ML17263A010

Research Information Letter 0061, Molten Sodium Interaction with Basalt Concrete

ML17263A010
Person / Time
Issue date:	10/11/1979
From:	Levine S Office of Nuclear Regulatory Research
To:	Harold Denton Office of Nuclear Reactor Regulation
References
RIL-0061
Download: ML17263A010 (6)

Text

• UNITED STATES NUCLEAR REGULATORY COMMISSION WASHINGTON, D. C. 20555 ICT 111919

• -MEMORANDUM FOR: Harold R. D.enton, Di rector .

Office of Nu~lear Reactor Regulation FROM: Saul Levine, Director Office of Nuclear Regulatory Research

SUBJECT:

. RESEARCH INF9RMATION LETTER NO . .6]_, MOLTEN SODIUM

. INTERACTION WITH BASALT CONCRETE

REFERENCES:

1. SAND 79-0938, Molten Sodium Interaction with Basalt Concrete: and Siliceous Firebrick ( NUREG/CR-0934)

2. Letter E. G. Case to S. Levine, March 7, 1978, RedirEfotion of Confirmatory Research Program to

--InvestigCJ.te Inher¢nt Retention Capability for Sodium and Core Debrfs (RR-NRR-76-2) -

Introduction and Summary Conclusions

. This mem.onmdum ~ransmi ts the results of :tests co_nducted at SanCii a Laboratories

.on the interactfons of molten sodium with basalt concrete and siliceous fire-brick (Refererice l}~ Th)s work was carried-out -in response to Reference 2 which req*uested .redirect.ion of the generic program on sodium-concrete interaction to support the NRR safety review of F~TF. Specifically,. emphasis and priority of testing with the FFTF types of cbnGrete aggregate, basalt and magnetite, were requested~ Thus, large scale sodium.:.basalt concrete interaction tests have been carried out and.brief reports were forwarded arid included in Supplement No. 1 to the*FFTF Safe~y EvaluatiOn Report. Ref~rence 1 documents these tests in detail and includes chemical phenomenology, water release data, and the results of smaller scale separate effects tests.

The.results of the basalt c1;mcrete tes~s from Reference 1 provide a firm basis for several conclu.sions ... An exothermic reaction will .result from a sodium spill onto concrete und*er certain conditions. of temperature and sodium-concrete contact. The. reaction will continu'e until the sodium is consumed at

• sustained liner' penetration rates of about 2.5 cm per hour. Products of the.

r~actioh po not plug defects. in liners (tes4 def~ct was 0.6 x 15 cm} but build up. with. sufficient force to readily deform the *steel liners. The siliceous firebrick also reacted readily with the sodium.

L

'I . .,,

Referi:nc~ 1, therefore, answers some of the.questions and substantiates the staff position in N!JREG..:03~8, August 1978, ,which questioned the lo.wer sodium-c.oncrete reaction rates and purported self-1 imiting concrete penetration .assumed by t.he FFTF contraetor*. The prineiple results have been forwatded~previously and were included as Appendices F through Hof NUREG.-0358, Supplement 1, May 1979.

Results

.The results*of the large-scale tests of sodium-basalt concrete interaction

,. are provided in final . report. fonri in Reference l .* There are no major *

• . . cbrrections to the fr~fotrnation ~upplied previous.ly for Supplement l *of

. NUREG-0358, but th_e more complete evaluatian*_of the test data may show

. nifnor numer_'i cal differences. Referen¢e 2 al.so requested tests on magnetite concr~te .. One such large-sc~le test .has peen completed, additional testing is in progress*, and the. resu*lts wi 11 be the subject of a future report.

• _The high temperature sodium-concrete facility used in these tests has a

• . sodium pour capacity of 250 kg and has a clata recording system for temperat.ure, pressure, g~s identificqtion, gas flow, moisture, strain

. and concrete penetratfon rate-' infdrination .. '.fhe test crucibles were poured i.n accord with F.FTF specifications with pasalt aggregate obtained

• from FFTF .. The firebrick was of.FFTF specifi'cation and was obtained

. from the FFTF suppliers for firebrick. Concrete strength was monitored by crushing cylindrical concrete test specimens on the date of the test.

Three large-sq1le sodium-b(lsalt .concrete tests were conducted. The fi*rst test* was sodium on bare basalt concrete. The test confirmed that an eriergettc reactfon results for basalt as Well as limestone aggregate.

the second test .was a crucible lined with two layers of firebrick and a steel insert to simulate the FFTF cell design. -The purpose of this test

• was to investigate the interact) on 'of sodium. and. firebrick. Post-test ofo;erva~ion~ of severe concrete era.eking in this second test and the possible influence of the crutible test geometry led to the design of the third test . . This* third t~st was "one dimensionalized" by insulating

. the crucible *walls with a steel cylindrical liner backed by MgO. This

. permitted attack 'Only Ori the bottom of the crucible in the downward direction.

