T e R ok A g e D R o A T et i “h it oy SRRt o s i N i s o : ar L "% i : ",3 . A "',:_VOAK RIDGE NA'"ONAL LABORATORY o o operated by L UNION__ CARBIDE CORPORATlON IR N UCLEAR DlVlSlON S o forthe ""U S ATOMIC ENERGY COMMISSION l o 7' f‘}ORNL TM 302"{ CARBIDF' fel COpy No -_:_. ""--—__'"-,'-:--_‘-_:_VDATE - June; 1970 . SPRAY MIST BUBBEES AAND FOAM IN THE MOL‘]IEN_}SALT REAC'I'OR EXEERIMENT - g.R. 'Engel,% P, N Hau enreich, : afld;,A . Houtzeel ~ L\ if;—}mThe fuel pump bowl in the fi;MSRET 'incorporated a. ring from whlch 50 gpm of salt was sprayed. through the cover gas and into the: salt: pool - The -device effectively removed xenon. from the fuel as intended but ‘also- produced several ineidental effects, - Although none seriously interfered with opere.tion, con~ - sidereble effort was spent on. elucldation of these effects.,__';.,7}'{_;7;;, - ' -a,i'The spray produced a mist ofrselt droplets some of which drifted into _the effgas line at a rate of a few grams per month, " The resultent salt de- po81ts required cleanout at intervals of six months to a year, The stripper Jets also drove bubbles several ‘inches into the salt pool, reduclng the "._,_ Caverage density'and raising the actual level sbove that indicated by the bub- bler level elements._ Some. salt transferred into +the overflow 11ne, apmarently -as froth although there was no evidence of persistent foam.,-Mbst of the bub- bles driven into the salt returned to the surface, but a small fraction was- ~drawn into the circulating loop, The situation was ‘such that small changes 1n o pump speed or. physical properties ‘of - the salt changed the depth of the bubble_ e “zone -enough to change the volume fraction of gas in the loop over the range o from 0 02% to 0. 7% - L L e e e e Keywords MSRE fused salts, reactors, operation, pmps,bubbles, S separatmn processes, reactivi-t" -, performance. R R T ...... - rLuborotory Atis subjecf to mvmon or correctnon and therefore does _nof represenf a hnuf reporf Sl e - L DMTION OF THIS DOCUMBNT IS UNLMI'ED i 3 § 1 i ; T g s g i | e ey s i e LEE;AL 'NOTICE This roport was proporod as on cccount ol' Govornmnt sponsoud work. Neither the Umtod S'ai-;, nor the Commission, nor any person ucting on behalf of the Commission: comp!ohness, or usefulness of the information contained in this report, or ihcfl the use of - ony information, apparatus, me'hod er pvocess disclosed in thu uport may not infrlngo . privately owned rights; or T . - ; . :B._ Assumes any liabilities with r.spod‘ to thc use of or for damages nsuhmg from tha use of any information, apparatus, method, or process disclosed in this report. " As used in the above, “‘person acting on behalf ‘of .the Commission"” includes any omployae of contractor of the Commission, or Omployo- of such controctor, fo the extent that such omploycc - . or contractor of the Commission, or employes of such contractor prepares, disseminates, or - _provides access to, any information ‘porsuant to hls ‘employment or contract with the Commission, | or his omploymenf with nu:h contractor, - - - A, Makes any warranty or representation, expressed -or implied, with respect |o the uccmgcy. I LEGAL NOTICE This report was pfcpurod as on account of Government sponsorod work. Neither the Umhd Sfato;, . nor the Commission, nor any person acting on behalf of the Commission; ) A, Mckes any warranty or representation, expussod or implied, with rospeci to the uccurncy,' ' complofeness, or usefulness of the mformuhon contained in ‘this report, or that the use of any mformmlon, apparatus, methed, or process disclosed in this report may not lnffinge‘ - privately owned rights; or : . . : : : . B. Assumes any liabilities with rupocf to 'Iho ‘vse af or for damoges resulting Irom iho use of . "any information, opparatus, method, or process dusclosod in this report. - As wsed in the above, “‘person acting on behalf of the Commission' includes any ofisployse or - ‘cantréctér of the Commission, or employse of such contractor, to the extent that such employse . or .contractor of the Commission, or smployee of such contractor prepares, disseminates, or . | . provides access to, any information porsuant to his employmem or contract mth the Commuss:on T S 11 - or his employment with such contracter, - o o 4 \ 1 | LEGAL NOTICE ‘ This report was prepared a8 an account of Government sponsored work. Neither the United - States, nor the Commiasion, nor any person acting on behalf of the Commisston: E | A. Makes any warranty or representation, expressed or implied, with respect to the accu- | * racy, completeness, or usefulness of the information contained in this report, or that the use .. iii ' * aniy {nformation, apparatus, method, or process diaclosed in this Teport may sot infringe | wvately owned rights; or .. - o L B. Assumes any liabilities with respect to the use of, or for damages resulting from the’ | use of any information, apparatus, method, or proceas disclosed in this report. - : | As used in the shove, “‘person acting on behall of the Commission*’ includes any em- o , . | plogee or contractor of the Commission, or empioyee of such contractor, to the extent that - CONTENTS oyee or contractor of the Commission, or employee of such contractor prepares, /\ ¢ such empl R 51 disseminates, or providesa access to, any fnformation pursuant to his employment or contract - " | with the Commission, or s employment with such contractor. L e | 7 - . - . . | , Lage @ ABSTRACT . l.' m’mochION- V. . . n . o * '. . * . 7. - l 2. IESCRIFTION OF FUEL FUMP AND CIRCULATION TOOP . . . ... .. 2 lep-o---coooo------.-..' . - 2 LOOP-:-...-.---... '.‘.‘.-"...'.8 3. SIRAY AND - - i . i MIST. - - . @ - . . e . & . . . . - * . * * - e - - ® lo - Observations in Development Facilities . . . . . 10 Salt in MSRE Offgas System , , . . . . ..........15 Possible Effects on Transfer to Overflow Tank. . . . . . . .18 COI],CluSionS. . . .' . _:' .. '._.4 . . s & 8 o s @ 019 L. BUBBIES AND FOAM IN THE FUMP BOWL , . e 20 Description of Bubbler Level Elements, , . | 20 Measurement of Absolute Densities, , ., . . : : e s e e '-Eh - * MEasurement of Relative Densities, , . ., , .- S sl Ei; - Variation of Void Fraction with Depth, cr st o7 —~ Effects on Reactor Operation , ., ., . . :': : : c s e e e 21 5. TRANSFER TO THE OVERFLOW TANK , . . . . . . . . . | 32 Initial Observations , , . . ' Experience with Fuel Salt . . . . . . . . . . .. .. " 3 ~ Experience with Flush Selt . . . . . . . . . o o oo 33 Relation to Other Operating Variables, e 39 Effects on Operatiom., ., . . . . . . . . R hg ‘6. GAS IN THE CIRCULATING LOOP . . . . . .. k2 , Ihdicétbrs e e T - SRR T | EXPerience ., . . . v .'p v 0 o . Tt T i ¥ 7. REMARKABIE EEHAVIR OF GAS AFTER THE 1968 SHUIDOWN, . . . . . 56 ~ Salt Condition at the Start of Run 15, , . . . . . . . . . 56 :Behavior During Flush Salt Circulation . . . L : :~' 61 -iggg:v?or-ggring'Firsz~Eériod of Circulation of Fuel Salt . 62 ~Behavior During Subséquent Beryllium Ex sures. in | * ~Other Redox Experiments, . . . . e e e Th » < !r‘g, 78. 7 HYMIESES AND CONCIUSIONS. .V .. '. Vl l > » . - L .. . » ’ . . | 79 APEENDIX A — MATERTALS REMOVED FROM FUEL FUMPBOWL , . . . . .. .81 APENDIX B — MSRE SALT IENSITIES . . . , . . . . . . LT " DISTRIBUTION OF THIS DOCUMENT IS ONLBNTED T e e v A BTN & «) ] > SIRAY, MIST, BUBBIES, AND FOAM IN THE MOLTEN. SALT REACTCR EXFERIMENT J. R. Engel, P. N. Haubenreich, and A. Houtzeel . 1. mmbmcrion One of the Durposes of the MSRE was to show that handling molten fluoride salts in a reactor is & practical matter. OFf course, before the MSRE ever operated there was & considerable_body of experience which said . that handling problems-: would,not be bad. But the question still had to ‘be answered whether after 1ong exposure of the salt to the reactor en- ;vironment with the concomitant changes in composition, there might not be some unforeseen problems with its phy31cal behav1or. The general conclusion from the years of MSRE operation is that the - salt is well-behaved, and the original premise that molten salts can be -handled in a reactor without much difficulty deepened into conviction that this is true, _ Thatrthe concluslon was. favorable 1s not to say, however that nothing punexpected turned up or that ‘there were no problems Salt mist in the fuel-pump bowl led to plugging in the offgas line; there was'salt transfer into an overflow Ppiye above +the salt surface by & mechanism that was never definitely established and there were changes that affected ‘the behavior ' {;of gas that was churned into the salt These matters were the sabject of ;‘Amuch discussion and study and have been described from time to time in in- | irformation ‘meetings and: progress reports. It has been difficult for an ';'dinterested person to form a clear overall picture, however because there '~”are 50 many facets and the experience has been spread out over such a 1ong flperiod of time, The purpose of this report then, is to bring tOgether bdthe available ev1dence in one place and to “tell what we have been able to | [deduce. . 2. IESCRIPTION OF FUEL FUMP AND CIRCULATING IOOP Many of the phenomena to be discussed in succeeding chapters of this report are closely related to the particular cemponents and their configu- ration in the MSRE, Although much detailed information is available from other sources, notably the MSRE Design Report,l a description of some parts of the System is presented below to provide a common base for further dis- cussion. Pump _ The spray, mist, bubbles, and foam in the fuel system all have their origin in the fuel circulating pump. This'is really a multipurpose com- ponent whose tank not only houses a centrifugal impeller and volute to circulate the salt through the loop but which also serves as the surge tank, the salt sample point, the uranium addition point, and the contactor for continuously stripping gaseous fission'products from the fuel salt. Figure 1 is a cross section of the fuel pump with details of construction omitted to emphasize the flow patterns. The motive force for the salt flow is provided by an 11. 5-1n.-d1am impeller driven at 1189 rpm by an induction motor. The impeller and its volute are installed in a 36-in.-diam tank or pump bowl. In the MSRE fuel system the impeller delivers 1250 gpm against a head of 55 ft of salt (~ 53 psi) Some 50 gpm is diverted as described below, so the flow out through the tangential discharge line that penetrates the side of the pump tank is 1200 gpm. o | Since fluid pressures inside the volute are high relative to those in the pump tank, some salt "1eakage"-occurs when there are unsealed fit- tings., The major ccmponent of this"ieakage" is the'sc-called fountain flow through the clearance between the impeller shaft and the top of the ;volute. Baffles afé previded around the shaft to keep the fountain flow R, C. Robertson, MSRE Design and Operations Report, Part 1 — Description of Reactor Design, USAEC Report GRNLrTM-728 Oak Ridge National Laboratory, January 1965 v ST 1 \\\,“. R S R \\\ SAMPLE CAPSULE " CAGE \\\\\\\\\\\\\\\\\\\\\\\\\\\\\.\ ) R = \\\\ N "N N) N & N L OVERFLOW | . "PIPE 100~ . 80 €0 40 20 LEVEL SCALE (%) 1. Flow o) 13335%% 1 %% R\ - SHAFT " _PURGE s . . SUCTION Patterns in MSRE Fuel Pump ORNL-DWG 69- 10172A o REFERENCE ‘BUBBLER ,-INE ! - v | w 1 b LY ' . DlSCHARGE is in the region”above and outside the skirt that extends out from the volute. This region is agitated by the jets and should be rather well-mixed Salt l enters the region under the skirt with an average radial (inward) velocity - . , . of only 0.11 ft/sec, accelerating to 1.7 ft/sec through the scalloped _ #@P openlng into the pump suction, Those gas'bubbles which rise out of the = salt after it passes under the skirt, together with the gas from the - QE,Q et mea s a1 B Tt - iobnbantn o ik S kA 4 e bk 401 eSS L LB ke bbb LB L R » “) «) “op on. bubbler level tubes,'can:move5along the bottom surface of the volute and up through vent holes at the juncture of the skirt and the volute. _ With the fuel pump off, both the spray ring and the volute support cylinder are practically empty. These regions fill with salt, at the ex- pense of the level in the pump bowl, when the pump 1s turned on, - Thus, even if. there are no gas "pockets" in the- loop, some drop in fuel-pump level occurs each time the ‘pump is started The salt volume associated - with these two regions is 0.27 ft3 * The volume calibration of the fuel pump is of interest in translating level changes to volume changes in the rest of the loop. Figure 2 shows the volume of salt in the sump region of the fuel pump (outside of the volute) as a function.of salt level. This calibrationdoes not include salt inside the volute.support cylinder or spray ring since the smount of salt in these regions does not vary with pump-bowl level when the pump is Another important-funCtion'of'the fuel pump is to provide the gas- liquid interface and a compressible surge volume for the fuel loop. Since the gas space in the pump ‘bowl (nominally 1.9 ft3) was Jjudged inadequate _ for a major salt-level excursion an annular 5.5-£t° overflow tank was pro- - vided below the pump'bowl around the pump suction line, Communication between the pump bowl and the overflow tank is through a 1-1/2-inch IPS vline that extends upward into the pump'bowl ahove the normal salt level. (See partial sectlon immediately to the left of the main drawing in Fig. l ) To minimize the intrusion of salt spray into the top of this open line, - the baffle on the spray ring has an. extension that forms a "roof" over the Luline. Four cover-gas streams normally flow into the pump‘howl Two of | .2these flow through separate 1nternal surge chambers to the” ‘dip tubes of '_ilthe bubbler level elements., The surge chambers prevent expu151on of salt "".into unheated gas supply lines during rapid pressure excursions. Normal "'gas flow through each bubbler tube is 0.37 z/min STP. Another minor gas £low (0.15 z/min STEO enters through the bubbler reference line.' The en- '7'trance of this line into the pump bowl is baffled as shown, to minimize the intrusion of salt spray. ' ORNL-DWG 70-5191 N N VOLUME (§13) V4 3 -2 A o 1 2- 3 4 5 & 7 ELEVATION ABOVE LOWER BUBBLER TIP (in.) - Fig. 2. Salt Volume as a Function of Level in Fuel Pump Suinp L «) " o i as s L AL A0 M A 10 a0 i3 8L o e e The prlncipal flow of cover gas enters the pump‘bowl through the an- ~ nulus between the rotating ‘impeller shaft and the shield Plug, The pur- ~ .pose of this entry p01nt is to prevent diffusion of radioactive gases _(and possibly salt mist) up the shaft annulus to the vicinity of the oil- ',lubricated bearings. The flow enters the annulus between the bearings and the 1mpeller and a small amount (~ 0.1 £ /min) flows upward to keep oil vapors from diffusing downward the remainder flows down into the pump bowl, _The normal gas flow rate £o the shaft annulus is 2 L z/min STP but values between 1.5 and 5 z/min have ‘been used Additional cover gas can be pmov1ded by flow down the sampler line but this line is normally closed at the sampler. During routine sampling (or enriching) operations,'a gas flow of 1 z/min STP down this line is maintained 5 L ' The cover gas normally 1eaves the pump bowl through two 1/2 in, IPS lines with baffled: bottoms.r The two lines penetrate the pump bowl h-l/2 in, apart and merge into a. single l/2-in. pipe within 8 in. The offgas flow . carries out gaseous fission products (and some salt mist) and provides a rmeans of regulating the system overpressure. .Two circular baffles are .Vprovided in the pum;rbowl gas space to minimize the transport of salt mist into the offgas lines. When the main offgas lines are restricted part or all of the pumpcbowl cover gas flows down the salt overflow 1ine and bubbles through. the salt 1n the overflow tank, It then enters the main fuel offgas line through the vent line on that tank The last device in the pump bowl is the cage for the sampler-enricher. TfThis cage consists of 5 vertical rods to confine the capsules that are _fflowered into the pump bowl However the ~cage has no bottom 50 capsules 'ffcan reach to the bottom head of the pump tank, This cage 1is surrounded ufby 8 spiral baffle whose purpose s to prevent salt spray from entering .. . and obstructing the opening 1nto the pump bowl. There 1is a narrow slot " between the bottom of the ‘affle and ‘the botton head of the pump bowl to _:permit circulating of liquid salt through the sampling chamber. The general configuration of the fuel circulating 1oop is shown sche- L matically in Fig. 3. Although many detalls have been omitted,lsome rela- tively minor items are deliberately inclnded because of:their importance to the discussion in subsequent sections of this report. In general, the components and connecting piping are arranged and pitched so that, under stagnant conditlons, gas bubbles within the loop will tend to migrate toward the gas-liquid interface in the pump. The k most notable exceptions to this are the top of the heat exchanger where the fuel salt enters the shell and a small region at the bottom of the larger core access plug in the reactor neck. The extent to which such - ~ bubble migration actually ;woceeds depends on & number of relatively in- | determinate factors 1nclud1ng the tendency for bubbles to slide along surfaces ' - An important factor in evaluating the behaV1or of circulating bubbles 'is the fluid pressure as & function of position in the loop. The table on Fig. 3 shows the calculated absolute-pressure at several points when “the ‘ pressure in the pump bowl is 5 psig and the salt flow rate is 1200 gpm ¢ (Ref. 2)., Other tabulated qnantities of interest are the salt volumes | between points and the transit times at 1200 gmm. _ . - Thermocouple TE-R52 will be referred to leter. This thermocouple is in a well that protrudes into the salt stream at the lower end of the ~ core specimen access plug (just to the fight of point 8 in Fig. 3). There is good reason to bhelieve, however; that the thermocouple junction was not actually inserted tcrthe very bottom of the well but remained up_inside the plug. At any rate its reading was several hundred degrees'below the tempergture of the salt leaving bhe core and seemed to be respcnsive‘to ‘changes in the salt level in the annnlus around the plug. 2. J. Kedl, internal communication, June 17, 196kL. - ORNL- DWG 70-5192 LOOP DATA AT 1200°F, 5 psig, 1200 gom = S DIFFF., ~ TRANSIT .. . : ! POSITION VOLUME TIME PRESSURE T | (%) (sec) - ({psio) ¥ 10 204 . -y 1.1 _a4y: 733 | 2 : 0.76 - : 0.26 " 69.7 Il_'l!