ML20114A520

Rev 1 to Final Deficiency Rept DER 83-49 Re Broken Impeller Blades & Diffuser Bolts Loose in C-E Reactor Coolant Pumps. Deficiency Not Reportable Per 10CFR50.55(e)

ML20114A520
Person / Time
Site:	Palo Verde
Issue date:	01/09/1985
From:	Van Brunt E ARIZONA PUBLIC SERVICE CO. (FORMERLY ARIZONA NUCLEAR
To:	Kirsch D NRC OFFICE OF INSPECTION & ENFORCEMENT (IE REGION V)
References
ANPP-31659-TDS, DER-83-49, NUDOCS 8501280234
Download: ML20114A520 (8)

Text

g.m Arizona Nuclear Power Project P.O. BOX $2034

• PHOEMX, ARIZONA 85072-2034 ,

} {) ]O b January 9,'tB9 f ANPP-31659 ;1N/TR3 7 .. ,?1'^

U. S. Nuclear Regulatory Commission Region V 1450 Maria Lane - Suite 210 Walnut Creek, CA 94596-5368 Attention: Mr. D. F. Kirsch, Acting Director Division of Reactor Safety and Projects

Subject:

Final Report Revision 1 - DER 83-49 A 50.55(e) Reportable Condition Relating to Broken Impeller Blades and Diffuser Bolts Broken / Loose la The Reactor Coolant Pumps.

File: 85-019-026; D.4.33.2

Reference:

A) Telephone Conversation between P. Narbut and R. Tucker on July 15, 1983 B) ANPP-Telephone conversation between T. Young and R. Tucker on July 18, 1983 (Interim Report)

C) ANPP-ANPP-27593, dated August 19,1983 (Interim Report)

D) ANPP-28734, dated January 27,1984 (DER 83-50 Interim Report, Revision 1)

E) ANPP-28733, dated January 27, 1984 (Interim Report, Revision 1)

F) ANPP-28313, dated November 29, 1983 (Time Extension)

G) ANPP-29713, dated June 11, 1984 (Time Extension)

H) Telephone conversation between P. Narbut and T. Bradish on August 14, 1984)

I) ANPP-30188, dated August 13, 1984 (Tine Extension)

J) ANPP-30520, dated September 14, 1984 (Final Report)

Dear Sir:

Attached is Revision 1 of our final written report of the deficiency referenced above, which has been determined to be Not Reportable under the requirements of 10CFR50.55(e).This revision clarifies the Condition Description and provides updated information on the Analysis of Safety Implications and Corrective Action which has been taken.

Very truly rs

, CQL r\t O E. E. Van Brunt, Jr.

h h4 S 0 09 528 APS Vice President Nuclear Production ANPP Project Director j EEVB/TRB/pik j Attachment  ;

cc: See Page Two I 3j l

> 1

.:r ,

k 1

Mr. D. F. Kirsch DER 83-49

'Page Two cc: . Richard DeYoung, Director' Office of Inspection and Enforcement U. S. Nuclear Regulatory Commission Washington, D. C. 20555 T. G. Woods, Jr.

D. B. Karner

.W. E. Ide D. B. Fasnacht A. C. Rogers L. A. Souza D. E. Fowler T. D. Shriver C. N. Ruano J. D. Houchen J. R. Bynum J. M. Allen A. C. Gehr W. J. Stubblefield W. G. Bingham R. L. Patterson R. W. Welcher H. D. Foster D. R. Hawkinson L. E. Vorderbrueggen R. P. Zimmereen M.~ Matt T. J. Bloom D. N. Stover L. Clyde D. Canady J. E. Kirby.

Records Center Institute of Nuclear Power Operations 1100 Circle 75 Parkway, Suite 1500 Atlanta, GA 30339 Mr. E. Licitra U. S. Nuclear Regulatory Commission Phillips Building 7920 Norfolf Avenue Bethesda, MD 20814

9 t

i FINAL REPORT - DER 83-49 DEFICIENCY EVALUATION 50.55(e)

ARIZONA PUBLIC SERVICE COMPANY (APS)

PVNGS UNITS 1, 2, 3 I. Description of Deficiency Af ter pre-core hot functional testing (HFT) in July 1983, the Unit 1 RCP (tag 1MRCEP01A) was disassembled to repair a previously identified linear indication in the pump casing circumferential weld. An inspection of the pump internals revealed four broken and two loose diffuser-to-casing retaining cap screws. Also, ten of the diffuser-to-suction pipe cap screws were found to be loose and the impeller shaft keyway bolt van broken, (Ref. NCR 2657). There was also slight cavitation damage on the leading edge of seven diffuser vanes.

