ML20082T557
| ML20082T557 | |
| Person / Time | |
|---|---|
| Site: | Millstone  |
| Issue date: | 12/31/1983 |
| From: | WESTINGHOUSE ELECTRIC COMPANY, DIV OF CBS CORP. |
| To: | |
| Shared Package | |
| ML19274C264 | List: |
| References | |
| NUDOCS 8312160055 | |
| Download: ML20082T557 (6) | |
Text
,
f Docket No. 50-336 Attachment 1 Millstone Nuclear Power Station, Unit No. 2 Evaluation of Operation with Broken Fuel Assembly Holddown Springs _
y Non-Proprietary 1 >
- December,' 1983 -
s
,?
i PDR_l ,1' l',
L
WESTINGHOUSE PROPRIETARY CLASS 3 i
Northeast Nuclear Energy Company in conjmetion with our fuel vendor for Millstone Unit No. 2 has evaluated plant operation utilizing fuel assemblies with broken talddown springs. It has been concluded that operations with broken holddown springs is acceptable and does not result in any undue risk to the public health and safety nor degrade plant . safety. The basis for this conclusion is provided in the following information which supports NNECO's cwrent plans to op-rate with nine f uel assemblies, each with one broken holddown spring.
Break Characterization All the spring fractures were single breaks with the broken spring sections f ully restrained by the guide post. The spring fracture occurred approximately one complete turn from the spring end with the exception of one spring on assembly F59 which broke near the center coil. Examination of the spring fractve surf aces indicated a typical f atigue f ailure due to the combined mean and alternating torsional stresses. The potential causes for the spring f atigue f ailures and the consequences of operating with broken springs are addressed in the following sections.
Cause of Failure The fuel assembly holddown spring f ailur es were caused by excessive vibratory motion durmg reactor operation. The alternating stresses in the holddown spring due to the oscillatory motion combined with the relatively high mean stress resulted in a classical f atigue type fracture. A visual examination of the f ailed springs showed that the fracture surface was along a plane approximately 45 degrees relative to the longitudinal axis of the spring coils. The f ailtre plane corresponds to the maximum tensile plane which is characteristic of helical coil springs or cylindrical rods subjected to pure torsional stress.
Although the source of the spring excitation has not been positively identified, two potential mechanisms exist; flow induced vibration and base motion. Flow induced vibration which has been identified as the catse of helical coil spring f ailures in other reactors is considered to be the prime cause of the f ailtres.
Since the f ailures correlate strongly with core locations adjacent to the shroud, the crossflow in that regions in conjmetion with axial flow is suspected to be the f excitation source. Both mechanical and flow tests were conducted to determine l
- the excitation mechanism and to verif y the spring dynamic characteristics.
A hydraulic flow test was conducted using a single holddown spring and guide post to investigate the fluid and mechanical interactions. The results obtained from this investigation confirmed significant spring vibration at reactor flow rates. The tests showed that axial flows excited simultaneous lateral and axial motion. However, the presence of cross flow was found to cause spring excitation at lower axial flow velocities and resulted in significantly higher ring vibrational amplitudes. The frequency response of the spring under the
, atmulated reactor flow conditions was approximately ( )b,c. Mechanical vibrational tests were also conducted on a top nozzle assembly containing four (4) holddown springs. Data obtained from these tests indicated that the fundamental frequency of the spring was ( )b,c and that the second mode was approximately ( )b,c. The second mode, which was observed to be the one that was excited during the flow tests,is a combination of simultaneous -
L
34 e , .-4)l . %
- 3 l
~
y WESTINGHOUSE PROPRIETARY CLASS 3 i
/
- <y
' lateral and axial motion. ( '
- N Q' }.
)b,c. . These results are consistent (with the fact that the spring ,
f ailures occurred in fuel assembl!es that were adiacent to the core shroud.
'% /, ,
i Post irradiation visual examinatiod of the guide post wear was also extensively
, conducted to provide additional information relative to the spring failure
- mechanism. A comparison of the guide post wear for Regions F and G did not
- reveal any substantial difference. This suggests that protracted fuel residence in the core interior does not significantly increase the probability of spring failure.
It was also observed from mechanical tests that two of the four springs in the nozzle assembly did not readily vibrate. The reduction in the spring dynamic response was attributed to the damping cause by spring contact with the guide post. This is consistent with the data obtained from fuel assembly visual examinations which showed little or no wear on the guide posts with failed springs while guide posts with more extensive wear had no broken springs.
