Escolar Documentos
Profissional Documentos
Cultura Documentos
OPERA-2D
USER GUIDE
Chapter 1
Structure of the User Guide
Road Map ................................................................................................ 7
Chapter 2
Program Philosophy
Introduction ............................................................................................. 9
Pre-processing ....................................................................................... 11
Analysis ................................................................................................ 14
Post-Processing ..................................................................................... 16
Chapter 3
Getting Started
Starting the OPERA-2d Pre and Post-Processor ................................... 17
Pre-processing ....................................................................................... 19
Solving the Problem .............................................................................. 43
Post-Processing ..................................................................................... 46
Chapter 4
A Permanent Magnet Motor Example
Introduction ........................................................................................... 51
Pre-Processing ...................................................................................... 53
Post-Processing ..................................................................................... 87
Chapter 5
Eddy Current Example: AC and Transient
AC Analysis Case 1 .............................................................................. 95
AC Analysis Case 2 ............................................................................ 102
TR Analysis ........................................................................................ 106
Chapter 6
Dynamic Analysis of a PM Motor
Introduction to Rotating Machine Analysis ........................................ 109
Analysis .............................................................................................. 122
Post-Processing ................................................................................... 126
Chapter 7
Stress Analysis Notes and Examples
A review of stress analysis .................................................................. 131
Example - A Simple Beam ................................................................. 143
Chapter 8
Thermal Analysis Notes and Examples
Thermal Analysis ............................................................................... 155
Example - A Simple Heat Bath ........................................................... 157
Chapter 9
Coupled Analysis: Static Fields and Stress, AC Fields and Transient
Thermal
Electromagnetic analysis coupled to Stress and Thermal ................... 167
Chapter 10
Space Charge Example
Introduction ......................................................................................... 189
OPERA-2d Pre-processor ................................................................... 191
Creating the Emitter Data ................................................................... 203
Post-Processing ................................................................................... 206
Chapter 11
Application Notes
Introduction ......................................................................................... 211
Flux Linkage Calculations in 2D Solutions ........................................ 213
Inductance Calculations in OPERA-2d .............................................. 216
Power and Energy Calculation in AC Solutions ................................. 219
Field Recovery Outside Meshed Regions ........................................... 221
Complex Material Properties .............................................................. 227
Setting up a Lossy Dielectric Problem ............................................... 231
Use of Command Scripts to Calculate Fourier Series ........................ 237
External Circuits in OPERA-2d .......................................................... 239
Shared Filamentary Conductors in OPERA-2d .................................. 247
3-phase Systems .................................................................................. 251
Delta Connected Loads ....................................................................... 254
Using OPERA-2d/DM to Magnetize a Magnet .................................. 257
Demagnetization of Permanent Magnets “In Service” ....................... 270
Linear Motion Models ........................................................................ 284
Steady State Modelling of Induction Motors ...................................... 293
Dynamic Modelling of Induction Motors ........................................... 300
Electromagnetic and Stress Analysis of a SRM ................................. 309
Stress Analysis Solver Validation ....................................................... 317
Particle Trajectory Intersections in OPERA-2d .................................. 326
Plasma Free Surface Emitter in 2D ..................................................... 328
Importing DXF Files .......................................................................... 332
Road Map
Program Philosophy
An overview is given about the underlying philosophy of the software - the
fact that models are created in a pre-processor including material defini-
tions and mesh generation, and the computed results viewed and processed
in the post-processor. Both of these functions are carried out in a single
module, called the Pre and Post-Processor, so that whilst viewing results,
changes can immediately be made to the model in order to optimise a
design.
Getting Started
This gives a detailed description of how a model is prepared and analysed.
New users are encouraged to spend some time going through this chapter,
as it will answer many questions that can otherwise arise when using the
software.
Application Notes
This chapter contains a number of useful techniques that can be used for
performing various tasks. If a question arises as to how to use the software
in a particular way, this chapter should first be consulted in case an answer
Tutorials
A series of examples using the software is included. Each attempts to high-
light a typical application using various analysis modules.
Introduction
The software uses the finite element method (FEM). Since much informa-
tion is required before the analysis may be performed, data entry is carried
out using a powerful interactive pre-processor. Using the graphical interac-
tion within the pre-processor, the model space is divided into a contiguous
set of (triangular) elements. The physical model may be described in carte-
sian or cylindrical polar (axi-symmetric) coordinates.
Once the model has been prepared, the solution is calculated using a suita-
ble analysis module. Several modules exist for analysis of the different
types of electromagnetic excitation conditions e.g. static, steady state. The
analysis program iteratively determines the correct solution including non-
linear effects if these are modelled.
Pre-processing
Aspects of Pre-processing
The geometry of a device to be analysed by OPERA-2d is presented to the
Pre and Post-Processor as a set of polygonal areas or regions on the 2d
plane. One region can be a ‘background’ region which covers the whole
problem space. The others are non-overlapping polygons which specify the
other materials in the problem. In magnetic devices, for example, a region
can represent one of the following:
• free-space;
• a conductor with a prescribed or induced current density;
• permeable material with a linear or nonlinear material characteristic;
• a permanent magnet.
A special gap region must be defined for the interface region between the
stator and rotor of rotating machines for OPERA-2d/RM. The gap region
may also be used beneficially for electrical machine models with other
analysis programs. For more information see the OPERA-2d Reference
Manual.
Within each region, finite element mesh generation is automatic, using the
subdivisions of the sides to control the mesh density. There are two classes
of regions shapes: quadrilaterals and general polygons. The mesh within
quadrilaterals is generated by transformation to a unit square and regular
subdivision. The mesh is therefore predictable and mesh generation times
are short. Elements with large aspect ratios necessary for small air gaps can
be generated. The mesh within polygons is generated using an algorithm
based on Delaunay triangulation. Internal nodes are added if necessary to
achieve element sizes which vary smoothly across the regions and element
shapes which are as near to equilateral as possible. Polygons allow large
areas of space, especially near a ‘far-field’ boundary, to be meshed with a
minimal number of elements.
Region data is entered with the DRAW and POLYGON commands, and
can be edited with the EDIT, MOVE and MODIFY commands. Regions
can be copied with the COPY command.
All the material properties and boundary conditions are stored with each
region. The exceptions to this are the nonlinear relationships between flux
density and field strength which are stored as tables of values, related to the
regions by material code numbers. Such BH tables are defined and edited
with the BHDATA command.
Regions are also used in post-processing as the smallest units of area over
which integration can be performed (INTAREA command).
OPERA-2d/PP also has commands for displaying the region data, numeri-
cally and graphically (LIST, RECONSTRUCT), generating and checking
the mesh (MESH, CHECK) and reading and writing files (READ,
WRITE). Before writing a data file, specific data needed for the type of
analysis to be performed should be added using the SOLVE command. The
SOLVE command allows the user to define such items as convergence tol-
erances, output time points, nonlinear iteration type. A data file containing
the model and the analysis specific data is then created. No further interac-
tion is necessary before running the analysis program, which can be started
without leaving the Pre and Post-Processor.
Analysis
Once the analysis has started no further user interaction is required. The
programs create results files which contain a copy of the data and the solu-
tion(s), as well as a log file containing diagnostics.
There are a number of analysis programs in the OPERA-2d Suite. They all
read data prepared by the Pre and Post-Processor and create results files
which can be read by the Pre and Post-Processor. The analysis programs
are:
Statics:
ST This handles both magnetostatic and electrostatic problems. It
assumes that the excitation current is constant with time, and
allows excitation by permanent magnet. The material properties
can be linear or nonlinear (i.e. the permeability may be a func-
tion of the field strength at each point in the material). Isotropic
or anisotropic material properties are also catered for.
SP This module solves for time invariant electric fields including
the effects of space charge created by beams of charged parti-
cles. Infinite XY or axisymmetric coordinate systems can be
used. A magnetic field may also be applied in addition to the
electric field.
Eddy Currents:
AC Steady-state ac eddy currents with linear or nonlinear materials
and either current or voltage driven sources. Permeabilities can
either be taken from the region data, looked up from a previous
ST or TR solution, or calculated from the maximum field in the
AC solution. In all cases the permeability can be complex.
TR This assumes the excitation current or voltage is of any form that
is described by a variation over time. The solution is determined
at discrete time steps. The transient waveform may be selected
from standard driving functions or may be explicitly defined in
tabular format.
VL Eddy currents induced by constant velocity motion of one part
of the model with respect to the rest. Only static excitation is
allowed.
Post-Processing
Further processing can take the form of integrations along lines or over
regions giving values for forces, stored energies etc., or particle trajectory
calculations.
The first request for a field value from the Post-Processor prompts the pro-
gram to perform ‘field averaging’. The finite element method used gives
potential derivatives which are discontinuous from one element to the next.
The field averaging process finds the average from each element which sur-
rounds a node and applies that average value to the node. (Physical discon-
tinuities are maintained.) The averaged values are used in all the post-
processing commands, but some can also access the un-averaged values.
Comparison of the averaged and un-averaged fields allows an estimate of
the local and global errors in the solution. These error estimates are calcu-
lated following the field averaging and can be displayed in the same ways
as the fields.
Microsoft On Microsoft Windows systems the OPERA Manager is started from the
Windows menu bar as follows:
Platforms
Start -> Programs -> Vector Fields OPERA -> OPERA 11.0
Alternatively the OPERA Manager can be started from the system icon tray
Unix Platforms On UNIX systems, the OPERA Manager is started from the command line
by typing:
$vfdir/bin/opera.sh $vfdir
where $vfdir is a shell variable set to the directory where the software is
installed.
The project folder can be changed from the default by the menu route:
Pre-processing
Select OK to finish.
To refresh the graphics screen click on the Refresh button . The axes are
now scaled in centimetres.
Building a Model
Figure 3 shows a three dimensional cut view of a coil, wound around an
iron “E” shaped former positioned close to a circular metallic disc.
Using the axial symmetry of the coil, former and disc, it is possible to
model this using a two dimensional section. The two dimensional represen-
tation is shown in Figure 4.
Select OK to close. This will then cause the UNITS and PROBLEM DATA panels
to the right of the screen to be changed and an appropriate message is con-
veyed to the user.
This has now set the model to use axisymmetry i.e. an R-Z coordinate
(cylindrical polar coordinate) system. The solution potential of MODI-
FIED R*A is a modified vector potential that should be used to improve
accuracy near the axis in axi-symmetric problems.
Refresh the display as before using the Refresh button and the axes
change to display the R-Z coordinate system.
Click on the Set the region to be a Polygon by clicking on the radio button at the top of the
Region prop- dialogue box. On the Material properties tab set the Material label to 3
erties for a and the Permeability to 100. Other values remain unchanged.
new region
button
Select OK.
The number of subdivisions on each side of the region will be used to create
the finite element mesh at a later stage. At this stage it is necessary to set
up how many subdivisions will be used during the mesh creation, though
these values can be modified later as required.
If the side properties are set at this point this will become the default for all
regions created until it is changed again. For this model a good starting
point is to specify 10 subdivisions (elements) per side.
Biasing can be used to concentrate or grade the elements towards either the
start of a side (bias values near 0) or towards the end of a side (bias values
near 1). Initially the bias will be left at 0.5 to give an even spread of ele-
ments.
Click on the Change the way the subdivisions are specified to be Specify subdivision by
Side proper- number and set the Number of elements to 10.
ties... but-
ton
Select OK.
In order to enter the points used to define the disc region more easily a grid
is set up which enables ‘Snap to Grid’.
Click on the Select Cartesian grid and set the Grid sizes
Grid button for Step in X and Y to 1 followed by OK.
Click on the The first side is created by clicking on the first two corners at (0,7) then at
Create new (10,7).
corner at
cursor (or
grid point)
button
Side three is created by clicking on the forth and last corner at (0,8).
Click on the Finally the forth side is created and the region is completed by clicking on the
Close poly- Close polygon button. Each region is identified by a number displayed at the
gon button centre. The region for the disc is identified by the number 1.
If while creating this region a mistake is made the last point and side cre-
ated can be erased by clicking on the Erase last side button .
Initially the material properties of the region will be set. The settings are a
material label of 1, a permeability of 1 and a current density of 2e6Am−2.
Then coordinates of the region and the side data will be entered.
Click on the Set the region to be a Polygon by clicking on the radio button at the top of the
Region prop- dialogue box. On the Material properties tab set the Material label to 1,
erties for a Permeability to 1, and the Current density to 2e6.
new region
button
Select OK.
The Coil region is entered as before by clicking on the grid to enter the cor-
ner coordinates.
The material properties will be set first, giving this region a material label
4 with a permeability of 300, and resetting the current density to 0.
Click on the Set the region to be a Polygon by clicking on the radio button at the top of the
Region prop- dialogue box. On the Material properties tab set the Material label to 4
erties for a and the Permeability to 300. Reset Current density to 0.
new region
button
Select OK.
For the point definition, selection of existing points from the conductor ver-
tices will be used to quickly enter the first 4 points. Four additional points
will be defined using the previous method, before closing the polygon.
Click on the Position the cursor near the existing corner at (1,6) and click to use this exist-
Pick an ing corner.
existing
corner button
Click on the The remaining 4 points are entered as before using the grid by clicking at
Create new points (5,6), (5,0), (0,0) and (0,6).
corner at
cursor (or
grid point)
button
Click on the The Core is completed by closing the region using Close polygon.
Close poly-
gon button
Initially set the material properties to air and define this to be a background
region.
Click on the Set the region to be a Background by clicking on the radio button at the top of
Region prop- the dialogue box. On the Material properties tab set the Material label
erties for a to 0 and the Permeability to 1.
new region
button
Select OK.
The background region can be entered in the same way as other polygon
regions. As the air region is much larger than the rest of the model, to
ensure that the far fields are captured correctly, the display needs to be
resized. In order to use the grid for creating the background region this also
needs to be resized.
Select OK.
With a larger area now visible the background region can be entered by
clicking on the grid to pick the four corners.
message box appears indicating that points are being added to allow conti-
nuity across the different region boundaries. Click on Yes.
The geometry of the model is now complete, but the mesh should be
checked to ensure that it is suitable and the boundary conditions for the
problem must be set.
The mesh generator now operates on the model data and reports on the size
of the mesh in a message box. The message box also reports on the checks
carried out by OPERA-2d on the model. Two warnings are given indicating
that boundary conditions have not yet been applied and that no BH curve
has been assigned to Material 4.
Close the message box to leave the model outline displayed. The outline
should be checked carefully. Region boundaries between materials of the
same type are not shown, boundaries between different materials are shown
in text colour and external boundaries are orange. If there are any orange
lines within the interior of the model, then the mesh is not continuous and
must be corrected before an acceptable solution can be obtained. If this
Click on the The mesh can be displayed on the model by clicking on Display mesh. This
Display mesh button acts as a toggle (so selecting again will remove the mesh from the dis-
button play).
sions along the axis, however a better method is to use biasing to grade the
mesh towards the centre.
Click on the Place the cursor just inside the background region, at about (1,30), and click
Modify side to select the edge (this does not have to be very accurate as the nearest edge
by picking will be selected).
button Set the Number of elements to 10, and the Bias to 0.05 so that the elements
are grouped towards the start of the edge.
Repeat with the edge along the negative z-axis at around (1,-30). This time
set the Bias to 0.95 so that the elements are grouped towards the end of the
edge.
Click on the The mesh is created and checked as before. The new mesh is displayed show-
Refresh but- ing the grading along the z-axis.
ton
To check the mesh around the coil former and disc it is necessary to resize
the display to be closer to the area of interest.
Click on the
Set view...
button
It can be seen that the mesh is not very good in the Disc (too coarse) or in
the air gap (varying too rapidly). This can be improved by changing the
number of subdivisions along the edges of the Disc and the background
region edge on the z-axis.
The four edges that make up the Disc also need to be changed. Select the inner
edge by clicking at (0.2,7.5) and change the Number of elements to 4.
Repeat for the outer edge by clicking at (9.8,7.5) and again changing the
Number of elements to 4.
Select the upper edge by clicking at (5,7.8) and change the Number of ele-
ments to 35.
Repeat for the lower edge by clicking at (5,7.2) and again change the Number
of elements to 35.
This completes the mesh refinement so the model can be re-meshed and
displayed.
Click on the
Generate
mesh... but-
ton
Click on the The refined mesh shows a much better mesh in the air gap and disc regions.
Refresh but-
ton
The boundaries are set by selecting the boundary condition type and then
clicking near the edge to be set.
The Error check display now shows the position and type of boundary
nodes as well.
Solution Methods
A solution method is needed for the example problem. The analysis pro-
grams available are listed below. Note that only analysis modules that are
licensed are available for use.
• The static analysis module (ST)
• The steady-state ac analysis module (AC)
• The transient analysis module (TR)
• The velocity analysis module (VL)
• The rotating machines module (RM)
• The linear motion module (LM)
• The transient demagnetization module (DM)
• The space charge module (SP)
• The thermal analysis module (TH)
• The transient thermal analysis module (THTR)
• The stress analysis module (SA)
• The lossy dielectrics module (LD)
The problem to be modelled has a constant drive current and hence can be
solved using the static analysis module (ST).
