e S e A B e TR }) "fié; I ot [ g, e e it s e S VBEACTOR POWER ME!ASUBEMENT AN.D HEAT TRANSFER IERFORMANCE '_OAKRIDGE NATIONAL LABORATORY - opersted by SR o UNION CARBIDE CORPORATlON o i ‘..j':”“’*:'NUCLEAR DIVISlON o - s for the U S ATOMIC ENERGY COMMISSION | L _,"ORNL TM-'3002 | : = o ”-.:COPYNO -109 :__"'-IJDATE-May 1970 o m THE MOL'I‘EN SALT REACTOR EXIERDENT AL - . Gabbara flUT'CE Tl-us document 'ccntmns” mformahon of a prelimmury ‘nature R and was prepored ‘primarily ‘for internal use at the Oak Ridge Nafionol - Laboratory. It is. subject"to. revision or . correction und therefore does ~ not reprcsent a fmul reporf : - LI R . : ms'rth}TfiN '6-‘:",'??“_5.".)'09-[], MENT I8 UNLTIIRED LEGAL NOTICE . This report was prepared as an account of-Gciavemmo;nr sponsored work. Neither the United S"tuf;s:,- nor the Commission, nor any person acting on behalf of the Commission: ’ ‘A. Makes any warranty or representation, sxpressed or implied, with respect to the accuracy, -~ completeness, or usefulness of the information contained in this report, or that the use of- any information, apparatus, mefhod or process disclosed in this report may not infrings privately owned rights; or “ B. Assumes any liabilities with respcct to the use of or for damages resulting from the use of' - any information, apparatus, method, or procoss disclosed in this report. '_As used in the above, “"person acting on behalf of the Commission®® includes any employee or - contractor of the Commission, or employse of such contracter, to the extent that such employes or contractor of the Commissnon, or cmployoe of such contractor prepares, disseminates, or provides access to, any information pursuant to hu employment or contract with the Commission, or his employment with such contractor., . [] (» 1 0 ) iii CONTENTS ABSTRACT . . . ¢ & v s et v v ettt s e e o s e e INTRODUCTION . . &« ¢« ¢ ¢ v 4 o ¢ o o o o o« MEASUREMENT OF REACTOR POWER . + v ¢ & o o o « & Normal‘Heat Balance Calculation s e e e v e e Coolant Salt Flow'Measurement'by Decay of Circulating Activation Products. Coolant System Pressure Drop . . . . . Air Heat Balance Calculation. . . . . . . . . . PERFORMANCE OF THE MAIN HEAT EXCHANGER AND RADIATOR. Design Review e e . . Primary Heat Exchanger . . . . Coolant Radiator . . . . . . . | Analysis of PerfOrmance . . . o« « o o o o . . . Primary Heat Exchanger . . . . . . . . Coolant Radistor . . . « « . . . . . . CONCIUSIONS AND RECOMMENDATIONS. . . . . . . . . REFERENCES . . . . . . o o o v v v v v s v v 0 0 s DISTRIB 10 15 16 16 19 20 20 2 27 29 UTION OF THIS DOCUMENT 1S UN LIMITER L £) e ol ) g e REACTOR POWER MEASUREMENT AND HEAT TRANSFER PERFORMANCE IN THE MOLTEN SAIT REACTOR EXPERIMENT C. H. Gabbard - ABSTRACT ‘The operating power of the MSKE was routinely determined by a heat balance on the fuel and coolant salt systems performed by the on-line com= puter. This gave a calculated full-power level of 8.0 MW. However, changes in the isotopic composition-of uranium and plutonium in the fuel salt indi- ~ cated a power lower than the 8 MW by about 7 - 10%. Attempts to resolve this discrepancy included a measurement of the coolant salt flow rate by " the radioactive decay of activation products in the coolant salt, a recal- culation of the coolant system Pressure drop, and a heat balance on the air 'side of the coolant radiator. These efforts to date have been inconclusive, ‘but the coolant salt flow rate was found to be the only potential source of significant error in the heat balance. A calibration check of the ‘differential pressure cells reading the coolant flow venturi will be made . during the scheduled post-operation examinations. ;s " The heat-removal capabilities of the fuel salt to coolant salt heat y ekchanger and coolant salt to air radiator were below the predictions of the original design calculations and limited the full-power output of the MSRE., In the case of the primary heat exchanger, the overestimate was due to the use of erroneous, estimated physical property data, the thermal ’ conductivities in particular, for the fuel and coolant salts. When ac- curate, measured values of physical properties were used with the heat transfer relationships for conventional fluids, the calculated performance of the primary heat exchanger agreed with the observed value. In the case of the radiator, the overestimate in the design was only partially ex- plalned by the fimproper selection of an air "film" temperature. ' There was no decrease in heat ‘transfer capabllity of . the two. heat Vrexchangers over more. than 3 years of operation. Keywords- . MSRE, heat balance, heat transfer, heat exchanger, o ffised salts, performance, operation, reactor, INTRODUCTION Operation of the Molten Salt Reactor Experiment (MSRE) from 1966 through 1969 provided a unique opportunity to measure heat transfer in equipment using molten fluoride salts in the temperature range of 100014 1200°F over an extended period of time, Analysis of these data has led to conclusions regarding the adequacy of conventional design. procedures and the change (or lack thereof) in heat transfer resistances in this Eg molten-salt system over more than 3 years of operation., This report. describes the problems associated with measuring the power of the MSRE 'gthen deals with predicted and observed heat transfer coefficients. Qtn; | The MSRE design and operation are described in detail 1n References 1 and 2. Figure 1 shows the layout of the important components. Fuel salt was - circulated at about 1200 gpm through the core, where 1t was heated by the fission chain reaction, then through a 279~ ft2, cross-baffled shell- ' and-tube heat exchanger where it transferred heat to & second salt flowing through 1/2-inch tubes. Heat was removed from the coolent salt and