‘ ll'hil'h’ll_'! 5 9.72 - - 3.63 —e 7 23.52 879 355 s 137 081 ... - Fig. 3. '_H'essures,_;VOJ.umes , and Transit Times in Fuel Circulating Loop 5 ok} 10 3. SERAY AND MIST During operationnof_the fuel pump there is a mist, or suspension of molten-salt droplets, in the g&s space of the pump bowl. There is'much ' evidenee for this, but none clearer than Fig. 4., This is a photograph of a 1/2-inch-wide strip of stainless steel that was exposed in the sampler cage for 12 hours, The 4-inch-long strip was positioned'so that the lower end was at the salt pool surface (as indicated by the bubbler level'elements). Therupper‘end was near the penetration of the sampler tube into the top of the pump bowl. (See Fig. l.) Although the size of the.droplets and the B amount of salt on the specimen in Fig..h are unusual (possibly'because of the long exposure or the condition of the stainless steel surface), quali- | tatively similar deposition was observed on. numerous other sample capeules and a set of graphite specimens3 exposed in the gas space., The origin of most of the mist in the MSRE pump bowl is undoubtedly .' the spattering and'splashing of the streams from the spray ring. _In some of the.pumpe that were operated during the deveiopfientprOgram5-the leakage up around the shaft also emerged into the gas space as a spray. The mist | may drift with the purge gas flow into the gas lines attached to the top of the pump bowl and freeze there. Frozen mist has been & problem in some pumps, not in others. Spray or falling mist has also been suggested as a possible contributor to the transfer to the overflow tank on the MSRE fuel pump, but, as will be explained below, this ean hardly be the meinflcause.. Observations in Development Facilities | One of the first moves after the MSRE was epproted in 1960 was the design and construction of a water loop for pump development 4 The pump tank and inlet pipe in this water loop were made of Plex1g1as to permit direct observation of flows. '"The hydraulic design of the test pump and - MR Program Semiann. FProgr. Rept., Aug. 31,.1967, 0RfiL-hl91, p. 131. *MSR Program Semiann; Progr. Rept., Julyrsl; 1960, ORNL-301k, p. 29. — &y " fr Fig. L. ‘” ; Y Pho‘!:Ograph of Salt Droplets on a Metal Strip Exposed in - Gas Space for 10 Hours | i s [-—l that of the reactor fuel pump were identical."® "Various baffles were de-- vised to confrol splash, spray, and gas bubbles caused by the operation of the bypass flows in the pump tank."® “Obsérvations of the.fountaifi- flow from the impeller upper labyrinth reveasled the need to control it; the slinger impeller was ceusing an undesirable spray., This spray wés contained and controlled by use of a cover enclosing the labyrinth and slinger impeller, and having drain ports located at its lower end.,"” Five different configurations. of xenon strippers were tried; the last was the toroidsl spray ring with two rows of holes and a flow of about 50 gpm, "Considerable splatter of liquid resulted from the impingement'of this flow onto the volute and volute gupport. Cpntrol of this srlatter was obtained through use of baffles installed on the stripper and on the volute support,"® The -prototype fuelrpump,lwhich wasrtésted by circulation'of salt for thousands of hours, was equipped with the same kind of stripper and baffles that had been in the final tests.with water. That the spray situation was adequately handled was indicated by circulation of salt for nearly 1k, 000 hours, During the first 11,000 hr, there was no trouble at all with the offgas, but the purge flow was quite low (<0.lL g/min). After the purge was increased to the MSRE design rate of 4 £/min, some minor difficulty was encountered with plugging in & needle valve about 15 £t downstream of the pump tank, After 2500 hr at the high purge rate a "hot trap“ ébn- sisting of an enlarged section of pipe which could be heated was installed ~in the offgas line near the pump tank, After this modification the pump was operated only 300 hr longer before it was shut down to make way for testing the Mark IT pump, but in that time there was no indication of Plugging.® SMSR Progream Semiann. Progr. Rept., Feb. 28, 1969, CRNL-3122, p. MT. ép. G. Smith, Water Test Development of the Fuel Pump for the MSRE USAEC (RNL-TM-T9, Oak Ridge National Laboratory, March 1962 p. L. 7Ibid., p. 22, 8Ipid., p. 27. ?A. G. Grindell, private communication, December;.1968. % . em ks e et e A YRS 1k 5 41 s a0 A1 58 bl 4 b 580 o ootk ) v 2k <) 13 A similar salt pump, the PK-P pump, was operated for & long period ' .of time withoutvtrouble., During e shutdown after 5436 h of operation .'fdifficulties vere caused.by an aerosol-type dispersion of salt particle “.1 A cylindrical baffle of sheet Inconel inserted in the sampder guide stopped | : the salt deposit in the capsule stop area, Afterrthe offgas line and the - --”'_bubbler reference. line were switched (so the offgas 1ine'was'baffled'and” '”there was no evidence of salt collecting in the offgas line or spark-plug risers in the pump tank. "10 This pump eventually operated a total of -23,500 hr with fluoride salts w1th no pdugging of the offgas system. The - gas purge rate was . only about a tenth of the MSRE design rate, however. - The pump in the Engineering Test Loop was & DANA pump 1eft over from the ANP, The hydraulic. performance of the DANA pump. is not the same as the MSRE pumps but the two pump configurations and the fountain flow which emerges into the gas space are very similar. "Some difficulty was experienced during the operatlon of the loop with. pdugging of the pump offgas line. Examination after the first 788 hr of operation revealed a collection of salt at the junction of the unbaffled 1/2-inch-pipe offgas connection with the DANA pumprbowl 1id."1' The bubbler reference line through which gas entered the “pump bowl was covered w1th an "internal - splash shield" or baffle and although there was some salt around the pipe opening it was not obstructed - There was however a potentially trouble- some deposit 6 in. above the liquid level at the capsule stop area of the _ sampler.enricher guide tube.__An.air-cooled copper tube inserted in the pump. bovl during 89 h of salt circulation at 1100°F accumulated a deposit of what appeared to_be_frOZen'droplets, No such deposit appeared on a cold finger in the drain tank and it was concluded that "the pump bowl . s. nl2 *fihi}the reference line was. not) the Pump was operated for several thousand "'1 hours without evidence of further Plugging 13 - 'alqmss reogram Semiann. reogr. Rept., Aug. 31, 1961 CRNI- 3015, P. h6 1M Program Semiann.'frogr. Rept., Aug. 31, 1962 RNL- 3369, p. 56. 12mid., p. 58. - o 1L L. Crowley, private communication, December, 1968 1 The Mark-II fuel pump was designed to'provide a greater'surge volume for fuel salt expansion.l* In the new design the spray fing and baffles were also changed with the intention of improving the xenon stripping, In the water tests of the pump tank models "large numbers of very small ~ bubbles were present in the pump tank liquid;"15 In the effort to reduce | ‘the bubble production in the liquid, a baffle was added which intercepted the spray jets. This created another Problem: impingement of the jets on this baffle produced much spray and mist. Another vaffle or cover over the spray area prevented direct splashing against the topAof the pump tank but mist could still drift out underneath the impingement baffle into the - region from which the offgas is drawn. o | The Mark-IT1 pump was never 1nstalled at the MSRE, but it was opera- ted for an extended period in the pump development faclllty. During the first 4000 hr of salt testing of the Mark II, there was far more accumu- lation of frozen salt droplets in the offgas line than there had been in ihe tests of the earlier model pump. A filter about 15 £t downstream from the pump bowl trapped salt particles (all 1oy or smaller) at a rate of about a cubic inch per 100 hours of salt circulation,2® A noticeable re- striction which built up in the offgas line was repeatedly (about once a | week) relieved'by.applying a torch to the vertical section of offgas line just above the pump bowl or by rapping at this location. Examination re- vealed a brittle deposit which crumbled easily and was made up'of sa1t beads up to 15y in diameter.l? After LOOO hr of operation, the salt level in the pump bowl was raised abeut 5 in., submerging the lower edge of the impingement baffle so that mist-laden gas could not flow so freely into the region of the offgas line. This drastically reduced the rate of accumu— lation of salt in the offgas line, although it did not altogether eliminate - it, | 14¢sR Program Semiann, Progr. Rept., July 31, 1963, ORNL-3529, P. 3. ~ 1°MSR Program Semiann. Progr. Rept., Jan. 31, 1964, ORNL-3626, p. 4l. 16A, G. Grindell, private communication, February, 1969. 17MSR Program Semiann. Progr. Rept., Feb. 28, 1969, CRNL- 1.L396, PD. 31 - 32. * 8’ ). .o o+) 15 Salt in MSRE'Offgas System From time to time throughout the operation of the MSME, solids have - been removed from filters, valves, and lines in the fuel and coolant off- _gas systems, Material from the coolant offgas system has shown only traces of salt constltuents, but many of the fuel offgas SOlldS have 1ncluded 'tiny beads of frozen salt evidence of the salt mist in the fuel pump bowl. | During the pmecritical operation, special side outlet inserts were installed in the offgas,flange nearest the fuel pump boul.r the first for a krypton-stripping eXperiment rthe second for drawing off gas to a fluo- o ride analyzer. When the flange was opened to install and remove. these in- _ serts the line appeared generally clean but small amounts of solids were found between the flange faces. These_contained tiny_glassy_beads which -_ were presumed to be frozen salt mist The behavior“of the fluoride ana- 1yzer during startup and operation of the fuel pump also suggested the presence of particulate fluorldes in the sample stream.18 L Frozen salt beads. were also found during the prenuclear operation ' far downstream at the fuel pressure control valve in the vent house. Near the end of the first run with flush salt fuel pmessure control became erratic and after the end of the run (March 1965) the small control valve was found to be partially plugged An\acetone rinse contained small | (1 - 5u) beads of a glassy_material. After a week of carrier salt circu- lation-in May 1965,_thejsmall control valve again began to plug. This time . it was removed and cut'openrfOrrexamination.' A depOsit on7the stem was 'g_:found to be ‘about’ 20% amorphous carbon and the remainder l-'to 5- u beads | _having the composition of . the flush salt 19 The beads ‘were glassy rather "'pthan crystalline, 1mplying rapid cooling of molten salt mist The carbon. e_f'was pmesumably soot from Oll thermally decomposed in the pump'bowl Salt was not an important constituent of the material that caused 1the severe plugging of the fuel offgas filters and valves when the reactor 187, q. Million, Analysis of the Molten Salt Reactor Offgas for - HydrOgen Fluoride, K—L—2079, September 1, 1965 19ysR Erogram Semiann. Pr0gr. Rept., Aug. 31 1965, ORNLP3872 p. 1k, 16 was first'operated‘atepower (Janusry - February 1966). Samples of the effendiflg material were nearly all radiation-polymerized organic, either _as & viscous liquid or a sticky solid, with traces of salt showing in only a few of the samples. In September 1966 when the first perticle trap was | removed, the material plugging the entrance stainless steel mesh was found to be organic with a very high fraction of barium and strontium (20 wt% Ba, 15 wt% Sr). There was only & minute emount of salt (0.0l wt% Be, 0.05 wtd Zr), ifidicating that if frozen salt mist was still leaving the pump bowl it was being stopped somewhere short of the vent house vhere the particle trap was located, , - Material chiseled and rodded from the offgas line near the pump bowl ' in November 1966 appeared to be mostly residual frozen salt from the over- £111 with flush salt that had occurred in July. The increase in pressure drop during operation must have been caused by some material gradually fill- ing the small blow-hole through the frozen salt, but the nature of this rplugging material was not determined. | The offgas line at the pump bowl was opened next 16 months later, in April 1968, to investigate the unusual pressure drop that had showed up in intervals of low-power operation during the preceding 6-m6nth run, A flexi- ble cleaning tool pushed to the pump bowl came out‘with material adhering to it that appeared to be a tar-like base with 8 fair amount of salt in it. The flexible jumper line and the flange Joints were found to contain llght deposits of soot-like material with a small amount of frozen salt droplets.®! In Decembervl968; the -offgas line was again cleaned from the nearest flange'to the'pump’bowl to relieve‘a restriction that had appeared'during the 2800-h of high-temperature operation in the 237y startup. An obstruc- tion was encountered gbout where the one had been at the end of the assy operation: Just above the pump. bowl where the temperature of the gas stream would be decreasing rapidly. The material that came out on the cleanout tool looked different, however., Instead of edhering, tarry ma- terial'generally distributed, there were a few bits of material having the . 20MSR Progrem Semiann. Progr. Rept, Feb, 28, 1967, CRNL-4119, p. 55. . 21MR Program Monthly Report for June 1968, CRNIL internal memorandum MSR- 68-98, p. 19. " " )} [1] <) - appearanceof salt.'_The;cleauout'toolVWas hollow and connected to an ex- _chaust pump- through a. filter;f'Thevfilter paper collected a blackish powder , 1 which appeared to contain h to 7 mg of fuel salt which had gotten there ~ since the 233 startup. | During the first 5 months of 1969 a signiflcant restriction again de- veloped near the pump bowl. As was the case with other restrictions, this vdiverted the offgas flow through the overflow tank. However, one week be- fore a scheduled shutdown on June 1, 1969, a restriction also developed in _the vent line from the overflow tank, Since the latter restriction was "essentially complete, the . offgas again flowed through the restriction in ”dthe normal offgas 1ine which by that time had ‘increased to sbout 10 psi fat normal flow, Some adjustments were made in operating parameters and .this condition persisted until the scheduled shutdown with no serious ef- 'fect on thevoperation. _This,time, however the 1ine was not opened to ‘determine the nature of the plug.. Instead, a heater was installed on- the - '_line section nearest the fuel pump and the restriction was cleared by - 3heating and then applying a differential Pressure toward the pump. bowl. The heater was left in place and connected to a pover supply for possible | future use, Tests showed that the overflow tank vent line was plugged in - a flanged section containing two valves. This was replaced and the plug - was found to be organic material,with very little, if any, salt located in & l/h*inch rort in one of the ‘valves,23 ‘A restriction again became detectable in the offgas line near the fuel -l'pump after another 2200 hours of salt circulation. This occurred on Decem- '5ber 8 1969, only four days before the scheduled final shutdown of the MERE. | '781nce the restrictlon was not,great enough to interfere with operation and o shutdown,'no effort was made to clear it. E?MSB Program Semiann. ETogr Rept., be 28 1969, 0RNL-h396, | pp. 143 - 1bs. 23R ETOgram Monthly'Report for July 1969, ORNL internal memorandum MSR 69-71 .10, 18 Possible Effects on Transfer to Overflow Tank The possibility of salt spraying or splashing intoathe top of the overflow line was considered in the pump design, and the baffles and shed roof over the overflaow 1ine were laid out to'prevent-this, - Observations in the transparent tank-of the water test indicated that the.baffles were effective, and there is no reason to believe that they were not effective in the MSRE pump. When the salt was first circulated in the MSRE fuel sys- tem,. there was only a very slow accumulation in the overflow tank when the indicated salt level was within 2.8 in. of the overflow point and none at lower levels, Thus the baffles were evidently pmeventing heavy splashing into the overflow line, If the accumulation had been due tofspray there would have been no reason for it to stop when the level reached a certaln proint, Therefore, there is no reason to attribute any of the transfer in 'this period of operation to spray, Neither can later response of transfer rates to changing fuel-pnmp level, beryllium additions, and other variables (including the striking difference between flush salt and fuel salt trans- fer rates) be explained by any hypothesis in which spray or mist is the . dominant mechanism, ‘ | N Although spray or mist can practically be ruled out as the primary _mechanism.for transfer, it is interesting to look at possible rates, Since the first beryllium addition in the #3%J startup, the rate of salt trans- fer to the overflow tank has normally averaged between L and 