.II - ?2 aim-n Thsi's?chncisgy 0f ?sis?iy VQLUME1 Reactar Physics and EDITORS T. J. Thompson J. G. Prepared under the auspices of the Division of Technical information U. SzAtomic Energy Commission TH MAI. Massachusetts of Technology - Cambridge, Massachusetts - .. . .. . . . I Hid}. SIM Ere: . 638 chamber, and the ion current from the principal chamber fell more rapidly than that from the gamma compensation chamber. Thus, with increasing flux the not current would appear to drop to negative values since the gamma compensation current is subtracted. The result eculd. and did, appear as a negative period when actually the period was positive. Comments, Couclusions, HeCOmmendations (1) There should always be at least two 011? scale neutrOn signals preferably from different types of chambers with different types of electrical circuits?for example, a iission chamber detector utilizing a battery power supply and operating a galvanometer, and an ion chamber with its own voltage supply and amplifier?recording unit. The reactor instrumentation System should satisfy the Principle of Diversity discussed in the chapter on Sensing and Control Instrumentation, Secs. 1.4.1 and 1.4.2.- (2) Safety System circuitry should not be made a part of an experiment or of a control unit. Use of instruments for both safety and control reduces the number of independent safety circuits and thus violates the Principle of Redundancy discussed in the chapter on Sensing and Control Instrumentation, Sec. 1.4.2. (3) Records should be kept of all changes made to any part of the system including the circuitry. Before any changes are made, a review {by someone or some grcIup completely knowledgeable of the given reactor and the points at issue) should be carried out of the effect that the proposed change will have on the existing system. This review should also consider any previous alterations which may affect the situation. (4) Reactor instrumentation should be designed to trip, or at least sound an alarm, 0n fast negative periods as well as fast positive periods, (5) The use of automatic controls to carry out a rise to power may be somewhat questionable unless the system has been repeatedly checked and found to be completely reliable in taking appropriate action in the event an unexpected input signal is obtained. In particular, in approaching a new and higher power level slow and careful manual operation by competent operators is likely to be the more conservative approach. Especially this is true when the instrumentation is untested in the desired operating range. Manual operation using slow steps would probably have given the operator a chance to catch the problem in time. It is difficult to build judgment into an automatic startup method which is as sound and broad as that of an alert and knowledgeable human unless the situation is quite routine. {The situation here contrasts somewhat with that in Sec. 3.6 where extremely rapid operations were re? quired. Even there, the system, when autoinated should have been fully checked out and tested and then operated on a series of small extrapolations in performance.) (6) Electronic circuitry and chambers should be tested under the operating conditions in which they will be used. If this is not possible, it should be recognized that they are being tested by the reactor pool, T. J. THOMPSON operations themselves and extra precautions, care, and alertness should be observed. Epler [41] says ?failure to test under planned conditions must result ultimately in an unplanned test.? (7) The safety system should have adequate menttoring and alarms or scrams On such items as circuit continuity, proper high voltages, line voltages, etc. (8) In general, it is better to utilize a smaller signal from an ion chamber further away from the neutron source with a suitable amplifier, than it is to overdrive an ion chamber and risk operating in a region where the chamber reopense is far from linear with increasing neutron flux. 3.9 m_ SEE Fuel Element Damage Acoide? [43?49] The Sodium Reactor Experiment (SEE) was built at Santa Susana, Calif. to aid in the develop? ment of the sodium?cooled, graphite?moderated reactor concept for civilian power use. This 20 Mw(t) plant went critical in April of 1957 and first generated electrical power on July 12, 1957. As shown in Fig. 3?10 the reactor core region is divided into hexagonal cells of zirconium?clad graphite 11 in. (2?.94 cm) across the ?ats and 10 ft (3.05 m) high. The outer units serve as reflectors and the inner ones centain fuel in a central cylindrical tube 2.80 in (7,11 em) ID. The fuel elements in the core in question (Fig. 3?11) [43] were made up of 7 rod clusters, each rod 3. 