dissi- pated to the atmosphere in an air-cooled heat exchanger ("coolant radlator") ~ with 3/h—inch tubes . MEASUREMENT OF REACTOR POWER The official operating power of the MSKE was determined by meking a heat balance around the fuel and coolant systems. This heat balance,‘ which vas routinely calculated by the on-line computer, is more fully described in References 3 and 4, The various nuclear power instrumen- tation systems {1linear chambers,-fission chambers,'and safety chambers) vere calibrated to agree with the nuclear power &s indlcated by the heat_ balance. . ' The heat balances calculated through March 1968 indicated 8 nominal full-power level of gbout 7.2 MW. There was some reason to suspect this value,,however, because data from reactor operation at different power levels strongly suggested that the coolant salt specific heat was a con- stantrrather than the temperature-dependent relastion that was?being usedh (, ) » ' o n - e ORNL-DWG 63-1209R REMOTE MAINTENANG CONTROL ROOM T ; ) | REACTOR CONTROL ROOM : & | S0y s . 6 ; y ‘3 ; - 3 : ;" - —.-';E#_‘--_ 2 ; T _ . | ‘ | 1 . = ‘. Vo i 8 = 1. REACTOR VESSEL 7 RADIATOR 77777 A 2. HEAT EXCHANGER 8. COOLANT DRAIN TANK N 3. FUEL PUMP 9. FANS " 5\ 4. FREEZE FLANGE 10. DRAIN TANKS 2 ‘ 5. THERMAL SHIELD 1. FLUSH TANK N 6. COOLANT PUMP 12. CONTAINMENT VESSEL Fig. 1. Layout of the MSRE . FREEZE VALVE in the heat balance.5 Two leboratories (ORNL and the National Bureau of 'Standards) independently measured the coolant salt specific heat, arriving at values in good agreement with each other but substantially higher (in the MSRE temperature range) than the previously used value.® The new . measurements also showed virtually no variation with temperature. The new, constant value of the specific heat was incorporated into the computer Prior to the beginning of operation on U-233 fuel in January 1969 The calculated full-power level was changed from 7.2 to 8.0 MW as a result of the specific heat revision._', ' | As results of precision isotopic analyses of the heavy elements in the fuel salt became available, independent determinations of reactor power were obtained from the'changes in uranium and plutonium isotopic ‘jcompositions. The most recent evaluations of these isotopic data yielded A full-power level of T7.34 + 0.09 MW (References 7 and 8). The heat balance calculation has been reviewed as described below in attempts to resolve the discrepancy between the measured heat balance power'and the pover indicated by the changes in isotopic composition of the fuel salt. Normal Heat Balance Calculation At significant power levels, the dominant term in the heat balance was the heat removed from thecoolant salt at the radistor. From Table I, which shows the relative importance of the various terms in the heat_r balance for operation at full power, it is clear that there was little opportunity for significant overall error due to the other terms; ' | The heat removed at the radiator was calculated from the mass flow rate, the speciflc heat of the coolant salt, and the temperature drop across the radiator Possible sources of error in each of these were examined | : _ The salt temperature drop across the radiator was measured at thermo- couple wells at the inlet and outlet. Three calibrated-thermocouples were installed in each well with two from each well being used in the heat balance. Although the laboratory calibrations of the' thermocouples indj- cated a AT error of about -O,3°F,.there was concern that larger systematic ») »n &) 0 ®) Table T Typical Values in MSRE Heat Balance at Full Power MW Heat removed from coolant salt at radiator ~ T.853 Heat removed by cooling water ‘ - 0.341 ~Heat removed by component cooling air 0.016 Heat removed by fuel pump oil N . 0.003 Unaccountable heat losses = = 0.011 - ~ Power input to electric heaters | - =0.175 Power input to fuel pump - | -0.035 .waer-input to space coolers . =0.007 - Power. 1nput to coolant pump impeller =~ -0.036 " Nuclear power generated i'i B - T.971 errors might exist in the,installed;condition. To test this possibility, in November, 1969 the reactor system was operated isothermally at 1210°F, . 1070°F, and 1010°F to. determine the error that would actually occur in .'ioperation over the . temperature range of the coolant system., The indi- ,tcated AT error for the coolant system at full power was +0. 23 B which 'would cause a 0.h4 overestimate in the calculated power.. A second test %o determine the 1nfluence of the radiator air flow on the thermocouple ”’reedings showed no detectable effect " The coolant salt den51ty is believed to contribute a ;'l%'nncertainty '1to the heat removal term, and the revised specific’ heat measurement had a stated uncertainty of +1, h% - The only remalnlng potentlal source of a significant error-is in the tmeasurement of the coolant salt volumetric flow rate. The volume flow ~ rate of the coolant salt -was measured at the radiator inlet by a venturi flow meter with two channels of readout Each readout channel consists of a differential pressure cell and the associated electronics to supply & linear flow signal to the computer.. The differential pressure cells are connected to the venturi pressnre taps and are isolated from the high- | temperature salt by'metal'diaphragm segls and NaKEfilled lines. A review ~of the ventnri manufacturer's calibration data disclosed an error in con- verting»the differential head from water to mercury which,had)causedta -2.9%_reduction in the measured flow rate® and in the radiator heat-removal term. Another possibility for error, which still exists, is in the cali- bration of the differential pressure cells. The flowmeter readings were checked for evidence of trapped gas or compressibility in the NaK?filled lines byfobserving the flow