15 1b/hr. This 1s 50 to 180 in3/hr. The cross section of the overflow line is 2.04 in%® So the observed transfer rates would be equivalent to a "rain" of 25 to 90 in. of_liquid per hour, far more than the hardest torrential downpours. B o . How dense is the mist in the pump bowl? The appearance of the salt 'droplets on specimens exposed in the sampler_gives the 1mpression of a fairly light mist or heavy fog. Fog and mist concentrations are typically around 1078 g/cm3. This is consistent with the transport of-roughly 8 gram a month of salt mist out of the pump;bowl with tne Offgas, which is not inconsistent with the amounts of frozen salt beads found-in the off- . . gas line., On the other hand, attempts to measure concentrations of uranium and ®5Zr in the pump bowl gas indicated much higher mist concentrations i o0 ) .plg (vhere the samples vere taken). - Apperent concentrations in early sample ~ attempts ranged up to 10"%g/cm> (Ref. 2L4). If this were typical of the . offgas leaving the pump bowl, about 500 g of salt would be lost each day. No such loss actually occurred. In an effort to explain th1s discrepancy, " Nichols suggested that high concentrations might be sustained,in the pump bowl by electrokinetic phenomene and that in the offgas nozzle a large fraction of the salt partlcles would lose their charge anfi fall back into the pump.?5 If a similar situation existed in the mouth of the overflow 1line, what transfer rate could result? One pound per hour is all the salt from 80 liter/min of'gas:at"a:cOncentration of 107% g/cm3 This gas volume is already unreasonably high, but it would have to be even higher since later, more representatlve samples of pump bowl gas taken in d0uble-walled evacuated, freeze-valve capsules gave much lower salt mist concentrations, - on the order of lO‘sg/cm:.lehuS'it.is'evident,that the rate of gas‘trans- port into and out of the overflow line that would be required to produce a salt transfer of a few pounds per hour from mist fallout is entirely too “high to be plausible. | - | | '»rgonclusions We are reasonably confident of the following conclusions. l. There is a mist in the pump bowl, produced largely by the jets from the xenon stripper ring.. 2. The concentration of'salt droplets in the gas leaving ‘the pump 'bowl is in the range of ordinary mists. | 3. - Frozen mist particles in—the NEBE fuel-pump offgas contribute to ~ gradual restriction of the offgas line near the pump'bowl which requires ";cleanout at intervals of a few months to &8 year or 50. L, Sprasy and’ mist are not significant contributors to ‘the transfer 'u-into the overflow tank 24MS.R Program Semiann. PrOgr Rept., Aug. 31 1967, csNL h191, p. 131. =5, P, NlChOlS Ibssible Electrokinetic Phenomena in the‘MSRE Pump - ,Bowl GRNL Internal,Memorandum, MSR 68-10, January 3, 1968 20 . BUBBIES AND FOAM IN THE RUMP BOWL The xenon stripper jets déscribed in the preceding chapters drive about 50 gpm of salt intb the surface of the salt pool in the pump bowl, These streams carry under copious amounts of blanket gas. The gas bubbles | tend to coalesce and float back up to the surface, but there is a region in which the density of the fluid is significantly reduced by the presence of the gas, This région always exists, but it appears from the few db; servations that can bé made directly that the depth of the léw-density region and the density profile in it depend on several variables. The condition in the MSRE pump bowl has sometimes been referred to -1oosely as "foam", It appears that under some conditions, the density near the surface is low enough to justify this appellation; It should be noted, however, that "foam" has a connotation of persistence that is neither sup- ported by the MSRE observations nor appears likely from laboratory experi- ments with similar salts. | | | ' The primary evidence for the presence of bubbles or "foam" in the pump bowl comes from analysis of the salt level indicationé, The purpose of | this chapter is to present this evidence and to draw some inferences, Description of Bubbler Ievel Elements The pump-bowl level indicators, shown schematically in Fig. 5, are . -based on the principle of a pressure differential between a referenqe line connected to the gas space above the salt and a bubbler tube submerged(in the salt, Theré are two bubbie; tubes (596 and 593) extending to different depths. During construction_of'the pump, the distances from the bubbler “tips up to the plane'of the volute éefiterline were measured to;be 3.510 ifi. for bubbler 596 and 1.636 in. for bubbler 593. Thus if the bubblers were - both submerged in a pool of fluid of uniform density, the pressures in the bubbler tubes (and the differenfiialé between the bubblefs'and'the reference line) would differ by 1,874 in. times the density of the fluid. The dif- ferential preséures_are measured by electric d/p cells having.two ranges that can be selected electrically. The span that is used when fuel éait is in the system is 22.k4 in. fieo; the other span, 19.4 in, HZ0, is used 4P . ) | | o °§ ’L_qq&fl’_!q' s ;1 e R 21 ORNL-DWG 70-~5193 1.636 in. 1.874 in. © [ [——-—voLuTE ¢ ——Je }DENSITY ZONE 'Fig. 5. Schematic Refireéentéfiioh 6f;Fue1 Pump Bowl Level and'Density_Indicators | 22 when flush salt is.in'the system, These spané are equivalent to 10 inches of fluid having densities of 140 1b/ft> and 121 1b/ft> respeétively. Read- ;out'instruments are in percent of span., To maké'the levél indicated by'fhe ‘two instruments agree, a zero shift of 19% (nominally 1.9 in, of fluid) was added to the instrument on the shallower bubbler, Thus if the fluid in the pump bowl is of uniform density equal to that assumed in setting the span, the two instruments wofild read the same, namely the distance from the tip of the lower bubbler up to the surface, on a scale of 10% per inch, 1 The idealized situation described above does not exist and the level instruments consequently do not generally indicate exactly the same level, One set of reasons is associated with the inétrumentation. ‘Gas flowing through the three lines produces appreciable pressure drops'betweén the - d/p cell connections and the pump;bowl, The normal pressure dro?s'aré compensated in the zero settings of the instruments, If the zero settings should drift (as they do over long periods of time) by different amounts ° or if the pressure drops should change by different amounts (as they some- ~ times do because changes in pumt#boWl pressure affect the three flows dif- ferently), then a difference will appear in the indicated levels, .Another: reason for difference in the two level indications can be the fluid in the zone between the bubblers'having a density other than that used in setting " the instrument span. The flush salt density was close to the 121 1b/ft> used in setting that span but the fuel salt density was above 140 1b/£t3 when the 235U - 238 mixture was in the salt, less then 140 1b/£t> after the 233 was substituted, Furthermore, the density that affects the 4if- ference between the level indications is the average density of the fluid, - which is less than that of salt if there.are any -gas bubbles in thersalt; How the various factors mentioned above affect fhe indicated levels can best be understood by looking at some simple_reiations, P Iy = I3+hs— | Ps Lg = L2+h6—5;;_ # %) » LIV Lx-Lg = A~ ¥ vhere L is the indicated level, | I° is the instrument zero, or the level that would be indicated if ~ the salt were below the bubbler tip, ' A is the offset between the zeroes of the two 1nstruments, h . is the depth of submergence of the bubbler tip in the fluid, P is the average density of the fluid between the bubbler tip and the surface,- 'p*, is the density used in setting the instrument span, and subscripts 6 and 3 refer to. level elements 596 and 593 respectively Since there may be a non-uniform distribution of gas in the salt in the bowl, b5 1s not necessarily ‘the same as ps If however, one assumes that density profiles as &a function of depth are the same beside 593 and - 596 (at least down to the depth of the 593 tip) then hsfig'ra- h,p + (h6 - h,) P, where is the average density of the fluid in the zone ‘between the tips - of 593 and 596 - / r'iThe difference in’ level indications is then R he h6 - 'h-_5 2h Measurement of Absolute Densities It 1s evident from the last equation that the difference in level ifidications could be used to measure the density of the fluid between the two bubblers, If the instrument factors A and o are known exzctly, and the level readouts are precisé, the accuracy in the calculated density is limited only by the accuracy with which hg =~ hx is known, (Diétortion or - tilt of the pump could make it different from the measured 1,874 in.) In ,fact, level measurements‘made at times shortly after the instrument zeroes were checked, with the pump off so that the salt in the pump bowl contained virtually no gas, gave densities for the flush salt and fuel salts in reaf sonably good agreement with those predicted or observed by other methods (see Appendix B). Up to 10% discrepancy in the densities measured for quiescent salt were encountered at other times however, apparently due to drifts in the differential pressure instrument factors. '(A_lo% change in calculated density would result from & drift in A of 1.9% of scale.) | Measurement of Relative Densities Although the measurement of absolute density is compromised by the long-term zero drifts, the bubblers are a most useful-indiéatiqn of changes in density with operating conditions over reasonable periods of time. ~ The most striking short-term changes in density are produced by ‘starting and stopping the fuel pump., When the pump is;off the salt in the bowl is virtually free of gas bubbles and the density in the bubbler zone is that of pure salt. When the pump is turned on, the difference between the levels indicated by the bubblers changes within a few mlnutes to a new value, with the deeper bubbler indicating a lower level than the other. The indication is that the zone between the two bubblers produces a smaller increment of pressure at the 1éwer bubbler than it did when the bump fias off and the zone was filled with gas-free salt; that is, the average den- sity of the fluid in the zone is reduced by starting the pump. | Further changes occur as the pump continues to operate and the amount of salt in the bowl varies due to the slow‘transfer of salt to the over- flow tank and periodic recovery from the overflow tank. As the amount of 1) . m " i -salt and the actual level_change,the difference between the two bubbler 'leveljindications‘varies'somewhat,_with greater differences observed at lover levels, . 7 7 - The behavior described above is easily ‘understood, qualitatively at least. The jets from the spray ring, impinging on the salt pool carry . substantial amounts of gas below the surface, ?he fraction of the ges in - the salt decreases with de;fih until only the very small_hubhles that can be dragged down With.the'Saltlflow are present, If the low-density region - were entirely above theltipS"of.both_bubblers, the difference between the level indications would be the same &s with the pump off (although both ‘would indicate a level lower'than'the actual'top'of'the gas-liquid mix- ture.) If the low-densityzregionrextends into the bubbler zone, the dif- ference between the level’indications'reflects the density reduction due to the gas, As the amount of salt in the pump bowl decreases, the gassy 3region moves down and more gas appears in the bubbler zone. An implication of this picture is that the density profile should change with pump speed. That is, at reduced speed and head ‘the velocity f,‘of the jets and the net downward velocity in the salt pool are less, S0 -there should be less gas at a given depth This was proved to be the case in experimental operation of the fuel pump at various rotational speeds in Runs 17 - 19 (February - Sermember, 1969) Figure 6 shows the effect of pump speed on-the apparent void frace- r'tion in the zone between the -two bubbler tips, (The errors due to zero *‘rshifts are cancelled in the calculation of void fraction or density rela- :(ftive to that of clear salt ) The data for this plot were ‘a1l obtained -_'with approx1mately the same 1ndicated salt 1evel The results clearly _'(;show the greater depth of the low—density region or- "gassy zone" at higher B pump speeds and higher salt jet velocities. o The points in Fig.‘6 also show that in this particular parameter, any '”difference due to changing from helium to argon. cover gas is not discern- 'Ti ible and the differences between fuel salt and flush salt have only & minor p'effect ‘Tt will be shown later (Chapter 6), however that vhen the gassy 'region extends sufficiently deep in the pump'bowl very small changes can . produce pronounced effects in the circulating loop. 25, PUMP VOID FRACTION (%) ORNL—-DWG 69-10543 9¢ 'SALT COVER GAS 0 FLUSH He 20 ® FLUSH Ar A FUEL He - Ao FUEL Ar A 15F o o 10 lE ' o 1o A . - | A Ao - | & 0 L—@ n- ' : — ‘ 700 800 900 1000 {1100 1200 1300 FUEL PUMP SPEED '(rpm) Fig. 6 Effect of Fuel Pump Speed on Void Fraction in Zone: Between Two Bubblers in the Fuel Pump ' , 0 ) #y 27 Variation of Void Fraction with Depth It is quite.clear that'there is-a‘substantial void fraction in the. . pump bowl, Its variatlon w1th salt depth (i.e. the density profile) is of interest but is not 50 clearly determinable, With some approximations and other observations, however, we can use the same data to develop at least a qualitative description. Because of the approximate nature of this treatnent, e will_applytit only to the normal flow condition. . Figure T shows results;obtained in Runs 15 and 16 (October - December, '.1968) which indicated the correlation between bubbler difference (void 'fraction)'and”pump-bowl 1eve1 already alluded to. This‘correlatiOn was approximated by a straight line obtained by least-squares treatment of the data, If we extrapolate this line to the bubbler difference for bubble- free, stationary salt (1.58% at ‘the time of these measurements) we obtain _the indicated salt level above which substantial bubbles. would not appear in the zone between the bubblers. This level is T7% on the shallow bub- bler (IE-593), which corresponds to 5 3 in, of clear salt above the tip of this bubbler. If we then assume that the total thickness of the bubble zone is in- 'Vdependent of pump level this zone will reach down to the deeper bubbler tip when that bubbler indicates 5.3 in. of salt (56. 14). At this salt | depth the correlation from Fig. 7 gives a reading of 61.5% on the shallow bubbler.r At this level, the average void fraction between the bubblers would ‘be 20%. o o | | Data on the rate of salt transfer to the overflow tank during the time it-of the above level observations showed 8 substantial drop when the - 1nd1ca- "} ted level on IE- 593 fell below ~ 60% This change may be associated with - a drop in the level of the top of the foam 10 a p01nt Just below the over- ~ flow pire. Using the level indication and the distance from the bubbler %o the overflow point (6.5 in.) we obtain an average void fraction of 6% ~in that zone, At the same. eondition, the average void fractlon between drthe bubblers is 259, Transfer ‘rates were relatively uniform at indicated "levels between 60 and 69% on IR~ 593. However, above this range much higher rates prevailed. This seems to suggest a region of very low density 28 ORNL-DWG 70— 5194 7 0—0-0—6 oo BUBBLER DIFFERENCE (%) . 3 55 A €0 ‘ €5 | 70 INDICATED SALT LEVEL, LE~593 (%) ' Fig. 7. Effect of Indicated Salt.Level in Fuel Fump on Bubbler Difference oY ” o v 29 material (fosm) &t the'top'of the”salt'with a region of higher density lrf_.belOW'it - The ‘very high transfer rates would then be assoclated with overflow of salt from the intermediate region, The several pieces of ‘information presented above were'used to gen— erate a void-fraction profile in the pump (Fig. 8). The characteristics ~ of this profile satisfy the behavior described above but it must still be regarded as only an educated guess i.e., a profile of this general shape Vrmust ex1st but the values assigned to it are relatively crude. This analy31s was carried one step further to indicate the relatlon between the actual top of the salt and the internal structure of the pump. | Figure 9 shows ~the "true"_level as a functlon of the indication on the ~ deeper bubbler (IE-596). This suggests that the thickness of the gassy zone is not as great as indicatedain'Fig. 8' or at least not uniformly this great S Ifiv were, the offgas line would normally ‘be submerged in ~the foam and substantial salt transport to the offgas would occur. It is, of course, possible that the hardware in the gas space’ could depress the - foam in some areas but a more’ likely explanation is that the thickness is oVerestimated Nevertheless 'it'seems clear that a substantial bubble head does exist under normal pumping conditlons and that this layer could 'easily reach as high as the overflow pipe. Effects'on Reactor Operation The existence of the gassy reglon in the pump bowl affects the opera- :_-:ition of the MSRE in several'ways._ The sometimes fairly rapid transfer of -7-:fdsalt into the overflow tank has as its most likely explanation the mount- ”lfing of the foam 1evel to the top of the pipe.‘ The taking of samples ‘and ?lthe exposure of materlals in the sampler cage were affected by the increase. ' Lfin 1evel due to the gas.' Most importantly the gas in the pump'bowl intro- u .duced gas into the clrculating loop. These effects will_be discussed in .Jlflater chapters o T T R e e VOID FRACTION IN SALT (%) 4100 - 60 40 20 0 30 ORNL-DWG 70-5495 | - \ \ \ ; \\ g \ 0 2 4 6 8 DISTANCE BELOW SALT INTERFACE (in.) 10 Fig. 8. Estimated Void-Fraction Profile in Fuel-Rump Bowl 2 n " " 30 . 