6 ft (1.83 110) column of 6 in. (15.2 cm} long uranium slugs in a 0.010 in. (0.25 mm} thick stainless steel tube. The 0.010 in. {0.25 mm) annulus was bonded by Nah?. and there was a helium?filled Space above. The outer sin rods were wrapped with a stainless steel helical wire to prevent rods from touching each other or the process channel. The flow of the sodium coolant is upward thrOugh the core and out to the heat exchanger. ?The design and construction philoSOphy of the system emphasized the use of conventional. commercially obtainable compenents wherever possible. To this end.the coolant circulating; pumps are simple adaptations of hot oil pumps. The stuff- ing box (see Fig. 3?12) [43] was replaced by a cooled annulus around the shaft to freeze sodium and thus seal liquid sodium in the pump casing from the supporting bearings on the shaft and the drive motors. The primary and secondary systems each have a 6?111. and 2vin. pump and three of these have had erratic operating experience. . .The 6?in. 1500 gpmi? pump in the primary the main offender in the fueleclement damage incident that occurred in July 1959. Difficulties had been experienced from time to time with binding similar to that of the 2?in. pump, and. on two occasions, the auxilary coolant (Tetralinl in the freeze seal has leaked into the main sodium coolant stream. This was detected by identifying hydrocarbon vapor in the atmosphere above the The first of these occurrences. *1500 94.5 liter/sec . v.31 . J. THOMPSON ecautions, care, 32d. Epler med conditions led test.? have adequate on such items . voltages, line ilize a smaller away from the .plifier, than it . risk operating esponse is far flux. int (SEE) was in the develop? ?ts?moderated rer use. This :?Lpril of 195? on July 12, the reactor renal cells of .94 cm) across the outer unite ones contain 2.80 in (1211 are in question rod clusters, 6 in. (15.2 cm) $.25 mm} thick n. {0.25 mm) there was a lier six rods helical wire to other or the edit-m: coolant m: to the heat philosophy of conventional, wherever :ulatlng pumps ms. ?hc stuff? replaced by a zoo sodium and using from the and the drive systems each nurse of these e. . .The G?ln. system. . . vas scent damage Difficulties to time with pump, and, int main sodium 93! identifying re above the occurr enc es a. . manm-mo. ?ham. ACCIDENTS AND DESTRUCTIVE TESTS ?3 ROLLER ROTATABLE SHIELD i AUXILIARY SODIUM INLET LINE mlmy I SAFETY ROD f" I -. If? v: 'l Ian MODERATOR ELEM EN I I CONTROL ELEMENT saw Roof?, J, . - fl. lift? scum? pLENqu I TETRALIN COOLANT hum-mau- az. u~ -- r-w 1.. when .. 639 FROZE ME TM. 5 AL. 5 MAIN SODIUM INLET LINE CORE TANK 1. . THERMAL SHIELD a OUTER TANK . . . :n 1 r. CAVITY LINER moms:an SHIELD GRID PLATE PLACIE SUPPORTWG CYLINDERS FIG. 3?10 Cress sectlon of the SRE. ROD Q's-e HELIUM FILLED EXPANSON SPACE ROD JACKET ?r SLUGS (I2) 0.090? (TYP) 9 In .3 as SEVEN ROD ELEMENT 0.0m in as. TUBE 0.010 in BOND 035 In. SLUG ma. FIG, 3- ll SRE fuel element. in April 1958, resulted in an insignificant amount of leakage before the freeze seal was repaired. This leak was caused by a pinhole in the freeze?- seal casting. The second such leak occurred in May 1359 and resulted in somewhere between 2 and 10 gal of ?l?etralin being admitted to the pri~ mary sodium stream This leak was traced to failure of a thermocouple well. The Tetralin?cooled shaft seal was then replacedwith a NaKucooled seal arranged in such a way that two independent barriers would have to fall before mixing of and the primary sodium could occur. As will be mentioned later, however, sufficient Tetralin had already been admitted to the system to create the condition that damaged the fuel assemblies Some comments on plant performancepertlnent to safety or to fission product retention taken from reference {43] are mentioned here although most of them had no direct bearing on the accident itself. . ?a 1' Ll-u.? - . gm -.-.. .13: .. - 640 THERMOCOUPLE TOP OF SEAL . Prh?? .- - v.or? any} J. THOMPSON 0 200 400 600 800 i000 TEMPERATURE TETRALIN COOLANT .J 4 a He FREEZE POINT i LIJ 2 cf 9'3 5an BOTTOM CI FIG. 3- 12 SEE pump shaft freeze seal. ??~Unalloyed uranium metal is an unsatisfactory fuel material for a high?temperature reactor bed cause of a tendency to swell. Fuel rods irradiated to about 1000K an increase in diameter of 2 to a mils. The.. annulus. .. had been occupied by the swelling uranium and the can distended.? ?Modification and maintenance of the sodium system have been accomplished with case and safety. Piping can be cut and welded by first freezing the contained sodium. There have been no sodium fires during any operatioa involving cutting or welding piping containing frozen sodium.- ?l?he cold traps and hot traps have periormed well in removing sodium oxide; no diificulty was experienced in maintaining the oxide concentration below 10 ppm.? ?The major difficulty with the instrumentation has been a series of spurious scrams caused by ?uctuations in voltage from the power supply.?* Reference [43] presents six pages of graphic op? erating and scram history for the reactor. Since the circumstances involved inthe accident extended over the entire period of time from November 29, 1958 to July 26, 195911; seems ap~ propriate to outline briefly the pertinent facts chronologically as is done in more detail in refer? ence Quotations are from that reference. Run 8 After a shutdown of about two months for repairs and modifications (involving several trans? fers ?of the sodium from the reactor to the fill tank which was known to have considerable Sodium oxide). the reactor was taken to 3.6 Mw on Nov? ember 29, 1858. The fuel outlet temperah1res,which usually showed a spread of less than showed much higher values (am-800%? or 259? This was attributed to high Oxide centent. *Frequent spurious scrams and warnings tend to dull the operators? sensitivity and to lead op? erators to ignore anomalous behavior. At the . same time, preventing serams is also a dangerous practice. Somewhere between these two extremes - there is a reasonable level for the rate of scrams per year, but it must be judged on an individual basis, depending on the stage in the reactor op? erations, the type and use of the reactor. conse? quences of a serum, complexity of the reactor. etc. On December 12. two elements which had been excessively but were removed and washed. Both had black material on them. ?Jiggling? by moving the element up and down one inch or less (in its position in the reactor) was found to improve heat transfer, butwashing helped even more. Power was increased to 12 MW (December 18) and 14 MW (December 19) and the run continued to December 23. After shutdown. 