Fretting Wear An analysis using the Archard wear equation was performed to determine the amount of fretting wear on the top nozzle posts caused by spring vibration. The projected maximum localized wear depth for three cycles of reactor operation is approximately (
~
Ja,c inches. A visual estimate of the maximum wear depth obtained for tivo cycles of reactor operation was ( )a,c inches which is consistent with the projected wear. Since the wall thickness of the -post is
( )a,c inches, the fuel assembly post will retain adequate strength af ter three cycles of operation to meet the normal and accident design conditions.
Plant Operation with Broken Springs The purpose of fuel assembly holddown -orings is to provide an axial force that precludes fuel assembly lif t-off ' rom ;he lower core plate under . normal operating conditions. An evaluation w0s performed to determine the functional and safety related consequences of operating Millstone Unit No. 2 with broken fuel assembly holddown springs. Post irradiation examination at the end of -
Cycle 5 indicated that all but one of the broken springs'~were fracttred at the beginning of the first active turn adjacent to a spring end. _Since'the number of active turns are only slightly ' decreased by such a break, the spring rate-and forces during normal operation are not significantly changed. One of the failed .
springs, located on assembly F59, was ' fractured at a central coil location relative to the spring ends.' A calculation of the spring forces for this type of L
-fracture also showed no significant loss in~ spring load if _ the spring' dces(not intermesh or the broken parts rotate relative to each other. Since the normal-operating preload 'in a spring broken at .the center 'is' relatively _ high t
( )a,c, the resulting friction between the contacting coils prevents relative rotation of the two broken ' spring ; sections.- This. hypothesis was confirmed by post irradiation examination of fuel assembly F59. _ This assembly experienced -two cycles of operation and no discernible relative : angular.
movement of the spring ends at the broken section was identified.: The springs in -
assembly F59 have been replaced with newly designed springs (
Reference:
- 2) for
- Cycle 6 operation.
WESTINGHOUSE PROPRIETARY CLASS 3 The holddown spring rate and forces were analytically derived for the fuel assemblies with broken springs and were subsequently verified by mechanical t ests. Functional test; of irradiated fuel assemblies were performed whereby the movable cross bar (flower) of the top nozzle was deflected approximately
( )b,c inch to assure acceptable and unimpeded operation. A total of ( )b,c assemblies were tested of which 10 assemblies contained broken springs. A comparison of the spring rates and preloads for the fuel assemblies with and without broken springs revealed no significant dif ference.
The maximum holddown force required to preclude fuel assembly lif t-off mder normal operating condition is ( Ja,c. Based on the holddown force requirements at normal operating conditions and the test data on broken springs, all four holddown springs for any given fuel assembly could be broken without any loss of imction. The holddown force requirement for startup (four reactor coolant pumps operating at 5000F) is ( Ja,c lbs. and can also be met with more tnan one broken spring. For the case of pump overspeed, the calculated design net lif t load is ( Ja,c lbs. This value is slightly greater than the nominal holddown force of ( Ja,c lbs for a normal spring pack (no broken springs).
Since the potential for fuel assembly lif t off is normally present during a pump overspeed condition, the impact of broken springs is negligible. The analysis of the broken spring and the supporting mechanical test data show conclusively that the observed spring failures do not significantly change the fuel assembly holddown force or impair the movement of the flower.
The effects of broken springs with respect to seismic and LOCA accident conditions were also investigated. Because broken holddown springs do not alter the lateral dynamic properties of the fuel assembly, the reactor core response during a seismic accident is unaffected. For the LOCA accident, the vertical response of the reactor core would change only if the spring preloads decreased appreciably. The small dif ferences in the calculated holddown forces between assemblies with and without failed springs would not significantly change the impact force at the top or bottom nozzles. Consequently, a small increase in impact force would not violate any design criteria or limits since the structural design margin at the nozzle locations is substantial.
Probability of Multiple Fractures An evaluation was conducted to estimate the probability of the occurrence of cdditional fractures in once broken springs. Both analytical and experimental studies were performed to obtain the dynamic characteristics and response of a typical f ailed spring. An examination of the failed irradiated springs indicated that the typical break location was (
Ja,c. Fatigue type fractures in unirradiated production springs were obtained by mechanically exciting a top nozzle assembly. The locations and orientations of the spring fractures were consistent with those observed in the f ailed irradiated springs and all the springs contained only one f racture.
