Proceedings of Space Nuclear Conference 2007Boston, Massachusells, June 24-28 2007

Paper 2030

Capabilities and Testing of the Fission Surface Power Primary Test Circuit(FSP.PTC)

Anne E. GarberNASA/Marshall Space Flighf Center

Huntsville, AL 35824Tel: 256-544-0665, Fax: 256-544-22J6, Email: Anne.E.Garber@nasa.gov

Abstract - An actively pumped alkali metal jlow circuit, designed and fabricated at the NASAMarshall Space Flight Center, is currently undergoing testing in the Early Flight Fission TestFacility (EFF-TF). Sodium patassium (NaK), which was used in the SNAP-lOA fission reactor,was selected as the primary coolant. Basic circuit companents include: simulated reactor core,NaK to gas heat exchanger, electromagnetic (EM) liquid metal pump, liquid metal jlowmeter,load/drain reservoir, expansion reservoir, test section, and instrumentation. Operation of thecircuit is based around a 37-pin partial-array core (pin andjlow path dimensions are the same asthose in afull core), designed to operate at 33 kWt. NaKjlow rates ofgreater than J kg/sec maybe achieved, depending upan the pawer applied to the EM pump. The heat exchanger providesfor the removal of them/al energy from the circuit, simulating the presence of an energyconversion syslem. The presence of the test section increases the versatility of the circuil. Asecond liqUid metal pump, an energy conversion system, and highly instrumented thermalsimulators are all being considered for inclusion within the test section. This paper summarizesthe capabilities and ongoing testing of the Fission Surface Power Primary Test Circuit (FSP­PTe).

I. INTRODUcnON

To expand the multi-mission technology base related tothe use of alkali metal systems for potential surface powerapplication, the Early Flight Fission Test Facilities (EFF­TF) tearn has designed, fabricated and filled a pumpedalkali metal (NaK) flow circuit. The circuit is comprisedof a core, heat exchanger, liquid metal pump, liquid metaIflowmeter, test section, lower (load/drain) reservoir, upper(expansion) reservoir, tubing, instrumentation, and spilltray. J The first test matrix is designed to characterize theperformance of the electromagnetic liquid metal pump,which the vendor predicted would operate at near 5%efficiency. Thirteen pump performance data points havebeen generated thus far, which paint a clear picture ofpump operations under a variety of conditions. A powerbalance of the system has also been generated, whichverifies the proper functioning of the NaK flowmeter.

II. PRIMARY COOLANT

Liquid metals are an attractive choice for space reactorsystems as a heat transfer medium due to their highthermal conductivities, high heat capacities and low vapor

pressures. NaK-78 is a eutectic mixture of 22% sodiumand 78% potassium by weight with a melting point of ­12.6°C and a boiling point of 785°C, making it a liquid atroom temperature. While freeze/thaw issues must beconsidered for space operations, this low melting pointmakes non-nuclear ground testing simpler as the liquidmetal need not be heated before it can be circulated. NaKhas good compatibility with stainless steel over thetemperature range to be tested (650°C maximum). NaKalso has a demonstrated service history for use in spacereactor systems, which include SNAP-lOA, TOPAZ I & 11,and the RORSAT series. One notable disadvantage ofNaK lies in its volatility. It reacts when exposed tooxygen, resulting in the formation of potassium oxide,potassium peroxide and potassium superoxide, whichcreate a crust on the surface of the NaK. Hydrogen isreleased when NaK is exposed to water, and the heat ofreaction causes it to burn.

ill. CIRCUIT HARDWARE

The key components of the FSP-PTC are identified inFigures 1 and 2. The test article is mounted on a 3 m by 2m aluminum tilt table with a 13 em-high lip around its

https://ntrs.nasa.gov/search.jsp?R=20070031971 2020-05-14T10:36:13+00:00Z

"

edge, which is sufficient to contlin all of the NaK in thesystem should a leak occur. The entire unit is installed ina 9-ft vacuum chamber for testing. NaK canspontaneously combust when exposed to oxygen, aphenomenon which is exaceIbated by elevatedtemperatures; thus, testing in a vacuum adds a layer ofsafety beyond the physical boundary of the chamber itself.Instrumentation, power connections, and various gases(for pump cooling, pump head pressure, valve control andheat exchanger operations) interface with chamberfeedthroughs. No NaK is routed oulside of the chamber.

