7 2" 24 0 50

N A S A TECHN ICALM E M O R A N D U M

NASA TM X- 68068

00NO

vO

IX

PARALLEL OPERATION OF TWO BRAYTON-CYCLE

ALTERNATORS WITH PARASITIC SPEED CONTROLLERS

by Dennis A. PerzLewis Research CenterCleveland, Ohio

TECHNICAL PAPER proposed for presentation atSeventh Intersociety Energy Conversion Engineering ConferenceSan Diego, California, September 25-29, 1972

https://ntrs.nasa.gov/search.jsp?R=19720016400 2018-09-04T00:52:02+00:00Z

PARALLEL OPERATION OF TWO BRAYTON-CYCLE ALTERNATORS

WITH PARASITIC SPEED CONTROLLERS

by Dennis A. Perz

Lewis Research CenterNational Aeronautics and Space Administration

Cleveland, Ohio

ABSTRACT

The experimental paralleling characteristics oftwo 1200-Hz Brayton-cycle alternators are presented.Since the Drayton power conversion system uses elec-t r i c ,s|icc<l (tmlrollcrH, Iho p j i r n l l c l i n ^ r o q u i r r m r n t Htiro homcwhal different, from lltone for conventionalground-based power systems. Results include thetransient effects of synchronizing the two alternatorswith various phase-angle, voltage, and frequencydifferences. Based on these results, the effects of

^ synchronizing differences can be defined, and adjust-ment requirements of the parasitic speed controllersduring synchronizing can be established. Data indi-cate that parasitically loaded alternators are able toparallel over a wide range of synchronizing differ-ences. However, equilibrium could not be reachedin extreme cases where alternator load differenceswere great and, at the same time, the phase-angleerror was large (150° or more).

THE DEVELOPMENT OF DYNAMIC POWER SYSTEMSfor space applications has reached the point where theuse of multiple generators operating in parallel is ofinterest. The range of space application of Hraytonsystems may be extended by using two or moren x i H i l i H - H rnlliiit* Hum j t iMl n Hlit|;lu mnchlnii. Tin- IIHOof multiple generators permits u modular approach U>power-system design with the resulting advantages ofincreased flexibility and reliability and reduced cost.

Much work has been done in the past on multiple-generator systems, both in aircraft and commercialpower applications; however, the advanced turbo-alternator systems currently under investigation toruse in space (1)* differ from conventional power sys-tems in several ways.

Space power systems use an electric speed con-toller in combination with a parasitic load instead ofa turbine valve and mechanical governor (2). Theelectric speed controller applies a parasitic electricalload to the alternator as a function of frequency inorder to absorb the constant generated power evenwhen the useful system load varies. The effect of theparasitic load on synchronizing and system stabilityafter paralleling must be determined.

Space power systems also use solid-rotor alter-nators. The solid-rotor machine does not requirebrushes for field excitation, is easier to cool thanwound-rotor machines, and is considered more re-liable. Solid-rotor alternators do, however, havelarger rotor inertias than wound-rotor machines forthe same output. Since rotor inertia is significantduring the synchronizing transient, the effect of thelarger value must be determined.

As an initial effort to study the paralleling char-acteristics of parasitically loaded turboalternators,two alternators from the 1200-hertz Brayton-cyclesystem (1) were tested. In addition, breadboardversions of system parasitic speed controllers, shuntvoltage regulators, and series field modules wereused (2).

Transient results of synchronizing the two alter-nators under several load conditions are presentedfor various differences in voltage magnitude, phase,and frequency. Of particular interest are (1) theability of the alternators to parallel in spite of syn-chronizing differences, (2) the magnitude of the syn-chronizing transients, and (3) the effects of the

^Numbers in parentheses designate Referencesat end of paper.

parasitic speed controllers on synchronizing.

PARALLELING CONSIDERATIONS FOR PARASITI-CALLY LOADED TURBOALTERNATORS

1 is a s imi i l i f iod block dinnrnni «if n nii; . n r r i i i u r j i i i r i i H l t l c i i l l v li>iiilt!«l |x>wur HyHlom. 'Iliinfigure sliows the significant power-transfer stages be-tween the prime energy source and the useful systemelectrical load.

