ML19260C930

Safety Evaluation Re NEDE-21821-02, BWR Feedwater Nozzle/ Sparger Final Rept, Suppl 2

ML19260C930
Person / Time
Issue date:	01/14/1980
From:	Office of Nuclear Reactor Regulation
To:
Shared Package
ML19260C929	List: ML19260C928 ML19260C930
References
REF-GTECI-A-10, REF-GTECI-RV, TASK-A-10, TASK-OR NUDOCS 8002070142
Download: ML19260C930 (27)

Text

.

• / %

SAFETY EVALUATIO'l FDP THE GEf'ERAL ELECTPIC TCPICAL PEPDPT "PPP FEEDPATEP f!0ZZLE/SPAPGEP FIflAL DEPOPT, SUPPL EPENT 2" (f!EPE-21P21 02)

PDEPAPED PY OFFICE OF flUCLEAP PEACTOP PEGULATION U. S. fillCLEAP REGtlLATORY C0f4'ISSION PEVIEWEPS:

R. P. Snaider, Systematic Evaluation Progran Franch, Division of Operating Deactors (Task f*anager)

R. E. Johnson, Engineering Pranch, Division of Coerating Peactors P. S. Hazelton, Engineering Branch, Division of Operatina Deactors P. P. Klecker, Engineering Pranch, Division of Operating Peactors J. J. Zudans, Engineering Branch, Division of Operatina Peactors P. K. f'attu, f'echanical Enaineering Branch, Division of Systens Safety f'. P. Hun, l'aterials Enaineering Branch, Division of Systens Safety S. D. l'acKay, Plant Systens Pranch, Division of Operating Peactors

?. 'l. Pandall, Structures and Conconents Standards Branch, Office of Standards

'eveloprent f'. Subudhi, Erookhaven ?!ational Laboratory L936 329

/V1 8002070

BWR FEEDWATER N0ZZLE/SPARGER FINAL REPORT SAFETY EVALUATION TABLE OF CONTENTS

3.0 DESCRIPTION

OF SOLUTIONS 3.1 GENERAL ELECTRIC TRIPLE SLEEVE SPARGER DESIGN 3.2 CLAD REMOVAL 3.3 MODIFICATIONS TO FLUID SYSTEMS AND OPERATING PROCEDURES 4.0 VERIFICATION OF SOLUTIONS 4.1 VIBRATION TESTING 4.2 THERMAL - HYDRAULIC TESTING AND ANALYSIS 4.3 MATERIALS TESTING AND SELECTION 4.4 THERMAL FATIGUE ANALYSIS 5.0 OTHER SPARGER DESIGNS 6.0 ULTRASONIC TESTING AND RECOMMENDED INSPECTIONS

7. 0 IMPLEMENTATION 8.0 SAFETY CONSIDERATIONS

9.0 CONCLUSION

S

}9bb b

1.0 IflTROPUCTI0ff r neral Electric Coroany (r-E)

Dy letter dated September 4,1979, the e

submitted for staff review a topical report entitled "Poiling Vater Peac-tor Feedwater flozzle/Sparger Final Peport, Suoplerent 2" OlEDE-21P21-02, Auaust 1979). The docurent incorporated in full the earlier recorts on the sare subject, NEDE-21821 (f' arch 1978) and t'EDE-21821-01 (January 1979), the related changes forwarded by GE letters dated f# arch 1,1979 and t' arch 20, 1970, and changes in response to the staff's recent com-rents on flEDE-21821-01. The report provides generic inforration relative to (1) the design of a rodified feedwater sparger and themal sleeve assembly; (2) testina and analysis of this design; (3) analysis of nozzle crackina, including the identification of the causes of such crackina and the safety implications; (4) analysis of other rodifications, such as nozzle clad renoval and system changes, which would prevent cracking or decrease the rate of crack propagation; and (5) discussion of non-destructive examination (flDE) rethods and recormended applications for inspection of BWR nozzles. The reports did not address the related control rod drive return line nozzle problem. This ratter is beina handled separately by the '!RC and GE.

The flPC topical report review included the ceneric desian and analyses of the rodified sparger/ sleeve assembly, descriptions and analyses of the other available solutions and verification of solution effectiveness (includina identification of the causes of cracking). 'he review also assessed the ability of the soarger and unciad nozzle reofons to withstand BUR environmental conditions during design lifetires, the capability and limitations of the proposed inservice insoection proaran and the proposed inspection freauency.

1936

$31

2.0 PACKGROUND Of the 23 operating BWRs in the United States with feedwater nozzle /

sparger systems (nomally 4 nozzles /spargers per BWR; nominal nozzle diareter 10 to 12 inches), 22 have been inspected to date and cracks have been discovered in the feedwater nozzle blend radius or bore of 18.

Although nost cracks have been relatively superficial, a few grew to 3/4" total depth (including cladding) and one vessel exhibited cracks which penetrated the base retal to a total depth of approxirately 1.5 inches.

Crack initiation results fran high cycle therral fatigue as the internal water temperature fluctuates in the themal sleeve-nozzle annular reaion durina periods of low feedwater terperature when the flow ray be unsteady and intemittent. Once initiated, the cracks are driven deeper by the larcer pressure and themal cycles associated with startup and shutdown.

Fracture analyses indicate that the cracks found to date in the feedwater nozzles constitute a potential safety problem because the observed rate of crack growth is such that the raroin of safety against fracture would be reduced below acceptable values unless the cracks are detected and recoved periodically.

In cases of severe crackino, repair by arindout could be repeated only a few tires before ASf'E Code linits for nozzle rein'orcerent were exceeded. Pepair by weld buildup in the crindout region has not been demonstrated as yet to be acceptable to the flPC.

In addition, inspection and renoval of cracks by arindir; involves radiation exposure to personnel and is deemed unacceptable as a long-tem solution.

Extensive and long-tem study of the causes of the problem and the efficacy of the rodified sparger design has been undertaken independently by the General Electric Coreany (GE) and the flRC staff.

The GE studies have been docurmnted in the report evaluated herein which represents a surmation of the engineering desian, test, and development effort under-taken and accorplished by GE. The NPC staff has worked closely with GE and others in the effort to understand and resolve the conolex safety issue. The f!RC staff published interin guidance in the fom of f!UPEG-0312, " Interim Technical Peport on RWR Feedwater and Control Pod Drive Peturn Line flozzle Cracking." This guidance will be superseded by the f!UPEG docurent to be issued at the conclusion of the f!PC Staff's generic study.

The safety objective of these efforts was to assure lono-tem reactor vessel integrity.

Effective sparper desian also would pemit an increase in the interval between in-vessel surface exaninations. Ue have concluded that the safety objective has been ret.

I936 337.

2

3.0 DESCRIPTION

OF SOLUTIONS 3.1 GENERAL ELECTRIC COMPANY TPIPLE SLEEVE SPARGEP CESICH l'ost of the PWPs in operation today went into service originally with icose-fittina sparger therral sleeves. After discovery, by inspection, of the various cracking problens caused by the loose-fit desian, licen-sees replaced the original scarcers with an interin, interference fit desian. No dorestic BWP is operating with a loose-fit desian today.

Although the interin interference fit desian can reduce bypass leakaae flow, its lona-tem effectiveness has been called to cuestion because the interference fit may degrade with tire.

