A PROTECTION BOHEME FOR ALTERNATOR FIELDS by James …
Transcript of A PROTECTION BOHEME FOR ALTERNATOR FIELDS by James …
A PROTECTION BOHEME FOR ALTERNATOR FIELDS
by
James Torey Lancaster
Thosia submitted to the Graduate Faculty of the
Virginia Polytechnic Institute
in candidacy for the degree of
MASTER OF SOIENOE
in
Electrical Enginoering
May, 1965
Blacksburg, Virginia
TABLE OF CONTENTS
CHAPTER
I.
II.
III.
IV.
v. VI.
VII.
VIII.
INTRODUCTION • • • • • •
THE REVIEW OF LITERATURE
THE INVESTIGATION ••••
• • • • • • • • • • • • • •
• • • • • • • • • • • • • •
• • • • • • • • • • • • • •
Object of the Investigation.
Method of Procedure •••••
Discussion of the circuit.
• • • • • • • • • • •
• • • • • • • • • • • •
. . . . • • • • • • •
Discussion of statistical analysis of data . . . Results. • • • • • • . . . . . . . . . . . . . . .
DISCUSSION OF RESULTS. • • • • • • • • • • • • • • •
CONCLUSIONS. • • • • • • • • • • • • • • • • • • • •
SUMMARY ••• • • • • • • • • • • • • • • • • • • • •
ACKNOWLEDGE:viENTS • • • • • • • • • • • • • • • • • •
BIBLIOGRAPHY • • • • • • • • • • • • • • • • • • ·• •
VITA • . • • • • • • • • • • • • • • • • • • • • • • •
APPENDIX • • • • • • • • • • • • • • • • • • • • • •
PAGE
5 6
7
7
7
7
1:;
24
28
29
:;o :;1 :;2 :;4
:;5
LIST OF FIGURES
FIGURE
l. SC I-ID;I!JATIO DIAGRAM OF RELAY CIRCUIT • • • • • . . . . . . . 2. BLOCK DIAGRAlt. OF TIMING TEST CIRCUIT ••• . . . . . . . .
PAGE
8
9
:,. SCHEMATIC DIAGRAM OF TRIGGER CIRCUIT •
4. TYPICAL DISPLAY OF OPERATING TIME •••
• • • • • • •••• 11
. . . . . • • • • • 12
5. REGRESSION LINE FOR TEN PER-OE.l~T UNBALANCED CONDITION. • • 18
6. REGRESSION LINE FOR FI FTEEl'i PER-CENT UNBALANCED CONDITION. 19
7. REGRESSION LI1"E FOR T\'lENTY PER-CENT UNBALANCED CONDITION • 20
8. PREDICTION LINES FOR TEN PER-CENT UNBALANCED CONDITION • • 25 9. PREDICTION LINES FOR FIFTEEN PER-CENT UNBALANCED CONDITION · 26
10. PREDICTION LINES FOR TWENTY ?ER-CENT UNBALANCED CONDITION. 27
LIST OF TABLES
TABLE PAGE
1. Observed Times of Operation For Ten Per-Cent Unbalanced Condition In Milliseconds. • • • • • • • • • • • • • • • 14
2.
:,.
4.
5.
Observed Times of Operation For Fifteen Per-Cent Unbalanced Condition In Milliseconds. • • • • • • • • • •
Observed Times of Operation For Twenty Per-Cent Unbalanced Condition In Milliseconds . • • • • • • • • • • • • • • •
Data For Calculation of Regression Line For Ten Per-Cent Unbalanced Condition. . • • • • • • • • • . • • . • • • . Data For Calculation of Regression Line For Fifteen Per-Cent Unbalanced Condition .• • • • • • • • • • • • • . • •
6. Data For Calculation of Regression Line For Twenty Per-
15
16
21
22
Cent Unbalanced Condition. • • • • • • • • • • • • • • • 2;
5
I INTRODUCTION
An unbalanced condition on a three-phaae power system can_·
generate a voltage ncroaa tne field of an alternator which 11 pro-
portional to the degree of unbalance and of twice the system fre-
quency. Thia voltage is due to negative-phaae-aequence currenta
in the stator winding of the machine during the unbalanced condition.
