User'S Manual Fld7 Complex Potential Electric Fields: O.W. Andersen
User'S Manual Fld7 Complex Potential Electric Fields: O.W. Andersen
Andersen
PROGRAM INSTALLATION
FLD7 is transmitted as a zip-file. It is extracted and installed in any directory (folder). The program can also
be installed on a memory stick and run from there.
Here all the Command Prompt commands and file names will be in capital letters. However, they are case
insensitive, and small letters can also be used.
RUN DEMO.INP
After a few seconds, a field plot with 9 equipotential lines appears on the screen. It has been drawn on a
Visual Basic Form. If the picture appears to be cropped or too small, adjust the file SIZESCR.FIL. At the
same time a bitmap picture file PLOTFILE.BMP has been produced. Close the form and enter command:
PLOT
The plot now reappears in a standard Windows program. The tic-marks to the left and at the bottom show
the positions of horizontal and vertical finite element grid lines. Electrodes are red, and insulation is green.
If it is now desired to print the field plot, crop the picture file first to remove empty space and save it.
Microsoft Office Picture Manager or Microsoft Paint can be used for that. Rather than printing it directly, it
is recommended to transfer the picture file to Microsoft Word. Here it can easily be resized and comments
added before printing.
Output from FLD7 is stored in file OUTPUT. To display it on the screen, enter:
FILE OUTPUT
Batch command FILE starts the standard Windows program NOTEPAD. It will be used here for viewing,
editing and printing text files. The first time it is invoked, it should be set to Courier New size 9, word wrap,
and to no top and bottom extra text when printing. The window should always be maximized.
-2-
INPUT
FILE DEMO.INP
What the numbers mean can be found on the input sheet, page 5. For an explanation of what else can be
done with the input file, copy it first to a new file with the command:
HEADINGS NEW.INP
FILE NEW.INP
The abbreviated headings on the input file also explain the numbers. With a little experience, that explanation
suffices to enter new numbers and to make up new input files.
Old input as similar as possible is first copied to a new input file. Then headings are introduced and the file
changed. Numbers always start in columns 1, 11, 21 and so on. They can be entered with or without decimal
point.
Before the new file can be run, the headings must be removed. Do this first with:
CLEANUP NEW.INP
A file without headings can have headings introduced and be viewed at the same time with:
HEADFILE NEW.INP
-3-
Input must be entered very carefully, following explanations on the input sheets and instructions elsewhere in
this manual. Small mistakes like a comma instead of a decimal point or a number starting in the wrong
column are not tolerated. Some mistakes are caught by the program and are explained on the output. Another
way to catch mistakes is by giving a command such as:
CHECK NEW.INP
The input must here be in the standard format, without headings. A picture similar to a flux plot, but without
flux lines, will be displayed on the screen. Mistakes with the geometry can be caught this way.
Headings can also be removed and the file run at the same time with:
CLEANRUN NEW.INP
Another command which does the same, and also modifies the input grid automatically, if necessary to
satisfy program requirements, is:
CORRECT NEW.INP
Modified input will replace the original input, in this case in file NEW.INP. If desired, the original input can
be retrieved from file INP1.FIL.
Before modification of the grid with CORRECT, it should be reasonably close to being right, and the number
of vertical and horizontal break lines should be less than the maximum 64 with a good margin. However,
even so, there is no guarantee of a successful modification.
-4-
PROGRAM DESCRIPTION
O.W. Andersen, "Finite Element Solution of Complex Potential Electric Fields", IEEE Transactions on
Power Apparatus and Systems, vol. PAS-96, July/Aug. 1977, pp. 1156-1161.
Program FLD7 calculates electric fields in configurations with both capacitive and resistive distribution of
potentials. It is an extension of the earlier and more widely used program FLD6, which was made for a
purely capacitive potential distribution, and which is described in the paper:
O.W. Andersen, "Laplacian Electrostatic Field Calculations by Finite Elements with Automatic Grid
Generation", IEEE Transactions on Power Apparatus and Systems, vol. PAS-92, Sept./Oct. 1973, pp. 1485-
1492.
Numerous improvements have been made since these papers were written. One of them is that the finite
element equations are now solved directly by Gaussian elimination, instead of as previously by iteration.
Iterative solution used to have significant advantages in terms of speed and storage requirements, but with
newer computers, it is more important that Gaussian elimination is safer and more accurate.
The grid generation has also been vastly improved, to give the best possible grid under all circumstances
with maximum numerical accuracy, and a minimum of restrictions on the input specifications.
An important use of program FLD7 is in configurations where the potentials have a purely capacitive
distribution, but where there are conducting surfaces with floating potentials. Such surfaces can be entered in
FLD7 with a surface conductivity which is high enough to equalize the potentials, but low enough to avoid
numerical problems in the solution. The results have been checked against theoretical values calculated from
the distribution of capacitance for some sample cases, and have been found to be extremely accurate.
