Text

Interim Rept on Palo Verde Nuclear Generating Station Reactor Coolant Pumps
	ML20100M126
Person / Time
Site:	Palo Verde
Issue date:	01/31/1984
From:	; ABB COMBUSTION ENGINEERING NUCLEAR FUEL (FORMERLY
To:	;
Shared Package
ML20100M077	List: ML20100M126;
References
	FOIA-84-212 CEN-271(V)-NP, NUDOCS 8412120216
	Download: ML20100M126 (175)
	v • d • e

{{#Wiki_filter:. CEN-271(V)-NP i + l l e 4 INTERIM REPORT ON PALO VERDE NUCLEAR GENERATING STATION REACTOR COOLANT PUMPS i g January,1984 e Og220216840612 BERNABEB4-212 PDR 1 POWER E SYSTEMS COMBUSTION ENGINEERING, INC. U a

LEGAL NOTICE THIS REPORT WAS PREPARED AS AN ACCOUNT OF WORK SPONSORED BY COMON ENGINEERING, INC. NEITHER COMBUSTION ENGINEERING NOR ANY PERSON ACTING ON ITS BEHALF: A. anarse ANY WARRANTY OR REPRESENTATION, EXPRESS OR IRWLIED INCLUDING THE WARRANYlES OF PITNESS POR A PARTICULAR PURPOSE OR RNRCHANTA88LITY, WITH RESPECT TO THE ACCURACY, COMPLETENESS, OR USEFULNESS OF THE INF RMATION CONTAINED IN THIS l REPORT, OR THAT THE USE OF ANY INFOR'.ATION, APPARATUS, AETHOO, OR PROCESS DISCLOSED IN THIS REPORT MAY NOT INFRINGE PRIVATELY l OWNED RIGHTS:OR l E. Amme ANY LIABILITIES WITH RESPECT TO THE (25 OF,OR POR DAMAGES RESULTING PROM THE USE OF, ANY INFORMATION, APPARATUS, METHOO OR PROCESS DISCLOSED IN THIS REPORT. 1 6 e 4 ( L

n 3 a ] t. o 8 INTERIM REPORT ON PALO VERDE NUCLEAR GENERATING STATION -UNIT #1 REACTOR COOLANT PUMPS w Prepared By: Combustion Engineering, Inc. Windsor, Connecticut a

TABLE OF CONTENTS INTRODUCTION ABSTRACT O ~ PART I DIFFUSER / SUCTION PIPE 80t. TING PART-II DIFFUSER CAVITATION PART III IMPELLER FAILURE D-

INTRODUCTION LICENSE AGREEMENT CE-KSB Pump Company, located in Newington, New Hampshire, has a License Agreement with Klein, Schanzlin & Becker A.G. (KSB) of West Gemany to design and manufacture Reactor Coolant Pumps of the KSB design. .The engineering resources of KSB have been utilized throughout the investiga- -tion and review of the concerns described in this report. Representatives from KSB have inspected the Reactor Coolant Pump components, both at the Arizona'Public Service Company (APS), Palo Verde Nuclear Generating Station (PVNGS) site and at the C-E Newington facility. KSB representatives have been involved in meetings with Combustion Engineering (C-E), CE-KSB, and APS, and have reviewed and approved all design modifications.as described in this report. INDEPENDENT CONSULTANT Dr. Elemer Makay, President, Energy Research and Consultants Corporation, Morrisville, Pennsylvania 19067, was retained ty C-E at the request of APS as an independent consultant. He has been actively involved in reviewing the . problems discussed in this report participating in the meetings, witnessing and advising on various test programs, and reviewing the test results. GENERAL DESCRIPTION OF THE PUMP The CE-KSB Reactor Coolant Pump is a vertical, single stage, diffuser-type centrifugal pump with bottom suction and radial discharge. The pump develops 3 a dynamic head of 111.25 m (365 feet) at a rated flow of 7.028 m /s (111,400 gpm) while operating at 1190 rpm. The general arrangement of the major pump components is as follows. The item numbers refer to FIGURE I-1. The pump rotating assembly and diffuser (1) are mounted in a spherical pump casing (2) of forged low alloy carbon steel and. ..-,~,-----e,,.c----

internally clad with stainless steel. The pump impeller (3) is double keyed and locked to the pump shaft. A water lubricated journal bearing (4), located iinnediately above the impeller, is mounted in the seal housing. Additional shaft support is provided by an oil lubricated combination journal and thrust bearing (6), located in the thrust bearing housing, which is mounted in the motor support assembly. Because the pump drive shaft penetrates the reactor coolant system pressure boundary, it is sealed by a hydrodynamic shaft sealing system. A rigid coupling (5) is located between the shaft seal and the combined journal and thrust bearing to connect the pump shaft to the drive shaft. This coupling allows separation of the two shafts for replacement of the seal assemblies and water lubricated journal bearing without removing the motor. The motor (8) is coupled to the drive shaft by a spacer-type flexible coupling (7), which minimizes dynamic interaction between the drive motor and pump. The pump components are described in detail in the following para-graphs. Major Components The major components under discussion in this report consist of the pump casing, diffuser and suction pipe, and impeller. Pump Casing - The Reactor Coolant Pump casing is designed and fabricated in accordance with the ASME Boiler and Pressure Vessel Code, Section III, Class 1, and is a spherical design forged of SA-508 Class 2 or Class 3 material. The casing internal surfaces are clad with type 308/309 stainless steel. The heavy walled, symetrical configuration of the pump casing allows all support loads to be taken through the casing wall. + Diffuser and Suction Pipe - The CE-KSB pump incorporates an eleven blade, removable diffuser. The diffuser is fabricated in two sections and assembled in a preferred orientation in the pump casing. The diffuser sections are cast from 13% Cr-4% Ni material. The suction pipe is fabricated from the same material, and is supported in the pump casing by the diffuser. 2

Impeller. - A six-blade, radial flow impeller is used. The number of impeller and diffuser blades is selected to minimize hydraulic forces. No splitter or guide vanes are required in the pump suction pipe. System Operation

~

Due to the nature of the C-E System 80 design the Reactor Coolant Pumps ' operate at two distinct points on their head capacity curves. The preferen-tial design point results when all four pumps are operating. However, in startup and cooldown there are restrictions that limit the number of pumps in operation to three or less. When a single pump operates there is the addi-tional flow path back through that loop's idle pump to the steam generator. As a result, the flow resistance is significantly lower and therefore this - flow (calledrunoutflow)ishigh. Cherearever7 slight differeticas from these values when the reactor coolant is cold (lower flowrates). Two-pump operation on the same steam generator provide essen-tially four-pump flow rates, whereas two pumps operating in opposite loops provide essentially single-pump flow rates. Finally there is the case of three pump operation, wherein the two pumps on the same steam generator are at the four-pump rate and the remaining pump is at the single pump rate. There are small variations in the above flow rates from case to case as there are also variations from pump to pump due to manufacturing differences. e 3-

^ k jd. g 8 .r,..- ,) 7* 4 7 y; .h h Uj h3 6 ..'9 I'! e vvC4 i k (rj. =F \\ l' l\\ i " " %}l',[g A.: N"' ! 1::~.~. *,"i b. lk.[%"Yk 3 1 .?. ** gN .q . gh f e,.ws. -.x - %N,5

cj ~'ph QW '~'l

,f fl .y GENERAL ARRANGE?tEftT OF REACTO.1 CCOLANT PUftP FIGURE I-1

'A85 TRACT The pre-core Hot Functional Post Test Inspection of Reactor Coolant Pumps Serial Numbers 1109-1A,1109-18,1109-2A, and 1109-28, operating in the Arizona Public Service Company's PVNGS Unit 1, identified three deficiencies. Each of these deficiencies is described in separate parts of this report. The three deficiencies are shown on FIGURE A-2, and are briefly described below: 1.- BOLTING - There are two bolted connections inside each pump casing. Several cap screws were found broken and/or loose at the diffuser / casing connection, and several cap screws were noted to have come loose in the diffuser / suction pipe connection, see FIGURE A-1. 2. DIFFUSER CAVITATION - All four pumps had one or more diffuser vane upper inlet tips with cavitation damage. The intensity of the damage varied among vanes in any one pump and also among pumps. This damage was in a highly localized area at the leading edge of the vane at the upper shroud of the vane interface in each pump, see FIGURE A-1. 3. IMPELLERS - Two of the four impellers have missing segments from the leading edge of the vane. One impeller had one broken vane, the second impeller had two broken vanes and one cracked vane, see FIGURE A-1. AnintensiveengineeringreviewwasinitiatedatC-E,C-ENewington(CE-KSB), and KS8 (West Germany) to determine the causes of the deficiencies, identify the necessary modifications, and to verify the modifications by model and/or prototype testing. The aforementioned program included the following: 1. Model tests with increased impeller / diffuser gap to evaluate the re-duction in pressure loadings within the pump. This modelling was done to optimize the gap size. 2. A metallurgical investigation of both the bolting and the impeller material to determine failure mode. A-1

3. Model tests to verify known impeller stresses and to identify additional loadings which may result from runout flows. 4. Refinement of the impeller acceptance criteria. 5. Prototype testing in the C-E Newington Test Facility to collect baseline data on the full size hydraulic components at operating temperature and pressure as originally designed and as modified.- Details of the recommended design modifications along with a description of the support design analysis and parallel studies, metallurgical investiga-tions, model hydraulic testing, and the final prototype test in the C-E Newington full scale pressure and temperature test facility are described within this report. A demonstration test is planned for Palo Verde Unit 1 to confirm the adequacy of the repairs to the reactor coolant pumps under operating conditions which are similar for purposes of testing to those during the pre-core hot func-tional test. During the test, data will be taken for various reactor coolant pump combinations at selected coolant system temperatures and pressures. The maximum pressure for the test is 2250 psi and the maximum temperature is 565'F. 4 O t 9 A-2

e e e' SuletARY OF IIEACTOR COOLANT PUNP INVESTICATION FOLLOWING COLD HYDROSTATIC AND PRECORE HOT FUNCTIONAL TESTS Initial On-Site Inspection - July 16-22, 1983 M ER GF D45A55DWLY 1 2 3 4 Total Run Hours / Number / 1136/16 686/18 986/18 1041/15 starts of PU W NUDBER 1A 18 2A 2B )lFFUSER BOLTING 2 loose /4 broken (3 out of 4 0 loose /1 broken (but 1 loose /1 broken 5 loose /3 were broken of broken bolts came free of captured) which 2 were chipped. locking sleeves & caused mech-Wedge - 0* - Tight anical damage (peening) Inside Wedge 0* - tight 180* - Loose Wedge - 0* - loose pump casing.) 180* - Loose 180* - very loose i Wedge - 0* - tight & cocked 180* - loose & cocked i Top plate worn; 3 o' clock key broken 2 8y WCTION PIPE BOLTING 10 loose. Some fretting of the All 16 in good condition. Good condition. 2 broken and partially backed y diffuser and suction pipe joint. out. NDE of Suction Pipe @ 1 Significant wear at lower end CE-KS8 showed crack in taper suction pipe on casing. Ilp which is junction of suc-J tion pipe to diffuser. [)lFFUSER VANE CAVITATION 7 vanes, vane #1 has the most 4 vanes, vane #1 has the 1 vane - very slight 7 vanes, vane #1 has'the most damage

20 men wide, 20mm damage 4 mm deep 20 sun mark.

damage *. ] down from the upper shroud. down from upper shroud.