Pecause of these prchlems, GE has desianed an "frproved interference fit sparger" as described in Chapter 3 of NEDE-21821-02. This sparger desian, also called the " triple sleeve scarcer" in this report, has been reccrrended by GE as a replacerent for the interin single sleeve desian rentioned above.

The improved interference fit sparcer desian was based on the service experience discussed above and on the themal-hydraulic test results described both in Chapter 4 of ffEDE-21821-02 and in Section 4.0 of this evaluation. The tests confimed the postulated crack initiation and growth rechanisrs and served as input in desiens to ritiaate such crackina.

The tests revealed that, for the oriqinal loose fit desian, the cause of themal cycling was:

"... prirarily... leakage flow passing between the themal sleeve and safe end. This leakace flow, which is at fee 6. ater temperature, rixes in a turbulent canner with hot downcomer flow in the annulus between the nozzle and themal sleeve. The mixing fluid impinces on the nozzle wall causina themal cyclina of the retal su rface.

It has been detemined by test and ' field'reasurerent at Millstone fl] and Brown's Ferry 2 that the retal temperature cycling, with leakage present, has a ragnitude of up to 50% of the difference in temperature between the feedwater and downcorer etcr.

This cyclina occurs with frecuencies between 0.1 Hz and 1 Hz and thus can initiate crackino rapidly. The exact tire to crack initiation depends on several factors includino the duration of operation with low feedwater terperature."

The triple sleeve sparger was designed to prevent the themal cyclina phenomena, thus significantly reducing the likelihood of crack initiation within the lifetire of the plant. The sparger utilizes three concentric themal sleeves, the innnemost of which conducts feedwater to the soarger ams. The ams are attached to the sleeve by a forged tee, fastened to the reactor vessel wall at their end points by brackets, and are designed to deliver feedwater unifomly to the annular area between the core shroud and the vessel wall.

In so doina, they provide subcooling for the jet puros and help raintain a unifom core power distribution.

1936 333 3

Bypass leakage flow in the feedwater nozzle bore will be reduced substantially by two piston-ring seals and an interference fit. Thus thermal cyclina will essentially be elininated.

The first, or upstrear, therral sleeve piston-ring seal foms the prirary seal between the innemost sleeve and the nozzle bore. Vater leakina past this seal would pass into the vessel throuch the annulus between the inner sleeve and t 'e "mid-themal" (CE teminoloovl sleeve, which is supported at its upstream end by a slotted attachrent to the inner sleeve. Attached to the " rid-themal" sleeve is an outer sleeve which is fitted tichtly in the nozzle bore at its upstrean end to prevent vibra-tory rotion of the sparger asserbly. The secondary piston-ring seal at that tight interference joint reduces potential bypass flow to nearly zero because the pressure drop is very low across the secondary seal.

The sparger arrs were rodified in the triple sleeve sparger design in one important respect.

Flow is no loncer discharged into the vessel thrauch holes in the sides of the sparger ams, but through elbows rounted on top. The elbows are fitted with convergina discharce nozzles. These features reduce temperature stratification in the sparger and flow separation around the periphery of the flow holes at low feedwater flow (It had been observed that the cold feedwater roved along the bottom of the pipe during low flow producing a very larae top-to-botton terperature dif ferenti al. The resultina therral stresses caused therral sleeve cracking. Sirilarly, the flow separation had caused flow hole cracking).

Sparger am cracking, which was another problen with the loose-fit design, had been solved earlier by use of a forced tee to replace the older tee box.

The NPC staff considered from two aspects the ability of the triple sleeve sparger to perfom its function. These aspects were its effec-tiveness in reducina themal cyclina of the nozzle bore and blend radius

~

and its durability. The two considerations influence both the criteria for inspection freauency and the inspection rethod to be recorrended in the forthconino NUPEG report. Effectiveness will be discussed in Section 4 Reaardina durability, the principal consideration is the loss of sealina ability as a result of wear or corrosion. There is sufficient experience to justify the assertion that the desian is acceptable in this reaard, althouah corrosion of carbon steel safe ends under the Diston rina seal is rentioned by GE as a cotential problen warrantina introduction of special claddina (high ferrite, unsensitized 308L stainless steel) on the sealing surface in new plants.

The staff also feels there is some cuestien about the durability and fatigue resistance of the triple sleeve assembly, especially at the slotted connection of the "nid-themal" sleeve to the inner sleeve. No specific weakness has been identified but past experience with feedwater sparger problems indicates that it will be prudent to ronitor the perforrance of the first units to be installed, and especially to ronitor for leakage.

1936 334 a

3.2 CLAD PEPOVAL The inner surface of licensed BWP reactor vessels, includino feedwater nozzles, was clad with stainless steel. The weld-deoosited overlay was oriainally installed for corrosion protection of the carbon steel vessel and to ninimize rust accurulation in the vessel water.

Powever, renoval of nozzle claddina coincident with installation of the inproved sparger design is now recorrended by the General Electric Com-pany. Analyses show that clad renoval results in about a factor of two reduction in cyclic therral stress at the surface of the retal. The net effect of clad renoval is to prolong the tire to crack initiation if the magnitude of tenperature cyclina is low. Peroval of the claddina also increases the nurber of startup/ shutdown cycles reauired to crow fatigue cracks to the liniting depth as specified by the applicable code. This results from the elimination of stresses due to differential themal expansion of the stainless steel and carbon steel near the surface.

Renoving the cladding also facilitates the interpretation of ultrasonic (UT) signals by elininating the clad-base-retal interface, a comron source of spurious indications.

It renoves any retal that ray have suffered faticue darage.

Pecause sore base retal is recoved along with the cla.1 (arounts rance fron 0.1 to 0.5 inches of base rmtal removal), a recneck of the cross-sectional area available for nozzle reinforcerent is reauf red to verify that ASPE Code rules will still be met. The limited experience to date has indicated that this should not be a serious problen, provided the clad removal machinina operation is perforced with full recard for the as-fabricated dirensions and alignnents of the nozzles and safe ends.

Stress corrosion cracking of the cladding and base retal was considered by GE with the conclusion that clad reroval had a cositive effect. The stainless steel clad in some vessels has a relatively lcw ferrite con-tent; iow enouah to render it susceptible to stress corrosion crackina.

Although no instances of feedwater nozzle crackina have been attributed to stress corrosion to date, CE believes it possible. On the other hand, it was stated that the chances for base retal cracking by the BWR envi-ronrent is slight.

Pitting or general corrosion of the exposed base retal is not expected to be a probler, because there have been no corrosion problers with partially-clad nozzles nor with the areas in existing nozzles where grinding to rerove cracks had reroved the cladding and exposed base retal.

The EWR reactor vessels for plants undergoing licensinc review contain unciad feedwater nozzles. The feedwater nozzle areas of future BWR vessels also will be unclad. Peaardina clad removal at existina plants, rachine tools have been developed by GE and others to remove the claddina from the nozzle blend radius and bore to prepare the seatina surfaces for the seals on the thermal sleeve. Typical clad thickness encountered in crinding out cracks was 0.25 in.

(The rance was from 0.20 to over 0.50 inches.)

1936 335 5

Pased on its own independent assessment, the staff concurs with de CE assertion that clad removal offers a net benefit toward the acal of nini-rizino the likelihood of crack initiation.

