. The simplest function is a linear variation with
position; this gives first-order elements, where the potential inside a triangular element is
obtained from the potentials at the three vertices or nodes. High-order elements use high-
order polynomials and additional nodes to represent the potential. The problem of solving
equation A-5 then reduces to the solution of a set of linear equations for the unknown
potentials at all the nodes. This must be repeated several times if the model contains non-
linear magnetic materials.
The accuracy of the finite-element solution depends on three factors: the nature of the field,
the size of the elements, and the element order. In regions where the direction or magnitude
of the field is changing rapidly, high accuracy requires small elements or a high element order.
In addition, the methods used to find the finite-element solution are iterative, with an
adjustable error criterion for terminating the process.
Constant values of permeability are chosen for each element, from the initial slope of the material
B-H curve.
The resulting linear equations are solved numerically for the magnetic potential, using the semi-
iterative conjugate gradient method.
The flux density values are calculated from the magnetic potential, and these results are used to
calculate new values for the permeability of each element.
The process is repeated until the element permeability values have converged.
CG steps
At each step in the conjugate gradient process, the change in the solution is monitored. The
process continues until the change is less than the CG Tolerance. For most static problems,
the default value of 0.01% should be satisfactory. Time-harmonic eddy-current problems
frequently require a value at least 10 times smaller for accurate results.
Newton steps
At each step in the permeability calculation process, the change in the solution is monitored.
The process continues either until the change is less than the Newton Tolerance, or until the
limit of Maximum Newton Iterations is reached. For most problems, the default values of 1%
and 20 iterations should be satisfactory. Examples of problems where these settings are not
sufficient are (a) permanent-magnet materials in contact with non-linear soft magnetic
materials, in a device where the flux density values are very high, and (b) closed magnetic
circuits in iron surrounding a current (like a closed slot in an electrical machine). For these
kinds of problem, the tolerance should be reduced to 0.01% and the iteration limit increased to
50 or more.
With some models, increasing the polynomial order is as effective as using adaption to
improve the solution accuracy. In most cases, however, good results will be obtained by
setting the polynomial order to 2 and using adaption as described below. In special cases, a
high order must be used in conjunction with adaption: see the case study on magnetic pull-off
force (page 103).
Adaption
Adaption is the process of automatic refinement of the mesh to improve the solution accuracy.
For 3D models there is a choice of two adaption methods: h-type adaption, where element
sizes are halved, and p-type adaption, where the element polynomial order is increased. For
2D models, only h-type adaption is available.
After each adaption step, the change in the calculated value of stored magnetic energy is
monitored. Adaption continues until this change is less than a specified tolerance, or the
specified number of steps has been reached. As the case studies demonstrate, the default
tolerance of 1% is generally too large for a high-accuracy solution. If the quantity of interest is
the force or torque, rather than an energy-related quantity such as inductance, a more
accurate solution may be required, and the change in the stored magnetic energy may not be
a good indicator. In such cases it is often advantageous to set the tolerance to a very low (but
non-zero) value, and control the mesh refinement by adjusting the maximum number of
adaption steps. The optimum setting can be determined by changing the number of steps and
monitoring the change in the force or torque value.
Maximum element size: the maximum element edge length. This can be increased to force
adaption to start with a coarse mesh, or reduced to give a fine mesh.
Curvature refinement ratio: a measure of the maximum deviation when a curved part of the model
is approximated by the straight-line edge of an element.
Curvature refinement minimum element size: limits over-discretization of tight curves when the
elements are refined.
These properties can be set for the entire model, or on individual components, surfaces and
edges. In addition, the mesh can be controlled by edge subdivision. This feature of Simcenter
MAGNET enables the user to specify the number of segments on a given component edge
(line or arc) when the initial mesh is generated. The subdivisions can be linear or logarithmic.
Details of the procedures for setting the mesh properties are given in the Simcenter MAGNET
help. For most purposes, the user is recommended to avoid using edge subdivision because
the other methods are more effective.
The case study on the linear synchronous motor (page 84) gives an example of setting the
maximum element size in the small airgap of the machine. In this case, the method is more
effective than adaption for getting accurate force values.
Definitions
When a magnetic field is established in a device by increasing the currents in the coils from
zero to some final value, energy must be supplied in addition to the i2R ohmic loss in the
conducting material. This energy is considered to be stored in the magnetic field; if hysteresis
is neglected, it can be recovered when the currents are reduced to zero.
The diagram below shows part of the B-H curve for a magnetic material. If point b represents
the final magnetic condition in a particular part of the material, then it may be shown [9] that
the energy density, or energy stored per unit volume, is given by:
w H dB [J/m3] (B-1)
B
1.6
b
a
energy.
dB
co-energy.
0
o0 dH c H1000
The energy density is the area between the B-H curve and the B axis, which is the area oab in
the diagram.
w B dH [J/m3]. (B-2)
The co-energy density is the area between the curve and the H axis; this is equal to the area
ocb in the diagram. Since the sum of the two densities, w + w', must equal the area of the
rectangle oabc, we have
w B dH B H w B H H dB [J/m3]. (B-3)
If the material is linear, then the energy density and the co-energy density are both equal to
half the area of the rectangle oabc. For a non-linear material the energy density is less than
this area, and the co-energy density is greater than this area; we therefore have
w ½B.H w (B-4)
The total energy stored in the device is just the integral of the energy density over the volume:
W w dv B H H dB dv [J]. (B-6)
When a device is solved in Simcenter MAGNET, the stored energy and co-energy values
displayed in the Energy tab of the Post-processing bar are the values of W and W' given by
equations B-5 and B-6.