Small hole in vane #1.

l

Small hole.

l ] IWELLER VANES NDE @ CE-KSB shows no cracks. I vane with missing segment. NDE @ CE-KS8 shows no 2 adjacent vanes with missing j NDE @ CE-KS8 shows no other cracks. segments. IOE @ CE-KS8 showed cracks. a second vane with a through thickness crack at the vane junction to upper shroud. 4

g l h 5 e' t j wwi l r Iff' 61 M lu //// e !w; u. l l I T j c / -Q h / o (h kM / h / / 1L / Diffuser / Casing Bolted N ff.;rs. 9.) 5,. Connection - PART I Kr ty's Sg.' g'Jh r N Diffuser Cavitation y atre(y {y PART II I b h Mt h simpeller Inlet Vane N PART III Diffuser / I$ / \\ Suction Pipe f Bolted / Connection, - l PART I L \\ C' 'l \\ T v i IDENTIFICATION OF FROBLEM AREAS FIGURE A-2 1 ' ' ' *,'. s/.., .'{ ,..r. . '..',, ; J ' /- 4. .I _ i.,' j p *, 'y .4... g '.i' 1 yll,-

ip,

.,g - ,,v:'*+y 4 g s :.,. ..+ .p 4 y... -

.s PART I o DIFFUSER / SUCTION PIPE BOLTING p p 0 m I-1

PART I DIFFUSER / SUCTION PIPE BOLTING TABLE OF CONTENTS

1.0 INTRODUCTION

1.1 Description of Problem 1.2 Safety Implications i. 2.0

SUMMARY

2.1 General Description of Inspections and Testing 2.2 Description of Original Design 2.3 Analysis of Results 2.4 Analysis of Deficiencies 2.5-Description cf Modified Design 3.0 INSPECTIONS AND EXAMINATIONS 3.1 Site Inspections 3.2 Component Examinations 4.0 TESTS AND ANALYSES 4.1 Metallurgical Tests s, 4.2 Failure Mechar. ism 5.0 DESIGN MODIFICA.ONS 5.1 Diffuser / Casing Connection 5.2 Diffuser /Suctica Pipe Connection 5.3 Analysis of Diffuser / Casing Connection 1-2

.PART I DIFFUSER / SUCTION PIPE BOLTING TABLE OF CONTENTS (Cont'd). 6.0 PERFORMANCE VERIFICATION 6.'.- C-E Newington Tests 6.2 Test Requirements 6.3 Instrumentation 6.4' Test Procedure 6.5 Results-and Evaluation ?-3

L .PART I-DIFFUSER / SUCTION PIPE BOLTING LIST OF FIGURES Fig. No. Title 2.2-1 Original Configuration, SYS80 RCP 2.2-2 Diffuser / Case Connection Original Design -2.2-3 Typical Locking Sleeve for Diffuser and Suction Pipe Cap Screw 2.2 Diffuser Wedge Assembly 2.2-5 . Suction Pipe / Diffuser Connection Original Design 3.2-1. Cap Screw Inspection Sunnary APS PVNGS Unit 1 Pump S/N 1109-1A 3.2 Cap Screw Inspection Sunnary APS PVNGS Unit 1 Pump S/N 1109-18 3.2-3 Cap Screw Inspection Sunnary APS PVNGS Unit 1 Pump S/N 1109-2A 3.2-4_ Cap Screw Inspection Sunnary APS PVNGS Unit 1 Pump S/N 1109-2B '4.1-1 Three Cracked Diffuser Cap Screws 4.1-2 Fractured Surface of Cap Screw D2A9 .4.1-3 Fractured Curface of Cap Screw D2A11 4.1-4 Fractured Surface of Diffuser Cap Screw D1810 4.1 failed Suction Pipe Cap Screw S2A1; 4.1-6 Microstructure of Cap Screw D2A9 (Tempered Martensite) 4.1 Microstructure of Cap Screw D2A11 (Tempered Martensite) -4.1-8 Microstructu.*e of Suction Pipe Cap Screw S2A1 (Tempered Martensite) 5.1-1 Diffuser / Case Connector Final. Design-5.1-2 Final Configuration, SYS 80 RCP 5.2-1 Suction Pipe / Diffuser Bolted Connection Final Design i 5.2-2 Suction Pipe / Diffuser Pinned Connection Final Design 6.1-1. Pump Test Loop l6;3-1 ~ Instrumentation Installed for RCP Baseline Testing at CE-Newington I-4 <2_._,

l .PART I DIFFUSER / SUCTION PIPE BOLTING LIST OF TABLES Table No. Title j 'I 4.1-1 Sunnary of Mechanical Tests with Diffuser Cap Screws 4.1-2 Chemical Composition of Diffuser Cap Screws - T i l i!.',. i. l l l I-5 l

PART I DIFFUSER / SUCTION PIPE B0LTING

1.0 INTRODUCTION

1.1' Description of Problem The' inspection of the four (4) APS PVNGS Unit 1 Reactor Coolant Pumps following Pre-Core Hot Functional Test identified anomalies in -the bolted connections at the diffuser / casing and the diffuser / suction pipe. The damage varied from pump-to-pump, but all fcur pumps exhibited some deterioration in the form of loose # broxen cap screws. FIGURE A-1 sumarizes the results of the inspection. Additional details of the anomalies are described in Section 3.0 of thisreport(Parti). The history of the operating time on each of the four APS PVNGS Unit F

1. Reactor Coolant Pumps is shown in FIGURE A-1.

Hours are shown for-both cold hydrostatic test operation and for pre-core Hot Functional Tests. 1.2 Safety Implications e C-E has reviewed the potential failure mechanisms associated with the diffuser / suction pipe cap screw deficiencies and their conse- .quences. C-E has determined that: During' four pump full power operation, hydraulic forces alone can - maintain the diffuser in place. Only during startup or coast down ,7 is there any potential for axial movement of the diffuser if the cap screws should fail. The design is such.that the diffuser / suction pipe assembly is captured radially regardless of any axial movement. The diffuser can not rotate because it is restrained by two keys whicn engage the mating pump casing-ledge. With these design features, the potential for. impeller birding is remote during start-up and coast down. I-6 +.

The radial gap 'between the. impeller and diffuser is small enough to prevent the escape of pieces from broken cap screws that are large enough to cause local flow blockage. This conclusion was verified in Reactor Coolant Pump 1109-1A in that the three broken diffuser cap screw heads which came free from their locking sleeves did not pass through the impeller / diffuser gap. e The total assessment of the effects of the cap screw deficiencies; shows that the pump maintains sufficient flow to satisfy the criter-ia of the safety analysis.. If the diffuser / suction pipe deficien-l' cies were to have remained uncorrected, they would not have adverse-ly affected safety at any time throughout plant life. i O k I-7 e, .,.,-..,n.,

4 .PART I DIFFUSER / SUCTION PIPE BOLTING 2.0

SUMMARY

j 2.1' General Description of Inspection and Testing w Following disassembly of the pumps, the broken cap screws were metallographically examined. Fatigue indications were evident on o the fracture surfaces. The properties exhibited by the cap screws met the material requirements. A review of the bolted connections was initiated as a result of the APS PVNGS Unit I cap screw failures. This review included all of the significant. loading conditions expected within the lifetime of. the pump, included hydraulic loads and thermal effects, and also' dynamic loads associated with heatup/cooldown and starts / stops over the. lifetime of the pump. A detailed prototype test program was developed and implemented to instrument the bolted connection at the diffuser / casing interface during tests which were conducted at C-E Newington. 2.2 Description of the Original Design 2.2.1 Diffuser / Casing Bolted Connection i O i i I-8 L

~ o 2.2.2 ' Diffuser / Suction Pipe Connection 4 m N ) 'I-9 u,-s Y

4 ORIGINAL CONFIGURATION, SYS 80 RCP FIGURE 2.2-1

F:~ ; j T i f S. 4 OlFFUSER/ CASE CONNECTION ORIGINAL DESIGN ' FIGURE 2.2-2 . _. g:

'&r 1 ,4 .4 g s. 4 T e-f- ,F4 2' r H ,.'s:. A h k i FIGORE 2.2-3 ~ l . Typical Locking Sleeve for Diffuser ~ and Suction Pipe Cap Screws t 1 s

9 FIGURE 2.2-4 Diffuser Wedge Assembly

r -y n s.: L [ B 4 1 SUCTION PIPE / DIFFUSER CONNECTION. ORIGINAL DESIGN FIGURE 2.2-5 =;

[ 2.3 A'nalysis of Deficiencies. Diffuser / casing cap screws and diffuser / suction pipe cap screws were found to.be broken or bent following the pre-core Hot Functional Test at APS PVNGS Unit 1. Fatigue has been identified as the failure mechanism of the diffuser cap screws. Fatigue has also been-l'* I identified as the most likely failure mechanism of the suction pipe cap. screws t,ut due to the poor condition of the fracture surfaces this cannot be metallurgically confinned. No evidence was found to indicate that failure initiated due to defective material, improper manufacturing procedures, embrittlement, or corrosion. The fatigue failures were solely due to overstressing the bolts under cyclic mechanical loading. Fatigue loading of the bolts is possible if the initial bolt preload is less than operating service loads or if the parts loosen or 1 settle because of other joint design deficiencies. Both the dif-L fuser/ case connection and diffuser / suction pipe connection original-ly employed multiple tight clearances and ambiguous load paths to the extent that these features may have detracted from the preload actually stored in the bolts at assembly. Thus, bolts which are tight could, after a short period of time, become -locse because of thermally induced shifts and normal mechanical vibrations. In the modified. design, the ambiguities in the bolted connection load paths' - have been removed and the clamping load increased such that the preload far exceeds operating loads. 'n, 2.4 Description of the Modified Design r l- - 2.4.1-Diffuser / Casing Bolted Connection i f-I-10

f. 9 d 2.4.2 Diffuser / Suction Pipe Connection m e ' e ' 4 4m I-11

2.4.3 Preferred Diffuser Orientation ~~ ~2.5 Analysis and Verification of Modifications The analytical basis for the modified design is contained in para-3raph 4.2. The design criteria applied is that the friction and ~ clamping forces must exceed the applied _ static and dynamic hydraulic L loads. The adequacy of the modified design has been demonstrated by a test . program on a specially instrumented production pump in the CE-l'ewington (CE-KSB) Test Facility. A series of four tests were 7 conducted (two on the original design and two on the modified ~ des'egn) to establish the behavior of both designs. Test results show substantial reductions-in cyclic loading on the cap screws to 'the extent that essentially none exist. Details on the test program and results are found in Section 6.0 of Part I. L, I-12

~ -PART I DIFFUSER / SUCTION PIPE BOLTING 3.0 --INSPECTIONS AND EXAMINATIONS 3.1 Site Inspections All parts from the diffuser and suction pipe connections were

~

labeled upon disassembly from each Reactor Coolant Pump. Wherever possible each part was identified as to unit and location within' the unit. A visual examination was made on the majority of parts removed from the Reactor Coolant Pumps. After all parts had been disassembled and cataloged, parts were distributed to C-E-(Windsor), CE-KS8 (C-E Newington), KSB (West Germany), or remained on site at APS PVNGS. The purpose of the dis-tribution was to: a) Expedite testing and examination h) Provide independent confirmation of test results c) Ensure tests were conducted utilizing the most up to date equipment. J 3.2 Component Examinations

l..

l 3.2.1 Diffuser / Casing Connection The pictoral summary of the broken or loose cap screws in each of the four APS PVNGS Unit 1 Reactor Coolant Pumps is-shown in FIGURES 3.2-1, 3.2-2, 3.2-3, and 3.2-4..Nine out of 64 cap screws in the-diffuser / casing connection were found to have broken in the four Reactor Coolant Pumps. Eight out of 64 cap screws were found.to have loosened. Forty-seven out of the 64 cap screws were found to L .be intact but bent. Several cap screws were selected for metallur-gical examination, and these results are presented in Section 4 of this report. I-I3

e ~ In Pump S/N 1109-1A, three.of the four broken cap screw heads came free fro;a the locking sleeves and caused peening on the inside of the pump casing. Inspection of the locking sleeves for those three screws indicated that they had not been completely staked into the grooves in the bolt head (Figure 2.2-3). No broken cap screw heads were observed to have fallen out of any of the other connections. Inspection of all other locking sleeves indicated that they had been properly staked. Those cap screw heads that did become free of the locking sleeves were rotated around the casing by the top shroud of ~* theimpeller,andtheirsizewasreducedfrom( ]inchtoless ~than( inch diameter. Most of the wedges described in Paragraph 2.2.1.6 were loose. Particularly those wedges installed at the 180' orientation from the discharge nozzle centerline. 3.2.2 Diffuser / Suction Pipe Connection The pictoral sunnary of the broken or loose cap screws in each of f the four APS PVNGS Unit 1 Reactor Coolant Pumps is shown in FIGURES 3.2-1, 3.2-2, 3.2-3, and 3.2-4. Two cap screws out of 64 were found ' broken while 10 were found to have loosened. Fifty-two cap screws were in the same condition as originally installed.. All cap screws were effectively secured in the suction pipe flange by the locking sleeves. A( mm crack was discovered as a result of the Non-Destructive Examination (NDE) at C-E Newington in the upper portion of the taper j_ lip on the suction pipe of Pump S/N 1109-2A-(see FIGURE 2.2-5 for the taper location). j' I-14 L

Pump S/N 1109-1A exhibited.the most loose and/or broken cap screws in both bolted connections. Evidence of wear and degradation of-other components was greatest in this pump. O 9 e \\ I-15 _. ~.... _ -.

PART I-DIFFUSER / SUCTION PIPE BOLTING I i w f i ( N -<g __o~s [ /6 s,b 6x \\ s ~ \\ f 0 dl l =' : \\o fyo) \\ ~ ~ y 9,/ \\ / N + \\. if E / \\ x l' ~~O-O/ ~ ~ l. 1 ORIGINAL CONFIGURATION, SYS80 RCP @LooseCapScrews 3roken Cap Screws CAP SCREW INSPECTION

SUMMARY

APS PVNGS UNIT 1 PUMP S/N 1109-1A l FIGURE 3.2-1 l

PA.RT I DIFFUSER / SUCTION PIPE BOLTING I i c f 1 ( 1 zo, < x - dx 's 16 \\ j \\ 's 6 a / \\ p"b f ,o +- I 5 \\ T 'Se '935ggJgj9' ~ O __Q j v ~ l l l ORIGINAL CONFIGURATION, SYS80 RCP

Loose Cap Screws CAP SCREW INSPECTION

SUMMARY

@3rokenCapScrews APS PVNGS UNIT 1 PUf1P S/N 1109-18 l l FIGURE 3. 2.