15 elements were washed and more cold?trappinth was done. The run continued again from December 27 to January 29 at 20 Mw maximum power with jiggling. On January 7 a sample of the cover gas showed the presenco of napthalene (and therefore Tetralin) in the system, something not Suspected before?? although there had been a prior leak ?in June (It was not known if any Tetralin had entered the primary sodium system from the earlier leak caused by a crack in the bearing heusing casting on the main primary pump.) The run was terminated on Jan? uary 29 ?because the desired exposure of 600 de was attained.? {hip} The run started on February 14 and cen? tinued until February 26. The reactor was run at 20 MW with continuing difficulties with fuel exit temperature spread leading; to shutdown and more washing and cold~trapping. ?Reactor op- erations were February 20. Examination of the records of the shim?rod positions (which was made after run 14) indicated that an increases in reactivity of 1/2% had occurred. Such an in? ereasc is expected because of the xenon decay during a shutdown of this length. However, a preliminary calculation of this effect indicated an expected increase in reactivity of It is be? lieved that this discrepancy is due to approximations made in calculations of the xenou correction. Similar discrepancies are noted in later rune. . ?There were two reactor scrams caused by to purify Na or NaKis discussed in the chapter on Chemical Reactions. evidence w0uld seem to indicate three leaks one in April. May, or June 1958 (the reports differ in the time), one in December 1958 during run 8 (no mention is made of the source of this one or its repair), and one during run 13 in May?June 1959. THOMPSON :gich had been washed. Both 1g? by moving it or less (in .nd to improve more. Power .18) and 14 MW to December re washed and run continued 29 at 20 Mw . January 7 a the presence :in the system, tough there hurl (it was not :1 the primary sak caused by en the main mated on an? cosure of 600 14 and con- actor was run ties with fuel Shutdown and ?Reactor op? Examination Jsitions (which let an increase Such an in? ;e xenon decay . However, a :ct indicated an It is be? approximations on correction. .nter runs. . ims caused by or: is disCussecl indicate three or June 1953 .c in December L-l made of the and one during hm" .. ACCIDENTS AND DESTRUCTIVE TESTS i3 excessive temperature drop across a moderator can and several scrams caused by power line transients. The reactor has a long history of scrams due to the latter effect.? Run 9 ?was terminated after the desired exposure of 125 de was achieved.? After shutdown, the fuel element in reactor channel 56 was examined. The orifice plate had a thin black deposit. The fuel element was washed and replaced in the reactor. Runs 10, 11, 12 (March 64?, March Iii-April 6, May 14?24) These runs showed some continued improvements in the fuel exit temperature spread. ?An examination of the records of shim rod position (made after run 14) shows that at the start of run 11 a loss of reactivity of had occurred. The loss may have been due to the replacement of a thimble.? In Run 11 afurther series of scrams due to flow fluctuations occurred and a reactivity discrepancy of 1/ 38h attributed to xenon occurred again. During Run 11 the radiation level in the main sodium gallery seemed high although this was not observed until ten days after the run. At the end of this run a filter was installed in the primary system which collected considerable car? bon containing material. During Run 12 a planned outlet temperature of was reached for 1 hour at 0 Mw power and steam was produced at A check after shutdown showed no meaSurable change in fuel dimensions. Run 13 (May 27wJune 3) Except for a sodium flow rate scram, the run was considered normal at" a power of 20 Mw until 09:00, May 30. Then, several abnormalities were seen including, a slow three?day rise in inlet temperature from to an increase in log mean temperature difference across the intermediate heat exchanger indicating impaired heat transfer, a rise in temperature over a 20~min period from in one fuel rod, some increases in exit fuel temperatures, a jump of at 22:30, May 30, for moderator delta compared to earlier fluctuations, and one or two other temperature probe effects. In addition, ?although it was not noted at the time becauSe the reactor was on automatic control, an examination of the record of shim rod position (made after Run 14) showed that a shim rod motion corresponding to a reactivity increase of about 0.3% had occurred. This change was gradual and extended over a period of about 6 hours. Following this, the reactivity showed a steady increase of about 0.1% over the next three days of operation.? By June 2, it was obvious that the heat transfer characteristics had been impaired in the primary system and the cause was believed to ban Tetralin leak. The odor of 'l?etraiinwas detected in the pump casing of the main primary pump. The run was terminated on June 3. and after a 10 day interval to allow radioactive sodium to decay. the pump was removed. A leak was discovsred in the wall of the thermocouple well of the freeze gland soul (see Fig. where a dislodged piece of hard plating material were through the wall as the shaft rotated. I ?Seventeen luel elements were visually ex- amined by means of a television camera and found . . -. .