Inductor with losses in the ATP

Orlando P. Hevia Gorostiaga 1483

CAUE - Comité Argentino de Usuarios del EMTP 3000 Santa Fe
GISEP-UTN Facultad Regional Santa Fe Argentina
Canadian/American EMTP User Group heviaop@ciudad.com.ar

1. Introduction

The SATURATION option of ATP produces flux-current output (in peak values) from
voltage-current input (measured RMS values). The measured current is assumed to be purely
reactive. That is, the inductor is assumed to have no losses, so the resistive component of
current can be ignored. This is reasonable as a first approximation.
Also, SATURATION may be troubled by a lack of homopolar (i.e., zero-sequence)
components of the measurements. SATURATION assumes that the inductor has been
measured as a single-phase device. But this generally is not true if inductors are part of a 3-
phase bank that is delta-connected. While the delta connection can be accounted for, the line
current --- what, in fact, is measured, typically --- will lack harmonics with numbers that are
odd multiples of 3 (i.e., 3, 9, 15, etc.). In some instances, the difference can be considerable.
Inductors connected in star (wye) with ungrounded neutral are comparably complicated. These
currents, too, will lack homopolar components.
LOSSY SATURATION is a new option of ATP that accounts for losses as well as the
connection of the three inductors. Losses can be represented by a non-linear resistance in
parallel with a non-linear inductor. This is not the same as an inductor with hysteresis, it
must be emphasized. Hysteresis is not being considered in the present analysis.

2. Consideration of inductor losses

If active (real) power is measured, it is possible to separate the RMS current into resistive and
inductive parts. This is using no-load losses. The resistive RMS current for each measurement
is obtained from the real power loss and the voltage (assumed to be sinusoidal) using
Ir − rms
k
= Pk / Vk (1)

Point number 1 is assumed to correspond to lowest excitation. For this small value, the
resistance is assumed to be linear, so peak current can be obtained by multiplying by 2 :
r
I1 = I− 1 2
r rms
(2)

Then, for the second point (or segment of the non-linear resistance), the instantaneous current
will be
 i2 (θ ) = (V2 sin(θ )) / R1 if θ < θ1
(3)
i2 (θ ) = I r + (V2 sin(θ ) − V1 ) / R2 if θ 1 ≤ θ ≤ π / 2
1

where
V2 = V rms2
2
R1 = V1 / I1
R2 = (V2 − V1 ) / ( I2 − I1 )

This can be generalized for k segments, with

θ1
θ ) = Vk sin(θ ) / R
ik ( 1 0

θ2
= Ir + (Vk sin(θ ) − V ) / R
θ1
(4)
1 1 2

L
π /2
= Ir + (Vk sin(θ ) − Vk − )/ Rk
k −1 1
θ k −1

with Rk = (Vk − Vk −1 ) / ( Ik − I k −1 )

The angles corresponding to a change of segment are determined by

θ = arcsin(V / V ), for j = 1,2,..., k -1
j j k

Except for the first section, all I and R are unknown. But they can be determined one pair
rj j

at a time, by recursion. The second segment can be determined from the known first segment;
then the third can be determined from the known first and second segments; etc.
Unfortunately, each step of the recursion involves iteration, as computation of parameters for
the second segment will illustrate.
Iteration for the current that ends the second segment, I , begins with its estimation. This r2

then is used to compute the current at many intermediate points as voltage varies sinusoidally
from zero to peak V . The square of these computed currents then is integrated by the
r2

trapezoidal rule over a quarter of a period (sufficient due to symmetry) to yield the RMS
current. Of course, this computed RMS current should equal the measured RMS current, but
it does not because I was only an estimate. So, I is modified in proportion to the
r2 r2

discrepancy. Etc. (the iteration is repeated until the difference between the measured and
calculated RMS currents becomes sufficiently small). Convergence is not a problem. A
difference of 10 in I typically can be obtained after a few iterations.
-12
r2

3. Consideration of the saturation

Assuming the supply voltage is sinusoidal, it can be shown that harmonic components of the
resistive current are orthogonal to harmonic components of the inductive current. Therefore,

Il −rms = 2
It − rms + I 2−
r rms
(5)

as demonstrated in the Appendix (Sect. 8).