PART I DIFFUSER / SUCTION PIPE BOLTING I w e i 1 ( N /0 . a Ox /b DN ~ / \\ / l 1 i Y I i

> O,l

\\ / O' r-Q~ /v ~ )~ l i I j ORIGINAL CONFIGURATION, SYS80 RCP

Loose Cap Screws

@BrokenCapScrews CAP SCREW INSPECTION

SUMMARY

APS PVNGS UNIT 1 PUMP S/N 1109-2A l l l FIGURE 3.2-3

~ PART I ~ DIFFUSER / SUCTION PIPE BOLTING I i i t I \\c N O-O' /0' a dx \\ / \\ / .s e 1 i 1 o,j \\o ', ~ ~ \\ p / 'xO_ _Q,w ' ~ ~ 1 ORIGINAL CONFIGURATION, SYS80 RCP

Loose Cap Screws

@BrokenCapScrews CAP SCREW INSPECTION

SUMMARY

APS PVNGS UNIT 1 PUttP S/N 1109-2B t FIGURE 3.2-4

.PART I DIFFUSER / SUCTION PIPE BOLTING 4.0 TESTS AND ANALYSIS 4.1 Metallurgical Tests 4.1.1 Introduction and Background The diffuser cap screws were ordered to ASTM Specification A193 Grade B6 (AISI Type 410) with additional Charpy impact testing requirements; the diffuser / suction pipe cap screws were also ordered to the ASTM Specification. The following seven failed cap screws were examined at C-E Windsor. c Cap Screw Broken at fillet shank to head, fracture oxidized 01A8*: and excessively worn. Cap Screw Broken through sixth thread from the threaded end ~ 01810: and fracture oxidized. Cap Screw Three threads broken off to centerline of_ cap screws, 02A9: fracture oxidized (FIGURE 4.1-1 and 4.1-2). Cap Screw -Broken through sixth thread from the threaded end, D2A11: fracture oxidized (FIGURE 4.1-1 and 4.1-3).- Cap Screw Cracked same as cap screw D2A9, but end not broken 02A12: off(FIGURE 4.1-1). Cap Screw Broken through third thread from the threaded end, D2B9: fracture and some threads completely worn.

Key:

Identification of Cap Screws: 0= Diffuser S= Suction Pipe IA, IB, etc. = Pump Number 8, 10, etc. = hole number (see Figure 3.2-1) I-16 l _a

Cap Screw Broken'through thirteenth thread from the threaded $2A1: end, fracture oxidized and a portion of it worn (FIGURES 4.1-4 and 4.1-5). ~4.1.2 Examination Diffuser cap screws from the pumps were all manufactured from the same lot. Mechanical testing and chemical analysis results are shown in TABLE 4.1-1 and TABLE 4.1-2, respectively. As can be seen, they are in compliance with ASTM Specification requirements, and are also_ very close to the values reported by the material certificates. Of the seven cap screws examined at C-E Windsor, cap screws DIA8 and D2B9 had the fractured surface damaged to such a degree that they were useless for investigation. The other five cap screws are shown in FIGURES 4.1-1 to 4.1-5. Cap screw D2A12 was cracked. This crack was in the same direction and location as the fracture of cap screw D2A9. Cap screws D2A9, D2A11, D1810, and S2A1 (FIGURES 4.1-2, 4.1-3,and4.1-4)wereselectedforfurtherdetailedexamination. Although the majority of cap screws examined provided no clear indication of the failure mode, D1B10 did show evidence of fatigue. The " beach and river" markings shown at location A in FIGURE 4.1-4 are evidence of this type of failure. I-17

L, v n :...:....:.: . ~ e.,: c .m... n . v a ; ;. - n,

w.

.m l 2: 9 - f 1.y.

f..

a h. -a'R y f J '. s The faceted fracture surface suggests that more than one fatigue .,f f crack had initiated. A possible second initiation point is shown at <.j* ? location B in FIGURE 4.1-4. The existence of INltiple Cracks is '. \\f.; ' ,,.{ p indicative of alternating loads. .n.:;,".p,. ,7 .j.; '.}/ 7.! The fractcre surfaces of the broken cap screws were examined on the (. '...* T I^ scanning electron microscope. There was no indication of embrittle- ? 1 ment (cleavage) or intergranular cracking. B. I f r '. > %.. u ..y. - c Cap screws which did not fail were bent. Twenty-nine diffuser cap ' [ {-3 n' .*9, screws and 34 suction pipe cap screws were measured for straight-

.j : :.y [.:,..;

ness. The degree of bending varied from 0.004" to 0.146". h.j,.I.: 7 h- .-7 ,i...... .[. ~ 4.1.3 Conclusions 5 l. ((.. ./ The results of the examination of chemical composition, mechanical y properties, microstructure, and fractured surfaces of the failed cap I screws confirmed that the cap screws were made in accordance with g} c.' 'l A: 7 Specificatiors and C-E requirements. There were no indications ,4 which would sugge 'efective material or improper manufacturing

'A "Y

.j procedures.

he cap screws broke by fatigue solely as a [ result of me . -:4

I.Q,.

0 I E.- The failure - w D1810 is significant in that this is .'..h ~ 3

g. -

the only cap b this pump (S/H 1109-1B). Based on n s. .. ~..i, V this and the 11 screw D1810 fractured surface with x,.. w. all others, h d that fatigue is *ne primary failure

..} [.g; p

D mechanism of tne Jf screws. $w.. T. M g% 4 .it. 6 The fact that all cap screws which did not break were bent would 'CW' E..:%. M7 indicata that the cap screws lost their pre-load during the test [h.,j $ < g 4V run, permitting the diffuser to move at the contact surface of the }c.y '

.i pump casing.

This rroven:t applied a bending load which caused the f:A' N f ty >+. bending and cracking of the cap screws. / D i. ~ (( . x..:.._,. N.? L.,,.

. }_;.

' 16,} - p.'. I-18 y '3,g .f h'...lqfh _3 .y, f [;W ll ^l. ; _.' T ; y ; ':: _ j,l* ;.a., [ _ ' y }.._ -j[ 4.g j.-[3. y..} v

v

',., J _f V j -l l ?$.;;.,.@& G.....+.

...M*= W & ~

.1.. % s.ls' :lo*).....c.,:L ?.ii h :T..,..,;.l: '. s k k._...^.h I.

.~ L

. ;hJa' c

L. y, l 1:. f.... N. This metallurgical investigation was performed by the C-E Hetallur- -[.c, :..T, gical Laboratories at Windsor, Connecticut. An independent metal-i;$i{.f] lurgical investigation has been conducted by KSB, West Germany, and . [ (.. ;,. is in agreement with the C-E investigation. ...L... + ~ ~. .: y l v.+.> .g... 73

~ 1. c: -.,: -

_..=.;..;

c...

,4,,: '.f ,? .sst:.- ... - y..

4... -

s.....

... i . ;. e.: ';_ ; ' s. ,. i. . f. '.^_',?.*'. ..,9;. ,e-f ' s' * *, -N stab. a 's b. g- .g

f. f._.s.f. :(..

., ;..... p . -j. - f>- -V... '.1. .q - .n. .-1.* 5 1 l t -,,, -...'y... i .g..- ..a ....dja;."; ~

: k ?;.

'3.- + 4 p. -9. .- ::. 7.. e w 1.. .'; -...-,s, p * - . f,._ ?. n: ~Q.' .s .,6: ss.99+.. J '.; . [f -... _, _ p:. ,,-e P. ,:.. = n,.' 8 ' .n I-19

3. > l..

a s e..

+9

[ 4'e a 1 ...r. <r ' y 9A. . -..[, 'g,j { h". . { y e -{' f. - ' *i } .,[F , [. 1.f. Th %. 8,, i, p>. p.' - - ;~.. ',,,'gl J. +, c r

yw-- "e o-Ju Gd y-. an A A - e .- A M Q-W. l - On, u - U M CL.' n LJ ~% e en : 3-M 46= r :- g.' w i Cll3 - M. Y M+e - LaJ d 3 k . in enW W l t-g U e E g .c U Q W .L l f -[~ ~ M i -a. L i I f t . f I ~ J-s ' r ~ "r-- .,n.,--_ TTY 9WW-v-%w,__,_,,

M 'O ~ ~ PART I DIFFUSER / SUCTION PIPE BOLTING TABLE 4.1-2 Chemical Composition of Diffuser Cap Screws (%) e + <f TABLE 4.1-2 ..e.

i Fig. 4.1-1 Three Cracked Diffuser Cap Screws e N -r .~

] ~ s Fig. 4.1-2 Fractured surface of cap screw D2A9 1 l l Fig. 4.1-3 Fractured surface of cap screw D2A11 i l = ..... _. ~... _ - -,.

g,

g: -

V-T .4 T

.4 1

..: / .-i s4 M i. {n

E'-

1 s e t I+ t . Fig. 4.1-4 Fracture surface of - Diffuser cap screw D1810 '.P-..

y -) 2 L f ='. s Wx .g "1 2-2 s.~ .3- -,;.J. ': t o.. n e f 6,, t-9 i e A k y tY b-', -l w k 's. "T r O -fig;-4.'l-5 Failed section pipe ~ caps: screw S2A1 u~. .c._ W s . ye 'r o y v.a..,.. -y__. Y Y VM7-vv y -rNc-w v4 yyy a +,w.

i m.-== .M a = l 1 _P Fig.:4.1-6 -Microstructure of. cap screw D2A9 (tempered martensite) r-E Fig. 4.1-7 Microstructure of cap screw D2A11 (tempered martensite) ..e

p y b b m.*; r ~ ie s .w ,~( ~ + J I \\ h \\ I a f .i., a s M T m, k

9 t,

t', s:s.. i. i + + m i Fif]. 4.1 Microstructure of suction ~ pipescap screw S2Al-(tempered martensite) ~ + 1 f (h a a M A Z L .{ i e d T g a(-b ' r '-+..,.#,,,,, ,,,',p,,,,....,,w_,,_,,,,~..,,,,, gym,,..,,,.5,.

j. 4.2 Failure Mechanism '4.2.1 Diffuser / Case Connection The original diffuser / case connection employed [ ]boltstogether with _ _,long ring segments to clamp.the diffuser to the casing _ ledge. The original configuration employed close radial fits in threa places simultaneously: a) Ring Segment I.D. to diffuser lip 0.D. b) Ring Segment 0.0. to Casing I.D. c) Diffuser 0.D. to Casing I.D. Spherical washers were installed under the bolt heads and the bolts weretorquedto{ ] final torque. The multiple close fits are believed to have detracted from the preload actually stored in the bolts during assembly. The effect is similar to a bolted flange where a soft gasket is employed. As the gasket compresses plastica 11y due to thermal effects and vibrations, the bolt preload is diminished. Cyclic loads then are transmitted through the bolt and fatigue of the bolt is possible if the cyclic loads exceed the fatigue strength of the bolt. This failure mech-anism is consistent with the results of metallurgical examination of failed diffuser bolts given in Section 4.1. 4 During normal operating conditions hydraulic forces alone are sufficient to maintain the. diffuser in place even if all diffuser / casing cap screws were to fail. The total weight of the diffuser andsuctionpipeis( ) During four pump operation the upward hydraulic force is[_ ] Radial motion is prohibited by the mechanical fit of the diffuser. Thus, no diffuser movement would occur. I-20

y 4.2.2 Diffuser / Suction Pipe Connection The suction pipe connection of the original design also contained ambiguous or multiple load paths. For example, a gap was left under the suction pipe flange at assembly and the bolt force is then opposed by the friction forces along a []taperfit,ratherthanby a flat conventional flange surface. The most likely cause for the bolts becoming loose is that the suction pipe, in the presence of mechanical vibrations and oscillating flow forces, settled ever so slightly along the shallow taper fit, to the extent that the preload on the bolts decreased or disappeared entirely. At this point, fatigue loading of the bolts was possible. 6 9 9 e I-21

i .PART I DIFFUSER / SUCTION PIPE BOLTING 5.0 DESIGN MODIFICATIONS 5.1 Diffuser / Casing Connection )ap[ screws;thisdivesaboltthatis,, ] ({ cap screws have been replaced by The ~ ] longer. This was accomplished by deepering the threaded holes in the dif-fuser and by thickening the ring segments. The portion of the cap screw which stretches upon application of the torque has been . increased by roughly a factor of two. This increase in length plus a change in the final torque from ]whichplacesthe ~ ~ cap screws at[ ]of the minimum yield strength results in an approximate four-fold increase in cap screw elongation at assembly. Generally it is an objective in bolted connections to maximize the length of cap screw so that thermal effects and other factors have the least effect on bolt preload (see Figure 5.1-1). The increased number of cap screws provides a greater total clamping load. The ring segments were shortened such that they cover an arc of only 3comparedtotheoriginalsegmentswhichspanned[). Shorter segments give a more uniform loading and increase the effectiveness oftheclampingsurfaces(. ). The inner step on the diffuser has ber removed (see FIGURE 5.1-1). The diffuser was machined to be nomii. iy flush with the top side of the casing ledge. The new ring segments were also re-designed to match-the diffuser. The radial clearance between the ring segments and the casing (see FIGURE 5.1-1) has been increased. I-22 s . p.