Storing a Problem
It is important to save the model regularly to back up your work while
building the model. This can be done at any time during the pre-processing
stage.
Two OPERA-2d files containing the model and mesh data are created when
a model is saved. These are disc.op2 and disc.mesh respectively. The .op2
file contains all the model data and should be kept. The .mesh file is less
important and can be deleted if storage space is limited (although for an
adaptive analysis, the mesh file must be kept as well as the .op2 data file,
as this mesh data cannot be reproduced from the model data). It is useful as
it avoids the need to re-mesh the model, which can be time consuming for
very large models.
The solution process has created a solution file, disc.st. This contains the
solution data from the static analysis for post-processing.
Post-Processing
Click on the The default settings for a Contour map are set
Contour to produce a plot of flux lines.
map... button
Select Draw contours button to give the con-
tour plot shown on the right.
The model error is calculated and displayed at the bottom right of the
screen on the status bar when the first field quantity is calculated.
Introduction
Since the motor has rotational periodicity (symmetry), only a single pole
pitch needs to be modelled (in this case - 60 degrees). This reduces the
amount of work needed to create the model and reduces the number of ele-
ments and nodes required. Figure 5 shows the complete model.
Pre-Processing
Click on Quit
The units displayed on the information panel of the graphics screen change
to reflect the new unit system.
Note: Make sure this is set correctly at this stage, as all subsequent data will
be defined in this unit system. Changing the units will not change
the overall dimensions of the model.
The axes dimensions also need to be set to be suitable for a stator radius of
5 cm. To do this:
Model information is given on the right hand side of the graphics display.
It can be seen that the default element type is linear. It is recommended to
solve a problem with linear elements initially, so that the model solution
can be validated. After this, quadratic elements can be used to obtain an
improved solution with an associated increase in computational effort and
time.
The material label to be used when drawing the stator regions will now be
set. Material label 5 and a relative permeability of 2500 will be used,
although material characteristics may easily be modified at a later stage.
Region prop- Click on the Polygon radio button and set the Material label to 5 and the
erties for a Permeability to 2500, followed by OK.
new region
Construction lines do not form part of the model but are useful for defining
points within the geometry.
Several other straight construction lines are needed and, as the dialog has
not been closed, the user can continue defining Straight lines.
Start at Start at End at End at
Rotation
X or R Y or Z X or R Y or Z
0 0 5.4 0 10
0 0 5.4 0 0
0 0 4.5 0 4
0 0 2.7 0 7.5
Arc shaped construction lines can also be drawn (these are mathematically
“straight” lines in polar coordinates). This is also from the same dialog.
Several other arc construction lines are needed and, as the dialog has not
been closed, the user can continue defining Arcs.
Origin Origin Start at Start at End at End at
X or R Y or Z Radius Angle Radius Angle
0 0 4.5 0 4.5 60
0 0 2.7 0 2.7 60
0 0 2.5 0 2.5 60
Click on Cancel to leave the Construction Line dialog. Figure 6 shows the
screen after entering these 9 construction lines.
Modify side by The discretisation of this region is poor at the moment. The number of ele-
picking ments on each edge will be modified; bias and curvature will be added where
necessary. Pick the edge near 5.3,0.4 which is then highlighted. Complete
the dialog box as shown and click on OK
.
Continue modifying the side properties of region 1 by selecting each side con-
secutively in a counter-clockwise direction.
Number of elements = 4, Bias = 0.5, Curvature = 0,
Boundary condition = None
These are:-
• Copy
This produces a new region (with a new number). Once created, the new
region may be modified or deleted without affecting the region from
which it was copied.
• Replication
This produces a clone of the original region. Any modifications (includ-
ing deletion) will affect the region from which it was replicated.
Two types of replication will be made. The material surrounding one slot
will be created by mirroring around 10 degrees. Two further copies will be
made by rotating with a 20 degree displacement.
points from region 1 (including its replications). The others are produced
by using COPY.
Region prop- Set the Material label to 1, the Permeability to 1 and the Conductor
erties for a number to 1.
new region
Close polygon
Modify side by Pick the side of region 2 near 2.7, 0.45 and
picking complete the dialog as shown.
Since region 2 has now been drawn, it’s area is known and can be used to
define a current density equivalent to a total current of 600 Ampere-turns.
Modify region Click inside region 2. Modify the Current or charge density to 600/
by picking area, then click on OK.
The second and third conductors may now be created by copying region 2.
The new regions currently have the same material properties as region 2.
The current density and conductor number for both need to be modified.
Modify region Select region 3. Set the Current or charge density to 0 and the Conductor
by picking number to 2.
Click on OK.
Similarly, select region 4. Set the Current or charge density to -600/
area and the Conductor number to 3.
Region prop- The easy direction of a permanent magnet material is defined by its Material
erties for a angle - the angle made with the x-axis in a counter-clockwise direction. In this
new region motor, the magnet is magnetized in the +Y direction. Set the Material label
to 6, the Permeability to 1.1, the Material or phase angle to 90 and the
Conductor number to 0, followed by OK.
Side properties The elements on each side of the magnet will be 0.1 cm long. Set the Element
size to 0.1.
Zoom in (rub- Use the rubber box zoom to make the region
ber box) where the rotor will be drawn fill the display.
The grid should now be shown at 0.1 cm inter-
vals.
Create new Click near the grid point at 2.2, 0.5. Note
corner at cur- that the bottom left corner now presents the
sor or grid coordinate of the nearest grid point, not the
point actual cursor position.
Close polygon
The second region of the rotor is the rotor core which is a soft magnetic
material. This region will be drawn using several different methods.
Region prop- Set the Material label to 4 and the Permeability to 2500. Because this is
erties for a a soft material it is unimportant what value the Material or phase angle is.
new region
Side properties The elements on the first side of the rotor core will be 0.05 cm long. Set the
Element size to 0.05 and click on OK.
Construction Draw two Arc construction lines at radii 2.45
lines and 0.8 cm. between 0 and 60 degrees,
entered on the origin.
Side properties The elements on the second and third sides of the rotor core will be 0.15 cm
long. Set the Element size to 0.15 and click on OK.
Pick an exist- Click near the lower right corner of the magnet
ing corner at (2.2, 0.5), followed by the other 3 corners of
region 5 in a clockwise order.
Side properties The size of the elements will be changed back to 0.05 cm and the curvature of
the next edge given. Set the Element size to 0.05, click on the Curvature
button and enter -1/2.45 in the Curvature box. Click on OK.
The next edge will form part of a periodic (symmetry) boundary and must
match with its paired edge. To match, it is important that the edges have the
same number of elements on both edges of the pair. Rather than using an
element size, the next edge will have the number of elements along it spec-
ified.
Side properties Click on the Set subdivision by number button, set the Number of elements
to 15 and click on the Straight button, followed by OK. It is not necessary to
set the Symmetry boundary condition as this will be set when the Rotating
machine air gap is defined.
Draw straight
sides
Close polygon
The next section to be created is the stainless steel shaft. Since a magneto-
static solution is required and the shaft is non-magnetic, it may be repre-
sented as an air region.
Close polygon
There are several ways to improve the mesh in the magnet. Examples
include:-
• Increasing the number of elements on the edges of the region. This may
be inefficient in this case as the element sizes in neighbouring elements
will increase.
• Subdividing the region into two or more regions. This requires modify-
ing the region geometry or deleting the region to allow two or more
regions to be drawn.
• Changing the shape code of the region (this is only possible within
quadrilateral regions).
In this case, it will be changed from a polygon shaped region (the default)
to a shape H region.
Refresh using
The air gap region was originally developed for the OPERA-2d/RM anal-
ysis. In this program, the rotor regions rotate by an angle corresponding to
their angular speed and the transient time-step. At each new position of the
rotor the mesh in the air gap reforms to make a contiguous mesh. However,
using the air gap region also has two distinct advantages for other OPERA-
2d analysis programs used to model electrical machines.
• The rotor and stator regions can be moved relative to each other by any
arbitrary angle and the mesh easily reforms
• Periodicity boundary conditions are correctly imposed irrespective of
the shape of the model
Building the Two regions with properties of air must be constructed, one on the rotor
other air gap side and one on the stator side. The outer radius of the rotor side region will
regions be 2.47 cm and the inner radius of the rotor side region will be 2.48 cm.
Zoom in (rub- Use the rubber box zoom to display the air gap
ber box) as large as possible.
Enter corner Enter the first point on the rotor side air gap
co-ordinates region in polar coordinates. Select the Polar
tab. Set Absolute values Radius to 2.47
and Theta to 60, followed by Corner and Can-
cel.
Side properties Click on the Specify subdivision by size button and set the Element size
to 0.05. Click on the Curvature button and set the Curvature value to 1/
2.47, followed by OK.
Close polygon
Side properties Click on the Curvature button and set the value of the Curvature to -1/
2.48. Click on OK.
Draw straight
sides
Close polygon
Rotating The air gap is completed by specifying the radius and symmetry for the
machine air gap OPERA-2d/RM gap. The radius must be a value that lies between 2.47 cm
and 2.48 cm as these were the radii of the curved surfaces of the other air
regions in the gap. The symmetry value has two roles - it specifies how
many rotational copies of an equivalent geometry there are in the complete
machine and how the fields are reflected in these copies. The model con-
sists of a 60 degree wedge (1-pole pitch) of the motor i.e. 6 copies of the
geometry make up the complete motor. The polarity of the magnets and sta-
tor circuits is arranged to produce alternate North-South poles i.e. the
model has negative symmetry. This means that the magnitude of the values
of magnetic vector potential on the boundary of the model along the X-axis
will be the same as on the boundary at 60 degrees but the opposite sign.
When the mesh is generated, the appropriate symmetry boundary condi-
tions will be applied.
There will be no change to the geometry but the information box on the right
side of the screen will now read 9 regions (+ Gap)
Both curves will be loaded from the directory, <Install Dir>/bh. BH curves
are a table of points with values of flux density against values of field
strength. A valid curve must contain at least 5 points, so before any curves
have been specified, the dialog says that the curve is currently invalid.
Mesh generation
The data is now complete and the mesh can be generated. Meshing per-
forms several check functions to ensure that the data to be submitted to the
analysis modules is correct.
The mesh is generated. The timer bar reports progress. The graphics dis-
play is a line drawing of the model. The outer boundaries of the model will
be drawn in the highlight colour (default is orange) with the interior model
drawn in white.
Any orange lines in the interior of a magnetic model indicate the mesh is
not continuous. These errors must be corrected before submitting the
model for analysis.
The display also shows the boundary conditions that are imposed on the
model. The rotating machine air gap region has imposed the symmetry
boundary conditions on the two radial boundaries of the model.
The information box that is also displayed should indicate 0 errors and 0
warnings. Any errors reported must be corrected, while warnings may
need to be corrected. Clear this box by clicking on Close.
The completed mesh, including the rotating machine air gap, can now be
viewed.
Refresh
and zoom
to bounding
box
Analysis Data
The final data that must be specified before the analysis can be run is the
analysis data. The first analysis of this model is a statics (ST) analysis. Data
specifically needed by the ST module includes:
• Linear or nonlinear material properties
• Number of iterations and convergence tolerance for nonlinear analysis
• Type of nonlinear iteration (modified Newton-Raphson or simple
update)
Close the information box. The model must now be saved for analysis.
Save Set the File name to pmm (or use the browser) and click on Save and analyse
now. This will also start the analysis automatically.
OPERA-2d then reports on the progress of the solution, and when com-
pleted, will close the window automatically.
Post-Processing
The units system in use at the time the pmm.op2 file was written will still
be set. The CGS electromagnetic units (gauss for flux density, oersted for
field strength, siemen/cm for conductivity etc.) are commonly used but it
is probably more convenient to modify the units of energy, power and force
to the SI system.
Set units Click on the Mechanical Properties tab. Select the newton button in Force,
joule in Energy, watt in Power and kilogram in Mass. Click on OK.
Contour map Click on Draw contours to accept the default style and component for contour
plotting.
The plot should be checked to see if the flux pattern agrees with the
expected magnetic performance of the motor. If errors appear to be present
then the model should be modified to correct the error and the analysis re-
run.
The contour map does not show many levels within the range of flux den-
sity magnitudes (about 0 to 30 kGauss). Also, the range represented by
each colour is not easy to compute. Both these can be improved so that each
colour will now exactly represent a 1 kGauss band.
Line Integral
To calculate the total force and torque on a body, in this case the torque on
the rotor, an integration surface around the body is defined. Working in two
dimensional space, OPERA-2d calculates the total force and torque on a
body by summing the results of a series of line integrals encircling the
body. For the case of XY symmetry, the calculation is based on a unit length
distance in the third dimension. The line integrals must enclose the body on
which the force and torque are to be calculated in a counter-clockwise
direction. The torque is evaluated with respect to an action point. In this
model, the x and y forces and the anti-clockwise torque for one pole of the
rotor are evaluated around the (default) action point (0,0).
Line integrals can be performed in two ways - using a fixed step between
each evaluation point or adaptively by ensuring that the error at each step
is below a user specified tolerance. In this example, the second method is
used with the relative error of the integral at each step set at 0.01%. The
evaluation points used in the integral are displayed on the screen.
Field integrals Set the Starting angle to 0, the Final angle to 60 and the Radius to
around a circu- 2.475. Set the Number of steps or relative error to 1e-4 and click on
lar arc OK.
An information box opens showing the results of the line integral. The val-
ues of force and torque computed are per cm. length in the Z-direction.
Note that the line integral also integrates the Expression in the dialog.
The torque due to the one pole of the motor modelled may be found by:-
• The product of the torque given in the message box (system variable
TORQUE) and axial length (L) of the motor (since torque is calculated
as per unit length) i.e.
In this example, the motor has six poles (one was modelled) and is 15.3 cm
long. Thus the total torque equation would be
Total motor torque = TORQUE*15.3*6
User variable Set the Name to #tt and the Value to torque*15.3*6 (no spaces). Click
calculator on Set constant.
An information box reports that the total torque for the motor is about 1198
N-cm.
The variable name can be up to 16 characters long, but the first character
must be the # symbol. The expression uses a FORTRAN-like syntax.
Hz, 6-pole motor), the mechanical output power can be computed from the
product of the total torque and the angular frequency (in radians/second).
User variable Set the Name to #power and the Value to #tt/100*1200/60*2*pi (no
calculator spaces). The division by 100 converts the torque to N-m, so that the answer is
in watts. Click on Set constant.
This concludes the magnetostatic analysis of the six pole permanent mag-
net motor example. In another worked example in the OPERA-2d User
Guide, the same motor is also modelled dynamically using the OPERA-2d/
RM analysis software.
AC Analysis Case 1
In this chapter the disc example discussed in Getting Started [page 17] will
be used again and analysed with the AC and the TR solvers. Two AC cases
will be discussed, one without and one with eddy currents, and also a tran-
sient case. The forces on the disc will be calculated. The forces can be com-
pared with the results of the ST solver as well.
The file of the model created previously will be used. For convenience the
file disc.op2 is available in the folder <Install Dir>/examples/2d.
Then refresh
For the first AC case the frequency is set to 100 Hz. If the “Getting Started”
example has been done, a file called disc.mesh has been created automati-
cally in the working directory. In this case the mesh does not need to be cre-
ated again. Otherwise the mesh has to be generated as follows.
The new analysis can then be prepared. It will be an analysis with linear
material properties and a frequency of 100 Hz.
Close the dialog by selecting OK and Close the information box that appears.
The model must now be stored for and the analysis started.
Save Set the File name to disc_ac100_nocond.op2 (or use the browser) and
click on Save and analyse now.
Post-Process Case 1
When the analysis is completed, the results in file disc_ac100_nocond.ac,
which was created above.will already be loaded.
This is now a steady state cosine solution with a frequency of 100 Hz. The
results can be obtained at any angle around the AC cycle. For example the
Force Calculation
To calculate the total force on a body, in this case the force on the disc
(region 1), an integration surface around the body is defined. Working in
two dimensional space, OPERA-2d calculates the total force on a body by
performing a series of line integrals encircling the body. The line integrals
must be in an anti-clockwise direction.
Field integrals On the X or R directed tab, define a line starting at r = 0 to r = 10.5 at z = 6.5.
along a line Set the number of steps to 1000 with a curvature of 0. Leave the Field com-
ponent Expression to be Bmod. Set the AC time to Time average. For the
Accumulator option (at the bottom of the dialog) set Zero then add. Click
on OK to accept.
A message box pops up and displays the integral of the z-component of the
Maxwell stress for the line defined above.
To find the total force, two more line integrals must be performed and the
results (integral of z-component of Maxwell Stress) must be added
together. The addition can be automatically performed by making use of
the Accumulator feature in the dialog which adds the results to a total after
each line integral is completed. The second line integral would be from
z = 6.5 to z = 8.5 at r = 10.5.
Field integrals On the Y or Z directed tab, define a line starting at z = 6.5 to z = 8.5 at
along a line r = 10.5. For the Accumulator option set Add. Click on OK to accept.
The third line integral would be from r = 10.5 to r = 0 at z = 8.5. Note that
the direction is important: the disc has to be surrounded anti-clockwise.
