Professional Documents
Culture Documents
Met Glas
Met Glas
by
Edward Zhi-Wei XUE
A thesis submitted to
The University of New South Wales
for
The Degree of Master of Engineering
1
UNIVERSITY OF N.S.W.
2 6 AUG 1993
LIBRARIES
CERTIFICATE OF ORIGINALITY
Signed By: .
2
ACKNOWLEDGEMENTS
3
Abstract
4
saturation 1{lux density which compare to the conventional
direction.
5
Contents
Chapter 1: Introduction....................................................................................9
6
Chapter 5: Causes Of The No-Load Power Loss In
Metglas Amorphous Ribbon Material ..................... 5 5
Chapter 9: Conclusion.................................................................................................137
Appendix I..........................................................................................................................139
References............................................................................................................................ 150
8
CHAPTER 1
INTRODUCTION
SOUR TRANSFORMER
BATTERY
ARRAY AND INVEFITER
BATTERY
CHARGER
BsD
DIESEL 240V USER
ORGRD 240V
50Hz
50Hz
13
significantly when it is in half or no load conditions. The economical
benefit are obvious.
*
Information Supplied By RAPS Reserach Group In The Power Dept.of N.S.W University
15
CHAPTER 2
Fig.2-1
17
Metglas Alloy 3% Oriented Silicon
3
Specific Gravity (g/cm ) 7.2 7.65
Table.2-1
Schematic Diagram of Allied's Process for Continuous Casting bv Rapid Quenching iCCL Process)
A Raw materials are melted down m E. Jet is quenched at 10* 'C per second to
induction furnace form an amorphous strip
B Melt is transferred to the feeder section F. Strip wdth & thickness are measured in
C Control of casting head line and fed back for process control
O. Thin film of melt is ejected onto the rapid- G Strip is gu'ded to the thread-up unit for
quenching wheel take-up
H Strip tjko-up
2605S2
H (A/cm)
21
5 605S2
22
Total annual core loss: 85W*24*3600*365= 2.687 GW
Total annual load loss: 170W*10*3600*365= 2.234 GW
Total annual losses: 2686560+2233800= 4.92 GW
Core loss contribute 54.5% of total annual losses
load loss contribute 45.5% of total annual losses
25
The mechanical bonding and plastic deformation give
POWErCORE strip many of its physical characteristics. The strip is
rigid enough to be handled in transformer cutting, punching and
stacking operations. It is tough enough to be coiled and recoiled and
fed through strip processing equipment. However, if bend around
too tight a radius, the strip will delaminate into its individual plies.
Also POWERCORE strip is very strong under tension with an
ultimate tensile strength of about 500-750 MPa. But, if the load is
not applied evenly through the strip thicness, it may fail in stages
at half these values. Finally, the local plastic deformation between
adjoining ribbons is consistent with the increased pack factor of the
strip.
26
METGLAS ALLOY POWERCORE
2605-S2 RIBBON STRIP
Specific Gravity (g/cm3) 7.2 7.2
Lamination Factor (%) 85 90
Young's Modulus (GPa) 180 150
Thermal Conductivity (W/m.K) 8 9
Electrical Resistivity (pO-cm) 137 -
Saturation Induction (T) 1.58 1.58
Remanent Induction (T) 1.30 1.20
Coercive Field (A/m) 1.60 3.20
Curie Temperature (K) 688 688
Core Loss 1.4T/60Hz (W/kg) 0.15-0.25 0.25-0.28
Exciting Power 1.4T/60Hz (VA/kg) 0.25-0.45 0.50-0.80
Table 2-2
27
However, while new opportunity is exciting, there still
remains much to be done in the way of material development,
material manufacturing, transformer engineering and transformer
manufacturing if the advantages of amorphous metal cores are to
be fully utilized.
28
CHAPTER 3
29
Hardness --- Amorphous metal is extremely hard ( diamond
point hardness of approximately 900). Its ultimate tensile strength
is more than four times that of conventional electrical silicon steels.
These characteristics, in combination with its thinness, make the
material exceedingly difficult to slit and shear; therefore
transformer design using as-cast widths or requiring only a few
cuts in the material ( i.e: jointless core ) are desirable..
30
additional stresses are introduced in the core manufacturing
process, amorphous metal must be annealed in the presence of a
magnetic field and an inert atmosphere, such as nitrogen, argon, or
helium to develop the best magnetic properties. The annealing of
the amorphous metal can serve three purposes:
1. magnetic orientation;
2. relief of residual stresses;
0>
Core Skin
Time (hours)
Fig.3-1 Typical temperature profile of anneal process
alignment) along the ribbon direction. The field level used through
the anneal process was 800A/m, which was chosen to ensure that
the core material is fully saturated during the annealing. To fully
develop this anneal-induced anisotropy, the core material must
approach magnetic saturation during the annealing. Fig.3-2 shows
the field levels required to ensure core material saturation at
various temperature.
32
to 250pm. Its stacking factor of about 90% is midway between
19
cn
ID
18 (/)
V)
17 Z)
<
16 o
0
_J
15
*
14 1
13
12
11
10
9
8
6
5
20 40 60 80 100120 140 160180 200 220240 260280300 320340 360 380400
temperature cq
Primary
.Secondary
Tore
36
CHAPTER 4
The outer member, the coil, in the cases of IA, IB and the core
in the cases of IC and ID, can not rotate around the inner member.
