ML20235A423
| ML20235A423 | |
| Person / Time | |
|---|---|
| Site: | South Texas  |
| Issue date: | 01/05/1988 |
| From: | HOUSTON LIGHTING & POWER CO. |
| To: | |
| Shared Package | |
| ML20235A387 | List: |
| References | |
| ST-HL-AE-2461, NUDOCS 8801120240 | |
| Download: ML20235A423 (2) | |
Text
_ _ . _ _ _ _ ._
l l-EPRATA
- to l ST-HL-AE-2461 (Page 1 of 2)
- 1.Section I.A. page 6, 2nd paragraph 4th line:
"An additional test (ITEPO7-AF-0002) was successfully performed"
'Should read: 1 i
" Portions of the system operating procedure were successfully 1 performed" ;
I
- 2.Section I.B. page 11, 3rd paragraph, lith line:
"The tee was cut out and replaced."
Should read:
"The tee was cut out and will be replaced."
- 3.Section III.C.2 page 36 1st paragraph 7th line "The test was then rerun."
Should read:
"The venting procedure was then rerun."
- 4.Section III.C. page 37, Section 4:
"The test was run on November 25" Should read:
"The test was run on November 26"
- 5.Section III.C. page 37, Section 5:
"This test was initially run on November 25" Should read:
"This test was initially run on November 27"
- 6.Section III.C. page 38, Section 6:
"This test was run on November 29, 1987."
Should read:
"This test was run on December I and 5, 1987."
NL.LER87Di6 8801120240 880105 PDR ADOCK 05000490 p PDR
I
(
ERRATA to ST-HL-AE-2461 (Page 2 of 2) !
- 7.Section IV.A. page 67, 3rd line:
"are being rebuilt with new parts including the plug, stem, yoke and actuator."
Should read:
"are being rebuilt with new parts including the stem, yoke and actuator."
- 8.Section V.B. page 78, Section 3:
"has been installed on Trains A, B, and C" Should read:
"will be installed on Trains A, B, and C."
- 9.Section V.B. page 78, Section 4:
"The valve actuator has been fitted with a needle valve in the airline to increase the stroke times."
Should read:
"The valve actuator will be fitted with a needle valve in the airline to increase the stroke times."
- 10.Section V.B. page 79, Section 5:
the FCV's have been limited....This was accomplished....:
Should read:
the FCV will be limited....This will be accomplished....
Note: The work indicated in items 2, 8, 9, and 10 hac ..
for the most part been completed. Documentation verification is ongoing at this time.
I i
NL.LER87016
. 9
/
The Light
[
company '
P.O. Box 1700 Houston, Texas 77001 (713) 228 9211 Houston Lighting 8e Power December 24, 1987 ,
ST-HL-AE-2461 File No.: G02 10CFR50 U. S. Nuclear Regulatory Commission Attention: Document Control Desk Washington, DC 20555 South Texas Project Electric Generating Station Unit 1 Docket No. STN 50-498 Auxiliary Feedwater Hydraulic Transients
- The attached report describes a series of hydraulic transient events which have occurred in the South Texas Project Unit 1 auxiliary feedwater system beginning in early November 1987; the extensive testing program to determine the root causes of these events; and the corrective actions undertaken by HL&P to eliminate the causes of these events, repair damage to the system and return the system to operational status.
The hydraulic transients have been of two distinct types: Short duration O water hamer events and longer duration vibration events. Because of the nature of the earlier events it was first thought that air in the system was the cause of the water hammers. After installation of additional high point vents and a revision to the venting procedure, the air induced water hammer _
problem was resolved. However, the system continued to experience vibration events. Subsequent testing and investigation led to the identification of the ,
cause of the vibration. When the flow control valves from Train A and D were '
4 in a highly throttled position they were found to generate a pressure fluctuation, the dominant frequency of which (24 Hz) matches one of the acoustic (i.e., hydraulic) natural frequencies of the piping system. It was also found that a portion of the piping system has a natural frequency vibration mode at or near the fluid oscillating frequency. The system therefore exhibits a very rare combination of both hydraulic and structural resonance. As a result, the magnitude of the vibrations was large enough to cause the damage which occurred.
Vent assembly connections and pipe support damage caused by the hydraulic
, transient events have been repaired. Nondestructive examinations which were repeated following completion of these repairs and after completion of root cause testing have confirmed the integrity of the piping and supports. The permanent plant instrumentation, pressure transmitters, and flow transmitters are being recalibrates. The crossover valves, which were subjected to the highest vibration effects, have been disassembled and inspected. These valves have not yet been reassembled. They will be tested under full flow conditions upon reassembly to confirm the valves stroke within the specified time. The results of these tests will be included in a supplement to this report.