A voltage induced in this manner may reach a magnitude of such order
aa to exceed the insulation level of the field winding and cause
tho inaulation to break down thua damaging.the machine.
Exiating protective lilYstema which arc deaigned to protect
an alternator against auch a condition arc often too alow in their
operation to be of much value. Speed ia essential in auch an
operation, .both for the protection of the machine and for the stability
of the entire power aystem.
6
II THE REVIEW OF LITERATURE
While the literature pertaining to the application, construc-
tion, and operation of relays is quite extensive, the literature
pertaining to this particular aspect of protective relaying is
practically non-existent.
Several sources listed methods of determining negative-
sequence impedance, and Henderson(;) outlines a negative-phase-
sequence scheme of protection for an alternator. Several other
sources listed protection against loss of generator field, but this
is only remotely related to this discussion.
Mason (7) mentions protection against rotor overheating
because of unbalanced stator currents, but makes no mention of
protection against voltages generated in the field due to an un-
balance. This also applies to Westinghouse (2) (8), Kimbark (6), and Skrotzki (lo).
7
III THE INVESTIGATION
Object of~ Investigc.tion
This study was set up with the following objectives in minds
(1) To design and construct a very fast, dependable, solid state
relay capable of protecting a synchronous machine from the voltage
induced in the field winding due to negative-phase-sequence currents
caused by unbalanced conditions; (2) To find a method of detecting
such unbalanced conditions as swiftly as possible; and(;) To test
this protective scheme in the laboratory.
Method of Procedure
Discussion of the circuit. Several devices were designed,
constructed, and tested before one was found which seemed to give
consistently good results. The circuit of the device which was
used ia shown in figure 1. The test circuit that was used to obtain
data is shown in block-diagram form in figure 2.
Since a synchro generator is very similar to a synchronous
machine, it was decided that such a device could be used as a de-
tector. The synchro generator was operated as a synchronous motor
and the relay was connected to the field by means of a current
transformer. The current transformer was necessary because the
power sypply used to excite the field presented a very low impedance
to the 120 cycles per second frequency of the generated signal
voltage, which tended to short-circuit the signal voltage.
A trigger circuit-was designed and built to provide positive
AJI capacitor ratings are in microfarads.
INPUT
LEGEND: C1=5 C2.C3=100 C4.C5=3 C6.C 7=0.001 C8=1
+22v
R1,R 2.R7 = 10 k R3,R10,R12 = 3,3 k R4= 6.8k R5= 22k R6= 5 k
R9:330 0=1N673 R13=1k R8= 1 k pot R11= 2 k pot
Figure 1
SCHEMATIC DIAGRAM OF RELAY CIRCUIT
O>
70 v 3(1) N:. Source FIELD DC Source
FAULT
TRIGGER ELAY LOAD OSCILLOSCOPE
\()
Figure 2
BLOCK DIAGRAM OF TIMING TEST CIRCUIT
10
triggering of the oscilloscope sweep. The oscilloscope was trigger-
ed by the unbalance, and the vertical input to the oscilloscope was
the voltage across the relay load. Thus the difference in time be-
tween the appearance of the unbalance and the relay pick-up was
displayed on the oscilloscope. A diagram of the trigger circuit
is shown in figure;. A picture of a typical display is shown in
figure 4. A Tektronix type 564 Storage Oscilloocope was used.
Existing and readily available equipment and supplies were
used in this study. It should be pointed out that the synchro-
generator used as a detector was not designed for this type of ser-
vice, and that much more consiatent results could have been obtained
along with faster operating times if a machine had been available
whicµ was more auited to the taak. Also, such a machine could have
resulted in the elimination of the amplification stages of the relay
circuit. Both the relay and trigger circuits had to be designed
according to the following criteriaa
l. The circuit must reapond to a fault which was either
positive or negative going at the instant of starting.
2. The circuit must be extremely fast in its operation,
preferably on the order of micro-seconds.
;. The circuit must be extremely stable throughout its
operating temperature range.