Thus, the outer member must be wound around the inner member
by inserting the end of the coil conductor or corestrip material into
the centre hole, pulling it out and tight around the inner member,
and repeating the process again and again. Currently available
production machines (toroidal winders) are not suitable for this
group because the numbers of turns which can be continuously
wound by such a machine is inversely proportional to the size of
the centre hole. For distribution transformers, the size of the
37
conductor may leave a choice of either using a large centre hole
which would need additional core material and conductor material
required to surround the large centre hole or having a large
number of joints in core or conductor, which will cost more in
labour.
Group three (III) results from the group two conversion of the
inner member from an overall circular shape to an overall
rectangular shape with the same type of cross-section (See Fig.4-3.).
But for those configurations where only one member could rotate in
group two (IIB and IID), this change eliminates the possibility of
rotation. For those where both members could rotate in group two,
one member still retains the possibility of rotation (IIIA and IIIC).
For assembly of the configurations where the possibility of rotation
is lost in this change, either joints must be introduced or the process
38
GROUP I
A (configuration 1) B (configuration 2)
c (configuration 3) D (configuration 4)
COIL
Fig.4-1.
GROUP II
Fig.4-2.
40
GROUP III
CORE COIL
CORE COIL
Fig.4-3.
41
involves the outer member being inserted through the centre
window and pulling tight around the inner member as described in
group one. In this group, one variation of configuration IIIB is the
popular rectangular coil core-form distribution transformer and one
variation of configuration HID is the familiar rectangular coil shell-
form distribution transformer.
43
Basic Configu Coil Cross Core Cross Core Type Configuration
rations * section section Number
44
4.2 Core-Coil Assembly Options
46
Fig.4-6 .
Jointed wound cores and stacked cores are usually
constructed separately from the coil, then assembled onto the
previously wound coils.
1 . 6, 10, 12, 15
2. 1, 2, 6, 8, 15, 18
4. 6, 8, 10, 21
50
Option Core-Coil Configuration
1. 12, 15
2. 2, 18
4. 8, 21
5. 23
52
GROUP IV
£ n
Flux___ _
r 1 1
1
hp
L j
1
1
Fig.4-7
53
Configuration 8 and 21 are the inverses of 12 and 15
respectively in that the core is rotated rather than the coil. Like 12
and 15, they have advantages in manufacturing techniques and
disadvantages of high material costs and losses. Because the cores
are not annealed after forming, the no-load losses of 8 and 21
would be among the highest of the configuration possibilities.
54
CHAPTER 5
55
anomalous loss
power loss
per cycle
classical eddy
current loss
Frequency
Fig.5-1.
57
ampere turn
meter
Ndj>
E= (volts)
dt
dB
Power at any instant= E i= H A L- ^ (watts)
dB
Work done in time dt= H-A-L- — dt (joules)
dt
entire cycle) is the area enclosed by the loop. The volume of the
ring specimen is A-L:
Work Done_____
Area of Loop = (joules)
Specimen Volume(m3)
58
produces an induced voltage around each closed path that encircles
lines of magnetic flux. Circulating currents induced in the conductor
by these voltages are known as eddy currents, and the resulting
heat losses are eddy current losses. In an effort to minimize these
losses in transformer cores, magnetic material are made in thin
sheets, or laminations.
(Ttd-f-B V
eddy w/kg (5-1)
59
From experimental measurement it has been shown the
anomalous loss is responsible for more than 90% of the total power
loss of Metglas amorphous ribbon materials at power frequencies.
This compares to the 30% - 90% range for grain-oriented silicon
steel(21). A convenient method commonly used to define the
magnitude of the anomalous loss is to compare to the classical eddy
current loss giving the anomaly factor, which is defined, as:
(pm ) (O)
60
Origin Of The Anomalous Loss
61
The most commonly observed domain structure on amorphous
ribbons is a bar structure which consists of alternate wide and
narrow domain, (as shown in Fig.5-3.)
63
LACK OF FLUX PENETRATION AND DOMAIN WALL
BOWING
64
The local variations of flux density have been measured(34) in
the amorphous ribbon and grain-oriented silicon-steel. It has been
found that the local value of flux density varies from the whole flux
density of the specimen by approximately ±10% for amorphous
ribbon and ±5% for the silicon-steel. There is no obvious change in
these values after strain-relief annealing. It is a quite large
localised variation in flux density in the amorphous ribbon, because
there is no grain-structure which causes the variations in grain-
oriented silicon steel. So the localised variations in amorphous
ribbon is probably caused by differences in its short range ordering,
surface condition, and most probably the localised stresses which
caused by the rapid quenching and quick solidification of the
amorphous manufacturing, and are not removed by strain-relief
annealing. Therefore the variation in localised flux makes a larger
contribution to the anomalous loss in the amorphous ribbon than in
grain-oriented silicon steel.
As Cast
Annealed
0 5 25 30
Tensile Stress MPa
Fig.5-5.