For sene reactors with hinh (420*F) operating feedwater terperatures, the corbination of clad rencval and a zero leakage triple sf eeve sparger ray be all that is necessary to suppress cracking within the design lifetire. Other reactors with lower feedwater terperatures nay require systems changes as noted in Section 3.3.

3.3 f'0DIFICATI0t:S TO FLUID SYSTEMS AND OPEPATING PROCEDL' PES 3.3.1 Objectives ffEDE-21821-02 indicated that the overall objectives of the various solutions for the PPR nozzle crackina problen were:

(1) to prevent the initiation of cracks and (2) to linit crack crowth to less than 10% of the wall thickness during the life of the plant, i' they do initiate. CE considers liniting crack growth to be the rore irpor-tant objective. However, these objectives nay not be ret for all operatino olants if the cl.addina were renoved and a sinole sleeve sparger with zero leakage were installed. This is particularly true for those plants with lower feedwater temperatures durina full power operation (such as 3a0 F rather than 425 F). These ob.fectives could be ret for these plants by clad reroval and with a triple sleeve sparger with zero leakage. However, the sparger perforrance is very sensitive to leakage and it is not certain that leakace would be avoided durina the life of the plant. Therefore, it is advisable to augrent clad reroval and scarcer redesign with rodifications to fluid systens and chances to operating procedures, in order to further reduce therral cycling within the feedwater nozzle.

Althouch clad renoval and sparcer redesion may not, by themselves, be accepted as a ceneral solution for all BWPs, CE analyses indicate that systen and procedural changes alone would not be sufficient to meet the overall objectives. Therefore, a particular solution must be derived for each BWR and will in rost cases consist of clad removal, a new sparcer design and sore systen and procedural changes.

3.3.2 Soecific System t'odifications 3.3.2.1 Low Flow Controller The low flow controller would be used to control feedwater ficw over a rance of flows fron 0.5% to 10" of rated flow for the purpose of reducina thermal cycling during periods of low feedwater flow and high subcooling. Analyses show that systen changes in ceneral do not rake a larae contribution to delaying crack initiation.

However, there is also analytical evidence which shows that a low flow controller would be necessarv to linit crack growth to less than one inch in 40 years.

1936 336 6

3.3.2.2 Deactor Water Cleanun Syster, (PWCll)

This systen rodification would involve rerouting the discharge of the PWCU to deliver the flow to each feedwater nozzle. Althouah flEDE-21821-02 shows that systen changes in general do not make a larae contribution to delaying crack initiation, it does show that rerouting the PWCll can decrease the usage factor with respect to crack initiation from.70 to.46.

This would represent a signifi-cant usage factor reduction in those plants where reroutina is feasible.

3.3.2.3 Other Systen f'odifications NEDE-21821-02 presented an evaluation of the low flow controller and the reroutina of the PVCU in tenas of linitina crack initiation and crack growth. Althouah only these two possible rodifications were evaluated, other solutions ray exist and are not excluded by this GE report.

3.3.3 Plant Oneratina Procedures f1EDE-21821-r 2 suggests that there are many improvenents that can be inclemnted to reduce therral cyclina in the feedwater nozzles. A

" Proposed Alternate Operating Procedure" conbined with sore systen modifications was evaluated and the results aiven in Table 4-31.

The proposed procedure consists of the following:

1.

PWCU flow would be direc ted to all feedwater nozzles at maxinun flow rate and exit tenn rature durina all low flow conditions prior to turbine loadir I.

Sore plant designs would reouire piping changes to achi-le this.

2.

The turbine would be accelerated, synchronized and loand at a reduced reactor pressure of 600 psia (instead of 1000 ( +1a).

l'ain stean bypass just prior to turbine acceleration would be the minirun compatible with that action (approximately 5",).

Operating plant procedure chances would be required to achieve this. To our knowledae, early turbine roll has not been atterpted yet at any operating facility.

3.

Turbine extraction heaters (at least the top heater) would he in service at the tire of, or before, turbine loading to 5".

f'ost feedwater train designs, including heater drain charac-teristics, are coroatible with this operation. Sore heatcr eouiprent chance night he recuired in a few cases to achieve this.

4 For start-ups and shut-downs, the feedwater control systen would be capable of low flow contr31 sufficient to elininate on-off feedwater operation and with sufficient controllability to preclude greater than 25 F peak-to-peak nixture tercerature variations durino steady denand. Though this feature contri-butes sore benefit toward reduction of high cycle fatigue, it I95b 33[

7

is the single rost effective feature applied to ritigate the icw cycle fatigue problen discussed in other sections.

5.

Plant operating procedures generally would be rodified to mininize the total tire spent at larce subcoolina and to reduce the subcooling experienced for long periods of time, particu-larly at high feedwater flow rates.

The evaluation presented in flEDE-2182102 showed that the laraest improverent in the crack initiation usace factor would be achieved by reroutina of the PPCU, i.e., a reduction in usaae factor fror.70 to.a6 The early turbine roll (flo. 2) and the early feedwater beatina (flo. 3) each would reduce the usaae factor by about 10 percent. The low-flow controller would have little effect on the cracF initiation usage factor but is necessary for limitino crack prowth.

Based on its review, the staff concurs with to GE assessment of syster rodifications and the benefits to be achieved ty their installation.

Plant-specific review will be necessary in order to determine what combination of modifications is acceptable and necessary.

}9bb

a.0 VEPIFICATION OF CE TPIPLE SLEEVE SPAPGEP DESIGN AS AN EFFECTIVE SOLilTION 4.1 Vibration Testina One of the problers associated with the oriainal loose-fit sparaer desion was flow-induced vibration.

Such vibration contributed to the formation of cracks at the junction of the sparger ams and the themal sleeve.

The vibration was induced by the flow of water through the cap between the themal sleeve and the nozzle safe end. The fatigue damage was aggravated by the gecretry of the original tee box junction between the themal sleeve and sparger ams.

The gap, and therefore the flow of leakage water, has been elininated in the interim interference fit desian by the tight fit between the nozzle and themal sleeve. The gecretry of the sparger/ sleeve junction has been nodified in alrost all existing plants by the use of the forged tee, which provides less flow resistance. However, the interin interference fit is not expected to retain its tightness against accurulated themal working of the nozzle and themal sleeve. Therefore, as stated in Section 3, GE has recommended the installation of the triple sleeve sparger, which utilizes piston rina seals in addition to an interference fit for the purpose of eliminating leakage flow over the long tem.

Testing was reauired to identify the vibration characteristics of the triple sleeve sparger design to verify that this sparger would not experience c"nditions similar to those which resulted in problems with the oriainal o ians. The experirental goal was to deronstrate that the sparcer was vie:ation-free during all operatino flow reaires, thus helping to assur9 lona sparger operating life.

As described in Chaptcr 4 of NEDE-21P21-02, CE's test facility was able to acccmmodate a full-scale sparger, and several different variations of the triple sleeve design were tested. The recirculatina loop providina water to the sparcer could deliver 5300 gpr flow at approximately 32 psid across the sparger.

The experirents involved flow sweeping (rodifyina flow slowly and stead-ily) fran minirun to raximum anticipated flow.