As a result of the above, the other three Unit 1 RCPs (tags

JMRCEP01B, C, D) were disassembled and inspected. In addition to discovering loose and/or broken diffuser and suction pipe 4 cap screws, damage was noted as follows (Ref. NCRs SM-2658,

2659, and 2660)

The pump casing internal surface above the impeller had sustained peening damage from the loose diffuser cap screws.

The pump casing-to-diffuser fit indicated surface fretting damage had occurred due to relative m)tiot between the parts.

l Pieces were missing from the leading edge of one impeller

, vane on RCP 1B and two vanes on RCP 2A.

There were broken and/or loose impeller key retaining screws, shaft protection sleeve key retaining screws, and l carbon journal hearing key retaining screws.

The shaf t seals showed wear on the stationary seal ring holders and heavy deposits of 0-ring lubricant was noted throughout the seal assemblies. Also, bolts in the seal cartridges were loose.

4 Leakage was observed between the bearing sleeve and the l seal housing, indicating seal ring failure.

The PVNGS type ROI Reactor Coolant Pumps are supplied by Combustion Engineering (C-E), designed by Klein Schanzlin &

Becker (KSB) of West Germany, and were manufactured and performance-tested by CE-KSB in Newington, New Hampshire.

l

JP .

s

^

Mr. D. F. Kirsch DER 83-49 Page Two Evaluation CE-KSB has manufactured and tested a total of 24 duplicate '

' type R01 RCPn at their Newington facility for use in System 80

, plants. The Palo Verde IB pump, the first of this group

. tested, was run approximately 600 hours0.00694 days <br />0.167 hours <br />9.920635e-4 weeks <br />2.283e-4 months <br /> in the test loop. All the other pumps were tested'approximately fifty hours each.

The damage described in section I above was incurred during 700 to 1100 hours0.0127 days <br />0.306 hours <br />0.00182 weeks <br />4.1855e-4 months <br /> of operation per pump during the HFT. There i are no RCPs of.this design in any operating plant. The pumps at Palo Verde Unit 1 are the first of this particular design

{. 'to be operated in other than a test loop, therefore, any design or teaterial deficiencies would be evidenced af ter the HFT at Palo Verde.

) The root cause of the capocrew and impeller vane failures is L fatigue, i.e., cyclic loading exceeded the material fatigue-strength. In the case of both the diffuser-to-casing and ,

i diffuser-to-suction pipe capscrews, it was determined that l loss of pre-load preceded the fatigue failures. The loss of l pre-load was due to a combination of cyclic loading, a i i relatively low capocrew working pre-load (approximately 45% of j yield), and ambiguous and multiple load paths.

! The proximity of the impeller outer diameter (OD) to the

! leading edge of the diffuser vanes (ID), in this case 11mm (2.3% of impeller radius), contributed to the magnitude of the

! dynamic forces which caused the fatigue failures and to the cavitation damage of some diffuser vanes. Fatiguing hydraulic pressure fluctuations are generated at numerous rates with

! effects dependent on location within the pump. The close gap j hetween the impeller and diffuser vanes increases the pulse intensity and the potential for' fatigue failures. Operation j at runout (single-pump) which is approximately 142% of derign ,

flow, produces the highest pulse intensities and therefore the '

highest stresses in the working parts.

KSB conducted model tests to evaluate the effects of increased l j impeller-to-diffuser clearance to preclude diffuser vane

, leading edge cavitation damage and reduce the magnitude'of '

dymanic forces, both at design and runout flows. The '

hydraulic model employed for this test program was the same as used during the original development of this particular RCP.