Figure I. FSP-PTC without support structure

Figure 2. FSP-PTC as constructed

All components are constructed of stainless steel. Thecircuit is all-welded save at a handful of locations whereVCR fittings have been employed: at either side of theliquid metal flowmeter, at the expansion reservoir inlet,and just downstream of the core inlet. These fittings arealso used to connect the pressure transducers and liquidlevel sensors to the tubing and reservoirs, respectively.Prior to the start of testing all components and tubing wereproof pressure tested to greater than 300 psi and leakchecked using gaseous helium to better than 10-9 standard

Proceedings of Space Nuclear Conference 2007Boston, MassacI1uselts, June 24-28 2007

Paper 2030

cc1sec. The potential loosening of the mechanicalconnections due to cyclic heating and cooling of the circuitis being monitored, and no evidence of leakage has beenfound to date.

mAo COn! and Thermal Simulators

The circuit was designed around the core, which ishaselined from a IOO-kWt Los Alamos design stud/. Inorder to simpliry the fabrication, integration, and testingof the hardware, the reflector drums and shields locatedbeyond the outer pressure sheU were eliminated_ Inaddition, a partial array of the core geometry wasconstructed This consists of the central three rings (37fuel pins), resulting in a II3-power version of the fullassembly, or 33 kW total. These reductions significantlyimpact the overall cost and complexity of the system (onlyone electromagnetic pump is needed, the requisite size ofthe heat exchanger and associated plumbing is reduced,and fewer thermal simulators are employed).

1f~~~aii~1!1!1!!tl!ml1}1!;~ ~ -r\- I Z1;~-..... 0....-.-... __ .~

Figure 3. Core assembly

All testing performed at the EFF-TF is non-nuclear innature, and thermal simulators are used to simulate theheat of fission'. The heater elemenls selected for thissystem are based on a graphite design that has beensuccessfuUy used on a number of prior EFF-TF projecls.Each element is 59.7 em long and 0.775 em in diameter.Alumina insulators are spaced every few centimeters alongthe length of each unit to electricaUy isolate the graphitefrom the stainless steel cladding that encompasses it. Toimprove thermal performance, a core face seal has beenincorporated into the core assembly. The face seal permilsa helium environment to be established between thegraphite and the steel cladding, significantly improvingthe thermal coupling between them.

FoUy instrumented thermal simulators (e.g. incorporatingthermocouples at mnltiple axial locations) are undergoingdevelopment.' If this effort is successful, the centralthermal simulator will be replaced with a fuUy­instrumented element for additional characterization ofthe test article during heated operations. Figure 4illustrates a thermal simulator prior to installation in ilssteel cladding.

Proceedings of Space Nuclear Conference 2007Boston, Massachusetts, June 24-28 2007

Paper 2030_-.-I- l lithium Oltlet

NaKlGN2 heal exchangerFigure 6,

_........: ~-~ =

~--"' -. .-"~

.~ '__ -...JIFigure 4. Thermal simulalor

Figure 5. Power zones

W.B. Heat Exchanger

The thermal simulalors are arranged in zones thaI areconnected 10 individual power supplies (one power supplyper zone). Each power supply is raled for 15 kW of DCpower delivered via 150 V at 100 A maximum. The 37­pin core assembly is divided inlO four control zones asdepicted in Figure 5. The grouping of healers in thesezones (series/parallel) is laid oul such thaI the equivalenlresistance attempts 10 maximize the power supply outpulcapability. In this configuration, a maximum of 44 kWcan be trnnsferred 10 the healers al 700°C. If additionalpower is required (10 simulale transients), the number ofzones could be increased 10 a maximum of twelve. Thiswould resull in 180 kW of available power.

( I)F=JxB,

where F is force, J represents the electric field and Brepresents the magnetic field. In this pump, nickel busbars have been brazed 10 a flattened section of the stainless

Electromagnetic pumps operale on the principle ofLorentz forces, or

Figure 7. Electromagnetic pump

Ill.C. Electromagnetic Pump

NaK is circulaled around the flow path by means of anelectromagnetic pump, which was selected 10 maintain theinlegrity of the all-welded flow circuil with no movingparts and 10 build experience with a pump type thaI isapplicable 10 a space power system. A general produCIsearch and open procuremenl in the commercial markelresulled in a single pump submission thaI met the flow,lemperature, and pressure requirements. This is a Style­VI two-stage AC conduction unit with a stainless sleel316duet originaUy markeled by Mine Safety Appliance andcurrently offered by Creative Engineers Inc. (design dalesback 10 the 1950's). Figure 7 shows the pump system(prolective shielding removed) before integration inlo thepump housing. It provides continuous operation allemperatures up 10 816°C, with flow control from 10"10 10100% (control provided by a variable trnnsformer).