Another thing needed to be mentioned is the variation of
power loss with applied stress. It was suggested the application of
stress should reduce the domain wall spacing and then reduce the
total power loss in the core (Fig.5-4 shows the domain configuration
changing under stress). Some experimental results(29) have shown
that power loss of as-cast amorphous ribbon can be improved by
application of stress but not for annealed amorphous ribbons. Fig.5-
5 demonstrates that the effect of introducing tensile stress at 50Hz
and 0.75T.
68
CHAPTER 6
BATTERY
CHARGER
BSD
DIESEL 240V USER
ORGRD 2 4 0V
' 50Hz
50Hz
Configuration 1.
SOLAR TRANSFORMER
BATTERY
ARRAY AND INVERTER
BATTERY
CHARGER
CHANGE OVER
SWITCH \
DESEL 240V USER
ORGRD 240V
Configuration 2.
SOLAR TRANSFORMER
BATTERY
ARRAY AND INVERTER
END
DIESEL USER
OR GRID 240V 50Hz
Configuration 3.
Fig.6-1
70
h 20 ms
H
Fig.6-la
71
6.1 Non-Sinusoidul Power Supply in Remote Area
f(t)
i i
20 ms
57C
3
-► cot
271
3
-E
Fig. 6-2
< t < X
ECO =
- <
5x
-E X < t <
< t < 2x
72
As we know, for a function f(t) which has periodic of perbd T,
if E(t) satisfy a set of rather general conditions. It may be
represented by an infinite set of sinusoids, the Fourier series:
ao V
E(t) = W + 2j [ancosncot + bnsin(ncot)]
n=l
a<)= ^ JE(t)-dt
| 2tc
2n 5n
1 3 3
an=“[ jE*cosncot-d(cot) + j* (-E)-cosncot-d(cot) ]
E , . 2nn 5njc
( sin~— - sm~z— )
nrc 3 3
73
bn= ~[ jE sin(cot)*d(cot) + J(-E)-sin(cot)*d(cot) ]
0 n
E 5n7t 2nn
= (1+ cos~r~ - cos - cosnii)
njr 3 3
For the first 40 items (n=l to 40), the calculated results for an,
bn and the effective harmonic value V an + bn are as follows:
an bn V an2+ bn2
VIE 2VIe
ai=
K
b,= ^
K 71
VIE u 3E
2VIe
a’"' 5k bs~ 5jc 5k
VIe u 3E
2VIe
1l
b,= 7 it Ik
VIe , 3E
2VIe
a“_' llTt b“~ lire 1 1 7C
VIe 1 3E
2VIe
a,3_ 13 it bl3_ 13tc 1 3 7T
VIe , 3E
2VIe
“17 ' 17tc b,,~ 17* \1k
VIe 1 3E
2VIe
a,,_ 19 it bl9 1971 1 9ti
b. V a„2+ b„J
V3E 2^13E
■ 3E
*“"23* b23_23jc 23k
V3E , 3E
2^13E
a“_25jt b2S_25Jl 25k
V3E 2a/Ie
1 3E
a2,_29* “29~29k 29 k
V3E , 3E
2^~3E
a’1_31 j: b,,_3l7t 31k
V3E , 3E
2^3E
aM-357t b33“35jc 35k
V3E 2^/lE
a5,"37)t b3,"37jc 31k
75
Amplitude
E-
Harmonic Spectrum For The Special
Square Waveform In Fig. 6-2
I i___i I___ I
0.1 0.2 6.3 0.4 6.5 0.6 6.7 6.8 <5.9 fo /.1 1.2 1.3 1.4 1.5 1.6 1.7 18
Frequency
Fig.6-3
Without Ann«a(u
77
Magnetization curve for Metglas amoiphous
metal core prototype transformer
l(mA)
Fig.6-4
78
and the temperature is applied together with an applied magnetic
field of 750A/m through out the annealing. The purpose of this
measurement is to find out the magnitude differences of the core
loss in the range from 0 to 0.7T flux density before and after anneal
process under the pure sinusoidual power supply.
Oven Temp.
11!ore: Skin
Time (hours)
U
and B =
4.44-f-N-A
_____ U_____
B= (6-la)
V2-4.44-f-N-A
79
Where: U is the applied voltage, f is the applied frequency, N
is the turns of the primary coil, A is the core cross-sectional area
and B is the flux density. In this measurement, pure sinusoidal
frequency is applied. From Eq.6-la, we can calculation the flux
density values through applied voltage U. For example, if we want
to make flux density B=0.1T, According to Eq.6-la:
U=4.44-f-N-B„,-A=4.44-50-130-V2-0.1-0.05-0.045=9.2(V)
The calculated flux densities and the measured results for the
core losses are shown in the Table6-1 following table and plotted
in curve in Fig.6-6.
Table 6-1
Flux Density (T) 0.1 0.2 0.3 0.4 0.5 0.6 0.7
Applied
9.2 18.4 27.6 36.8 46.0 55.2 64.4
Voltage (V)
80
50 Hz core loss of Mstglas 2506-S2
before and after anneal
before anne il
after anneal
B (T)
Fig.6-6
W/Kg 0.24 0.45 0.72 1.04 1.36 1.71 2.05 2.41 2.79 3.17
4.0
3.5
3.0
2.5
2.0
O 1.5
1.0
0.5
0.0
0 100 200 300 400 500 600
Frequency
Fig.6-7
82
From the measurement, it shows that at 0.55T flux density,
the loss in the Metglas 2605S-2 amorphous core is nearly linear
increased with frequency in the range from 0 to 500 Hz. This would
indicate that very large portion of the core loss are hysteresis loss
(since hysteresis loss is proportional to the applied frequencies).