Analogous to the preceding resistive computation, there will be a reactive computation.
Voltage is replaced by the flux, and resistances are replaced by inductances
Lk = ( Φ k − Φ k −1 ) / ( I k − Ik −1 ) (6)

A comparable recursive and iterative solution then determines the inductive currents I k . Along
with the associated flux values Φ k , these characterize the non-linear inductor that represents
the saturation.
If either a single inductor is being considered, or if the measurements are for a group of
inductors connected in star, the task is ended. Delta-connected inductors are more
complicated, however (see next section).

4. Delta-connected inductors
For inductors connected in delta, measured current lacks harmonics that are multiples 3 ---
the homopolar (zero-sequence) components. These currents circulate solely within the closed
triangle, if the three non-linear branches are exactly equal. The preceding general procedure
remains applicable, but analysis must account for the missing harmonics. The current
calculated according to (3) and (4) is to be decomposed into its Fourier components in order
that the homopolar harmonics can be omitted. Then the current can be reconstituted.
A more efficient, and possibly more exact, method is used by the author. The current for each
angle, and for the same angle plus 23 π , is calculated. By Kirchhoff’s current law, the sum of
these two currents must equal the current measured. Of course, line currents are assumed to be
3 times device currents (balance is assumed).

Both resistive and inductive currents are processed this way. The end result is a representation
having losses and saturation connected in delta. Punched output will consist of 3 Type-98
pseudo-nonlinear inductors and 3 Type-99 pseudo-nonlinear resistors.

5. Format for the data

Input data for LOSSY SATURATION generally is compatible with that of old
SATURATION, although extensions are required to accommodate additional data. These
begin with the miscellaneous data card, which now includes parameters that indicate the
connection of the windings (if three rather than one), a column-80 request for branch cards,
and bus names for those branch cards (produced by a $PUNCH request).
The most important new parameter is integer KCON, which indicates the winding connection.
A value 0 or -1 indicates a single-phase inductor. In this case, input voltage, current, and
power loss are used without modification.
KCON = 1 implies a balanced, 3-phase system of inductors connected in star (wye). In this
case, input voltage will be divided by 3 , and the power loss will be divided by 3. This is
because measurements are assumed to be line voltage (i.e., line-to-line voltage) and total, 3-
phase power.
KCON = 3 implies a balanced, 3-phase system of inductors connected in delta. In this case,
input current will be divided by 3 and power will be divided by 3. This is because
measurements are assumed to be line voltage, line current, and 3-phase power. Currents in
both elements will be adjusted after eliminating the harmonics that are multiples of 3.
Following the miscellaneous data card will be the table of current, voltage, and power loss.
The format of SATURATION has been retained, and extended by the addition of a new, third
column for the power loss.
If all powers are zero, the program will produce a non-linear inductor assuming that the
current is purely inductive, just as SATURATION would.
The routine verifies that the slope of the 2nd or later segment of each characteristic is less than
the slope of the preceding segment, just as SATURATION does. Yet, because any violation is
not necessarily an error, only a warning message is associated with the check.
The routine also verifies that each pair of voltage and power loss does not correspond to an
active current that is greater than the measured current (clearly impossible).
Finally, the routine verifies that the v-i characteristic of the non-linear resistance is monotone
increasing. If for some point it is not, that point will be discarded, and a warning message will
be issued.
In other respects, the new option operates as SATURATION has for decades.