The integral of Maxwell stress is shown each time the line integral is used
as well as the accumulated integrals when the accumulator is used. It is not
necessary to integrate along the Z-axis since the three integrals describe a
cylinder that completely surrounds the plunger.
Comparison The total force on the disc can also be calculated for the ST solution in Get-
ting Started [page 17]. The same steps have to be followed as explained
above. The total force on the disc for the static case is Fz = -7.1895 N. This
is twice the value of the AC solution. The reason for this is that the peak
current in the AC solution is the same as the value used in the ST solution.
1
The RMS value of the current is therefore ------- of the value in the static case,
2
and for a linear problem the force is proportional to the square of the cur-
rent.
Tip For a different model, for example if the disc is touching the surface of the
iron, the integration path has to be defined right on the interface between
the disc and the iron. In this case a message box will appear with a warning
that the “integration path is not all in air”. The program correctly performs
the integration by evaluating the Maxwell stresses as if the line is a vanish-
ingly narrow cut in the material. It is recommended that line integrals be
performed in air regions where possible, for improved accuracy.
AC Analysis Case 2
The second case features an electrical conductivity in the disc. The other
material remains unchanged. The induced eddy currents in the disc are
opposite to the primary field and therefore weaken it. It is to be expected
that the total force on the disc will be smaller.
With the file disc_ac100_nocond.ac still loaded the second case can be pre-
pared. The material properties of the disc (region 1) are changed from
material label 3 (default colour dark blue) to material label 9 (default colour
khaki). This makes it easier to recognise the model with the conducting
disc.
Modify region Pick the disc region with the mouse (see right).
by picking Change the Material label from 3 to 9. Change
the Conductivity from 0 to 6 E+06.
Click on OK to accept.
6
With a conductivity of 6 ⋅ 10 S ⁄ m , a permeability of 100 and a frequency
of 100 Hz there is a skin depth of 0.205 cm. The equation for the skin-depth
δ is:
1
δ = ---------------------------------------- (5.1)
π ⋅ f ⋅ µ0 ⋅ µ ⋅ σ
For the disc example as discussed above the mesh has to be refined to meet
these requirements. Region 1 (the disc) will be changed from a “polygon
shape region” to a “regular quad shape H region”. Note: this can only be
done if the region is bounded by 4 sides.
Modify side by Click at (0.1, 7.5) and set the Number of ele-
picking ments to 8 and the Bias to 0.9. Close the
menu with OK.
The case 2 model featuring the conducting disc will be solved with the
same analysis options (linear solution with 100 Hz). Therefore no further
options will have to be set before saving the model.
Post-Process Case 2
The file disc_ac100_withcond.ac, which was created above, will be auto-
matically loaded.
The total force on the disc is evaluated in the same way as before in case 1,
using time average values.
The total force on the disc for the second case is Fz = -3.02 N. The value is
smaller compared to the first case due to the eddy currents.
This finishes the AC example. The current session should be left open with
the model loaded ready to run the TR analysis.
TR Analysis
Modify region Pick the coil region with the mouse (see right).
by picking Change the Conductor number from 0 to 1.
The modify operation has deleted the mesh. Therefore the mesh should be
generated again, before the analysis options will be set.
Generate mesh
It is recommended to run the TR analysis for a few AC cycles until the ini-
tial transients have decayed, and then compare the results with the previous
AC results. In this case 4 AC cycles are set up.
Model -> Analysis Options -> Transient Time Varying Analysis (TR)
Ensure the default setting of Linear material
analysis is selected.
Having closed the “Transient analysis” window, the model can be saved
and then solved.
To end the current session now and close the Pre and Post-Processor:
When a final design has been obtained, it is often necessary to examine the
motor’s performance dynamically. The OPERA-2d/RM analysis has been
designed for this. A time-stepping solution to the transient electromagnetic
equation is obtained allowing the rotor to rotate by the appropriate angle at
each time step. For more information see the OPERA-2d Reference Man-
ual.
Set units The pmm.op2 model was constructed in CGS units. For this example, it is bet-
ter to use the more familiar SI units for mechanical properties. Click on the
Mechanical properties tab. Select the newton button in Force, joule in
Energy, watt in Power and kilogram in Mass. Click on OK.
Zoom bound-
ing box
External circuits
The magnetostatic model used defined current sources in the stator slots to
provide the armature mmf in the model. In the RM model, the armature
windings will be connected to a 3-phase power supply using external cir-
cuits. OPERA-2d/RM will determine the currents in the windings by solv-
ing the circuit equations in conjunction with the electromagnetic field
equations. The first task is to remove the defined current sources from the
stator winding regions.
given a conductor number. To make it easy to define the circuits, these can
be displayed instead of the region numbers.
Set view Click on the Region labels tab and the Con-
ductor numbers button (coordinate limits
were set when zooming to the bounding box
earlier so do not require changing), followed
by OK.
Three circuits are to be defined for this model (representing the three
phases). The coils will be considered to be filaments i.e. there are sufficient
turns in each coil that the distribution of current in each turn can be consid-
ered uniform and eddy current effects can be ignored. In a Cartesian coor-
dinate model, the length of the circuit in the Z-direction must be given (in
this case, the same length as the motor = 15.3 cm) and the connections of
the conductors making up the circuit to other conductors in the images
implied by the boundary conditions. In this motor, we shall assume that
there is only one “parallel” path in each phase i.e. each phase has a conduc-
tor connected in series in all the images. The values of the passive compo-
nents and voltage supply external to the model are also given as global
parameters for the circuit.
When these global parameters for the circuit have been defined, the user
also specifies the characteristics of each conductor making up the circuit.
These are the number of turns, the resistance / unit length of the wire and
the direction (go or return).
Set the Conductor number to 1. Set the Number of turns to 5, the Resistance
per unit length to 2e-3 and click on the Go button.
Click on Add followed by Cancel. An information box appears saying that the
first circuit has been stored. Clear this box to define the second circuit.
The global parameters for all 3 circuits are identical - so the user may click on
Create new circuit immediately. Note that the expression 110*sqrt(2)
has been evaluated.
Using the pull down button on the right of the Conductor number box, select
2. Click on the Return button and set the Number of turns to 5 and the
Resistance per unit length to 2e-3. Click on Add followed by Cancel.
List
Click on the Output times tab. Click on Change to specify the times when a
complete solution of the model is to be output to the solution file.
The motor will be close to synchronism after about 0.1 seconds. Several other
output times will also be specified. This allows the analysis to be restarted
from any of these times if, for example, a different mechanical load was
required after a certain time. Set the New output time to 0.005. Click on
Add. Repeat for times 0.01, 0.02, 0.05 and 0.1. Click on OK.
Click on the Drive tab. The three circuits will be specified as cosine drives at
60 Hz. with different phase angles. Click on Change to set these.
The phase angle of circuit 1 is 150 degrees i.e. the voltage applied to the circuit
will reach its peak positive value at ωt = 150° . Click on Circuit 1 and the
Cosine Drive type button. Set the Frequency to 60 and the Phase angle to
150. Click on Apply to update the definition of the Function for Circuit 1.
Circuits 2 and 3 are similarly defined except that their phase angles are -90
and 30 degrees respectively.
Click on the Logging tab. The currents in the 3 circuits will be monitored dur-
ing the transient, as well as the electromagnetic torque, the speed of the rotor
and its angular position. Click on Change to add these system variables to the
logged values.
The name of the log file will be derived from the name of the analysis data
.op2 filename. Click on the From data file name button. In the Component
fields 2 through 7, enter i1, i2, i3, rmangle, rmtorque and rmspeed.
Click on OK.
Within the last category, two methods can be used to define the rigid body
dynamics equations. Using the user specified moment of inertia, the angu-
lar acceleration of the rotor is evaluated by the program from the sum of the
computed electromagnetic torque and the applied load, friction torque and
variable speed torque. Alternatively, the user can create a command input
(.comi) file which computes a user defined constant called #ACCEL using
any available system variable such as RMTORQUE, RMSPEED etc. The
latter option allows sophisticated control of the machine such as speed gov-
ernors or time varying load torque. In this example, however, the simple
torque / angular acceleration equation will be used.
The motor will be started without any applied load. Only a friction load
torque will be specified. The friction load always acts in the opposite direc-
tion to the rotation. Note that the torque values specified are total values for
the motor.
Click on the Rotation tab. Click on the Mechanic coupling button. Set the
Magnetic length of the rotor to 15.3 (cm), the Moment of rotor iner-
tia to 1 (kg cm2) and the Frictional torque to 200 (N-cm). Click on OK.
The analysis data is now complete. The data file can be saved and the anal-
ysis executed.
Save Set the File name to pmm rotating (or use the browser). Click on OK.
The file name should already be in the dialog. Click on OK to start the analysis.
The Pre and Post-Processor can still be used while the analysis proceeds.
Analysis
The transient RM analysis will start. As the voltages in some of the circuits
will be non-zero at time t = 0, the program solves an initial DC nonlinear
problem using a current in each circuit computed purely from the total cir-
cuit resistance and the instantaneous value of the applied voltage.
At any time during the analysis, the log file which is monitoring the cur-
rents in the circuits and the mechanical properties can be examined. The log
file is a text file with columns of numbers showing the instantaneous values
of the monitored expressions. It can be copied into a spreadsheet or docu-
ment. It can also be examined using the graph drawing facility in OPERA-
2d/PP. The columns of the file are addressed by system variables COL1,
COL2 etc. (up to COL7 in this case). The default is to draw COL2 against
COL1, which in this case will correspond to drawing the current in circuit
1 (i1) against the transient time (ttime).
The graph shows the current in circuit 1 up to about the first 2 msec. of the
transient. The GRAPH command loads all columns from the log file when
it is executed. Any other expression involving the COLn system variables
can also be plotted by executing another Graph data in text file
instruction. However, the .log file must be re-read if the variables are to be
updated to the transient time which the analysis has reached.
Figure 8 shows the currents in the 3 circuits for the duration of the analysis.
The waveforms over the last cycle or so are becoming almost identical in
each phase indicating that the transient effects from starting are decaying
and the motor is approaching synchronous operation.
This is further demonstrated by plotting the angular speed (in RPM) and the
angle of rotation (in revolutions) against time, as shown in figures 9 and 10
respectively. The rotor is settling to its synchronous speed (-1200 RPM)
and the angle of rotation is tending to increase (in a negative direction) lin-
early with time. If the motor had attained synchronism instantly, it would
have complete 2 complete revolutions in the first 0.1 seconds but the initial
transients - especially the small excursion in the opposite direction at the
beginning of the transient - mean that it has only completed about 1.8 rev-
olutions. l
Post-Processing
The graphics display shows the displaced position of the original rotor pole
at this time during the transient. Note also that the speed (in RPM) is dis-
played in the information box.
Flux plots
Note how the periodicity boundary conditions have also been enforced azi-
muthally in the air gap so that contours leaving the stator side at the gap re-
appear at the rotor side. In fact, the rotor side contours are the negative of
the stator side because of the negative symmetry condition.
Contour map
Note that the rotor and stator are now completely detached from each other
but the rotating machine gap region still enforces the periodicity.
Currents in circuits
The pmm rotating.res file created during the OPERA-2d/RM analysis con-
tains the currents in the circuits at the output times. The values for the
loaded case can also be obtained in the Pre and Post-Processor by listing
the circuits.
List
The program also translates the currents into current density values in the
regions making up the circuits.
Although the current in circuits 2 and 3 are of opposite sign, the current
density is the same sign because the conductor in circuit 2 is a Return
while circuit 3 is a Go.
Stress
Stress is force per unit area. There are three different forms of stress. In
each of the following cases we assume unit thickness normal to the paper.
F
α
W F F
Y
L δ L δ
W
X
(A) Tension (B) Compression (C) Shear
Figure 11 Definitions of stress
(a) Tension
f
Tensile stress (force per unit area) σ x = ----
w
σx
Strain (dimensionless) ε x = ------
E
Extension δ=Lεx
By convention tensile stress is positive.
(b) Compression
f
Compressive stress (force per unit area) σ x = ----
w
σx
Strain (dimensionless) ε x = ------
E
Compression δ=Lεx
By convention compressive stress is negative.
(c) Shear
f
Shear stress (force per unit area) σ xy = ----
w
σ xy
Shear strain (radians) α = --------
G
In a right-handed axis system, positive shear is defined by the right-
hand rule, i.e. α is positive anti-clockwise measured from the undis-
placed shape. The shear in the diagram is negative.
Plane stress and Plane stress may be seen by considering Figure 11 (a) for tension.
plane strain
problems If the rectangle represents a thin sheet of isotropic material which is not
prevented from contracting in the direction normal to the paper when it is
stretched, then a state of plane stress exists.
If the thickness were infinite in the direction normal to the paper, then each
plane parallel to the paper would be constrained to remain plane, and a state
of plane strain would exist. Stresses would be developed in the direction
normal to the paper, their magnitudes being determined by the Poisson’s
ratio of the material.
In practice plane stress and plane strain rarely exist in isolation and the sit-
uation will be somewhere between the two extremes.
Note:
• Plane strain will usually result in higher calculated stresses.
• Plane stress will usually result in higher calculated displacements.
Poisson’s ratio Poisson’s ratio is defined as the ratio between transverse strain and longi-
tudinal strain when plane stress exists:
δ2/2
W L δ1
δ2/2
Figure 12 Definition of Poisson’s ratio
δ1 δ2
From Figure 12, it may be seen that ε 1 = ----- and ε 2 = ----- . This gives
L W
ε2
Poisson’s ratio as ν = -----
ε1
Assuming the rz plane lies in the surface of the paper, then circumferential
stress and strain are normal to the paper.
where
The applied forces may be specified, together with any known fixed dis-
placements.
The global stiffness matrix and vector of forces are then formed.
Modelling
Stand alone Using the stress analysis as a stand alone module allows the mesh devel-
analysis oped to be refined where the gradients of the stress (and displacement) are
expected to be large.
Coupled If electromagnetic results are to be coupled to the stress and thermal anal-
problems ysis modules, it is necessary to transfer information such as Lorentz forces
and Joule heating from electromagnetic solution to the structural and/or
thermal model. In these cases the mesh developed for the electromagnetic
analysis is used for the other two analyses.
The electromagnetic mesh may have been refined in order to obtain accu-
rate estimates of magnetic fields. This may not be adequate in regions
where high stresses prevail, such as holes or notches which may not be
highly discretised for the electromagnetic solution. A compromise is usu-
ally needed.
This means that the effects of the way in which a force or constraint is
applied tend to be localised. It is this principle which allows a force to be
applied as a point load, for example, since in theory a point load would
inflict an infinite stress at the point of application (in practice it is impossi-
ble to apply a force at a point).
Finite Element The finite elements used are triangles, with linear or quadratic shape func-
Mesh tions as for the electromagnetic analysis.
Elements with large aspect ratios are to be avoided. The automatic mesh
generator attempts to form elements with small aspect ratios.
Aspect ratios greater than 10 should be used with care. Check the solution
carefully to see whether it is a realistic result.
The problems caused by high aspect ratios are that the elements may be
incapable of correctly representing local changes in strain and stress. Also
an ill-conditioned set of equations may result. When solved these will give
a completely incorrect set of displacements.
This simple beam is being bent by moments applied to its ends, and which
is subdivided into a single layer of triangular elements as shown in
Figure 14.
compression
strain
tension
The strain in such a beam varies linearly across the depth of the beam, as
shown in the right hand diagram of Figure 14. A linear triangular element
is only capable of representing a constant rate of strain within it (which is
why it is often called a constant strain element)
It cannot represent the variation of strain across the beam. Ten or more lay-
ers of elements may be needed in such cases.
Material Properties
The material type determines how much information must be provided to
the Pre and Post-Processor.
Anisotropic Materials may have different properties in three orthogonal directions (the
materials principal axes, set at some angle to the x, y and z axes of the problem).
νij is defined as the ratio of strains in a plane ij under plane stress (ratio
strain in direction j to strain in direction i for stress in direction i).
The following types of material may be defined in the Pre and Post-Proc-
essor.
Material properties can vary quite widely depending on the history of the
material. When a metal ingot is cast, its Young’s modulus and yield
strength are usually quite low; they improve dramatically when the mate-
rial is worked (for example, rolled or drawn). Values quoted in material
standards are usually minima for material which has been subjected to nor-
mal working.
Constraints
Sufficient constraints must be applied to the model if a valid solution is to
be obtained from a structural analysis. In 2D only 3 degrees of freedom are
possible for a rigid body: translation in x and y directions, and rotation
about the z axis; therefore the minimum number of degrees of freedom
requiring constraint is three. However, these must be chosen such that the
translations and rotation are prevented. A 2D model with 1000 nodes will
have, after minimal constraint has been applied, 1997 degrees of freedom
and will give rise to a set of 1997 simultaneous equations.
Pressures
To specify pressures (which are normal to surfaces), the method used by
OPERA-2d is general.
The surface traction is specified. These are the components in x and y (or r
and z) directions of forces per unit area of surface. The force may be at any
angle (even parallel) to a surface.