During the flow sweep, instruments recorded strain, acceleration, and displacement concurrently with differential pressure across the sparger. The instrurentation included accelercreters (radial, vertical, and circunferential), iendina strain gauges (vertical, Forizontal, and radial), and displacement trans-ducers to sense vertical, radial, and circumferential notion.

The intent of the progran was to sirulate all loadings that the sparger would see during all phases of reactor operation, including self-generated and externally applied loads. To obtain a conservative range of results, leakaae flow was an active test variable in the five rcckups, varyina From essentially none to substantial flow.

In general, the vibration levels of the triple sleeve scarcer were acceptably low for all flow and load variations tested. Spectrum analyses were performed where strain or displacement sensor amplitude readings 1936 339 e

were higher than nomal. These analyses were to determine the values of strain or displace.r,ent at single fundamental response frequencies. Care also was taken te allow sufficient time during sweep testing so that any tendency for vibrat'an amplitude build-up from structural resonance would be noticed.

In ali cases, the experimental values of strain and displacement were low.

The tests also indicate that, assuming all external driving forces were modeled in the testing, the triple sleeve sparger is free of vibrational problems. We conclude that the tests were representative, the results acceptable, and that this design has solved one of the problems recog-nized in the original loose-fit design and potentially present in the interim interference-fit design. Therefore, the design is acceptable from this aspect.

4.2 Thermal-Hydraulic Testina and Analysis Although the cause of crack initiation was generally assumed to be thermai fatigue resulting primarily from leakage flow passing between the thermal sleeve and safe end, extensive testing was considered necessary to characterize the flow instability and to test the various design solutions under consideration.

The open tank used by GE for vibration testing of full size feedwater spargers was modified to provide a 100 F temperature difference between the simulated separator downcomer flos and the feedwater flow. The nozzle area was instrumented for temperature measurement by the same type of sensors as those used in operating reactors (Millstone 1 and Browns Ferry 2). The natural bypass leakage around the themal sleeve was pre-vented by 0-ring seals and controlled leakage was introduced at taps around the safe end circumference. The tests run in this two-temperature test (2T2) facility provided the basis for the explanation by GE regard-ing the causes of feedwater nozzle cracks, and the 2T2 facility was the proving ground for various proposed sparger design alternatives to stop the thermal cycling.

A typical test of a given configuration required several runs at different feedwater flows.

From the temperatures taken at a given location during a 4-minute time interval, the peak-to-peak amplitude was measured and reported as a percentage of the available temperature difference at that instant.

In the first tests, the facility was fitted with a loose-fit T-box sparger sleeve like the original Millstone i sparger.

The pattern of temperature cycling was found to be similar to that at Millstone 1, and the amplitude of thermal cycling was in proportion to the difference in the available aT (difference between reactor and feedwater temperatures).

Substantially greater AT exists in an operating reactor.

Some of the significant test results were:

1.

For large bypass leakage flow and low fredwater flow, the cyclic aT of the water near the nozzle blend radius was nearly 100% of the available aT.

10

2.

The amplitude of the cyclic temperature at the blend radius metal surface was approximately one-half of the water temperature amplitude.

3.

Cyclic AT midway along the nozzle bore was dependent on the leakage flow rate, which apparently determined where the mixing of hot and cold water was taking place.

For the forged tee sparger, the cyclic AT results for the bore and blend radius were about equal at low to medium leakage rates. Above the medium rate, the values for the bore dropped off sharply.

4.

With no leakage, the cyclic AT was about 20 percent at low sparger fica, increasing to 30-40 percent at high flow. A concentric double themal sleeve reduced this to 10 percent.

To confirm that themal fatigue was the cause of the feedwater nozzle cracking, tests were run on large rectangular specimens containing a cen-tral hole through which hot and cold water flowed alternately to produce themal cycling while the specimen was under a tensile load. Cracks were initiated by this method. The number of cycles required to produce a crack was reduced when the hole surface was clad with stainless steel, and was least when the clad had been heavily coldworked by a chamfering operation.

As noted above, various design alternatives were tested in the 2T2 program, such as:

1.

A vortex suppressor, consisting of a vertical plate fastened beneath the forged tee in a plane containing the axes of the nozzle and the vessel, was tested to see if it would reduce thermal cycling of the blend radius region by preventing changes in the leakage flow path from one side to the other. The vortex suppressor was effective in reducing thermal cycling when there was significant leakage flow, but was considered unnecessary with the triple-sleeve design, which minimized leakage.

2.

A flow baffle, consisting of a disk placed around the sparger sleeve at the vessel ID to close the annulus opening, was tested but the idea was abandoned when the baffle was found to cause severe stratification of the water in the annulus region near the safe end.

3.

A' hot flushing concept was tested to detemirs whether the introduction of hot water at the safe end to flush cold water from the annulus would be beneficial. Results indicated that the required flow of hot water would be quite high and other concepts were deemed to be preferable.

4.

A concentric double thermal sleeve design was tried in varous configurations

- the evolution of the final configuration of the triple thermal sleeve design.

Finally, the 2T2 facility yielded heat transfer data useful for calculating operating conditions other than those simulated in the tests.

Comparisons also were made between the in-reactor measurements at 1936 341 3,

Millstone 1 and Browns Ferry 2.

Based on comparison of the relationship of metal temperature thermal cycling to that of the water in the annulus, the heat transfer coefficients were considered to be surprisingly high.

Confirmatory tests of the effectiveness of the triple sleeve sparger design were performed in a GE test facility near Pacific Gas & Electric Company's Moss Landing Power Plant in California. Feedwater and super-heated steam were obtained to provide test temperatures that matched BWR operating conditions. The test was full scale with the exception of sparger arm reduction in length to fit the test vessel. The Moss Landing test was required because the 2T2 temperatures (70 F feedwater,160 F reactor water) did not provide a sufficient density difference to simulate the cold feedwater stratification in an operating plant. The Moss Landing facility did accurately reproduce the temperature fluctua-tions found in operating reactors. Thermal cycling was shown to be reduced to acceptable levels with the triple-sleeve sparger design. The testing also showed that the thermal hydraulic performance of the triple-sleeve sparger design is acceptable.

In addition to the above confirmatory tests of thermal-hydraulic characteristics of the triple sleeve sparger, the new design also was subjected to a thermal shock test. This test in a separate Moss Landing fixture, was intended to verify behavior of the seals and interference fit and to verify the mechanical integrity of the tested components. A total of 110 thermal shocks was imposed by heating the sparger sleeve seal region to 550 F, then quenching with 70 F water. The results revealed some tendency of the piston rings to bind in their grooves and malfunction. As a result, minor design changes were made. The inter-ference fit relaxed from 0.023 to 0.010 inch in 20 cycles (before housing rebore to simulate corrosion of the safe end sealing surface).

The NRC staff believes that the testing done in the two-temperature te:t facility and at Moss Landing demonstrated that the thermal-hydraulic phenomena that caused feedwater nozzle cracking have been reproduced in the laboratory. The test results are sufficiently quantitative to pro-vide an adequate basis for analysis of new designs. With regard to the triple-sleeve sparger design, the staff has concluded that the test results demonstrate that it should be effective in reducing thermal cycling of the feedwater nozzle bore and blend radius areas.