While eddy current loss are negligible (since eddy current loss is
proportional to the square of the applied frequencies). This matches
the physical properties of this material which is very thin in
thickness and has large electrical resistivity.
83
Table 6-3
^w/kg\|W 10 20 30 40 50 60
150Hz
250Hz
,350Hz
l450Hz
550Hz
Voltage
From the above data table and graph, it can be seen that at a
certain applied voltage U, the losses are different for different
frequencies (frequency range from 50Hz to 550Hz). It shows that at
the same applied voltage, lower frequency causes higher losses that
84
because of the lower applied frequency gives a higher flux
density(see Eq.6-1).
= Vl-0875x2
x= 0.95 U (6-2)
Table 6-4
Applied
Voltage U(V)
10 20 30 40 50 60
Fundamental
X(V) 9.6 19.2 28.8 38.4 48.0 57.0
5th Order
X/5 (V)
1.92 3.84 5.76 7.68 9.60 11.5
7th Order
X/7 (V) 1.37 2.80 4.11 5.48 6.85 8.2;
11th Order
X/11(V)
0.87 1.78 2.61 3.49 4.36 5.2^
85
It means that for the modified square waveform in Fig.6-1, a
10V applied voltage is equal to the superposition of a applied 9.6V
with 50Hz plus 1.92V with 250Hz, 1.37V with 350Hz and 0.87V
with 550Hz (neglect other higher orders harmonics, see Table6-4
inlOV column). Then we measured the losses at those voltages of
the above four different frequencies and superposed them to get
the losses of the special square waveform. Their corresponding
losses are shown in Table.6-5 and plotted in Fig.6-8 compared with
the losses of 50Hz pure sinusoids:
Table 6-5
LossVw
10 20 30 40 50 60
Harmonics
Fundamental 2.354
0.178 0.435 0.742 1.202 1.722
X(V)
11th order
0.000 0.001 0.002 0.005 0.008 0.011
X/11(V)
86
some more power loss in the Metglas metal transformer core (see
Fig.6-9), but not seriously.
B<T>
0.22 0.33
Special Square
0 .25 - Vaveform
0.20-
0.15-
Sinusoidual j
Waveform
0.10 -
0.05-
Voltage (V)
Fig.6-9
87
Department, School of Electrical Engineering, UNSW, which is able to
simulate various shape of waveforms by programming two EPROM's
through a micro-processor to control the width of pulses. The
sinusoidaul Pulse Width Modulated Output Waveform and principle
of the controller diagram is shown as following:
20ms
Over-current Output
Sensing Switching
Logic
Over-temperature
Sensing
Fault
Battery MICRO Indication
Momitoring PROCESSOR Display
Singal
Conditioning
Solar
Cell
Input Watc hdog
88
We did try to apply this new system to our measurements to
measure the loss of the modified square waveform, but we couldn't
get the repeatable results since its stability is still not quite able to
match our requirements. But it will provide a direct and convenient
method for the similar experiments and measurements after its
stability is improved.
89
CHAPTER 7
90
where qx is the heat flow rate along X direction in (watts). A
2 dT
is the area normal to the direction of heat flow in (meter ); ^ is
^ = -tVT (7-2)
92
value, in general, is not uniform over the surface. Thermal
convention relies almost entirely on empirical data to provide the
magnitude of the heat transfer coefficient.
H=c(T4-T2) w/m2
93
7.2 Several Dimensional Heat Conduction Equation
3T
(qx+vqz)+ q(dxdydz) = (q*+dx+Vdy+ciz+dz)+ cp(dxdydz) —
where the rate of heat generation per unit volume q , and the
temperature T, are in general functions of the three coordinates x ,y
,z as well as of time t, the specific heat c and the density p are
independent of temperature.
94
Fig.7-1
The rate of heat conduction into the element across the left
side in the X direction, q can be written as :
Qx =(- ^ ^ ) dxdy
a aT
9x - qx+dx = ^(* dxdydz
95
and similarly for the y and z directions :
d dT
% ■ v<iy = ^ ^) dxdydz
d dT
q* - qz+dZ = ^ ^ ) dxdydz
a 3t a 3t a 3t aT
^(*^)+^(*^)+^(*^) + q = cp*
a'T a2T q i 3t
«. 2 + •% 2+-n2+» —
dz k a X-^ (^*5)
ax dy at
k
where the constant a = — is called the thermal diffusivity and
cp
has the unit meter2/sec. in SI system. Equation (7-5) is known as
the general heat-conduction equation and governs the temperature
distribution and conduction heat flow in a solid having uniform
physical properties.
96
If the system is steady, but heat sources are present, equation
(7-5) becomes the Poisson’s equation :
tfr cfr
= 0
dy2
97
In the following, the above analysis will be applied to the
problem of heat distribution and transfer in a core made of Metglas
amorphous material. Because the thermal, electrical and special
characteristics of Metglas differ from normal silicon steel cores, the
thermal response will also quite different.