readings as the coolant system,overpressure was increased from 5 to 65 psi during a coolant-system-pressure test. There were essentially no indicated flow changes on either;channel during this test. The actual range calibration of the differential pressure cells cannot be checked until later this year when the coolant piping will be cut so that known pressure signals can be applied to the cells Two independent attempts to determine the coolant salt flow rate are described below. However, the results of these two efforts vere incon- clusive and the final assessment of the flow rate will belmadedfrom the calibration check of the differential pressure cells. | | Taking the nominal full power of 8.0 MW and spplying the 2.9% flow - error and the 0.4% AT error, the heat balance would indicate a power level - of 8.2 + 0.16 MW as compared to 7.34 * .09 MW indicated by the isotopilc analysis of the fuel. If all of thiS’discrepancylwere'asaigned»to the coolant salt flow rate in the heat balance, the flow rate would have to be lowered from the nominal value of 850 gpm to 770 gpm [ Coolant Salt FlOW'Measurement by Decay of Circulating Activation Products o Because the accuracy of the differential pressure cells reading the coolant salt flow venturi could not be checked until some months.after . the end of reactor operation, an attempt was made during the laatpower nm in December 1969 to measure the coolant salt flow rate by the decay of activation products in the salt. Nitrogen-l6 and-fluorine-20vwere produced » ( —r n 8] » in the coolant salt by neutron reactions with fluorine in the heat ex- changer, These activities then decayed with half-lives of T.4 sec and 11.2 sec respectively as the salt was pumped around the coolant loop. We were also hopeful of finding long-lived activities from impurities in the salt that would have been useful in making a geometry calibration of the equipment. A high resolution gemma spectrometer with a 4096-channel analyzer was available for detecting and -counting the various energy peaks that might be present. | | Two holes were drilled in the high-bay floor to the coolant cell so that the coolant salt piping could be scanned at two locations while the reactor was‘operating-at full power, These two locations were separated by & total circulating salt volume of 20.9 £t3 which included the radiator, the coolant pump, and the EOS;line. This volume would give decay times of 1 and 1.5 half-lives respectively for the ®°F and 16N at the design flow rate, Correction factors were estimated to account for the effects of mixing by the side stream through the relatively stagnant coolant pump tank and for the effects of the line-205 flow that bypassed the radiator volume. The effects of possible flow variations through different sections of the radiator tube bundle were found to be negligible.,, Preliminary data showed that the background count was obscuring the count from the coolant plplng Lead bricks were stacked around the de- tector to reduce the background count and the diameter of the collimator used to aim the detector was increased from l/8-in to 1/2-in. to increase the count rate from the coolant line The background remained high com- pared to the count rate from the coolant cell, but the ability to resolve | 'the count rate into discrete energy peaks and the limited time avallable for the experiment led us +o begin the. actual data collection. Although ,“ the background count was higher than desired for the experiment ‘the radi- ation wa.s - below 2 mR/hr and was not a biOlOgical hazard. o Data were collected on magnetic tape for a total active counting - _i_time of. about 11 hours at each of the locations on the ‘coolant piping. . Two sets ofrdata,were-takenaat_the_second-location with_about six inches of lead shielding between the detector and the hole to the coolant cell. The first count measured the background count in the high bay, and the second count was taken with a small S6Co source to calibrate the gamma . energy to the channel number of the analyzer. N o | The data were analyzed by a computer program which gave the integral count rate for each energy-peak in the spectrum; 'The baCRgrofind count appropriate for each Particular peak was. automatically subtracted by the - .computer program. A comparison of the two gamma spectra with and without the lead shielding between the detector and coolant piping indicated that 'only-the-l.63FMev energy peak from 2°F_woold'be useful in calculating the coolant flow rate. None of the other peaks, including two that corre- sponded -to the 6.13 and T.12-Mev gammas from 1°N, were attenuated by the - 8ix inches of lead shielding. These energy peaks that were not attenu- ated were believed to be capture gammas‘from_the'reaotor‘cell'shielding. . The detector was located near the ends of the top shield'blocks'at the southwest corner of the high bay and there was a field of gamma photons ~and fast neutrons from beneath the top shield blocks. - The coolant salt flow rate as calculated from the decay of the 1.63-Mev photons from 2°F was 610 gpm as compared to 850 gpm indicated by the flow venturi. Although'the 610 gmm is probably 20 to 30% below the actual flow and the measurement was not useful for its intended pur- pose of resolving the power discrepancy, this method of7measuring the flow rate appears to be feasible if the proper firecautions are'taken in setting up the experiment., The largest source of'érrOr,in7the-experiment was probably in the geometry differences between the two scanning points. This type error could possibly be eliminated by a more careful design of the counting stations to ensure low background and similar countingige— ometries, or by spiking the cOolanf sélt'fiith a 1onger?lived-activityi* that would be essentially