- . . - . . . .. .. "Sula; . -.. ?133.322.: -- - ?nimr'im'l' a- -- 641 to be dirty, but in good condition.? An attempt was made to wash one element, but during the operation a pressure excursion occurredwhich severed the hanger (see Fig. 3-11) and lifted the shield plug out of the wash cell. It is believed that hydrocarbons from the breakdown of Tetralin could cause sodium to be trapped in the holddown tube on the hanger rod by blocking the sodium drain holes. This sodium then reacted with the wash water. ?As a result of this incident, no further washing was done.? It was decided to ?strip? the Tetralin and organics which would volatilize by passing nitrogen gas through the sodium system. The process had been used on October 12, 1958 to remove Tetralin from the eye? tem after the first leak. The stripping began June 17 and continued until July 5. The sodium System temperature was 350?]? initially and was raised to by the end to help the removal. In all 400,000 n3 (11,300 1113) of nitrogen were used and 3 pints (1.42 liter) of ?l?etralin and 1500 cm3 (91.5 in?) of napthalene were removed. The system was then purged for ten hours with 4700 ft3 (133 m3) of helium and argon. The primary pump was reinstalled with a h'aK?cooled freeze seal in place of the Tetralin? cooled seal, and the System was prepared for operation.* Run 14 (July 12?26, 1959) The run was begun with the anticipation that the situation would be similar to that experienced in Run 8. The reactor was made critical at 08:50, July 12. At 08:35 as the reactor was slowly increasing in power to 0.5 MW large fluctuations of were noted on the moderator delta?T recorder. Nor? mally, even at 20 Mw these were less than The fuel exit channel temperatures started to show a spread of about Operation con? tinued at less than 1 Mw until 11:42. when a. scram occurred due to loss of auxiliary sodium flow. Criticality was reestablished at 12:15 and op?~ crations continued at slowly increas ing power levels with fuel exit channel temperatures from 510 to (255 to Fluctuations of in the moderator delta??1? at 1.5 Mw were observed. At 15:30 reactor room air monitors showed a sharp increase in activity. The radiation level over the sodium level coil thimble in channel 7 rose to 500 mr/hr. Air filter and stack activities increased. The reactor cover gas pressure was lowered from 2 psig to less than 1 ps?ig in an *After the final accident during Run 14, the use of nitrogen at this point caused considerable concern about the possibility of nitriding of the stainless steel and zirconium and thus promoting fuel and moderator can failures. Tests seemed to indicate that nitriding will ocour in preference to carburizing at bicarbon?bearing sodium with a nitrogen cover gas. Apparently it will even ocCur after ahclium purgehus supposedly swept all of the nitrogen out. The nitrogen evidently is held by the carbon and calcium impurities. Measurements, coupled with the lmowu solubility of carbon in sodium [till] showed that the system had been saturated in carbon ever since Run 8. mum. swim-.7. . . a Justin's?? - Jen-d vs: 'f - .4 642 effort to reduce the level. By 17:00 the radiation level over core channel 7 reached 25 r/hr. Accordingly. at 17:30 power reduction began. at 20:57 the reactor was shutdown, and the sodium probe was removedfrom channel 7 and replaced by a shield plug. The reactor was brought to criticality at 04:40 July 13 with exit fuel temperature scram set? points lowered to At 13:30 it was observed that the moderator delta-T followed a rise in the sodium outlet temperature and that the moderator temperature did not respond prop? erly to an increase in sodium flow. It was con? cluded that little sodium was leaking across the grid plate for moderator coolant. At 17:28 a planned increase in reactor power from 1.6 Mw began in order to deliver heat to the electrical substation. ?At the start the power level per? sisted in rising somewhat faster than expected even though control rods?were being slowly in? serted in an attempt to hold it back.? (The solid curve in Fig. 3-13 [43] indicates the course of the power trace during this time.) By 18:07 with the power at 2 Mw, a negative period of about 45 see was observed and power fell to 2.4 Mw in about 3 min. Control rod withdrawal was started, the reactor was critical at 18:11 and power rose to 3.0 Mw by 18:21. Then, as the power was increasing more rapidly, rod insertion began, but, in spite of this, power continued to rice. At about 18:24 three positive transients were observed with about 50?330 periods and at 18:25 a 7?1/2 sec period was indicated. The reactor should have started an automatic power setwback at a lO-sec period, but did not, and the operator scrammed it manually. The automatic electronic period scram did not actfive second The peak power indicated was 24 Mw.