Thermal Expansion
It is possible to specify a thermal expansion integral for a material. This is
the positive value given by:
T2
∫T1 α dT (7.3)
Failure Theories
There is no one theory which may be used to predict failure of a stressed
component. The choice of theory will depend not only on the type of mate-
rial, but on the way in which it is used, the nature of the loading etc. Mate-
rials which are subjected to cyclic stresses (and electromagnets may be in
this class) may fail due to fatigue; this type of failure is extremely complex
and may require a great deal of information about the material and the
stresses, temperatures and even environment to which the component is
subjected (for example, the presence of alkali or acid or even just air may
affect the stress and/or time at which failure occurs). Particularly in brittle
materials (or materials made brittle locally by temperature or fatigue) fail-
ure may start at a notch or crack. For steady stresses the common theories
are:
• maximum stress failure occurs when the maximum stress (usually ten-
sile) in the material reaches a limiting value.
• maximum strain failure occurs when the maximum strain reaches a lim-
iting value.
• maximum shear failure occurs when the maximum shear stress reaches
a limiting value.
• strain energy failure occurs when the strain energy reaches the strain
energy at failure in a tensile test.
• shear strain energy (or Von Mises theory) failure occurs when the shear
strain energy reaches a limiting value determined by a tensile test.
The limiting value may sometimes be taken as the stress (or strain energy)
when the material fails completely in a tensile test, but it is probably more
useful to use the value at yield (which is when the material ceases to be
elastic, and some deformation is permanent).
The stresses computed by OPERA-2d may not be the maximum. The max-
imum stress will be the larger of the two principal stresses, which are
obtained from the x and y stresses by:
2 2
σ x + σy σ x – σ y 2
p = ------------------ ± ----------------------- + ρ xy (7.4)
2 4
The maximum shear stress is half the difference of these two principal
stresses.
In axisymmetric problems the circumferential stress is the third principal
stress, and is likely to be greater than that calculated above.
Much of the above assumes (tacitly) that materials fail in tension. This is
not always the case, since high compressive stresses lead to high shear
stresses which may cause failure. This may be the mode of failure for a
short column.
However, a long column or strut will fail due to instability (i.e. applying an
axial load to a thin wooden or metal strip, when the load is increased the
strip will suddenly bow).
The designer avoids sharp re-entrant corners by introducing fillets into the
design. However there may be little control over scratches and surface
imperfections. The reason that these imperfections have little effect is that
in ductile materials (most metals) a high local stress causes local yielding,
and this local yielding relieves the high stress. Because it is tensile stress
that causes failure in this way, highly stressed components are often sub-
jected to surface treatments such as polishing, rolling or shot-preening,
which by reducing scratches and introducing compressive surface stresses,
offset the tensile stresses at imperfections.
This example of a simple beam will illustrate the use of the Pre and Post-
Processor with the stress analysis module. Methods of building the model,
solving and post-processing are given including:
• Mechanical material properties
• Constraint conditions
• Load conditions
• Tables for pre and post-processing
• Display of deformation
The beam is assumed to be infinitely long in the third (z) direction e.g. a
shelf or platform, fixed at the y axis and loaded at the far end. The model
is made up of a single region (see Figure 16).
Pre-processing
Setting the Pre Launch OPERA-2d and enter the Pre and Post-Processor in the normal
and Post- way. In this example the default SI units will be used.
Processor
Environment
Entering the The material definition for the beam is as shown below:
polygon region
Select the Select Polygon option, and set the Material label to 3.
Region Prop-
erties button
Enter the last vertex of the polygon and subsequently close the polygon as
instructed below:
Enter the subdivision data one by one as shown below. The user can select
the appropriate subdivision or click on Other to enter this manually.
Having entered the appropriate subdivision for a particular side, hit Pick
to select the side, based on the information given in the Table below. Hav-
ing entered the data for all sides, select Quit. The subdivision of each
side is shown below:
(0,7) (10,7) 50
(10,7) (10,8) 3
(10,8) (5,8) 16
(5,8) (5,8.5) 3
(5,8.5) (0,8.5) 16
(0,8.5) (0,7) 5
Generate the finite element mesh by clicking on the Generate mesh button,
, and note that there is a warning relating to potential settings.
Entering the It is necessary to define the mechanical material data constants for the
mechanical beam. To do this select
material data
Enter values for Young’s modulus (2.1e11) and Poisson’s ratio (0.29).
A message box appears informing the user that material data has been
checked successfully. The Shear Modulus is automatically calculated and
now appears in the box. Select Cancel twice to exit all dialog boxes.
Post-Processing
Loading the Read in the solution file beam.sa. You will notice that the solution tables
results into the available are also read in automatically. A message box is displayed which
Pre and Post- informs the user of the new variable names included in the solution file.
Processor
Set the viewing options as follows:
Select the Set Click on the Region Style tab and select No
View button filling, material outlines:
Displaying the The deformation due to loading may be viewed as a deformed mesh.
results as a
deformed mesh
Thermal Analysis
There are two thermal analysis modules available: steady-state and tran-
sient.
The transient thermal analysis module will start from a specified tempera-
ture distribution and will model the change in temperature with time as a
heat source is switched on. The output from the transient analysis at differ-
ent times can be stored and studied within the post-processor.
Modelling
Once the geometry of the model has been specified, there are areas where
additional information may be needed.
• Material properties
The thermal conductivity needs to be given for each material included
in the model. The thermal conductivity may be anisotropic and oriented
in any direction if required. For transient analysis problems, the heat
capacity and density must also be given. Any air regions (material 0)
within the thermal model are ignored by the analysis and the tempera-
ture within them will be undefined.
• Boundary conditions
The default boundary condition is a thermal insulation boundary, imply-
ing that no heat will pass through the surface. The thermal insulation
boundary condition will be imposed on all interfaces adjoining air
regions, such that no heat passes from one material, through the air, to
another material. Two other boundary conditions are available: fixed
temperature condition, and heat transfer conditions. The heat transfer
condition allows a heat flow dependent upon heat transfer coefficient
and temperature difference with an ambient temperature, combined
with a fixed heat flow through the boundary. The boundary conditions
do not allow true modelling of losses through convection or radiation.
• Heat source density
This is given in a table of values at the centroid of each element within
the model. This allows a variation of heat source density to be input, e.g.
calculated from the eddy current and iron losses in an electromagnetic
simulation.
• Initial temperature distribution
This is only needed by the transient module, and is given in a table of
values at the nodes of the finite element mesh. This can be generated
from a previous steady-state or transient analysis, where a change in the
heat sources is being modelled.
This example of a simple Heat Bath will illustrate the use of the Pre and
Post-Processor with the steady-state thermal analysis module. Methods of
building the model, solving and post-processing are given including:
• Thermal material properties
• Fixed temperature conditions
• Thermal flow conditions
• Display of temperature distribution
The bath is assumed to be infinitely long in the third (z) direction. The ther-
mal analysis is driven by a fixed temperature at the right hand end. The
model is made up of two regions as shown in Figure 19.
Pre-processing
Setting the Pre Launch the Pre and Post-Processor in the normal way. In this example the
and Post- default SI units will be used.
Processor
Environment
Entering The material definition for the bath is shown below for the two regions of
polygon region 1 the model. Region 1 is defined as follows:
Select the Select the Polygon option, and set the material label to 3.
Region Prop-
erties button
Click on OK to confirm.
Draw -> Next Corner in Polygon -> Corner and Side Properties
Fill in the cartesian coordinates X and Y (0,7):
For subsequent vertices, the coordinates are entered together with the side sub-
divisions.
The next vertex will be at (10,7), and the number of elements set to 50. Click
on Corner to enter the new vertex.
After entering the last vertex of the polygon, set the subdivision for the
final polygon edge as follows:
Select the Select the Polygon option, and set the material
Region Prop- label to 5.
erties button Click on OK to confirm.
(9,9)
(9,8)
(0,8)
Following this, define a new vertex and close the polygon as instructed
below:
Draw -> Next Corner in Polygon -> Corner and Side Properties
Enter the new vertex coordinates (0,9) and set
the number of elements to be 5:
Select Corner.
Finally, Close polygon with 20 subdivi-
sions.
Generate the finite element mesh in the normal way, and note that there are
warnings relating to potential settings and BH data.
Entering the It is necessary to define the thermal material constants for the bath. To do
thermal material this select
data
The Stress and Thermal Material Properties dialog appears and should be
completed as shown below:
Select the Thermal option, and set Thermal conductivity values k11 and k22
to be 63, followed by OK.
Hit Cancel having confirmed that the material was successfully checked.
Repeat the procedure for Material number 5, assigning a Thermal Conduc-
tivity of 400 W/m/K for both x and y components (k11 and k22).
Entering the A fixed temperature boundary condition may then be entered as follows:
Thermal
boundary
conditions
Now select Pick Side and mouse click at (5.0, 7.1) and (9.9, 8.0) to pick
the bottom and right hand side of the bath.
Select Heat transfer and complete the parameter boxes with heat flux (50),
heat transfer coefficient (1) and temperature (120) values and apply to side at
(5.0, 8.9) - top face of the melt.
Completing the The Thermal Analysis data can now be prepared and analysed. .
Pre-processing
In this chapter the disc example discussed in Getting Started [page 17] is
used again. Two examples of coupled analyses are shown:
• mechanical stresses caused by electromagnetic forces
This example uses the same finite element mesh for both types of anal-
ysis. The results from an electromagnetic analysis are used to calculate
a table of body forces; the material properties and boundary conditions
for a stress calculation are then defined with the same finite element
mesh.
• temperature distribution caused by joule heating
This example shows how a different mesh can be used for each analysis.
The TABLE EXPORT command is used to create a file containing the
positions where the heat sources are required, this file is then PROC-
ESSed using the electromagnetic solution to calculate the joule heat-
ing. The heat source table is then IMPORTed into the thermal model.
The model created previously will be used. For convenience the file
disc.op2 should be copied from the folder <Install Dir>/examples/2d.
Calculating the Launch the OPERA-2d Pre and Post-Processor and open the model.
electromagnetic
solution
Calculating the The magnetostatic solution can then be used to calculate the body forces
electromagnetic acting on the plate.
body forces
Refresh
OPERA-2d Tables are used to transfer body force data to the stress analysis
module. In this example the ferromagnetic plate above the E-core will be
treated as a diaphragm restrained on its outer circular edge, it will therefore
be bent downwards towards the E-core by the electromagnetic forces.
The Stress analysis module looks for body force information in tables that
must be called XLOAD and YLOAD for XY symmetry models; or RLOAD and
ZLOAD for axisymmetric models. The Tables menu contains options to
Create Standard Tables of loads (Lorentz forces, J × B ) and Maxwell
Stress for the complete model. However in this example the dialog is used
to illustrate the parameters used in creating the tables.
This example uses the same mesh for both the magnetostatic and stress cal-
culations. OPERA-2d tables of radial and axial body forces are calculated
from the magnetostatic solution, and the additional material and boundary
condition data is added for the stress simulation. The stresses are only
required for the plate, therefore the tables only need to be created for region
1.
The units defined for tables should correspond to the units of the OPERA-
2d variables used to calculate the values in a table (i.e. if the expression is
J**2/SIGMA the units should be DENSU**2/CONDU).
Defining the To calculate mechanical stresses the physical properties of the plate mate-
data for stress rial must be defined (note that the material is defined as Axisymmetric, and
analysis. that the in-plane and circumferential data could be different).
Select OK to confirm that realistic values have been entered, then Cancel to
return to the material dialog and Cancel again to exit.
Saving the model The model can now be saved for stress calculation and the mechanical
and calculating stresses calculated
mechanical
stresses.
Displaying stress When the stress analysis has completed, the file containing the results can
results. be viewed
Refresh
As the file is opened the additional results from the stress calculation are
reported in an information box. Contours of the stress (SIGMAR, SIG-
Calculating the In this example the power loss caused by eddy currents will be used as a
Electromagnetic heat source for a transient thermal analysis. Launch the OPERA-2d Pre and
fields Post-Processor and open the model.
Select refresh
The plate above the E-core is defined as a solid block of conducting iron in
this example. A steady state AC electromagnetic field analysis is per-
formed to calculate the distribution of power loss in the plate.
Modify region Select the Modify region by Picking tool button and then pick the plate
by picking (region 1).
Specify its Conductivity as 6e6.
The mesh may need to be generated (if it was not stored with disc.op2).
Creating the Export, processing and import of tables will be used in this example. The
model for model for thermal analysis could therefore be completely different. To
transient thermal show this, erase everything except the plate (region 1) from the electromag-
analysis. netic model (the discretisation could also be changed if desired).
Select OK to confirm that realistic values have been entered, then Cancel to
return to the material dialog and Cancel again to exit.
Creating the The facilities for exporting, processing and importing tables will be used to
Heat source create the heat sources for this transient thermal analysis. The four steps
tables. required to complete this are:
• Export the table of element centroid coordinates from the thermal model
• Create a table of time average power dissipation using the steady state
AC results. Note that calculating the power dissipation as a nodally
averaged table only in lossy materials is the most accurate representa-
tion
• Process the element centroid table, evaluating the time average power
dissipation at each element centroid
• Import the heat source table into the thermal transient model and store
the model with the heat table.
(The initial temperature conditions can also be specified using tables.
The thermal transient analysis program looks for a table called TEMP.)
Exporting the Export the table of element centroid coordinates from the thermal model
position where
the heat loss is
required
Calculating the Create a table of time average power dissipation using the steady state AC
power loss results. Note that calculating the power dissipation as a nodally averaged
table only in lossy materials is the most accurate representation.
Evaluating the Process the element centroid table, evaluating the time average power dis-
power loss at the sipation at each element centroid
centroids
Select Process to read the Input table of points, calculate the heat sources and
output the new table of heat sources. Then select Cancel to exit.
Importing the Import the heat source table into the thermal transient model and store the
heat sources. model with the heat table. l
The initial temperature conditions can also be specified using tables. The
thermal transient analysis program looks for a table called TEMP, if this
exists it uses the values it contains to define the initial temperature, other-
wise the initial temperature is set to zero.
Select Create table to create the table, Close the confirmation dialog.
Finally Cancel to exit.
Zoom bound-
ing box
The value of time for this solution is shown in the right hand information
panel, it is also available as the system variable THTRTIME. The temper-
ature distribution in the plate can be displayed as a contour map.
Introduction
Space Charge models can include the effects of magnetostatic and electro-
static fields or electrostatic fields alone. This example illustrates electro-
static fields only with potentials assigned on the electrodes. Thus only air
regions are required (with an additional conductor region being added to
terminate the beam). The complete model is shown in Figure 21.
OPERA-2d Pre-processor
The change in potential type has invalidated the units of some quantities. A
message box details these quantities and states the units to which they have
been reset.
Set units Set length unit to millimetre, and leave the rest as default values.
Click ok.
Region 1
Draw the region A separate region is used near the emitter surface so that the mesh can be
well structured to fit the beam pattern leaving the emitter. This is very
important for the convergence of the solution and the accuracy of the
results obtained. The default material definition is accepted for region 1
since this is a vacuum. The region coordinates, subdivisions and curvature
are listed in the table below. XY coordinate input is suggested in this case.
Continue the XY input for the following coordinates that make up the
region in front of the emitter.
X Y
0 5
25 8.17542
25 3.17542
After the last point has been created, close the region with the close pol-
ygon button, then Cancel to close the dialog.
Modify side by Pick first side from (0,0) to (0,5), and set
picking Number of elements to 6.
Continue with the other sides (in a clockwise order) setting curvature and
subdivision as follows:
Side Curvature Subdivision
2 -1/100 20
3 0 6
4 1/100 20
Region 2
Region 2 is also a vacuum region. It is created in a similar manner, using
Draw -> Next Corner in Polygon -> Corner and Side Properties
with the following coordinates and edge subdivision (number of elements).
The first 3 points listed in the table below could also be selected using
mouse and the Pick Existing corner option. The faces are all
defined in an anti-clockwise sense (refer to Figure 21).
Number of elements
X Y Curvature
on side
0 5
25 8.17542 -1/100 20
25 3.17542 0 6
25 0 0 4
27 0 0 2
27 3.71397 0 6
33 10 0 8
33 25 0 8
43 25 0.178885 8
43 0 0 8
75 0 0 5
75 50 0 8
30 50 0 8
30 60 0.2 8
75 60 0 8
75 80 0 4
35 80 0 8
25 100 0 8
25 150 0 15
0 150 0 10
0 100 0 15
0 60 0 20
0 50 0 5
0 20 0 15
Close Polygon 0 20
Region 3
The final region is a conductor region (material 1). It is a feature of the anal-
ysis module that as a particle trajectory enters a material 1 region (conduc-
tor region), the trajectory terminates and this makes analysis more efficient.
The Material type is set prior to drawing the final region as follows:
Region prop- Set the option Polygon and set the Material label to 1.
erties
The region coordinates can then be entered with the following values, using
Draw -> Next Corner in Polygon -> Corner and Side Properties
as before.
X Y Curvature Subdivision
0 150
0 160 0 1
25 160 0 10
25 150 0 1
Close polygon 0 10
Refresh the
display
Regions created from the menus are polygons by default. Polygon regions
can have many sides and are appropriate in most cases. However four sided
polygon regions can be converted from polygon shape regions to H or Q
shape regions. The advantage of H or Q shape regions is that they have a
regular mesh.