Zone 1 - 7 Heater ElementsZone 2 - 12 Heater ElementsZone 3 - 12 Heater ElementsZone 4 - 6 Heater Elements

•••••••••••••••••••••••••••••••••••••

Once the NaK is heated by the core assembly il passes inloa heat exchanger 10 remove up 10 40 kWt. The baselinefor this heal exchanger was a "battleship" or facility-styledesign thaI provides significanl robustness in this initialflow circuit. It is a 0.6 m long counler-flow design withNaK confined by the ouler jackel and the secondarygaseous nitrogen coolanl flow confined by 107 lubes (0.8ern outside diameler) thaI pass through the NaK flow pool.The exchanger is equipped with temperature and pressuremeasurements 10 monilor boundary and fluid conditions.Nitrogen flow rales in excess of 0.4 kg/sec can beachieved, and a gas prehealer raises the gas inIellemperature up 10 490°C. The healed exhaust gas isvenled 10 the atmosphere.

steel pump duct. NaK is a conductor of electricity, sowhen power is applied to the pump, current flows fromone bus bar to the other through the NaK-filled duct. Atthe same time a magnetic flux is generated by low­temperature commercial windings on two U-shapedmagnetic iron cores (perpendicular to the bus bars). Theinteractions of the electric and magnetic fields results inthe exertion of force on the NaK and causes fluid motion.The fields have been aligned such that the maximumpossible force is developed in the direction of the flowpath (per the right hand rule). The commercial magnetcoils are temperature-limited, and operation of the pumpat the upper end of its range or at elevated NaKtemperatures results in the gradual heating of the pumpmechanics. The test article is operated under externalvacuum pressure conditions of 10,3 torr or lower,eliminating natural convective cooling. The pump hasbeen installed in a stainless steel enclosure with a gaseousnitrogen cooling purge to prevent overheating.

IIJ.D. Reservoirs

The test article contains two reservoirs. The lower(load/drain) reservoir serves as a storage tank for NaKbetween tests and was used as the receptacle during thefirst Jill of the circuit. It is separated from the flow pathby means of a remotely-{)perated valve. During NaKtransfer operations, applying argon pressure to the surfaceof the liquid metal forces the NaK up and into the tubes ofthe circuit via a dip tube, provided that the pressure in thelower reservoir is higher than the pressure in the circuit.This tube extends to 0.32 em above the bottom of thereservoir. The applied pressure need only be very low ­just two or three psi greater than circuit pressure - toachieve a controiled NaK transfer.

Fignre 8. Lower reservoir

The function of the upper (expansion) reservoir is tocontain the expansion of NaK as it is heated At 650°C,NaK will expand to 17% above its volume at 25°C. Theupper reservoir has been sized to accommodate theincrease in volume of the entire circuit at 650°C withroom to spare. This reservoir is also used as the

Proceedings 01 Space Nuclear Conlerence 2007Boston, Massachusetts, June 24-28 2007

Paper 2030

argonlvacuum interface with the closed flow path. Duringtesting, a slight argon pressure is applied here to providethe pump with head pressure (and prevent NaK fromsimply oscillating back and forth in the empty volumerather than circulating).

lll.E. Instrumentotion

Approximately 75 rype-K thennocouples (rCs) are spot­welded directly to the stainless steel circuit, encompassingevery component save the electromagnetic pump housingand support structure. TCs are spaced approximatelyeight inches apart on each section of tubing. Thethermocouple wire is a high-temperature, glass-jacketedvariety capable of operation up to 704cC. Thisinstrumentation not only tho.roughly monitors thetemperature of aU circuit components but can also serve asa rough backup for other instrumentation, such as thelevel sensors.

There are six pressure transducers on the NaK flow path:at the inlet and exit of the pump, core, and heatexchanger. Single transducers are located at both theexpansion and fill/drain reservoir to monitor argon gaspressure, and multiple pressure sensors are employed onthe nitrogen line that feeds the heat exchanger. The sixsensors on the NaK flow path measure a range of 0-75psia, and aU wetted parts are stainless steel. They aretemperature compensated and must remain below 162°Cto operate correctly. Each of these particular transducersis set on an 18 em standoff to prevent overheating;however, when the circuit is operating at hightemperatures (above approximately 450°C), the excessiveradiation from the uninsulated circuit can cause them tooverheat. The circuit is scheduled to be insulated in thenear future, which should drasticaUy reduce heat lossduring testing and hence prevem transducer overheating.

NaK level measurement is required in both the upper andlower reservoirs. This is accomplished using powerfeedthroughs that terminate in weld lips that are welded toVCR fittings. These are then mated to complementaryfittings on the reservoirs. The level sensors work bycompleting a circuit (resistance measurement) when thelevel of the NaK comes into contact with the stainless steelpin of the feedthrough. As shown in Fignre 9, the sensorarray in the lower reservoir is designed to indicate a rangeof discrete volumes. The sensor heights were chosenstrategicaUy.