Turboalternator power systems with parasiticloads are designed to operate at a single constantpower level. Thus the energy source has a constantoutput, and no provision is made to control the powerinput to the turbine. As a result, the total alternatorelectrical output must also be constant so that nomi-nally constant speed and frequency can be maintained.The useful system load is assumed variable, however,so a frequency-sensitive waste, or parasitic, load isadded. The parasitic load, which typically operatesover a 2-pcrccnt frequency range, offsets any changein the useful load so that the alternator frequencyremains nearly constant.

For alternators to be synchronized, they musthave the same phase sequence. In addition, any dif-ferences in the voltage magnitude, phase, or frequencywill produce transients in the form of current, power,and torque surges in proportion to the size of thedifferences. These transients arise out of the naturaltendency of alternators to self-synchronize (i.e. , tominimize magnitude, phase, and frequency differ-ences) when tied to a common bus.

In parasitically loaded turboalternator power sys-tems, alternator frequency varies with the amount ofparasitic load applied. As a result, when two parasiti-cally loaded turboalternators are to be paralleled, afrequency difference will exist before paralleling ifthe two machines have unequal amounts of parasiticload. This is likely to be the case.

A further consequence of using a parasitic speedcontroller is the effect on voltage regulation. Asnoted in reference (3), a characteristic of presentlyused speed controllers is that alte'rnator voltage mag-nitude varies with the amount of parasitic load applied.For the 1200-hertz Brayton-cycle power system,voltage variations of about 3. 5 percent of rated voltagewere measured over the operating range of the speedcontroller.

Because of the existence of voltage-magnitude andfrequency differences, the question arises regarding

the feasibility and practicality of paralleling withoutregard to synchronizing errors (r.indom paralleling).The advantage gained is that special synchronizingcircuitry would not be required. The < > l n »>us draw-backs arc that (1) the alternators may l"% nimble toHynchronlXf! for vory Innjf phiiHo mul frequentlyc r n n - H , and (2) the peak transient currents and lowersmay be excessively large and potentially harmful tothe machines. j

In order for the range of acceptable conditions tobe determined for multigenerator operation, the sys-tem must be experimentally subjected to the varioustypes of errors possible during synchronizing. It isalso of interest to examine any variations in conditionswhich may occur when the machines are paralleled.For example, the useful-load power factor may bevariable, or the total generator loads may be unequalas during system startup.

APPARATUS AND PROCEDURE

The components tested were built as part of the1200-hertz Brayton-cycle program currently underinvestigation by NASA. Figure 2 shows the majorcomponents used for this test. Two alternators,each with its own voltage regulator and parasiticspeed controller, were tested.

ALTERNATORS - The two alternators used werethe Brayton rotating unit with rolling element bearings(BRU-U) (fig. 3) and the alternator research package(ARP) (fig. 4). The BRU-R (4) consists of a turbine,alternator, and compressor mounted on a singleshaft. The alternator is a modified Lundell designwith stationary field coils.

The ARP was built to be electromagneticallyequal to the BRU-R alternator. Its original purposewas for experimental determination of alternatorcharacteristics. These data are presented in ref-erence (5). The nominal three-phase rating of thetwo alternators is 14. 3 kilovolt-amperes at a j0.75 power factor. Rated voltage is 120/208 volts,rated current is 39.7 amperes, and rated power is10.7 kilowatts.

The ARP was coupled to a facility turbine, andboth it and the BRU-R were operated with a facilityair supply. A servovalve was used at each turbineinlet and was automatically controlled to maintain apreset valve position. Thus, constant power wasprovided to each turbine for a given valve position.

For the purposes of this test, the compressor

wheel on the BRU-R was replaced by a solid disk ofequal inertia. The compressor inlet was then sealedto minimize windage losses.