uniform throughout the coolant loop. This activity could then be used to;provide a geometry calibration factor be- :tween the two counting'stations. Other important sources of error could be in the effective salt volume between the two stations or in the effects of the high background. The reactor was shut down -a short time after the data wer; taken and there was no opportunity to refine the experiment. ‘C » . Coolant System Pressure Drop The coolant salt flow rate of 850 gpm measured by the flow venturi was within the range originally predicted from the calculated coolant system pressure drop and the performance characteristics of the coolant pump. A range of 850 to 940 gpm had been originally predicted by allowing + 10% variation on the calculated pressure drop and a *+ 5% flow vari- ation on the coolant pump}: A flow rate 10% below the design range would require a large.error'in the’head'loss’calculation Oor an unreasonsbly poor pump efficiency. - '_ | There is no convenient way to check the performance curve of the installed coolant‘pump, but the system head loss was recently recalculated by the writer at the design flow rate of 850 gpm. Table II gives the revised head losses of the various components of the coolant system. Allowing for a + 15% uncertainty in salt viscosity, the calculated coolant system head loss ranged from 9h to 99 feet of ‘salt as compared to the original design value of 78 ft Actually, a somewhat greater uncertainty band is probably required to account for the selection of friction factors and coefficients for entry and exit losses. The head loss of 99 ft would give a predicted minimum flow of about 800 gpm based on the coolant pump wvater test data and allowing for & 5% lower flow in the MSRE than in the water test pump. However, -this cannot be taken as a precise flow calcu- ‘lation because there are a large number of assumptions in the pressure drop calculation. and because the actual coolant pump characteristic curve 'might not be within the 5% margin. The minimum predlcted flow rate of - 800 gpm is still above the TTO gpm required to resolve the discrepancy ‘in reactor power. 10 Table II .CalculatedJHead Loss of MSEE Coolant Systems Components - At 850 gpm | - ~ Head loss ‘Ttem - (ft of salt) " Line 200 S i13.8. | - LineVEOl ‘ - __13_8 , ~ Line 202 - - 11.8 - Heat Exchanger - - 28.0 Radiator . S : : 29.0 . Total D 9% Air Heat Balance Calculastion The radiator air system provided an opportunity to make an inde- rendent measurement of the operating power of the reactor. .Héat baiances - on the air system were completed in May 1966 and these were in general agreement with the salt heat balances using the revised value of the coolant salt specific heat. The stack,airuoutlet.temperature*for these heat balances was measured at a,single point near the stack wall and the air flow measurement was based on the reading of a'pitot-#enturi'flowmeter ' at the center of the stack. The relation between the flowmeter reading | and the total stack flow had been previously determined at several flow rates frbm velocity profiles takén'with a hot wire anemometer. These velocity profiles were taken at ambient air temperature. This measure- ment related the total stack air flow directly torthe computer readout of the flowmeter output and did not involve the.manufactuferfé butput Vs velocity calibration data. 11 The only point at which air velocities could be measured was at a 1ocation about 50 feet up the TS-ft“high 10-ft-diameter air stack., This location would give upstream and downstream L/D ratios of 5 and 2.5 respectlvely. Both upstream and downstream distances are insufficient to ensure a normal flow distribution, and flow disturbances could be intro- duced by either the sharp 90° corner at the bottom of the stack or by wlnd effects at the top of the stack. R Since the air stack could also have a temperature distribution as well as a velocity distrlbutlon and since the velocity distribution might change under actual ofierating conditions, two air heat balances were completed in the fall of 1969. For these heat balances, the mounting of thewpitot-venturi-was-modified!and-a thermocouple was added so that welocity and temperature;traverses could be taken onitwo perpendicular diameters across the stack while the réactor was operating at power., A new velocity calibration°was:alsotcompleted'on.the pitot-venturi prior to using it in running the ‘traverses, The new pitot-venturi calibration gave air flow rates below those obtalned prev1ously | Elgdre_Q_shows_a_comparison of the new calibration and flow equation with the'previonsly used stack flow.relationship and with*the manufacturer's*calibration data. The new calfbration, which was in general agreement with the manufacturer's data, was adopted. ~ Air heat balance data vere taken at two operating conditions of the reactor, one at the nomlnal,full-power_condition and the other at highest _power.attainable withoneutlower;j-The results of the two air heat balances _are shown in Table III.";Préwious heat balances had given higher results more in agreement with the salt heat balances. The.main-difference in _the a1r heat balances was in the lower air flow rates indicated by the new calibratlon of the- pitot-venturi Figure 3 shows- ‘the velocity and "temperature d1str1butions across the radiator stack for the full-power | ,:'condition The average veloc1ty for the two traverses ‘shown was 3270 fpm '“ras compared to: 3h75 fpm which would have been dbtained with the previously used flOW'measurement The temperature distribution is shown as & tempera- " ture rise sbove ambient air temperature because the ambient temperature changed during the time data were being taken. Similar distributions were | (x10%) AIR VELOCITY (ft/min) 12 ORNL-DWG 70-5358 14.3 psiu,f'f?