LS1 LS2 LS3 LS4 LS5 LS6

Figure 9. Level sensors in lower reservoir

IV SIMULATION

A model of the FSP-PTC has been constructed using theGeneralized Fluid System Simulation Program (GFSSP).This is a general-purpose, NASA-developed program foranalyzing steady-state and transient flow rates, pressures,temperatures, and concentrations in a complex flownetwork. It is capable of modeling phase changes,compressibility, mixture thermodynamics and externalbody forces (such as gravity and centrifugal forces). Theprogram contains subroutines for computing "real fluid"thermodynamic and thermophysical properties for twelvefluids. Other, less commonly used fluids (such as NaK)can be added by the user, so long as the requiredproperties are known. User-written subroutines may alsobe incorporated within the software. As an example, thepump performance data provided by the vendor has beenincluded in this model as a user subroutine. When a fullset of experimental pump performance data has beengathered, it can be tabulated and used in place of thevendor data.

Figure 10 shows the graphical representation of theGFSSP model of the FSP-PTC. The numbered boxes arenodes, which are connected by branches. The userspecifies what type of object lies between each node~tubing, orifice, etc.). The working fluid in the closed loopIS NaK, and the fluid in the open loop is gaseous nitrogen.The Pum~' reactor core, heat exchanger, and tubing are allmcluded m the open loop. The boundary conditions thatmay be set on the actual system are the same as those thatmay be set within the model. GN2 inlet conditions at theheat exc~ger (temperature and pressure),electromagnebc pump voltage, and applied core power areall set prior to running the simulation. The boundaryconditions at the exit of the heat exchanger open loop havebeen set to ambient.

Proceedings of Space Nuclear Conference 2007Boston, Massachusetts, June 24-28 2007

Paper 2030

III =€ -8'-::::-[;] e -[;] - -:d"~i--[;] - :;]~~~- •• •• - ~I~-~'-t Pump Input = 140 v ~ "'

_ltZI NaK Flow RlJt:t: _ 1.06 kg/s • • Core Q.ltJel

ric, tlilrogen Row RaIl!. 0.14 kgJs __ ~r une J

--, Tolz'System .......... Drop-5.19psl IfJ ~_ 'i"~_ Core Inlet Temperatl.re = 594 ·C t;:o....~- ';:0.l1l:I-'1-~

Core Outlet Tempcrilture • 62"J OC 0 0 ."1"".....- I i- $

/8 -<;orelnlet ! 'f •. ,.. "~.... 0- . :til ... HXlnlet

~---J::~e>--0~M---0-.....--.B~---4I'"~ -'/ .. - " ."

Pumo HX Ovttet

Figure 10. GFSSP model graphical representation

GFSSP employs a finite volume formation of massmomentum, and energy conservation equations i~conjunction with the thermodynamic equations of state forreal fluids. Mass, energy, and species conservationequations are solved at the nodes while momentumconservation equations are solved in the branches. Thesystem of equations describing the fluid network is solvedby a hybrid numerical method that is a combination of theNewton-Raphson and successive substitution methods4

,5.

The original purpose of this model was to calculate thepressure and temperature distribution in a liquid metalcooled reactor system and thereby assist in the placementof instrumentation. The model performed this taskadmirably, but now that testing has begun, a fewrefinements are being considered. At present, heat isadded to the model by directly applying it at a single nodewithin the core. A thermal simulator could beincorporated into the overall model. More important thanthis, however, is the need to better match the simulationwith the actual thermal environment in the chamber. AsSection VB explains, a good deal of the power applied tothe core is lost to the chamber walls through radiation (thecircuit is uninsulated). GFSSP is overpredicting the NaKflow rate at a given D.P by nearly twice the experimentalvalue, which is almost certainly due to the lack of real heatlosses in the model. Simulation results can be bettercompared to actual test data when l) radiation losses areincluded in the model, or 2) the circuit is insulated. Aswas previously established, preparations to insulate thetest article are underway.

V TESTING

A series of experiments are being conducted on the FSP­PTC, beginning with the execution of a pump~rformance test matrix. The results will be comparedWith performance data provided by the vendor. Following

the completion of this stage, a series of integrated systemperfonnance tests will be performed, which (inconjunction with the pump perfonnance data) can be usedto validate the GFFSP model. Finally a series of transientcases will be examined, at which point the team will beready to incorporate new hardware components into thesystem via the test section. Before the test section can beremoved, the interior must be cleaned of NaK. To thisend, the team has procured an alkali metaI cleaningsystem from Creative Engineers, Inc., which mixes drysteam with the NaK-wetted interior of the circuit to formhydroxides. The machinery may also be used to react thebulk of the NaK, which would be done within theload/drain reservoir. Once the circuit has been cleaned thetest section may be removed at its mechanical connectionpoints. The hardware components being considered foruse includes a Stirling engine heat exchanger or aninduction-style liquid metal pump. A second, smaller testarticle designed to characterize the flow of NaK around asingle resistance heater may also be added to the testsection.