VOLTAGE REGULATORS - As shown in figure 2the voltage regulators used consist of a shunt fieldrn*ul:ilnr which HCIIKPH lond I M I H vollngo :iml 11 HcricBI ' i r l i l controller which S C I I H C H l ino c u r r e n t . Knellexcites separate windings in the alternator. Bread-board versions were used, and the operating charac-teristics are reported in references (3) and (6).

Two modifications to the shunt regulators wereincorporated into the original design (2). First aconstant volts per hertz reference circuit was addedas described in reference (7). The second modifi-cation was the addition of reactive load compensationcircuits for paralleling. The circuit design used wasbased on the aircraft system application describedin reference (8).

PARASITIC SPEED CONTROLLERS - The speedcontrollers are breadboard units and are described inreferences (3) and (6). The speed controllers adjustthe load dissipated in a resistor by means of phasecontrol. Phase control is a technique by means ofwhich a solid-state switch (SCR) can be made to con-duct current over a controlled portion of each halfcycle of the alternator voltage. The effective currentis varied, by varying the conduction interval, therebycausing the resistor's parasitic load to change. Thespeed controller senses alternator frequency, whichis proportional to rotational speed, and adjusts theSCR conduction interval accordingly. The two para-sitic speed controllers used were adjusted to havesimilar loading characteristics.

TEST PROCEDURE - The total power on eachalternator was set by adjusting the trubine servo-valves. Alternator voltage adjustments were madeby potentiometers provided in the shunt regulators.Synchronizing was performed by observing the realtime relationship between the two alternator voltageson an oscilloscope. The paralleling contactor wasopened and closed manually. All data were recordedon an oscillograph. Accuracy of the data is 5 percentor better. All loads were nominally balanced three-phase.

RESULTS AND DISCUSSION

The data presented here are based on the differ-ences in voltage magnitude, phase, and frequencybetween the two alternators at the moment the paral-

leling contactor is closed. For the sake of brevitythe following definitions are made:

Af frequency differenceAV voltage magnitude (rma) differenceA</> phase

For each data point, the maximum rms alternatorphase current, maximum average single-phase alter-nator power, and maximum and minimum rms load-bus voltage variations are presented. The currentand power peaks are measures of the torque loadsplaced on the machines during synchronizing. Thevoltage swings are indications of the effects the syn-chronizing transient can have on the useful load.

PHASE-DIFFERENCE EFFECTS. Zero-PowerTransient - Figure 5 presents a synchronizing tran-sient with zero load on each machine: AV = 0,Af = 0, A<p = 145°. When the paralleling contactoris closed, the phase difference immediately goes tozero, as only one voltage can exist on the bus.

The current and power traces show that the sys-tem oscillates at its natural frequency and is dampedin this case. When the machines are paralleled outof phase, the rotors must adjust their angular posi-tion relative to each other in order to reach equilib-rium. Depending on their relative angular positions,one machine will motor off tho other machine whichoperates as a generator. Rotor inertia causes anovershoot, however, so that the functions are thenreversed — the generator becomes the motor and themotor the generator. This process continues untilsystem damping causes the transient to die out andthe proper phase relationship is attained. (By con-vention, generator power (output) is positive andmotoring power (input) is negative.)

Figure 5 also shows the transient effect on theload-bus voltage (see fig. 2) for the same phase errorof 145°. An initial voltage drop occurs, followed bya rise as the alternator currents oscillate and theshunt voltage regulator adjusts. The voltage is modu-lated as the current surges between machines. As aresult, peak voltages well above the 120-volt ratingare realized.

Effects of Voltage Magnitude Errors - In figure 6the combined effects of phase error A<p and voltageerror AV are shown; Acp ranges from 0° to 180° andAV from 0 to 5 percent. The alternator load is zero.The effects on the extreme values of alternator current,alternator power, and load bus voltage are shown.

Two observations can he made. First, caehcurve has an extreme value occur nopr 180°. (Referto fig. 5 for peak locations.) This in expected becausethe greater the A<p, the greater tin- transient.