* (No particularly high tem? peratures were recorded during the transient.) Later examination of the periodsct?backmech? 31118111 (a mechanical actuation by means of a cam in the period recorder) showed that it worked properly only if the period decreased at a slow rate, but would not operate if the period decreased rapidly. ?Recovery from the scram was made cautiously. Criticality was attained at 19:55- Approximately 2? 1/ 2 hours after the scram reactor power reached 2.0 Rod positions were now 52 in. (132.1 em) out rather than 49.5 in. (125.7 cm) as before the scram, but the difference was, at the time, at? tributed to xenon. The rods returned to 50.5 in. (128.3 cm) by 02:00 at a power of 4.0 Mw. It was decided that the excurSion had not affected the reactor adversely. Operations continued until 13:00 when a scram was caused by a short?circuit introduced into the demand circuit for the primary pump being prepared for a flow oscillatior. test. The reactor was made critical quite rapidly {13:11) and operations continued. *The final report [4321] corrects this value to about 14 Mw, noting that ?The 24 Mw value was obtained by a linear extrapolation of the log recorder chart from power levels of about 2 Mw, and linear el?tr?im13t10? is not valid.? - til-lies. a T. J. THOMPSON I EXPECTED WITH NORMAL OPERATING CONDITIONS ACTUAL POWER mac1.51m?: I TIME (sec 1 FIG. 3?13 Machine calculation 1 of the SRE power excursion. Zero time is 17:23 on July l3I 1959. It was decided to pressurize and vent the re? actor atmosphere once to reduce the radioactivity caused by the xenon in the cover gas. At 05:50 July 15 the pressure was reduced from 1.8 psig {1.12 atm) to 0.6 psig (1.04 atm), repressurizcd to 3.0 psig (1.20 aim), and then reduced to 1 psig (L067 aim). Lowering the core pressure caused an increase of about 0.0195,- in reactivity and raising the pressure had the reverse effect. This is not a normal effect on this reactor. ?Operation was continued at a power level of approximately 3 A review of fuel exit temperature spread on July 15 showed that it would be useless to try ?to get the Edison turbine generator ?on the line? since the maximum power level attainable would probably be less than 4 This would not permit operation at the desired high inlet tern? peratures while circulating through the steam gen? orator. Alterations Were made and on July 10 at 07:04 the reactor achieved criticality at a red position showing a substantial loss of criticality since the beginning of the run. intermittent opera? tion continued at less than 2 ltiw until July 20. Several tests on preSSure effects, plugging tem? peratures, and sodium level vs riations were carried out. On July 18, the motor-generator set which supplies the vital bus stabilized power failed and operations were resumed with the unstabllized Edison supply. 011 July 20, the reactor pou er was increased to 2.5 Mw to raise loop temperaw Lures gradually to On Jul): 21 at 02:10 a scram was caused by a fast period indication. (Apparently it was attributed to un? stabilized power.) The reactor was critical again at 02:25. At 06:45 radioactivity in the reactor be~ gun to build up. At 09:45 flow was lost in the main secondary loop causing a scram. The secondary loop was restored to service, the vital bus put back on the repaired motor?generator set, and operations continued at power levels up to 4.5 Mw, sodium flow rates up to 1500 (34.5 llterfsec), and reactor outlet temperatures to On July 23, it was decided to shut the reactor down and it was scheduled for 17:00 July 24. Until 09:00 July 23, reactor outlet temperatures were kept between T00 and 800"]? (371 to 42T?C)although a few reached the 000 to (482 to THOMPSON excursion. Zero vent the re? radioactivity gas. At 05 :50 from 1,8 psig repressurizcd luced to 1 psig assure caused it}; and raising t. This is not Operation was matcly 3 Nil; are spread on useless to try Ir ?on the line? :ainablc would his would not igh inlet tcm~ the steam gen? . on July 16 at lily at a rod 5 of criticality mittont opera- until July 20. plugging tom? wcre carried itOI' set which . power failed is unstabilized Teactor power loop tempera- . On July 21 a fast period rihuted to un- critical again he reactor be- ast in the main The secondary i vital bus put rotor set. and up to 4.5 MW, spinal-fees), mt the reactor July 24. Until eraturcs wore 127?C)although (482 to 4. 41:. 185.. ACCIDENTS AND DESTRUCTIVE TESTS ?3 range. At 09:50 July 23, a reactor scram was caused by a fast period indication. It was at? tributed to an electrical transient and the reactor was critical again at 10:15. Between 00:00 and 08:00 on July 24 it was noted (while jiggling elements to dislodge foreign material) that the elements in channels 10; 12, 35, and 76 were stuck while it was known 10 was free on July 22. A scram was caused by a last period indication at 12:50 on July 24 and was attributed to an instrument transient and the reactor was made critical at 13:14. Accidental loss of auxiliary primary flow caused a reactor scram at 15:40. The reactor was critical again at 15:56. Cold? trapping was put back in service when the outlet temperature reached 510? (266 and the primary plugging temperature gradually dropped from 455 to (230 to within about seven hours. On July 26 it was noted that the fuel in channels 12 and 35 were no longer stuck and 76 was somewhat free, but 10 was still stuck. The