4.3 Materials Testina and Selection Section 4.6 of NEDE-21821-02, entitled "Sparger Life," describes the materials selected for the triple sleeve sparger and summarizes the stress analysis and fatigue analysis that accompanied its dcsign.

Inconel 600 was chosen for the piston ring seals and the upstream end of the thermal sleeve to obtain a close matrh of the coefficient of expan-sion of the seal, the sleeve and the safe *.nd.

The triple sleeves are Type 316L stainless steel. The sleeves cannot be solution treated after welding, hence the low carbon Type 316 stainless was chosen to prevent stress corrosion cracking.

As noted in Section 3.1, the staff feels there is some question about the durability and fatigue resistance of the triple sleeve assembly, 1936 342 12

especially at the slotted connection of the mid-thermal sleeve to the inner sleeve.

Although no specific weakness has been identified, past experience with the feedwater sparger problems suggests that early inspection of the first installed units may be prudent.

However, with the exception of the possible problem with the slotted connection, the staff has found the selection of materials to be satisfactory.

4.4 Thermal Fatigue Analysis 4.4.1 Objective General Electric performed an extensive fatigue analysis as part of the triple-sleeve sparger qualification process.

The purposes were:

(1) comparisons of the various sparger designs; (2) determination of advantages accrued by clad removal, and; (3) determination of effects of system changes proposed to mitigate thermal cycling.

The primary analysis concerned crack initiation resulting from high cycle fatigue.

Fatigue crack growth from an assumed initial 0.25-inch deep flaw also was analyzed.

The driving force for the low-cycle fatigue crack growth was assumed to be related to startups, shutdowns, and plant transients.

4.4.2 Crack Initiation Analysis The first step in the quantification process was the development, from the many records (operating reictor and previously-described 2T2 and Moss Landing tests) of temperature versus time, of what GE termed a thermal fatigue " load" spectrum.

Thermal cycles were counted during a 240 second period, utilizing the " ordered overall range" approach described in NEDE-21821-02.

The result was a " load" spectrum in which the ordinate (vertical) axis was the screening (minimum) amplitude expressed as a percentage of the peak-to peak amplitude, and the abscissa (horizontal) was the number of half-cycles with amplitude greater than the screening level.

Because each individual spectrum seemed to have similar frequency content, GE used a single envelope curve which included all of the spectra.

The " ordered overall range" approach provided results which were different than the 1.0 Hz frequency assumed at the peak amplitude in the original GE analyses.

As an example, the envelope spectrum amplitude at 1.0 Hz was only 20% of the peak-to peak amplitude, and the large amplitude thermal cycles (greater than 95% of available aT peak-to peak) occurred only about once every 100 seconds.

The next step n the GE analysis was the extension of ASME i

Section III fatigue S-N curves to cover the GE region of interest and to serve as design basis curves.

The resulting curves developed by GE extended Figures I-9.1 and I-9.2 of ASMF Section III beyond 108 cycles to 10 ' cycles.

Using the modified curves, a linear cumulative damage rule, and the load spectrum determined from the " ordered overall range" approach, GE derived the cumulative fatigue damage (usage factor) per 1000 hours0.0116 days <br />0.278 hours <br />0.00165 weeks <br />3.805e-4 months <br />.

The derivation included predictions at various peak

}9)b 13

'alternatino stresses and was accanolf sbed for both the nozzle clad-dina (stainless steel) and base raterial (low alloy carbon steel).

The reak alternatino stresses used in the derivation were obtained directly frcn reasured values of cyclic temperature difference as a percentage of the available temperature difference (T reactor -

T feedwater). The tercerature difference data care from the Poss Landing test results.

Finally, it was necessary to know the arount of time scent at various reactor terperature differentials.

Two tire-terperature flow raps vere aiven in NEDE-21P21 n2. The first was a " reference procedure" characteristic of the present operatina rode of BPPs and the second a " proposed alternate fsicl operatino procedure" con-tainino certain systen chances and procedural chances as described in Section 3.3.

The probability of crack initiation was evaluated in tems of the fatfave usace factor for several corbinations of scarcer desian, feedwater tenperature and coerating procedure.

These were surrarized in FEDE-21P21-02.

The conclusions drawn hy CE from comparison of tabulated fatiaue usace factors for the various corbinations of proposed solutions included the followino:

1.

Predicted crack initiation tires were in general agreement with crackina observations at plants which had the original loose-fit sparcer design. This provided assurance of the reasonableness of the analytical rethod.

2.

As anticipated, leakaae bycass flow is an extrerely irportant variable, as is the terperature difference T reactor - T feed-water. For exarple, in the case of the triple-sleeve sparcer installed after clad removal, GE predicted no crack initiation in 40 years with full power feedwater tercerature of 420 F and leakace held to a maxinur of 1 gor.

If, however, leakaae exceeds 1 con or the feedwater tercerature durino power opera-tions is as low as 340 F, crack initiation is predicted durina the plant's desian lifetime.

3.

An unciad nozzle with the welded sinnie sleeve sparcer desicned for zero leakace should onerate for 40 years without crack ini-tiation if operatino feedwater tenperature is 320 F.

The faticue usace factor on such a nozzle could be reduced fror 0.77 to 0.a6 by adoptina GE's proposed alternative coeratina procedures, a.d.3 Crack Growth Analysis As a first step in the fatigue crack arcwth evaluation, nozzle stresses were calculated using a finite elenent rodel. This allowed a systeratic evaluation of the effect of chances in the beat trans-fer coefficient produced by chances in sparger desicn. A turbine rell event, involvina a steo chance in feedwater tercerature 'ron 550 F to 100 F at 25% of rated feedwater ficw, was used to rodel a t'

1936 244

thermal transient. Maximum metal surface stresses developed typically in 2 to 4 minutes, but longer times were experienced when the heat transfer coefficient was low. Therefore, in order to determine maximum values, it was necessary to compute stress intensity factors as a function of time for each value of assumed crack depth. The initial flaw was assumed to be a semicircle 0.25-inch deep.

Based on recent data concerning reactor thermal operating history, GE made a refinement in the original model of the low frequency stress cycles. That model had defined a startup/ shutdown cycle as the combination of one pressure cycle (0 to 1050 psi and return to

0) and six thermal cycles in which feedwater temperature cycled between 100 F and 550 F.

The revised model comprises three scrams to low pressure hot standby and return to power for each startup/

shutdown cycle. A reactor lifetime is considered to include 130 startup/ shutdown cycles and 390 scram cycles. A scram is assumed to include a on-off feedwater flow cycles during which feedwater temperature aries from 100 F to 300 F and 12 such cycles with temperature variation from 100 F to 430 F.

The new model for crack growth was compared with known data from operating reactors. Specifically, the growth of a 0.25 inch crack was compared with the crack growth observations at Pilgrim, Nine Mile Point and a foreign reactor when each utilized the original loosefit sparger. A best fit curve was used for the relationship of da/dN (crack growth rate per cycle) as a function of effective stress intensity factor. Good agreement was obtained with predic-2 tions based on a heat transfer coefficient of 2000 BTU /HR-ft - F for the original loose-fit sparger.

The predictions of fatigue crack growth were then used in the evaluation of sparger designs and the determination of the need for a low-flow feedwater controller. Results are described in Section 3.3, above.