-5–
The ability to handle non linear resistive materials has been added after the IEEE paper was written.
Complete solutions must be made about 12 times in succession, with adjustments of the conductivities,
depending on the calculated field strengths. Therefore, for the running time to remain reasonably short in
such cases, the grid should be made fairly coarse, if possible with the number of nodes limited to about 5000.
The solution is still made by complex potentials, and therefore ignores the higher harmonic components in
the current between the electrodes due to the non linearities.
The adjustments of the conductivities are made as explained for the example in the figure.
1/G
10 log scale
= 2.35(kV/mm)2.79
3
2
0.3 3
0.1 1
0.03
0.01
0.003
E2 E1
0.001 E kV/mm
0.03 0.1 0.3 1 3 log scale
-6-
An initial value 1 of the conductivity is chosen. At the first iteration, the calculated field strength is E1. This
corresponds to the higher conductivity 2. But if 2 is chosen for the second iteration, it may reduce the field
strength to E2, and the result may be even worse than at the outset. Therefore, a relaxation factor R is applied
in the formula:
ln 3 = ln 1 + R(ln 2 - ln 1)
For the curve shown here a good value for R has been found to be about 0.4. Various schemes have been
tried for the program to adjust the relaxation factor continuously, based on the results from preceding
iterations. But none of them has worked as well as the simpler scheme with a constant R, which has finally
been adopted.
Characteristics, codes and recommended relaxation factors are given on the next page for four non linear
coatings and one non linear bulk material. Points on the characteristics are in file NONLIN.FIL, where the
data can be changed and new data added very easily.
COMPLEX POTENTIAL ELECTRIC FIELDS PROGRAM FLD7 INPUT SHEET
Numerical data are entered with the first digit in columns 1,11,21 etc., as indicated. (Does not apply to coordinate lines.)
Decimal point is optional.
SPECIFICATION OF INPUT
LINE 1. The identification can consist of up to 80 characters, including blanks. Most combinations of
letters, numbers and special symbols on the keyboard can be used.
LINE 2. The drawing is always oriented with the x or r-axis horizontally, and with the y or z-axis vertically.
The origin for input coordinates can be anywhere as long as all the coordinates come out positive, but output
coordinates always have:
Axi-symmetric problems always have the z-axis to the left. The minimum radius is the true radius from the
axis of symmetry. It will be equal to zero if the axis of symmetry is the left field boundary.
The scale of the input drawing (a smaller drawing has a smaller scale) should nearly always be specified as
one. In any case, it should be between 0.1 and 10, for the output to contain a sufficient number of significant
digits. If in fact the scale is outside of this range, it should be multiplied by an appropriate factor, for
example 0.1 or 10, to make it come out right. The specified minimum radius (col. 11) and the output
information must then also be modified accordingly.
LINES 3 TO 10. Line numbers are entered for vertical grid lines, where the grid density changes (grid
density break lines). In the example in figures 3 and 4 of the IEEE paper "Laplacian Electrostatic Field
Calculations --- ", they are 1, 13 and 25.
Field strengths are assumed to be constant within each triangle in the numerical solution, and the grid should
be fine enough for this to be true within acceptable accuracy, especially in critical regions. In any case, the
grid lines must be close enough, so that contour line points are at least as far apart as the grid lines in the
same region.
LINE(S) 11. One or more lines which contain x or r-coordinates of the grid density break lines are entered
here. Up to 10 coordinates are entered on one line. Each coordinate consists of a five digit number with one
digit after the decimal point. Then there is a blank before the next coordinate. When a coordinate is less than
1000 mm, it is entered with leading zeros. In the example, the coordinates are:
LINES 12 TO 19. In the example, horizontal grid density break line numbers are 1, 9 and 39.
LINES 21 TO 23. The detailed output information increases the bulk of the output and the computing time
significantly, and should therefore be limited to those areas where it is of real interest. The field strengths in
triangles adjoining those with the highest calculated values will give valuable information about the accuracy
of the results.
If only one area of detailed output is desired, lines 22 and 23 are filled out with zeros.
LINE 24. The scale of the field plot is unimportant for the plot on the screen. It is used for the plot on the
printer or plotter, but if the value is given too large, it is reduced automatically by the program. The plot is
automatically tilted 90 degrees, if this permits a better utilization of the paper.
LINES 25. It is important that the contour lines are entered in a proper sequence, since the grid is fitted to
line segments that are common for two or more lines only the first time they are entered. This usually means
that electrodes should be entered first, so that they receive the proper potentials and codes throughout.