Select first region and last region to be 1, and option Regular quad (H).
The same procedure was applied to region 3, although the mesh in this
region is unimportant as the beam will terminate as it enters any region that
has material label greater than 0.
Boundary Conditions
Once this has been completed, the Boundary Conditions (Potentials) of the
model are to be set.
Set Boundary Set the option Fixed potential, and set the
condition Potential value to 0.
The same is done to apply other potential values to other edges. Zoom out
to the previous size, and assign the following potentials (refer to Figure 23):
X Y Potential
55 49 50000
24 55 50000
55 61 50000
55 79 100000
30 90 100000
24 120 100000
After the program finishes meshing the model, the outline of the model is
drawn on the screen. A text box appears which displays the results of the
mesh generation. The user should scroll to the bottom of the text box and
Generate mesh
verify that there are no errors and/or warnings. Any errors or warnings will
need to be corrected before running the model.
Before running the analysis module, the user needs to provide the program
with information which describes the emitter. This is accomplished by
defining the emitter parameters in the pre-processor using the model geom-
etry. The definition must be saved with the same filename as the model, but
with the extension .emit. (in this case space.emit)
Space charge Highlight Emitter 1 (created by selecting the Create emitter button) and set
emitters Number of charges to −1, Max ray spacing to 2, Temperature to 1273, Work
function to 1.82, and Emission constant 350.
Select Set emitter parameters to save these settings.
Select Create
emitter but-
ton.
Green arrows will be displayed on the line segment showing the directions
of emission. In this case in the -Y direction. This is wrong for this case, so
must be reversed.
Select Trajec- Set values Maximum trajectory step length to 2, Absolute tolerance to
tory parame- 0.01, and Normal sampling distance to 0.1, then select Set Global param-
ters tab eters button.
A summary of the parameters set above for this emitter can be seen by
selecting List emitter data, and is as follows:
Close the information dialog by selecting Close, and then close the Emitter
dialog by selecting Cancel.
The accuracy of the solution is dependent on the mesh and the emitter
matching correctly. Calculation of the currents generated in each beam is
dependent upon the space charge calculated near the surface of the emitter.
Hence the SHAPE H region was used to generate a uniform mesh over this
area with the number of elements matching the number of beams being
emitted from the emitter surface.
Now that the space.op2 and the space.emit files exist, the model can be
solved with the Space Charge analysis module.
The user should enter the filename space when prompted and Accept the
defaults for the other prompts.
Post-Processing
Read the model into the post-processor. The file to be read in is space.sp.
In addition to the solution file, the user should note that the table RHO con-
taining the space charge distribution is automatically read in. (If the mesh
has not been saved, it will be necessary to remesh the model.).
The potential distribution can be seen by displaying contours with the com-
ponent set to POT.
Contour map Set Field component to POT, and select the Coloured zones option.
Set Number of contours to 100, then Draw contours.
There are several ways of displaying the tracks calculated by the space
charge analysis program. This includes a display of the tracks over the
geometry, while another is to display a three angle projection.
The charge distribution can be seen by plotting contours with the compo-
nent set to RHO. The results are shown in Figure 12.4.
Introduction
This chapter contains a set of Application Notes which are provided to help
users with specific types of applications of OPERA-2d. The examples
given do not contain complete instructions but should be used in conjunc-
tion with other sections of the User Guide and the OPERA-2d Reference
Manual to obtain details of the commands.
Cartesian Problems
In a model with XY symmetry, the cross section of a long device is ana-
lysed. There will be X and Y components of flux density and Z directed
currents.
Consider two points in the cross section. The flux per unit length linking
lines parallel to the Z direction, through the two points, is simply given by
the difference in the vector potential (Az) between the points.
For a line in the x-direction the flux linked in the x-z plane is given by:
z= 1 x = b
φ = ∫z = 0 ∫x = a B y d x d z (11.7)
Φ = ∫ nφ dx dy (11.9)
coil area
For a coil with uniform turns density, in terms of the vector potential Az,
the mean flux linkage is:
1 1
Φ = ------------------------
coil area a ∫ Az ( a ) dx dy – ------------------------
coil area b
- ∫ A z ( b ) dx dy (11.10)
coil area a coil area b
where conda and condb are the region numbers of the conductors to be
inserted as appropriate.
Axi-symmetric Problems
For an axisymmetric problem, solved in 2-dimensions, the flux linking a
loop radius a is:
θ = 2π r = a
φ = ∫θ = 0 ∫r = 0 Bz r dr dθ
(11.11)
a
= 2π ∫ B z r dr
0
a
Since curl(A)=B, and only Aθ is present (i.e. ∫0 B z r d r = rA θ ), then:
φ = 2πr A θ ( a ) (11.12)
Φ = ∫ nφ dr dz (11.13)
coil area
For a coil with uniform turns density, in terms of the vector potential Aθ,
the mean flux linkage is:
1
Φ = --------------------
coil area ∫ 2πrA θ dr dz (11.14)
coil area
Note that the solution type modified rA gives the best results for all types
of axisymmetric geometry.
1 I
Φ ( I ) = ---- ∫ L ( i ) di (11.15)
N i=0
(neglecting hysteresis effects), where Φ is the flux linking the coil, L is the
inductance, and N is the number of turns in the coil.
Using the flux, Φ, and the current in the coil, I, the self inductance can be
calculated as:
NΦ
L = -------- (11.17)
I
A similar equation can be used for the mutual inductance, using the flux
linking one coil due to the flux generated by the current in another. This is
calculated using:
Ni Φi
M ij = ------------ (11.18)
Ij
Using the solution to the model with one coil switched on will allow you to
find the self inductance of the coil, and the mutual inductance of other coils
with respect to that coil.
for a model with n-coils. Similar equations can be generated for Φ2 etc.
Note that all the inductances will be functions of the currents.
The model must then be altered so that the current in any one of the coils is
changed by a small amount (∆I), saving the model, and generating a new
solution file (using restart run to speed matters up).
The small change should not greatly affect the field, but a small change of
∆ΦI will occur for each coil. For example, for a change of ∆I1:
N 1 ( Φ 1 + ∆Φ 1 ) = L 1 ( I 1 + ∆I 1 ) + M 12 I 2 + M 13 I 3 + … + M 1n I n
N 2 ( Φ 2 + ∆Φ 2 ) = L 2 I 2 + M 21 ( I 1 + ∆I 1 ) + M 23 I 3 + … + M 2n I n (11.20)
…
Hence taking the difference between equation 11.20a and equation 11.19,
we can calculate L1:
N 1 ∆Φ 1
L 1 = ----------------- (11.21)
∆I 1
Equation 11.20b and equation 11.19 will give M21, and so forth. Modifying
I2 would allow similar calculations for L2 etc. Therefore, finding the self
and mutual inductance for all n coils requires n models to be solved.
This method can also be used if other energy sources are present, e.g. per-
manent magnets or external driving fields, as it works on the basis of a
change in field.
Power Calculations
If a current and voltage waveforms vary as:
i = I cos ( ω t )
(11.22)
v = V cos ( ω t – φ )
where
VI
A = ------ cos ( φ )
2
VI
B = – ------ cos ( φ ) (11.24)
2
VI
C = – ------ sin ( φ )
2
π
P ( 0 ) + P ---
2 VI
P tav = ------------------------------- = A = ------ cos ( φ ) (11.26)
2 2
Energy Calculations
Since nonlinear materials are not modelled, in a magnetic circuit, B and H
are sinusoidal. Replacing the i and v with B and H, the same arguments fol-
low, such that the time average energy density is given by:
π
E ( 0 ) + E ---
2 B⋅H
E tav = ------------------------------- = A = ------------- cos ( φ ) (11.27)
2 2
General
P is the system variable POWER, and E is the system variable ENERGY,
set after the INTA command, and ωt is set by the TIME parameter (in
degrees). Further notes are available under the relevant commands in the
Reference Manual.
9
10.0 UNITS
Length : cm
Y [cm] Flux density : T
9.0 Field strength : A m-1
Potential : Wb m-1
-1
Conductivity : S m
8.0 Source density: A m-2
Power :W
Force :N
Energy :J
7.0 Mass : kg
6.0
7
5.0
PROBLEM DATA
4.0 C_Magnet.st
Linear elements
XY symmetry
Vector potential
3.0 Magnetic fields
6 Static solution
Scale factor = 1.0
2.0 1585 elements
834 nodes
5 10 regions
2 1 3
1.0
4
0.0
0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 12.0
X [cm]
C_Magnet_pic1.ps
OPERA-2d
Pre and Post-Processor 8.830
Figure 27 Flux lines in a C-core magnet; only 1/2 of the magnet is modelled
Nodal Fields
By default the magnetic field can be evaluated only in meshed regions. If
the field is requested outside the mesh an error message Point outside
mesh is displayed. Figure 28 shows the meshed regions.
The field in the lower half of the magnet (y < 0) can be evaluated after ena-
bling the field symmetry options. Field symmetry is set using:
Options -> Field Options
For the example above, the dialog should be used to set the reflection of the
field in the X-axis such that only the normal component exists, as shown in
Figure 29.
20.0 UNITS
Length : cm
Y [cm] Flux density : T
18.0 Field strength : A m-1
Potential : Wb m-1
Conductivity : S m-1
-2
16.0 Source density: A m
Power :W
Force :N
Energy :J
14.0 Mass : kg
12.0
10.0 9
PROBLEM DATA
8.0 C_Magnet.st
Linear elements
XY symmetry
Vector potential
6.0 7 Magnetic fields
Static solution
Scale factor = 1.0
4.0 1585 elements
10 8 834 nodes
10 regions
6
2.0
2 1 3 5
4
0.0 -4.0 0.0 2.0 4.0 6.0 8.0 10.0 14.0 18.0
X [cm]
04/Mar/2003 13:44:54 Page 50
C_Magnet_pic2.ps
OPERA-2d
Pre and Post-Processor 8.830
For further information see the SET command in the OPERA-2d Refer-
ence Manual.
It should be noted that the lower half of the magnet is not displayed on the
screen after the field symmetry options have been enabled. A consequence
is that only the POINT command and all line commands (i.e. LINE, CIR-
CLE, INTLINE, INTCIRCLE, HARMONICS) can access the lower half
The TABLE command can also be used to recover values in the symmetric
images. However only system variables (e.g. POT, HX, EZ etc.) correctly
apply the sign changes implied by the field reflection (for example in the
C-core magnet the correct symmetry is Hx(y)=-Hx(-y)). User defined tables
evaluated from expressions involving system variables, or imported with
the solution, will not include the necessary sign changes. The magnetiza-
tion vector tables MAGBX and MAGBY (or MAGBY and MAGBZ) com-
puted by the DM analysis will not have the correct symmetry sign.
Therefore using reflections in the DM analysis should be avoided.
Integral Fields
The integral field option can be enabled using:
Options -> Field Options
and clicking on the Integral fields button, as shown in Figure 30.. Note
that the Total field option has also been selected meaning that the field
will be computed from the integration of the magnetization and coil cur-
rents.
In order to obtain the correct integral fields it is very important that the field
symmetry options have been enabled as discussed in the previous section.
Example
The following two figures show a line plot of the magnetic flux density in
the air gap of the C-core magnet. The line is at y < 0, shown in Figure 31.
Figure 32 then shows the magnetic flux density both with nodal fields
(default) and with integral fields. The variation of the fields is similar, but
there is an offset in the nodal values due to truncating the mesh at the outer
boundary. To reduce this difference, the outer boundary should be moved
further away.
6.5 UNITS
Length : cm
Y [cm] 6.0 Flux density : T
7 Field strength : A m-1
Potential : Wb m-1
5.5 Conductivity : S m
-1
4.0
3.5
3.0
PROBLEM DATA
C_Magnet.st
2.5 6 Linear elements
XY symmetry
Vector potential
2.0 Magnetic fields
Static solution
1.5 5 Scale factor = 1.0
2 1 3 1585 elements
834 nodes
1.0 10 regions
0.5 4
0.0
0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0
X [cm]
Component: BMOD 04/Mar/2003 18:34:17 Page 67
C_Magnet_pic3.ps
0.0399234 0.216808 0.393692
OPERA-2d
Pre and Post-Processor 8.830
0.09 UNITS
Length : cm
Flux density : T
0.08 Field strength : A m-1
Potential : Wb m-1
Conductivity : S m-1
0.07 Source density: A m
-2
Power :W
Force :N
0.06 Energy :J
Mass : kg
0.05
0.04
PROBLEM DATA
0.03 C_Magnet.st
Linear elements
XY symmetry
0.02 Vector potential
Magnetic fields
Static solution
0.01 Scale factor = 1.0
1585 elements
834 nodes
0.0 10 regions
X coord 6.5 6.5 6.5 6.5 6.5 6.5 6.5 6.5 6.5 6.5
Y coord -0.8 -0.6 -0.4 -0.2 1.1921E-08 0.3 0.4 0.5 0.6 0.7
OPERA-2d
Pre and Post-Processor 8.830
Discussion
Nodal Nodal interpolation is used to compute the field at any specific point of
interpolation interest by interpolation of the potential (and resulting fields) at the mesh
nodes surrounding the specific point. Therefore, the result obtained
depends not only on the overall quality of the solution, but also on the qual-
ity of the solution locally - i.e. in the area of interest. The quality of the
solution, in turn, depends on the mesh quality.
Integral fields Integral fields will evaluate the field at the specific point of interest by inte-
grating the magnetization in magnetic materials, the current densities in the
conductors and the eddy currents in conducting materials. This is a very
complicated task, so if the fields are required at a set of points (say fields
on a surface), this could be very time consuming.
This method can be more accurate as, although it depends on the overall
quality of the solution, it does not depend on the quality of the solution
locally - it is therefore more likely to obtain a smoother field distribution
from an integral field computation. Integral fields can be inaccurate in
cases when there are large cancellations between the source and magneti-
Integral fields are not only useful for computing fields inside the mesh in
regions where the mesh quality is poor, but can also be employed for the
computation of the fields at any point outside the finite element mesh.
Nodal vs. In some cases the two different methods (Nodal and Integral) may give dif-
Integral Fields ferent results, depending on the actual mesh. Nodal Fields can be used
effectively to compute fields in areas of interest provided these areas are
sufficiently well meshed (a natural requirement anyway). Integral fields are
very useful for computing fields at a distance where the mesh is often less
well defined.
Integral fields inside a mesh work fine in air and in source conductors,
however they give completely wrong answers in all magnetized materials
such as iron. Therefore integral fields inside a mesh should only be used if
the region of interest is at a sufficient distance from magnetized materials.
Introduction
A simple model of hysteresis can be represented by a complex material per-
meability. This note reviews the approximation and shows how the losses
can be calculated.
– iα
with µ̃ = µ ⋅ e . (11.30)
or
UNITS
B Length :m
Flux density : T
Field strength : A m-1
Potential : Wb m-1
8.0 Conductivity : S m-1
Source density : A m-2
Power :W
6.0 Force :N
Energy :J
4.0 Mass : kg
2.0
0.0
-0.1 -0.08 -0.06 -0.04 -0.02 0.0 0.02 0.04 0.06 0.08
Complex hysteresis loop H
OPERA-2d
Pre and Post-Processor 7.092
Hysteresis losses
The consequence of using a complex permeability as above is that there is
a phase shift between B and H.
H = H 0 ⋅ cos ( ωt + ϕ ) (11.33)
B = µH 0 ⋅ cos ( ωt + ϕ – α ) (11.34)
The hysteresis energy loss per cycle is given by the area of the ellipse:
W = π ⋅ sin α ⋅ ∫ µH 0 dΩ
2
(11.35)
The second step is to specify that complex material properties are used in
the AC analysis. Complex permeability can be enabled by setting the
appropriate option on the AC analysis options menu which is presented
when Steady state harmonic (AC) is chosen using:
MODEL ↓
Analysis Options → Steady State AC
In keyboard mode the same is achieved using the SOLVE command with
the sub-command: DATA +CMU
UNITS
B Length : cm
Flux density : gauss
Field strength : oersted
Potential : gauss-cm
20000.0 Conductivity : S cm-1
Source density : A cm-2
-1
Power : erg s
Force : dyne
Energy : erg
10000.0 Mass :g
0.0
PROBLEM DATA
hist_test4.ac
Linear elements
-10000.0 XY symmetry
Vector potential
Magnetic fields
Cmplx mu ac solution
Frequency = 50.0 Hz
-20000.0 832 elements
467 nodes
4 regions
-4000.0 -2000.0 0.0 1000.0 2000.0 3000.0 40
BH Curve H
Phase angle 20 degrees at 20 kGauss
Phase angle 20 degrees at 25 kGauss
19/Aug/1999 17:35:05 Page 17
OPERA-2d
Pre and Post-Processor 7.092
Figure 34 Using different phase angles for different parts of the BH curve
Figure 35 below illustrates the layout of the Medium Voltage Cable Termi-
nation. The following components can be identified:
• Aluminium conductor
• Insulation of the cable
• Semiconductor of the cable
• Zero volt conductor
• Stress control
• Insulation sleeve
• 2 Sheds
Sheds are insulating truncated cones placed around the termination prima-
rily to increase the length of the creep path for fields that track down the
side of the termination due to dirt, moisture etc.