‘F | V=1254.78 A PO-53092 1184 2 © LABORATORY CALIBRATION b N ® MSRE STACK CALIBRATION | 4 MANUFACTURERS CALIBRATION . | | 1 o 4 2 3 &4 5 & | - AP (in.H20) | - Fig. 2. Calibration of Pitot-Venturi Air Flow Meter g 13 ‘Table III “ Results of MSRE Air Heat Balances Heat Refidfed‘by RadiatorAir Flow (MW) 7‘-" 7.335 Heat Removed by Cboling¢Water (MW);:E" 0.343 Heat Removed at Component Coollng PUmp (MW) 0.0166 Electric Fower Inpit - -0.679 Nuclear ‘Pover by Air Heat Balance (MW) | 7.01 Computer Heat Balance waer (MW) : T.96 Ratio of Air Heat ‘Balance waer/Salt S Heat Balance waer CLHEE V_Vfl-iif‘,;¢ | 0.861 II 4.93 0.295 0.017 -0.421* L.82 6.31 0.764 These valueé ineiude the power.lnput to the radiator heaters and to the main and annulus blowers in. addition to the electrlc power input appllcable to a salt heat balanee AIR TEMPERATURE RISE (°F) AIR VELOCITY (fpm) 14 . ORNL— DWG 70— 5357 130 l ' === 0 NORTH — SOUTH @ EAST—WEST 120 ' "o w00 L _ 3800 ' /8\ ' \ \B | 3600 . AVERAGE VELOCITY PER 2/ o = 7pT — JMSRE STACK CALIBRATION _ __ / | __ _ 3400 II \ £ [ <. averace | /| L BN _vevoomy | s YT 3200 L N ‘ 2 4 \ o \ "‘""'/5 \, / 3000 N~ \ s ° I . 2800 , 0 2 4 6 8 40 S AND W N ANDE STACK POSITION (ft) Fig. 3. Air Velocity and Tempera‘bure' Distribution in MSRE Radiator Stack at Nominal Full Power a o - =t 15 obtained for the partial power operation. The large variations in these velocity distributions are an indication of the difficulty in obtaining an accurate flow measurement. _ B - At the present time, the accuracy of the salt heat balance must be given precedence over the air heat balance for the following reasons. 1. The two air heat balances taken at different power levels were inconsistent with each other as indicated-by the ratios of air heat balance power to salt heat balance power shown in Table IIT. The salt heat balance power at various power levels was in agreement"withatheneutron flux power indication from the com- pensated ion-chamber and mas therefore proportional to the | actual power. ' 2. The difficulty in obtaining the true air flow and temperature rise with the large varlations as shown in Figure 3. 3. Unaccounted heat losses and air leakages from the radiator enclosure. The construction and instrumentation of the radiator air system were not intended for precisionrmeasurements'as-required for a heat balance and therefore the difficulties encountered were not surprising. PERFORMANCE OF THE MATN HEAT EXCHANGER AND RADIATOR The 1nitial escalatlon of the ‘MSRE power level in Aprll and May of 1966 showed that the heat transfer capabllity vas below ‘the de31gn pre=- diction for both the primary heat exchanger and the radiator. ‘With the reactor system operating W1thin its design temperature range, the maximum 'i:power 1evel of the reactor as calculated by the computer heat balance was ':limlted by these components to about 7 2 MW as compared to ‘the nominal fffull-power rating of 10 MW Slightly higher power could have been achieved 'p'by ralsing the fuel temperature, but the large temperature increase re- 'f}fquired to obtaln only a small power ‘increase made this lmpractical The original designs of both the heat exchanger and the radiator 'were rev1ewed to determlne the cause of the lower than expected per-' - formance. The actual operating performance was also carefully monitored 16 to determine if the reduced performance was caused-by some factor associ- - ated with the operation. A more complete discussion of the initial evalu- ation of this problem is given in Reference 10. Design Review Primary Heat Exchanger | 7 The primary heat exchanger is a conventional cross-baffled, U-Tube exchanger as shown in Fig. 4. Fuel salt circulates on the shell side at ; 1200 gpm and coolant salt circulates at about 850 gpm through the tubes. The exchanger now contains 159 half-inch tubes on a triangular pitch. For a more detailed description, see Reference 1. | - The methods used in the design of the MSRE heat exchanger are those commonly followed in designing heat exchangers of this type. The tube- side coefficient was computed from the Sieder-Tate equation, and the shell-side coefficient was computed from a correlation by Kern.,ll Im- » plicit in the use of these procedures is the aSsumption that the fused salts behave as normal fluids, Previous heat transfer work on fused salts . had shown this to be a valid assumption for both flow inside tubes and on the outside of tube bundles.Z 13 _ | The design calculations tend to give a conservatively low prediction of the heat-transfer capability (effective UA) for four reasons. First, the correlation for shell-side coefficient by Kern is conservative, i.e., his design curve falls below the data points rather than through the mean, This would tend to make the predicted shell-side coefficient low by O - 20%- Since the shell-side resistance is about a third of the total, the ~ effect of this conservatism on the predicted overall coefficient, U, is about O - 64. Second, the predicted coefficient would also be low because an additional resistance of about 11% was added arbitrarily to allow for scale. This was done even though it has been shown, both ifi_and out of o pile, that the salts do not corrode or deposit scale on Hastelloy-N under MSRE operating conditions. The third conservative approximation was in the definition of the effective heat-transfer surface area. Here no - credit was taken for the bent part of the tubes, i.e,, the active length .