The second observation is that l i l l l u change occursIn r ii AV ol Ii perreii l r u l n l l v i ' l«> » AV if 0 percent.Since nil v o l l n ^ c va r i a t ions < l i i f to tlio parasitic loadsare less than 5 percent, it was concluded that AVerrors arc not significant.

Kffccts of Alternator Loads - The variation oftransient peaks with phase error for a range of equaltotal alternator loads is presented in figure 7. Ineach case the total alternator load was 100 percentparasitic (/ero vehicle load). Each machine had itsown parasitic speed controller, as shown in figure 2.

Results show that maximum alternator rms cur-rent and maximum average alternator power occurnear 180° and increase with the magnitude of the load.Maximum measured phase current is 113 amperes for6-kilowatt loads with approximately 180° A<p. Peaksingle-phase power for the same condition is 12 kilo-watts (30 kW total). These peaks are about six timesthe original loads and about three times the machineratings.

Figure 7(c) indicates that voltage maximums andmin imums are not affected by magnitude variations inequal alternator loads.

Two other significant points are (1) the machinessuccessfully synchronized for even the most extremephase errors, and (2) the two-machine paralleledsystem remained stable even with two parasitic speedcontrollers operating simultaneously and independentlyon the same bus.

Figure 8 presents the transient effects of syn-chroni/.ing alternators with unequal loads. In eachcase., one alternator was unloaded (0 kW) and theother had only a parasitic load ranging from 3 ID7 .:» kilowatts. The most significant result is thatalthough paralleling was successful under most condi-tions, equilibrium could not be reached for the mostcNlreme phase-angle differences. For a 6-kilowattload difference, the maximum possible AV is about}<iO°; for a 7. 5-kilowatt load difference, the maximumA<£ is l.r>0°. In order to determine the cause of thislimitation, the test for the 6-kilowatt load differencewas repeated several times with the following vari-.ition.0:

(1) The loads on the two alternators were reversed.(2) The load was changed from phase-controlled

parasitic to linear useful at 1.0 and 0.75 lagging power

factors.(3) The load levels were changed from 0 and 6

to 3 and 9 kilowatts.(4) The reactive load compensators were removed.

In each case, the 160° AV limitation appeared. Thus,It nppeni-H Ih i i t the CI-OHH i i l lor i inlor load difference inthe source of the l imitation.

Figure 9 is an unstable synchronizing transientfor initial machine loads of 0 and 7.5 kilowatts, AV = 0,Af = 0, and AV = 153°. In contrast to the transient infigure 5, the current and power swings are not dampedbut are actually increasing in time. The parallelingcontactor was manually opened as soon as this condi-tion was determined.

Examination of the phase error trace after the *"paralleling contactor was opened shows that AV has afrequency of about 35 hertz. This indicates that thetwo machines have been driven apart in speed at thispoint compared to the initial Af = 0.

EFFECTS OF FREQUENCY DIFFERENCES - Theeffects of paralleling the two alternators with differentfrequencies was first looked at for the zero-load case.The open-circuit voltages were equalized (AV = 0) andthe frequency difference was adjusted with the turbineservovalves. Results show that for frequency differ-ences up to 40 hertz, the synchronizing transient isdamped and stable. The 40-hcrtz difference repre-sents a 3. 3-pcrccnt error for a 1200-hertz system,and this is greater than the 2-percent frequency regu-lation of the parasitic speed controller.

To determine the effects of frequency differenceson loaded machines, each turboalternator was connect-ed to its parasitic load and the turbine valves wereadjusted to give about 6 kilowatts per machine. Onevalve was then closed in small increments. Each ad-justment had the effect of reducing that alternator'sload and frequency. After each adjustment, the alter-nators were synchronized randomly several times.The first case of unsuccessful paralleling occurred formachine loads and frequencies of 6.0 kilowatts at1216 hertz and 3.6.kilowatts at 1210 hertz. Thus, afrequency difference of only 6 hertz coupled with aninitial power difference of 2.4 kilowatts is enough toprevent synchronization. Paralleling was attemptedfor this condition five times and was successful fourtimes. *•

SUMMARY OF RESULTS

Two 1200-hertz parasitically loaded turboalter-

„

nators were subjected to a variety of synchronizingtransients in which voltage magnitude, phase, andfrequency differences were varied. Results were gen-erally favorable in that the machines were able to syn-chronize over a wide range of conditions. Some limi-liiUoiiM do c x i s l , however .