40 50 60 70 ACTUAL LEVEL (in. above volute ¢) a L Fig- 9.7 e | o o 31 ORNL-DWG 70-5496 INLET OF / L OFFGAS LINE | 7o F— OVERFLOW lNDICATED LEVEL LE~- 596 (%) Actusl Salt Level as a mnction of Indicated Level from Estimated Dens ity Profile 32 5. TRANSFER TO THE OVERFLOW TANK The more or less continuous transfer of salt from the pump bowl into - the overflow tank has been alluded to in the discussions of mist and foam in Chapters 3 and L, This chaptef rresents the data on transfer rates, The most significant point about the transfer is that it océurs at all when the punp is operated, as'in usually is, with the‘indicated sait level from 2 to'h inches below the open top of the overflow line. ‘This is what first drew attention to the conditions which are the subject of this re- port, Other intriguing features of the transfer fate.behavior that will _ be brought out are as follows: | | | 1. . the dependence {or sometimes the lack thereof) of the transfer rate on indicated salt level, 2. the shift in transfer rate from one interval of accumulation to the next, | | 3. the changes that have occurred very gradually or between runs, hQ the comparison of rates W1th flush salt and with fuel salt 5. the effect of beryllium additions, and - ' 6. the effect of salt pump head (speed). Initial Observations Tfansfer to the overflow tank was first observed during prenuclear testing with flush salt in February 1965 (Ref. 26). After overflow during a deliberate overfill had been observed to oceur at an indicated level of 92% with the pump off (as predicted), the operating level was set at T70%. - At thils indicated level, howéver salt transferred to the overflow tank at a rate of about 0.7 1b/hr during pump operation, The transfer stopped at an indicated pump level of 6&% and in continued operation at that level or below for another two weeks there was no measurable transfer Fuel - salt was circulated in May and June, 1965 for a total of 1000 hr, always 26MsR Program Semienn. Progr. Rept., Feb. 28, 1965, CRNL-3812, p. 1h. C " n " " 5 33 with the pump-bowl level at or below the "threshold" of 644, which was then prescribed as the maximim normal operating level. Over the 1000 hours only 86 1b .of salt accumulated in the overflow tank.&7 Ekperiencerwith Fuel Salt When operation:resumed'in December 1965, the-raterof fuel salt trans- '-'_fer was still very low.- As7shown in Fig. 10, the rates observed‘through February, 1966 are practically zero, - This figure shows that as operation continued, however, a measurable transfer rate developed, increased gradu- " ally, and appeared to level:out near 1 1b/hr. After about a year in this range, the transfer rates shifted downward by about a factor of two and " remained there for the final six months of 235y operation, Then, just after the resumption of fuel circulation folloW1ng salt processing and Jinitlal loading of 233U there was a drastic change in behavior involving much higher transfer rates than ever seen before. _ The transfer rates that are plotted in Figs, 10 - 12 were computed from rates of rise of overflow tank level. Rates based on the decrease in pump-bowl level are in general agreement but are less accurate because of effects of changes in loop temperature -and gas, volume, The rates shown are averages measured from one overflow tank emptying to the next, so each covers a range of fuel—pump levels. A remarkable '-'fact is that through Run 14 the rate in any interval of overflow accumu- ~ lation (usually 1 to h days) was practically independent of pump4b0w1 level - between the normal operating 11mits of 644 and 50% It is also remarkable that although the rate was practically constant between any o emptyings | -,:of the overflow tank it rarely was the same in any two succe381ve intervals. There is no cause known for the very noticeable shift in rates be- - tween Runs 12 and 13 shown clearly in Fig. 11. The rates 1n Run 12 were | "slightly over 1 Ib/hr and had been relatively steady for some time. f-’Throughout the 3 days of fuel circulation in Run. l3 and the 6 months in | Z7MSR Progrem Semiann. Progr. Rept., Aug. 31, 1965, CRNL-3872, p. 16. 100 50 - N o o o, N FUEL SALT TRANSFER RATE (1b/hd) o o 0.2 0.1 JAN ORNL~DWG TO-5197R 5 RUN NUMBER FEB MAR = APR MAY JUNE - JULY AUG SEPT - OCT NOV DEC 1966 : o Fig. 10. Measured Rates of Salt Overflow from Fuel Pump — 1966 S’ 1e 100 ' ORNL-DWG 70-5198R E BERYLLIUM ADDITIONS .50 20 10 RUN NUMBER 'FUEL SALT TRANSFER RATE. (Ib/hr) 05 o2 o L 1 _ 1 _ S 7V UAN . FEB. MAR APR MAY JUNE JULY AUG. SEPT OCT NOV DEC T 1967 | - - ‘ F_i‘g;.:..ll.- Measured Rates of Salt Overflow from Fuel Pump - 1967 GE 100 | IUM ADDITIONS 50 n O o 14 RUN NUMBER FUEL SALT TRANSFER RATE (1b/ hr) O o 0.2 © 1968 JAN FEB . MAR APR . MAY JUNE JULY AUG SEPT oCT. 9.4 'Fig., 12. Measured Rates of Salt Overflow from Fuel Pump — 1968 A ORNL=DWG 70-5199. NOV . DEC 9 5 bkl L S et et et v S 1 b . € AR 4 2 8151 AT 8 01 b [)] n » 0 ¥ 37 »Run!lh, however,,the observed fuel salt transfer rates were consistently ~ lower by about & factOr-of;two., Otherwise the characteristics of the transfer were unchanged, that is, there was virtually no dependence on ' pumpcbowl level in the operating range. "~ The whole behavior of the fuel transfer changed drastically on Sep- tember 15, 1968, during a beryllium exposure a few hours after the begin- - ning of fuel c1rculation following the long shutdown for salt processing. As shown in Fig. 12 some very. high rates were measured in October and November, As will be described in Chapter T, the highest rates occurred only Just after some additions of beryllium. Because of the apparent cor- relation with beryllium additions subsequent to the salt processing and 233y 1oading, the times and amounts (in grams) of all ‘beryllium additlons »through 1968 are 1ndicated on Figs. 10 - 12, Some of the effects of the ,beryllium were only temporary, ‘but the transfer rate remained s1gnif1cantly higher after the firstrberyllium addition in Run 15 than it had been before, Anotherrdifference,was}that the rate became strongly dependent on pump-bowl level, The continuous variation of overflow rate with level makes a continuous chronological plot of the rate in Run 15 and thereafter impractical The transfer rates plotted in Fig. 12 for this period simply -indicate the range of rates that were encountered, Usually the rate de- creased from 10 - 1k lb/hr tol -2 lb/hr on each cycle of salt transfer to the overflow tank and return to the pamp bowl, ~This level dependence ~persisted through. the remainder of the MSRE operation, but 1t was found ) - to vary with time. Figure 13 shows the effect of indicated fuel-pump . level on transfer rate'for two time periods separated by about 1 year. :f-flSmall Zero shifts in the fuel-pump level element (l - 2%) cause some un- .certainty in the relative positions of the two sets of data, but these do not affect dependence of the overflow rate on level - The slopes of the o two sets of data differ‘by about a factor of 2. Norreason{haS'been deter- | ;fmined for this change.-*rjf}j*,~~3 38 * ORNL—DWG 70-4807 14 — — / , ® o OCT-NOV, 1568 ® OCT-DEC, 1969 12 Y o FUEL SALT OVERFLOW -RATE (Ib/hr) 45 50 55 60 65 70 INDICATED FUEL PUMP LEVEL, LR-596 (%) . Fig. 13. Effect of Indicated Fuel-Pump Level on Salt Transfer to the | - Overflow Tank During 233y Operation | ) ¥ [ 1] » » 39 Experience with Flush Salt During the b-year period from 1966 throuéh-l969, there were 9 occasions | ‘on which flush salt masicirculated for sufficient time to provide meaning- ful data on theiraterof;transfer of this salt to the overflow tank. These data are summarized in Table 1, along with the reagon for flushing and the : major'events that preeeded;the flush. The transfer rates through'June, l967;arerpractically thersame as those observed with fuel salt., However, the 4 measurements over the 6 days of flush-salt. circulation in September- _1967 gave values that were 31gn1f1cantly higher than the prev1ous values ~with fuel salt, Even more astonishing is the comparison of these high flush-salt rates with'subsequent fuel transfer rates which were a factor of 2 lower than those before the flush, The later flush-salt transfer _rates (with the possible exception of the final measurement) were again reaSOnable consistent'with,the early fuel-salt data, The flush-salt rates never reached the very high. values experienced with the 233 fuel salt. The precision for the final flush-salt transfer measurement is poor, but there is some suggestion of an unusually high transfer rate. Relation to Other Operating Varisbles High transfer rates went with high bubble fractions in the pump bowl. .In experiments-in Runs 17”- l9iwhen the fuel pump was oberated at‘lower ' tspeeds to reduce. the amount of gas churned into the salt, the overflow i;rate showed a. striking drop.z For example in September, 1969, the fuel ' ,f'salt transfer rate went from 3 h 8. h lb/hr with the pump at 1189 rpn to 0.k - 1, O lb/hr with the pump at 600 rpm. This is, of course, consistent afwith the hypothesis advanced 1n Chapter L, that the transfer is caused fih:by "foam" rising to. the top of the overflow pipe.__‘ Co ' | | There was never any detectable transfer with the pump off (except . -:during deliberate overflows) This implies that there was no leak into ; r;the overflow line below the surface of the salt pool The shifts in transfer rate between success1ve 1ntervals of overflow accumulation cannot be correlated with any other observable change. ko Tsble 1 Flush Salt Transfer to Overflow Tank Measured S S Prior -+ Transfer Rate ~ Time Interval Occasion Activity (1v/hr) 9/25/66 - 10/3/66 Start Run 8 Replace gore - 0.47,0.81 - | . sample assenbly | | 11/2/66 - 11/5/66 End Run 8 - Reactor operation 0.78 - 12/11/66-12/12/66 = Start Run 10 Rod out fuel pump’ 1.16 ' | ; S offgas line - 5/11/67 - 5/12/67 End Run 11 Reactor operation 1,20 ~ 6/16/67 - 6/17/67 Start Run 12 Replace core - 1.50 ' ' Sample assembly : _ 9/8/67 - 9/14/6T Start Run 13 Retrieve sampler 1.24,2,58,1.98, \ - latch 3.20 8/14/68 - 8/16/68 Sample flush U-recovery from xk . 0,46 : salt for U flush salt, replace \ core sample assembly 8/11/69 - 8/15/69 Start Run 19, Replace core - 1.17,2.24,1.35 cover gas ~ sample assembly , experiments ' | 12/13/69-12/14/69 End Run 20, Reactor operation, 3.6 Final flush leak *_ | | Low confidence in this value because of variations in system . gverpressure, : ' o Ll ¥t ) [} I " 1) no :'tEffects'on Operation In order to place the matter of salt transfer in proper perspective, ' it should be noted that although the transfer was an unexpected phenomenon' it caused fev problems in the operation of the reactor. Normally, it was :a simple matter to pressurize the overflow tank and return salt to the li_dfuel pump - to maintain the pump-bowl level in the desired range. The only - period when this recovery operation was much trouble vas in Run 9 (November - 1966) while the main reactor offgas 1ine (L~522) was completely Plugged :‘near the fuel pump, Since all ‘the offgas ‘then flowed out through the over- flow- tank the act of pressurizing the overflow tank blocked the reactor offgas., This required that the reactor be made suberitical and the fuel pump stopped 50 that the helium flow into the pump could be stopped while - salt was being returned from the overflow tank This rather awkward mode of operation was adopted only as’ a temporary expedient and - the system was _soon shut. down to clear the main offgas line, 42 6. GAS IN THE CIRCULATING LOOP | 'From the evidence describéd in Chapter k4, it is clear”that iarge quantities of gas are driven intb-the salt in the pump bowl. ~Some of this gas is drawn into the circulating loop with the 65 gm of salt that flows from the_pump bowl into the pump suction., This was dbServedfin'the pump development tests and was therefore expected in the MSRE. This chapter deScribes the ways in which gas has been detected and measured in the de- velopment loops and in the MSRE, then goes on to present the results of the observations over the years ofIMSRE operation. Indicators - There are several independent indicators of gas in the MSRE fuel ioop. _‘One'used only before the beginning of power operation.fias:a densitométer based on gamma-ray penetrétion. Changes ifi pump-bowl level attending cir- culation, behavior during pressure-release experiments, accumulatlon in the access nozzle annulus, direct effects on reactivity, effects on xenon stripping, and effects on neutron noise are indicators useful durlng later_' operation, Densitometer' Early in the pump development program it was rec0gnized that some of the gas churned into the salt by the xenon stripper spray would enter the circulating'loop and a program was started to measure the density of cir- culating fluid by gamma-ray attenuation.®® Eventually, a sensitive, sta- ble densitometer was developed, using a 40-curie 37Cs source and an elec- tron multiplier phototube positioned on opposite sides of the 5-inch salt pipe.#® This device of course could not be used in the reactor after the beginning of power operation because of the very high gamma radiation from 2SMER Program Semiann, Progr. Rept., July 31, 1963, CRNL-3529, P. 50. | 29MSR Program Semiann, Progr. Rept., Feb. 28, 1965, (RNL-3812, pp. 51 - 52, 9 » . B " o L3 the salt itself Results obtained with the densitometer under various '-conditions in the pump test facility and in the prre-pover testing of the MSRE are described later in this chapter under "Experience". '.Egmp Bowl Leve;,Changes- :Theindicated salt ievelrin;the pump - tank decreases when the pump is started and rises when the pump is stopped. As"explained in Chapter 2, ‘part of this difference is because salt fills some parts of the pump above the level of the main body of salt only when the pump is running, However, the level change on starting or stopping the fuel pump has always been ' greater than could be accounted for by salt holdup in these regions, At :least part of this excess volume change occurs immedlately upon a pump ~ start or stop, (even on the first start after £1lling the loop with salt ‘that should be free of bubbles), evidently as salt moves into and out of a trapped gas volume when'the pump-head and the pressures around the loop change, FPossible candidates for such volumes are the spaces between gra- =phite stringers and the annuli at the reactor access nozzle.. When circulating bubbles are introduced by starting the pump, they dlsplace salt from the loop 1nto the pamp tank cau81ng a level rise, If there are bubbles when the pump is stopped and the loop pressures decrease, their expansion causeS‘more_of an immediate level rise followed by a slow level decrease as the gas'findsfits way from the loop into the pump, If while the pump is running there is 'a sudden change in the rate at which --sgas 1s being drawn- into the 1oop, there should follow & change in average ~loop bubble fraction (and pump-bowl level) with a time constant about equal .f:to 520 gal/65 gpm. 8 min. Changes with tlme constants almost this short ;"have, in fact, been. observed 7n fV,fIressure-Release Expgriment | The level change produced by a sudden change in pumpebowl pressure fi";would seem to be a sim;fle, direct indication of the amount of bubbles in i:the loop., of course, as 1ndicated above, compressible trapped pockets of gas would have the same effect as clrculatlng bubbles and would be included in the calculated volume_of gas. But what really complicates interpretation Ly of pressure-step experiments.is the presence of gas in the pores of. the ficore graphite. Salt is not moved into or out of these pores to any ap- preciable.extent by pressure changes of the attainable magnitude; gas is; but the rate varies widely depending on the nature of theipressure change. On a suddefi pressure inéreése the pores would have very little immediéte effect: although gas would begin to flow into the pofés, the rate would é | ~ be limited by the transport from the salt stream to the graphite.surfaces; : In addition, the total amount of gas in the graphite is iimited_by the partial preséure in the flowing_liquid.rr(This can be an important factor if the salt in the core is not saturated with gas.) However, after the gas - pressure-in the pores has equilibrated with that in the salt, a sudden decrease in system pressure that leaves the total pressure in the core . ‘liquid below this gas pressure will éllow gas to fizz from'the_graphite - pores into the salt, This would have the same effect on pump-bowi leVel as the expansion of a large volume of circulating bubbles. The effect can, in some cases, be further magnified by gas coming out of solution in the salt. Thus a pressure-release test tends to be misleading if the .graphite pores have been charged with extra gas before thé release, Pressure-release tests, although not an accurate measure of circula- ‘ting void fraction, are believed to provide an indication of whether there are some bubbles circulating.through the core or none. In & typical test the pressure is brought up by about 10 psi over a period of an hour or so, then is dropped back in a few seconds to near the 6riginal‘pressure. Since'gas transport between bubbles, 1iquid, and graphite 1n;the MSRE is quite rapid, it is likely that near-equilibrium conditions prevail except during and just after the release. Thus the important factor in a pressure - release is the gas partiasl pressure in relastion to the liquid static pres- sure in the cofe, | When there are circulating voids throfighout the core, the gas partial rressure and thg liquid stati; pressure are approximately equal. (The gas pressure may even be preater if the bubbles are small enough to be affected by the surface tension of the liquid.). In this case, any pressure release . will cause gas to escape from the graphite and the amount that escapes is prdportional to the pressure change. w » ) n o) » b5 At the opposite'extreme'is the case-where the amount of,available gas is so low that it is completely dissolved in the liquid with a partial irpressure much lower than the liquid static pressure, Again-'the gas pres- f 1sure in the graphite will be near the gas partial pressure in the liquid In this case no gas will escape from “the graphite (and there will be no - associated pumpclevel or reactivity effect) until the static pressure is '_reduced below’ this gas partial pressure. Since the usual pressure release was only about 10 psi, no excess gas would be expected to appear in the *circulating stream under the conditions described in this paragraph Con- versely, the absence of_such“gas would indicate rather conclusively‘that | . these cbnditibns did prevail;dl Iet us now'consider the-intermediate case where sufficient gas is " available so that the liqfiideould be very nearly saturated at equilibrium, The gas pressure in the graphite would then exceed the 1iquid Pressure : after the release and gas bubbles could appear in c1rculation even though - none had been present before.. However, the NEEE 1s a dynamic system and, to reach this condition, gas 1ntroduced at the pump suction ‘would have to dissolve before the liquid;arrived at-the_core.