6. Example
Consider a 3-phase transformer with the measured windings connected in delta. To obtain the
same results in all the cases, voltage and current have been scaled according to the type of
connection.
C
C
LOSSY SATURATION
C
C SINGLE PHASE WITHOUT LOSSES
C
C FREQ--VBASE---SBASE---IPUNCH--KTHIRD--KCON----IOUT----BUS1--BUS2--BUS3--BUS4--
50. 0.001 1.0E-6 -1 1DOMAR DOMBR DOMAL DOMBL
C
C IRMS----------VRMS------------P--------------- I P
1.7378 50414.80
2.5346 53693.57
4.2259 57504.09
8.3427 61292.08
45.1892 64570.86
93.0631 67918.08
9999.
C
C SINGLE PHASE WITH LOSSES
C FREQ--VBASE---SBASE---IPUNCH--KTHIRD--KCON----IOUT----BUS1--BUS2--BUS3--BUS4--
50. 0.001 1.0E-6 1 1DOMAR DOMBR DOMAL DOMBL
C
C IRMS----------VRMS------------P--------------- I P
1.7378 50414.80 16880.0
2.5346 53693.57 20213.33
4.2259 57504.09 25253.33
8.3427 61292.08 31653.33
45.1892 64570.86 38000.00
93.0631 67918.08 45120.0
9999.
C
C STAR CONNECTION WITHOUT LOSSES
C
C FREQ--VBASE---SBASE---IPUNCH--KTHIRD--KCON----IOUT----BUS1--BUS2--BUS3--BUS4--
50. 0.001 1.0E-6 1 1DOMAR DOMBR DOMAL DOMBL
C
C IRMS----------VRMS------------P--------------- I P
1.7378 50414.80
2.5346 53693.57
4.2259 57504.09
8.3427 61292.08
45.1892 64570.86
93.0631 67918.08
9999.
C
C STAR CONNECTION WITH LOSSES
C FREQ--VBASE---SBASE---IPUNCH--KTHIRD--KCON----IOUT----BUS1--BUS2--BUS3--BUS4--
50. 0.001 1.0E-6 1 1DOMAR DOMBR DOMAL DOMBL
C
C IRMS----------VRMS------------P--------------- I P
1.7378 50414.80 50640.0
2.5346 53693.57 60640.0
4.2259 57504.09 75760.0
8.3427 61292.08 94960.0
45.1892 64570.86 114000.0
93.0631 67918.08 135360.0
9999.
C
C DELTA CONNECTION WITHOUT LOSSES
C FREQ--VBASE---SBASE---IPUNCH--KTHIRD--KCON----IOUT----BUS1--BUS2--BUS3--BUS4--
50. 0.001 1.0E-6 3 1DOMAR DOMBR DOMAL DOMBL
C
C IRMS----------VRMS------------P--------------- I P
3.01 29107.00
4.39 31000.00
7.51 33200.00
14.45 35387.0
26.09 37280.00
53.73 39213.00
9999.
C
C DELTA CONNECTION WITH LOSSES
C FREQ--VBASE---SBASE---IPUNCH--KTHIRD--KCON----IOUT----BUS1--BUS2--BUS3--BUS4--
50. 0.001 1.0E-6 3 1DOMAR DOMBR DOMAL DOMBL
C
C IRMS----------VRMS------------P--------------- I P
3.01 29107.00 50640.0
4.39 31000.00 60640.0
7.51 33200.00 75760.0
14.45 35387.0 94960.0
26.09 37280.00 114000.0
53.73 39213.00 135360.0
9999.
BLANK
BLANK

Next, consider ATP output corresponding to the preceding data:

Alternative Transients Program (ATP), GNU Linux or DOS. All rights reserved by Can/Am user group of Portland, Oregon, USA.
Date (dd-mth-yy) and time of day (hh.mm.ss) = 13-Apr-01 11:47:21 Name of disk plot file is elostt.pl4
Consult the 860-page ATP Rule Book of the Can/Am EMTP User Group in Portland, Oregon, USA. Source code date is 03 January
2001.
Total size of LABCOM tables = 4929880 INTEGER words. VARDIM List Sizes follow : 6002 10000 10000 340 20000 5000
13140 120000 1450 1900 360 900 36400 255 64800 384 45 254 400000 100000 3000 12000 40000 120 30000 10000 600
126000
--------------------------------------------------+---------------------------------------------------------------------------
-----
Descriptive interpretation of input data cards. | Input data card images are shown below, all 80 columns, character by
character
0 1 2 3 4 5 6 7
8