98
CHAPTER 8
99
8.1 Heat Transfer In The Cross-Section Of Amorphous Core
Fig.8-1
100
Here, we choose the as-cast rectangular unjointed wound core
as an example for analysis (referring to configuration IIB in Chapter
4). In Fig8-1. is the cross-section of such a core which is made by
amorphous metal 2605S-2. A rectangular coordinate system is set
for the analysis.
y
T4=f4(y)
Fig.8-2
32T 02T q
= 0 (8-1)
dx2 + dy2 + ^"effective
T^fjCx) T2=f2(y)
T3=f3(x) T4=f4(y)
101
The problem here is to find out the effective thermal
conductivity in this cross-section area and then the solution of
equation (8-1).
102
scalar; it represents the ability of heat transfer of a certain solid
body. In what follows, we shall restrict ourselves to the discussion
of laminated structures and calculate the heat fluxes normal and at
right angles to the laminate. Although the heat fluxes not only
transfer normal to or at right angles to the laminate, heat flux is a
vector and it can be divided into two components along X axis and Y
axis respectively. Only steady-state conduction is discussed.
Fig.8-3
103
Here, T , Tb, , , , are the temperatures at the internal planes of
contact and :
m
8=18;
j=l
qs, qs2
so + )
k1 +k2
Si+S2+.... +5m
and k
n S1
+ 52 Sm Jl,
ki
, + ..... + r1
k2
X
V 8-
j=l
ki
(8-3)
This equation looks like resistance connected in series and
consists of a sum of reciprocal conductivities because it can also be
written as :
...... h
kn kl k2 km ^ ki
j=l
k.„k
air^ metal
then kn - (8-4)
8(1-X) k.,,X + ka„. i(l-X)
-------+ —;---------
' metal
104
Heat Flow Parallel to the Layers .. Now we consider
another general circumstances, the structure isshown in Fig.8-4,
which consists of m layers of thickness, Sj, s2, s3, , , , sm ofmaterials
whose thermal conductivities are kx, k2, k3 , , ,km. The total
thickness is S..
Fig.8-4
105
here kis the effective thermal conductivity paralleled to the
layer. We substitute the expressions from (8-3) and (8-4), obtain:
X¥i
(k\s\ + ^2S2 + + kmlm ) i=l__ (8-7)
5l + 52 + + S,
kmcMXS + *,lr(l-X)S
then £p = ^ + £.ir(l-X) (8-8)
5 6
T1 =
in
s v 5i
xvi k.
j=i j=i j
T1 =7^ = (8-9)
w 5j_ 82
(& 18 ^ + ^282 +
)(k,+ k2 +
106
multiplication in Equation (8-9). That fact indicates heat transfer
along the lamination direction faster than it transfers along the
thickness of the laminated body.
k.A
air™- m ctal
n= (8-11)
[^metal+M^air^metal)] f ^air-*" ^(^metajair) ]
__ air'*' metal
Here, we define :
107
Since k * is symmetric with respect to the indices the metal
and the air, it means that interchanging these two different
materials changes nothing in the properties of the compound.
Form Equation (8-4) and (8-8), we can find that the effective
thermal conductivities kn and £p are determined by the lamination
stacking factor X, the thermal conductivity of air &tir and the thermal
conductivity of Metglas metal kmtltl. So, if for a Metglas metal
transformer core with:
= 6.804 W/mK
K 0.21
*1 ~fr~= ft gQ4 “ 0.03 ; (kp is about 33 times of ^n) (8-1 lb)
= 0.49 W/mK
108
*P = + Ul-M= 8.0 x 0.95 + 0.0257(1-0.95)
= 7.602 W/mK
= 0.5 W/mK
=23.8 W/mK
109
K_ 0.5
0.021; kp is about 47 times of kn. (8-1 Id)
kp 23.8
Test points
x6 x7 x8 x9 xlO
Fig.8-6
Table 8-1
AB=p=r|j ; AO=8
1 12
then the effective thermal conductivity along x' and y' axis become:
k'=k-
p p
a2r a2r
,2+,-.*+k =o (8-12)
ay' effective
U lr
. __ , __ _________ ^air^mctal
where: effective n k ■ k
'metal (1-A.)
* ^ 7200kg/m3 3
Q= 0.247W/kg x 8.63kg x~8~63kg = 1778-4 W/m
According to our measurment, the boundary conditions are:
f3(y')=f4(y’)=-2873200y’4+271750y’:>-27286y’2+937y’+24.9 (8-13)
a0= 24.9
1 14
36
12
0.05
Y axis
Fig.8-7
Now the conditions for solve the Poisson’s Equation 8-12 are
completed. The differential equation is:
92T' <fT q _
3x'2+3y'2 + ive = °
f3(y')=f4(y')=-2873200y’1+271750y’3-27286y’2+937y’+24.9
8zip 9z(p
and (8-15)
✓
I
V—
U>
1 16
Then the temperature distribution of the core cross-section is:
+ (j>(x’) (8-l5a)
from (8-14), the solution for the problem d(x) can easily be
found to be:
d24> _ q
tlx ^"effective
dd _ q
then. j i~ k x + C\
^effective
qp
C,= rr----------- C2=0
Xeffective
q*x’ ,
So: <j)(x')= (P-x ) where p=rja, x'=T|x (8-16)
effectie
d2<Pi a2<p,
In easel: rr + r^T-O with the boundary conditions:
dx dy
1 17
3‘cp; 32(p2
In case2: + with the boundary conditions:
dx dy
32<p3
In case3: dx,2 + 9^_^ with t^ie boundary conditions:
a:(p4 a2<p.