‘ The rate of gas dissolution tends to decrease as saturation is'approached .and it is strongly influenced by bubble diameter.3° (For bubbles larger than ~ 2 mils in dismeter, the available time is insufflcient for complete dissolution.) Thus, it is not, pos51ble to attain the ‘no-void condition in the core with the liquid nearly | saturated However, the void fraction would be quite low., The response 'jof the system to a pressure release would thus accurately reflect the 'ofhpresence of voids. . The principal value of the pmessure release is in distinguishing be-- -']tween the complete absence of voids and the presence ‘of only a few. Higher 'f:void fractions (>0. 2% or so) are readily measured by their level and reac- . gthty effects. | ~ 3%4eR Program Semiann. Progr. Rept., Aug. 31, 1969, CRNL-bikg, pp. 8 - lo. L6 Access Nozzle Annulus - Variations in temperatures in the reactor access nozzle annulus also afford some information on the presence or sbsence of circulating gas bub- bles in the fuel loop. In the Engineéring Test Loop (where theré‘were no entrained bubbles) helium trapped in the annulus was gradually removed by dissolution in the salt circulating pastfthe lower end.?l' Less—soiuble argon was removed much more slowly.. Therefore provisions were made for .freezing salt in the annulus to prevent molten salt from rising to the flange seal at the top of the annulus, When the MSRE was started up, how- ever, it proved difficult, if not impossible, to freeze a dependable plug of salt low in the annulus,>2 Nevertheless, the thermocouple readings did indicate that molten salt did not rise very high in the annnlus, suggesting that some mechanism was maintaining the amount of trapped helium, Subse- quently it was observed that the salt level inferred frnm’the temperatures varied and that the level varistions were correlated with variations in circulating bubbles indicated by other, independent evidence.?” It clearly appeared that collection of circulating bubbles was delivering gas to the annulus, dfiving the level down closer to the circulating stream until the removal by dissolution increased to balance the input by the bubbles. The - equilibrium level then afforded a semi-quantitative indication of the rate of separation of bubbles from the circulating stream. Direct effect on Reactivity Gas in the core reduces the amount of fuel and thus‘decreases the re- | activity. Hence during nuclear operation a change in the bubble fraction - in the core must be accompanied by a change in some other véfiable to keep the reactor Just critical, The'dénsity'coefficient of reactivity of the 2355 fyel salt was 0.18; for the 237U fuel it is 0.45. That is, a 1% de- crease in fluid density (an‘increase of 1% in bubble fraction in the salt 31MR Program Semiann, Progr. Rept., July 31, 1963, ORNL-3529, p. 4l. 33MSR Program Semiann. Progr. Rept., Feb. 28, 1965, CRNL-3812, p. 9. 3MSR Program Semisnn. Progr. Rept., Feb. 29, 1968, ORNL-L254, pp. 7 - ) wn [ )] b7 in the core) produces.a O.fi5%-decreaSe in reactivity. This is a very use- ‘ful end precise indleator, but it has its limitations: it can be used “,;only when the reactor is critical and it may be confused hy concurrent changes in reactivity from other causes. A distinction is that it measures ':.an average void fraction in the core rather than the'density at a point, as measured by the densitometer, or the average in the whole loop indicated by the pump-bowl level changes. - o Xenon Stripping The presence of blanket gas bubbles circulating with the fuel salt strongly affects the fate of the 135Xe produced during power operation. A few bubbles dispersed-in_the.circulating salt_can contain far more of the low~solubility xenonrthan'all the salt, and provide an avenue by which the xénon can escape fronlthe'salt into the offgas stream. "On the other hand, bubbles of xenon-ladenigaSVchurned into the salt.in the pump bowl reintroduce xenon., The;partialfipressure:of the xenon can increase drasti- cally in a. bubble if'most'of‘the diluent gas goes into solution after com- pression in going through the pump. ‘Helium and argon behave differently in this respect. In summary, the xenon poisoning is sensitive to the bub- ble fraction in the fuel but the complexity of its dependence severely ”i.limits its usefulness as a clear indicator._ Noise Anal sis B When the reactor is critical the ;resence of compress1ble gas in the ‘ Ecore affects the observable fluctuations in neutron flux. Thus there is. potential information on’ the bubble fraction 1n the neutron flux noise, 7'_;and sophisticated noise analysis techniques have been developed to take 'f'advantage of this. | | | "Noise Analysis," as used here, refers to the examination ‘of the '=:frequency ‘spectrum of small statistical variations in the neutron flux (or "f pover 1evel) of the reactor.; The flux variations. result from small reac- -7rtivity perturbations which are, in: turn, caused by small variations in ~ other ‘system parameters For example, if there are circulating voids in - the MSRE, the effective fuel-salt density and, hence, the reactivity will 48 be affeéted by variations in core pressure. Analytical studies34 showed that the dbsolute amplitude of the noise spectrum should be pmoportional to the square of the circulating void fraction. Although this relatibnsfiip nominally-applies to the entire noise spectrum, the spectrum. itself may be more sensitive to voids at one or more frequencies either because of the spectrum of the input (pressure) disturbance or because of the response (gain)-of the reactor system aé a function of frequency. Measurements of the inherent noise spectrum in thé MSRE showla peak‘near 1 Hz that appears to be closely related to the circulating void fraction. Changes in the void fraction.have produced pronounced changes in the noise spectrum, par- ticulafly at 1 Hz and, in fact, the power spectral density around this frequency was shown to vary approximatély as the square of thevvoid frac- tion, as jmedicted by the analytiéal model.>> Based on these results an instrument was built td give a direct readout firoportional t6 the root- mean-square of the neutron noise around 1 Hz. This instrument, located in the reactor control room, is & convenient and immediate indicator of changes in the fuel void fraction. It is affected to'somelextent by changes in other variables, however, and must be calibrated by some other indepen- dent measurement of void fraction. An experimenfal measurement of an absolute value for the void fraction ‘can be obtained by a methbd véry closely related to noise analysis, This is the "sawtooth pressure" experiment.”® The pump-bowl pressure is cycled on a LoO~sec period by opening and closing a valve to vent gas to a drain tank, Taped records of the pressure and neutron flux signals are then ana- lyzed to obtain the actual neutron-flux-to-pressure freguency-response funétidn. The equivalent freéuency-response function of the analytical model is adjusted to the best fit by varying the void fraction assumed in the model. | 34p, N. Fry, R. C. Kryter, and J. C. Roblnson, Measurement of Helium Void Fraction in the MSRE Fuel Salt Using Neutron-Noise Analysis, C(RNL- T™-2315 (August 1968) 35Ib1d 367, C. Robinson and D, N. Fry, Determination of the Void Fraction in ‘the MSRE Using Small Induced Pressure Pbrturbations, ORNL-TM-2318 (beruary 1969) . | | 9 .} . I . was turned off and .on at 1150°F and reduced level, there was some indica- 'zero-power experiments in June and July. After the dens1tometer showed ”17 no detectable voids in the loop during pumprstop experiments we tried to : detect the presence of voids by varying the fuel-system overpressure with h9' '1'Ex;§rience ‘The first dependable measurements of circulating,void fraCtion in'the pump test facility were obtained in 1965. (Earlier measurements were ques- tionable because of inadequacies in the densitometer.) With a nominal 69% . level in the pump bowl, the densitometer'showed 4.6 vol'% voids in the f loop, when the level was raised to 79% (submerging the Jets) the void frac- tion in the loop was reduced to 1.7% (Ref 37). In these tests, however, the pump had a 13-inch impeller which sent 1615 gpm through the loop and - 85 gpm through the stripper,. MEasurements in 1966, with an 11.5-inch im- peller (as in the MSRE pump)_showed only 0.1 vol % voids in the loop.>® Flush salt was first circulated in'the MSRE fuel loop in January 1965. On several occasions in February and March the pump-bowl level rise when the pump was stopped 1ndicated some compressible gas in the loop., There was more on some occasions,than on others, indicating some bubbles. On '_March 5.the'temperature.uas'reduced to0 1030°F to bring the level down from 70 to 50% and there was clear evidence of bubble ingestion in the lower range of levels and temperatures. | The fuel carrier salt, with 0.6 mole % depleted uranium in it was. ' circulated in May 1965 The densitometer, which had been moved from the punp test facility and installed on the line between the heat exchanger and the reactor vessel showed no detectable bubbles (<0.076 vol%) when ~the pump was started and stopped under normal conditions.39 When the pump tion of bubbles. s , Attempts to measure a circulating void fraction continued during ‘the pp- 62 - 65. . 3fiMSR Irogram Semiann. PTOgr Repm., Aug. 31 1965, ORNL 3872 3R Program Semienn. PTOgr Rept., Aug. 31 1966 ORNL hO3T, p. 81. 39MSR Program Semiann. Progr Rept., Aug. 31, 1965, ORNL-3872 p. 65, 50 the salt circulating. Three tests were performed in which the overpres- sure was first increased tollO‘- 15 psig and then rapidly decreased to ~ 5 psig (Ref. L40). We reasoned that the rapid pressure decrease would allow any circulating voids to expand and.produée an observable change in ~the densitometer reading. The first two tests were carried out at normal fuel-system temperature (~ 1200°F)5with the normal operating level (~ 60%) of salt in the fuel-pump bowl. Tn neither case was any changerdbServed, either-in the densitometer reading or in other system parafieters that should have responded to a change in salt density. The third test was performed at an abnormally low pump-bowl level (~ 50%) - that was obtained by lowering the salt temperature to lQ509F. This time the densitometer responded‘dramaticaliy to the.rapid pressuré decrease,-indicating a de- crease in salt density or an increase in the circulating void fraction. In addition the pump-bowl level increased and the system reactivity de-. 'créased, lending further supporf to the densitdmeter evidence. At first we (erroneously) assigned all of the observed effects to simple expansion of gas already in circulation. Evaluation on this basis indicated that the circulating void fraction was 2 - 3 vol %. As other evidence was accumulated, we concluded,thaf_much of the'gés that appeared in circulation immediately after a pressure release came from,a-h%n- circulating reservoirw—'prbbably the pores of the graphite — and that the steady-state circulating void fraction was actually much smaller. (See discussion on pp. 43 - 45.) : B The densitometer was removed after the initial zero-power experiments as part of the preparation for reactor operation at power, ‘(Since its operation was based on a collimated y-ray beam and & high-sensitivity detector, it could not be used in the high radiatibn fields produced by povwer operation of the reactor.) The most important conclusion to be drawn from the densitometer étudiés is that; at normal loop cdnditions, the circulating void fraction in the MSRE initially was near zero. 401pid, pp. 22 - 23, " » [ b " » 51 On the basis of this informatlon, the early power operation was com-. '_pared with the expected behavior with no circulating voids. The most ob- vious disagreement between the predicted and observed behav1or was in the 135Xe Ppoisoning. A model used to describe xenon behavior with no bubbles o predlcted a Xenon p01soning of 1 08% Sk/keat full power while the observed - poisoning was only 0.3 to O, h% Reevaluation of the mathematical model indicated that the low poisoning values could be explained only if circu- lating voids were present to greatly enhance xenon removal to the offgas system. In addition, each tlme the overflow. tank wvas emptied, small re- activity perturbations appeared that raised some suspicions about cireu~- lating voids, - Detailed analyses41 showed that an average loop void frac- tion of 0.1 to 0. 15 vol % vith a high probability of exchange in the pump" “bowl (50 to 100% per pass) was required to explain both the steady-state 'iand transient behavior of xenon in the reactor.n As a result of the above observations a series of pressure-release | experiments was performed with fuel salt in July 1966 to see if any direct -evidence could be found for the postulated voids.*® Six tests were per- : formed and substantial voids were observed in each, Although the quanti- , tative evaluations that were made then are now believed to be incorrect, there was no doubt that the. behavior was markedly different from that ob- served in earlier tests, (It may be of incidental interest to note that the change in circulating void fraction occurred during the same general time period in which significant salt transfer to the overflovw tank was -'developing, see Fig. 9,) - Two additional sets of pressure-release experiments were performed '?in October 1966. The first set was performed with flush salt and none of . .6 tests showed any ev1dence of circulating v01ds._ The second set with . fuel salt ‘showed c1rculating voids in each of 6 tests. -~ Another experiment was - also performed in an effort to obtain a quanti- | f.ftative indication of ‘the- circulating void fraction.43 When the reactor was 41MSR Ir0gram Semiann. Progr. Rept., Aug. 31 1966 ORNL h037, pp. 13-21. 4271pid, pp. 22 - 2k, | - 4R Program Semiann. Irogr. Rept., Feb. 28, 1967, ORNL hll9, p. 17 52 filled with fuel salt for the start of Run 8 (October 1966), the salt had been in the drain tanks for 11 weeks and should have been free of all voids. This was verified by observing the lack of compressibility-of the loop con- . tents immediately after filling. We then measured critical control-rod configurations with the fuel pump off snd after it had been on for some time, The react1v1ty ;oss associated with pump operation was somewhat greater than that expected (and previously observed) from circulation of the delayed-neutron Precursors, The discrepancy, when attributed to cir- culating voids indicated a core void fraction of 0.1 to 0.2 vol %. Al- . though the confidence in this value was not Veryrhigh‘(becauSe the amount of reactivity involved was only ~ 0.02% 8k/k), it was in at'least general agreement with the void fraction required to describe the xenon behavior. The c1rculating void fraction remained essentlally unchanged for The - remainder of the reactor operatien with 2357 fuel. A few isolated pressure- release experiments provided no new information. However, between Decefiber 1967 and March 1968, an extensive series of tests was performed to determine the effects of small changes in reactor operating parameters on the circu- lating vold fraction and xenon poisoning.** The parameters varied were system temperature, overpressure, and fuel-pump level and significant changes in the core void fraction were detected. The minimum void frac- tion, for the range of parameters studied, occurred at the highest core outlet temperature (1225°F) and the lowest helium overpressure. (3 psig); there was no discernible level effect between 5.3 and 6.2 in, The abso- lute change in void fraction as conditions were changed to the lowest temperature (1180°F) and highest overpressure (9 psig) was 0.15 to 0.2 - vol 4. o | The results reported in the previous paragraph are based on reactivity measurements at zero power but considerable supporting evidence was ©b- tained from neutron-noise measurements?5;45 and related pressure-fluctuation 44MSR Program Semiann., Progr. Rept., Feb, 29, 1968, CRNL-L25hk, pp. L - 5. 4SMSR Program Semiann, Progr. Rept., CRNL-434k, Aug. 31, 1968, pp. 18 - 19, 46D, N, Fry, R. C. Kryter, and J. C. Robinson, Measurement of Helium Void Fraction in the MSRE Fuel Salt Using Neutron-Noise Analysis, USAEC Report ' GRNLPTM'2315, Oak Ridge National Laboratory, Aug. 27, 1968. : KN - [ L y » 53 tests.