A Subsidiary of Houston Industries Incorporated 5145a/0185a b[
- gMto f 'l1 -
Houston Lighting & Power Company ST-HL-AE-2461
, ~ 'T File No.: G02 t _) Page 2 i
Physical modifications to the auxiliary feedwater system have been made.
These include:
a) The installation of positive stops on the flow control valves and adjustment of the limit switches so that these valves cannot be positioned such that flow would be less than 50 gpm. This prevents )
the valves from being operated at the highly throttled position that 1 generates the critical pressure pulsations. A modification of the valve will be developed and tested on Unit 2 to determine if the need for the positive stop can be eliminated. This modification will then be installed on Unit 1.
b) The air diaphragm operators on the Train A, B, and C crossover valves have been replaced using stiffer springs, c) The operation of the Train D crossover valve has been slowed down by the installation of a needle valve in the air operator, d) Five additional high point vents have been installed.
e) Modifications to a number of the piping supports have been made and f I,'} additional supports have been added to the crossover piping header to mitigate the consequences of any future water hammer event caused by V
equipment failure or operator action.
Confirmatory testing of the operability of the modified flow control valves and crossover valves has not yet been completed. These valves will be reassembled and testing conducted the week of December 28, 1987. Following completion of this confirmatory testing, HL&P will issue a supplement to this report to document the test results.
If you should have any questions on this matter, please contact me at (512) 972-8116.
C '
(J e M. R. Wisenburg Managor, Engineerin censing MRW/sig
Attachment:
Auxiliary Feedwater System Report: nyestigatio raulic Transient Events V
5145a/ 3185a
Houuon 1.ighting & Power Compans ST-HL AE 2461 O. File:No.: CO2 V Page 3 cc:
' Regional Administrator, Region IV Nuclear Regulatory Commission 611 Ryan. Plaza Drive, Suite 1000 l Arlington, TX 76011
- N. Prasad Kadambi, Project Manager U. S. Nuclear Regulatory Commission 7920 Norfolk Avenue Bethesda, MD 20814 Dan R. Carpenter Senior Resident Inspector / Operations c/o U. S. Nuclear Regulatory Commission P. O. Box 910 Bay City, TX 77414 J . 'R. Newman, Esquire Newman & Holtzinger, P.C.
1615 L Street,' N.W.
Washington, DC 20036 R. L~. Range /R P. Verret Central Power & Light Company P..O'. Box 2121 -l Corpus Christi, TX 78403 l R.. John Miner (2 copies) ;
Chief Operating Officer City of Austin Electric Utility i 721 Barton Springs Road .
Austin, TX 78704 l R. J . Costello/M. T. Hardt City Public Service Board P. O. Box 1771 San Antonio, TX 78296 Rufus S. Scott Associate General Counsel Houston Lighting & Power Company
+
P. O. Box 1700 'l Houston, TX 77001 I Revised 11/20/87 O L2/NRC/f
O =
AUXILIARY FEEDWATER SYSTEM REPORT I
l l
l INVESTIGATION OF HYDRAULIC TRANSIENT EVENTS l
SOUTH TEXAS PROJECT DOCKET NOS. STN 50-498, STN 50-499 HOUSTON LIGHTING & POWER COMPANY I l
l DECEMBER 1987 e'W d
' SOUTH TEXAS PROJECT.
[]
's.
~.-
AUXILIARY FEEDWATER SYSTEM REPORT jl TABLE OF CONTENTS i EXECUTIVE
SUMMARY
INTRODUCTION I BACKGROUND AND DESCRIPTION OF EVENTS A. . Chronology of Recent. Events B. Fracture Analysis and NDE 11 SYSTEM DESIGN BASIS A.- Description and Design Basis of System III INVESTIGATIONS TO ESTABLISH CAUSE OF HYDRAULIC TRANSIENTS- l l A. Design Review'of Water Hammer Potential B. . Review of Pre-operational and Hot Functional Testing
- 1. Pre-operational Testing .l D
d 2. Pre-core HFT Operation .
- 3. Post-core HFT Operation
~C. Additional Testing Performed to Isolate Cause of Transients D. Review of Potential Causes of Hydraulic Transient Events
-Causes
- 1. Data from Other Plants
- 2. Fast Valve Operation
- 3. Entrainment of Non-Condensible Gases
- 4. Pump Instabilities
- 5. Check Valve Closure
- 6. Pump Trip
- 7. Steam Void Collapse
- 8. Review of Valves in System
'O i
0148i/0005i
SOUTH TEXAS PROJECT
N AUXILIARY FEEDWATER SYSTEM REPORT N] l TABLE OF CONTENTS (Dant'd)
IV SAFETY SIGNIFICANCE A. Impact of Fluid Transients on System Integrity V CONCLUSIONS OF INVESTIGATIONS A. Root Cause
- 8. System Modifications C. Confirmatory Testing APPENDIX
- 1. Piping and Instrumentation Diagram
- 2. Piping Isometrics
- 3. Valve Drawings
/5
\s l
!s -
j ii 01481/00051
l SOUTH TEXAS PROJECT.