4. The operating temperature range must be broad.
5. The circuit must be rugged and reliable.
11
A-C INPUT
___ ...:;~----~---e--o+30VDC
R1
LEGEND: C=1 microfarad R1=1 k pot R2=1 k
C
R3= 130 R4= 47
D= 1N673 UT= 2N489A
Figure 3
UT
SCHEMATIC DIAGRAM OF TRIGGER CIRCUIT
PULSE OUTPUT
0
12
-.... _,_ -_ ... _,_
-_,_ _,_
----...
I I I I I I I I I I I I I I I I I 1 I I I I I I I I I I I I I I I I I I I I
I I I I I I I I I I I ' I I I I I I I I I I I I I I I I I I ------
-
----... _,_ ... _,_
5 10 15 20 25 TIME IN MILLISECONDS
Copy of oscilloscope trace of voltage ncroas relay load. Sweep= 0.5 milliseconds/division. 100 volts across load. 15% unbalanced condition.
Figure 4
TYPICAL DISPLAY OF OPERATING TIME
The circuits shown accomplish these conditions, being temperature
compensated over a range of -4o0 o to 100°0 (9). An additional test was performed to determine whether the
relay was too zensitive. This test consisted of operating a syn-
chronous motor connected to a D. 0. generator which served 1u a load.
The relay was placed in parallel with the motor. Load was suddenly
removed or added, and the operation of the relay was observed.
Discussion_£! statistical analysis_£! data. It was decided
to prepare a atatistical study of the data obtained from the pro-
cedure given above, in order to predict the maximum operating times
of the device. These data are given in tables 1, 2, and, for 10%,
15%, and 20% unbalance, respectively._ Attempta were made to obtain
data for 5% and 25% unbalances, but difficulties were .encountered.
The 5% unbalance resulted in erratic behavior due to the very poor
wave form present in the laboratory three-phase supply. This wave
form also created some inconsistencies in the data used, notably
in the 10% unbalance. The 15% and 20% unbalances were large enough
to compensate £or the poor wave form. The 25% unbalance resulted
in the machine losing synchronism. It also resulted in very fast
operating times, on the order of one millisecond or less for the
majority of operation&. It may safely be said that a 25% unbalance
will result in operating times that are fast enough to eliminate
all relay time for practical purposes of system stability.
It was found that the data used fitted a linear regression
14
Table l
Observed Times of Operation For Ten Per-Cent Unbalanced Condition
In Milliseconds
0.15 ;.oo 0.95 0.05 0.05 0.05 1.10 1.15 1.40 1.35 0.05 0.05 0.05 4.20 0.05 0.05 0.05 1.30 1.70 0.05 0.10 2.65 0.05 1.20 1.20 0.05 2.10 2.90 1.20 0.05 0.05 1.20 1.15 0.20 0.05 1.45 0.05 0.05 5.10 4.30 0.05 0.05 0.75 5.30 0.15 0.05 1.45 0.05 4.10 0.10 0.05 1.10 0.05 1.60 4.25 0.05 2.90 1.05 0.90 0.95 0.05 0.05 0.05 0.05 1.00 0.10 0.20 0.10 1.25 1.35 3.30 0.05 0.10 0.05 0.10 0.05 0.10 0.10 1.15 1.15 3.15 3~45 3.80 1.25 0.10 0.05 0.15 2.40 1.25 3.05 0.15 1.00 1.90 0.05 0.90 0.05 0.05 0.10 1.60 0.55 0.15 0.05 0.20 0.10 0.,:05 0.10 1.00 4.60 ;.10 0.05 o.ao 0.05 0.05 1.;o :t.15 0.05 0.10 2.;o 0.95 0.05 0.05 0.20 0.05 0.05