*? These measurements showed & change by about a factor of-7 between - the extreme conditions*and that the minimum core void fraction was around ~0.02 to 0.0k vol 4. Thus, there 1s good agreement on the change in void fraction with operating parameters and some apparent disagreement on the minimum value. However, thenuoid fraction required to satisfy the xenon ‘behavior (0.1 to 0.15 vol $) is an average over the entire loop while the above measured minimum value;applies only to the core, Because of its solubility in fuel salt, helium cover gas can lead to a void fraction that varies widely with position in the fuel loop. So these two values are not necessarily incompatible.r The single reactivity-based core void measure- ment (0.1 to 0.2 vol % in Run 8) was made at ~ 1180°F whidg again, reflects _agreement with the n01se measurements. A dramatic change - occurred in the system behavior shortly after the start of reactor operation with 233U fuel when the normal circulatlng void fraction increased rather abruptly by more than a factor of ten, into the range of 0.5 to 0,7 vol % (Details of this transition and other related - observations are described in ~subsequent chapters of this report we will deal only with measured v01d fractions here.) Following the. increase in void fraction, other phenomena were observed notably power blips48 and variations in xenon poisoning,‘f*9 which prompted additional 1nvestigations into the behavior of v01ds in the fuel loop., Of primary- 1nterest in these experimental studies were the effects of fuel- pump speed and cover-gas solubility on the circulating void fraction. Since all circulating gas must originally enter the loop from the 7‘fuel~1ump bowl it is clear that conditions in the Iump'bowl mst have a fg_significant effect on the circulating void fraction. We have already de- A'ftailed the effects of - fuelf;ump speed and cover gas on conditions in the ) __pump bowl (Fig._6) Figure lh shows the corresponding effects on the cir- 7”:i culating void fraction. The data_for flush salt are based on fuel-pump . 473, C. Robinson and. D.N. Fry, Determinatlon of the Vold Fraction in the MSRE Using Small Induced Pressure Perturbations, USAEC Report ORNL- TM-2318, Oak Ridge National Laboratory, Feb, 6, 1969 | “48MsR Program Semiann, Progr. Rept., ORNL-4396, Feb. 28, 1969, pp. 16-21. 4SMSR Program Semiann, Progr. Rept., ORNL-Lhhg, Aug. 31, 1969, p. 10. 54 ORNL-DWG 69— 10544 LOOP VOID FRACTION (%) ] i . SALT COVER GAS | 0.7— o0 FLUSH He 1: ® FLUSH Ar 8 FUEL He | A FUEL A 0.6 0.5 04 0.3 0.2 0.1 0 A L: 700 - 800 900 1000 1100 1200 FUEL PUMP SPEED (rpm) 1300 Fig, 14, Effect of Fuel-Rmp Speed on Void Fraction in Fuel Loop K st pa e RO b L AL R ». o » 22, :]level,changes and;'hence, represent average,void fractions‘for the entire ::fuel_loop. _Asoshown by the,error’flags, the uncertainties in these wvalues ~ are relatively large. However it is important to note that these data | flrepresent the only times that circulating voids were observed in the flush isalt The fact that circulating voids could be pmoduced in flush salt with : helium cover gas by running the pump at speeds only slightly higher than rnormal (12&5 vs 1190 rpm) indicates that the normsl condition was very ' close to the bubble threshold Although only two data points were obtained with this combination, the respective absence and Dresence of circulating - voids were confirmed by pressure—release tests, The higher void fractions with argon cover gas in flush salt result from the fact that argon 1s less . soluble than helium by a factor of ~ 10. Thus, for low rates of gas in- gestion at the pump, all the helium dissolves while the argon remains as Vbubbles The data for fuel salt are based on reactivity effects whlch give a | more precise measure of themvoid fraction. Consequently, these data repre- Asent average core'(rather-thaniloop)'void fractions;”'The fuel-salt points “show the lower speed threshold that would be expected from the slightly higher void fractions observed in the pump bowl (cf Fig. 6) ‘They also 'rshow the. same gas-solubility effect as the flush salt, Experiments vere 2lso performed to study the sensit1v1ty of the hlgh void fraction Obtained with helium at normal pump speed to changes in sys- _ tem temperature, overpressure and pump level. As before, the most signifi- - cant parameter was temperature but the effects were much less dramatic than - at low void fractions, the maximum change observed was - 1ess than a factor lof two. . Although there appear to have been minor. variations, under normal con- f‘-;ditions the clrculating void fraction remained relatively stable at about 0.5 vol % throughout the remainder of the operation with . 233U fuel - Sig- . _nificant, short- term changes were observed during special experiments that ",fchanged the oxidation-reduction condition of the salt, These temporary | effects are described later in this report. e 56 7. REMARKABIE EEHAVICR OF GAS AFTER THE 1968 SHUTDOWN Significant“changes in the behavior of the salt and gas in the MSRE fuel system occurred shortly after operation was resumed in the fall of 1968, following the processing of the salt and the loading of the first batches of ®% enriching salt. The first, most conspicuous change fias a rather-sudden, large increase in the afiounf of gas_bubbles circulating . with the fuel salt which occurred se#eral hours after the beginning of fuel circulation. Thereafter the void fraction remained genefally-high . but displayed at times some remarksble behavior. This behavior contri- buted to the understanding of the MSRE operation (although it may'seemrto héve raised more questions than it answered). Ferhaps more importantly it focussed the attenfiion of reactor chemists and designers upon the inter- relation of the chemical state cf the salt, its physical properties, and behavior ct gas=-salt-solid interfaces, Our observations and correlations during operation with 2337 fuel are described in this chapter. Salt Condition at the Start of Run 15 As will be described later in detail, during Run 15 the amount of gas cifculating in the fuel loop was observed to change;during-énd after exposure of beryllium or zirconium metal to the salt. No such response had been detectable when reducing agents had been added during earlier operation, Thus consideration of the changes in the“chemistry of the fuel salt between previous operation and this run may afford a clue to the causes of the phénomena. During August and September 1968, between Runs 1k and 15, both the ~ flush salt and fuel salt were processed.5® This on-site processing con- sisted of fluorination to remove uranium, followed by reduction of cor- rosion product fluorides to the metals, and, finally, filtration of the processed salts as they were being returned to the reactor tanks. The changes in salt composition were small, by far the largest beingfthe SOMSR Program Semiann, Progr. Rept., Aug. 31, 1968, ORNL-L3kk, pp. L - 11, i ) » o7 V‘renoval of 0.9 mole %TUF4_from,the-fuel. ‘Some fission products (such as ~ tellurium and iodine) were'removed by the fluorination, but these were present only to 8 few ppm'tO'begin with, Immediately after the filtra- ~tion, salt samples showed container metal concentrations not very different from what they had been during earller operation of the reactor, Although there was no gross change in'catlon composition, it-appears that the state of reduction of the salt at the beginning of Run 15 was different from what it had been earlier, _’l | In discussing theprocesSing and later additions,of beryllium, it will'be-necessary'to-refer to the relative reducing power of the various o elements. Table 2 lists the standard free energies of .formations of fluorides at 13LO°F in molten LiBeF, (Ref. 51). - The point to be made - from this table is that exposure of an active metal to a mixture contain- ing these specles will result inlthe formation of its fluoride end the re- duction of some species below it on the list. The lowest species present in the mix tends to be,reduced'first, but in the vicinity of a strong re- | ducing egent fractiOnS'of'seferel species can be'reduced. Another point is that reactions may be slow in reaching equilibrium, Prior to the MSRE'salttprocessing, experiments had shown that in N molten fluoride mixtures, a hydrogen sparge will reduce NiFy easily, FeFo ~with difficulty, and CrFa to a negligible extent in a practlcal time period.®2 Extension of these experiments showed that a practical process for reduction of . FeF5 and. CrFa in simulated MSRE carrier salt was addition of pressed slugs of 21rconium metal turnings, followed by hydrogen sparging. l1After the addition of an amount of zirconium equal to 2. 8 times the num- Vrrcber of equivalents of Fng and Cng originally present, filtered samples : 'f5showed % of the original iron and 6% of the original chromium concentra- _'tlons. Although one interpretation 1is that this was dissolved iron and *l~chr0mium fluorides that had. not been reduced, it was recognized that at _cglleast part was probably reduced metals that—had gotten through the sample "filter. L sy, R Grimes "Moltenxéelt'Reactor Chemistry," Nucl; Apfil, Tech., _8, 137 (970). | s S2MSR Erogram Semiann, Erogr. Rept., Feb. 29, 1968 ORNL h25h pp. 155-157. - Tgble 2 Standard Free Energies of Formation for Species in Molten LigBeF, at 1000°K #* Material® S _-AGf (kecal/mole) LiF 125 BeF | 107 UF o 99 UFy | | 9T ZrF, o 97 CrFo . | 75 FeFo 66 NiFo i 55 “The standard state for LiF and BeF> - is the LigBeF4 liquid. That for the other species is that hypothetical solution with the solute at unit mole fraction and with the activity coefficient it would have at infinite dilution. In the MSRE processing campaign, the flush salt was processed before the fuel salt., Corrosion product concentrations reportedS> for samples of the flush salt taken during and after the processing are listed in - Table 3. The fuel salt processing followed fi'similar pattern, but the ~ fluorination took longer and wore corrosion yroducts appeared in the salt, Sample results for the fuel are also shown in Table 3. . o The objective of the hydrogen sparging before the zir¢onium additions - was to reduce all the N12+, and the amounts of zirconium turnings dropped into the processing tank were in excess of the total number of eqfiivalent 'S3R. B. Lindauer, Processing of the MSRE Flush and Fuel Salfs,'USAEC Report ORNL-TM-2578, Oak Ridge National Laboratory, August 1969. 8 . " . 59 Table 3 .._Structgral'Metal“Qongentrafiions in - MSRE Flush and Fuel Salt Concentration ( ppm) +10 Fe + 40 Ni + 15 o Flush Salt? - In reactor system during Run: lh (averages) ,Before fluorination - After fluorination (6. 6 hr) After 10.8 hr of Hg sperging _After 604 g of Zr and 9 hr of Hz sparging After 1074 g of Zr and 25 hr- of H2 sparglng In drain tank after filtration - Fuel Salt” In reactor system during Run pL (averages) Before fluOrination , After fluorination (h6 8 hr)A '~After 17 1 hr of Hg sparglng i After 33.5 hr of Hg sparging o After 51.1 hr of Hp sparging‘;“..“.. 'Vd After 5000 g of Zr and eh hr of H2 sparging ;e_After 5100 g of Zr and 32 hr of HE sparging | | In drain tank after filtration. 6 150 52 104 .- - 133 210 516 No sample taken 100° 1t soP No sample taken 76° w1 26 85 130 60 170 131 36 420 400 840 uoo® L3P ¢ 420 - hoo® 5200 w60 - 380° 180° 100° 110° * (both solu- ble), with only a small amount of CrZt reduction also’_mcfirfing.s4 But on this first beryllium addition in Run 15 (described later in this chapter ‘and in Appendix A) there was a heavy coating in which the Fe:Cr ratio was 113:1, A further indication of the salt condltion came from the corr051on rate. During the first three months of fuel salt circulation its chromium concentration rose from 34 ppm to about TO ppm, indicatinglcorrosion of the Hastelloy-N fuel loop surfaces by some oxidizing agent; One hypothesis that might Dbe advanced to explain this corrosion would be that the fuel ' salt came back from the processing containing some 3 g-atoms (168 g) of iron as ]§‘e2+ A difficulty with this is that the effectiveness with'Which the chromium was reduced in the final stages of the salt proce581ng seems to preclude the likelihood that 51gn1ficant quantltles of Fe2t were de- _11vered to the fuel circuit, It has been suggested that the iron came back “as Fe° all right, but was oxidized to Fe®* by some contaminant_(prdbably oxygen) in the fuel circuit.5® The FeZt then reacted with the Hastelloy to leach the chromium, | 5‘*.MSR Program Semiann. Progr Rept., Aug. 31, 1967, ORNL-L4191, pp. 110 - 115. SSMSR Program Semiann. Progr. Rept., Feb. 28, 1970, CRNL-hsu8 Section 10.l. » 6L Iniany case 1t is elearfthat-very early in Run”15 an unprecedented\ condition ex1sted in which exposure of beryllium served mostly to produce " reduced metal. As detailed in Appendix A, examination of the addition 'capsules showed signiflcant amounts of finely divided metal that seemed to collect on thersurfacerof the salt. Magnets lowered into the pump bowl revealed that some of these materials persisted in the pump bowl after the 'reduction. The amount of 1ron powder that could have been 1nvolved is substantial, possibly on the order of lOOg. (This figure is based on the ~ amount of corrosion and also on the amount of. beryllium that was added be- - 'fore the Fe:Cr ratio in the deposits on the cages reached a low level.) Another significant observation in Run 15 was ‘that each of the beryl- lium addition assemblies removed from the pump bowl showed evidence that fuel salt had wetted the nickel cage ‘and had adhered to it.5® (None of - 'the beryllium additions during earlier operation had showed such wetting.) - - Sample capsules taken betweenrberyllium additions showed the normal non- wetting behavior, so the conclusion was that although major changes in the salt-metal interfacial tensions were induced by the exposure of the beryl- 9lium, they were transient in nature. 'In additlon to the foregoing differences the fuel salt density was "reduced roughly 5 percent by the substitution of 38 kg of uranium (85% x 2343) for the 221 kg of uranium formerly in the fuel. In summary: when- operation was resumed at the ‘beginning of Run 15, j) the fuel salt was somewhat less dense and, for the first time, was in a . 'state such that exposure of strong reducing agents could produce sub- - ';fstantial quantities of insoluble solids and also affect (at least tempo- "_-rarily) the gas- liquid-solid interfacial behavior.s' Behavicr.naflng}Figsh Salt Circulation ~ On August 1k, 1968"‘éf£eriéemp1és of flush salt from the drain tank ireshowed that the chromium Was accepmably low, the salt was transferred in- " to the fuel loop and circulated for a Lo-hour trial. There was no per- - '.ceptible‘dlfference in behavior,of the ‘salt between_this and earlier periods S56MSR Program Semiann, Progr. Rept., Feb. 28;,1969, ORNL-4396, p. 135. 62 of operation. The two pump-bowl bubblers indicated practicaily no gas between the bubblers or in the loop. The rate of transfer to the over- flow tank was steady at 0.k 1b/hr, very close to that which had existed in the fuel salt operation in the preceding run. (The rate was substanti- ally less than in the last lengthy period of flush salt circulation;,which was before Run 13.) | Behavior During First Period of Circulation of Fuel Salt After the fuel salt was processed, LiF-UF, (73-27 mole %) enriching salt containing 21 kg of uranium was loaded in through FD-2 Then, on Se ptember lh the fuel loop was filled and circulation started. The pur-~ ,pose of the circulation was twofold: to obtain a sample of the thoroughly _mixed salt and to permit the exposure of a beryllinm rod to reduce part | of the U*t to U, (The production process for the enriching salt had left practically all the uranium as U** and it was desirable to reduce some so as to attain a reducing envirorment that fiould prevent corrosion.) When fuel circulation was started, chart records showed no abnor- ‘mality. As shown in Fig. 15, when the pump'first started, the bowl level dropped 0.8 inch, then very slowly increased, As explained in Chepter 2, this drop was due to filling the spray ring and volute support cylinder and compressing the normal gas pockets in the loop. The slow increase could be explained entirely by a slight rise in loop temperature that was ‘observed during this 2-hour period, Proof that the rise was not due to gas building up in the loop came when the fuel pump was stopped at 08L46: the amount that the bowl level rose was the same by which it had dropped on the initial start. The same behavior was seen eleven hours later when the pump was again stopped and started, By the time the berYllium rod in its nickel cage was lovered into the p&mp bowl at 2037 on September 1k, the fuel had been circulating for 1k hours, with no anomalous behavior and with a very low rate of transfer to the overflow tank. | For some two hours after the beryllium entered the pump'bowl every- thing apparently continued normally. Then the indicated salt level began to rise, At the same time the trace on the recorder chart became much iy » - ORNL~DWG 70~ 5200 63 I..fl'.l *I-.ll..’ll. |lfl|l A m 1 1 1 illll.%.l.lld 1 o e e ; - e ,;MM EEEEE T ..w: St EE No. 83002 .”#HWQ.TllTI o GRAPHIC no.‘mno..m CoRP, 50\ !.Ifi..t.alfi. -l.- - — — Sttt w ._.. .,‘... —3ot——1——T — T8t i = g o T oy T o Ly e o e - @l T ............ - ——— . —— el - et ot e — ] - - —_ - L A - - 1= Y T T T T es [~ - An_clv 23t s Coprmpmdaspespop e g SE b 2 - W_M - 207" al ~ PRESSURE, 0-50 psig :ltl M { 2 i Qi T ot | il L i ;: ; i ..' ] Rl T8 Nt | gl E.H_Fa b 0730 896}' b} "1d3S *INIL . OorR LEVEL,0-100 % During First Two Pump Starts with Fuel Salt in Run 15 . Fig. 15. Recorder Chart of Fuel-Rmp Level Indicator (IR-59) & more noisy, as shown in Fig. 16. After graduelly rising sbout 0.4 inch, the indicated level curved over and began to decrease at the relatively high rate of 0.08 in./hr, implying an abnormally high rate bf-salt trans- fer to the overflow tank. (Thé'computer was unfortunately not fully opera- tive during this period of circuldtion, so a detailed record of the over- flow tank level does not exist ) | | - When the remains of the beryllium rod were removed after & 12-hour exposure,* there was no immediate effect on the apparent smount of gas in the loop or in the overflow rate. Both continued high for the remaining 10 hr that the fuel was circulated. .Quite clear evidence of the high gas fraction shortly before the fuel was drained is shown in Flg. 1T When the fuel ;ump was stopped, the indicated level rose 1.3 in., compared to the 0.8 in, change the first time it was stopped. The difference was the effect of expansion of more gas in the 100D as the pressure fell with the stppping of the pump, Immediately afterwards; the level bégfin to subside as some of the gas found its way}ofit-of the loop.. When the pump was re- started after about 13 min., the level dropped about 1.6 in., indicating a remarkably large amount of gas Stlll in the loop where it could be com- pressed. After the start, more gas apparently was ingested, causing the level to rise gradually. (ContraSt'Figs.rlT and 15,) Additional evidence ~~ of a high gas fraction was the response to a pressurerdecrease shown in o Fig. 17. Before the salt was drainéd the drain tank vents vere opened,' dropping the fuel loob pressure (the lower trace on Fig, 17) quickly by 1 psi. Expansion of the gas_in'the loop caused the sharp le#ei rise of about 0.15 in. When the drain valve fias thawed; the pump was allowed to run until the dropping level cut it off through a control interlock, which explains the sbrupt drop at the end. | | | Although it was not noticed at the time, the behavior of thermocouple TE-R52 changed sharply T hr after the beryllium went in, (See p. 8 for a description of this thermocouple location.) Up until then the recorder chart shows it was quite steady. Then it began to fluctuate in the manner ¥ | - : o ‘Later examination showed that the rod weight had diminished by’ 10.1 g during its exposure. h 6 m - z.mv. 1 1 1 3 3 ) , 1 \ \ \ { \ 1 X \ — Y X { \ _ \ O 1 i 1 i i 1 { 1 ! \ { \ \ | | 1 i . | 1 1 _I“iuu 1 X 1 3 1 1 1 1 1. 1 1 B \ { 139 1 { { 1 - , — 1 1 | 1= O 1 AT I 1 1 { 1 - - 1 1 1 1 % ~ 3 Y I 1 X 1 X \ I 1 { 1 X ] 1 1 i 1 - 1 3 { ] ] W - — 1 I~ 1 < k 1 1 1 “\v— 1 1 1 1 . 1 V 1 “\rn 1 1 1 1 1 1 i ' 1 { R\ 1 { 1 { o 8ot 1 m I M) 1 — _ d o Y e ' M i 1 1 1 1 1 | 1 1 ~J “ __ _ ,r .II+ > o IN-— __ ] 35 —1 1 1 . m , .O.. e R T N o P - — -0 , ) N 30 30 B m —— 25 25 - u6< ...... i S T e A - A - - w W 20 ; = , - W e - SO W R * " Ty E - 13 , - B ——15 ] e T = I . I-.- I 1 ] —— ¢ ——I ] 1 T - Qe i m—— ——— o i I—1 1 | %..1 T — i i i 7 ——1 i bt W 3 fr e -!.*...,,.lqu.... e e e e T e e e e e e e e = — : _ o 1107 — , R TRTIN gt 1 ] =1 —— & — i j—— — Tl e e g e I 1 I ] ¥ ] ¥ I _.N un i i —— ia_-m..[.L. .~ . M -m d g —f ]3] f—F——F— , . , , , ,_ _ ] ST o e e e e AT 1 I—Qd—7T 1—1 1 1 1 — I f f f—o——f— o Q o o Q o o O .Mw QO M o 0 - M . N o o O © I mes b , 8961 °G) "1dIS INIL , . 5 w N1 o ' 1 ] Fuel-B;lmpIevel Trace During In’it'ial‘_ Buildup of | 'Circulafiilg Voids in Run 15 Fig. 16, 66 . "o . . # 4 .I.flGOITII*I a -n..i...n nozio_.u ooav M I—.L.w@lfl| +!. uc.“fl!b. » e : _ “mmi— Sy 1 1 = I-flnl- A uu 4 e e e e 1 3 1 __ 4_._ ii*!il.l#.: mH e ememe e e - ;,nr*t-a“....w.a i 1 i © m:..,.tT o e e u_T. - e T -l - - — - - —fm - - - e e e —-——— - i " m—— ORNL-DWG 70- 5202 - - . - - e . . - . o . 35 . S35 —— el et B Jommal s b e 29 > T T -k a4 4. — o B rps e Lt wor v— —— E e —— el o s v i i - —— e s = — e o e = - — e ™ - —— - - L — e f——— e - . E A W » % T . P~ ] - U e md 3 g . s i e e om e } e e s 4 e o ] eeimie ol M T e e L v . sy, A fm, ~ PRESSURE ,0-50 psig 8961'GH "100* AWIL 84 bt e R i A Wl 1 5 e b | OR - 'LEVEL,0-100% Fuel~Pump Level Response to & FPump Stop and Restart with Circulating Voids Present Fig. 17. L) » » 1gof reactivity effects, 67 typically observed later during pover. blips.57 This fluctuation persisted through the remainder of the circulation period. ~In summary, noteworthy points about the first period of circulation of the fuel after processing are as follows, ' o ‘ 1., There was: practically no gas in the loop'forrthe first 16 hours; 2. ‘Then, after a beryllium rod had been in the" pump bowl for two "hours, gas began to build up in the loop.‘ 3. In less than an hour the loop gas had levelled at about - 0. 5 vol %, where it stayed for the remaining 18 hours of circulation. 4L, The ;ump-bowl level ‘trace became rough at the same time the " loop gas increased, '5.fv TE-R52 began to show fluctuations in the salt level at the bot- . tom of the core -sample. access plug about 5 hours after the loop gas started to increase, - 6. ' The transfer rate, Which had been low before, rose to roughly 4 1b/hr while the beryllium was in and remained at about that rate after ’the beryllium was removed Behavior During Subseguent Beryllium Exposures in Run 15 Beryllium was ekposed:tofthe'fuel salt in the IumD‘bOwl on three - Subsequent'occasions in Run 15. - Some effect on ‘the gas in the fuel loop. "was evident 1n each case but there were some 51gnificant differences in ‘the behavior. T '1l£57iSecond Berxllium Expgsure The second beryllium exposure in Run 15 was on October 13, 1968 By r_f5this “time enough 233U had been added to make the reactor eritical -and while : ;the beryllium.was in the pump bowl the reactor vas. on- sServo control at ‘”low power. Thus the regulating rod p051tion was available as an indicator . 5TMSR Program Semlenn. Progr, Rept., Feb. 28, 1969, CRNL~4396, pp. 16 - 21. - 68 Figure 18 is a reproduction of the fuel-pump level record spanning the 10.5-hr exposure of the beryllium (during which time the rod lost 8.3-g Be). At the outset the void fraction wes high, as it had been since the first beryllium addition. 'In less than a half hour after the beryl- lium went in, the level beéan to decrease and the reactivity to increase,- ipdicating a decrease'in the amount of gas in the loop an& in the core, This trend was soon reversed and from then until the beryllium was rémoved, the void fraction went through the gyrations indicated by the level chart. The control-rod movements associated with each level change were consistent with changes in the core fuel density expected from the suggested void- fraction variations. Except for the return of salt from the 0verflow tank, no operator actions were performed that should have influenced-the behavior in the loop. :Thus, the observed level changes were spontaneous and sug- gestive of major changes in the circulating void fraction., The transfer rates observed just before the Be expoSure and in the intervals during the exposure when the pump level and circulating void fraction were low were between 2,5 and 4 1b/hr. Hofiever as shown in Fig. 19, at higher levels fi and just after the exposure, rates as high as 30 lb/hr were observed, The very high rates weré nearly independent of indicated salt level betweén 6.1 and 6.9 inches (on IE-593). Unusually high transfer rates persisted for some 100 to 120 hr of salt circulaetion time and then returned to the 3 - 5 1b/hr range. _ | A , Comparison of the transfer rates with the densities indicated by the differences between the two bubblers revealed what seemed to be paradoxical behavior but which may actually be a clue as to what was happening in the pump bowl during this beryllium exposufe. The comparison of the two bub- ‘bler readings while the beryllium was in the bowl showed less voids in the bubbler zone_than usual at that indicated level. This is consistent, of course, with the clear evidence that less gas was being drawn into the loop. It clashes, however, with the inference from the unusually high transfer rate that more than the_usual amount of gas was causing froth to spill into the overflow line at a high rate, These indications could be reconciled by postulating that, while the beryllium was in, there was the unusual condition of a thick blanket of persistent foam that produced the “ PUMP BOWL LEVEL (% - »k Fig; 18. OCTOBER 13,1968 . - Lol .._‘.._mfimfi Fuel—fhmp Level Behavior During Second Beryllium.Exposure in | 233[1 Fuel Salt 69 50 D o TRANSFER RATE (1b/hr) O Q 20 10 ORNL-DWG 70~ 5203 60 x FUEL PUMP OFF r COMPUTER NOT IN SERVICE ~s— BERYLLIUM IN oL x » } : . | Efii " ~+—BERYLLIUM OUT | c ~ ] v . | . ) , _ | ", 14 15 16 A7 18 DATE : OCTOBER, 1968 i Qi 10 k) 12 Fig. 19. Transfer Rates During and After Second Beryllium Exposure 19 ‘_.’_ & L4 and control rod began to}reeover'and-the-initial conditions were restored in only 15 min, The beryllium was immersed sgain and decreases in pump level and rod position began immediately. The exposure was interrupted s ~only minor effects were observed The pump level then dropped rapidly to _'Vproduce a total change of 0 85 in. The remaining 8 hr of the exposure were '_jfunlnterrupted with the ‘salt level remaining relatively low. However, there '"r;'perlod the pump level had risen O 3 in. and the. difference ‘between the -"two bubblers had returned to practically the same value that was observed "with the beryllium out | - ‘ ' ' | transfer to the overflow ‘tank’ proceeded at only 1 - 2 lb/hr ‘Tn the interim periods with the beryllium. out, the rate was ~ 4 1b/hr. However, shortly Tl high transfer rate and at the same time ;revented the jets from driving ‘gas deep enough into the underlylng pool to reach the bubbler zone and the circulating stream. Third Beryllium Exposure The third beryllium eXPbsure was made on November 15, 1968, The total immersion time was 11 hr with 9. L g of Be dlssolving. However, as described '-deelow ‘the exposure was not continuous -This exposure was: started with a lower 1nd1cated salt level in the pump bowl to reduce the risk of salt transport into the offgas line. (Some evidence of a restriction:in'line 522 had already been observed,) In con- ~trast to the previous eXpOsure,'this capsule spent some 2~l/2 hr in the | pumplbefore any effectsfvereldetectable; then the'level'began‘to decrease -‘anddthe regulating rod began to insert as bubbles disappeared from circu- ‘lation, The beryllium'vaSWithdrawn with the level down‘by 0.64 in, and Cstill dropping becauseithe level had gone below thejspecified operating range for the pump. As soon as the beryllium was withdrawn, the salt level again because of the low salt level and, again, the effects were qulckly reversed At this point some salt was transferred from the overflow tank to alleviate ‘the concern about the low level. When the berylllum was immersed again, a Lo-min wait ensued in which were variations (= 0.1 in ) around this low level. By the end of the 8-hr Whlle “the pump level was low with beryllium metal in the salt the 2 ~ after the end of the exposure, the transfer rate jumped to 72 lb/hr., At least part of this high rate was due to an unusually high (70%) fuel-fiump | - level. However, éven at more moderate levels, transfer rates were in the range of 5 - 10 1b/hr, After about 48 hr more, the transfer rates were back to 5 1b/hr or less at normal levels, Fourth Beryllium Exposure A fourth assembly containing beryllium was exposed to salt in the pump bowlion November 21, 1968, The previous three exposurés had resfilted in the deposition of considerable "crud" on the perforated nickel container for the berylliufi rod, Since it abpeared that this was.reduced corrosion products, the fourth assembly contained several high-temperature permanent magnets to collect any magnetic material that might form as the beryllium dissolved. The presence of the magnets substantialiy decreased the beryl- lium surface area in this assembly so that only about 1 g of beryllium dis- solved in the T-hr exposure, The circulating void fraction decreased some- what during the exposure but the effect was much less pronounced than in earlier exposures, There was no detectable effect on the salt transfer rate, either during or after the exposure., Only a small amount of material was attracted to the magnets. :Comégrisofis : In all four beryllium exposures described above, there wés some ef- fect on the circulating void fraction as indicated by level changes and/or reactivity effects. However, the effects were not entirely reproducible, even qualitatively. The first beryllium exposure apparently contributed to the establishment of a high yoid fraction., Once these voids began to .appear, some 2 hours after the start of the exposure,'thefe was no evi- dence of a decrease in the wvoid fractibn, either during or after the ex- Posure, Tn all other cases, there was at least a tendéncy for the void fraction to decrease while beryllium was in contact with the salt and to return to the higher value after the beryllium had been removed. During the second exposure, the void fraction was quite unstable, exhibiting both decreases and increases while the beryllium was present., The 't # " & o 73 instability appeared almost immediately upon insertion of the beryllium and stopped as. soon as it was removed, On the third exposure, there was & 2-1/2-hr delay before the void fraction started to decrease. However, 'no additional 1nstabilit1es were Observed while the beryllium was present, In addition, the response of the void fraction o removal and the reinser- tion of the beryllium rod was immediate. The fourth exposure was appar- ently similar to the third but the effects were much smaller. We also. observed that the rate of salt transfer to the overflow tank was abnormally high for as long as 100 hr after ‘some beryllium exposures, implying that the beryllium produced some kind of abnormal condition from V_Which the system recovered only very slowly, Power Blips in Run 17 -The observations described so far in this chapter were gll made in Run 15 with the reactor'subcritical or at very low power, Power operation with 2337 fuel Was started;withpRun 17 in January, 1969. One of the more ‘unusual aspects of the early power operation was the appearance of small, positive disturbances in thelnuclear power (blips), The blips will be “ treated at length in another report®® but some discussion is presented here because of the interrelation between blips, circulating voids and 'beryllium exposures, ~ When first observed, the blips were occurring with an irregular fre- guency of 10 to 20 per hour. A typical blip involved a reactivity increase ~of 0.01 to 0. 02% ak/k for l.to ho seconds, resulting in a temporary power . increase of 10 to 15% The size and frequency of the blips decreased with g;tlme and by the end of March 1969, they could. no longer be distinguished - from the continuous noise in- the neutron flux, Although the blips were ib:not recOgnized unt11 January, When the reactor power ascension had reached . ’75 MW a detailed search of data records revealed that disturbances in other . system parameters_associated;with the blips began occurring in September | 58P. N Haubenreich and J. R. Engel Reactivity Disturbances ("Blips") in the MSRE, USAEC Report RNL-TM, Oak Ridge National Laboratory (in preparation) " when the high circulating void fraction first developed. -Figure 20 shows the typicalL small disturbances in fuel-pump pressure and level and in the temperature indicated by TE-R52 that accompanied the blips. | Since the blips were associated with circulating voids (this was i» proved later by running the fuel pump at reduced speed to eliminate wvoids and blips) and the voids were affected by the presence of beryllium, the blip behavior waé carefully Observed'during a beryllium exposure made On January 24, 1969, The beryllium had the expected effect in reducing the circulating void fraction and, for 2-hr 20-min while the void fraction was low, there were no blips. As soon as the beryllium was removed, the void fraction incréased and the blips returned, Figure 21 shows two segments from the record of linear nuclear power (as indicated by a cqmpensated'ion. chamber) before and during the beryllium exposure. These observations were used to support later conclusions about bubbles and blips. . Other ReGOX“ExIeriments_ Although the power blips gradually disappeared during Run 17 and the void fraction became stable, a number of other reduction-oxidation experi- % ments were performed, Table 3 summarizes all the tests with 233 fuel, including'the five beryllium exposures already discussed, These experi- ments had a variety of objectiveé which ‘included the effects of milder re- ducing'agents (Cr and Zr), observation of changes in interfacial tension of the salt with beryllium present, and the effect of the salt redox state on fission-product behavior. Since this report is concerned primarily with spray, mist, bubbles, and foam, only those aspects of the experiments will be discussed, To provide a uniform basis for comparison, Table 3 also lists the amount of reduction or oxidation, in gram-equivalents; accomplished by each exposure. The chemical effects of these actions have‘been‘evaluated elsevwhere,>® L a | 59SR Program Semiannual Progr. Rept., ORNL-4396, Feb. 28, 1969, pp. 133 = 135. | : | - 75 i } i i i | i 1 i i " ORNL-DWG 69-5373 ! R I 5 ‘4’ -~ ';' P : - s : o R e T Dot g , Ca . ?'i_ ! '2 lr'\w s'puh i E oy i T -—| l- 10 sec 'x i - - i i ' e T e e e . ! r’P mzs:we.fi. ' N e ! R v 1 . Por ety SP)\ ‘Ph“:-‘.; ;|'.<1A'§ 8T tog i ! i ! R i ! g e la b e L b TE A : oL T | oL . 7 . l : 1 [ P s r o e dT ey '[I ;-l isi ;!;l-,;'.-,l Pl P oy o !"' o L e e e ot b B + R ' 1 ' : oo b Lo R Tt e D : ¥ L . T . l T ’ . - ! : ’ : Voot i e g JE '-!. . : P STt " - . A . y - - St . ! : P . Lo L . v'f:-fi- -AJ;,,,: wbeTped g PR e e s e D . - . e [ : - e b e T M o ) T 1 i 3 . S . :_:jg_— Do ' N : R . T a . ’i- A I i FEEEE T 7: . t. . i L : i . . o . : . ' i F’P Ls.vzl_ b ?o.':. AL BERE : SR g iy . b et ! e P ! ! ceemn R meas — . . . - —-——— _..