reactor was finally shut down on July 26 at 11:20 after logging 16 de in Run 14. Post?run examination of the core showed that 10 of the 43 assemblies in the core had undergone severe melting of the cladding. - The top and bottom halves of these ten elements were separated. The zone of failure was between one-third and two?thirds of the length of the ele- ment measured from the top. The accident showed that iodine released from the elements was very effectively retained in the sodium coolant. In fact. no activity except the noble gases was detected in the cover gases In reference 147] Fillmore has considered the transient which OCCurred on July 13 in detail. The general features of the transient include a slow but steady rise in power partially corn- pensated for by control rod insortion, a sharp drOp in power, followed by another short interval of slow but steady rise, andafinal fast rise termin? ated by scram. These features are shown by the solid line in Figs. 3?13 and 3?14 On these 643 graphs 0 at 17:28 hours, July 13, 1959. The dotted line of Fig. 3?13 shows the course the power trace would have been expected to take ifihc normal reactivity coefficients and control rod worths were in effect. These calculations were made using the AIREH IBM Code. By introducing seven ramp and step changes in reactivity, Fillmore was able to get the agreement shown in Fig. 3-14. The general features can then be explained at least semiquantitativcly. The slow but steady rise at a rate of 0.04% in a 3 minute ramp in Spite of gradual control rod insertion and the negative Doppler effect is at? tributed to an abnormal rise of the temperature of the moderator which has a reactivity coefficient of; 1.7 3.1 and perhaps also? to some sodium vapor formation in partially plugged channels. (The sodium void coefficient is positive although the fuel temperature coefficient was considered to be ?1.1 or 2.5 The fast negative excursion is attributed to rod insertion reducing the increase in power rate and causing the collapse of sodium bubbles and void regions, thus improving coolant contact with the fuel. This cooling in turn reduced fuel temperature and caused again of reactivity because of the Doppler effect, and perhaps sodium void collapse, so that the control rods needed only to be withdrawn a little to again start the reactor up on a slow steady rise. This was followed bya fast transient which added +0.13% in a 5 to 10? sec ramp. It is postulated that this was caused by the more or less simultaneous voiding of about 10 partially plugged fuel elements. Study of the damaged fuel elements [48] seems to Show that thermal cycling occurred at temperatures above the a?B phase transition of uranium which would lead to fuel ruptures and aISo steel?uranium eutectic formation. It is therefore postulated that several channels Luiderwent one or more cycles of heating and sodium vapor formation followed by void collapse. The cycles in the various elements 5.0 RELATIVE POWER {N/Nol TIMElsec [03) mg 3.14 Machine calculation of the SEE power- excursion. Numbers indicate the following sequence: start ramp 3mm of 0.11 thus introducing total dip of 0,0077%; (2) hold firsc ramp constant at 0.0079; and start second ramp 3 pf?t oi a 0.3: 10?4k?fsec; hold first two ramps: constant, Start third ramp of 0.1 10"3/53c. thus introducing total on of 0.02%; hold all three ramps con- sum: (in 0,0077%. - 0.00693 and 0.02%, respectively), introduce step 3,0 .- 0,067?: (5) introduce step op a 0.029;; introduce mop dp 0.00595; (7) introduce step Ag: 0,03%. Zero time 17:28 on July 13, 1959. .a . .. .. ~sj_iismt?rrisass u. - 644 are postulated as acting independently at first with periods of the order of 2 minutes. But the. rise in' power to 5 Mw was sufficient so that ?all cycling channels were affected? and voiding in all occurred more orless simultaneously. Further, one can postulate that there was a reactivity inter-? action between channels and that voiding in one led to heating and voiding in others and so on.* Comments Conclusions, Recommendations (1) Systems should be designed utilizing com? patible components, materials, and ?uids in all possible cases. In this particular case, the choice of Tctralin for cooling a bearing which could leak into the sodium system was a primary cause for the accident. if this is not possible, then?Where materials are potentially incompatible under nor? mal service conditions, means must be provided to separate them. Where there are com? patibility changes with abnormal operating cen- ditions, e.g. temperature, special attention is re? quired for monitoring and controlling Such con? ditions,? [48a] (2) At the time the choice of fuel had to be made for this reactor the knowledge of fuel per? formance was not so good as it is now and there were not so many choices. It was known, however, that. uranium metal had an 0?8 phase transition with accompanying swelling at about and that uranium formed an eutectic with stain? less steel at about 1340"}? Thus, while the use of a stainless steel-uranium metal ele? ment may have been the only choice, nevertheless phase transition and eutectic formations greatly lowered the temperature at which fuel element failure and melting occurred, Designers should always try to select materials and combinations of compatible materials that will stand up as well as possible under abnormal conditions. During the