The results of the first test, bare basalt concrete, show that: an

. energ~ti c reaction result~,; al i the sodium was (:onsumed; the crucible cracked~ upon heat i!'lg (not upon ~ool ing) ; and t.he w~ 11 s of the cracks w~re attacked signiffcaritly by the soqium fanning a multipointed, star-shaped, '.cross section .in the cylindrical cavity region of the crucible.

The results of the second tes_t, two layers of firebrick and a defected steel liner, show that: the firebricks were consumed-by the hot sodium; the concrete was then att_acked by the sodium; and the defected liner was not p1u.gged by the products of 'the so di um-firebrick or concrete in

Harold R. Denton . T~e third test was one.dimensionalized for linear penetration rate

. me9suren.ient by irisl!lati'ng a ste.el cylindrical* 1iner with MgO from the

• * .tyl i ndri cal concrete walls of the crucible. and by two layers of firebrick below the liner. The bottom of the liner was tack welded in place so the products 'of the sodium .. concrete.irit~raction could break it free and push 1t.upwar~without.destroy,ing *the.1ntegrity of the cylindrical wall liner. The s odi urn (239 kg' at *9.73.°K) was duinped into the crucible and

. rehea.ted

• oy

• immers~ on heaters.* The exathermi c *sodium-firebrick reaction

.. caused the thermo'<:ouple lo.ca.fed between"the firebrick* layers to exceed

. the sodjum pool temperature .110 minutes into. the test. The thermocouple loca~ed one dn belciw the cqilcrete surface indicated a rapid excursion at 170 min. A similar *excursion *6 cm. below tile surface followed at 280 minutes. These temperature excursions within the concrete became self-1imited* at ab_out f,000°K. The* concrete J)enetration proceeded at a uniform r~te, fOrf!1.ing layered products of react.ion, through 25 cm of concret'e .. Only 13 cm 'of con~rete remained when gross cracking dropped a Small amount (< 5 kg) of sodium out of the crucible and the test was terminated.

The results of this third test clearly indicate: an energetic reaction

• results from the sodium-basalt 'concrete interaction; the sod1um-firebrick reQction fs *not inhibited by .a . defected steel liner; the penetration rate fotc) the, firebrfck is a*s large as *the penetration rate into concrete; the penetration rate into .the b.asalt concrete is very linear and is not dim.inished by lay~red products of the reaction; the reaction continues

• until th¢ sodium 1s*consumed; and the basalt concrete exhibited no significant spallihg under the conditi6ns of the test.

Reference l also summarizes studies of the chemi.cal p*henomenology involved.

The process _involves temperature,*ton$tituent concentrations and kinetics of the reactions. Separate effects' tests* on small and intermediate scales.. are continuing to define the processes for basalt, limestone and magnetite concretes.

From the results of the three large-scale tests, however, the following appear to be substant1ated by . *cheinital analysis of the products of reaction,. the' qbserved exothe~ic reactions, and the gas evolution. The initial r:-esponse* of t.he conc.rete to t~e hot sodium is the migration of water. This water migrates dawn the .thermal gradi.ent from the heated concrete surface> and simu1t.aneously down the concentration gradient toward the hot surface. The water re~cts (at temperatures of 750 to l ,000°k) with so'c;liJ.1m forming sodium hydroxide and free hydrogen. The hydrog~n formation h:as peen confirmed by the compo~ition measurement of thE! evolved ga~;'es .. However, the sodium hydroxide i*s very soluble in sodium metal at 800° to l,000°K and does not react with the concrete until th.e sodium has (?ecome*saturated with sodfom hydroxide. Additional formation of sodium hydroxi'de results in the precipitation of a liquid layer. .This liquid 1ayer of sodium hydroxide. reacts with the s i1 i ca and al1Jlnina components* of the con.Crete. Water iS released and reacts rapidly with .the sodfom to continue the production of sodium hydroxide*and hydrogen at the reacting interface.