I 1.05
15
Table 2
Observed Times of Operation For Fifteen Per-Cent Unbalanced Condition
In· Mi 11 is econds
4.oo 0.01 ,.15 ;.45 ,.so 0.25 0.15 0.65 0.75 0.85 2.,5 1.15 0.01 1.95 0.85 2.70 1.15 0.95 ;.60 0.95 1.35 0.01 1.75 0.80 0.01 0.65 2.65 0.70 2.00 o.80 0.05 1.15 1.50 0.85 ,.80 1.,5 1.20 3.60 3.55 3.75 2.45 0.10 0.75 0.75 1.55 0.75 1.75 2.25 0.75 0.80 1.50 0.15 0.75 3.30 1.50 1.05 0.01 1.05 1.95 4.10 0.95 1.05 1.15 2.65 3.75 0.95 0.01 2.65 0.95 0.75 4.10 2.15 1.05 0.01 0.15 1.00 0.05 ,.80 4.95 4.75 0.85 0.05 0.01 2.05 3.15 0.10 0.05 5.15 0.75 0.90 0.15 0.10 o.80 0.95 2.70 1.00 0.25 3.60 0.70 0.85 0.95 0.70 2.50 0.85 2.00 2.25 0.01 3.53 1.30 1.20 o.80 0.75 5.25 ,.40 3.90 4.10 0.65 1.90 1.45 1.15 0.01 0.70 0.15 4.35 1.70 1.60 0.95 2.35 2.50 3.80 0.95 2.00 4.80 ,.25 1.10 1.10 ;.25 1.20 1.55 ;.20 2.00 1.95 o.80 0.80 0.95 0.75 1.00 1.25 3.65 ;.15
16
Table;
Observed Times of Operation For Twenty Per-Cent Unbalanced Condition
In Milliseconds
1.;o 1.25 0.05 0.10 o.40 0.15 1.15 1.00 0.65 0.60 1.05 0.05 1.00 0.75 0.70 0.75 1.10 5.00 0.05 1.;o 1.75 0.10 0.75 1.45 ;.55 0.15 ;.55 1.80 1.25 1.15 0.60 1.60 4.50 0.20 2.45 1.;o 0.25 0.10 1.65 0.50 2.65 0.65 0.10 0.05 o.;5 0.95 4~40 0.85 ;.05 0.50 0.75 0.10 1.60 ;.65 2.90 2.40 ;.60 ;.15 0.05 4.45 2.90 ;.55 1.00 2.40 0.05 0.05 0.80 0.85 2.20 0.20 1.10 0.05 ;.20 ;.20 1.20 0.10 0.15 5.,;o 0.85 0.05 ;.40 1.;o o.;5 0.05 2.60 1.50 2.50 0.05 o.;o o.;o 0.80 4.80 2.40 1.85 0.15 0.10 0.05 1.;o 1.05 0.05 1.05 1.10 0.10 0.75 0.05 2.;o 1.15 0.15 ;.10 0.85 1.15 2.80 o.;5 :;.;o 0.20 0.20 0.55 0.85 1.;5 0.10 0.85 0.05 0.60 2.80 0.10
17
in each case, according to the theory presented by Spiegel (11).
The plots of these regression lines are shown in figures 5, 6, and 7. These regression lines were obtained from data given in tables 4, 5, and 6 in the following manner:
X=g N
- IY Y=N
where N = number of operations observed.
According to Hoel (4) a regression line is given by:
Y'-Y= bCx-x> , where b=I:(xi-x) Yj
I(x. -x) 2 I
The chi-square test was used t9 determine the goodness of
fit for each of tho three unbalanced conditions used •. According to
Spiegel (11): 2 k (o- - e-) 2
X=r J 1 j=1 ej
The f'it was determined to be good by comparison Qf x 2 f'or the do.ta.
with X 2 as given in tables for six degrees of' freedom.
The standard error of' estimate, Sx~ , and the coefficient of
correlation, r, were determined f'or each case by use of' the equation
from Spiegel (11):
where tho negative sign indicates negative linear correlation.
100
80 Cf) z 0 I-
60 w Q._ 0 LL 040 n:: w d) 2 :) z 20
18
"'-.
' ......
" " "'-~I" r--....
1 2 3 4 5 TIME 1N MILLISECONDS
Figure 5 REGRESSION LINE FOR TEN PER-CENT UNBALANCED CONDITION
50
40 Cf) z 0 I-
30 w 0.. 0 LL 0 20 0:: w co 2 :) 2 10
"
19
"' ' "' '"" '""-
I~ 1 2 3 4 5
TIME IN MILLISECONDS
Figure 6 REGRESSION LINE FOR FIFTEEN PER-CENT UNBALANCED CONDITION
100
80
(/) z 060 I-<( 0:: w 0... 040 LL 0 0:: w ~20 :::, z
I~
20
I~ I"' ~. "'. II,.