planned run (Run the central fuel element temperatures must have been quite close to the cr~B phase transition point in a number of elements, IL is not good practice to use, even as temporary additives, materials or gases which may remain in the system and cause deleterious effects later, Thus, the use of nitrogen as a stripping agent for Tetralin led to a concern over nitriding later. The total cost of thctime spent and the tests run was probably greater than if a ehem~ ically inert noble gas had been used in the first place. (4) There were several instrumentation prob? lems, The principal one involving a period sct~ back has been described. The practice of incorpor? ating such an important item of equipment as a period setback as a part of a recorderis question? able. Recorders require easy acouss for main? tenance and ink?refilling, and often have very *This description, it correct, illustrates the type of reactivity effect which can be very serious since the mechanism depends upon a positive internal feedback, increasing without limit, until terminated by unless terminated first by an external means such as control rods. win-sure .t 'iaa?zh- T. J. THOMPSON poor safety Characteristics, For instance, if the slidewire breaks, the recorder goes to one ex? treme of its travel. This can lead to an unsafe situation, (especially if the recorder is being used to drive an automatic control unit for the reactor? as happened in one case). In addition, there was evidently much to be desired in the normal be? havior of the system. Reference [43] mentions several times the fact that ?the reactor has a history of spurious serams due to apparent period transients?. This problem is often encountered during the initial tests of a reactor and is often due to either improper grounding of electrical com-- ponents, electrical noise from other equipment on the same power line, or power surges, or all three, Usually it can be eliminated in a short time without sacrificing period trip sensitivity, This instrumentation system had been used since 1957 and by 1959 the problem should have been solved. The fact that the operators are described as achieving criticality after scrams in time intervals of 15 to 25 minutes would. seem to in? dicate that no thorough investigation of possible causes could have been made. intact, if Fillmore?s explanation oi the fast transient effect is correct, there may be some reason to believe that some of the. several pcriod-scrams which occurred in Run 14 were genuine, although ?No evidence was uncovered to in die ate that there were indeed genuine scrams other than during the transient at 18:25 on Jul}r 13? McDonald and DeVan [49] give the opinion that, ?the reactor instrumentation under the immediate surveillance of the operators was inadequate to indicate excessive fuel element temperatures, the blocking of coolant passages, and fission product leakage, As a result, the operators did not con? sider such indications (where they cxistcd)serious enough to warrant shutting down the reactor, Since the SRE is a ?developmental facility built to investigate fuel material-5? it would appear that additional instrumentation, as well as closer tech? nical management, might have reduced the damage to the SRE core,? (5) Frorn the evidence available in this accident it would appear that fission products other than the noble gases are retained well in sodium. This, in a sense, affords an additional safeguard unless the Sodium should then become exposed to air in such a way as to become a fire hazard. (6) The Circumstances which eventually led to this accident began as early as spring, 1958, when the first ?l?etralin leak occurred. A second leak occurred in Run 8 on November 29, 1958, and problems continued until July 24, 1959. During that time so many difficulties were encountered that, at least in retrospect, it is quite clear that the reactor should have been shut down and the problems solved properly. Continuing to run in the face of a known Tetralin leak, repeated scrams, equipment failures, rising radioactivity releases, and unexplained transient effects is difficult to justify. Such emphasis on oonthiucd operation can and often does have serious effects on safety and can create an atmosphere leading to serious ac? cidents. It is dangerous, as well as being [also economy, to run a reactor that clearly is not functioning as it was designed to function. - b?nwu, .-: . . . - bum in" mum n+4 J. THOMPSON instanceunsafe 2r is being used or the reactor-e tio'n, there was the normal be? [43] mentions . reactor has a 7:.ppar9nt period an encountered and is often due electrical com? 21? equipment on surges, or all Lied in a short rip sensitivity. men used Since said have been a are described :rams in time :ld seem to in? ion of possible it, if Fillmore?s feet is correct, love that some ch occurred in evidence was =3 indeed genuine 3.3111: at 18:25 on he opinion that, the inn-oediate inadequate to nperaturos, the fission product did not corr- zxistcd) serious :1 the reactor. ll facility built uld appear that as closer tech? :ed the damage in this accident icts other than Lsodium. This, lifeguard unless exposed to air azard, ventually led to spring, 1958, Tell. A second nbei? 29, 1958, a, 1959. During re encountered is quite clear shut down and inning to run in posted so rams, :ivity releases. is difficult to Ll operation can :5 on safety and to serious no? as being false clearly is not action. - no - .4wmlui-gm .. -. am am. ?lulu: Jeri-1;; - - ACCIDENTS AND DESTRUCTIVE TESTS ?3 In the long run, reactor economics as well as reactor safety will demand adequate centinuing maintenance at all times, and this will include early shutdown and proper action whenever there is the least doubteomerningthesituatiou. Manage? ment can and must establish this sort of attitude. 