012345678901234567890123456789012345678901234567890123456789012345678901234567890
--------------------------------------------------+---------------------------------------------------------------------------
-----
Comment card. NUMDCD = 1. |C data:elostt.dat
Comment card. NUMDCD = 2. |C
Comment card. NUMDCD = 3. |C
Orlando Hevia fits params for lossy saturation. |LOSSY SATURATION
Comment card. NUMDCD = 5. |C
Comment card. NUMDCD = 6. |C SINGLE PHASE WITHOUT LOSSES
Comment card. NUMDCD = 7. |C
Comment card. NUMDCD = 8. |C FREQ--VBASE---SBASE---IPUNCH--KTHIRD--KCON----IOUT----BUS1--BUS2--BUS3--
BUS4--
Misc. const. 5.000E+01 1.000E-03 1.000E-06-1 0 1| 50. 0.001 1.0E-6 -1 1DOMAR DOMBR DOMAL
DOMBL
Comment card. NUMDCD = 10. |C
Comment card. NUMDCD = 11. |C IRMS----------VRMS------------P--------------- I P
(I,V,P) point. 1.73780E+00 5.04148E+04 0.00000E+00|1.7378 50414.80
(I,V,P) point. 2.53460E+00 5.36936E+04 0.00000E+00|2.5346 53693.57
(I,V,P) point. 4.22590E+00 5.75041E+04 0.00000E+00|4.2259 57504.09
(I,V,P) point. 8.34270E+00 6.12921E+04 0.00000E+00|8.3427 61292.08
(I,V,P) point. 4.51892E+01 6.45709E+04 0.00000E+00|45.1892 64570.86
(I,V,P) point. 9.30631E+01 6.79181E+04 0.00000E+00|93.0631 67918.08
Special termination-of-points card. | 9999.

SINGLE PHASE INDUCTOR

Derived saturation curve gives peak current as a function of flux :
Row Current [amp] Flux [volt-sec]
1 0.0000000000 0.0000000000
2 2.4576203287 226.9463350790
3 5.3560613238 241.7059857186
4 9.1661340450 258.8593523638
5 19.8437304842 275.9112983760
6 130.8582464266 290.6709940315
7 231.6425155570 305.7387779303
9999

Next, check the derived curve by independent reverse computation. Assuming sinusoidal voltage (flux) at the level of each
point,
rms current is found numerically. This curve should be equal to the original I-V points inputted.
Row Active current P.U. Reactive current P.U. Total current P.U. Voltage P.U. Losses P.U.
2 0.00000000 1.73780000 1.73780000 50414.80000000 0.00000000
3 0.00000000 2.53460000 2.53460000 53693.57000000 0.00000000
4 0.00000000 4.22590000 4.22590000 57504.09000000 0.00000000
5 0.00000000 8.34270000 8.34270000 61292.08000000 0.00000000
6 0.00000000 45.18920000 45.18920000 64570.86000000 0.00000000
7 0.00000000 93.06310000 93.06310000 67918.08000000 0.00000000

Comment card. NUMDCD = 19. |C

Comment card. NUMDCD = 20. |C SINGLE PHASE WITH LOSSES
Comment card. NUMDCD = 21. |C FREQ--VBASE---SBASE---IPUNCH--KTHIRD--KCON----IOUT----BUS1--BUS2--BUS3--
BUS4--
Misc. const. 5.000E+01 1.000E-03 1.000E-06 1 0 1| 50. 0.001 1.0E-6 1 1DOMAR DOMBR DOMAL
DOMBL
Comment card. NUMDCD = 23. |C
Comment card. NUMDCD = 24. |C IRMS----------VRMS------------P--------------- I P
(I,V,P) point. 1.73780E+00 5.04148E+04 1.68800E+04|1.7378 50414.80 16880.0
(I,V,P) point. 2.53460E+00 5.36936E+04 2.02133E+04|2.5346 53693.57 20213.33
(I,V,P) point. 4.22590E+00 5.75041E+04 2.52533E+04|4.2259 57504.09 25253.33
(I,V,P) point. 8.34270E+00 6.12921E+04 3.16533E+04|8.3427 61292.08 31653.33
(I,V,P) point. 4.51892E+01 6.45709E+04 3.80000E+04|45.1892 64570.86 38000.00
(I,V,P) point. 9.30631E+01 6.79181E+04 4.51200E+04|93.0631 67918.08 45120.0
Special termination-of-points card. | 9999.