"'· "'~ "' "' 1 2 3 4
TIME IN MILLISECONDS
Figure 7 REGRESSION LINE FOR TWENTY PER-CENT UNBALANCED CONDITION
5
21
Table 4
Data For Calculation of Regression Line For
Ten Per-Cent Unbalanced Condition
Time (X) Operations (Y)
X-X (X - X)Y (X - X)2
0.80 70 -2.40 .. .:.168.00 5.76 1.60 ;; -1.60 - 52.80 2.56 2.40 5 -0.80 - 4.oo o.64 ;.20 7 o.oo o.oo o.oo 4.oo ; o.80 2.40 o.64 4.80 5 1.60 8.00 2.56 5.60 2 2.40 4.80 5.76
22.40 125 o.oo -209.60 17.92
22
Table 5
Data For Calculation of Regression Line For
Fifteen Per-Cent Unbalanced Condition
Time Operations x-x (X - X)Y (X - X)2 (X) (Y) o.ao 48 -2.40 -115.20 5.76 1.60 .',9-5 -1.60 - 72.00 2.56 2.40 17 -0.80 - 1;.60 o.64 ;.20 12 o.oo o.oo o.oo 4.00 19 o.80 15.20 o.64 4.80 6 1.60 9.60 2.56 5.60 ; 2.40 7.20 5.76
22.40 150 o.oo -168.80 17.92
Table 6
Data For Calculation of Regression Line For
Twenty Per-Cent Unbalanced Condition
Time Operations X-X (X - X)Y (X - x,2 (X) (Y)
0.80 57 -2.40 -1;6.80 5.76 1.60 ;; -1.60 -217.80 2.56 2.40 9 -0.80 - 7.20 o.64 ;.20 1; o.oo o.oo o.oo 4.oo 7 0.80 5.60 o.64 4.80 4 1.60 6.40 2.56 5.60 2 2.40 4.80 5.76
22.40 125 o.oo -;45.00 17.92
24
Figures 8, 9, and 10 were then constructed to form time pre-
dictions on the basis of the observed data and the statement:
••• if we construct lines parallel to the regression line of Yon X at respective vertical distances a x' 2s , and 3Gy x from it, we should find, if N is largeyenough;•trui.t there would be included between these lines about 68%, 95%, and 99-7% of the sample points. (11).
The same reasoning holds true for horizontal distances plotted for
The actual calculations involved in this study will be found
in the appendix.
Results
The actual results are given in graphical form in figures 8,
9, and 10.
The load-changing test on the synchronous motor resulted in
a slight decrease in the sensitivity control, but the relay was still
sensitive to unbalance.
100
~80 0
t5 60 Q_
0 LL 040 0:: w aJ 220 :::> z
,~ "i
25
G' ~~--
'¾ I
2 4 . 6 8 10 TIME l N MILLISECONDS
No Predictions Possible.
Figure 8
PREDICTION LINES FOR TEN PER-CENT UNBALANCED CONDITION
12
100
(/)
Z 80 0 I-<(
f5 60 0... 0 LL 0 40 cc w (l)
2 20 :) z
26
', ' ' ' ' '~
" ' " "'' ' '" ~, '" '" ' ' " > ...
... 0
... ,~ ' ' ' ' ' ' ' ' 2 4 .6 8 10 12
TIME IN MILLISECONDS
Figure 9
PREDICTION LINES FOR FIFI'EEN ?ER-CENT UNBALANCED CONDITION
100
(./') z 80 0 I-<(
5 60 0... 0 LL 0 40 er: w m
20 :::::> z
\ \ \
I--
27
\ \ \ \ ' \ \ \ \
\ \ \ \\ \ \ , t,
1,, -;; 0
'f \ \ \ \ \ \ \ \
\ \ \ \ ' \ ' \
2 4 · 6 8 10 12 TIME IN MILLISECONDS
Figure 10
PREDICTION LINES FOR TWENTY PER-CENT UNBALANCED CONDITION
28
IV DISCUSSION OF RESULTS
It was shown that a more pronounced unbalance resulted in
shorter operating times, and tho.ta definite negative correlation
exists between time and number of operations. Statistical predic-
tions were made, which indicated that 99.7% of all operations would
occur below 10.94 milliseconds for 15% unbalance and below 8.51
milliseconds for 20% unbalance, which is a distinct improvement over
existing relays.