+

This testing was also used to verify that the original design ,

j requirements for head-capacity, horsepower, and NPSHR would t

i 1

.+-mmvw,, ..wn-n-. v- ,nr-n, ---,..,,r-~ --,-----r, -, +,r,n-,----n---v-.nv- m,,-e,. ,-n:,c.,,, --n-n--, -,--.---n, . , . -

Mr. D. F. Kirsch DER 83-49 Page Three not he compromised by this modification. From these tests, it was determined that increasing the radial gap from 11mm to 28.5mm (6.0% of impeller radius) by cutting back and re-profiling the leading edges of the diffuser vanes would produce the desired results of reduction in the magnitude of the dynamic forces and elimination of cavitation damage.

These model tests also verified that head reduction at the design capacity due to the increased gap could be recovered by l hackfiling the impeller vanes and that the resultant change in runout performance would be acceptable.

Analyses of material properties of both capscrews and impellers by CE-KSB verified that the materials conformed to specifications were not deficient,'and that failures were due to fatigue.

A detailed survey of the three broken impeller vanes and nineteen intact vanes (the two remaining vanes were used for metallurgical examinations) revealed that the broken vanes were thinner than the intact vanes in the critical area at the leading edge to upper shroud junction. Both finite element stress analysis and strain gage tests were correlated to the known fatigue failure stress in the failed vanes, including variablen such as casting imperfections and stress concentrations. Using these values, the nnalysis of an intact vane with greater thickness and greater fillet radius than the failed vanes results in peak stresses well below the fatigue failure stress. .This design analysin by CE-KSB concludes that all the impe11ern now installed have a safety factor of at Jeant 1.5 over the impe11ers originally used in the Unit 1 RCPs. (For impeller modifications, see section III, paragraph 2). The increased impeller to diffuser gap further reduces loads, therefore, the current safety margin is now approximately 1.75 compared to the thickest vane which previously failed.

II. Analysin of Safety Implications C-E investigated various failure mechanisms of the RCPs and consequential damage to the system including locked rotor, degraded pump costdown and core flow blockage. In all canen it was determined that the internal failuren would not constitute a safety hazard, and that the reactor could he safely shutdown if any internal pump breakage and associated movement of broken parts were to occur. The safety requirement of the RCPn in to maintain the Reactor Coolant System (RCS) pressure boundary integrity. C-E has determined that if the conditions had remained uncorrected, the integrity of the RCS presnure boundary would not he compromined.

Mr. D.' F. Kirsch DER 83-49 Page Four Base.d on the preservation of the reactor coolant pressure boundary integrity and the fact that fragments of RCP internals have been predicted to not damage the reactor, have adverse affect on heat transfer to fuel elements, or interfere with control rod movement, the project evaluates the RCP deficiencies as not reportable under the requirements of 10CFR50.55(e) and 10CFR Part 21; since, if these conditions were to remain uncorrected, they would not represent a significant safety hazard.

III. Cor(ective-Action A comprehensive program was completed by C-E, CE-KSB, and KSB confire the root cause of the deficiencies and define and implement specific modifications. Additionally, programs were established to verify the modifications by analysis, model testing, prototype testing, and full-scale field testing.

These programs and their results are contained in Reference (1).

C-E's corrective action program and disposition of the referenced NCRs are summarized as follows:

1) The diffuser-to-casing fit (#1) and the suction pipe-to-diffuser fits (#2) were redesigned to eliminate eehiguous load paths and to minimize the nusher of multiple load paths, i.e., elleinstion of tapered fits, use of shorter ring segments at fit #1, and addition of a support ring and four reinforcing level pins at fit #2.

Additionally, the nusher of capocrews was increased (from 16 to 29 for fit #1 and from 16 to 24 for fit #2) and the design torques increased to provide working stresses of 85%

of yield. Also, the fit #1 capscrews were increased in length by 54% to make them less susceptible to loss of preload due to thermal and cyclic loading affects and the length of the threaded inserts (helicoils) in the diffuser was increased to guarantee complete thread engagegent. A re-analysis was conducted to verify that the capocrew preloads will exceed operating loads under all variations of operating conditions.

A total of 351 hours0.00406 days <br />0.0975 hours <br />5.803571e-4 weeks <br />1.335555e-4 months <br /> of testing was conducted in the CE-KSB test loop to generate baseline data on an unmodified pump and comparshle data af ter modifications. Accelerometer and strain sage data indicated that the original loss of preload could be duplicated within 100 hours0.00116 days <br />0.0278 hours <br />1.653439e-4 weeks <br />3.805e-5 months <br /> of runout operation. After modification, data verified that there was no change of capocrew strain after 150 hours0.00174 days <br />0.0417 hours <br />2.480159e-4 weeks <br />5.7075e-5 months <br /> of operation at runout.