4.4.4 Staff Evaluation and Conclusions The staff has reviewed the GE analyses discussed above and has concluded that the methods used and the results are acceptable. We have further concluded that the results of the crack initiation and growth analyses may be applied in establishing generic inservice inspection requirements.

1936 345

,s

5.0 OTHER SPAPGEP DESIGNS The GE report briefly discusses three alternative scarcer designs. The first of these, the welded themal sleeve, is in use at three operating reactors (Duane Arnold, Brunswick Unit No.1, and Hatch Unit Fo. 2) and has been installed in two reactors (Zirrer & WPPSS-2) under inital licensing review.

The staff generally acrees with the GE assessrent that a confiauration with the therral sleeve welded to the nozzle safe end provides sone assurance of protection acainst crack initation if feedwater temperature during operation is at least 420 F.

However, as GE noted in the report, there are several drawbacks to this particular design. flot noted is the lack of suitable inspectability of the thermal sleeve-to-nozzle weld.

The staff's concern is that weld failure after several years could result in substantial bore cracking prior to the appearance of cracking on the accessible areas of the blend radius. The staff is still devoting effort to the resolution of the inservice inspection issue, as noted in the introduction and in Section 6.0 of this SEP. However, dye penetrant ins-pections of accessible nozzle areas (an inspection techniaue acceptable to the NRC staff) perfomed already, at Duane Arnold and Prunswick Unit Fo.1, demonstrated the efficacy of the weldcd design early in the plant life in that no indications of cracking were found.

In addition, a lin-ited visual inspection of the sleeve-to-nozzle weld was perforred at Duane Arnold, where sparger desian allowed such insoection. The weld was reported to be intact. Althouch these early inspection results indicate satisfactory weld inteority, the inspection program will still reouire

/

examinations to assure continued integrity later in plant life.

The second desian cited by General Electric is the sinole piston ring design, which is simply an augrentation of the interference fit sleeve desian and would serve similarly to the interference fit as an interim "fix" until its efficacy has been deronstrated by field experience. GE acknowledoes this in their statement that the "... interference fit will not be le.t during the limited design life of this component." The only operating plants with an installed sparger/ sleeve similar to this are the Ponticello Fuclear Generating Plant and Browns Ferry Unit 1.

The staff will continue to review nozzle inspections at these plants in order to determine the efficacy of this particular design.

The third design discussed by GE is the interference fit themal sleeve design, which was the first counter-reasure to the cracking resultina fron loose-fit spargers and is the interin solution rentioned herein.

Experience has shown that the interference fit can prevent crack initic-tion but its loncevity is linited as it relaxes with tir e.

Therefore, although it is acceptable on an interim basis, the staff does not recard it as a long-tem replacerent without relatively frecuent inservice inspection.

Not rentioned in the report, since they are beyond GE's responsibility, are other approved and currently operatino desions at Nine Mile Point and Oyster Creek. The staff, while acceotina the GE triple-sleeve, double piston rina desion as an effective long-tem solution, will review other prooosed designs %r acceptability.

)9 b

16

6.0 ULTRA 50flIC TESTIflG Af!D RECOMMENDED IflSPECTION 6.1 Introduction Chapter 6 of NEDE-21821-02, entitled " Ultrasonic Testing," describes the General Electric UT procedure and results of feedwater nozzle exanina-tions perfonred by GE.

Chapter 7, entitled "Pecorrended Inspections,"

proposes an inservice inspection (ISI) progran for plants with cither the triple sleeve sparger or the welded sparcer. The proposed ISI proaram differs from the recorrendations of the f;PC's interim cuidance document

'1UPEG-0312 by proposing to elininate licuid penetrant TPT) exaninations and to substitute UT examinations at less frecuent intervals.

6.2 Staff Evaluation and Conclusions The selection of conservative flDE nethods and appropriate inspection intervals is dependent upon the nature of the flaws under investication.

Themal fatiaue cracks detected in the feedwater nozzle blend radius and the bore region generally have been as deep as 1/2" t, 3/a" total depth (includina claddina) and up to twelve inches in length.

Sore have been deeper.

The effectiveness of UT inspection is adversely affected by the conplex gecretry, relatively long exanination retal paths, and cladding interference encountered in feedwater nozzle inspections.

Currently, the only acceptable rethod for conclusively detecting, locating and characterizina existing flaws is PT of the inner surface and reroval of cracks by local grinding. However, PT inspections and renoval of cracks by grinding have resulted in significant personnel radiation exposure and plant shutdown time. An objective of current NDE technology progrars is to develop a reliable and effective UT procedure that can be perforced from the vessel exterior surface.

Section XI of the ASPE Code recuires ceriodic voluretric exarination of the feedwater nozzle region. However, a specific recorrended procedure has not yet been published. To irplerent t!UPEG-0312 as recuired by the staff, licensees are performing auamented ISI programs at designated intervals of operation, including suoplemental PT and UT inspections during scheduled outages. There are currently many different UT proce-dures in use. Evaluation of olant specific practices has t'een necessary because the differences in nozzle gecretries corbined with certain inspection variables can influence the exarination results. To date, no specific UT technicue is acceptable to the flPC as a sole rethod of characterizing fatigue cracks. However, there is an extensive effort underway to develop such a technicue.

As a result of there being no repeatable, reliable UT technicue which has reauisite sensitivity, the staff does not at this tire accept UT exanina-tion as the sole reans of assurina nozzle intecrity. Ve are concerned, of course, that continued PT exaninations result in significant radiation exposure for licensee and inspection ccrpany encloyees. Therefore, we will adopt a realistic procram which provides credit for licensee actions to minimize the possibility of crack initiation and growth. Such actions 17 1 O } [)

47

will extend the tire between PT inspections. The program is discussed briefly in Section 6.4 and will be published as part of the forthcoming fl0 REG document.

The staff has concluded that Chaoters 6 and 7 of f!EDE-21821-02 are unacceptable in their entirety and may not be cited as a reference in licensing actions by either licensees or applicants.

For such licensino actions, the staff will provide specific guidance, usina fiUPEG-0312 until the forthcoming ffUREG is issued.

6.3 Surrary of Continuino Major Activities Inspection companies, EPRI, and an ASPE Code workina aroup are continuing major programs to develop a reliable and effective UT procedure for the nozzle inner radius examination. The primary objective has been directed toward the detection and location of flaws.

Current exanination tech-ninues are not sufficiently developed to characterize the dirension or shape of indications with an acceptable degree of accuracy. The only interpretation presently a;ailable is that the minirun threshhold depth of crack has been exceeded.

The relatively lona ultrasonic wave retal paths, inherent beam spread, and cladding rake cuantification of the indication dimensions extrerely difficult and could reouire advancerents beyond the state-of-the-art technology. The difficulty will be eased somewhat because cladding will be renoved as part of the installation of the trinle-sleeve sparger.

A comprehensive correlation of UT indications on actual thernal fatigue cracks compared with PT verification could demonstrate that all cracks that could affect the structural integrity of the nozzle are detectable in any location. The distribution of actual flaws compared to recorded indications and the mininum detectable flaw depth for a specific UT pro-cedure also would be established. This type of data may becore available during planned clad removal pro.iects.