It was further determined that the relay could be desensitized
slightly, and that it would not be sensitive to sudden load changes
within the·normal range of loads.for the machine, but that it re-
mained sensitive to unbalanced conditions at its terminals. The
design is such that it will allow the operating voltage to be
changed, and its sensitivity to be controlled.
29
V CONCLUSIONS
The detection circuit used in this study could be vastly im-
proved by the design of a machine specifically for this purpose.
It is the autr~r•s belief that this method of detection and relay
circuit show promise of both increased speed and sensitivity over
existing types. Also, the relay circuit is quite rugged and reliable,
with no moving parts. The detection devico has the disadvantage
of being a rotating machine, but there is no reason why a machine
designed for this purpose could not be made highly reliable.
It is also the author's belief that consistent results could
have been·obtained for all conditions of unbalance from very small
up to the point where loss of synchronism occurred if the laboratory
wave form had been pure sine waves of the type normally encountered.
It has been shown by Thompson (12) that wave form distortion also
affects the operation of conventional relays.
;o
VI SUMMARY
Voltages appearing across the field of an alternator due to
negative-phase-sequence currents in the stator may be of such magni-
tude as to cause insulation failure in the field. A fast method of
detecting and relaying such a condition was devised.
Readings were taken of operating times for various conditions
of unbalance. Statistical time predictions were made for various
percentages of total operations.
It was concluded that the scheme offers the possibility of
faster operating times than existing methods, with no loss of de-
pendability. Also, the relay portion of the study offers increased
simplicity and ruggedness over existing types.
VII ACKNOWLEDGEMENTS
The author wishes to express his appreciation to Professor
George R. Powley for his guidance and encourQgement; to Professor
George C. Barnes, Jr. and Dr. Mansell H. Hopkins for their interest
and suggestions; to Professor Ralph R. Wright, Head of the Electrical
Engineering Department, for making available facilities in the
Electrical Engineering Laboratories; and to Dr. Albert L. Duke for
the loan of special equipment.
The author also wishes to express his appreciation to Mr.
Rudolph P. Hensley, M.r. Fred Bower, and Mr. Marvin Surface for their
assistance·.
Finally, the author wishes to express his gratitude to his
wife, Patricia P. Lancaster, for her inspiration, cooperation,
patience, help, and typing.
1.
2.
4.
VIII BIBLIOGRAPHY
Clarke, Edith. New York:
Circuit Analysis :lf_!:::::2._ Power Systems, Volume II. John \·iiley and Son:s, 19.50. Pp. 328-362.
Electric Transmission and Distribution Reference Book. Fourth Edition. East Pittsburgh, Pennsylvania: Westinghouse Electric Corporation, 1950. Pp. 349-352.
Henderson, John. Automatic Protective Gear for!:£ Supply S¾stems. London: Sir Isaac Pitman and Sons, 1934. Pp. l 1-143.
Hoel, Paul G. Edition. 128.
Introduction to Mathematical Statistics. New York: JolmWiley and Sons, 1954. Pp.
Second 127-
5. Hopkins, Mansell H., Jr. 11A Method of ~1easuring Negative-Phase-Sequence Currents in a. Three-Phase System." Unpublished Master's Thesis, Virginia Polytechnic Institute, Blacksburg, 1958. Pp. 26-28.
6. Kimbark, Edward Wilson. Poi1er Svstem Stability, Volume III. New York: John Wiley and Sons, 1956. Pp. 220-226.
7. Mason, O. Russell. The Art and Science of Protective Relaying. New York: John \•/iley and Sons, 195b. Pp. 221-225.
8. Silent Sentinels. Newark, New Jersey: Westinghouse Electric Corporation, 1949. Pp. 42-43.
Silicon Controlled Rectifier Manual. Third Edition. Auburn, New York: 'General Electric Company, 1964. Pp. 55-60.