STAR CONNECTION
Voltage and flux values will be scaled dividing by sqrt(3)
Derived nonlinear curve gives peak current as a function of voltage :
Row Current [amp] Voltage [V]
1 0.2733812747 41163.5118281551
2 0.3439894245 43840.6163228035
3 0.4159304519 46951.8928743602
4 0.5066202731 50044.7737579487
5 0.5770362029 52721.8864175629
6 0.6487978951 55454.8801031758
9999

Derived saturation curve gives peak current as a function of flux :

Row Current [amp] Flux [volt-sec]
1 0.0000000000 0.0000000000
2 2.4423677771 131.0275276495
3 5.3456214954 139.5490159194
4 9.1567134913 149.4525167695
5 19.8376652602 159.2974623899
6 130.8579404716 167.8189766497
7 231.6413633881 176.5183657398
 9999

Next, check the derived curve by independent reverse computation. Assuming sinusoidal voltage (flux) at the level of each
point,
rms current is found numerically. This curve should be equal to the original I-V points inputted.
Row Active current P.U. Reactive current P.U. Total current P.U. Voltage P.U. Losses P.U.
2 0.19330975 1.72701482 1.73780000 50414.80000000 16880.00000000
3 0.21734765 2.52526378 2.53460000 53693.57000000 20213.33000000
4 0.25354748 4.21828691 4.22590000 57504.09000000 25253.33000000
5 0.29816346 8.33737020 8.34270000 61292.08000000 31653.33000000
6 0.33977107 45.18792264 45.18920000 64570.86000000 38000.00000000
7 0.38355095 93.06230961 93.06310000 67918.08000000 45120.00000000

Comment card. NUMDCD = 32. |C

Comment card. NUMDCD = 33. |C STAR CONNECTION WITHOUT LOSSES
Comment card. NUMDCD = 34. |C
Comment card. NUMDCD = 35. |C FREQ--VBASE---SBASE---IPUNCH--KTHIRD--KCON----IOUT----BUS1--BUS2--BUS3--
BUS4--
Misc. const. 5.000E+01 1.000E-03 1.000E-06 1 0 1| 50. 0.001 1.0E-6 1 1DOMAR DOMBR DOMAL
DOMBL
Comment card. NUMDCD = 37. |C
Comment card. NUMDCD = 38. |C IRMS----------VRMS------------P--------------- I P
(I,V,P) point. 1.73780E+00 5.04148E+04 0.00000E+00|1.7378 50414.80
(I,V,P) point. 2.53460E+00 5.36936E+04 0.00000E+00|2.5346 53693.57
(I,V,P) point. 4.22590E+00 5.75041E+04 0.00000E+00|4.2259 57504.09
(I,V,P) point. 8.34270E+00 6.12921E+04 0.00000E+00|8.3427 61292.08
(I,V,P) point. 4.51892E+01 6.45709E+04 0.00000E+00|45.1892 64570.86
(I,V,P) point. 9.30631E+01 6.79181E+04 0.00000E+00|93.0631 67918.08
Special termination-of-points card. | 9999.

STAR CONNECTION
Voltage and flux values will be scaled dividing by sqrt(3)
Derived saturation curve gives peak current as a function of flux :
Row Current [amp] Flux [volt-sec]
1 0.0000000000 0.0000000000
2 2.4576203287 131.0275276495
3 5.3560613238 139.5490159194
4 9.1661340450 149.4525167695
5 19.8437304842 159.2974623899
6 130.8582464266 167.8189766497
7 231.6425155570 176.5183657398
9999

Next, check the derived curve by independent reverse computation. Assuming sinusoidal voltage (flux) at the level of each
point,
rms current is found numerically. This curve should be equal to the original I-V points inputted.
Row Active current P.U. Reactive current P.U. Total current P.U. Voltage P.U. Losses P.U.
2 0.00000000 1.73780000 1.73780000 50414.80000000 0.00000000
3 0.00000000 2.53460000 2.53460000 53693.57000000 0.00000000
4 0.00000000 4.22590000 4.22590000 57504.09000000 0.00000000
5 0.00000000 8.34270000 8.34270000 61292.08000000 0.00000000
6 0.00000000 45.18920000 45.18920000 64570.86000000 0.00000000
7 0.00000000 93.06310000 93.06310000 67918.08000000 0.00000000