The following rowulls hold for equal frequenciesand equal open-circuit voltage magnitudes:

1. The two machines successfully paralleled withequal loads of 0, 3, and 6 kilowatts for any phase dif-ference.

2. When initial machine loads are unequal,synchronizing is not always successful. For a loaddifference of 6 kilowatts, the maximum permissiblephase error for successful synchronizing is 160°. Fora load difference of 7. 5 kilowatts, the maximum erroris 150°.

3. Paralleling the two machines, each with itsown parasitic speed controller and for equal loads of3 and 6 kilowatts, results in stable synchronizing andsteady-state operation with any initial phase difference.After paralleling, the two-machine system operatingwith two independent parasitic speed controllers re-mains stable.

4. For equal loads of 6 kilowatts and a synchroniz-ing error of 180°, the worst transients resulting insuccessful paralleling occurred. The measured rmsphase current was greater than 110 amperes and thepeak three-phase power was 36 kilowatts. These peaksare about 600 percent of the levels before parallelingand about 300 percent of the alternator ratings.

5. Additional testing showed that open-circuitvoltage errors up to 5 percent and frequency errorsover 3 percent did not significantly affect the no-loadsynchronizing capability of the two machines.

REFERENCES

1. J. L. Klann, "2-to-10 Kilowatt Solar or Radio-isotope Brayton Power Systems," IECEC, Vol. I,IEEE, pp. 487-415, 1968.

2. B. D. Ingle and C. S. Corcoran, "Developmentof a 1200-Hz Alternator and Controls for Space PowerSystems," IECEC, Vol. I, IEEE, pp. 437,438, 1968.

3. B. D. Ingle, H. L. Wimmer, and R. C.Bainbridge, "Steady State Characteristics of a VoltageRegulator and a Parasitic Speed Controller on a 14. 3-Kilovolt-Ampere, 1200-Hz Modified Lundell Alterna-tor, " NASA TN D-5924, 1970.

4. B. B. Heath and R. A. Luther, "The Designand Fabrication of the Brayton Rotating Unit Operatingon Rolling Element Bearings (BRU-R)," Rep. HPS-5327-R, NASA CR-72642, AiResearch M.-inufiicturingCompany of Ari/x>na, Sept. 19G9.

5. D. S. llopaH mid 11. A. Kc lk in , "I'erformnncoCharacteristics of a 14.3 Kilovolt-Amperc ModifiedLundell Alternator for 1200 Hertz Brayton CycleSpace Power System," NASA TN D-5405, 1969.

6. H. L. Wimmer and B. D. Ingle, "TransientCharacteristics of a Voltage Regulator and a ParasiticSpeed Controller on a 14. 3-Kilovolt-Ampere, 1200-Hertz Modified Lundell Alternator," NASA TN D-5925,1970.

7. H. L. Wimmer, "Experimental Evaluation ofa Volts-Per-Hertz Reference Circuit for the IsotopeBrayton System," NASA TM X-2502, 1972.

8. H. A. Kahle, "Parallel Operation of AircraftA-C Generators: Reactive Load Division and Selec-tive Protection," AJEE Trans., Vol. 75, p. 2, #25,July 1956.

GASLOOP

ENERGYSOURCE

TURBINE POWER = CONSTANT

P + P =ru rpCONSTANT

FIG. 1. Power flow in turboalternator power system with parasitic

speed controller.

VOLTAGE REGULATOR

i

ALTERNATOR

SHUNT- FIPI n

REGULATOR

nSERIESnn n

CONTROLLER

VOLTAGE REGULATOR

ALTERNATORi i

SERIESFIELDCONTROLLER

1

SHUNTFIELDREGULATOR

11-11

11i

Fig. 2. Major electrical components of a two generator power system.