In case4: + with boundary conditions:
solutions.)
nttx n?ty
118
Similarly, solution for case 2 is:
nn x n7t(5-y X
n oo sinf )-sinh[ 1
, , 2Tor[l-(-l)l ^ K s J T\s
(p:(x,y) =------------------ 2- —
n-n n=1 nrcS
sinh( )
r\s
nrcy nn\
oo sin(-r-)-sinh[-r-]
<P.(*,y) = I -------------- —^ ' g(n)
sinh(~~)
where:
5 n+l 4 3 2
g(n)= [(-1) (a48 +a38 +a28 +a18+a0)+a0] -
nrc
8 8
(—H(-lf !-(12a4S2+6a38+2a2)+2a2] + (—)5'24a4[ 1 -(-!)“]
n 7T nn
119
The constant temperature: TAB=?
The constant temperature: Toc=?
The boundary condition coefficients: a4, a3, a2, a! and ao=?
The effective thermal conductivity kcffecuve = ?
The heat generate rate: q=?
The coordinate transform coeffiecient rj = ?
120
T« C»
T= fjiy1
15cm. -* 5cmi T= 24 S *C
Fig. 8-9a
Fig.8-9b
121
8.4 Comparison of Temperature Distribution Profiles
Between Metglas Metal Core and Conventional Silicon Steel
Core
following assumptions:
6. Assume all the iron loss convert into heat as the heat
source.
122
Case 1: We calculate the Metglas 2506-S2 amorphous metal
core temperature profile with the same analysis in episode 8.3 but
it is force cooling at 30 °C (303K) enviroment:
T44 2
“c
>2
Fig.8-10a Fig.8-10b
cross-section of points where the
Metglas metal core temperature calculated
= 0.21 W/mK
= 6.804 W/mK
123
n = k p = AIL
6.804
- 0.03 ;
q*Tlx
(1) 4>(x)=2^------ (^a-qx)
effectie
n oo sin( )-sinh( )
<»»»*>i-..t *
" 1 sinh(-----)
v rja y
n7tx n7t(b-y)
n oo sin( )*sinh[ ]
2T2[l-(-l) ] ^ qa
(3) <p2(x,y) =
^•n n= 1
nrcb
sinh( )
v rta '
124
nrcy nTi(Tia-x)
sin( g )-sinh[ b
n7iria
sinh( ^ ')
nny mix
sin(-^-)*sinh(-^-)
nrcpa
sinh(_b~)
125
Ti :Ci
Fig.8-10c
x(cm)
Fig.8-10d
126
Case 2: We calculate the silicon steel core temperature
distribution profile.
Fig.8-1 la Fig.8-1 lb
10kg silicon steel with loss 0.63W/kg at 1.35T, Xsttel= 95% and
= 0.5 W/mK
= 23.8 W/mK
127
0.5
= 0.021
23.8
7650kg/m3
q= 0.63W/kg x 10kg 4819.2W/m3
10kg
kH = k = 0.5W/mK
T1=T2=T3=T4=30°C=303K
q-qx
(!) ------- (Tla_Tlx)
effectie
nrcx nrcy
sin(----- )-sinh(----- )
2T,fl-(-l)°l v rja y v -qa y
(2) cp,(x,y) =
7tn n= 1 nrcb
sinh( )
n7ix n7i(b-y)
sin( )-sinh[ ]
2T,ri-(-l)‘l
(3) tp2(x,y) =
7t-n n= 1 njib
sinh( )
128
nrcv n7r(Tia-x)
~fcT>sinh[ b“
njiTja
sinh( 5 )
nny nix
sin( —g ^)-sinh(—)
nTtrja
sinh( b )
129
T' :C
Fig.8-1lc
Fig.8-1 Id
130
Case 3: If we also consider the same silicon steel core to be
operated at its maximun flux density which is 1.7 T, the core loss
under the flux density is l.l\V/kg(20> and the heat generate rate:
Fig.8-12a
131
T( °C)
Fig.8-12b
^0839320
Fig.8-12c
132
Case 4: If we want a METGLAS metal core transformer to
have the same power handling capacity (able to handle same
applied voltage) as in Case3, then it must increase cross-section
area to compensate its lower magnetic saturation flux density to
ensure it can handle the applied voltage (E=4.44-f-N*BmA). Since the
width of the METGLAS amorphous 2605-S2 is fixed at 5cm, so we
can only increase the laminations (the core thickness) to increase its
cross-section area.
a-TT5a= l26A
thus, the Metglas metal core with the same power handling ability
as in Case3 should have the dimensions( see Fig.8-13a):
a=0.05 m
b=0.05 x 1.26= 0.063 m
7200kg/m3
q= 0.26W/kg x 10kg xl.26 x 10kg xl.26 = 1 872W/m3
133
y(cm)
T2
Fig.8-13a
section as before and their locations are shown in Fig.8-13b and get:
1'“ t" “ H-
134
If we do the calculation of 2500 temperatures inside the
cross-section area and plot out them graphically in a three
dimentional coordinate system as before, it shows as following by
also rotate X-Y (horizontal) plane 135°C clockwise when viewed
from above (see fig.8- 13b) and tilt X-Y (horizintal) plane 45°C
V °C)
Fig.8-13c
135
distribution along the core width derection because heat tranfer in
that direction have to go through many layers of air gaps which
resist the transfer and cause the temperature gradient. On the other
hand, temperatures change sharply in the interfaces between core
and ambient along the core width direction that because heat is
generated in the core laminations uniformly and heat transfer
within the metals are much fast than it into the ambient.