Vibrational tests were performed on two of the fatigued f ailed springs in order to obtain the dynamic properties of a broken spring. The results of the testing showed that the fundamental frequency of the failed spring was i
WESTINGHOUSE PROPRIETARY CLASS 3
(
)b,c. This effect would be significantly beneficial in detuning the spring driving force of certain types of fluid excitation mechanisms such as vortex shedding. Additional vibrational tests were also performed in which unbroken and f ailed springs were sinusoidally excited at their respective '
fundamental frequencies and at a constant acceleration level. The maximum displacement amplitude ) ( b,c than that of the f ailed spring. It should be noted that the failed springs ( )b,c when randornly inserted onto the nozzle post. Extensive adjustments were required to properly center the spring
( )b,c. Under normal reactor operation, it is expected that a broken spring would be shif ted against the post and thus result in higher resistance to vibration. Based on the results of the vibrational tests it is ,
concluded that the ( )b,c of the failed spring would
( )b,c and the potential for a subsequent fracture.
The probability of multiple fractures in once broken springs will be further reduced by locating those assemblies with f ailed springs in core interior positions. Results f rom hydraulic testing indicated (
)b,c. A fatigue analysis of the spring stresses, showed that
( ya,c was required to mduce spring f ailure. Based on mechanical and hydraulic tests it is concluded that spring multiple f ractures have a relatively low probability of occurrence (less -
than one multiple failure during Cycle 6 operation).
Loose Parts A postulated break in a previously broken spring can occur at either end of the spring. A break which occurs at the opposite end of the spring from the-first break will result in broken coils which will be captured by the nozzle extension posts.
~
The minimum size of a broken spring coil that will be retained on the nozzle extension post was calculated to be a ring segment of ( . )a,c degrees. From mechanical tests and stress it was found that
( )b,c is analysis required of to aattain broken spring, he or develop t maximum stress value. Examination of failed springs indicated that the spring force at the broken end is applied at two discrete points, namely the tip of the broken end and ( a,c from that point. Since all the observed spring failures -
occurred a minimu)m of ( ~
)a,c, it is highly probable that the size of the second fracture piece would be 3Mficient to be retained by the post.
An evaluation of loose' spring parts has been completed, in the unlikely event that a section of spring became detached. ,
l The detached section of spring could be carried upward into'the upper internals of the reactor vessel by the reactor coolant flow. The detached spring section :
would most likely migrate to and come to rest in a low velocity plenum of the primary system such .as the. steam generator or.. areas in the upper internals package. A detached section of . spring could enter 'the Control Element
.tp m
-s , s - - ,~s + ~ , < w r ss e r-
WESTINGHOUSE PROPRIETARY CLASS 3 Assembly (CEA) shroud assembly region. It can theoretically become lodged +
- - between a CEA finger or hub and the CEA shroud. In this unlikely scenario, downward movement (insertion) of the CEA would tend to dislodge or push the
- spring fragment back out. The probability of a spring fragment catsing a CEA to l be stuck in position is low. Additionally, periodic CEA exercise surveillances
, would identify such a condition. Furthermore, only three of the nine assemblies i with broken springs will reside mder CEA's during Cycle 6.
Jamming of a control rod in the fuel assembly could not occur since the annular clearance between a CEA rodlet and the. fuel assembly guide thimble is much
- sn: aller than the minimum dimension of the coil fragment (ie, 0.087 vs. 0.218 inches respectively). Entry of a spring fragment into a guide . thimble or an .
Instrument tube in a non-CEA location is prevented by the c!cse fit of the nozzle extension with the blind holes of the upper core alignment plate. Since the size and mass of the fragment is small, damage due to impacting by the fragment is of negligible concern.
Conclusion The most probable catse of the broken spring was_ flow induced vibration
- resulting in a f atigue failtre. Test and operational data indicated that
( )b,c was required to produce failures. The top
- nozzle holddown performance is not altered or degraded by the presence of
, broken springs. Additional spring failures are not expected during subsequent reactor operation.
The potential for loose parts as a result of ' additional spring fratures -was evaluated. Multiple failures of the same spring are unlikely due to the change in spring characteristics following the initial break. In the unlikely event .that a -
loose part is generated, no impact to plant operation is expected.
l I
f i
4
( -
- a u-h 1 i
og.
. p .4 , v
- y, , y i>- } e5 N}- t T e w f 9' 7