In addition, a full size nozzle rockup with laboratory-induced thernal fatigue cracks recently has been fabricated under a project sponsored by the Electric Power F:: search Institute (EPPI). A systematic investigation of the various UT proce-dures on such a mockup could identify the rost effective procedure in tems of detectability, reproducibility, and efficiency to minimize personnel radiation exposure. The results of a sirilar survey, usino another nozzle rockup with nachined notches, are beino evaluated by EPRI.

The availability of these full size test specinens is irportant for procedure and personnel cualification and training of operators under sirulated plant environments. The rost reliable plant inspection procedures use a nozzle rockup as the calibration block.

The ASPE Code Section XI established a Task Group to define reauf rements for a UT procedure applicable to inner radius exaninations. The availa-bility of a full size rockup and additional suororting data on the UT response from actual characterized fatigue cracks should provide the basis for revisions to the ASFE Code. The staff will follow the UT development programs directed toward demonstrating an effective UT procedure and will issue appropriate further guidance beyond that to be included in the flVPEG document which ccrpletes A-10 study.

1936 '48

6.4 Pecorrended Insoections The staff will continue to evaluate plant-soecific leak test and nondestructive exanination recuirenents on a case-by-case basis.

PT of the inner surface and, if necessary, renoval of cracks by orindino, is the only demonstrated method to conclusively detect, locate and charac-terize flaws. The installation of an acceptable feedwater sparcer and modifications such as clad reroval and systen chances should prevent crackino or decrease the rate of crack propacation. Therefore, the current insoection interval between PT exanination will be increased for plants which have incorporated acceptable modifications.

The staf# will provide further guidance on acceptable inservice leak test and flDE rethods and inspection intervals.

/

o 6

19

\\

7.0 It'PLE!'ENTATION Chapter P of NEDE-21821-02 is a brief discussion of the GE reconrendations regarding the implerentation of physical and operational rodifications.

For plants still under construction, GE recorrends that all changes be impler.ented prior to startup. However, where completion of physical changes would delay startup, GE recorrends allowing operation for the first fuel cycle with operational changes only.

Physical changes would be made dur'nq the first refueling outage.

For operating plants, GE recomrends implementation of changes as soon as it is convenient, preferably during a long outage.

For plants with welded thermal sleeves, GE recommends installation of the system chanaes only.

For those plants under construction where systen changes would delay the startup, GE recommends delayina the rodifications until an apprceriate outage.

Peaardina the implerentation of physical nanges for plants still under construct.on, we will require such changes to be accomplished prior to initial reactor operation. Radiation exposure for nodification personnel thus could be avoided.

With regard to operatino plants, the staff has concluded that applicable hardware and syster chances should be implerented as soon as practicable.

Based on the experience of plants which have already reroved nozzle cladding and installed advanced design spargers, this work ray be deferred until the first lenathy outaae. Ve raintain our position as stated in the interim report, NUREG-0312 (July 1977), that, prior to corpletion of all physical rodifications, credit will be aiven for other actions taken to prevent crack initiation and growth.

Uhile the staff concurs that operating plants with welded themal sleeves (3ee Section 5.0) need only implement systen changes at this tire, we will reserve comment on the ability of the weld and themal sleeve /

sparger to remain intact throughout the design plant life, pending confimatory inspections.

1936 $50 20

8.0 SAFETY CONSIDERATIONS 8.1 Summary of Chapter 9, NEDE-21821-02 The approach taken by the General Electric Company in Chapter 9 to evaluate the safety significance o' nozzle cracks was to determine the margin of safety (against rupture of the pressure vessel) using fracture mechanics calculations. An assumption is made that the maximum flaw size will be limited to that permitted by Section XI of the ASME Code. By those rules, for example, nozzle corner blend radius cracks could pene-trate base metal to a depth of 0.95 inches in a typical case before repair would be required. Justification for this assumption is presented in the report.

Fracture mechanics calculations are presented for a nozzle corner crack, a nozzle bore crack and (just in case a very large crack remains undetected) a hypothetical through-wall crack extendirl both above and below the nozzle.

Experimental data are cited in m.

i to justify the reliance on linear elastic fracture mechanics (LEFM) and in particular the extension of its use to elastic-plastic conditions.

The salient features of the GE analysis include:

1.

Stresses considered to act on the nozzle corner and bore were pressure stress and thermal stress. Residual stress due to the cladding was included but other residual stresses were considered to be small. Stresses related to pipe reaction forces were omitted without comment.

2.

The first step in the safety analysis used a straight LEFM approach in which K r " fracture toughness" was compared to Kg(pressure).

IR An upper shelf value of 200 ksi M was used for K for 3.

temperatures 180Fgreaterthanthereferencetempbhature,RT Itwa$DT' which was assumed to be +40 F for the nozzle material.

stated, without supporting evidence, that 200 <si E represents a

" minimum upper shelf toughness for reactor vessel grade low alloy s teel."

4.

A thennal-hydraulic analysis showed that temperatures at the tip of postulated cracks exceeded 220 F under all conditions of normal operation when pressure and thermal stresses were high.

5.

Flaw size assumed in the LEFM analysis was 0.95 in for the nozzle corner crack and 0.71 in, for the nozzle bore crack. These~ values are 10 percent of the nozzle wall thickness (0.lt) in the direction of the advancing cracks.

6.

To obtain a solution for K, (pressure), GE used a three-dimensional analysis published by GilmAn and Rashid, supported by the solution for an edge crack in a circular hole and by photoelastic test results published by Smith.

1936 351 21

7.

Safety margins for pressure act1.ig alone were calculated in terms of:

K

= 200 ksi M IR Ky (pressure)

To be conservative, the entira pressure stress, including the peak stress caused by the geometric discontinuity effects of tha nozzle The value of K opening, was considered to be primary stress.

(pressure) was 62 ksi M for the "Section XI Flaw," hence thd safety marg:n on internal pressure was approximately 3.2.

8.

The second step in the safety analysis was to consider the margin of safety when, in addition to the pressure stresses, thermal stresses caused by a turbine roll event were included.

A step decrease from 550 F to 100 F of the water flowing in the annulue between the sparger sleeve and nozzle bore was stated to be ~.ne most severe themal transient for normal or upset conditio s.

Calculations were presented for two times in the transient - 90 seconds and 30 minutes.

From the report, the ratio of K 200 ksi E ) to K (thermal) is about 1.5 to 1.6, but GE does nd$ u(=se this ratio in tbeir safety analysis.

9.

To express the margin of safety for cases of combined pressure and thermal stress, GE introduced a new term, " Fracture Mechanics Margin," A.

A=

IR K7p + 0.5 KIT Its use is justified by an analysis based on a report (0RNL-TM-5090) regarding experimental data obtained on the intermediate test ves-sels in the Heavy Section Steel Technology (HSST) program at Oak Ridge National Laboratory. Two of these vessels had nozzle corner cracks. The facts that thermal stresses are secondary and that through-thickness yielding occurs prior to fracture at the tempera-tures of interest are cited to show the conservatism of this approach.

Also, a precedent from Appendix G of the ASME Code is cited, i.e.,

the requirement that 2Kyp + KIT KIR The values of "A" are somewhat greater than 2.0 for the cases cited.

10.