c The role of the wedges and keys was re-evaluated for whatever effect they may have had on the pump problens as a whole. Basically, both the key and the wedge are useful in assembly of the pump internals . to the casing. The key provides guidance as each' diffuser half is positioned within the casing. The wedges are used to push the diffuser halves radially outward against the casing I.D. A second-ary function of the wedge was initially to form a bridge across the diffuser splits and to limit relative motion of the diffuser halves. It is shown by analysis in Section 5.3 that diffuser motion is not predicted in the final configuration. This analytical conclusion would be the same whether or not the wedges and keys were in place. Because.these devices are useful in assembly of the pump, and because they are trapped securely in place and represent no liabil-ity during operation, the wedge and key designs were left unchanged. The role of these two components is strictly one of facilitating assembly. O I-23

E" *:. ;-' '. - J 3 .(' 'rz-J 5 5 T t a b A.. s b t' %,n 5 h ' 8 7 1." k , q w 4 r A", s l '.1", D e 3 d s DIFFUSER / CASE CONNECTION FINAL DESIGN l J FIGURE 5~1-1 ~. ,I . 3' t,. O ,,.___...m

jfif -\\ 4., ^-n i. . 3 j,, ) I ~ 1 4 + u, J. n' a M + I s N w t: s A - i ) I .6 t s v s E FINAL CONFIGURATION, SYS 80 RCP FIGURE 5.1-2 4-Y' 1 ~ 4

e-l ~ 5.2

Diffuser / Suction Pipe Connection Theoriginal[

)capscrewswereretained,however,thequant-itywasincreasedfrom( ) The torque value was increased from( ]whichplacesthecapscrewsat[.)f theminimumyieldstrength(seeFigure5.2-1). The taper fit and the thin lip on the suction pipe flange have been eliminated in favor of a flush fitting conventional bolted flange configuration. This improvement eliminates the potential for shifting of-the parts once the parts are properly clamped together at assembly. Tapered dowel pins have been added, one pair straddling each of the two diffuser splits to further reinforce the diffuser at its lower end and to assist in re-assembly operations following routine pump inspections (see FIGURE 5.2-2). The suction pipe ring serves three purposes. First, it restores the elevation of the cap screw heads to the original elevation, thereby permitting re-use of the original length ( [ cap screw. Secondly, it adds hoop stiffness to the entire connection. Thirdly, it provides a sure means of trapping the tapered dowel pin. O I-24 1 m

T:u-r,

. r. ..a. N T i t 4 4 w 4 SUCTION PIPE / DIFFUSER BOLTED CONNECTION FINAL DESIGN 1 4 FIGURE 5.2-1 4 1 ' r L A-, ,.. + +,,, 4 - ,,, - ~,,. .mmn-..,.--,,,an.,.+,--ne-,~wre--n-ww-4--e,-mw.-~~~,wm

= 'fA we, Ja a a un -

%4

-m,- _ i n ^ ~, .x o I ,F T - g 1 6,, / ' v b_ 1 ~- .t[ "i 6 4 s f E a t 4 P s_ 1 4-F i s , e.. ~ SUCTION PlPE/ DIFFUSER PfNNED CONNECTION FINAL DESIGN Y r FIGURE 5.2-2 5 r (

1 i

[. :.,. s: :n ..n__-r,.. ...u e, ..r .,~.,,.c,.... .r.,.-,....-n.,....,,,,-, .<n-n,,.,...,-+..,, ..-~,,nn... ,,...,., - -. ~.. .,,,,...,,. - ~,

7 5.3 Analysis of Diffuser Case Connection An analysis of the new configuration of the diffuser to case connec-tion has been perfomed to demonstrate that the design is adequate. + In the analysis, all hydraulic loads acting on the diffuser and suction pipe are compiled and sumed to determine the clamping force that must be generated in the bolts to prevent diffuser motion and subsequent fatigue of the bolts. Besides insuring sufficient preload, the analysis shows that the bolt stresses are within limits and also that the pump case ledge stresses are below allowable values for hot and cold operation, as well as for transients such as heat up and cool down. 5.3.1 Loads on Pump Internals An objective in the design of all bolted connections is that the preload established in the connection exceed operating loads, such that tha bolt stress is constant, and at some known value below a limiting value. The loads include component weights, thermal loads, hydraulic steady state loads, and estimated dynamic hydraulic loads. Weights d .~ Themal Expansion 7 (a) Heat-up and cooldown can produce a ytemperature differen- ~ tial between the pump internals and the pump case ledge. (b) Normal operating temperatures and differential thermal expan-sion between case, bolt, and diffusers is considered. I-25 w

Hydraulic Loads Hydraulic loads are those associated with flow plus an addi-tional_,on load for conservatism. ' There are several contributions to the total hydraulic load. a) A lateral force acts on the suction pipe due to flow maldistri-bution caused by the inlet elbow. It acts in the direction of the inlet pipe run at an elevation one third of the way into the suction pipe. (b) The sum of vertically acting thrust loads results in a load acting upward against the bottom side of the case ledge. (c) There is a lift force on each diffuser vane. The force varies from vane to vane depending in part on its location with respect to the discharge. The forces on the vanes can be vector subined to find the net side acting force on each dif-fuser half. (d) A side acting force in the direction of the pump discharge is produced by flow generally moving toward the discharge. (e) A radially inward acting hydraulic force is generated due to pressure recovery of the flow as it passes through the dif-fuser. (f) Dynamic hydraulic loads are estimated based on experiments at KSB as plus or minus( )of the steady state loads at the pump blade passing frequency of h I-26

l 5.3.2 Adc<uacy of Bolt Preloads The above hydraulic loads were combined for each diffuser half into a moment and shear force at the case ledge. This moment and shear must be overcome by the bolt clamping force. The following table illustrates the adequacy of the bolt preload by showing that the friction force produced by the bolts exceeds the lateral ~1rce at the case ledge and that the restoring moment produced by the bolts about a theoretical pivot axis exceeds the overturning moment due to the hydraulic loads acting in a plane below the case ledge. Bolts are stressed to ]oftheminimumspecifiedyieldstrength, which corresponds to a torque of ]atassembly. Left Side Right Side ShearLoad(lb) FrictionLoadfromBolts(lb) OverturningMoment(in-lb) Restoring Moment from Bolts (in-lb) 5.3.3 Bolt Stresses Bolt working stresses on the order of 75-90% of yield are recommend-ed for stiff metal to metal joints and for bolts made from ductile material. The final Palo Verde configuration employs these fea-tures. I-27

A torque coefficient K, of is used, where T=KDF D= Major thread dia. = a F = -Load, lb. T= Torque, ft-lb. The bolt material is specified with a yield of ] minimum. Apreloadwhichdevelops[]oftheminimumyieldisselected. The torque which produces a stress in the shank equal to of yield is b ] 5.3.4 Case Ledge Shear Stress and Bearing Stress The Inconel 600 case ledge is clamped between the diffuser and the ring segments by the bolts. An analysis was performed to verify that the case ledge bearing stress and shear stress are both below allowable values. The allowable bearing stress is taken as the yield strength of the material. Allowable shear is 0.6 S,. The following table illustrates that the ledge is not overloaded by the bolt clamping load. Ledge Bearing Allowable Bearing Ledge Shear Allowable Shear Stress (psi) Stress (psi) Stress (psi) Stress (psi) J 1-e 5.3.5 Thermal Effects Thermal effects will cause changes in bolt preload, stress, and clamping force. An analysis was performed for steady state hot operating conditions and for thermal transients such as heat up or cooldown where the temperature of the pump internals may be higher I-28

or lower than the casing ledge temperature. Transient temperature

differentials of t

were assumed. The different coefficient of -s expansion properties between the case ledge and the pump internals and the change in Young's Modulus of the bolt material as a function of temperature, were factored into the analysis. The results are expressed in terms of multipliers to be applied to the original cold assembly values of preload and stress. ~ The following table gives the factors which results from the anal-ysis. Factor to be Case Ledge Bolt & Diffuser Applied to Condition Temperature (*F) Temperature ('F) Cold Value Heat-Up i Hot Operating Cooldown ' From.the Table, note that the preload in the diffuser / case connec-tion is reduced slightly during heat-up and steady state hot operat-ing conditions, and increased slightly during cooldown. The following tables summarize the thermal effects on bolt stress, clamping force, and ledge case bearing and shear stress. I-29

l l mus' e e Yb e O a=== n*RO I-30

l .PART I. DIFFUSER / SUCTION PIPE BOLTING 6.0 PERFORMANCE VERIFICATION .i 6.1 Test Facility Description The CE-Newington pump test facility was designed and built to perfonnance test centrifugal pumps of vertical, single stage, single ~ suction design in accordance with the requirements of ASME Perform-ance Test Codes. The facility is capable of testing pumps of this design driven by electric motors with a voltage of 13.2 kV and power ratings as large as 11,930 kilowatts (16,000 horsepower), flow capabilities of up to 11.7 m/second-(185,000 gpm), operating temper-atures up to 343 C (650 F), and pressures up to 17.2 MPaA (2500 psia). The primary loop consists of horizontal runs of 915 m (36 inch) diameter carbon steel, stainless steel strip clad pipe approximately 21,300 m (70 feet) in length joined by a trifurcated return elbow as shown in FIGURE 6.1-1. The two outside branches incorporate butterfly valves for flow control while the center branch contains a throttling orifice that establishes minimum loop resistance. In order to cover the complete performance range, more than one orifice plate is required. Flow through the loop is measured by means of a calibrated 915 m by 640 m (36 inch by 25 inch) universal venturi tube. Pressure and temperature sensors are located at various positions around the loop. The fluid in the loop is~ deaerated, deionized water. All of the test parameters measured are logged through an integrated computer system. This system provides for logging test data contin-uously on a magnetic tape for processing at a later time, as well as recording selectable parameters on multipoint or multipen chart recorders for instantaneous analysis. A log is maintained correlat-ing events to test running time. I-31

Cooling systems provide temperature control of the primary loop water and cooling water for the pump and motor. A portion of the primary loop water is circulated through a water to air heat ex-changer. :The minimum primary loop water temperature that can be obtained is a function of heat load and environmental conditions. The closed cooling water system temperature is controlled by an evaporative cooler. The minimum temperature is dependent upon heat load and environmental conditions. A closed system provides for control of a wider temperature range and water quality. A charging system is provided to control the primary loop pressure and provide water for seal injection. A water heater provides temperature control of the seal injection system. I-32 { l . r- \\

6 / BUTTERFLY VALVE (2) \\ f}[ j e nn ,y y f-m i uu uu 45 20" I. D.. l I 2 !P P 70 FEET ~ r = PUMP CASING ~ l 30,, I. D. DISCHARGE l g l [ r' k [(.' ir ~ l \\ ( l_ i \\ l v ] } /-- 36" x 30" CONC. RED-36"I.D. l (/ )l l 1 ..i l l VENTURI UCTION 30"I. D PUMP TEST LOOP l j i

N+@p+k o (([ g$ i?g IMAGE EVALUATION 4 pff l. 9/ TEST TARGET (MT-3) 4 g//,4y' &g<.y# 'h W ,,f + r, 1.0 dEM Da S l8 EE !? m u 18 1.25 1.4 1.6 = =, 4 150mm 6" lk++//// 4{ y>,,,,

////p y

A J

4h o

1. %@/

/////, $ IMAGE EVALUATION 4 m1 1A-E1 < 1-e > + 1.0 EBM Eu [@Ha \\ II E m HE u

1.25 1.4 1.6 150mm A

%e

/4

$[5,),,,,,

3$djh4 7 y A 1 u

6.2 Test Requirements The Reactor Coolant Pump was tested in accordance with ASME Perform-ance Test Code for Centrifug-1 Pumps to demonstrate mechanical and hydraulic performance and operability using one production-type j casing in the C-E Newington (CE-KSB) Test Facility. Four tests were conducted in the C-E Newington Test Facility to confirm that the design modifications as described in this report were adequate to prevent further failures. These four tests were identified as: TEST #1 Original APS PVNGS design. Flowrate tested:90-135% of design flowrate......... 50 hours TEST #2 Original APS PVNGS design. Flowrate tested:135-150% of design fit.wrate........ 100 hours TEST #3 Modified Design Flowrate tested:135-150% of design flowrate........ 150 hours TEST #4 Modified Design Flowrate tested:90-135% of design flowrate......... 30 hours During all of the above four tests, performance data was obtained to verify the hydraulic characteristics (dynamic head and flow rate) and confirm the mechanical integrity of the total pump with the modified structural changes described in this report. 1-33 n

d Additional data was also obtained with the instrumentation described in Section 6.3. -Tests-#1 and #2 were conducted to establish a baseline with the original design and to correlate the effects - identifiec after pre-core Hot Functional Tests at APS PVNGS Unit 1. 6.3 Instrumentation A fully integrated analog-digital instrumentation system was utiliz-ed for recording data during the test, retrieving the data after the test, and statistically reducing data for use in analyzing and reporting test results. The system consisted of three major compon-ents: instrumentation, instrument. signal conditioners, and a comput-er logger. The test loop instruments consisted of measurement sensors and ' transmitters which provided signal to the instrument signal condi-tieners. The sensors measured pressure, differential pressure, temperature, flowrate, speed, shaft displacement, velocity of frame movement, current, voltage, and electrical power. The signal conditioners converted signals to analog signals for control room indicators and digital signals for the computer. Signals from vibration monitoring sensors were directed to the following equipment: an oscilloscope, frequency analyzer with tunable filter, tachometer, and display stations. The digital portion of.the computer system was used as a data logger during testing, and a data reducer for' analysis after completion of testing. The computer continuously logged data on a magnetic tape recorder and provided data to a teletype, CRT display screen, trend plotters, and/or an X-Y Plotter. I-34