Figure 36 below shows how the same device of Figure 35 appears in the
Pre and Post-Processor. Note that the cross-section of the sheds in the
radial-axial plane is actually very thin, such that they appear almost as a
line in Figure 36. This is, of course, made worse by the poor aspect ratio of
the picture.
All other materials are included in the model, and their permittivity and
conductivity is specified under the material (region) properties menu. Fig-
ure 38 shows the material properties assignment for the semiconductor of
the cable.
When solving for devices with conducting and dielectric properties, a wide
variation in conductivities in the materials may exist. For example, in the
present device, the semiconductor of the cable has a conductivity of 1 S/m
whereas the stress control has a conductivity of 1E-06 S/m. The conductiv-
ity of the aluminium is 2E7 S/m. OPERA-2d/LD can solve for a relatively
Choice of Solver
Having created the geometry in OPERA-2d pre-processor and assigned the
correct material properties and boundary conditions, the user can proceed
with the selection of solver. A static, steady-state a.c. or transient analysis
can be performed, and the appropriate solver is selected via the menu route:
Model -> Analysis Options -> Static
Steady State AC
Transient Time Varying
Within each of above analysis options the user must enable the Lossy Die-
lectric option, as shown in Figure 39.
Field Solutions
Figures 40 & 41 show representative field solutions, depicting the electric
potential in the cable termination, obtained using the OPERA-2d Electro-
statics solver (which solves only for dielectric, non-conducting materials)
and the OPERA-2d LD Electrostatics + Current Flow solver respectively.
in this material. However, this is only true along the length of the more con-
ducting stress control material. The 0kV conductor is sandwiched between
relatively high conductivity materials (the stress control and semiconductor
of the cable) across which little potential is dropped. The combination of
the above effects means that the conductivity of the stress control material
has a significant impact on the voltage stresses around that area. The cor-
rect choice of stress control conductivity leads to effective control of the
field distribution in the termination.
Reference
1. 'A transient formulation for the design of electric insulating/conducting
components: EU Project ADETEC', P. Alotto et al.
The results of this analysis will be the components of the Fourier series
described by
F ( x ) = a 0 + a 1 cos ( 2 π xt ) + a 2 cos ( 4 π xt ) + …
(11.37)
+ b 1 sin ( 2 π xt ) + b 2 sin ( 4 π xt ) + …
The.comi file above is a general purpose command file and must be modi-
fied for use in the correct context. If only part of the model has been created
within the software, the commands above must be changed to reflect the
partial nature of the results of the integral command, i.e.
F(x) is an Odd Function: An = 0
F(x) is an Even Function: Bn = 0
Only 1/4 period: Even harmonics e.g. A2, A4, B2, B4 are zero.
Introduction
Many electromagnetic devices are operated while connected to a voltage
driven electric circuit. For dc operation it is usually quite simple to deter-
mine the current that will be carried by the coils, as this will only depend
on the resistance of the winding and any external resistive circuit elements.
However, under time varying conditions, the inductance of the coil and
external circuit may also be significant and can be time dependent due to
magnetic saturation and eddy currents. Consequently, the current in the coil
will not be known. To allow problems of this nature to be solved OPERA-
2d supports external voltage driven circuits in the AC, TR, RM, LM and
DM solvers.
considered as being constructed from filamentary wires such that skin and
proximity effects in the turns are ignored. Windings, where these effects are
important, are discussed later in this document. Consequently, RFE has two
components – RDC, the DC resistance of the winding, and REC, the equiv-
alent resistance of the eddy current circuits in the problem referred to the
series circuit. The user specifies the values for V, REXT, LEXT and RDC,
while the program computes the values of LFE and REC.
10.0 UNITS
Length : mm
Y [mm] Flux density : T
9.0 Field strength : A m-1
Potential : Wb m-1
Conductivity : S m-1
Source density : A m-2
8.0 Power :W
Force :N
Energy :J
7.0 Mass : kg
6.0
5.0
PROBLEM DATA
circ_1.tr
4.0 Linear elements
XY symmetry
Vector potential
3.0 Magnetic fields
Transient solution
Time = 0.0001 s
2.0 1043 elements
555 nodes
4 regions
1.0
0.0
0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 12.0
X [mm]
18/Aug/1999 17:30:19 Page 3
OPERA-2d
Pre and Post-Processor 7.092
The model is created as for a normal current driven problem with two
exceptions: the current density in the coil is assigned to zero, and the coil
region is assigned a conductor number through the tool button.
For this model, the coil region is assigned conductor number 1. After mesh
generation, the user defines the external circuit through the
menu. In the first dialog, global parameters for the circuit are specified.
Element Value Explanation
Circuit type Filamentary Skin and proximity effects are not
included
Symmetry 2 Circuit currents return outside the
model
Length (XY only) 50 The coil is 50 mm long in the Z
direction
Peak voltage 1.5 Voltage applied to circuit, V
Phase angle 0 Used in AC solutions only
Resistance 0.02 Value of REXT in Ω
Inductance 0 Value of LEXT in Henrys
Capacitance 0 Value of CEXT in Farads (0 = no
capacitance)
Initial voltage 0 Voltage on capacitance at time = 0
sec. for TR/RM/LM/DM solutions
(in Volts)
After completion of the dialog and clicking on the Create circuit button,
a second dialog allows the user to specify the elements (conductor num-
bers) that build up the circuit. In this model, only one conductor is speci-
fied.
Item Value Explanation
Conductor number 1 The next series element to be
included in the circuit
Sense of conductor GO The current will flow into the plane
of the model
Number of turns 150 The number of series turns in this
conductor
Resistance/unit length 0.0012 Resistance of the wire used to con-
struct the coil (=1.2 mΩ/mm)
OPERA-2d has a number of facilities to enable the user to visualise the cir-
cuit and list its parameters. Functions are also available to make correc-
tions. In order to label the regions that belong to a circuit, the user can Set
view and select Circuit number under the Region labels tab.
When the screen is refreshed, only regions which form part of a circuit will
be labelled with their corresponding circuit number. To list the circuit, the
user can select the List external circuits tool button.
When the circuit has been entered correctly, the .op2 file for the transient
analysis is created specifying a step function drive for circuit 1.
Figure 45 shows the exponential rise of the current in the winding, dis-
played using the File -> Graph Data in Text File facility. Note that it
approaches its dc value, V / (REXT + RDC) = 83.2 mA.
UNITS
Length : mm
Flux density : T
Field strength : A m-1
Potential : Wb m-1
Conductivity : S m-1
Source density : A m-2
Power :W
Force :N
Energy :J
Mass : kg
PROBLEM DATA
circ_5.tr
Linear elements
XY symmetry
Vector potential
Magnetic fields
Transient solution
Time = 0.0001 s
2806 elements
1450 nodes
5 regions
OPERA-2d
Pre and Post-Processor 7.092
The “GO” part of the circuit is assigned conductor number 1 and the
“RETURN” conductor number 2. The external circuit is defined similarly
as for the symmetric case except that “Symmetry” is assigned to “1”, since
no currents return outside the model, and a second element must now be
added to the circuit.
Item Value Explanation
Conductor number 2 The next element to be included in the
circuit.
Sense of conductor RETURN The current will flow out of the plane
of the model
Number of turns 150 The number of series turns in this
conductor
Resistance/unit length 0.0012 Resistance of the wire used to con-
struct the coil (=1.2 mΩ/mm)
The circuit number to which each conductor belongs is shown in Figure 48.
Note that the +/- sign preceding the circuit number refers to a “GO” or
“RETURN” conductor respectively.
Figure 49 shows the asymmetric plate problem now driven from a four turn
coil with each turn having a 0.4 x 0.4 mm cross-section. In this example,
the voltage source is an AC supply at 10 kHz. As the skin depth of copper
at this frequency is about 0.7 mm, some redistribution of the current in the
coil is seen, shown by the coloured contours in the turns. Consequently,
REM≠RDC and can be evaluated from the power dissipation and current in
the conductor.
Introduction
Shared filamentary conductors can be defined in OPERA-2d, to allow for
modelling of interconnected and polyphase circuits. This note describes
how to set up such circuits and post-process the results to extract the
required parameters for each circuit.
Two voltage driven circuits were defined across the shared, lumped 2 Ohm
component. The first circuit is defined with a phase angle of 0, and a second
circuit with a phase angle 90 (by definition - see Figure 50). The circuits
were arranged such that the GO conductor in one loop also featured as a
GO conductor in the other. The dialog of Figure 52, which is accessed via:
In order to specify a shared circuit component, one of the two circuits needs
to be edited by selecting:
The Update button confirms the user entry. It is only necessary to define the
resistor in one loop, as the second will be updated automatically. A full
description of the resulting external circuits definition is listed in Figure 55,
together with the resulting current values, i1 and i2.
Kirchoff’s Voltage Law states that the algebraic sum of the voltage drops
in any closed path in a circuit is zero. Considering the paths taken by i1
and i2, and ignoring the minimal OPERA conductor resistance and
inductance, a system of 2 equations is formulated as follows:
100 = 3 2 i 1 (11.38)
– 100j 2 8 i2
Multiplying both sides by the inverse of the 2x2 matrix above results in the
following:
N.B. Applying Kirchoff’s 2nd Current Law to either node implies that the
current through the shared resistor is:
i PHA SE = i 1 + i 2 (11.40)
3-phase Systems
When setting up the model, attention must be given to the setup of GO and
RETURN conductors, in order to imply the correct circuit orientation
(local current directions) and thus drive the model correctly. Two current
loops are required (and hence two circuits), and both must be oriented such
that the current travels in the same direction across any conductor shared
by both circuits. Figure 56 shows the connections of a 3-phase winding
(with phases A, B and C) around the stator while Figure 57 shows the sim-
ple 3-phase generator geometry with the conductor numbers marked
around the stator.
The diagram above shows phase C to be common to both circuits. The ori-
entation of both circuits must be such that currents i1 and i2 both have
the same direction along Phase C. In order to satisfy this constraint conduc-
tors 3 and 4 are set to be GO in circuit 1, with 6 and 1 being RETURN con-
ductors. In circuit 2, conductors 2 and 3 need to be GO, with 5 and 6
specified as RETURN. Since Phase C is the common phase, values for
resistance, inductance and capacitance across conductors 3 and 6 should be
entered as shared components, with all other impedances defined in the
corresponding circuit in the usual way.
Using Kirchoff’s 2nd law, the currents in the 3 phases can be extracted from
the OPERA model using the relationships below:
iA = i1
iB = i2
iC = i1 + i2 (11.41)
The user may define a Delta connected load without using shared conduc-
tors. The three current loops shown in Figure 59 can be described by three
independent OPERA circuits.
Kirchoff’s 2nd voltage law states that the sum of all potential changes
equals zero:
i 1 1 + j 2π – ---
1
π
100
i 2 2 + j π – ---
2
100 ( cos ( – 120° ) + j sin ( – 120° ) ) = (11.42)
π
100 ( cos ( – 240° ) + j sin ( – 240° ) )
π 4
i 3 3 + j --- – ---
2 π
i1 2.734 – 16.307j
∴ i 2 = – 30.846 – 4.667j (11.43)
i3 – 13.669 – 30.223j
The result is in agreement with the OPERA values listed in Figure 61.
Example Problem
Figure 62 shows the field solution from an OPERA-2d/ST model of a typ-
ical loudspeaker. The model is axisymmetric and the permanent magnet
(the blue material) is consequently a ring. Ideally the permanent magnet is
magnetized perfectly in the axial (Z) direction. This application note shows
how the OPERA-2d/DM solver can be used to create the true magnetiza-
tion distribution.
Figure 63 shows the magnetization fixture that is used to magnetize the per-
manent magnet ring. Again this is an axisymmetric model. The solenoid
coil (in red) is connected to a 1 mF capacitor at 1.8 kV which discharges
through the coil. As current starts to flow in the coil magnetic flux is
induced in the magnetic circuit (in green) and through the permanent mag-
net. However, since both the magnetic circuit and the magnet are electri-
cally conducting the OPERA-2d/DM solver must solve the transient field
equations as eddy currents will try to prevent the flux being induced. Note
that the magnet is placed asymmetrically between the poles of the fixture.
Materials The BH characteristic for the magnetic circuit is a normal soft magnetic
material requiring no special treatment. The BH curve for the permanent
magnet material is the virgin curve for the material shown in Figure 64 and
a set of polynomial expression that specify the demagnetization character-
istics as shown in Figure 65. The polynomial expressions are created and
Circuit The solenoid coil has been labelled as conductor number 1. The capacitor
is connected to the coil by specifying a circuit in the Model menu. Clicking
on External circuits gives a dialog box to specify the external compo-
nents of the circuit as shown in Figure 67
After accepting the completed dialog the conductors making up the circuit
are selected, shown in Figure 68.
• the magnetization resulting from the flux density at the particular instant
in time during the demagnetization process may rotate with the direc-
tion of the flux density in the element (Rotate with B)
• the magnetization resulting from the field strength at the particular
instant in time during the demagnetization process may rotate with the
direction of the magnetic field strength in the element (Rotate with H)
In the Demagnetization analysis dialog box, open the Drive tab and set
the circuit 1 drive function to DC.
OPERA-2d/DM Analysis
The OPERA-2d/DM analysis proceeds similarly to an OPERA-2d/TR
analysis. Results are stored at each Output time specified by the user and
a .log file may monitor the current in the circuit or any other user defined
time dependent parameters. Figure 70 shows the value of the current in the
solenoid coil as the capacitor discharges.
Figures 71, 72 and 73 show the flux pattern at different times during the
magnetization process. The contour intervals are the same in all plots and
the effect of the eddy currents during the process can be seen.
If the user were to plot the values of MAGBR and MAGBZ later in the magneti-
zation process, the pattern would be very similar to Figure 75. The current
in coil reaches a peak close to 30 ms, and apart from some eddy current
effects, the elements in the magnet will not experience higher fields during
the magnetization process.
Creating a Table The table of element centroids in the loudspeaker model is generated after
of Element the model has been created and meshed but before it is solved. Under the
Centroids Tables menu, select the option Export a table. The dialog box shown in
Figure 76 allows the user to specify which regions the table of centroids are
to be evaluated and the name of the table file. Figures 62 and 63 show that
the magnet is in a different position axially in the two models. The dialog
box also allows the user to specify a translation and/or rotation that will
bring the centroid coordinates within the magnet in the OPERA-2d/DM
model.
Processing the The final case in the solution to the OPERA-2d/DM file is loaded and
Table Process a table is selected from the Tables menu. As shown in Figure
77 the user selects the name of the new table file and the additional tables
to be exported at the centroid coordinates (MAGBR and MAGBZ). Further trans-
formations can be made during the processing - for example, if the magnet
were moved to a new position in the fixture but the original element cen-
troid table file was still used.
Importing the Return to the unsolved loudspeaker model and from the Tables menu
Table select Import a table. When the first table file is imported either Over-
Write or Add can be chosen. However if several copies of the magnet
existed in the application model it is important to use Add to build up the
table of the magnetization in all the copies of the magnet during later
imports. Before running the analysis it is recommended that the user checks
that the tables have been successfully imported as shown in Figure 78
Note that the coarser discretisation of the magnet in the loudspeaker has
slightly reduced the range of plotted contours. The evaluation points from
the element centroids have not included values from the elements near the
top left and bottom right corners of the magnet where the highest and low-
est values were calculated by OPERA-2d/DM (see Figure 75).
The static OPERA-2d/ST analysis can now be run. The only limitation
compared to a normal ST solution is that mesh adaption is not allowed.
Figure 80 shows one such case, where the magnetization of 2 ring magnet
sections in a Permanent Magnet Direct Current (PMDC) motor was pre-
dicted using OPERA-2d/DM and the model was subsequently solved using
OPERA-2d/ST to model the field distribution, with the magnet rings now
surface-mounted on the PMDC stator back-iron.
The extended model starts as before with a virgin BH curve and sets of tem-
perature dependent demagnetization curves, as shown in figure 81. Addi-
tional data defining the recoil behaviour of the material is also now being
provided and the necessary field history of the material is stored so that de-
magnetization in the application device, e.g. the PMDC electrical machine,
can be modelled.
Implementation
In OPERA-2d the following steps need to be observed.
Demagnetization The user may define these as quadratic polynomials, as before (see Using
curves OPERA-2d/DM to Magnetize a Magnet [page 257]). However, quadratic
polynomials cannot accurately represent the demagnetization characteris-
tics of soft, easily demagnetized materials. OPERA-2d allows the user to
define any function to describe this characteristic, in the form of a series of
points. As an example, the user may need to define a variation equivalent to
α + H
B = α1 H 2 + α 2 H + α 3 + α 4Tan −1 5
α6 (11.45)
$function filename.table
which reads the text file containing the variation of flux density B with
magnetization, M, at a set of temperatures and remanent flux densities, Br.
The command automatically generates a functional variable, which is the
name of the final column (in this case M_F3BG).