3.10 Westinghouse Test Reactor Accident [50?53] The Westinghouse Test Reactor is a light? water-cooled and ~moderated reactor utilizing highly enriched uranium fuel. The fuel element consists of three cylindrical layers surrounding a central thimble. Each layer is made up of 0.052 in. (1.32 mm} of aluminum?uranium alley,cladwith 0.0365 in. (0.927 mm) of aluminum on either side. At the center of each fuel element is athimble which permits the insertion of a sample. Unless a special experiment is involved,thecentral sectiozis are loaded with either aluminum or cobalt slugs. The use of these aluminum and cobalt slugs helps with the Shimming of the reactor for control purposes. A diagram of the core is shown in Fig. 3?15 The nine control rod positions are indicated by the black circles. Each control rod consists of an aluminum?clad cadmium cylinder with a normal fuel element follower. Thus, on the full withdrawal of a control rod from the core a regular fuel element containing 199 of U235 is left in that position. In Fig. 345 the number before the dash hidientes the top to bottorn row starting at the top of the diagram, and the number after the dash indicates the position numbered from left to right in an}r given row. The number beneath the line indicates the number of grams of U235 estimated to be present at the time of the accident. Each of the fuel elements originally contained 199 of U235 . The core has a 36 in. (0.914 111) active height and a main active diam? eter of approximately 28 in. (0,711 m) with an alunihium?water ratio of l. Atthe original licensed power of 20 MW the average thermal neutron flux in the cere was approximately 5.2 1013 neutrons/ emB?sec. The fuel element that failed (Position 6?5) contained a Special experiment designed to moni? tor fast neutron fluxes. The experiment Consisted of a set of '7 hairline nickel wires, each sep? arately encapsulated in a quartz capsule and held in recesses of a 3/8 in. (9.52 mm) diameter aluminum rod. The rod in turn was encased in a 1/2 in. mm) OD aluminum tube with 1/8 in. (3.18 mm) weep holes drilled through the wall at 4 in. (10.16 cm) intervals. This assemny was placed in the Central thimble of the fuel element in the normal manner. A flow orifice at the bottom of the thimble section limited the flow to that required for cooling this red. Other experiments were located in Positions 5?6, 1-3. 7?6, 845 and 11-1. Other fuel elements contained aluminum and cobalt, as indicated in Fig. 3?15. The coolant flow is down through the core and up through the thermal shield. The design Specifi? cations required a flowof15,000gpm (9501iler/sec) at 20 Mw with a core outlet pressure of 93 psia Iii-213M mama-Li? man?s?- 645 FIG. 3?15 core cross section. Normal circle-s: indicer fuel elements, heavy circles 3-3, 3?6, etc.) indicate control rods. Circles enclosing an are plugged, those enclosing a are high- pressure The contents of the irradiation volumes: are denoted by darkened areas at top and bottom of individual fuel ele? ments; black strip across top ul?lldal?d V?basket with cobalt, black half-strip across top solid aluminum V?basket, black half- strip at bottom experiment. Upper numbers indicate posidon, lower numbers denote number of grams of U235 estimated to be present at the time of the accident. (NOW: l?or simplification letter prefixes used in reference [50} Loiclentify positions have been omitted in the above figure.) (8.4 aim) and a core pressure drop of 15,511sia (1.05 atm). The core inlet and outlet ten'iperatures were 115 and (45.1 and The normal surface heat fluxunder these Conditions was estimated to be'1.13 105 litufhr?it2 (35.6 watt/emQ) with a flow velocity of 21.3 ft/sec (6.50 mfsec). These numbers were calculated. for 57 fuel assemblies in the core. At the time of the accident there were 78 fuel assemblies in the core including control rods- The site ventilating systems, the various waste collection tanks, the evaporator?condenser tanks, and several other Components all vent their non? condensablc gases to a 100 ft (30.5 m) vent stack located between the Process and Reactor Service Buildings. . At the time of the accident the reactor staff was engaged in a series of tests to determine the feasibility of increasing the power gratin-all}r to 60 Ill-fwd). Reference provides a detailed chronology of events. The quotations in the follow? ing abbreviated account are taken from there. According to the license amendment of January 8, 1960, to permit operation atamaximum ofGO Mw(t), certain restrictions were imposed. These restrict? ions were [50] Westinghouse shall retain the bubble for? mation apparatus and the special detection channel described in the application in the reactor during the power escalation program until stable operation at 60 Mw(t) power level has been established; 2. The ratio of the maximum heat ?ux in the reactor to the burnout heat flux shall never exceed one?half; 3. The reactor shall not be operated in each a.