Comment card. NUMDCD = 46. |C

Comment card. NUMDCD = 47. |C STAR CONNECTION WITH LOSSES
Comment card. NUMDCD = 48. |C FREQ--VBASE---SBASE---IPUNCH--KTHIRD--KCON----IOUT----BUS1--BUS2--BUS3--
BUS4--
Misc. const. 5.000E+01 1.000E-03 1.000E-06 1 0 1| 50. 0.001 1.0E-6 1 1DOMAR DOMBR DOMAL
DOMBL
Comment card. NUMDCD = 50. |C
Comment card. NUMDCD = 51. |C IRMS----------VRMS------------P--------------- I P
(I,V,P) point. 1.73780E+00 5.04148E+04 5.06400E+04|1.7378 50414.80 50640.0
(I,V,P) point. 2.53460E+00 5.36936E+04 6.06400E+04|2.5346 53693.57 60640.0
(I,V,P) point. 4.22590E+00 5.75041E+04 7.57600E+04|4.2259 57504.09 75760.0
(I,V,P) point. 8.34270E+00 6.12921E+04 9.49600E+04|8.3427 61292.08 94960.0
(I,V,P) point. 4.51892E+01 6.45709E+04 1.14000E+05|45.1892 64570.86 114000.0
(I,V,P) point. 9.30631E+01 6.79181E+04 1.35360E+05|93.0631 67918.08 135360.0
Special termination-of-points card. | 9999.

STAR CONNECTION
Voltage and flux values will be scaled dividing by sqrt(3)
Derived nonlinear curve gives peak current as a function of voltage :
Row Current [amp] Voltage [V]
1 0.8201438240 41163.5118281551
2 1.0319687497 43840.6163228035
3 1.2477914306 46951.8928743602
4 1.5198609701 50044.7737579487
5 1.7311083147 52721.8864175629
6 1.9463937089 55454.8801031758
9999

Derived saturation curve gives peak current as a function of flux :

Row Current [amp] Flux [volt-sec]
1 0.0000000000 0.0000000000
2 2.3167351571 131.0275276495
3 5.2617554738 139.5490159194
4 9.0809132976 149.4525167695
5 19.7891409527 159.2974623899
6 130.8554921364 167.8189766497
7 231.6321462214 176.5183657398
9999

Next, check the derived curve by independent reverse computation. Assuming sinusoidal voltage (flux) at the level of each
point,
rms current is found numerically. This curve should be equal to the original I-V points inputted.
Row Active current P.U. Reactive current P.U. Total current P.U. Voltage P.U. Losses P.U.
2 0.57992926 1.63817914 1.73780000 50414.80000000 50640.00000000
3 0.65204307 2.44929316 2.53460000 53693.57000000 60640.00000000
4 0.76064253 4.15688029 4.22590000 57504.09000000 75760.00000000
 5 0.89449047 8.29460850 8.34270000 61292.08000000 94960.00000000
6 1.01931321 45.17770243 45.18920000 64570.86000000 114000.00000000
7 1.15065285 93.05598626 93.06310000 67918.08000000 135360.00000000

Comment card. NUMDCD = 59. |C

Comment card. NUMDCD = 60. |C DELTA CONNECTION WITHOUT LOSSES
Comment card. NUMDCD = 61. |C FREQ--VBASE---SBASE---IPUNCH--KTHIRD--KCON----IOUT----BUS1--BUS2--BUS3--
BUS4--
Misc. const. 5.000E+01 1.000E-03 1.000E-06 3 0 1| 50. 0.001 1.0E-6 3 1DOMAR DOMBR DOMAL
DOMBL
Comment card. NUMDCD = 63. |C
Comment card. NUMDCD = 64. |C IRMS----------VRMS------------P--------------- I P
(I,V,P) point. 3.01000E+00 2.91070E+04 0.00000E+00|3.01 29107.00
(I,V,P) point. 4.39000E+00 3.10000E+04 0.00000E+00|4.39 31000.00
(I,V,P) point. 7.51000E+00 3.32000E+04 0.00000E+00|7.51 33200.00
(I,V,P) point. 1.44500E+01 3.53870E+04 0.00000E+00|14.45 35387.0
(I,V,P) point. 2.60900E+01 3.72800E+04 0.00000E+00|26.09 37280.00
(I,V,P) point. 5.37300E+01 3.92130E+04 0.00000E+00|53.73 39213.00
Special termination-of-points card. | 9999.