V.A. Pump Performance

The pump performance test matrix is intended tocharacterize the developed NaK pressure and flow rate asa function of pump power over a range of NaKtemperatures. The aforementioned variable transformerallows the test engineer to speci.IY precisely how muchvoltage is applied to the pump, with 235 V being themaximum. Table I shows the data points to be tested;boxes in yellow indicate data that have alreadY beenobtained. The matrix is designed to be run one row (NaKtemperature) at a time. The numbers in each box indicatewhich pump setting is to be run first, second, etc. at thegiven temperature. This introduces a degree ofrandomness into the proceedings and helps to preventunintentional trending due to a linear progression of datapoints. In Table I, "NaK Flow Temp" is the temperatureof the NaK within the liquid metal pump. Because thecircuit is not insulated and the heat exchanger liesbetween the core and the pump, there is some differencebetween the NaK temperatures at these locations.

Proceedings of Space Nuclear Conference 2007Boston, Massachusetts, June 24-28 2007

Paper 2030

TABLE I

Pump Perfonnance Test Matrix

EMPum Vortaae100v 140V 200V 235 V

35O"C 4 2 3 137S'C 3 1 4 2400"C 1 3 2 4

NaK 425'C 2 4 1 3Flow 450"C 4 2 3 1

Temp 475'C 3 1 4 2500"C 1 3 2 4525'C 2 4 1 3538"C 4 2 3 1

In a typical pump perfonnance test, the desired NaKtemperature is selected first. This can be any of thetemperatures listed along the vertical in Table I. GFSSPis used to generate a set of ballpark system settings neededto bring the NaK up to this temperature (applied corepower, GN2 temperature at the inlet of the heat exchanger,and GN2 flow rate). It is undesirable to rapidly apply agreat deal of power to a cold core, and gaseous nitrogencannot be heated instantaneously, so these settings areincreased over a span of time until the final values arereached. (For example, power is applied to the core at amaximum rate of I kW/min). The pump voltage is set towhichever value the test matrix dictates should be first forthe given NaK temperature. Core power and heatexchanger GN2 settings are adjusted until the NaKtemperature appears to be settling at approximately thecorrect temperature. The NaK temperature is allowed tocome to steadY state, at which point the pump voltage ischanged to the second proscnbed value. This procedure isfollowed until all data points in the row have beenobtained (or until the test must be shut down, usually dueto time constraints). The system is declared to havereached steady state when the NaK temperature at a givenlocation has changed by no more than 4·C over a span ofone hour. To detennine the veracity of this definition, asimple test was run in which the test article was allowed torun at a single set of conditions for more than six hours.Figure 11 shows the NaK temperature at various locationsaround the circuit, the NaK flow rate, and pump voltageduring this test. The sudden change in the NaK flow rate(represented by the blue line) corresponds to anunanticipated 'blip' in the power that was supplied to theEM pump.

Proceedings of Space Nuclear Cooference 2007Bastoo, Massachusetts, June 24-28 2007

Paper 2030

~"'_"TIoOll:~

Figure II. NaK temperature, flow rate and pumpvoltage during steady state test

".---6 • ._-..- '-~ --. -- "- -- --.....-._-._--_. -.o--_-~-_-_-_-_---~

o 5 10 15 ~ ~ • ~ ~

Aow~(GPIIIII)

Figure 12. Pump performance data

Pump performance data is plotted in Figure 12, with NaKflow rate as measured by the liquid metal flowmeter on thex-axis and the pressure developed by the liquid metalpump on the y-axis. Data provided by the pump vendor,which was taken experimentally in the 1970s, IS

represented by the continuous lines. Since this is an ACelectromagnetic pump, the applied power represented bythese lines must be assessed carefully. However, due to thelocation where the current and voltage were measured inthe 1970s tests, the VI approximation can be made for thepurposes of comparison with FSP-PTC test data. InFigure 12, FSP-PTC data points have been grouped byapplied pump voltage, with the average applied pumppower for each group also indicated. Pump performancedata for each of these power levels was extrapolated fromthe vendor's curves and are represented by the dashedlines. (Note that 3159 W is the average pump power ofallthe data points in the upper right group.)