• Harold R. Denton Evaluation

_.The test results show that the sqdi_um-basalt GOncrete reaction does

• occur at temperatures of 750° to l;000°K. Thus, a sodium spill must be isOiated ftom bas'alt c:cmcrete qr, provision made for acco~odating the produtts of the *reattt'on. Defects: of the order *of 10 cm in a steel liner provid*e adequ*ate access to the concrete for continued sodium-concrete interaction~ The interaction is -self limiting w.ith respect to temperature. This self limiting effect may be responsible for the very uniforri(rate of p~n:¢tr_-ation which -r~sults. The penetration rate is

'V 1.5. tm/l:tr for basaltunder the defected liner and appeared to be less for tne *sihgle :bare basalt test. The.total concrete penetration is about 40% to 50% of the initial sod.ium pool depth. Thus, the reaction cou:ld be terminated by providing *sufficient sacrificial concrete.* On the other *hand, v_ery deep pools of sodium may not become saturated with so9ium. hydroxide before :the pool temperature drppsto a value (apparently about 700°K) where the .reaction will not occur.* The understanding of this chemit;:a1 phenomenology is fncompfote and- continued effort is required in tl:tis area~ The* concept of the saturation requirement for sodium hydroxide ~pp¢ars well_ fourn:led from th,e longer time required for the exothermic' reaction to initiate when the whole pool is in contact with the concrete-. The overal 1 reactions for 1imestone, basalt and magnetite indicate s1milar chemistry.*_* Only one magnetite test has been run, but the penetration was similar. The most significant difference in the magnetite te_st was the greater structural integrity of the crucible after the test. Only- very fine cracks in the walls were apparent. In addition to the basic chemistry and intermediate scale separate effects tests the 1arge scal_e magnetite ~e~ts wi i,. be completed and the generic

• prc>'gram re_siJmed. Since the 10 cm _defect provides full access for the sodium to the concrete, tests 'are planned using a series of smaller more realistic defects which would simulate various crack sizes in cell liner piates. The NRR staff has noted the desirability of the resolution of this crack size effect.

The crucible does not simulate an actual reactor cell; however, crucibles with insuiated walls do provide very satisfactory tests for the determination of -Cme dimensional penetration rates~ Compar~tive structural analyses of cell geometries and possible test geometries are now in progress.

The work reported in the enclosure documents the information on the sodium-basalt concrete interaction in suffiCient detail to evaluate the

. bo_undirig behavior of bare 2oncrete or conc~ete with* a steel liner and a defect large enough (10 cm ) to assure sodium-concrete contact. Finer

1-Harold R. Denton cracks,.which simulate poor welds, will be evaluated in the future, but

• .*such *te$tS wil.l require ~nalysis and design to assure that deformations of *the liner will not increase the cra<;:k areas beyond the test size.

Generic studi'es *of other concrete *aggregates wi 11 al so be continued.

For further information on these results, their use, and the continuing researdi in this area, pl ease contact Dr. T. J ~ Walker of my staff.

~,

0 *~ Director Office of Nuclear Research

Enclosure:

SAND 79-0938 (NUREG/CR-0934)

j.-;_

- _,;;* ........ ~, ........ ' _J Harold R. Denton ... .' * '. ~- *!

crack~:;:,,~1*ch simulate poor welds. w111 be evaluated in the future. but such tests will require analysi~ and des.ign to assure that defonnat1ons ' .. * .* ~ ;. .

.. . . '!~ **: ...

of the 11ri~r wi 11 not increase the *crack areas beyond the test size. * *

... Generic studies of other concrete agg*regates will also be _continued.

• For further 1nfonnat1on on these results, their use, and the continuing research in this area, please contact Dr .. T. 4. Walker of D\Y staff.

• e~-;srn~F ~;~-:~ * ~:J s{.',~~k Le*>>"HiO Saul Levine. Director Office of ~uclear Research

Enclosure:

SANO 79-0938 (NUREG/CR-0934)

DIST:

Subj -.1 ._.*.

~ .. - . .

Circ *,

Chron L :.

. Wa 1ker: ROG Silberberg:CY Kelber:CY Murley:CY Budnitz:CY Levine:CY Larkins:CY

'\**,

. ~;

_-, ... S.EE PREVIOUS YELLOW FOR CONCURRENCES .',"" :~ ~

• . .-: ~~!

• -~----r..--------r------...--------.------.------~--- . .--:. - **:'I SU::~::: ~~~!!~~ :~~!~~=f~: -~-:~:-~'.~~

DA TE .... _Ll2LZ9_____________ __9_ll'JLZ9_ ___________

~~-;ir;~~:~:::_ ~:-~::~;~~  :=-~l;t_=

---

~Lg§L?.~---------- ____ J_Q/.J)_LZ~----- ---~J_Q!!LlZ~----- ____ __]_QlJl_l~~-- ' .

~*