DELTA CONNECTION: There are no triplen harmonics in measured current.

Current values will be scaled dividing by sqrt(3)
Derived saturation curve gives peak current as a function of flux :
Row Current [amp] Flux [volt-sec]
1 0.0000000000 0.0000000000
2 2.4576547086 131.0275350719
3 5.6003513279 139.5490290044
4 10.5077587126 149.4525084821
5 22.6709043675 159.2974673993
6 42.8729564831 167.8189613317
7 94.5085603564 176.5205185273
9999

Next, check the derived curve by independent reverse computation. Assuming sinusoidal voltage (flux) at the level of each
point,
rms current is found numerically. This curve should be equal to the original I-V points inputted.
Row Active current P.U. Reactive current P.U. Total current P.U. Voltage P.U. Losses P.U.
2 0.00000000 3.01000000 3.01000000 29107.00000000 0.00000000
3 0.00000000 4.39000000 4.39000000 31000.00000000 0.00000000
4 0.00000000 7.51000000 7.51000000 33200.00000000 0.00000000
5 0.00000000 14.45000000 14.45000000 35387.00000000 0.00000000
6 0.00000000 26.09000000 26.09000000 37280.00000000 0.00000000
7 0.00000000 53.73000000 53.73000000 39213.00000000 0.00000000

Comment card. NUMDCD = 72. |C

Comment card. NUMDCD = 73. |C DELTA CONNECTION WITH LOSSES
Comment card. NUMDCD = 74. |C FREQ--VBASE---SBASE---IPUNCH--KTHIRD--KCON----IOUT----BUS1--BUS2--BUS3--
BUS4--
Misc. const. 5.000E+01 1.000E-03 1.000E-06 3 0 1| 50. 0.001 1.0E-6 3 1DOMAR DOMBR DOMAL
DOMBL
Comment card. NUMDCD = 76. |C
Comment card. NUMDCD = 77. |C IRMS----------VRMS------------P--------------- I P
(I,V,P) point. 3.01000E+00 2.91070E+04 5.06400E+04|3.01 29107.00 50640.0
(I,V,P) point. 4.39000E+00 3.10000E+04 6.06400E+04|4.39 31000.00 60640.0
(I,V,P) point. 7.51000E+00 3.32000E+04 7.57600E+04|7.51 33200.00 75760.0
(I,V,P) point. 1.44500E+01 3.53870E+04 9.49600E+04|14.45 35387.0 94960.0
(I,V,P) point. 2.60900E+01 3.72800E+04 1.14000E+05|26.09 37280.00 114000.0
(I,V,P) point. 5.37300E+01 3.92130E+04 1.35360E+05|53.73 39213.00 135360.0
Special termination-of-points card. | 9999.

DELTA CONNECTION: There are no triplen harmonics in measured current.

Current values will be scaled dividing by sqrt(3)
Derived nonlinear curve gives peak current as a function of voltage :
Row Current [amp] Voltage [V]
1 0.8201437775 41163.5141599937
2 1.0361612752 43840.6204335659
3 1.2619241541 46951.8902707868
4 1.5524214089 50044.7753316967
5 1.7758197311 52721.8816052690
6 1.9995225965 55455.5564213362
9999

Derived saturation curve gives peak current as a function of flux :

Row Current [amp] Flux [volt-sec]
1 0.0000000000 0.0000000000
2 2.3167716441 131.0275350719
3 5.5231355110 139.5490290044
4 10.4634895815 149.4525084821
5 22.6722126236 159.2974673993
6 42.8985522573 167.8189613317
7 94.5609677083 176.5205185273
9999