Recoil BH curve A constant dB/dH needs to be defined (typical values range from 1.01 to
1.05)
Activating the The extended demagnetization model is activated by defining a user varia-
Model ble #RECOILMU equal to the incremental recoil permeability to be used in
materials that have been specified with demagnetization characteristics.
The original demagnetization model will be used if #RECOILMU is not
defined or set to zero. All OPERA-2d solvers activate the model if this var-
iable is stored in the data file.
armature currents set back to zero) and comparing these with results before
the onset of rotation (and with no armature excitation). Sections of the per-
manent magnet have been permanently demagnetized by approximately
5%, as shown in figure 85.
Figure 85 B-field inside magnet before and after rotation (no currents)
8500 0 -20 0
9000 0 -20 0
9500 0 -20 0
10000 0 -20 0
0 8000 20 2750
500 8000 20 3220
1000 8000 20 3650
1500 8000 20 3950
2000 8000 20 4100
2500 8000 20 4200
3000 8000 20 4225
3500 8000 20 4250
4000 8000 20 4250
4500 8000 20 4250
5000 8000 20 4250
5500 8000 20 4250
6000 8000 20 4250
6500 8000 20 4250
7000 8000 20 4250
7500 8000 20 4250
8000 8000 20 4250
8500 8000 20 4250
9000 8000 20 4250
9500 8000 20 4250
10000 8000 20 4250
0 6100 20 1900
500 6100 20 2370
1000 6100 20 2750
1500 6100 20 3000
2000 6100 20 3150
2500 6100 20 3250
3000 6100 20 3275
3500 6100 20 3300
4000 6100 20 3300
4500 6100 20 3300
5000 6100 20 3300
5500 6100 20 3300
6000 6100 20 3300
6500 6100 20 3300
7000 6100 20 3300
7500 6100 20 3300
8000 6100 20 3300
8500 6100 20 3300
9000 6100 20 3300
9500 6100 20 3300
1000 0 20 0
1500 0 20 0
2000 0 20 0
2500 0 20 0
3000 0 20 0
3500 0 20 0
4000 0 20 0
4500 0 20 0
5000 0 20 0
5500 0 20 0
6000 0 20 0
6500 0 20 0
7000 0 20 0
7500 0 20 0
8000 0 20 0
8500 0 20 0
9000 0 20 0
9500 0 20 0
10000 0 20 0
0 8000 60 2400
500 8000 60 2870
1000 8000 60 3300
1500 8000 60 3600
2000 8000 60 3750
2500 8000 60 3850
3000 8000 60 3875
3500 8000 60 3900
4000 8000 60 3900
4500 8000 60 3900
5000 8000 60 3900
5500 8000 60 3900
6000 8000 60 3900
6500 8000 60 3900
7000 8000 60 3900
7500 8000 60 3900
8000 8000 60 3900
8500 8000 60 3900
9000 8000 60 3900
9500 8000 60 3900
10000 8000 60 3900
0 6100 60 1550
500 6100 60 2020
1000 6100 60 2400
1500 6100 60 2650
2000 6100 60 2800
Command Script
Command file to automate multiple static solutions, to include demagneti-
zation effects in permanent magnets.
/+++ Open Stream to Write Results In +++/
$OPEN 1 Results.dat write
/+++ Open Stream containing rotor currents +++/
$OPEN 2 Currents_Armature.txt read
/+++ Set Constants +++/
$CONS #ANGLESHIFT 30
$CONS #STEP 3
/+++ Enter Loop to Rotate Rotor and Solve +++/
$DO #R 0 #ANGLESHIFT #STEP
$CONS #T (#ANGLESHIFT/#STEP)*(#R/#ANGLESHIFT)
/+++ Read the file in and advance the rotor angle+++/
$STRING sT %REAL(#t)
$IF #R EQ 0
READ PMDCmotor_ini.OP2
$ELSE
READ RotorAT%int(#T-1).st
MODI REG1=ROTOR XCEN=0 YCEN=0 ANGLE=ANGLE+#STEP
$END IF
$CONST #T &sT&
RECO XMIN=-20 20 -20 20
/+++ Read Generate mesh and set solution type+++/
SET ELEM=QUAD
MESH
RECO LABE=YES MESH=NO CLIN=YES BACK=YES
/+++ Check for existence of tables +++/
$EXIST 'All_tables.table'
$IF fileexists eq 1
$OS del 'All_tables.table'
$END IF
/+++ save magnetizing field tables +++/
EXTRA
EXPORT FILEOUT='All_tables.table' REG1=1 REG2=* SOL3=DEMAGBX
SOL4=DEMAGBY,
SOL1=MAGBX SOL2=MAGBY NODAL=NO TRANSX=0 TRANSY=0 ROTATION=0,
NODEDUMP=NO NELD=NO
QUIT
/+++ Assign Currents (which deletes Tables)+++/
$READ 2 #INDEX #SC0 #SC1 #SC2 #SC3 #SC4 #SC5 #SC6 #SC7 #SC8 #SC9 #SC10
#SC11
$DO #J 0 11
MODIFY REG1=SLOT%int(#J) DENS=((#SC%int(#J))/(21.24*2))*1E5
ERASE=NO
$END DO
/+++ Now re-load Tables +++/
EXTRA
IMPORT FILEIN='All_tables.table' OVERWRITE=YES
QUIT
/+++ Write File and Solve +++
$STRING t1 'RotorAt%INT(#T)'
$STRING t2 '.st'
WRIT FILE='&t1&' SOL1=MAGBX SOL2=MAGBY SOL3=DEMAGBX SOL4=DEMAGBY
$OS operaanl ST '&t1&' FORE
$END DO
$CLOSE 1
$CLOSE 2
Introduction
This application note shows how to use the Linear Motion (LM) solver to
model a moving coil motor from a loudspeaker driver. This application
note will not cover the details of setting up the particular model but is
intended to show the user how to set up this type of Linear Motion problem
so that they can achieve good accurate results efficiently.
Loudspeaker Driver
The example being modelled is a simple ring magnet and iron yoke system
with the coil sitting in the pole gap.
The yoke is mild steel and the magnet is a NdFeB ring with a vertical mag-
netization. The coil has a defined current density of 10A/m^2.
Electromagnetic Model
As the driver motor is rotationally symmetric about its axis this model can
be defined as a 2D Axi-symmetric model. The model is built up using a
number of regions to define the yoke, the magnet, the coil and the surround-
ing air.
Once the basic items have been created, as in a static or transient model, the
additional items required to convert this model to a Linear Motion model
are very simple. These are the creation of the group of moving objects (the
coil) and the group where the moving objects are allowed to move (the air
in the pole gap), also called the Fluid Group.
Fluid Group One or more AIR regions are defined in the pole gap, extending above the
yoke and down into the yoke gap, which will allow the coil freedom to
move up and down. These are then formed into a GROUP using the
Region Groups tool button.
The Fluid Group can contain as many regions as required to define the area
where the Moving Group is allowed to move. It should be noted though that
as the Fluid Group is re-meshed at every time step if this area is too large
then this will slow down the analysis.
Moving Group The moving group in this model consists of the coil and a group of air
regions around it to improve accuracy, as is discussed below. Again these
are formed into a named GROUP using the Region Groups tool but-
ton.
As the Moving group and Fluid group are allowed to overlap in this special
situation it can become difficult to see and select the various parts of the
model. It is advisable to define the regions that will be used in the Moving
group after having defined the regions of the Fluid group so that they are
always drawn last and so are always visible.
Figure 88 Region groups for the moving and fluid groups of the model
Once the Fluid and Moving groups have been defined they are assigned
within the model using the LM gap region dialog from the
Model -> Air Gaps -> Linear Machine Air Gap
menu route, as shown in Figure 89, and clicking on the Moving mesh but-
ton.
At each time step a line integral around the outside of the Moving Group is
taken to compute the Maxwell Stress (the force in the z direction in this
example). For models like this with a coil or for models with a moving iron
mass (Solenoid Actuators) the accuracy of the line integral is greatly
increased by moving the line away from the coil or iron.
Where a coil is the moving part the very rapid field changes around the
ends of the coil will be difficult to model, unless a very large number of ele-
ments are used, so by moving the line integral into a region of lower field
gradient a more accurate answer is obtained.
So in this model four additional AIR regions have been defined around the
coil and these are included in the Moving Group. As these four AIR regions
and the Coil have all been defined as quadrilaterals they can be converted
to type H regions (regularly meshed, as can be seen in the figure below) for
a further improvement in the accuracy of the computed field.
Finally as the moving parts will actually move within the model space dur-
ing the simulation, it is important to ensure that the mesh is consistent
throughout the range of motion. This can be checked by changing the Local
Coordinate System for the moving group and re-meshing to view the mesh
as the moving part is shifted over the range of movement.
Logging It is not practical to store very many full field solutions at specific time
points as the solution file becomes very large. However it is desirable to be
able to monitor the motion and certain key components linked to the
motion at many more time points during the simulation. Logging is used to
allow both the progress of the analysis to be monitored as the solution is
being run and also so that the motion can be analysed separately from the
full field solutions stored at the output time points.
In this example the output time point (TTIME) is stored as the first column
of data. This is followed by the system variables z-force (lmzforce), z-shift
(lmzshift), z-speed (lmzspeed) and the user defined constant z-acceleration
(#accelz), as shown in Figure 93. If the coil had been driven by a circuit
then the drive voltage (v1) and the circuit current (i1) could also have been
included.
Coupling The mechanical coupling of the spring constants, masses and damping is
provided with a Command Input file (*.comi). In the example below these
characteristics have been derived from the Thiele-Small parameters for this
driver plus a set of end stops to limit the motion to within the defined fluid
region.
/ Linear Motion Speaker example
/ Driver parameters which describe the motion
/ All in SI units
/ Total moving mass (Kg)
$constant #Mms 0.050
/ Driver suspension stiffness (m/N)
$constant #Cms 0.0002
/ Losses due to the suspension (N.s/m)
$constant #Rms 2.0
/ Define End stops
$constant #lmminzshift -0.0048
$constant #lmmaxzshift 0.0048
/ Motion from F=ma or a=F/m
$constant #accelz (lmzforce-lmzshift*(1/#Cms)-lmzspeed*#Rms)/#Mms
These end stops are useful for ensuring that the moving parts do not leave
the defined region of motion (which would then halt the analysis).
The Command Input file is created using the OPERA-2d/PP command file
editor or other text editor and then assigned while setting up the analysis
using the Mechanical coupling option under the Command file tab.
Post-processing
As well as the usual range of post-processing facilities available once the
model has been solved such as plotting the fields and flux lines (as shown
below). It is possible to GRAPH the log file results.
The GRAPH command is available from the menus under File ->
Graph Data in Text File and allows any file containing columns of text
to be loaded and then to plot any component (column) against any other
component (column).
Introduction
OPERA-2d and, in particular, the AC and RM solvers have been success-
fully used for the modelling of asynchronous (induction) motors. The AC
solver is capable of accurately modelling the currents induced in the cage
rotor, resulting from the rotating 3-phase magnetic field generated in the
stator. The induction motor performance can be characterised at a range of
rotor speeds and/or stator winding excitation levels. Typical motor per-
formance parameters such as the variation of torque with slip frequency,
induced rotor currents and associated power losses are readily computed in
OPERA-2d AC. In addition, the Rotating Machines (RM) Solver, a tran-
sient eddy current solver, can be used to model the rotational effects, hence
predicting transient effects at start-up and torque ripple (this is discussed
further in Dynamic Modelling of Induction Motors [page 300]).
In the AC code, the use of the slip frequency requires further adjustments
to be made to any external circuits which may be attached to the motor. The
use of the slip frequency to drive the model results in values for variables
such as the back-emf which are reduced by the ratio of slip frequency
(fslip) to synchronous frequency (fsync). This can be remedied by scaling
the length of the circuit (in the External Circuits section). However, the
resistance per unit length must then also be adjusted, such that the true cir-
cuit resistance value is maintained. End region inductance figures will also
need to be adjusted for solutions at different speeds. The following is an
example of such a calculation for a 4-pole, 60 Hz induction motor:
Motor Data
Synchronous speed = 1800 rpm (equivalent to fsync of 60 Hz)
Rotor speed = 1740 rpm
Slip speed = 60 rpm
Slip frequency, fslip = 2 Hz
Data for a solution at starting (fslip=fsync)
R/mm = 1.4815 E-05 Ohms/mm (true value)
Machine length = 95 mm
End winding length = 270 mm
End winding resistance (lumped value) = 0.004 Ohms
Hence, Resistance / turn = (1.4815 E-05 * 2 * 95) + 0.004 = 0.0068 Ohms
Adjusted data for a solution at 1740 rpm
Adjusted machine length = 95 * fsync / fslip = 2850 mm
R/mm = 1.4815 E-05 * fslip / fsync = 0.04938 E-05 Ohms/mm
Hence, Resistance / turn = 0.0068 Ohms
Note: Lumped resistor values need no adjustment.
The air-gap of the machine should have at least 3 layers of elements and
should preferably be constructed from H or Q type regions. Figure 98
‘homes in’ on the gap region of a 6-pole, deep slot fractional slot winding
induction motor. The aluminium rotor bars are also finely discretised, in
order to correctly capture the induced eddy currents.
The current density in the aluminium slots can also be plotted, and the cur-
rent flowing down each slot can be evaluated by performing a surface inte-
gral of the current density over the selected aluminium regions.
The forces acting on the rotor, as well as torque, can be computed accu-
rately with special built-in functions such as the Maxwell Stress Integral.
Field Integrals Around a Circular Arc computes the X-Y force
components acting on the surface of the rotor, as well as the resulting
torque around the specified pivot point. Completing the dialog as shown in
Figure 99 gives the following output below:
per mm
Integral of x-component of Maxwell stress = -6.691E-04 N
Integral of y-component of Maxwell stress = 4.53265E-04 N
Integral of torque around (0.0 ,0.0 ) = 202.9261552 Nmm
Integral of POT
The line integral describes a circle around the rotor, half-way through the
air-gap between the stator and the rotor. The resulting torque figure of 203
Nm/m, which for the present machine of 95 mm translates to 19.285 Nm.
This figure is very close to the machine rated torque of 20.88 Nm. The dis-
crepancy is attributed to end winding inductance effects which cannot be
modelled completely with two-dimensional software.
Note that here, the component setting to J**2/SIGMA was left from the
previous contour plot. The program will evaluate the time average value of
J for each element (= 0) before evaluating the expression. The true time
average integral of J**2/SIGMA is one of the pre-set integrations carried
out by the program each time INTA is executed. The software returns a
value of 1.22667 W/mm which, for the present machine length of 95 mm,
translates to 116 W.
The length of the time step which the software uses between each solutions
is crucial to good convergence and computation times. In general, it is dif-
ficult for the user to assess the correct time step for any model. It is there-
fore considered good practice to choose the ‘adaptive time step’ option,
especially in the early stages of simulation. The software continuously
reports the length of the time steps it is using as it progresses through the
solution, and these can be recorded by the user.
As the software converges on the use of a (relatively) fixed time step, the
user can stop and restart (or run further simulations) using the time step pre-
viously observed. It is important to stop the simulation soon after an ‘output
time’, so that no computing time is wasted.
This Application Note introduces the user to the Rotating Machines Pro-
gram (OPERA-2d/RM). OPERA-2d RM is a Transient Eddy Current
Solver, extended to include the effects of rigid body (rotating) motion. The
solver also provides for the use of external circuits and coupling to mechan-
ical equations.
The faces of the regions that define the inner radius of the stator should be
circular, as should those of the outer radius of the rotor. Better results will
be obtained from the RM program by associating a layer of air elements in
the air gap with the rotor and stator regions. The Gap Region then fits
between these two layers and ensures that at least 3 elements are used radi-
ally across the air gap. This is important for the accurate calculation of
torque when coupling to motion.
The figures below show the air gap of a 4-pole induction machine, showing
the separate rotor and stator regions with additional air layers and then the
inclusion of the gap region which was specified at a radius of 0.0727 m.
∂A
∇ × --- ∇ × A – H c = J s – σ -------
1
(11.46)
µ ∂t
in which the current density has been split into the prescribed sources, J s
∂A
and the induced currents, σ ------- . In two dimensions, only the z components
∂t
of A and J s exist. Equation (11.46) can be simplified to:
1 ∂A z
– ∇ ⋅ --- ∇A z – ( ∇ × H c ) z = J s – σ -------- (11.47)
µ ∂t
External circuits
Transient solutions can be excited either by current sources or by voltage
sources connected via external circuits to the model. The voltage driven
option, employed in the modelling of this induction machine, allows a set
of coils to be defined as a circuit which is connected to an external voltage
source in series with an external resistance, capacitance and inductance. A
coil is the set of regions that have the same conductor number N.