136
CHAPTER 9
Conc^a4-ton4:
for transformer core are lower core losses and easy manufacture
the core losses for this material was also discussed. It was
suggested that more than 90% of the total core loss of amorphorus
137
Therefore, the causes of these xnconsxstent domain wall motions
and variational flux densities must be fully Investigated and
to be achieved.
138
Appendix I
Fig.A-l
dimensions:
32T 32T
~ 2+ 2 —0 (A-l)
ox dy
139
If the boundary conditions of the fin are:
T(x,y)=X(x)Y(y) (A-2)
<?X dV
Y7T+X7T (A-3)
dx dy
1 d2X 1 d2Y
(A-4)
' Ydx2 “Xdy2
d^X
^T- + yX(x) = 0 (A-5)
140
d2Y
- YY(y) = 0 (A-6)
A=B=0
X(x)=Asinhocx+Bcoshax, Y(y)=Csinhay+Dcoshay
T(x,y)=(Asinhax+Bcoshax)(Csinhay+Dcoshay) (A-8)
141
Situation 3: If y > 0, the general solutions of Eq.(A-5),(A-6)
are of the form:
nk
d
, n=l, 2, 3, •••• (A-ll)
142
(A-l) which satisfy the boundary conditions T(0,y)=0, T(a,y)=0 and
T(x,0)=0 can be written as:
In general,
f^x) * Ensinanx sinnhanb
oo
oo
d2X
-^r + yX(x) = 0 at X(0)=0; X(a)=0;
143
This is a Sturm-Liouville (see reference 30, p 129-p 131) problem
values:
nrc
\|/n= sinctnX and , n=l, 2, 3, • • • •
a
nrcx
Therefore, f^x) can be expanded in terms of sin and, in fact,
Eq.(A-14) is the Fourier sine expansion of f^x) over (o,a), where the
a
an
2
a
J f1(x)-sinanx-dx
o
Thus,
a
a-sinhoinb o
2
J f1(x)*sinanx dx
written as:
n7ix nrcy
2 00 sin( y )-sinh(“^—)
~ ~
E =
a-sinhcub
J fi(x) nttx t
sin--dx =
2T0
mi
[l-(-l)°]
sinhanb
144
Substituting this result into Eq.(A-13), we find:
n:cx n7ty
sin( —)-sinh(—)
. 2T0
T(x,y) = — I f 1 -(-1n )n]
n= 1 nrcb
sinh(_a~)
a2T a2T
_ 2 + _ 2 =0
T(x,y) = X(x)Y(y)
sinhab
D=-C
coshab
sinhab
T(x,y) = A C sinax (sinhay - coshay) (A-16a)
coshab
A C
where E = Application of the boundary condition at x = a
coshab
yields:
E sinaa sinha(b-y) = 0
which results in
sinaa = 0
or
n;c
a”=T,
d
n=l, 2, 3, • • • •
146
T(x,y)= ]TEnsinanX sinnhou(b-y) (A-1 7)
n= 1
expansion of f2(x) over (0,a), and the expansion coefficients an's may
be written as :
a
2
a„ J f2(x)-sinoux-dx
a
o
nrcx njt(b-y)
sin( —“)*sinh[ r n7tt
T(x,y) = - I J f2(t)- sin(—)-dt
a n= 1 nrcb
sinh( )
and
n7ty nxx
sin( )-sinh( ■)
~b~
T(x,y)
nrca o
sinh(—)
nny nTt(a-x)
sin(-£-)-sinh[ g----- ] nxt
T(x,y) T)-dt
nrca
sinh(-^~)
nnt
So, g(n) = (a4t4+a3t3+a2t2+a!t+a0)-sin(~g~)-dt
D n7t 4 3 2 1 I b
= “ [-cos(-g~ t)(a4t +a3t +a2t +a4 +ao) I 0 +
b
nnt 3 2 nTtt
cos(-g~)(4a4t +3a3t +2a2t+a1)*d(-g-)]
o
148
n7i [(-l)n+I(a4b4+a3b3+a2b2+a1b1+a0)+ ao)] +
(n^)2[(4a4t3+3a3t2+2a2t+a J*sin(/^) | *
nrct nrct
(12a4t>6a3t+2a2)-sin(~j—)-d(~j^~)]
n7i
[(-l)n+1(a4b4+a3b3+a2b2+a1b1+a0)+ ao)] -
nn
n;it njct
(—y (12a4t +6a3t+2a2)-sin(~^~)]
nji
(r~)3 [(-ir'(12a4t2+6a3t+2a2)+2a2] -
nn
n7tt nrct
(—)5 (24a4)-sin(-y)-d(-g“)
me
= nK [(■l)n+1(a4b4+a3b3+a2b2+a1b1+a0)+ ao)] -
3 n 4-1
(~) [(-1) (12a4b +6a3b+2a2)+2a2] -
(^-)5-(24a4)-[ 1-(-I)*]
149
Appendix II
150
DATE: 28-2-92
1
5
CO
I
of 4 1
•M
s
PAGE:
Uj
H-
ci
jjj
♦-» *3 J
CO
01
« O* ..