Finally, GE utilized an LEFM leak-before-break analysis in which very large cracks, one at the top and ore at the bottom of the nozzle, are postulated te have escaped detection and to have grown through the wall to form a through-wall crack with the nozzle opening at its midlength. GE calculated that K would not be exceeded under nomal operating pressure until he hypothetical through-wall crack extending above and below the nozzle reached a critical length of 29 inches.

1936 352 22

11.

GE cited experimental evidence obtained by Japanese investigators in what are called the JAERI* pressure vessel tests.

Nozzle corner fatigue crack growth was measured at a cyclic pressure stress (hoop) of 0 to 29 ksi at 75 F.

The cracks grew through the wall by fatigue causing a leak to occur instead of a fracture.

The fatigue crack growth rates also were analyzed by GE for evidence to support the fatigue evaluation of the feedwater nozzles, especially the predict-ability of crack growth for deep cracks.

There was remarxable agreement between predictions and observations.

The conclusion reached by GE regarding safety considerations is that the recommended solutions to reduce cracking and to improve inspec-tion methods will result in a significant reduction in the maximum expected flaw size in an operating reactor.

GE believes this reduc-tion to te so significant that, even with the presence of the maximum expected flaw in the nozzle of an operating reactor, the margin against failure c ee steel will be the same as that inherent in the design, by AtME Code, of an unflawed reactor vessel.

8.2 Staff Evaluation and Conclusions The staff agrees with the overall conclusions reached by GE and noted in 8.1, regarding safety considerations. We particularly note the mitiga-ting circumstance, shown by analyses and testing of feedwater nozzles, that whenever pressure stresses and tnermal stresses are high, the temperature of the metal in the path of an existing crack is generally high enough to provide optimum crack propagation resistance.

The exception is during a vessel hydrostatic test, for which special limits are required.

Further assurance of adequate safety margins has been provided by tests of several six inch thick vessels, as part of the HSST Program at Oak Ridge National Laboratory.

The models contained nozzles similar in design to the feedwater nozzles of boiling water reactors.

The test results showed that the vessels exhibited greater resistance to crack propagation than had been predicted by conservative fracture mechanics analyses.

Although the staff has concluded that each step of the GE analysis is acceptable for the purposes of the generic study, we have comments regarding certain steps:

Point 3 of the GE analvsis - The assumed value of 200 ksi,/In for K of thenozzlesteelatuppershelftemperaturesisinsufficientlysubsdnti-ated by valid data.

The highest measured value considered valid, according to A erican Society for the Testing of Materials (ASTM)

Standard E-399, is 148.55 ksi S.

The result was obtained from a tweive-inch thick specimen.

^ Japan Atomic Energy Research Institute (Refer to S. Miyazono et al.,

" Fatigue Behavior of Nozzles of Light Water Reactor Pressure Vessel Model", Third International Conference on Pressura Vessel Technology, ASME) 1936 353 23

Experiments on nozzle steel have shown that its resistance to fracture, measured in tenns of the maximum load ar,d corresponding displacement, increases sharply with temperature above that at which the valid result noted above was measured. Because the reactor vessel is operated at temperatures higher than that at which the ASTM value was obtained, we eelieve GE's use of 200 ksi /Ili in their conventional fracture mechanics analysis to be satisfactory.

Point 5 of the GE analysis - Maximum flaw sizes of 0.lt it is nozzle wall thickness) were assumed G the GE analysis and justified simply on the basis that 0.lt represents a limit in Section XI of the ASME Code. The staff accepts the 0.lt limit, but bases its acceptance on knowledge o#

BWR operating experience with regard to the nozzle problem. Substantial data have been gathered regarding the correlation of startup/ shutdown cycles with cracking severity.

In addition, there is extensive service experience with the interference fit sparger serving as an interim solu-tion prior to clad removal and installation of the triple sleeve sparger.

From this body of knowledge, the staff has conservatively predicted the time for an assumed 0.25 inch flaw to grow to 0.lt and has predicated its recommendations for inservice inspection on this prediction. The staff has gained additional confidence from the GE analyses and testing related to the development of tra triple sleeve sparger which will substantially limit crack initiation and powth. From these factors, the staff has concluded that the assumed 0.lt maximum flaw size is acceptable.

Point 9 of the GE analysis - The " fracture mechanics margin, A,"

introduced for this analysis, is a new concapt. The NRC staff accepts its use for the purpose of treatment of thermal stresses at the feedwater nozzle. GE justifies use of the concept by referring to tests of flawed six-inch thick vessels in the HSST program. The vessels, prior to burst, undersent greater through-thickness yielding than was predicted by con-servative fracture mechanics analyses. The staff 5elieves the empirical factor A is useful and gains confidence from the knowledge that in the case of feedwater nozzles, cracks generally advance into a lower-stress region where there is high temperature and therefore greater resistance to crack propagation.

As stated above, the staff accepts the use of the GE factor A for this analysis. However, the staff recognizes the empirical nature of this factor and may require additional justification for its use in the future, when more test information will be available.

Point 10 of the GE analysis - The staff accepts the conclusion of GE's leak-before-break analysis only when it is assumed that the flaw is a through-wall fatigue crack subjected to normal operating pressure.

Leak-before-break cannot be assured, by the GE analysis, for a crack which propagates through the wall under higher than normal orerating pressure. Nor is leak-before-break assured when the metal temperature in the path of the crack is lower than that (upper-shelf) whien yields optimum resistance to crack propagation.

1936 354 24

9.0 CONCLUSION

S Based on our review and evaluation of the infomation pr)vided by GE, the staff finds the GE topical report NEDE-21821-02, with specific exception of Chapters 6 and 7, acceptable as a reference in R1!R licenstaa actions for both the rodification of licensed plants and the licensing of future plants.

Inservice insnection rethods and freouency will be addressed by the NPC in a separate docurent.

Separate correspondence ray be recuired during the course of licensing actions to obtain plant-specific inforration from the licensee or applicant.

We have concluded that the prcoosed GE sparaer modification, when coupled with the removal of the stainless steel claddina and feedwater system chances when necessary, provides a substantial improvement over previous CE designs. A reactor vessel thus codified will be able to operate for an extended period of tire between surface examinations.

Extending the interval between surface examinations will substantially reduce the radi-ation exposure of plant staff and contractor personnel. Powever, we conclude that this specific sparger configuration is not the only accept-able design. We have approved the installation and use of different con-figurations by other designers at two operatina plants and have also approved welded configurations designed by GE. Any approved design will require some in-service verification of its continued acceptability throuah inspections, ior facilities now under review for an operating license, the combination of the proposed GE sparger modification, an unclad nozzle and appropriate system changes is an acceptable design. Subjects requiring further con-sideration and review are inservice inspection intervals and the use of leak testing and certain NDE techniques, particularly ultrasonic testing.

While the NRC staff recognizes that there have been improvements in NDE technology in the past few years, we have not accepted fully the industry evaluations regarding improved flaw detection, because the effectiveness and reliability of nozzle-related UT procedures have not yet been demon-strated adequately under field conditions for real cracks. The forth-coming NUREG report (to be published at the completion of the ongoing NRC generic program related to BWR nozzle cracking problems) will define acceptable 'nterim N9E methods and inspection intervals. The staff recognizes and encourages industry-sponsored.NDE programs to. demonstrate the reliability of UT techniques and will give credit for favorable results.

3 1936 355 25