10. Skrotzki, Bernhardt G. A. Electric System Operation. J',~cGrs. w-Hi 11, 1954. ?p. 64-65.
New York:
ll. Spiegel, Murray R. Theory and Problems of Statistics. New York: Schaum, 1961. Pp. 202~42-244.
12. Thompson, Frederick W. 11The Effect of Voltage Wave Form on the Operation of Two 1",rpes of Current Overload Relays. 11 Unpub-lished Master's Thesis, Virginia Polytechnic Institute, Blacksburg, 1951. P. 17.
13. Transistor Manual. Seventh Edition. Syracuse, New York: General Electric Company, 1964. Pp. 300-528.
14. Wagner, C. F. and R. o. Evans. York: McGraw-Hill, 19:;;.
Symmetrical Components. New Pp. :;0-99, 289-292.
Wright, Ralph R. and H. Richard Skutt. Electronics Circuits and Devices. New York: Ronald Press, 1965. Pp. 140-145, ;9:;-;95.
The vita has been removed from the scanned document
APPENDIX
CALCULATIONS
Regression line: y 1 - y = b(x - i) where b = I (X - X)Y
I (X - x) 2
Chi-square test for goodness of :f'i.t:
,,.2 K (o· - e->2 A =L I I
j=1 e j
where the number of degrees of freedom vis given
by v : k - l
For ten per-cent unbalanced condition:
Time Observed Estimated
0.80 70 46 1.60 ;; ;6 2.40 5 27 ;.20 7 18 4.oo :; 9 4.80 5 - l 5.60 2 -10
125 125
For fifteen per-cent unbalanced condition:
Time Observed Estimated
o.80 48 41 1.60 45 .?.? 2.40 17 25 :;.20 12 17.5 4.oo 19 10 4.80 6 2 5.60 -2 - 6
150 122.5
For twenty per-cent unbalanced condition:
Time Observed Estimated
0.80 57 64.5 1.60 33 49 2.40 9 33.5 3.20 15 17.5 4.oo 7 2 4.80 4 -15 5.60 2 -28
·· 125 125.5
For ten per-cent unbalanced condition:
b = -11. 7, from datu in table 4.
Regres3ion line is y 1 = -11. 7x + 55.44 -whel"e y = 18.
The statistic X 2 = 9.0
2 From tables, X =18.5, .995
Since O .676 < 9 .O < 18.5, degrees of fl"eedom.
and 2 X .0 5 = 0.676.
the fit is good for six
For fifteen per-cent unbalanced condition:
b = -9.4, from data in table 5. Regres3ion line is y 1 = -9.4s + 48.1
where y = 18. 2 The statistic X = 12.0
Since 0.676 < 12.0 < 18.5, the fit is good for six degrees of freedom.
For twenty per-cent unbalanced condition:
b = -19.3, from data in table 6.
Regression line is y 1 = -19.;x + 79.76
where y = 18.
The statistic X 2 = 17 .05
Since 0.676 < 17 .05 < 18.5, the fit is good for six degrees of freedom.
Standard error of estimate and coefficient of correlation:
For ten per-cent unbalanced condition:
sx.Y = o.o r = -1.0
Therefore hypothesis is statistically significant.
For fifteen per-cent unbalanced condition:
sx.Y = 1.98
r = -0.146
Therefore hypothesis is statistically significant.
For twenty per-cent unbalanced condition:
SX.Y = 1.4;7
r = -0.50
Therefore hypothesis is statistically significant.
.A. PROTECTION SCHEME FOR ALTERNATOR FIELDS
James Terry Lancaster
Virginia Polytecbnic Institute
Department of Electrical Engineering
MASTER OF SCIENCE THESIS
Abstract, 1965
A synchro-generator and a unijunction transistor oscillator
in conjunction with a silicon-controlled rectifier form a detection
and relay circuit for voltages appearing across the field of a three-
phase alternator generated by negative-phase-sequence currents due
to unbalanced conditions.
Readings were taken of the times required for operation of
the relay for various conditions of unbalance. Statistical studies
were made to enable time predictions to be offered.
It was concluded that the scheme offers the possibility of
faster operction, and hence better protection and system stability,
than is possible with existing types. In addition, the scheme will
result in no loss of dependability over existing types, and the relay
portion offers outstanding ruggedness and sensitivity.