(MJ -M1 to 5
I a >( ca>
28 - 2-1992
C >»■ M CM «-
CO -t\l „ J < < < <
■ "|£ §" TJ t- «-
x o l. a> t-
i sf - e“ -C _c
*-<*-*
C ■« C
to^>«^o
T3
•* t_ -o
4)
C.
"D
t-
O
01
XJ
t_
10 «- t_
S’ _C £ U 0) 4-» *■< 4-> 0.4-1 C
oi-»« 3«X£«
■- O •»
^~n ^
•* O •*
^
C ^ c ic |4->^--oro CfM4-<T-
° £ 6 • cv'*'l v aj 0)JD04->04Jt_(tit_tootoy)<o
■— x:
■a *-> — a* O ad 0 4 3 4-4 04 t_ ad
CL « «-> O" CT> c °* .5 .<51 - C='
% 1 2 5-55r55.'55-*55-55-55-55-
m u *“ V V
a o' « X = X := X : X := X : X: X
x CO = X: X:
ISSUE: 0 - CAL.C
< UI ■OJt ws w: W = ws WS W z ws
>- I
* g “ «
■M » —•
<U CD Si
c c cccccccccccccc
UJ
_II—
ac LU
I—
UJ
—I X *-X 4-» —- to ■*- (0 •*— <0 ■*“ (0 (0 CD •*"• (D •*— CD
X JQ O
I- o C.UC-UC.OL. u c- U c- u c- u
uu < o u. a. a to Q. </> Q. C/J Q. C/> £L </i CL X Q. (/) 0l C/i
« « JmiZ'
2
CQ <0 <0 03
J«N
II 5 o o o o £
S * *********** *% 5
%
Cl.
28-2-92
w
I
>2. DATE:
*to
CL
%
I
5
2
«
's.
Ol
I
s>
* *-
£ J
s- ..'i
PAGE: 2 of 4
H-C 01!_E 0>u
w
^lo,1
3 01 —
I— Ol
w Ol C
* C —
•»
Csf J) ^
z-'
« M + -Qa
23 01E o I *■H f\l•«
-8*r- +
w 01 H « c
II II II U >
_J " ?zl'n jj
ll Sm J2 c "5
•M
L. X l_ />
01
0)
•—
ai x 01 >- >(0 01 <MO —< I
c 0(1 c °* uI J
~p<U
> .°
c -' *c >L
-' 01 =:
-C3 V
•M *4—
^ E
4->CU Jc 2!
O
(J »*H- •*
U~ ^ 0^4)^ - X
O CTH- 0) O U
O <U 0> Q" h- j* O -* *4- w >4- w
0)
Lod pj Ood o£
C 'c N
•—C 'c« x
«- o
•#- ai
i- o >4-
S
(_
5tft ik'St a in a. ia •— E
ISSUE: 0 - CAL.C
^
■» ■»
oo
H- 0)
H-wH-wH-wH-w
4-» H- P 4- 4-»H- 4-» H- 8= -V
c£c£c£c£ II II II I— ■?0
ISISISIS
a tfl a. (A Q. a. CA
ca O’ X > 01
cn
xT Kl
£ ■W
**
I >o
Sr
a 1*b ) + pow (b/(n*P I , 2 )*< 12.*a 4*pow(b,2 )
co
xT fNJ
i. £
Sr
r\j Ql
w c
X
)
Q.
I
C
X
w
I
♦
%
+ a3*pow(b, 3 ) + a2*pow(b, 2 ) +
Q.
JO
I X jD
>w'
w
| Q.
i I
I R.
-C
H-
S)
X
^"N
Kl
xT
xT
w
I
a.
X
CL-
I
X
%
I Q-
%
t c4 = t c4 + single term;
rel_val ; fabs(single_term/t_c4);
n • n + 1;
}
ISSUE: 0 - CAL.C
ISSUE: PAGE: 4 of 4 44 DATE: 28-2-92
28-2-92
References:
151
Baghurst, C.P Foley. Bicentennial Electrical Engineering Conference
1988, Melbourne 11-15 April, 1988
152
12. "Amorphous Materials in Distribution Transformers." K Frewin,
GEC Distribution Transformers LTD, Kent, UK
153
19. "Susceptibility of Amorphous Magnetic Materials Over a Wide
Frequency Range." Y. Murata and K. Shirae IEEE Transactions on
Magnetics, Vol. Mag-20, No.5, Sept. 1984
21. "A comparision of the power loss and domain structure of wide
and narrow amorphous ribbons." M.G Bludell and K.J Overshott
IEEE Transactions on Magnetics, Mag-16, ppl224, 1980.
25. "The variation of power loss and domain wall spacing with
applied stress in wide amorphous ribbon materials" K.J Overshott
and M.G Blundell. IEEE Transactions on Magnetics, Vol-16, 1980.