The voltage specified should be the phase voltage amplitude. The external
resistance and inductance entries could be used to account for the end
winding resistance and inductance, otherwise not modelled in a two-
dimensional finite element program. The following was entered for each
phase of the induction machine:
Element Value Explanation
Circuit type Filamentary Skin and proximity effects are not
included
Symmetry 2 Circuit currents return outside the
model
Length (XY only) 0.15 The coil is 0.15 m long in the Z direc-
tion
Peak voltage 460 Voltage applied to circuit, V
Phase angle 0 Used in AC solutions only
Resistance 0.5 Value of REXT in Ω
Inductance 1.4 E-03 Value of LEXT in Henrys
Capacitance 0 Value of CEXT in Farads (0 = no
capacitance)
Initial voltage 0 Voltage on capacitance at time = 0
sec. for TR/RM solutions (in Volts)
OPERA facilitates the display of conductor regions labelled with the circuit
number which they belong to, as shown in figure 105. A positive number
denotes a GO conductor whereas a negative number denotes a RETURN
conductor.
Rotor speed
The simplest assumption in OPERA-2d/RM is that the rotor speed is con-
stant during the duration of the transient analysis. The user specifies the
rotation speed in RPM. Alternatively, the user may specify the speed of the
rotor as a function of time during the duration of the transient analysis.
where T is the torque, J is the polar moment of inertia and θ the angle. The
torque is the sum of the electromagnetic torque, computed by the program,
and the friction, load and linearly speed-dependent torques, defined by the
user. This (more advanced) option was selected for the RM analysis of this
example induction machine. A command file (*.comi) was prepared, which
included the following:
• Definition of the voltage sources driving the external circuits defined. A
functional 3-phase sinusoidal excitation was defined (#PHASE1,
#PHASE2 and #PHASE3), with a gradual increase of the voltage from
0 to full volts during the first 2 ac cycles.
• Characterisation of the load, increasing as the function of the square of
the speed, from 0 to a maximum of 7000 W.
The command file made use of the following system variables (a full list of
the system variables available may be found under the Rotating Machines
solver description in the OPERA 2d User Guide):
For example, in this 4-pole Induction machine, the source voltage gradu-
ally increased in amplitude during the first two cycles, before assuming
the normal sinusoidal variation with time. The .comi file contained the fol-
lowing instructions:
$CONS #FS 50
$CONS #AMPL 1
$CONS #timepoint 0.04
$IF TTIME LE 0.04
$CONS #PHASE1(TTIME*25)*#AMPL*SIN(2*PI*#FS*TTIME)
$ELSE
$CONS #PHASE1 #AMPL*SIN(2*PI*#FS*TTIME)
$END IF
The resulting waveform is shown in figure 106. Note that the time at which
the defined driving function is discontinuous was specified as a constant
with the name #timepoint. This, although not mandatory, helps the
time-stepping process, ensuring that enough sampling points are inserted
around this time. In order to define the variation of the load characteristic
with speed, the following instructions were required:
$CONS #MAXP 7000
$CONS #SPDPWR #MAXP/157/157
$IF RMSPEED GT 0.0
$IF RMSPEED LT 157.0
$CONS #LOADTORQUE #SPDPWR*RMSPEED
$ELSE
$CONS #LOADTORQUE #MAXP/RMSPEED
$END IF
$END IF
Output log file The transient analysis programs store their results at a set of times defined
by the user. However, it is possible to monitor the progress of the solution
by logging the output from all time steps to a file in the format required by
the GRAPH command. A log file of the Induction motor contained the
time (TTIME), rotation angle (RMANGLE) and speed (RMSPEED), the
torque (RMTORQUE) as well as the currents in the external circuits (I1,
I2 and I3) so that these may be graphed as the solution progresses or during
post-processing (the full list can be found under the SOLVE command in
the 2d Reference Manual).
Post-processing
Figure 107 shows a flux density distribution in the Induction Motor, with
the equipotential lines (which can be thought of as ‘flux lines’) superim-
posed. The plot was taken during the accelerating phase, and therefore
some flux penetration into the rotor body is evident.
Introduction
This note serves to introduce the user to more complex, so called “coupled
problems”, where results from an electromagnetic analysis are coupled to
the stress analysis module.
The model chosen in this instance is a switched reluctance motor. The dou-
bly salient, deeply saturated structure makes electromagnetic finite element
analysis a necessary tool for evaluation of the machine's performance.
Stress analysis performed on switched reluctance machines is important, as
the pulsed nature of torque production induces vibration (and hence acous-
tic noise).
This example illustrates the use of the Pre and Post-Processor with the stat-
ics and stress analysis solvers. The method of transferring electromagnetic
analysis results to the stress solver, as applied to an electric motor, is given.
Initially the electromagnetic problem is solved. The results are read into the
Pre and Post-Processor and examined before being transferred with addi-
tional mechanical information to a data file for input into the stress analysis
solver. The results of this analysis are then examined by further post-
processing.
Examining the The average torque produced by the motor, assuming a constant excitation
solution current (this can only be true at very low speeds) can be computed simply
by evaluating the change in co-energy at the 0o (un-aligned) and 90o
(aligned) rotor positions. In order to obtain the co-energy of the motor at
each position, the area integral command (INTAREA) is used, having set
the component to HDB. The software computes the H.B/2 and H.dB inte-
grals from which the co-energy value can be extracted. The average torque
produced by the motor was estimated to be 0.0525 Nm/mm stack length.
Figure 111 illustrates the field distribution at the 45o position. This is a high
torque producing position, as the rotor teeth tend to be pulled into align-
ment with the excited stator teeth. The instantaneous torque at this position
is expected to be much higher than the average torque computed earlier.
Instantaneous torque can be readily evaluated be performing an arc integral
(INTCIRCLE) in the middle of the air gap, a command which returns the
X and Y directed Maxwell stress force and the resulting torque about a
specified pivot point. The torque at this position was computed to be 0.139
Nm/mm.
The 45o (overlapping) and 90o (aligned) positions will be examined further
with the stress analysis solver.
A sequence of two dialog boxes was completed. Figure 112 shows the first
which specifies the material number. Clicking on the Edit button produces
a second dialog, shown in Figure 113, specifying the physical properties for
the material. For each new material, the material type, in this case Iso-
tropic, was selected first.
Entering the The definition of appropriate boundary conditions really depends on how
mechanical the motor is fixed to its frame, and therefore many valid combinations of
boundary boundary conditions can exist.
conditions
In this model, the stator was mechanically fixed at 4 points around the outer
edge, behind the centre of each pole. The centre of the shaft was also con-
strained in both X and Y. The condition was applied by selecting
Modify -> Stress and Thermal Boundary Conditions
completing the dialog as shown in Figure 114, selecting Pick node and
clicking on the 5 points described above followed by Quit.
Alternative The geometry of the machine can sometimes be such that the constraint
boundary cannot be expressed simply with movements in X and/or Y, as it could in
conditions (for the present model. In such cases a skew condition could be applied. A skew
reference) condition with a specified angle can be interpreted as a constraint on spec-
ified points lying on this angle, allowing these points to move radially but
not azimuthally. The skew constraint is also an alternative way to constrain
points lying on the X or Y axis. For instance, a Skew defined with Angle =
0 has the same effect as a point that is Constrained in Y (=0) at any point
on the X-axis.
Creating tables The stress analysis module requires data from the electromagnetic solution,
for mechanical relating the loading of the problem to the finite element mesh, to be avail-
analysis able as tables. The integrals of the Maxwell Stress on each element in the
X and Y direction (XLOAD and YLOAD) were created very simply by
Tables -> Create Standard Tables -> X and Y Loads
Having all data in place, the .op2 file must be written to include the geom-
etry, mechanical materials and boundary conditions as well as the Tables of
the Maxwell Stress Integrals. This was done by selecting
Model -> Analysis Options
selecting the Stress Analysis (SA) solver and opting for Stress. The
analysis was, consequently, then made using Plane Stress. This assumes
that the lamination is a thin sheet which is not prevented from contracting
in the normal - Z - direction. Plane Strain would assume an infinite thick-
ness, where each plane parallel to the paper would be constrained to remain
plane. Essentially, neither of these two cases exist in isolation, so the cal-
culation could be repeated with the alternative choice, and the ‘worst case
scenario’ used for design calculations.
When saving the .op2 file , it is necessary to select the XLOAD and YLOAD
tables to include them in the .op2 file.
Displaying the The excited stator teeth act as magnet shoes, pulling the rotor into a posi-
results as a tion of minimum reluctance. As the rotor moves towards alignment, forces
deformed mesh of attraction are acting on the excited stator and rotor teeth. In the aligned
position, the two bodies would still tend to close the gap that separates
them, in an attempt to minimise the reluctance of the magnetic circuit. The
rotor is therefore in tension while the stator is in compression.
The deformation due to loading may be viewed using the deformed mesh
feature. To do this, the regions which need to be examined must be selected,
and the displacement vectors scaled. The following sequence shows the
options selected for the stator deformation plot shown in Figure 115:
• Contour Map
• Set No contours option
• Under the Regions tab, select Include all materials, and set Mate-
rials to include to 3.
• Under the Other Options tab
A similar plot can be obtained for the rotor (refer to Figure 116 below).
A contour plot of the values of displacement can also be plotted. Figure 117
illustrates the values of DISPY in the aligned position, confirming the
trends described earlier: DISPY is negative in the stator (indicating a neg-
ative Y - downward - movement) and positive in the rotor.
45.0 UNITS
Y [mm] Length : mm
Flux density : T
Field strength : A m-1
40.0 Potential : Wb m-1
Conductivity : S m-1
Source density : A mm-2
Power :W
35.0 Force :N
Energy :J
Mass : kg
30.0
25.0
10 PROBLEM DATA
9 SAangl45_new.sa
20.0 8 Quadratic elements
XY symmetry
7 Vector potential
Magnetic fields
6 Static solution
15.0 5 Scale factor = 1.0
4 10560 elements
21217 nodes
132 regions
10.0
3
5
10
-20.0 -10.0 0.0 5.0 10.0 15.0 20.0 25.0 30.0
9 X [mm]
Component: DISPY 24/Aug/2000 16:18:52 Page 30
Minimum: -8.0E-11, 22Maximum: 1.0E-10, Interval: 2.0E-11 8
1
OPERA-2d
7 Pre and Post-Processor 7.507
Figure 117 Contours of stator and rotor displacement in the aligned position
Conclusion This Application Note has shown how OPERA can be used not only for the
full electromagnetic analysis of a switched reluctance motor, but through
close coupling of the ST and SA solver, also assessing the stiffness of the
motor structure and characterising its deformation behaviour under load.
Introduction
Many electromagnetic devices introduce significant internal forces in rigid
magnetic or conducting bodies. To determine the effects of these forces
OPERA-2d includes SA, the Stress Analysis Program, which solves for
static stresses and displacements in plane stress, plane strain and axisym-
metry. Forces may be determined from the electromagnetic analysis, but
the module may also be used with user-defined loads. The purpose of this
note is to give guidance on how to obtain the best results with the SA mod-
ule.
Like all Finite Element programs, the stress analysis solver is sensitive to
the element aspect ratio. The aspect ratio is defined as the ratio of the larg-
est to the smallest side of the triangular element. Elements of high aspect
ratios must be avoided. The aspect ratio should ideally be 1, although
aspect ratios up to 10 have been proven to perform well. The problems
caused by higher aspect ratios relate to the elements difficulty to represent
local changes in stress and strain. Ill-conditioned matrices may also result,
which lead to a completely incorrect set of displacements. With quadratic
elements, oscillatory behaviour in the field representation results.
Boundary conditions
External constraints such as zero displacement in X or Y (R or Z in axisym-
metry) or both can be applied to nodes or faces of a region, hence facilitat-
ing the modelling of common stress analysis problems such as a simply
supported beam or a cantilever. In addition, different types of loading can
be assigned to a beam. Point loads are represented as a load applied to a sin-
gle node, while distributed loads are applied by selecting a face of a region.
Theory The simply supported beam is 1m long, 0.1m in thickness and of unit depth.
One end of the beam is fixed with a PIN-type connection and the other end
is fixed with a ROLLER-type connection. A uniform load W/ unit length of
5 N/m is applied along the length of the beam.
where:
where:
d is the depth of the beam (m)
h is the thickness of the beam (m)
Model The model has one bottom end node constrained with zero movement in
both X and Y to represent the PIN type connection, while the other bottom
end node is constrained only in Y, hence representing the ROLLER type
connection. A uniform loading was achieved by applying a load of
(0,-5) N/m to the top face of the beam, noting that the negative sign signi-
fies a downward Y-directed force. A discretisation of 100×10 (length x
thickness) subdivisions was chosen, resulting in a fine and regular mesh of
unity element aspect ratio, as shown in Figure 118.
Figure 118 OPERA-2d model of a simply supported beam showing the regular
mesh and boundary conditions applied
Discussion on It is advisable to always construct regular meshes with aspect ratios ranging
Aspect Ratio from 1 to 10, so as to avoid inaccuracies in the solution. The combination
assignment of high aspect ratio geometry and elements, together with the use of quad-
ratic element representation can lead to oscillatory behaviour in field rep-
resentation results. One such example constructed was a simply supported
beam 10 m in length and 0.1 m in thickness, with a discretisation of 10x25,
resulting in a fine mesh along the beam thickness but a crude mesh along
its length.
Theory The cantilever is 1m long, 0.1m in thickness and of unit depth. A point
force of 5N/m is applied at the centre of the cantilever, i.e. 0.5m away from
the clamped end.
Model The equivalent model is set up in plane symmetry (plane stress) and has one
Preparation side clamped i.e. one face of the beam is constrained with zero movement
in both X and Y. A regular mesh was constructed with 50×25 subdivisions
and a resulting element aspect ratio of 5, as shown in Figure 121. A load of
(0,–5) N/m was applied at the top centre node of the cantilever model. Lin-
ear elements were selected for the analysis.
Theory The thin disk has a diameter of 22m and a thickness of 0.2m. Note that the
model aspect ratio is high, and therefore care must be taken in defining the
mesh. The disk is clamped around the whole of its periphery. A distributed
load of 2 N/m2 is applied to the disk.
The maximum deflection occurs at the centre of the disk and can be
expressed as
4
pa
w max = ---------- (11.54)
64D
where:
a is the radius of the disk (in m)
where:
h is the thickness of the plate
v is the poisson’s ratio.
The maximum stresses occur at r=a and z=±h/2. They are expressed as
2
3 pa –2
σr = ± --- --------- = 4537Nm (11.56)
max 4 h2
Model The model is set up in axisymmetry and has its whole side at R=11 m con-
preparation strained in both R and Z (this implies an R,Z constraint all around the
peripheral surface of the disk). The model must also be constrained so that
the centre axis of the disc (at R=0) does not move along the R-direction,
hence creating a ‘virtual hole’ in the model. The mesh chosen is of 55×10
subdivisions, yielding an aspect ratio of 10. The axisymmetric model set up
is shown in Figure 123.
and selecting the tracks file through the dialog shown in Figure 124.
and then complete the dialog shown in Figure 125 to define this intersec-
tion line by its end coordinates. In this case, a graph of current density
crossing the intersection line as a function of X has been selected.
Figure 125 Dialog for specifying intersection line and displaying results
Figure 126 shows the simple structure. The emitter is in the lower part of
the model, and the electrons are travelling upwards towards the anode. The
trajectories are calculated only in free space. As soon as a particle hits any
material, it will stop. In most cases the anode is being used to stop the beam.
We assume that the plasma mostly exists below Z=0 and that the emission
surface will be somewhere in the cone-shaped opening in the electrode
above the plasma (0 < Z < 2), see Figure 127. Note that the shape code of
region 1 is set to Q, which will give a regular mesh.
The emitter is specified by giving a curve (in this case, a straight line) that
will be anchored at its first point but is free to move until a self-consistent
solution is obtained. By self-consistent, what we actually mean is that the
normal electric field to the emitter surface remains constant along the sur-
face. Figure 128 shows the contours of EMOD and the trajectories, where
it is seen that although the surface has a constant normal electric field, it is
not at a constant value of EMOD. Figure 129 shows contours of EZ and the
trajectories.
Radial The particles are initially starting from a straight line, and all particles ini-
component of tially have only a z-component of the velocity (the x or r-component of the
velocity velocity being zero). At the end of the iteration process the emitter will
have a curved surface, and there will be a velocity in the radial direction.
Figure 130 shows the relative radial component (VELX/#VELMOD) of the
velocity. Mostly the radial component is about zero (which is why it
appears to be emitted from a constant EZ contour) but at the outer edge the
radial velocity at the start of the track is negative.
The absolute value of the velocity has been specified with the expression
$PARA NAME=#velmod,VALUE=sqrt(velx**2+vely**2+velz**2).
The files required to experiment with this example are on the distribution
CD: plasma_2d.op2 and plasma_2d.emit.
The OPERA-2d Pre and Post-Processor can import DXF data files to either
create construction lines or polygon regions.
Input Options
The import facility is accessed by selecting
The dialog as shown in Figure 131 requires the user to specify some (or all)
of the following information:
The data type Points in the DXF file may be represented in two ways:
• Construction line crosses i.e. pairs of construction lines which
intersect at the point. If this is chosen, then the Size of crosses (the
length of the construction lines) may be specified.
• A Sequence of construction lines i.e. each line connecting an adja-
cent pair of points.
Solid and Trace data types in the DXF file may be included as Rectangu-
lar strips (SHAPE=H) in the same way as the Polyline option above. For
these regions the number of subdivisions and material number must be
given.
-Z-
zone contours ................................ 47, 90