The NaK flow rate and developed pressure increase in aroughly parabolic fashion as applied pump power isincreased. (The two 200 V data points that are not locatednear their main group will be discussed later in thissection.) As was previously stated, the real pump powerfor each of the vendor curves is only approximated fromthe given current and voltage; therefore, the test engineershave been matching the voltages from each of thesecurves. To illustrate: the applied current for the group of140 V data points (green squares) was approximately 7.8amps as opposed to the 12 amps shown on themanufacturer's curve. This group had an average pumppower of 1134 W and logically falls below the 140 V, 12 A(-1680 W) line. When the extrapolated 1134 W line isadded to the graph, the data points taken at this powerlevel fall very close to it The same is true for the 100 Vgroup and most of the 200 V group. It should also benoted that the vendor data were generated at a NaKtemperature of 566°C. The maximum temperature thatcan currently be attained in the FSP-PTC is 527°C. This

2lleV,1''''(MOOW)

•

..,

..,-- -

~ / •If f/ •T~ / •

NilKAowRln •.111 / •

......"'- ••

•~... ,... ,... - ,... •

-

Data is averaged over the length of the steadY statewindows to produce the data points used in the pumpperformance graphs below (Figures 12 and 13). For threeof these thirteen data points, steadY state data was notavailable for a full hour. In those cases the average wastaken over the span of time for which the temperature atthe core exit did oot vary more than 4°C (the time spanwas at least 2500 seconds for all three points). Noaverages were taken when greater temperature transientswere observed.

..

As Figure II indicates, NaK temperatures stabilize fairlyquickly once final core power and pump voltage levelshave been reached. Data from all tests have shown thatswface temperatures around the circuit level off together,so when the temperature at one location has come tosteadY state, all NaK temperatures can be assumed to be atsteadY state as well. In the all-{!ay steadY state test, theNaK flow rate did not stabilize as quickly as thetemperatures did. NaK pressures, not shown on thisgraph, also took longer than the temperatures to stabilizealthough not nearly as long as the flow rate. Due to timeconstraints on the project, it is not possible for the testengineers to wait until the all these values have come tosteadY state before taking a data point Doing so wouldlimit the team to collecting ouly one data point per day.For the purposes of these tests, it has been determined thatthe pressures and temperatures are sufficiently stable towarrant declaring "steadY state" when the aforementionedconditions have been met

Proceedings or Space Nuclear Conrerence 2007Boston, Massachusetts, June 24-28 2007

Paper 2030

VOlumetricFlowRate(mls') x IiP(Pa)EfJ(%) = (2)

PumpPower(W)

may result in some differences between the FSP-PTC dataand the vendor curves. One would expect that the datawould not fall directly on the curves due to theapproximations and differences described here.

Figure 13 is a plot of developed NaK pressure versuspump efficiency. To illustrate the effect of NaKtemperature on pump performance, the data points havebeen grouped by temperature rather than appliedvoltage/power. Pump efficiency is calculated per thefollowing equation:

(4)

(3)

Q.- is comprised of energy losses due to conduction andradiation, but conduction has been omitted from thefollowing assessment for several reasons. First, the flowcircuit is supported by a steel structure, but the connectionpoints are few and cover a small surface area. Second,there are no thermocouples on the support structure tomeasure the .1.T needed for conduction calculations.Finally, as temperature increases, the losses due to

In a well insulated heat exchanger the energy coming inwould be equal to that removed. Since the FSP-PTC is notinsulated, the energy input to the system is equal to theenergy removed by the heat exchanger plus thermal losses:

V.B. Power Balance

The execution of the pump performance matrix providesan excellent opportunity to evaluate the liqnid metalflowmeter, which is difficult to calibrate. The only way todo this is to pass a known volume of NaK through it overa set amount of time. This has not been done for thisparticular piece of eqnipment; however, the vendor hasprovided a correlation based on previous flowmeterexperience. A power balance across the heat exchangercan be used to get an idea of whether or not the valuesbeing reported by the flowmeter are in the right ballpark.NaK flow rate can be determined through use of the heatexchanger and the following relationship:

It is difficult to determine just how much of the separationis due to temperature-improved efficiency as opposed tothe increase in overall power draw (and also since themost efficient 200 V data point was not the warmest of thefive 200 V points). Additionally, it is not known preciselyhow long it takes for the pump to "warm up"; the nitrogengas that flows through the pump housing is cold, and willvie with the heated NaK to affect the overall pumptemperature. Still, Figure 13 does appear to indicate ageneral trend toward higher efficiencies as temperatureincreases in each group of points. In the case of the datapoints taken at 235 V, developed pressure rises withincreasing NaK temperature, but pump efficiency does notincrease a great deal. This seems to suggest that the pumpis already operating at near-best efficiency when themaximum possible power is applied, leaving less room foran increase in performance as NaK temperature rises.However, only three data points have been taken at thisvoltage setting, and more data will help paint a clearerpicture of pump behavior at 235 V.