The number of points (col. 1, line 25) is used only to enable the program to read the input properly. It
actually means the number of pairs of five digit numbers, entered in coordinate line format. This should be
kept in mind when preprogrammed shapes (see separate instructions) are used to describe the contour lines.
Conducting surfaces with floating potentials can be given a surface conductivity 9000 1/G. This is usually
sufficiently high to equalize the potentials, and sufficiently low to avoid numerical problems.
COORDINATE LINES
A line can contain one, two, three, four or five pairs of coordinates, codes, etc. When more lines are needed
because there are more than five pairs, only the last line can have fewer than five.
Each coordinate usually consists of a five digit number, with one digit after the decimal point. Then there is a
blank before the next coordinate. The x or r-coordinate is entered first, then the y or z-coordinate.
When a coordinate is less than 1000 mm, it is entered with leading zeros. Example:
The preparation of the input drawing for transfer of coordinates should be done with great care.
2. If the boundary conditions make it desirable, estimated flux lines can be drawn in, deleting parts of the
section from the calculations (Fig. 10 of IEEE paper "Laplacian Electrostatic Field Calculations -").
This is rarely necessary.
3. Positions of grid density break lines (Fig. 3 of the IEEE paper), whose coordinates are to be entered on
input lines 11 and 20, are marked on a horizontal and a vertical line on the drawing. The spacing between the
lines can also be put in, and the line numbers which are to be entered on input lines 3-10 and 12-19. During
this process it is important to make mesh sizes reasonably in accordance with the requirements in the various
areas, with the finest meshes in the regions of the highest field strengths. It is also important to observe
requirements 8 and 10 in the "Instructions for Entry of Coordinates". Grid lines must always be spaced at
least as close to each other as the contour line points in the same region. If they are not, the situation may be
remedied by running program CORRECT (see page 3).
4. Now all the points along electrodes, dielectric materials and estimated flux lines, which are to be entered,
are marked on the drawing with their coordinates. This is done in accordance with "Instructions for Entry of
Coordinates".
1. The position of the reference point (origin) is unimportant, except that coordinates should always come
out positive.
2. The grid lines are horizontal and vertical, and the drawing should be oriented accordingly. For axi-
symmetric problems, the axis of symmetry must be in the vertical direction (the same as the y-direction in
flat, two dimensional problems), with the axis to the left.
4. Contour lines can be open or closed curves. Single points can also be entered. For closed curves, the last
point must have the same coordinates as the first. Only for closed curves, relative permittivity can be given
different from zero. Whether the curves are entered in a clockwise or counter-clockwise direction is
unimportant.
5. When points are supposed to have the same coordinates, or a point is supposed to be on a line segment
belonging to another contour line, a maximum error of 0.01 mm is permitted, to allow for slight inaccuracies.
- 12 -
6. When the same line segment is entered twice, the grid is only fitted to the line segment the first time.
Therefore, whenever electrodes have line segments in common with other contour lines, the electrodes should
be entered first, to make sure that all the points on the electrodes receive the proper potential and code.
7. A maximum of 150 points is permitted per contour line, for a maximum of 150 lines, but the total number
of points must not exceed 1500. If more than 150 points are required for a contour line, it must be divided up
into two or more parts.
8. The points should not be closer to each other than necessary, and with a spacing at least equal to the mesh
size in that region (except when the points coincide).
9. If a contour line is put in outside of the section boundary, the program will move it to the section
boundary.
10. When part of a contour line follows a section boundary, this part must be entered first, and points inside
of the boundary must be further away from it than half a mesh size. This avoids problems that otherwise
might arise due to grid points being unavailable on the boundary when this part of the line is fitted.
11. If a portion of the section is to be excluded from the calculations, a closed curve around this portion is
specified with code=1. Permittivity, potential and conductivity should be given as zero. Equipotential lines
will enter at right angles.
12. Conducting bodies with floating potentials can be treated as dielectric materials with a high fictitious
permittivity, such as 9000. For conducting surfaces, a high fictitious surface conductivity is specified, such
as 9000 1/Gohm.
13. Problems have sometimes arisen when three lines meet in a point at sharp
angles, as shown in the figure. In such a case, the middle line must not be entered
last, because grid points may then no longer be available for the line fitting.
14. An insulated electrode is entered with the outer contour first, with specifications for the insulation, and
then the inner contour, with specifications for the electrode.
If only one contour line is used to describe the electrode, the program will still work if point 2 is put in twice,
also as an intermediate point between points 1 and 3. But the program running time will be longer than
necessary, and the field strengths inside of the electrode might not come out exactly zero, because of
numerical errors.
Intermediate points are only put in automatically by the program when they come from other contour lines,
as in the example in the section "Preprogrammed Shapes".