\i; 17 - -ORNL-LR-DWG 32036R2 FUEL INLET A THERMAL -BARRIER PLATE | _ S e QP gpr” CROSS BATRLES | " TUBE SHEET — \ ’ ' “ .COOLANT INLET . COOLANT-STREAM .. SEPARATING BAFFLE / COOLANT OUTLET - FUEL OUTLET - Fig h : MSBE-,'Pi'imary Heat Exghanger- »y 18 of the tube was taken to be the straight'portion between the thermal barrier near the tube sheet and the last baffle. This approximation was made in recognition that the thermal efficiency of the return bends might be less than that of the straight portions. Nevertheless, this region contains 7 to 8 percent of the total tube area and will transfer a signifi- cant amount of heat. Finally an additional 8% of active heat- transfer ‘area was added to the computed requirement as a contingency factor.- The net result is that a deliberate margin for error of over 20% was 1ncluded I1n the heat exchanger design. Between the design and the operation of the heat exchanger some | modifications were made.' When the heat exchanger was hydraulically tested withrwateribefore being instailed in the reactor, the shell-side pressure drop was excessive and the tubes vibrated. To reduce the high pressure | drop;‘the~outermOSt row of four tubes was removed and the corresponding holes in the baffle ‘plates were plugged. To alleviate the tube vibration ‘problem, an inpingement baffle was placed at the fuel salt inlet. In addition the tubes were "1aced" with rods next to each baffle plate to restrain the lateral movement.of the_tubes. A 1aced structure-was also built up in the return bend to make these_tube.projections'behave as 8 unit, and the tubes essentiallyflsupport_eachother;sNo:attempt was made to measure the overall heat-transfer coefficient,_hnt,it’does not appear that these changes were enough to affect the conservatism.in'the original design. The effect of the rods and impingementrbaffle was probably negli- gible. The loss in heat transfer by the removal of the four tubes was also relatively small. The heat-transfer area of the'remOVed tubes was only about 2.5% of the total; the effect on capacity was probably less because these particular tubes, by virtue of their proximity to the shell, would be expected to have heat-transfer coefficients below the average. a At the time the design review was completed in the'summer of 1966, we concluded'that the design methods were appropriate the assumptions conservative, and that subsequent modifications should not have used up the margin of safety believed to be provided in the design. 19 -~_1The three remaining possible causes of the low heat transfer were: 1. . That a buildup. of scale was occurring on the tubes even though | , this was believed impossible. _ - 2. That the-tube-surfaces were being blanketed with a gas film. 3. - -The physical properties of the fuel and coolant salts used in the design were ‘not the correct values. ,Subsequent operation of the ‘reactor, as discussed later in the report, . showed that the heat. transfer was constant with time, indicating no buildup of scale and that there,was-no,evidence»of gas filming, However, a re- evaluation of the physical properties showed that the thermal conductivity of:both the fuelfand‘QOOlant salt5was-sufficiently below the value used in the design to account for the overestimate of the overall coefficient.. ~Table IV on page723;of’this‘report shows a comparison of the original - physical property data to the latest values and.shows the effect on the ‘calculated heat transfer coefficients. Coolant RadiatOr' - The heat transfer surfaces of the radiator consist of 120 unfinned 3/h-inch tubes, each about 30 ft long.r The S-shaped tube bundle, con- sisting of lO staggered banks of 12 tubes each, 1is located in a horizontal '; air duct 50 that air blows across the tubes at right angles Doors can be lowered just upstream and downstream of the tubes to vary the air flow over them A bypass duct with a controlled damper and the option of using either one or two blowers provide other means of varying the air pressure 'drop across the radiator A detailed description of the . radlator and 1ts ';enclosure is given in Reference 1. - As in the des1gn of the primary heat exchanger the Sieder-Tate ~equation was ‘used to calculate the ‘heat- transfer coefficient on the inside | of_the,tubes. The same comments as. to: validity of method and accuracy of salt properties apply 4in both designs. ‘In the radiator, however, only 2% _;of the calculated: heat-transfer resistance was inside the tubes, 80 no .conclusions with regard to accuracy of :the inside film calculations can - be . drawn from the -observed performance. 20 - Over 95% of the resistance is on the air side. This coefficient was ~calculated using an equation by Colburnfrecommended9by McAdams .14 This equation is well-prdven for cross-flow geometries identical in all es- sentiale to the MSRE radiator. The difficulty with applying. the equation to the MSKRE design is the very large difference between'the'tube.tempera- ture and the bulk temperature of the air. The phySical-propérties of the air vary so much over this range that relatively large variations in the heat-transfer coefficient can be calculated depending on which temperature is selected for the evaluation of the physical properties. The MSRE de- "'Jsign,calculaticn‘used air -properties at,the.temperaturejofithe outside " surface of the tubes. The procedure recommended by McAdams is to evaluate the properties at a "Pilm temperature” defined as the average of the sur- face and the bulk air temperatures. Had this been done, the outside film coefficient .