Next, check the derived curve by independent reverse computation. Assuming sinusoidal voltage (flux) at the level of each
point,
rms current is found numerically. This curve should be equal to the original I-V points inputted.
Row Active current P.U. Reactive current P.U. Total current P.U. Voltage P.U. Losses P.U.
2 1.00446689 2.83745419 3.01000000 29107.00000000 50640.00000000
3 1.12937162 4.24224230 4.39000000 31000.00000000 60640.00000000
4 1.31747158 7.39353560 7.51000000 33200.00000000 75760.00000000
5 1.54930290 14.36670319 14.45000000 35387.00000000 94960.00000000
6 1.76550243 26.03019595 26.09000000 37280.00000000 114000.00000000
7 1.99296490 53.69302553 53.73000000 39213.00000000 135360.00000000

Blank card terminating "SATURATION" data cases. |BLANK

As an application of the preceding results, consider ATP simulation of an unloaded, 3-phase
transformer. Only excitation current is drawn. Both wye (star) and delta (triangle)
connections will be considered, with the neutral of the wye grounded.
Consider the grounded-wye alternative first. The following shows line current for phase “a”
on the left, and residual (neutral) current on the right. Both resistive (R) and reactive (L)
components are shown, with the former magnified greatly (factors 40.2 and 87.8, respectively)
for easy visibility. Note how line current on the left is dominated by the third harmonic
whereas neutral current on the right seems nearly sinusoidal:

A Fourier series of the line current shows presence of the triplen harmonics. All odd
harmonics (1, 3, 5, 7, 9, etc.) are present:

Alternatively, consider the delta connection. On the left below are shown resistive and reactive
currents within a branch of the delta. On the right are shown the resistive and reactive currents
for one line --- waveforms that are quite different due to lack of the triplen harmonics
(numbers 3, 9, etc.):
A Fourier series of the line current confirms absence of the triplen harmonics (3, 9, etc.), which
are confined to the delta:

For both grounded-wye and delta cases, resistive currents are very small with respect to
inductive currents. As an approximation, the resistive part might be represented by linear
resistance, or ignored completely.

7. Conclusions and Warnings

LOSSY SATURATION calculates the saturation curve taking into account losses as well as
the connection of the three-phase system of inductors (transformer windings) that was
measured. Results in the form of branch cards can be employed directly by ATP for the
simulation of transients.
Yet, approximations should be understood. Error comes from many sources: different
saturation of the columns of a three-phase system, irregular flux distribution in the laminations,
distortion of the flux due to saturation, etc.
For a transformer having a modern core material, magnetizing current might be less than
capacitive current (the latter due to stray capacitance of the windings).
Final warnings: Data for LOSSY SATURATION is not readily available. If current is to be
measured experimentally, it is very important that instruments indicate true RMS. This is
because current is highly distorted.

8. Appendix
The current in the non-linear resistor, as well as the current in the non-linear inductor, will
contain odd harmonics only. This is because of the assumed odd symmetry of the
characteristics. As a consequence, these currents can be expressed as Fourier series that lack
even harmonics:
ir ( L
θ ) = a1 sin θ + a3 sin 3θ + + an sin nθ

and
il ( θ ) = b1 cos θ + b3 cos 3θ + +b cos nθ L n

where n=1,3,...
Adding these two, and simplifying, the total current is:

it ( θ) = 2
a1 + b12 sin(θ + ϕ 1 ) + 2
a3 L
+ b32 sin(3θ + ϕ 3 )+ + 2
an + b2 sin( nθ + ϕ
n n
)

where ϕ n = arctan( bn / an )

From these three expressions for θ ), il (θ ) and it (θ ), the associated RMS values are
ir (

calculated to be:

Ir − rms = a1
2
L
+ a32 + + an2

Il −rms = b1
2
+ b32 + +b2L n

and

It −rms = a1
2
L
+ b12 + a32 + b32 + + a 2 + b2
n n

respectively. From these last three equations, it is seen that

2
It −rms = 2
Ir − rms + I 2−
l rms
9. References
[1] Washington L. A. Neves, Hermann W. Dommel, "On modelling iron core nonlinearities,"
IEEE Transactions on Power Systems, Vol. 8. No. 2, May 1993.
[2] Washington L. A. Neves, Hermann W. Dommel, "Saturation curves of delta-connected
transformers from measurements", IEEE Transactions on Power Delivery, Vol. 10, No. 3, July
1995.
[3] ATP Rule Book, 830 pages, can be obtained from any ATP user group.