0001 00 200 3.00 ..00 5.00-_ ....Figure 13. Pump efficiency

230V

I _~.

2OJ~VY.~-----

T_,"

-"'-"'VC_ ~ Ell pulllp.210 - 2IlO"C _.:DJ.35D'C

'OOV m·3IillS"C- ••10·445"C -e«lO·15lXl"C ,

•

,.00

where M' is the NaK pressure rise and pump power is thewattage that is applied for pump operations. The data inFigure 13 clearly shows that the pump performs moreefficiently when operating near its maximum rating. InFigure 12, attention was drawn to two 200 V data points(red diamonds) that fell outside of their main grouping.At these particular points the temperatures of the NaK inthe pump were 428°C and 493°C, respectively - hotterthan the rest of the points taken at this voltage. Notemperature data is taken on the pump mechanicsthemselves (most temperature measurements within thepump housing have proven to be either noisy orunreliable), but test data shows that the pump was drawinga higher current at these two points than at the other three,resulting in a higher overall applied pump power.

The vendor has stated that the electromagnetic pump willoperate more efficiently at NaK temperatures greater than427°C. (At 358°C and below, nickel has a ferromagneticquality" which can negate some of the flux generated bythe pump's electromagnetic windings.) The twooutstanding 200 V points were taken above this NaKtemperature threshold, while the other three were below it.

,.,.

.."

'. Proceedings of Space Nuclear Conference 2007Boston, Massachusetts, June 24-28 2007

Paper 2030

radiation far outweigh conductive losses due to the 4tb

power temperature factor in Eq. (5):

QN;"_o.O," Q.,K.OulFigure 14. Heat exchanger power balance

VI. CONCLUSIONS

Figure 15 shows power and temperature data from onepump performance test. The maximum power applied tothe core was 25 kW, which brought the NaK temperatureat the exit of the core (the holiest location in the circuitsave within the core itsel/) up to 527°C. 11 is inunediatelyobvious that there is a significant thermal loss in thesystem, represented by the gap between the power appliedto the core (dark blue) and the power removed by the heatexchanger (yellow). At 15000 seconds, when the NaKtemperature had reached steady state, the differencebetween the two is almost 11 kW, and the flowmeter wasreporting a NaK flow rate of 13.36 GPM (0.64 kg/sec).Using an emissivity of 0.4' (oxidized stainless steel at 800K) an estimated chamber wall temperature of 25°C and ahe:.t exchanger surface area of 0.2917 m2

, Eq. (5) yieldsthe energy lost by radiation at the heat exchanger as 258 IW. Using the experimental data for nitrogen ten:peraturesat the inlet and outlet, calculated specific heats , and themeasured nitrogen flow rate, the energy removed by theheat exchanger is found to be 15108 W. When these twolosses, NaK temperatures and their respective specificheats' are incorporated into Eq. (6), the calculated NaKflow rate is 13.89 GPM (0.66 kg/sec). Considering thatconductive heat losses have been omilled, the emissivity ofstainless steel (at this temperature) ranges from 0.24 to0.67', and other approximations, it appears that theflowmeter is indeed reporting reasonable values.lnsulating the circuit will improve this assessment byreducing the uncertainty contribution associated with theradiation losses. Steps are also being taken to acqwretemperature data on the test article support structure,which will allow for the quantification of heat loss byconduction.

(5)

QNitrogen,lnQ ••dQN3K,In

Let us consider the heat exchanger as a control volume(Figure 14). All energy that enters the volume mnst alsoleave the volume (if there is no energy storage). As Eq.(4) states, energy enters the heat exchanger via heatedNaK and is dissipated through the nitrogen flow andradiation. The inlet and outlet temperature of the gas andthe gas flow rate are measured, and the specific heat isknown for the gas. For simplicity, the swface temperatureof the heat exchanger is taken as the average of the NaKinlet and exit temperatures. NaK mass flow rate isconstant throughout the circuit. Therefore,ifQ, = Q.J the NaK mass flow rate can beo.ues r.... •

determined from Eq. (6):

E is the emissivity of stainless steel, (J is Bolzmann'sconstant (5.67xlO" W/m-K), A is the radiative swfacearea, T. is the surface temperature of the test article andT~ is the temperature of the vacuum chamber wall.