A The sensors were scanned every second by the computer and every fifteenth scan was' recorded on magnetic: tape through the entire - test..The operator was able to change the recording interval to every scan when areas of interest were identified. A running clock time, generated by the computer, was part of every ~ record on magnetic. tape for identification and correlation of records with the test event schedule. The following is a. description of the added instrumentation used in the Reactor Coolant Pump Test Loop for this study-(see FIGURE 6.3-1): One displacement probe was mounted in the pun $p casing at the area of the suction pipe piston rings in line with the pump discharge to monitor suction pipe movement. Two pressure transducers were mounted in the pump casing suction and discharge safe ends at the pressure tap locations to monitor pres-sure pulsations. Three accelerometers were mcunted in the diffuser at the split' located 180* from the discharge. One on the five vane half, and two on the six vane half. These monitored radial, tangential, and~ vertical' acceleration of the diffuser halves. A' biaxial and a vertical accelerometer was mounted on the outside of the pump casing j:-- to monitor pump vibration, i Five ring segment cap screws were strain gaged tr, monitor cap screw - loadings. These were (1) the cap screw on each side of the diffuser split at discharge and at 180*, and (2) the cap screw near the diffuser key on the six vane half. Each cap screw had a thermo-couple mounted to it and above it in the water to measure transient thermal conditions. Four thennocouples were mounted on the top of the pump casing flange outside the pressure boundary to monitor thermal distribution of -this surface. I-35 x-

F ~ 4 t .. g 2 ath a v QC 3 - a 3 O w w Z e .Q W W U J $ e A g , c o y w Y Zw W 3 2 N o 3 .s W In D

2. I

~ - Fig. 6.3-1

-c 6.4 - Test Procedure Tests #1:through #4, as described in paragraph 6.2.1 were performed . using the test loop instrumentation. 6.5 - Results and Evaluation The test results are in the process of being finalized for review ~ and evaluation. Preliminary test results indicate that the cap screw strains in the modified design were reduced by a factor of - and in some cases up to The vertical accelerations of the - diffuserhalveswerered'uce byafactoreff}Thehorizontal accelerations (tangential and radial) were reduced by a factor of at least')ndinsomecasesupto Hydraulic performance character-istics'(head / capacity) were confirmed as meeting the design require-ments. The final report will include a complete discussion of the test data and results. g - 6 O I-36

1 b i PART II DIFFUSER CAVITATION c e 11-1

E-1 PART II DIFFUSER CAVITATION TABLE OF' CONTENTS

1.0 INTRODUCTION

1.1 General Description of Pump Operation 1.2 Description of Problem 1.3 Safety Implications 2.0 ' SIM%RY 2.1 General Description'of-Inspections and Analysis of Results 2.2 Description of Design Modifications ~2.3 Design Analysis 3.0 INSPECTIONS AND EXAMINATIONS 3.1 Site Inspections .4.0 TESTS AND ANALYSIS 4.1 Model Tests at KSB 5.0 DESIGN MODIFICATIONS 5.1 Description of Change - 6.0 PERFORMANCE VERIFICATION 6.1 C-E Newington Tests II-2 .n

~.. .PART II DIFFUSER CAVITATION LIST OF FIGURES Fig.' No. Title 2.2-1 Observed Cavitation Damage After System Cold Hydrostatic Testing 2.2-2 Diffuser Vane Rework (6% Clearance) 5.1-1 Modified Diffuser Design LIST OF TABLES - Table No. litle 2.2-1 Key to Cavitation Damage -O 4 II-3 n

PART II DIFFUSER CAVITATION

1.'0 INTRODUCTION 1.1 General Description of Pump Operation Single-pump operation at las system pressure is inevitable whenever the reactor coolant system is cooled.down. Before system pressure can be increased the temperature has to be elevated to prescribed levels for minimum pressure and temperature (MPT) considerations.

It is this operation that can lead to cavitation either at the inlet to the impeller or at the diffuser. The net positive suction head required (NPSHR)_ curve provided for these pumps is a composite curve. At the four pump flow rate or design point the impeller controls the shape of the NPSHR curve (it is more sensitive to cavitation) whereas at the single-pump flow rate the diffuser conditions dictate. Diffuser cavitation is a very localized pheno-menon and any degradation would not affect subsequent design point operation. 1.2 Description of Problem Cavitation refers to a condition within the pump where, owing to a local pressure drop, cavities filled with water vapor are formed. . These cavities collapse as soon as the vapor bubbles reach regions of higher pressure on their way through the pump. Cavitation can appear along stationary parts of a hydraulic machine as well as along a moving vane. Isolated vapor filled cavitation bubbles that collapse adjacent to a stationary wall can cause local high pressures and velocities at the wall. The magnitude of the fluctuation forces on the wall are sufficient to erode any material at a rate depending upon material characteristics. II-4 v,en-m-~~ ,,ww,a

pu7 Cavitation of centrifugal pumps in pressurized water reactor coolant systems is not a problem during normal operation because the system pressure is considerably above that required to guarantee avoidance of the vaporization-cavitation phenomena during nonnal operation. However, there are short periods of operation in the start-up mode where the combination of localized high fluid velocities, with the low system pressure and temperature, can create the environment for cavitation to take place. Evidence of cavitation was first noticed on the APS PVNGS Unit 1 l diffuser vanes during the post-system hydrostatic test inspection in-November, 1982. The cavitation damage varied from a shiny surface indication to actual removal of material in a highly localized area of the diffuser tip leading edge at the junction of the upper shroud and the back side of the diffuser vanes (see Figure 2.2-1, Part II)~. The four APS PVNGS Unit 1 diffusers were again inspected after the completion of the pre-core Hot Functional Tests. Again, there was evidence of cavitation damage as described in FIGURE A-1. KSB model tests confirmed that the damage as observed after the pre-core Hot Functional Tests was primarily a result of localized high fluid velocities at runout flowrates combined with a narrow gap 'l between the diffuser and impeller vanes. 1.3 Safety Implications The diffuser tip leading edge cavitation erosion has been evaluated as having no adverse safety implications. Any particles generated ~ from the diffuser vane cavitation erosion would be microscopic in size and would not lead to core flow blockage. The diffuser cavita-tion observed at APS PVNGS Unit I was minor and no material was found in the reactor coolant system which could be identified as diffuser material. No degradation in pump hydraulic performance will occur with this type of cavitation nor will the structural integrity of the diffuser be affected. II-5

c k i. PART II DIFFUSER CAVITATION 2.0 ' SUMARY 2.1 General Description of Inspections and Analysis of Results ~ All four diffusers for APS PVNGS Unit I wore inspected by represent-atives from C-E Newington (CE-KSB) at the APS PVNGS site after system cold hydrostatic test. A detailed examination was made on the diffusers after the cold hydrostatic test and the results are contained in the following paragraph. Since the cavitation which occurred during the Hot Functional Testing was similar to, but not as significant as, the cold hydrostatic test cavitation, a less detailed inspection report was produced (see Figure A-1). The cold hydrostatic test related damage was identified on a scale of intensity from 10:1, with 10 representing no observed impact from cavitation (seeTABLE2.2-1). Each diffuser vane was observed and a notation documented as to the cavitation damage. The results of this investigation are shown in FIGURE 2.2-1. In addition, it was noted that the diffuser tip inlet vanes had not been profiled in accordance with the design drawing. Review of FIGURE 2.2-1 indi-cates that the majority of the damage occurred on Reactor Coolant Pumps S/N 1109-1A, 1109-18, and 1109-2A. The heaviest cavitation damage occurred on the vanes in the immediate vicinity of the ^ discharge nozzle in the pump casing. The Reactor Coolant Pumps have been operated at the site for both the hydrostatic test and the pre-core Hot Functional Test. The hydrostatic test was conducted at low temperature and pressure (210-250'F and 350 psig), conducive to cavitation, while the pre-core Hot Functional Test was conducted at normal system temperature II-6 n

m q and pressure (E65'F and 2235:psig). Pump operations for the hydro-static tests were in combinations resulting in runout flow rates from each pump in operation, also conducive to cavitation. Opera-tion for the pre-core Hot Functional Test was with normal flowrates (4 pumps) for approximately half of the test and runout flowrates (less than 4 pumps) for the other half. Based on the flowrates and inspection results the following can be ~ conclu'ded. (a) Cavitation damage was caused by localized high fluid velocities i at runout flowrate ccnditions. (b) Cavitation damage was not related to water or material quality. (c) Diffuser vane tip irregularities at the intersection of the upper shroud induced premature cavitation. 2.2 Description of Design Modifications After the cold hydrostatic test each of the diffuser tip inlet vanes were re-profiled in the immediate vicinity of the cavitation damage. T'iis brought the vane profile within the tolerance of the design drawing. The length of inlet vbne requiring re-profiling was approximately 19 m long. The lack of proper surface contour in this area led to an accelerated rate of erosion at this surface compared to the rate of a properly profiled vane, f KSB, in cooperation with C-E Newington (CE-KSB), initiated a hy-draulic model test program at the KSB Frankenthal Works, West Germany. The objective was to determine the necessary diffuser modifications that would be required to minimize the potential of cavitation over the design life of the Reactor Coolant Fump. i II-7 I m

L The KS8 model tests which were performed after pre-core Hot Func-tional Testing indicated that increasing the impeller vane outlet to diffuser inlet vane tip clearance from , of diffuser I.D. would significantly reduce the potential for cavitation in this wea as well as reducing pressure pulsations in the pump. This change has been made to the diffusers which were returned to APS PVNGS. V O II-8 m

PART II DIFFUSER CAVITATION KEY TO CAVITATION DAMAGE CODE

DESCRIPTION 0F CAVITATION 10 No observed damage 9

Between 8 and 10. 8 Beginning of discoloration, positive identification. 7 Between 6 and 8. 6 Heavy discoloration, area 1 /2 square inch. No mater-1 ial removal. 5 Heavy discoloration, signs of beginning material removal. 4 Between 3 and 5. 3 Material removal jL /8 inch diameter. 1 2 Heavy material removal 3:,1/4 inch diameter. 1 /4 inch diameter approxi-1 Heaviest material removal 1 r'ately 1/16 inch deep.

Correlation to FIGURE 2.2-1.

See values inside diffuser vane diameter. TABLE 2.2-1 n

n. PART II DIFFUSER CAYITATION 08 SERVED CAVITATION DAMAGE AFTER SYSTEM COLD HYDR 0 TESTING. Discharge heaale DischarTe Noaale Ariaene 1 Ar13ene 1 [b h e e% ( v' >Q g/, 'I \\ 6 g / i f \\. . 3,/ e .a ? O Vane # Of ffuser Split Line O Vane f (TopView) Pump $/ nil 092)s RCP Vane Orientationi - Of ffuser $plit Line 3 RCP Vane Orientation {Il09 2A (Top View) Pump $/N Discharge Nozzle Disenerge Mozzle Aritois 1 Art.lone 1 / 0 0 s \\, h, , g k /,6 g g i / .;G W: l0 \\ W= I i zg.~ / \\ g. .,./ ~ 3 A \\ Mo A.N%c f Q lane i --. 'Of ffuser Split Line .O ' lane f Offfuser Solit Lire 3 ~ J I ACP vaee Orientationi,109.lA RCPVereOrientation[I (Too view) P w o $/N (Top View) PWo $/N 10111 See TABLE 2.21 for Code. FIGURE 2.2-1

O e PART II DIFFUSER CAVITATION E M DIFFUSER IMPELLER f \\ 8 i / ? i,+-- GA P " B " i \\' I 4

=q@%

O!FFUSER VANE REWORK [( ] Clearance) FIGURE 2.2-2 n - -.

i 'l L PART II DIFFUSER CAVITATION 3.0 INSPECTIONS AJO EXAMINATION ~ 3.1 Site Inspections An inspection was conducted on the four APS PVNGS Unit 1 diffusers after system cold hydrostatic test at the APS PVNGS site by repre-sentativesfromC-ENewington(CE-KS8PumpCompany). The results of the detailed inspection are shown in FIGURE 2.2-1. The diffusers were removed from thcir respective pump casings and were positioned to permit a thorough visual inspection of all surfaces impacted from cavitation. The Reactor Coolant Pumps had been operated at the site to fulfill the requirements of the Hydrostatic Test. The hours of operation for each pump were: I PUMP HOURS 1109-1A 9.2 1109-18 17.4 1109 2A 7.8 1109-28 15.7 Based on information from site testing it was shown that 70% of the Reactor Coolant Pump operation was with two pumps (single-pump flowrates). Most of the Reactor Coolant Pump operation during this phase of testing was with the pump fluid temperature between 210*F-250*F with systempressureat350psig.(Iowsystemtemperatureandpressure). !!-9 n

- =

t. :L A visual inspection of the parts removed from each Reactor Coolant Pump indicated that the C-E requirement for cleanliness had been met. Areas such as the 0.D. of the suction pipe inside the pump casing had no indication of foreign matter which could cause cavita-tion.