The usual definition of inductance is the flux linkage per ampere (see page 139). If the stored
energy is calculated in the usual way as ½Li2, this will differ from both the stored energy W
and the co-energy W'; in fact it is equal to the equivalent linear energy
The co-energy finds its application in the virtual work method of calculating forces and
torques; this is quite different from the method used in Simcenter MAGNET and described on
page 138. If the current is held constant and a part of the device is given a displacement s in
any direction, it may be shown that the component of force in the direction of s is given by
W W
Fs (B-8)
s s
If two models are constructed which differ by s in a distance s, then the force may be
calculated from equation B-8 by taking the difference of the co-energy values. A similar
equation holds for torque in terms of an angular displacement :
W W
T (B-9)
Lorentz force
The force per unit volume on a current-carrying conductor in a magnetic field is
f JB (B-10)
which is another form of the Lorentz equation given in equation 1-1 of chapter 1. The total
force is therefore given by the integral
F J B dv (B-11)
and the total torque about a point P is
T r J B dv (B-12)
where r is the radius vector from P to the volume element, and the integral is taken over the
volume of the conductor. These equations are known to give good results when the flux
density B is determined numerically by the finite-element method, so Simcenter MAGNET
implements equations B-11 and B-12 to calculate the force and the torque on current-carrying
regions in the model, provided the material has a relative permeability r = 1.
Maxwell stress
Equations B-11 and B-12 can be used only for current-carrying conductors; they do not give
the force and torque on magnetic materials. If a device includes magnetic materials as well as
current-carrying conductors, Simcenter MAGNET uses a method of force calculation based on
the Maxwell stress concept [9, 10] which gives the stress, or force per unit area, directly in
terms of the magnetic flux density. If Bn and Bt are the components of flux density normal and
tangential to a surface, and n and t are the corresponding components of stress, then:
Bn2 Bt2
n (B-13)
20
Bn Bt
t (B-14)
0
Provided that the surface is closed, and passes entirely through air, the total force and torque
may be determined by integrating the stresses over the surface. This result is completely
general; it is independent of the nature of the objects inside the surface, which may include
currents, soft magnetic materials or permanent magnets.
If the Maxwell stress method is used to calculate forces from a standard numerical solution for
the field, it is difficult to get accurate results. The integral for the force or torque may be
unreliable if it comprises terms which alternate in sign, leading to an accumulation of
numerical errors. It is also very sensitive to the accuracy of the numerical solution. Users
require considerable skill and experience to get good results by this method.
Simcenter MAGNET avoids this difficulty with the conventional Maxwell stress method by
implementing a novel tunable method [11]. This requires no skill on the part of the user, and it
gives accurate values for the force and the torque.
The Lorentz J B forces and torques are calculated for any current-carrying regions where
the material has a relative permeability r = 1. These quantities are not displayed in the
Results window, but are given in the Solver Log file. Since these forces and torques are
included in the total force and torque on a body containing the current-carrying region, the
force and torque on the magnetic material alone may be determined by subtraction.
Inductance calculation
Flux linkage
The magnetic properties of any pair of coupled coils may be described in circuit terms by self-
inductances L1 and L2 and mutual inductances M12 and M21. If coil 1 carries a current i1, the
flux linkages with coils 1 and 2 are given by
2 N 22 L2 i2 , (B-17)
1
L1 (B-19)
i1
2
L2 (B-20)
i2
12
M12 (B-21)
i1
21
M 21 (B-22)
i2
For a linear system, all four coefficients are independent of current, and M12 = M21 = M. When
N1 = N2, leakage inductances l1 and l2 may be defined as follows:
l1 L1 M , l2 L2 M (B-23)
For devices in which the magnetic materials are linear and soft, there is an alternative method
of calculation based on energy.
Energy methods
For a linear system, the stored magnetic energy in a pair of coils is given by
W ½Li
1 1 ½L2i2 Mi1i2 .
2 2
(B-24)
Alternative methods of calculating mutual inductance and leakage inductance can be derived
by equating this expression to the stored energy given by equation B-5.
To calculate mutual inductance, consider two situations: (a) identical currents, so that
i1 = i2 = i; (b) equal and opposite currents, so that i1 = –i2 = i. The stored energy values are:
Wa ½Li
1 ½L2i Mi
2 2 2
(B-25)
Wb ½Li
1 ½L2i Mi
2 2 2
(B-26)
from which it follows that
Wa Wb
M . (B-27)
2i 2
Equation B-23 for the leakage inductances may be unsatisfactory when the values are given
by the small differences between large quantities. An energy method can be based on a
simulation of the transformer short-circuit test, where the two coils carry equal and opposite
currents. In terms of the leakage inductances, equation B-24 becomes
2Ws
l1 l2 . (B-30)
is2
As with the physical short-circuit test, this method can determine only the sum of the leakage
inductances, not their individual values; but it may give more precise results for the sum than
equation B-23.