(and.the overall coefficient) calculated for the MSRE radiator would have been lowered from 60 to 51.5 Btu/(hr-ft=-°F), Even-icwer'values would have resulted if the physical properties had been_evaluated_nearer the temperature of the bulk of the air. S The heat-transfer coefficient calculated using the recommended air £ilm temperature was still greater than the observed value by about 20% A contingency factor of this magnitude would not be unreasonable when the large air-to-tube surface temperature difference is coneidered and when the ':unconventional geometry of the tube bundle within its enclosure is con- " sidered. However, the original radiator design had included only a h% 'overdesign. Analysis of Performance Zrimary Heat Exchanger The heat transfer performance of the main heat exchanger has been. monitored throughout the power operation of the MSRE by two methods. The overall heat transfer coefficient was evaluated by-a'procedure-described in References 10 and 15 which was developed to eliminate the effects of | certain types of thermocouple errors. The overall coefficient was calcu- lated frdm the equations: | | 21 1d(a +_B) A d(a-p) - ° o= g/ (Eccc © cve £Cr where | o and B Temperature parameters evaluated by on-line computer, U Overall heat transfer coefficient, A Heat exchanger surface area, Fe,F, Mass flowrrates of fuel and coolant, and CesCa Specific heats of fuel and coolant. The value of the derivative d(o + a)/d(a - B) was determined from the slope of the line obtained by plotting (a + B) vs (@ - B) at several dif- ferent power levels Thus a shOrt perlod at steady-state operatlon at several power levels was requlred for ‘each measurement at the heat trans- fer coefficlent | ;,( ; 1 , A more convenient method’of monitorlng the heat exchanger for changes in performance waslthe heat transfer 1naex (HTI) The HTI was evaluated at full power and -Was- deflned -as- the ratlo of the ‘heat: balance power to the temperature difference between the_fuel and’ coolant salts entering the heat exchanger.m Figure 5 shows thrlfiTI and overall gt of the heat exchanger taken over the life of. the reactor.~ These plots indicate that “there has been no deterloration of performance ovenfithe llfe of the re- actor and that there has been no detectable tube fouling or scale buildup 'in a period of about.3 1/2 years of reactor operation This would imply that the total operating life of the reactor 1nclud1ng about 26 000 hours J.______ of salt clrculation,_has been without scale buildup.%hfi%;_flf ~ The c1rculat1ng gas volume 1n,the fuel salt has varled from 0 to 0 6% during different periods of‘reactor operation.“ A test was conducted during the early power operation to determine if gas f11m1ng ‘of the tube ..... surfaces could be cau51néithe lower than predictéd ‘heat transfer. The test was conducted by rapidly venting gas overpressure from the fuel 22 (do/MW) X3ONI ¥IASNYHL IV3H ORNL-DWG 70-3174 (j-do z_4+ _J4 ME) INIDIII300 YIISNVHL LV3H 0 < " ~ - Q Q © 9 o G o O o o 0 . = _ _ o /2] 1 AT, = mean temperature difference. ¥4 oy 25 The difficulty in applying this equation was in the measurement of the outlet air temperature. A direct measurement at the top of the air stack could be made only when the bypass damper was completely closed and then a correction was required to account for the air flow from the annulus blowers. For other conditions, the ocutlet air temperature was calculated from a salt and air heat balance across the radiator. The air flow and temperature measurements were made with the original instrumentation and stack flow calibration discussed in the AiereatlBalancelsection of this report' ‘No attempt has been made to reevaluate”the radietor heat trans- fer based on the recent flow or temperature traverses because of the dis- crepancies that still ex1st in these measurements. ' The radiator overall heat transfer coefficients vs air pressure drop assuming a nominal full*power of 8.0 MW are shown in Fig. 6. The dis- continuity when the second'hlower_was energiied was originally believed to be the result of direct air impingement from the second blower. How- ever, this could be a result of flow or temperature errors and the calcu- lational procedures. The observed overall coefficient evaluated at full pover was L2.7 Btu/(hr- -5 °F) ais compared to the corrected design value - 51.5 Btu/(hr- £t2-°F), In all other respects;fthe performance of the radiator was completely Satisfactory., The heat transferuremained constant through the life of the reactor sndfthere'fiere'no.salt7lesks or other difficulties with the radi- ator itself. There were some early difficulties with ‘the radiator en- closure whlch are discussed 1n another report 16 These difficulties were eliminated by modiflcations to the enclosure and doors. 100 N OVERALL HEAT TRANSFER COEFFICIENT (Btu/hr - 112 °F ) 10 ~ ORNL-DWG 67-2155A 0 o _ TWO BLOWERS _|_| 0" N o =T | e | | ___.,J-—-fl"'—-T' | | -o—%ONE BLOWER > 5 0 AR PRESSURE DROP (m H20) Fig. 6. Observed Perf‘ormance of MSRE Radiator ‘ :2()_.7@5 | 9e -y up 2T 'CONCLUSIONS AND RECOMMENDATIONS _ The MSKE heat'exchanger'and-radiatorfiperformed completely satis- ,factorily except that the heat removal capability uas less than intended, In regard to the overall operation of the MBRE the power limitation im- ) 1posed by the heat removal system was not a serious problem. All of the ~ goals of the MSRE were successfully achieved at the attainable power level,. The analysis of the prlmary heat exchanger performance ‘showed that .the conventional heat transfer correlations are applicable to molten salts: - the initial overestimate of “the heat exchanger performance was completely ‘resolved by the revised physical property data for the fuel and coolant salts, Operation for more than 3 years showed no loss in hegt transfer capacity with time as a result of corrosion, scale, flow bypassing, or : flow restrlctlons In the case of the radiator, a discrepancy still exists between the calculated and observed performance. The cause for this discrepancy has not been deflnltely established It would appear that (l) the air-side heat transfer correlation was not completely suitable for the large- surface-to-air temperature difference that existed in the radistor, or (2):there were air flow