In addition, the impeller wear rings showed no signs of erosion. The results shown in FIGURE 2.2-1 indicate that the majority of the damage occurred on Reactor Coolant Pumps S/N 1109-1A.1109-18. and 1109-27 It should be noted that in addition to the individual vane inlet tip profile deficiencies noted earlier in this report, that the diffuser half orientation ([ []Thetwodiffuserhalves are interchangeable physically, however, the preferred orientation isasinstalledl[ [} The impact on diffuser half orientation was not identified until the model tests i were conducted at K58. P b d 9 e ff.16

~ L.t PART II DIFFUSER CAVITATION 4.0 TEST AND ANALYSIS 4.1 Model Tests at KS8 A[ )scalehydraulicmodeloftheReactorCoolantPumpimpeller and diffuser was used by KSB to determine the design modifications required to minimize the potential of cavitation on the prototype diffuser vane tips and other hydraulic characteristics described in the other parts of this report. The hydraulic model tests pennitted through the use of an endoscope and strobescope, observation of the region on the diffuser vane surface that had been subjected to cavitation at APS PVNGS Unit 1. The formation and/or collapse of bubbles in this region could be observed and documented to provide a means of evaluating the effect of design modifications. The gap between the impeller vane outlet and the diffuser vane inlet wasincreasedfrom[ 3ofdiffuserI.D.andtestsconducted to determine the potential fce reducing cavitation damage. The hydraulic model flewrate was varied from the equivalent of 50% to 150% of the design flow. The increase of the impeller vane to diffuser vane gap from( ]of diffuser 1.0. demonstrated a significant reduction of the potential for cavitation damage on the diffuser inlet during opera-tion at the low temperature runout flowrate condition. Data was obtained on the model tests to determine the performance impact upon the Roactor Coolant Pump. Results from the model tests !!-11 m

t. a i

permitted predictions of the impact on the head capacity character-istics both at the design flowrate and at 150% of design flowrate conditions. Data from the model tests, combined with the observations of cavita-tion initiation, also permitted confirmation of NPSH requirements of the Reactor Coolant Pump with the increased impeller and diffuser vana gap. Analytical results from the model test are presently being translated from German and will be included in the final report. A test to confirm the design modifications will be conducted at APS PVNGS Unit 1. The results of this testing will be included in the final report. 0 0 11-12

l. t,. PART II DIFFUSER CAVITATION 5.0 DESIGN MOSIFICATIONS ? 5.1 Description of Chance The original diffuser vane tip diameter of( ]waschangedtoa diameterof[ ]andthevaneleadingedgere-profiledasshown in FIGURE 5.1-1. This change resulted in increasing the diffuser / impellergapfrom[ ] The results of opening the diffuser / impeller. gap to[ ]are(1)the surface area of metal on which earlier cavitation damage occurred has been removed and (2) the ASB model hydraulic tests showed a significantly less intense bubble formation and collapse activity. Y 6 O !!-13

PART II DIFFUSER CAVITATION FLOW DIRECTION ~ -y .C,l [ Radius h' - Material Removed N N N N. NNN N Diffuser Diffuser Vane Top View Modified Diffuser Design FIGURE 5.1-1 ~

6.0 PERFORMANCE VERIFICATION 6.1 C-E Newington Tests Examination of the full size impeller following the results of the C-E Newington tests, conducted with the modifications, indicated that there was no evidence of cavitation as a result of running Test

3 for 150 hours and Test #4 for 30 hours as described in Section 6.2 of Part I.

The diffuser inlet vane surfaces were thoroughly inspected with the diffuser halves still positioned in the pump casing. (A more detailed investigation will be conducted after the diffuser has been removed from the test loop casing, and these results will be docu-mented in the final report.) G 1 II-14 t e

-~- ~ ~. - ~ ~ - ~.. - ~. ~.. A o PART III O INPELLER FAILURES e we w O e III-1

PART III IMPELLER FAILURES TABLE OF CONTENTS ~

1.0 INTRODUCTION

1.1 Description of Problem 1.2 Safety Implications 2.0 SUMARY 2.1 General Description of Inspections. Testing. and Analysis of Results 2.2 Description of Design Modifications 2.3 Design Analysis 3.0 INSPECTIONS AND EXAMINATIONS 3.1 Site Inspections 3.2 Component Inspections 4.0 TEST AND ANALYSIS 4.1 Metallurgical Tests 4.2 Design Analysis 4.3 Experimental Stress Analysis 4.4 Model Tests 5.0 DESIGN MODIFICATION 5.1 Description of Modified Design 6.0 PERFORMANCE VERIFICATION 6.1 C-E Newington Tests !!!-2

PART III IMPELLER FAILURES LIST OF FIGURES Fig. No. Title 3.2 Comparison of Vane Thicknesses 4.1-1 Palo Verde Impeller B005 Pump No. 1B 4.1-2 Palo Verde Impeller A002 Pump No. 2A 4.1-3 Impeller B005 Close Up of Broken Vane No. 5 4.1-4 Impeller A002 Vane Nos. I and 6 4.1-5 Impeller A002 Close Up of Broken Vane No. 6 ~ 4.1-6 Impeller A002 Extended Crack on Vane No. 6 4.1-7 Impeller A002 Extended Crack on Vane No. 6 4.1-8 Sketch of Impeller Failed Vane Segment: Impeller B005 Vane No. 5 4.1-9 Sketch of Impeller Failed Vane Segment: Impeller A002 Vane No. 6 4.1-10 Location of Test Specimens: Impeller B005 Vane No. 5 4.1-11 Location of Test Specimens: Impeller A002 Vane No. 6 4.1-12 MicroStructureofVaneNo.5ImpellerB005(Tempered Martensite) 4.1-13 Micro Structure of Vane No. 1 Impeller A002 (Tempered Martensite) 4.1-14 Micro Structure of Vane No. 6 Impeller A002 (Tempered Martensite) 4.1-15 Zirconium Oxide Inclusions in the Micro Structure of Vane No. 5 Impeller B005 4.1-16 Montage of Vare No. 6 Fracture Surface 4.1-17 Montage of Yane No. 5 Fracture Surface 4.1-18 SEMPhotographofFractureSurface(Vane 5,ImpellerB005)with Fatigue Marks 4.2-1 Finite Element Mesh of Vane 4.2-2 Stress Contours in the Vane 4.2-3 Modified Vane Thickness Profile III-3

LIST OF FIGURES (Cont'd) Fig. No. Title 4.2-4 Effects of Thickness on Vane Stress (At Location No. 1) 4.2-5 Finite Element Model of Fillet Geometry Near Hub 4.2-6 Stress Concentration Factors at the Vane to Hub Junction 4.2-7 Fillet Radius vs. Leading Edge Thickness Relationship T O F III-4

,. _ ~ w PART III IMPELLER FAILURES LIST OF TA8LES Table No. T1tle 4.1-1 Mechanical Results at Room Temperature 4.1-2 Chemical Composition of Vanes No. 5, 1, and 6 s e a M m% 9 e' e O III-5