-*.... bt bR v s B . T ¥ ! i ¥ i i i ; ot ' ih e mmem - e et el . . . i 1 ik s :.. + i Ly - s i 1N ! i Ce b b g ! ; i - | : ¥ o ~ .Fig. 20. Recorder Cha.rt Showing Nuclea.r Power, Fuel-Pump Pressure, R Indicated Fuel-Pump Level and Themocouple a,t Bottom of Access Plug During Blips ‘ " 76 S " ORNL-DWG 70-5204 40 - - ‘ ) t ‘ L 35‘ L = - ‘ i - 1 30 { | ‘l 3 1 25 ‘ T i E ! g 14 - o 20 : > ke ) < 1 - w o« 15 B {0 5 0 , ' — L | 0 0o 20 30 40 0 0 20 30 40 BEFORE DURING POWER (% OF {5 Mw} - : Fig., 2]_.." Linear Fower Record Before and ‘Dur»ing & Be;'yllium" Exposure ty " " ® TT Table 3 " MSRE Redox Experiments with 2370 Fuel ey | Incremental . Active . Reduction | I Date Agent Attained Nominal Purpose (erem eguivalents) 9/14 /68 Be ‘2,24 Reduce some U to U3 - 10/13-14/68 Be 1.85 | Reduce some U4+ to U-F 11/15/68 Be 2.08 Reduce some Ut to U 11/21/68 Be(a) C0.22 Recover magnetic reduction products - 1/24/69 Be | '1-90_ Reduce some U** to U3 - 1/29/69 Cr : O 18 Observe effects of mild reductant: L/15,17/69 Zr 0,89~ Observe effects of mild reductant L /2L4-25 /69 Zr ~1.05 Observe effects of-mild‘reductant 5/8/69 FeFo 0.6k Oxidize some.U3+ to U4+ 5/15/69 ~ Be 1.26 ° Reduce some U* to U3 5/20/69 Be(b) : O.ZQ- Measure interfacial tension of fuel 9/12-26/69 7(€) 0,08 ‘Six Pu additions to fuel salt 10/2/69 Be(b) 0;56'ff'. Measure interfacial tension of fuel R , o L salt with Be® rresent - 10/7-8/69. Be l 09 ,- " Reduce some U4 to U3 . 10/21/69. Nb | <0.01 Observe salt wetting characteristics | :l_li/29/69i. Be - “1.55 Reduce some U to u . 12/9/69 . Be :__; ;572520157_- Reduce some U** to U3 . .Be(b) " 0.67 Measure interfacial tension of fuel - salt with Be® present’ ( )Assembly contalned several magnets for- purpose indicated. ( )Assenbly contalned a graphite body for purpose 1ndicated Iresent as w1ndow covers on Ni capsules. oyt T8 With the exception of the two major zirconium exposures, there was .no.discernible effect on the void behavior in any of the last experiments. There may have been some decreasse in the void fraction in'the pump bowl but the effect did not carry over into the circulating loop. Within two hours after the start of the first zirconium exposure (4/15/69) the circulating voidrfraction began to decfease significantly and then recovered to its original value. The entire transient'réQuiréd, less than 1 hr for & decrease and reéovery of 0.4 vol % in void ffaction. Within 30 min after this transient, the exposure was intérrupted by an unscheduled drain of the fuel salt @ue to a local electric power outage).. The exposure was continued after the loop had been refilled but this time there was no discernible effect. The second zirconium exposure (L/2L4/69) ‘began to affect the circulating void fraction within 35 min; a decrease of . 0.4 vol ¢ occurred and remained as long as the zirconium was in the salt, This zirconium assembly was withdrawn once and reinserted during the ex- posure, The circulating void fraction responded promptly to both actions, increasing when the zirconium was removed and decreasing when it was rein- serted, This behavior was similar to that observed in the third beryllium exposure described earlier. L y ® & 79 8. HEYPOTHESES AND CONCIUSIONS . Although the primery purpose of this report is to record, under one cover, the observations and experience related to a particular aspect of 'theiMSRE operation, some conclusions can be drawn sbout the mechanisms that contributed to this'experience. Some of these conclusions are relatively . flrm while others must be regarded as hypotheses.' ‘ It is clear from the appearance of objects exposed in the gas space in the fuel pump that there was 8 fairly high concentration of mist inside the shield around the sampler oage. Tt must have been higher out in the pump‘bowl although it could hardly have been high enough to contribute signiflcantly_to the observed transfer to the overflow tank, A small ' '(perhaps surprisingly_small) amount of the mist was drawn into the offgas line where itvcontributedvto'the plugging near the pump bowl. The problems | engendered by the mist in the MSRE pump were not intolerable, but they did '1llustrate the need for effective mist protection in molten-salt systems. The pumpobowl bubblers indicated rather conclusively that, whenever the pump was running, for several inches below the surface there was a region that contained a high (up to 50 vol %) void fraction, This high " void fraction apparently raised the actual salt level in the pump bowl high enough to cause some spill-over into the overflow line, which probably _ accounted for most of the salt transfer ‘to the overflow tank. That only - a small fraction of the ;umpvbowl gas appeared in the c1rculating loop as ' bubbles is due to the fact that the ‘normal. bump speed was just below a 'sharp threshold for bubble 1ngestion. The gas ingestion and “the general f'behav1or of the circulating voids are all readily explained by ordlnary 7fhydrodynam1c princlples prov1ded that gas solubility effects are included, The explanation for the shift in the circulatlng v01d fraction from e-a very small value with 235U fuel to a higher value with 233U fuel is 'fifmost 1likely small differences in the physical properties of the two salts. . We know that the salt densities were different and small dlfferences in 7; viscOS1ty and surface tension would not be sur;rising. Since the thres- dihold for bubble ingestion was very steep, only small differences would be | -required.to produce & pronounced effect on the void fraction. However, 80 in the absence of Precise physical-property data, we can only s;eculate about the significance of such effects. Tt appears from the various correlations that the redox condition of the salt may have been & significant factor in the bubble behavior, Presumably through changes in surface tension and/or viscosity. Although ' recognltion of the possibility of such a correlation has led to ‘some studies, a thorough understanding of- this area has yet to be galned g et e i ?) " 81 APEENDIX A MATER TALS RECOVERED FROM FUEL FUMP BOWL ‘Many of the capsules used for special exposures in the pump bowl were "examined in detall in hot cells after their'recovery.__The principal find- ings of these examinations have been reported elsevhere.»® TInasmuch as -the wetting of the capsules by the liquid the presence of floating solids, " and indications on . the capsules of salt level and foam are clues to the " conditions in the pump bowl which are the subject of this report, some ._pertinent information which“iS'nOt fully described elsewhere is'pmesented in this section. During the years of operation with 235U fuel, practically all of the ' _fobjects_exposed in the fuel pump came out relatively clean except for very "1ight deposits of what appearedgto be droplets of salt. The perforated capsules,used,to expose beryllium were;unuSually radiocactive due to fission rroduct deposition and some contained deposits high in chromium but the outsides were relatively clean in appearance.B_ After the fuel processing . and 233 loading, the experience with ordinary sample capsules vas similar, in that no unusual deposits vere observed but the appearances of the cap- sules used t0 expose reducing agents were markedly different. Figures A-1 through A-5 are. representative of the appearance of cages used for beryllium exposures after the fuel processing and the 233y load- '-;ing. ‘The first two show overall views of the cages used for the second B and fourth exposures (Oct 13 and Nov. 21, 1968) Both show very heavy ‘.__deposits on the’ outside and Fig. A-2. shows the presence of a considerable "ijamount of white salt In general the deposits were predominantly salt ip\with a covering of dark scum., The tops of both of these capsules were d”frelatively clean, . suggesting that they were not totally immersed in the ."salt ~However, - the absence of 8 distinet "Water mark" indicates that the 'p;salt-gas 1nterface was not sharp and constant even inside the sampder shroud. IMSR Program sémiahn.~rfogr; Rept., Feb. 28, ‘1969,;0RNL 4396, pp. 133-135. AR Program ‘Semiann, Irogr. Rept., Aug. 31, 1969, CRNL-4ik9, pp. 109-112. JMSR Program Semiann, Progr, Rept., Aug. 31, 1967, CRNL-4191, pp. 110- 115 82 The next three figures afe closeup views of three beryllium cages, fFigure A-3 shows the inside of the cage used in the second beryllium ex- posure, The deposit is at least as heavy as on the outside. The cage in Fig. A-}4 was used for the third béryllium.exposu:e. This view clearly - shows the white salt deposit and the overlying scum. The relatively smooth dispersion of the salt phase suggests wettifig of the nickel surface — a condition that did not normally prevail, The top of the cage used for.thé ~ fourth beryllium exposure is shown in Fig, A-5. The defibsits on the cap suggest residues from cruddy bubbles that burst on the surface. " The deposits on the cages were leached off and analyzed chemically. Such analyses confirmed that the bulk of the depositsrwas fuel salt, but large amounts of structural metals were also found. For example, the de- posit on the cage used for the first beryllium exposure contained 8% Ni, 0.1% Cr, -and 124 Fe. The relative abundances tended to eliminate foreign materials (e.g; stainless steel) as sources of the metals. In addition,. the high concentrations of Fe and Ni suggested either that theisalt was in an off-equilibrium redox state or that significant ambunts of the free metals were present in the salt. | | - Some characteristic results of attempts to recover frée metallic'par4 ticles from the fuel salt are shown in Figs. A-6 through A-10. Figure A-6 is the result of the first attempt using a l/2-in.-diam x L-1in, long Alnico-V magnet in a copper capsule. Although some magnetic material was trapped in the 5-min ekposura, the amount was not impressive, probably less than 1 'gram. Figures A-T and A-B‘are two views of another attempt shortly after - a beryllium exposure. This time, somewhat more material was collected but | the difference was not enough to be significant. A cepsule containing | both magnets and some beryllium was also exposed., The result (Fig. A-9) was much like any other beryllium exposuré but little magnetic material was collected. Another stack of short magnets was exposed atfithe start af Run 16, just after the loop had been filled but prior to any salt cir- culation. The presence of some magnetic particles (Fig. A-10) suggested that at least some of this material was floating on the salt surface. As indicated in the body of this report, some relatlvely mild re- ~ ducing asgents were exposed to fuel salt in the pump bowl. Flgures_A-ll 1) » 83 - and A-12 show a_ghromiumjroé after about 6 hours' exposure torthe'sélt; | There was little attack on the Cr but both kinds of deposit, salt and crud, appeared., In this case; hofiever, the salt appeared to be non-wetting, The fractured surface 1n'Fig;ffi412‘(broken'as part of the-pOSt-eXPOSure exami- nation) shows a black deposit on the lateral surfaces, 8L & n Nickel Cage from Second Beryllium Exposure in % Fuel Salt Fig. A-1, , 86 B | _ R-45807 | Fig. A-3. Inside of Nickel Cage from Second Beryllium Exposure in 33 Fuel Salt e bk ot A U A 5 Fig. A-5. Top of Nickel Cage from Fourth Beryllium Exposure in 2337 Fuel Salt ‘N it i Wi .‘ ’ - Fig, A-6. . ‘ cT _ 7,7 Metallic Particles on Copper Capsule Used to Expose a Magnet in 2 ruel a1t . L Upper End of Magnet Capsule . A-T. Fig i Fig. A-9. Capsule Used to Expose Beryllium and Magnets Simultaneously Capsule Coritaining Several Short Magnets Fig. A-10. i 2337 Fuel Salt b 4T sttt ot & - "w 97 APEENDIX B MSRE 'sALT lfitNSI'IJIES As was explained in Chapter L, the fuel-pump bubblers could be used vhen the pump was not runnlng to measure absolute salt densities and re- sults were obtained that compared favorably with those produced by other methods., Table B-1 summerizes the densities for the various salts used | 'in'theimmfiE'as determined by several different methods."Sinee'only the bubbler-difference technique is described in detail in this report, a - brief discussion iS'given below_for'each of tne other methods, ~-éalt-Charging' - The amounts of flush salt and fuel carrier salt initially loaded into the MSRE were carefully weighed during the'charging'operation.r Since both -salts were loaded into FD—2 it was possible to use the two level probes in that tank as a volume measurement. The weight of - salt required to fi11 " the space between the probes (62 38 ft3 at 1200°F) was used to calculste’ the salt densities., Fill of Primary Ioop | - After the primary'lOOPVWas filled the preSSures in the fuel pump and “drain tank were balanced to hold a constant salt 1evel while the loop drain _'*valve (FV-103) was frozen., The pressure difference required to support a | ’column of, salt ~ 25 ft high then gave a measure of salt density. 'f,Addition of Mblar Volumes 7, Effective molar volumes have been developed for the various pure com- 1d'ponents of molten fluoride mixtures.' These volumes are weighted with the 'r:component mol fraction and added to obtain mixture densities at temperature. 1g, Cantor, Reactor Chem. Div. Ann. Er0gr Rept., December 31 1965, | USAEC Report CRNL-3913, P 7 - 29 98 _Table B-1 | | ) : | o \u) Densities of MGRE Salts o o Fuel ’ . 235 ' 233y Salt Mixture Carrier Flush Fuel ~ Fuel Method S Density at 1200°F (1b/ft7) Salt Charging 0.5 124° I Fill of Primary Loop | 121 - 123 140 135 ~ 136 - Fuel Pump Bubblers | 126 - 128 145 - 147 141 - 143 Molar Volumes® - 121 - 122 1h2 - 1b6° Method of Mixtures 121 o 1hs 135 Laboratory Measurement - 12% 146.5 - ®This method was also applied to the coolant salt which has the same composition; a density of 121.8 1b/ft> was obtained at 1239°F. bDensities obtained by this method are generally regarded as the most accurate, at least in the absence of direct measurements. “A more recent (2/25/69) calculation using better data gave 140 1b/ft>, | | , T L 3 99 . Method of Mixtures® This method iszbased ohfan empirical'correlation between measured densities of liquid fluorlde mixtures and calculated room-temperature densities. The room-temperature densities are obtained by addition of actual densities of the pure solid components, weighted by their mol fractions. Laboratorv Mbasurements These are results of direct measurements made in the 1aboratory under carefully controlled conditlons. '~Discuss1on o Each of the above approaches to- density evaluation has limitations which become apparent when the details of the method are: examined How- ever, it is not our purpose to evaluate or. crlticize the various methods Nevertheless, the table illustrates that the highly prec1se values needed d for accurate 1nventory control were not readlly available during the opera- | 'tion of the MSRE. B | | 23, T, Cohen andfT;5N;?Jones;7A>Summary of Density Measurements on - ,,{'.Molten Fluoride Mixtures and a Correlation for Predicting Densities of " Fluoride Mlxtures, USAEC Report 0RNLF1702 Oak Ridge National Laboratory, - July 19, 195k, - 3B, J. Sturm and R. E. Thoma, RCD AR, USAEC Report oam,-3913, - Oak Ridge National Laboratory, Dec. 31, 1965, pp. 50 - 51 »; k. s ko, - - » ® * L H e WhHO 1k, 1 = oW 18, n S 21"22 . 23, 2k, 25. 26. 27. 8., -~ 30. 31, 32, 33. _ 3h. 35. 36 LO - ., o, 43, L, 45, - 6. 50, w:oo-Jcny1¢ruin>+4' . - . >fim5qvaszu?qmmwzqmmwfimz J. L, R. F. C. Fo' C. E. C. J. - He F, S. E, e S, F. G. E A, B. D. » - . .. M. H. F. e . 2 . . ® . * . - - - ® - mm»qwgwwfiew .s. - Davis .Jo Ro : ce . Ronald Feit, AEC*Washington D. E, L. M. Ja. B, A. P, J. K. D. N, - C. H. " R. B. L. O. 101 ~ Internsl Distribution Anderson Apple - Baes Bamberger Barton Bauman Beall . Bender 7 Bettis Blankenship Blumberg Bohlmann . Boyd Braunstein Bredig' Briggs Brunton Cantor Chandler Chapman o Cope, AEC-OSR . Compere Cottrell Corbin Crowley Culler Cuneo Dale Ditto Doss Dworkin Eatherly ' Elias, AEC-Washington : Engel ' Ferguson _ Ferris - - o Fox, AEC-Washington Fraas S Franzreb - Fry - Gabbard Gallaher , Gilpatrick Degrazia, AEC*WEshington' Sl. 52, 53 -5k, 29 56. 5T+ 58. - 59. - 60.. 61, 62, 63, 6L, 65, . 66, 67. 68, 69. TO. Tl T2. T3. Th. T5. T6. 77 . T8. 79, - 80. 81.. 82. 83. 8L, 85. 86 -87. 88. 0. 91. 92. - 93. - ol 95. 96. QWABEnEmE R D -« * 89. Fmebbubmfinmmuwwzwq ORNL~-TM- 3027 G. Goidberg W. R, A, G. ‘E. D. R. H. H. N. _W. L. I. R. J. T. B9 ey o * . . * - . @ ch:n:k*fidifilz: Bo' Grimes — G. M. ‘Watson Grindell Gupton Guymon Norton Haberman, AEc-Washington Harley Haubenreich Hoffman Houtzeel Hudson Johnson Kaplan "Kasten Kedl Kelley -Kelly Kerlin . Kerr Kirslis Kohn Krakoviak Kress Kryter . Lamb Kbrmit Laughon, AEC- OSR * . .9 - .. » - » -« . » @ - > * * . &« » . - ' e * * @ 8 . zwzfifimwmbzmdmflwsz Low1n : Lindauer Tandin - Lyon ‘MacPherson Manning McCoy = McCurdy MeGlothlan - ' MCIntosh AEC-Washington Mclain McNeese McWherter ‘Meyer Miller ‘Moore Moulton 'Nicholson Perry 98. 99. 100. 101. 102, 103-105. 106. 107. 108. - 109. 110. 111. 112. 113. 102 Internal Distribution 129-130. Central Research Library (CRL) 131-132. Y-12 Document Reference Section (IRS) (continued) G. L. Ragan 11k, J. L. Redford 115. D. M. Richardson 116. M. Richardson 117. - D. R. Riley, AEC-Washington 118, J. C. Robinson 119. M. W. Rosenthal - 120. A, W. Savolainen 121, T. G. Schleiter, AEC-Washington 122. J. J. Schreiber, AEC-Washington = 123. Dunlap Scott 12k, J. H. Shaffer 125, M. Shaw, AEC-Washington 126, M. J. Skinner | 127. A. N, Smith 128, TURNUPOSYN GO L. G. ORNL-TM- 3027 Smith Spiewak A. R. E. Bc Co S. M. R. E. C. ,.J'. v. Sundberg Tallackson Thoma - Trauger Ulrich Walker Weinberg Welir _ Whatley White Whitman, AEC-Washington Wilson Gale Young 133-135., Laboratory Records Department (IRD) 136. Laboratory Records Department, Record Copy (IRD-RC) External Distribution 137. R C. Steffy, Jr., TVA-303 Power Building Chattanocogs, Tennessee, 3TLOL - 138-152, Division of Technical Information Extension (DIIE) : 153, ILaboratory and University Division, ORO | 1 iy