Figure 15. Power and temperature, 527°C test

Testing has begun on the Fission Surface Power PrimaryTest Circuit. A maximum NaK temperature of 527°C hasbeen reached with flow rates in excess of 1.1 kg/secreported. The first set of data points from the pumpperformance test matrix indicate that the liquid metalpump operates in the approximately 1.5 - 5% efficiencyrange. Efficiency improves as greater power is applied tothe pump and also with increasing NaK temperature,although the laller effect is not evident at low pumpvoltages. Thus far, results indicate that the increase inefficiency does occur near the vendor-reported temperatureof 427°C. This matrix is not yet complete, and it is hopedthat more tests will establish a beller understanding ofpump performance at the highest possible applied powerlevels. A basic power balance indicates that the liquidmetal flowmeter, which has not been calibrated, isreporting reasonable values. It is expected that the flow

(6)

....""E

•"I

I..-

mON2Cp.GN2 ATGN2 + sa4V" 4 -Tmr4)

Cp.NaK ATNaK

I Apped "'-t 1Nc:rn}-Ccn~<C*)

jr Hlt"-r<C*)HIIK TIIII\p G 0DnI Ea:

J oJ

d\YJY ~

,I'"o '7 ... ,... ,... .... -

0'--

mNaK

'"

.,

rates reported by the meter will better match the calculatedvalues once the circuit has been insulated. Insulating thecircuit will also prevent pressure transducer overheatingand will enable the test engineers to begin comparing testdalJl against the results of the GFSSP simulation. Uponcompletion of the pump performance matrix, integratedsystem and transient cases will be performed, followed bythe eventual inclusion of new hardware components in thetest section.

ACKNOWLEDGMENTS

The author would like to acknowledge the contributions ofThomas Godfmy, Roger Harper, Dr. A10k Majumdar, SIJInMcDonald, Gene Fant, Jason Berry, Dr. Kenneth Webster,James Martin, Melissa v.m Dyke, and Dr. ShannonBragg-Sitton for their many hours ofeffort on this project.

The work described within this report was supported byNASA, in whole or part, as part of the administration'stechnology development and evaluation activities. Anyopinions expressed are those of the author and do notnecessarily reflect the views of NASA.

NOMENCLATURE

Proceedings of Space Nuclear Conference 2007Boston. Massachusetts. June 24-28 2007

Paper 2030

V 'vUltsW Watts

REFERENCES

I. Anne E. Garber and Thomas 1. Godfroy, "Design,Fabrication and Integration ofa NaK-Cooled Circuit",Paper 6331, Proceedings ofICAPP 2006, Reno, NY

2. David I. Poston, "A 100-kWt NaK-Cooled SpaceReactor Concept for an Early-Flight Mission", SpaceTechnology and Applications International Forum ­STAIF 2003, American Institute of Physics (2003).

3. S.M. Bragg-Sitton, J. Farmer, D. Dixon, et ai,"Development of High Fidelity, Fuel-Like ThermalSimulators for Non-Nuclear Testing", SpaceTechnology and Applications International Forum ­SDUF 2007, American Institute of Physics (2007).

4. Majumdar, A. K., "A Generalized Fluid SystemSimulation Program to Model Flow Distribution inFluid Networks", Paper no. AIAA 98-3682, 34'"AIAAIASMEISAEJASEE Joint PropulsionConference and Exhibit (1999).

AACB°CCp

ccemDCFGN,GPMJKkgkWkWtmtilNaKPpsipsiaQssecTTC

AmpsAlternating currentMagnetic fieldDegrees CelsiusSpecific heatCubic centimetersCentimetersDirect currentForceGaseous nitrogenGallons per minuteElectric fieldKelvinKilogramsKilowattsKilowatts thermalMetersMass flow rateSodium-polJlssiumPressurePounds per square inchPounds per square inch, absoluteEnergySecondsSecondsTemperatureThermocouple

5. Majumdar, AX., "A Second Law Based UnstructuredFinite 'vUlume Procedure for Generalized FlowSimulation", Paper no. AIAA 99-0934, 37'" AIAAAerospace Sciences Meeting Conference and Exhibit(1999).

6. Davis, J.R, ASM Specialty Handbook: Nickel, Cobaltand Their Alloys, ASM International, Materials Park,OH (2000).

7. F. P. Incropera, D. P. DeWitt, Fundamentols of Heatand Mass Transfer, Third Edition, Chapter 12, JohnWLley & Sons, New York (1990).

8. Cenge!, Y.A., Boles, M.A., Thermodynamics: AnEngineering Approach, p. 764-765, McGraw-HilI,New York (1989).

9. Burdi, G F., SNAP Technology Handbook /-Vlume I,Liquid Metals, Atomic International Report No. NAA­SR-8617 (August, 1964).