^^ ^ ~~ " ~ ' ^ ~ ^ ^ ~~~ . ~. -.. - PART III IMPELLER FAILURES

1.0 INTRODUCTION

1.1-Description of Problem The four (4) Reactor Coolant Pump impe11ers were inspected at the completion of the pre-core. Hot Functional Testing at the APS PVNGS Unit 1 plant. This inspection took place during the period July 16 through 22, 1983. A brief description of the inspection is shown in FIGURE A-1. Two impellers were found to have no broken or cracked vanes. These impe11ers were identified as S/N A003 (fr. stalled in Pump S/M 1109-1A) and S/N B006 (installed in Pump S/N 1109-28). Impeller S/N B005 (installed in Pump S/N 1109-1B) was found to have one vane with a missing segment from the leading edge. Impeller S/N A002 (installed in Pump S/N 1109-2A) had segments from two vanes missing. One of the broken vanes had a through crack at the vane to upper shroud interface. Impe11ers used on SYSTEM 80 pumps were supplied domestically from Atlas Foundry and Machine Company and from Fisher Cast Steel Produc-tions Inc. Foreign supply was from Schmidt & Clemens. The impel- .lers used in PVNGS were supplied by Atlas. Domestically supplied 'i-impeller material is SA-487, Class CA6MM which is a high strength cist martensitic stainless steel developed to give good castability. weldability, and toughness. Schmidt and Clemens impe11ers were cast fromWbl410G-X5CrNi13.4(1.4313) material which is the West German designation for SA-487 Class CA6NM material. III-6

~~~ ~ ^ '~ ^~ 1.2 f Safety Implications The design modifications described in Section 6.0 were selected to. eliminate the possibility of damage to the pump impe11ers. The-CE-KSB pump tests and subsequent inspection have demonstrated the adequacy of the design modifications. Therefore, it is considered highly improbable that impeller failures will occur during plant operation. The safety implications of the APS PVNGS impeller failures, if they had occurred while the plant was at power, with regard to possible blockage of a core channel by an impeller segment or degradation in pump flow rate are discussed below. ~ A review of the consequences of segments breaking out of the impel- ~ 1er vanes shows that core flow blockage would not have resulted. The vane segments which broke off the APS PVNGS impellers were too large to pass through the reactor vessel flow skirt and into the core. This is confirmed by the inspections performed at APS PVNGS which revealed that the broken segments were trapped in the bottom of the reactor vessel outside the flow skirt. In addition, the crack propagation found in the damaged impe11ers suggest that addi-tional segments, if they broke off, would also be too large to pass through the flow skirt. The effect on core flow of the pump impeller damage observed at APS PVNGS Unit I was minor since the undamaged pumps would have made un most of the difference by providing more flow. Therefore, no significant degradation in plant safety would have occurred. 4 4 k III-7 L

~ ^ .. ~. I PART III IMPELLER FAILURES 2.0

SUMMARY

2.1 General Description of Inspections. Testing, and Analysis of Results 2.1.1 Inspections The inspection conducted on the four APS PVNGS Unit 1 impellers was partially performed on site at Palo Veroe with a more thorough inspection beir.g conducted upon receipt of the impellers back at the C-E'Newington site. . Upon return of these four impe11ers to C-E Newington, the following inspections were conducted and are summarized in Section 4.0 of this part: (a) Non-Destructive Examination - 100% water washable dye penetrant examination of wetted surfaces. (b) Dimensional inspection of selected areas. (c) Visual inspection of all surfaces. t (d) Metallurgical examination of selected surfaceh and broken vanes. These impe11ers, at the time of inspection, had accumulated the cherationalhoursasshowninFIGUREA-1.forthepre-coreHotFunc-tional and Ccid Hydrostatic Tests. 2.1.2 Testing ind Analysis Concurrently with the inspections being carried out as described above, the following analytical efforts were initiated. 1 III-8 .I

^~ ~ ^ ~ ~~~ \\ I (a) Stress analysis of the impeller. 1 (b) Strain gage tests on the model impeller to determine vane ~ loadings empirically. (c) Analytical review of impeller resonances supplemented by . additional empirical model testing. -(d) Review and upgrade of impeller NDE acceptance criteria. 2.2 Description of Design Modifications The physical changes perfomed on the Schmidt and Clemens impe11ers ~ that have been shipped to the APS PVNGS Unit 1 site, consisted of backfiling the impeller vane exit surface, minor weld repair of indications that failed to meet the upgraded NDE acceptance criter-ia, and subsequent re-heat treatment, and a re-balancing of the impeller as required by the original specification. Backfiling is a standard pump manufacturing technique used to obtain small increases in pump head or flow and was used in this case to regain the head lost by modifications of the design, i.e., gap increase as discussed in Section II of this report. Each Reactor Coolant Pump impeller shipped to APS PVNGS has been inspected in accordance with a revised impeller acceptance criteria. The revised acceptance criteria identified areas of special concern on the impeller vane surface. As a consequence a more stringent NDE acceptance criteria was adopted for the high stress areas. 2.3 Design Analysis A de atled Finite Element Stress Analysis of the impeller was conducted by KSB in cooperation with the Stress Analysis Group within C-E Windsor. Some of the topics that were reviewed included: III-9

(a) Relative Stress Distribution. (b) ~ Point of Maximum Stress. (c) Influence of Wall Thickness. ~ The detailed analytical approach was supplemented with model tests using strain gauges at the appropriate locations to obtain empirical data on: (a) Influence of wall thickness. ~ s (b) Influence of flowrate. (c) Influence of Gap "B" between diffuser vane inlet and impeller vane outlet as shown on Figure 2.2-2 of Part II. (d) Influence of reverse flow. The conclusions reached after review of this design analysis is that all modified impe11ers returned to APS PVNGS have a safety factor of at least[ [ over those impe11ers that were originally used in APS - PVNGS Unit 1. This safety factor will assure satisfac. tory perform-ance of these impellers in service. s a e III-10

L1 c PART III IMPELLER FAILURES 3.0 INSPECTIONS AND EXAMINATIONS 3.1 Site Inspections The site inspection of the impe11ers consisted of a visual check and sketches of the areas with. missing segments. Detailed inspection of all impe11ers and parts was conducted after the impe11ers were returned to CE-Newington. 3.2 Component Inspections A 100 percent Liquid Penetrant (LP) water wash of the wetted sur-faces of the PVNGS Unit 1 impe11ers was performed at CE-Newington prior to the start of the dimensional inspections. The result of the water wash confirmed that there were no further cracks in these impe11ers. Dimensional inspections were conducted on the impe11ers originally supplied to APS PVNGS as well as some impellers provtjed, by a different supplier, for other SYSTEM 80 plants. This_ inspection-consisted of vane thickness measurements, vane-to-shroud radius measurements as well as thickness-measurements at the leading edge of the vane. FIGURE 3.2-1 shows the comparison of vane thicknesses along the length of the vanes in Atlas (A) and Schmidt & Clemens (S&C) impel-lirs. Thickness measurements indicate that Schmidt & Clemens vanes are thicker.than Atlas vanes near the leading edge. However, at i measurement locations further away from the leading edge, there is less difference in thickness measurements between the two manufac-turers. The dimensions of the Schmidt & Clemens vanes gives them a flatter profile. This flatter vane profile has a beneficial effect III-11

.on the loading experienced.by the vane. This effect combined with greai:er thickness near the leading edge in Schmidt & Clemens impel-1ers provides. increased vane strength in the area of high stress. The increased thickness of the Schmidt and Clemens impe11ers occur-red as a result of a preference by the foundry to cast thicker blades to facilitate improving the flow of metal into the casting ~ molds. This also gives an improved metal quality (less porosity and shrinkage). The thicker blades have been examined at CE-KSB and . accepted based on experimental data showing that the required head / flow characteristics could be obtained. It should be noted that although the Schmidt & Clemens impellers have a. thicker vane than the Atlas impe11eis, except for isolated cases, all Atlas vane thicknesses are within the minimum drawing requirements. d [ o l-III-12

O 9 4 9 .g. O O e -g' O FIGURE 3.2-1 O

I PART III IMPELLER FAILURES 4.0 TEST AND ANALYSIS ~4.1 . Metallurgical Tests This section describes the metallurgical investigations performed by - C-E Windsor. Additional investigations were carried out by KSB in West Germany and are still under review. The results of these investigations will be included in the final report. ~ 4.1.1' Introduction and Background The impe11ers supplied for Palo Verde Unit #1 were ordered to ASME . Specification SA-487 Class CA6HM from Atlas Foundry and Machine Company. The failed impe11ers with the Serial Musber B005 (Pug S/N 1109-18) and Serial Number A002 (Pump S/N 1109-2A) are shown in FIGURES 4.1-1 and 4.1-2, respectively. As can be seen, a segment of vane number 5 has broken off of impeller S/N B005 (FIGURE.4.1-1), and impeller S/N A002 (FIGURE 4.1-2) has two vanes, number 1 and 6 damaged. FIGURES 4.1-3, 4.1-4, and 4.1-5 are close up photographs of the damaged areas. There was also a crack along the fillet between vane number 6 (impeller S/N A002) and the shroud. This crack is hardly visible in FIGURE 4.1-5 (bottom view), but quite predominant on the other side of the fillet (see FIGURES 4.1-6 and 4.1-7). The defective vanes were cut out at C-E Newington and shipped to C-E Windsor for examination. 4.1.2 - Examination 4.1.2.1' Radiography ^ The three cut-outs received at C-E Windsor were first x-rayed for casting defects. The radiographs did not reveal any unusual degree ~ of inclusions, porosity, or shrinkage, but showed imperfections which can be oxpected in castings. III-13 h

'4.1'. 2. 2 Magnetic Particle Examination After cutting the three vanes to smaller pieces for easier handling i'iuorescent magnetic particle examinations were performed. No additional cracks were detected; the crack in the fillet of vane number 6 was found to be about[ ( inches long. Vane number 5 (impeller S/N B005) and vane number 6 (impeller S/N A002) were selected for more detailed examination. FIGURES 4.1-8 and 4.1-9 are sketches of these two vanes showing the location of porosity and shrinkage as revealed by radiography and the eittent of the crack in vane number 6. Also indicated are.the locations and extent of repair welds as reported in Atlas' quality assurance ~ -records. 4.1.2.3 Mechan'ical Tests The specimens for tension test were removed at locations as indi-cated in FIGURES 4.1-10 and 4.1-11. The results of the tests are listed in TABLE 4.1-1. The yield and ultimate tensile strength values are acceptable, but the values for elongation and reduction of area do not meet the minimum values specified in SA-487 for Class CA6MM casting. The much higher ductility values reported by Atlas can be explained by the fact that Atlas took the tension specimens f from a separately cast test block, a practice which is allowed by the material specification. Such a_ test block is much cleaner than a casting which has a complicated shape and usually contains minor shrinkage and porosity (sometimes not visible on x-ray films). However, those minor cast imperfections do reduce the ductility values in a tension test. 4.1.2.4 Chemical Composition Samples were removed from vanes number 5 (impeller S/N B005),1 and 6 (impeller S/N A002) for chemical analysis. The results are listed in TABLE 4.1-2. The chemical composition of both impellers is in agreement with SA-487 requirements for Class CA6MM castings and with the values reported by Atlas. III 4.1.2.5 Microscopic Examination The microstructure of all three impeller segments was checked on microsections removed at locations marked M in FIGURES 4.1-10 and 4.1-11, and was found to be tempered martensite as can be expected after a hardening and tempering heat treatment in accordance with SA-487 requirements for Class CA6MM material (FIGURES 4.1-12, 4.1-13,and4.1-14). ~ During this routine examination of the microstructure, microscopic-ally fine non-metallic inclusions arranged in a network form could be observed (FIGURE 4.1-15). These particles were identified to be zirconium oxides. Zirconium is used in steel making as a deoxidiz-ing agent, combining readily with oxygen, nitrogen, and sulfur. If the precipitation of these deoxidation products is confined to the former austenitic grain boundaries, forming a network, they can have a degenerative effect on the mechanical properties.of_the casting, especially.on ductility and toughness values. This could be another reason the ductility values shown in TABLE 4.1-1 are low. 4.1.2.6 Examination of the Fracture Surfaces The fracture surfaces of all three broken vanes were covered with a black oxide layer, and it was difficult to see any details which could clearly indicate the type of failure. At most locations, the fracture surface was smooth without secondary cracking or plastic deformation. This, in general, would suggest failure by fatigue. III-15

/ FIGURES 4.1-16 and 4.1-17 show the fracture surface of vane number 6 of impeller S/N A00i! and vane number 5 of impeller S/N B005. The crack pattern depicted in FIGURES 4.1-6 and 4.1-7 indicates that crack initiation occurred near the leading edge at the vane / shroud '~ intersection. Examination of the fracture surface, particularly on vane number 5, indicates that the crack initiated on the suction side. This is evidenced by the (a) flat fracture face on this side, (b) shear lips on the pressure side, and (c) longer vane to shroud crack length on the suction side. The fact that a slight shear lip exists at the tip of the vane fracture surface indicates that the crack did not initiate there, but instead at some location behind it. The damage to the fracture surface makes it impossible to locate the exact initiation point. The edges of the fracture surfaces were examined to determine whether any surface defects were evident in the general initiation area which may have caused the vane to fail. Although no surface defects were found on the suction side, vane number 6 (impeller S/M A002) did exhibit a shrinkage cavity on the pressure side which was i open to the vane surface. No direct evidence exists to indicate i that cracks started at this point. Vane number 5 exhibited no observable defects on either side, or for that matter, on the fracture surface itself when magnified up to 40 times The only l indication of any possible stress riser was the existence of some relatively shallow grinding marks. Only a few of these marks were oriented in a direction which could be associated with the fracture. I Although both the cavity and grinding marks could result in cracks starting at these points, it has not been demonstrated that they are the cause of failure. Applying the scanning electron microscope (SEM), fracture by htigue could be identified. Figure 4.1-18 is a SEM photograph (impeller S/N B005) covering a portion of fractured surface indicated with an arrow marked SEM in Figure 4.1-8. The presence of arrest marks [ confirmed that the fracture was by fatigue. Arrest marks were also (- found on the fractured surface of vane No. 6 (impeller S/N A002). I III-16 t

These two locations were the only ones with fatigue markings. Other locations, i.e. the intersection of the laading edge with the shroud, were damaged to a degree that SEM examination could not identify the type of fracture. 4.1.2.7 Repair Welds A review was also made of the extent and location of repair welds made on the failed vane Nos. 5 and 6. As verification of the quality assurance records, the areas adjacent to the fractures were macroetched to indicate the repaired areas. As shown in Figures 4.1-8 and 4.1-9 there were no fractures associated with-the repairs. j 4.1.2.8 Non-Destructive Examination Comparisons l A review was made of the NDE records for the four Palo Verde impel-1ers to determine whether any significant differences existed between the overall quality of failed and unfailed vanes. Specific-ally radiographic records and films were reviewed to determine i whether failed vanes had contained unacceptable defects or whether the failed vanes contained a disproportionate number of acceptable 4 indications. ~ The results of this review indicated the following: (a) In comparison to impe11ers S/N A002 and S/N B005, which exper-l ienced vane failures, impeller S/N B006 contained a larger number of discontinuities and did not experience failures. 4 (b) Comparison of radiographic indications on individual vanes showed that vanes containing the same type, number and severity i of discontinuities will not necessarily experience failures nor will they necessarily be immune to failure. 4 f i III-17