leakages, bypassing, or air flow variations in this particuler installation that caused the low heat tramsfer. Regard- less of the reason for the overestimation of the radiator Performance, it is clear that the L, contingency factor in the orlginalldesign was in- - suffic1ent The unusually" low contingency factor for heat transfer equip- ment occurred because the ba31c radiator design was completed early in the | fMSRE design when the nominal design power was still 5 MW, The radiator tr'and main heat exchanger were designed for 10 MW to ensure adequate per- girformance When the nominal power rating of the MSRE was later raised to - 10 MW .the. radiator had only" a h% overdesign. An obvious but often ig- 'nored design principle, would be to evaluate the performance over the "maximum possible range of the physical property data and operational o _variables and then use . the resultlng performance range as part of the basis in selectlng a contlngency factor. A 1arger degree of overdesign would have been indicated for.the-radiator if this procedure had been followed. 28 Thus far we have been unsble to explain the discrepahcy'between the power indicated by the salt heat balance and that indicated by long-term changes in the isotopicrcomposition of ‘the fuel salt. The air heat balances indicated a lower power than the salt heat balance but the ac- .curacy of these measurements is not sufficient to override the salt heat balance. The coolant salt flowmeter is the only element that has not yet been checked as thoroughly as possible. A lower cooleht ealt'flow'rate was indicated by the decay of circulating activation products and also‘by e_the'recalculatibn of the coolant system head loss. However, neither of these methods is very accurate. The most relisble value of the flow rate should be determined from the planned calibration of the flow element dlfferential-pressure cells, , % ! ) 10. 11. 12, L -1k, - 1. 16. 29 REFERENCES R. C. Robertson, MSRE Design and Operations Report, PFart I, De=- scription of Reactor Design, ORNL-TM-728, (January 1965). P. N. Haubenreich and J. R. Engel, '"Experience with the Molten-Salt Reactor Experiment," Nucl. Appl. and Tech., 8, 118 (1970). R. H., Guymon, MSRE Design and Operations Report, Part VIII, Operating Procedures, ORNL-TM-908 Vol. II (January 1966). G. H. Burger, J. R. Engel and C, D. Martin, Computer Manual for MSEE Operators, internal memorandum MSR- 67-19 (March 1967). C. H. Gabbard, Specific Heats of MSRE Fuel and Coolant Salts, internal memorandum, MSR-6T7-19 (March 1967). MSR Program Semiann. Progr. Rept., August 31, 1968, ORNL-434k, p. 103. MSR Program Semisnn. Progr. Rept., Feb. 28, 1970, ORNL-4548, Sect. 6.2.1. MSR Program Semiann. Progr. Rept ., Feb. 28, 1970, ORNL-L5L8, Sects. 10.3 and lO b, MSR Program Semiann, Progr Rept. , Aug. 31, 1969, ORNL-kl4k9, p. 12. C. H. Gabbard, R. J. Kedl and H. B. Piper, Heat Transfer Performance of the MSREE Maln Heat Exchanger and Radiator, internal memorandum ORNL-CF-67-3-38 (March 1967). D. W. Kern, Process Heat'Trsnsfer, McGraw-Hill Co., Inc., New York, 1st Ed., p. 136, (1950). H, W. Hoffman and S, E, Cohen, Fused Salt Heat Transfer — Part ITI, Forced Convection Heat Transfer in Cilrcular Tubes C'ontaining the Salt Mixture Namogrnamo3-xuo3, ORNLr2h33, (March 1960) R. E. MacPherson and M M. Yarosh Developnent Testlng and Performance ‘Evaluation of Liquid Metal and. Molten Salt Heat Exchangers , internal - memorandum ORNL"CF-60-3 16l+ (March 1960) . | W. H. McAdams, Heat Transmission, McGraw-Hill Co. s Inc .y New York, 3rd Ed., P. 272, (195h) . | “H. B. Piper, "Heat Transfer in the MSKE Heat Exchanger, o Trans Am, ‘Bucl. Soc., Vol. 10 (supplement covering Conf on Reactor Operating ‘Experience), p. 29, July, 1967. R. H, Guymon, MSHE Systems and Components Performance, ORNL-TM, to be published. C W b 4 M Y - W OO\ e 26. 28. 29. h6 _ HHPY SN eRENRaNMEHARNEOS .!...O.‘...!F‘P?‘wzmqg 'bmmmwbswozqq> ORNL-TM-3002 INTERNAL DISTRIBUTION . L., Anderson F. Baes F. Bauman E. Beall Bender E. Bettis ~Bettis Billington Blankenship - Bohlmann . Borkowski . Boyd Briggs Carter Claffey Collins Cooke Cottrell . Cope, AEC-ORO ox o L., Crowley :&':‘:f—f!:"tdt‘ic—lflhjmm O"-Eltti L. Culler J. Ditto P. Eatherly Elias, AEC-Washington R. Engel _ E. Ferguson . M, Ferris P, Fraas K. Franzreb H. Frye - K. Furlong H. Gabbard B. Gallaher R. Grimes G. Grindell : H. Guymon - H. Harley - N. Haubenreich - W. Hoffman - Houtzeel L. 48, L9, 50. 51, 52. 53 5k, 25 56. 2T 58 . 59. 60. 61-62 63. 6k, 65. 66. 67 . 8. 69. T0. 1. T2 73. Th. 75. 76~78 19. 80. 81. 82. 83- . 8k, 86. . 88, 89. e uaquzwwzmwb?mzw>Qfieomfiwwz t1b+=s§=:gifi . - L, Hudson . R. Kasten Kedl . Kelley Krakoviak Kress Lane - it Laughon, AEC-OSR Iandin Lyon MacPherson McCoy McCurdy McGlothlan McIntosh, AEC-Washlngton McNeese McWherter Miller . MOOI‘e . Myers . Nichol . Nicholson oy - * s ahlds e @ RpmpHaHS RQEE S 4 E B 0N E > » . * e * o . . . * s . H = NP 5 5 0 < o L. Ragan Richardson R. Riley, AEC-Washington C. Robertson W. Rosenthal M. Roth, AEC-ORO P. Sanders W. Savolainen _ G. Schleiter, "AEC-Washington J. Schreiber, AEC-Washlngton nlap Scott M. Scroggins, AEC- Washington -Shaw, AEC-Washington J. Skinner . N. Smith . Spiewak . A. Sundberg FRRAPREEDIES 9l. 2. 93. ok, 95. - %.. 97. 9. 99. 100-101. 102-103. 104-106. 107. - 108. 109-123. 124, | ORNL-TM-3002 INTERNAL DISTRIBUTION (continued) R. E. Thoma D. B. Trauger A, M, Weinberg J. R. Weir . M. E, Whatley J. C. White G. D. Whitman L. V. Wilson Gale Young Central Research Library (CRL) ¥-12 Document Reference Section (DRS) Laboratory Records Departnent (ILRD) Laboratory Records Department —- Record Copy (l’..RD-RC) ORNL Patent Officé EXTERNAL DISTRIBUTION Division of Technical Information Extension (DTIE) Laboratory and University Division, ORO . v