- 13 -
ERROR MESSAGES
The program aborts and prints out one or more of the following error messages, if it is necessary to modify
the input.
6. The permittivity is given different from zero, and the last point on the contour line is not equal to the first.
7. The line fitting was interrupted for line segment - near -. (Coordinates define the line segment and the
point where the line fitting got stuck.)
8. This point is too close to the preceding point -. (The point is defined by its coordinates.)
9. The program has moved node no. - from one line to another line which is too close. (The node is defined
by its number and its coordinates.)
10. The area came out negative or zero for triangle with -. (The triangle is defined by its node numbers and
coordinates.) Change line sequence or move grid lines.
11. Coordinates of grid density break lines are not in increasing order.
12. Open lines with resistive coating go through more than 1000 nodes.
- 14 -
PREPROGRAMMED SHAPES
Two preprogrammed shapes that can be entered in a simplified manner are used very frequently, circular arcs
and rectangles with horizontal and vertical sides. The way they are entered will be explained by means of the
example in the figure.
mm electrode
50
30
25
=2 =3
10
electrode
0 30 45 65 95 110
The bottom electrode is specified only by its two end points. The intermediate points are put in automatically
by the program.
The region with = 2 includes a circular arc at the upper left corner. It is somewhat roughly drawn, but is
supposed to have 5 points, between which straight line segments are drawn by the program. It can be
specified as follows:
0030.0 0000.0
0065.0 0000.0
0065.0 0025.0
9000.0 0001.0 Codes
0005.0 0015.0 Number of points and radius
0045.0 0010.0 Coordinates of center (can be negative)
0090.0 0180.0 Starting angle and finish angle (degrees)
0030.0 0000.0
An x-coordinate 9000 mm is recognized by the program as a code, telling it that a preprogrammed shape of
some kind follows. The value 1 in the corresponding y-position indicates that the shape is a circular arc.
Angles are measured counter-clockwise from the horizontal, in the usual way. If the contour line progresses
through the arc counter-clockwise, as in the example, the finish angle must be greater than the starting angle.
Conversely, if the progression is clockwise, the finish angle must be smaller than the starting angle.
The region with = 3 is a rectangle with horizontal and vertical sides, which can be specified as follows:
Right angle insulation barriers with sharp and rounded corners have also been preprogrammed, because of
their use in insulation structures of large transformers and reactors.
X2, Y2
9000.0 0004.0 Codes
X1 Y1
X2 Y2
T1 T2
Th’k = T2
Th’k = T1
X1, Y1 X1, Y1
X2, Y2
The angles are similar to the ones shown above, except that the right angle corner is rounded. This makes
T1=T2=T, and the specifications are:
In the region of the rounded corner, it is important that the grid is sufficiently dense. It will usually be
satisfactory to have the line spacing no more than T or (outside radius)/5, whichever is smaller.
7 contour line points are generated along the inside radius, 8 along the outside radius.
- 16 -
POST PROCESSING
After the main program has been run, the run identification and all the essential calculated and input
information are in file FOR.FIL, and can be retrieved for further processing. This includes all the potentials
and node and contour line coordinates.
A post processor is supplied with the program, both in source code and in executable versions. It draws a
graph of field strengths along or perpendicular to a specified line, and is started with the command:
GRAPH
If an open contour line is drawn only for the purpose of calculating field strengths along it, permittivity,
potential and code should all be given as zero. Points along it must be specified wherever it crosses or
touches other contour lines.
- 17 -
The Command Prompt window should be maximized and the size adjusted to fill the screen after right
clicking the top title bar. Cursor size small and letter size 12x16 pixels are recommended. If Command
Prompt goes into full screen mode by an application, it can be brought back with Alt-Enter.
Since many PC users are not familiar with Command Prompt, here are some hints and frequently used
commands. The commands are examples and may be modified in obvious manners. Large and small letters
are interchangeable.
Other commands:
C: Moves to unit C or another unit.
CD\ Changes to base directory.
MD GRAPHICS Makes directory GRAPHICS.
CD\GRAPHICS Changes directory to GRAPHICS, just below the base directory.
COPY OLD.INP NEW.INP Copies old file OLD.INP to a new file NEW.INP.
COPY /? Explains options available for command COPY.
REN OLD.INP NEW.INP Renames OLD.INP as NEW.INP.
DEL OLD.INP Deletes OLD.INP.
DIR *.INP Lists all files in the directory with extension INP.
DIR *.I?? Lists all files in the directory with three letter extension starting with I.
START NOTEPAD OUTPUT Invokes Windows program NOTEPAD with file OUTPUT.
START PLOTFILE.BMP Starts a standard Windows program to process the bitmap file.