(c) Comparison of impeller vanes that failed to thosa that did not fail showed that'the failed vanes do not ccrrespond to the vanes with the greatest number or severity of radiographic discontinuities. (d) Relatively few indications were observed near the vane leading edge in the vane to shroud junction. The three failed Atlas vanes showed no indications in this region. Based on the original criteria no unacceptable indications were found. 4.1.2.9 Dimensional Inspection of Failed and Intact Impeller Vanes Dimensional inspections, consisting of thickness measurements and radius measurements, were perfonned on broken vanes number 5 (S/N 8005), number 1, and number 6 (S/N A002) and on nineteen remaining ~ intact vanes on all. four Palo Verde Unit 1 impe11ers located at CE-Newington(S/NA002,A003,B005,B006). Vane thickness measure-ments were taken at locations on the vane proper and along the crack path (where applicable). Radius dimensions were taken at the junction of the vane and the upper shroud where possible. These radius dimensions were taken for the first six (6) inches from the leading edge of the impeller vane which is the high stress area. The results of these measurements indicates that: T O mas e III-18

/ ( F ~ i ~ 4.1.2.10 Conclusions The conclusions of this investigation indicate the following: (a) The impeller castings met the requirements of specification SA-487, Class CA6NM. 4 (b) Although tensile specimens taken from the actual vanes did not meet minimum specification requirements for ductility (even though control samples from separate casting test blocks did. meet the ductility requirements) such disparity is not unusual for samples so taken. This effect is not contributory to the failure. (c) Radiography revealed only porosity, inclusions, nd shrinkage within the original acceptance criteria of the castings. (d) Fluorescence magnetic particle examination did not reveal any additional cracks. (e) Metallographic examination of the cast material and repair weld locations showed the microstructure to be tempered martensite as can be expected for this type of material. (f) Examination of the fracture surfaces indicated that the failure of the vanes was by fatigue. III-19

t - (g) Crack initiation appears to be on the suction side of the vane within the first one inch from the leading edge. (h) There was no evidence that repair welds were responsible for initiating fatigue cracking. ~ (1) Dimensional analysis of the four impe11ers indicates that the three failed vanes were the thinnest of the twenty-two vanes examined. (j) A shrinkage cavity was found in the vicinity of crack initia-tion on Vane No. 6. Inasmuch as a casting defect was observed on only one of the three failed vanes, it is suggested that poor casting quality was not the prime cause of failure. However due to the worn and damaged condition of the fracture surfaces the existence of such defects cannot be precluded. t 9 e 9 e h a b III-20 d .,-w-. r-v--y me -e me e --- r- -m-- , e -e -w w w ., w-r v e, w w w e-+-m e -,--*-t-- v e-- wev - -- ~+-m w - se. we-- h e -~ ew sme w-m= w om m -e -+ s, e

TABLE 4.1-1 - Mechanical Test Results at Room Temperature 8 e O ~. 9 9 F e

p S

m N w 1 = g-M e N O N 8 m M G C e Y O laJ %O

8

.M s= m m e= 4 0 m 4 W e 1 O J 4 J e

o Fig. 4.1-1 Palo Verde Impeller B005 Pump No. IB

4 Fig. 4.1-2 Palo Verde Impeller A002 Pump No. 2A

m ) i e i i 1 Y Fig. 4.1-3 Impeller B005 Close Up of Broken Vane No. 5 i

~ Fig. 4.14 Impeller A002 Vane Nos. I and 6

) P Fig. 4.1-5 Impeller A002 Close Up of Broken Vane No. 6

Fig. 4.1-6 Impeller A002 Extended Crack on Vane No. 6 i w.

9 ~ h Fig. 4.1-7 Impeller A002 Extended Crack on Vane No. 6

8 mm

- L b#

@ O WC &Z n@ @ 3m h W Qk C UT g MG Ge MP m 80,@ 6 6 L_ O ~ e S e SS w_

l l em NOO4@ .. b WO e 2ll .W W ' c. C Ee M lll> 6 *I ..M e w L&e o e O O e sin m

al' l l N'm p p'l'r-a '_r = pu a m R ' n 9' m u 'u 4 i aN pu 4 ' 'I p m Idi9mp i i Pa=mu -m.-um n u u h e

M g.W 8

=m (U O e== 2: e @c

6 M >

O e9 4 M e D r= La. e i eW \\ s 4' t +e O 4 9 -e se egene e

A e t e e OO eC W 5. @ O c= 2: F@@ I:L c E ses W> e-o P4 8 e=e i e o Y %'.\\ E - 9 1 4 W e J. i _cg e a ,,-,---n--,,, -.------m-m, e--- ,--~n- ..,,e--,e. +v----. ..e -,,n.r- ---,n--

D s s 4 -- ~ fig. 4.1-12 Micro Structure of Vane No. 5 Impeller B005 (Tempered ' Martensite) NY 6 e t 9 ,.,.-.,----.....-.,,.-,y.,--w,., y.c,,. ,e,r----e,-,_,-% n-.,,,-

. -.a. = -.. .P-Fig. 4.1-13 Micro Structure of Vane No. 1 Impeller A002 (Tempered Martensite). 9-u Eig. 4.1-14 Micro Structure of Vane No. 6 Impeller A002 (Tempered Martensite) .wwvm-w-=

~. _ _ _. ~ Fig 4.1-15 Zirconium 0xide Inclusions in the Micro Structure of Vane No. 5 Impeller B005 l l

w --- t 9 W A G-GU C4 @ %= > b'3 %= t/) C .a ! MU 86 1 W e.< 8 ew 4 Y l .h e- $4e b I D 4 e c y a +. - -

p. .u z,. - ~ -,... ...w a T e, O O .f. e-E. +W > b3 % 4A O

a..t M U.

b La. N@ f M e Y E e = .O M M

~ Jig. 4.1-18 SEM Photograph of Fracture Surface (Vane 5 Impeller B005) with Fatigue Marks v 4 m-wm,,-- 4 -r-.s...w..

l 4.2 f Desian Anal.ysis 4.2.1 Introduction Stress analyses were performed to better understand the impeller failures and to enable a basis to be established for assuring that new impe11ers would have an adequate margin against failure. The analyses were performed using the finite element program (ANSYS). The stress distribution in a vane which had thickness comparable to the vanes which failed was computed assuming a simplified uniform static load. The peak stress determined from the analysis was located at the failure region thus verifying the appropriateness of the model. Subsequent analyses of vanes with greater thickness near i the leading edge produced significantly lower stresses. Analyses of various sites of fillet radius between the vane and hub (upper shroud)regionwerealsoperformed. Combining the beneficial effect of greater thickness and the beneficial effect of p eater fillet j radius, an acceptance criterion was developed which enables the selection of impe11ers to meet a given safety margin compared to the failed vanes. 4.2.2 Finite Element Model of Impeller Vane ~ Figure 4.2-1 shows a projected view of the finite element mesh for an impeller vane. The edge representing the connection with the hub (upper shroud) is assumed to be rigidly fixed and the edge connected to the lower shroud is prevented from rotation but free to translate so that the hub and shroud can rotate relative to each other. A uniform pressure load is assumed to act normal to the vane surface. This load is intended to represent the dynamic hydraulic forces. Constant forces such as centrifugal forces are not considered since the cyclic fatigue loads have been identified by metallurgical examination to be the cause of failure. The stress distribution in the vane thickness and fillet radius similar to a failed vane resulting from the above loading is shown in Figure 4.2-2. The peak III-21

~ ^ ^ T ::: ^ ^ ~ T ^ ^ ~ ~ ^ ^ ' ^ 1 3 I stress is computed near the leading edge in the fillet area of the hub (upper shroud) connection.- This is the location of the cracks which lead to the vane failure. This coincidence indicates that the loading and boundary conditions used in the analysis are adequate to illustrate how changes in the vane will impact the stress. The same analysis was performed for vanes with different thickness and different fillet radii. The resulting stress distribution patterns were similar but the magnitudes were smaller when increased thickness or increased fillet radius were considered. 4.2.3 Evaluation of Effect of Thickness 1 The vane thickness was presumed to be thicker than the failed vanes for the first[ lbackfromtheleadingedge. A typical thicker [. vane profile is shown in Figure 4.2-3. These analyses led to the recognition that the thickness at station 1

nei back from the leading edge) was the most characteristic thickness for the deter-i 7

mination of the peak stress, that is, the thickness in this location could be used to categorize vanes as thick or thin. Figure 4.2-4 shows the effect of thickness at Station 1 vs. peak i stress in the vane for a given fillet radius and hydtaulic load. 4.2.4 Evaluation of Effect of Fillet Radius A finite element model of the fillet region is shown in Figure 4.2-5. This two dimensional plane strain model was used to deter-mine the additional concentration due to the actual vane to hub fillet shape. The effect of fillet radius on stress concentration is shown in Figure 4.2-6. l III-22

^ ~ ~ ^~ 4.23 Safety Marcin During the pre-core Hot Functional Test at Palo Verde, some impeller Vanes failed and some did not fail. The thinner vanes failed and the thicker ones did not. Therefore the stresses on the vanes must have been near the loading which would cause high cycle fatigue failure. The analysis of a failed vane results in a stress which can be scaled to the fatigue endurance limit by adjusting the magnitude of the assumed load. Using this load, the analysis of a vane with greater thickness and greater fillet radius results in peak stresses well below'the fatigue endurance limit. Reduction of the loading will of course also decrease the stress in the vane. Figure 4.2-7 shows the ~ thickness and fillet radius required to' provide a safety margin of ( [on the peak stress relative to the vanes which failed. Implicit in this relationship is the recognition that impeller / diffuser blade passing 1cadings have been reduced by at least[ - ] The actual force reduction can be seen in Section 4.3 to be[ '] This yields a safety margin even greater than( ] This reduction was accomp-lished by an increase of the impeller / diffuser gap as discussed in Part II of this report. Since the failed vanes experienced stresses near the endurance limit, any vane meeting this geometric criterion will experience stress well below the endurance limit and therefore will perform satisfactorily. 9 O e III-23

4 -1m 8.-d __ dweed ha,,,, ' * " ***W-== 3 9 e e 1 i G s e e a /' 4 7 FINITEELEMENTMESHOFViNE Fig. 4.2-1 ,...e,-- ,---..,,,,,,,.,-,--.n n___.._,

g,., , we ew... - ~ ~ ~ '~ G S S S 4 n Fig. 4.2 2. STRESS CONTOURS IN THE VANE

IC - l % %=.s .;w, pwoma em h, '..a.% .,wm+.ww.w.,%. .,,w-, r+ ,a 4 9 e e .J wW M M baJ ~ v oe E >== N cw n== b tO

== MI N. m k -s..._ e O L

T L. h O 5 e

9 ame S

FIG.4.2-4 EFFECTS OF THICXNESS ON VANE STRESS (AT LOCATION NO.1) 9 9

lrr =-- p9d

M.

g.v' h s 4 e 4 9 O

9

+ r i 9 9 e we 0 0 O e ~ Fig. 4.2-5 FINITE ELEMENT MODEL OF FILLET GEOMETRY NEAR HUB

a,M = m ( w e T 5p' s OW W 4 n. s i:5 E M m e.. of R m ND

v e t I . FILLET RADIUS VS LEADING EDGE fiffCKNESS RELATIONSHIP Fig. 4.2-7

p 4.3 / Experimental Stress Analysis In August 1983, an experimental stress analysis of the CE/K58 System 80 pump impeller was conducted on the model pump at KS8 in Franken-thal, West Germany in conjunction with the gap investigations conductad at the same time. For this investigation a pump model impeller having a dimensional scale factor of[ 7comparedto the prototype was prepareo for testing by thinning one of the vanes to[ ]mi and by mounting strain gages to both sides of the vanes near the leading edge at a position near the maximum stress posi-tion. { Wire leads from the strain gauges were routed out through the shaft of the impeller to a small radio transmitter mounted on the rotating shaft. The signals were sent by radio telemetry to a stationary receiver and power spectral density plots of static and dynamic strain were recorded as a function of gap size, rotational speed, vane thickness and coolant flow rate. ~ ~ ] The rules of scaling and structural analysis require the correcticn of this model to full scale pratotype as follows: = (h) [a,x(S.F.)2) ~- op o =stressontheprototype(psi) p o, =stressonthemodel(psi) 5.F. = Scaling factor =[ ] N,

Rotationalspeedofprototype(rpm)

N, = Rotational speed of model (rpm) =[ em me e !!!-24

~ In general, the conclusion of this experimental work is that the stresses in the impe11ers presently at APS PVNGS Unit 1 are below the endurance limit of the CA6MM casting material. 4.4 Model Tests In order to confinn the estimated impact on the Reactor Coolant Pump performance characteristics, caused by the design modifications described in Parts I and II, model tests were conducted at KSB, West Germany. These model tests provided the base data to determine the impeller outlet vane backfiling necessary to adjust the impel,1.e.r performance to meet the required head / flow characteristic of the SYSTEM 80 NSSS. These model tests consisted of: 0 (a) Testing with the diffuser / impeller gap (Figure 2.2-2 of Part II) at 2.3 and 6.0 percent as discussed in Part II of this report. (b) Testing with the model impeller Lackfiled to confinn the impeller would meet the required flow / head conditions. !!!-25

1 / The actual model test data.and results are presently being incor-porated into report format and a condensation will be included in the final copy of this report. The model hydraulic tests indicated that increasing gap "B" (Figure 2.2-2ofPartII)from2.3to6.0percentwouldresultinadecrease in developed head at the design flow rate. Backfiling the model impeller and ratesting confirmed that the performance characteris-tics at the designed operating point would be improved on the prototype impeller. Prototype Reactor Coolant Pump performance tests in the-CE-Newington test facility confirmed that the backfiled impeller did meet the ~ requirements of flow and head. ~ O 's e G III-26

I PART III IMPELLER FAILURES 5.0 DESIGN MDDIFICATION 5.1-DESCRIPTION OF MDDIFIED DESIGN Schmidt & Clemens impe11ers were returned to APS PVNGS. These impe11ers had the vane exit surface backfiled as necessary to obtain the required head / flow characteristics. This backfiling was neces-sary due to the impeller / diffuser gap increase discussed in Part II of this report. This gap increase also resulted in a reduction of impeller vane stresses. Each impeller was inspected and passed by a revised NDE testing criteria. This revised criteria now allows for no linear indica-tionsintheareaboundedby( 3fromthevaneleadingedgeand [ from the upper shroud. TI.is minimizes the presence of surface defects which otherwise could give rise to crack initiation. O . 0 o e 4, i r III-27

] / PART III IMPELLER FAILURES 6.0 PERFORMANCE VERIFICATION 6.1 C-E Newington Tests e A Schmidt & Clemens impeller was used for all re-testing of the baseline data and structural modification data in the C-E Newington -Reactor Coolant Pump test facility. The accuculated operating time on this impeller is identified below: Previous Cortract Testing 72 hours Test #1 - Design Flowrate 50 hours Test #2 - Runout Flowrate 100 hours

Test #3 - Runout Flowrato 150 hours
Test #4 - Design Flowrate 30 hours
Test #5 - Design Flowrate 21 hours TOTAL 423 hours
Tests #3 through 5 were with the design modifications installed as described in this report.

A' complete NDE examination was made (100% water washable dye penetrant examination of all wetted surfaces) on impeller S/N C003 prior to Test #1, after Test #2, after Test #4, and after Test #5. There were no changes in observed indications on any vane surface between before and after test NDE inspections. III-28

6.2 f Site Test A demonstration test is planned for Palo Verde Unit I to confim the adequacy of the repairs to the reactor coolant pumps under operating conditions which are similar to those during the pre-core Hot Functional Test. During the test, data will be taken for various -reactor coolant pump combinations at selected coolant system temper-atures and pressures. The maximum pressure for the test is 2250 psi and the maximum temperature is 565'F. r Based on the successful completion of this test and a completion of model and prototype test data analyses impe11ers for APS PVNGS Units 2 and 3 will-be modified to facilitate proper performance character-istics. 6 0 9 4 III-29 -.}}

ML20100M126

Contents

Text

1.0 INTRODUCTION

SUMMARY

1.0 INTRODUCTION

SUMMARY

SUMMARY

SUMMARY

SUMMARY

SUMMARY

1.0 INTRODUCTION

1.0 INTRODUCTION

1.0 INTRODUCTION

SUMMARY

Rotationalspeedofprototype(rpm)

Navigation menu

ML20100M126

Text

1.0 INTRODUCTION

SUMMARY

1.0 INTRODUCTION

SUMMARY

SUMMARY

SUMMARY

SUMMARY

SUMMARY

1.0 INTRODUCTION

1.0 INTRODUCTION

1.0 INTRODUCTION

SUMMARY

Rotationalspeedofprototype(rpm)

Navigation menu

Search