-72-953

CT-N£>U1

Figure 3. - Brayton rotating unit (BRU-R).

IC-72-951

Figure4. - Alternator research package (ARP).

-!

Q_

0

90

180

270

360

120

80

400

145° PARALLELING CONTACTOR CLOSED

I I I I I I I

(a) PHASE DIFFERENCE ACROSS PARALLELING CONTACTOR.

-78 A (MAX)

0 .2 .4 .6 .8 1.0

TIME, SEC

1.2 1.4 1.6 I.!

(b) RMS ALTERNATOR PHASE CURRENT.

Figure 5. - Synchronizing transient for zero alternator load,zero voltage magnitude error, zero frequency error, and145° phase error.

Q_I—

o 3P: ^0. CC

=2 <OQ >j

O O< >O

160

120

80

400

i GENERATOR POWER

°TMOTOR POWER

(0 AVERAGE SINGLE-PHASE ALTERNATOR POWER.

,-152 V (MAX)

140V (MINII I J I I0 .2 .4 .6 .8 1.0

TIME, SEC

1.2 1.4 1.6 1.!

<d) RMS LOAD BUS PHASE VOLTAGE

Figure 5. - Concluded.

o.1/15Of

O

100

80

60

40

20

0

10

VOLTAGE MAGNITUDE DIFFERENCE,AV,

iCT

(a) PEAK PHASE CURRENT.

Ofo

cc:

<ca.

o:

> 0

200

160

120

CO

oo

1/1SOf

40

(b) MAXIMUM SINGlf-PHASE ALTERNATORPOWER.

MAXIMUM

) 40 80 120 160 200PHASE DIFFERENCE, A<P, DEG

(c) VOLTAGE EXTREMES.

Figure 6. - Peak transient effects of parallelingalternators with 0 kilowatt loads and unequalphase voltages.

(a) PEAK PHASE CURRENT.

(b) MAXIMUM SINGLE-PHASE ALTERNATOR

2GQ r POWER"

o

120

CL.

(S)=3CO

ao

!̂

I

MAXIMUM

MINIMUM

40 80 120 160PHASE DIFFERENCE, A^ DEC

200

(c) Voltage extremes.

Figure 7. - Peak transient effects for paralleling al-ternators with equal loads.

120—PARALLELING

/ LIMITS

icr-sOL.n

r-(a) PEAK PHASE CURRENT.

J1^PARALLELING' LIMITS

o

200r-

160-

120

80-

40

(b) MAXIMUM SINGLE-PHASE ALTERNATOR

POWER.

MAXIMUM'

MINIMUM

II 40 80 120 160 200

PHASE DIFFERENCE, A¥>, DEG

(O VOLTAGE EXTREMES.

Figure 8. - Peak transient effects for paralleling al-ternators with unequal loads.

ceSo

oo

CL

0

90

180

270

360

h*153°, PARALLELING

CONTACTOR CLOSED

PARALLELINGCONTACTOR OPENED

AVERAGEFREQUENCY,-

35 Hz

1 1 1 1 I(a) PHASE DIFFERENCE ACROSS PARALLELING CONTACTOR.

to <'. 200

1000

131 A (MAX)

.1 .2 .3 .4 .5TIME, SEC

.6 .7 .8 .9

(b) RMS ALTERNATOR PHASE CURRENT.

Figure 9. - Synchronizing transient for 7. 5-kilowatts para-sitic BRU-R load, 0-kilowatt ARP load, zero voltage magni-tude error, zero frequency error, and 153° phase error.

-11.5kW(MAX)

^GENERATOR POWERMOTOR POWER

I

(O AVERAGE SINGH-PHASE ALTERNATOR POWER.

•170 VIM AX)

1 .2 .6 .7.3 .4 .5TIME, SEC

(d) RMS LOAD BUS PHASE VOLTAGE.

Figure 9. - Concluded.

NASA-L.wi.-Com'l