Mr. D.- F. Kirsch DER 83-49 Page Five

2) The Atlas (foundry source) impe11ers for the Unit 1 & 2 RCPs were replaced by Schmidt and Clemens (S&C, foundry source) impe11ers which were available from other System 80 l plants. The S&C impellers have thicker vanes with a flatter profile than the Atlas impellers near the leading edge where the failures occurred. At sections further away from the leading edge, there are less differences in vane thickness between the Atlas and S&C impellers. The Unit 3 impe11ers were replaced with Atlas impe11ers where the vane leading edges were re-profiled so the leading edge thicknennen are equal to or greater than the replacement S&C impellers installed in the Unit 1 & 2 RCPs.

A11 impe11ers were backfiled (to compensate for head losa due to cutting back the diffuner vanes) and rebalanced.

One S&C and one modified Atlas impeller were retested in the CE-KSB tent loop in Newington to verify the hydraulic performance of the modified parts.

3) Examination of the broken impeller key retaining screws indicated that the sockets in the heads were formed too deep. This resulted in insufficient wall thickness in the head to shank area. These screws were redesigned with a larger head to accommodate the socket. These screwn will be installed with retaining sleeves and staked in place.

The shaf t protection sleeve and carbon journal bearing key retaining nerews are smaller screws and their looseness was attributed to vibration. The method of torquing these screws and staking them was revised.

4) The bearing nieeve to seal housing seal ring van replaced with a flexitallic type ganket. The size of the bolta in the joint was increased to insure that the gasket is l properly neated.

5) Regarding the shaf t seal deficiencien, instructions will be issued by C-E on the proper une of "0" ring Jubrication.

Precautions will also be taken to insure that the seal cartridge bolts are properly torquad.

The Unit 1 seal cartridgen have been designated an test sets and have been installed in the Unit 2 RCPn for une during hot functional tents (HFT). They will subsequently be used for Unit 3 HFT. Unit 2 seal cartridges have been installed in the Unit 1 RCPs, reference NCR dinpositions.

i l

.. .. .e Mr. D. F. Kirsch

. DER 83-49 Page Sir

' 6) Special demonstration tests during July / August 1984 for the four modified Unit 1 RCPs, with subsequent disassembly of the 2B pump for a detailed inspection, have reconfirmed the adequacy of the corrective actions. The test sequence included the same conditions as those which resulted in the original damage and totaled 737 hours0.00853 days <br />0.205 hours <br />0.00122 weeks <br />2.804285e-4 months <br /> of running time on the 2B RCP (Reference 2).

O The post-test inspection of the 2B pump confirmed that all the diffuser and suction pipe cap screws were properly secured. Alan, there was no evidence of cavitation on the re-profiled. leading edges of any of the diffuser vanes.

Liquid penetrant inspection of the impeller vanen showed no unacceptable indications,-(Reference 3). Some evidence of minor cavitation was noted on the convex side of three of

'the six impeller vanes near the ' vane to lower shroud interface, approximately 35 mm back from the leading edge.

CE-KSB and KSB evaluated this evidence as acceptable, (Reference 4).

7) The required corrective actions for Unit I were implemented prior to the above-mentioned demonstration test via the following Design Change Packages (DCPs).

ISM-RC-107 ISM-RC-109 ISM-RC-113 Additionally, these same changes will be implemented in the Unita 2 and 3 RCPn prior to their fuel load via DCPs as followns 2SM-RC-107 3CM-RC-107 2SM-RC-109 3CM-RC-109 2SM-RC-113 3CM-RC-113 These DCPn provide the corrective action disposition of NCRs SM-1657, -2658, -2659, and -2660.

IV. References

1. Letter V-CE-30867 August 30, 1984 C-E Final Report (proprietary) on Palo Verde Nuclear Generating Station Reactor Coolant Pumps, CEN-271(V)-P, August 1984.

2. Letter V-CE-21757, August 14, 1984.

3. Letter V-CE-21756, August 13, 1984.

4.~ Letter ANPP-30677, September 27, 1984.

7

$