AUXILIARY FEE 0 WATER SYSTEM REPORT i EXECUTIVE
SUMMARY
This report describes the hydraulic transient events which have occurred on-
- the South Texas Project Unit 1. auxiliary feedwater system. The events began j in early November, 1987 while'the plant was going up into Mode 3 in
. preparation for pre-criticality testing. The hydraulic transients resulted in some damage'to the piping:and supports. During the course of testing to determine the cause of the damage it was determined that air was trapped in the system. Additional high point vents and an enhanced venting procedure' have solved this problem. This was not the cause of.the damage to the system. The remainder of the transient events caused sustained vibration.in
-the system. A series of tests established that the vibration occurred only when the Train A and D flow control valves were in a highly throttled, near-seat' position. The geometry of the internal parts of these two particular valves created pressure pulsations at'a frequency (24 hz) that J matched one of the hydraulic frequencies of the piping system. This problem.
has been resolved by installing mechanical stops on the flow control valves to
. prevent them from being operated at the highly throttled position that generates the critical pressure pulsations. The investigations'have also -
i concluded that the hydraulic transient events have not degraded the integrity
. of the. system components; the system is ready to be put back into operation..
l I
l l
l 1
pG i 1
- 01481/00051 1
I
SOUTH TEXAS PROJECT
(]
G AUXILIARY FEEDWATER SYSTEM REPORT INTRODUCTION The purpose of this report is to document the events on the South Texas Project which have caused several hydraulic transients in the auxiliary feedwater (AF) system. The transients have been of two distinct types: short duration water hammer events and longer duration flow induced vibration events. This terminology will be used throughout this report to differentiate between the two types of occurrences.
Section I of the report includes a chronology and brief description of the events which have occurred thus far including a summary of the damage sustained by 1-inch vent connections, instrument taps and pipe supports; and a discussion of the non-destructive examinations and metallurgical evaluations performea which then led to repair work.
Section II of the report provides an overview of the auxiliary feedwater system. The narrative covers the basic design functions of the system and j performance data for the major components. This section includes a schematic diagram of the system and a layout drawing showing the relative elevation of major components and piping.
Section III of the report discusses the investigations which have been underway to establish the cause of the transient events. There is a
( discussion summarizing how the system was designed and analyzed in recognition
( of the potential for hydraulic transients. The method of operating the system during modes 3 and 4 is compared to the system operation during pre-operational testing and hot functional testing to illustrate why the problems were not discovered in this earlier testing. Because of the nature of the early events, it was first thought that air in the system was the cause of the water hammers. After the installation of additional high point vents and a revision to the venting procedure, the water hammer problem was resolved. However, the system continued to experience. vibration events.
Section III covers the testing programs which have been executed to collect data on the hydraulic response of the system, and a description of the investigations leading to the identification of the cause of the vibration phenomena. '
Section IV of the report covers the safety significance of the hydraulic transient events in the context of potential failure of the system due to either an acute dynamic event or longer term cyclic fatigue.
The report ends with Section V which discusses the conclusions of all of the investigations. It has been found that when the A and D Train flow control valves are in a highly throttled position they generate a pressure fluctuation, the dominant frequency of which matches one of the acoustic (i.e., hydraulic) frequencies of the piping system. It was also found that a ;
portion of the piping system has a natural frequency vibration mode at or near I
v 2-01481/00051
the fluid oscillating frequency. Therefore, the system exhibits a very rare combination of both hydraulic and structural resonance. As a result, the
()
's._/
magnitude of vibrations is large enough to cause the problems which o'ccurred.
Additional supports are being added to the crossover piping header to mitigate the consequences of any future water hanmers caused by equipment f ailure or operator action. This section discusses other modifications which have been made to the system and describes the confirmatory tests that will be performed during Modes 4 and 3.
/ \
%_/
1 I
l
/s \
v) i 01481/00051 1
l
l O
l SECTION I BACKGROUND AND DESCRIPTION OF EVENTS 1
l 1
O
~4' 01481/0005i !
I
I.A. Chronology of Recent Events h
d This section provides a chronology of the hydraulic transient events which have occurred since early November,.1987. The transients have been of two distinct types: short duration water hammer events and longer duration vibration events. Figure I.A.1 shows the location of damage to the system as described in the following text.-
On November 5, 1987 a 1" vent line with two vent valves (AF0188 and AF0187) broke off the auxiliary feedwater pump discharge line in -
Train A. The vent assembly fractured at the nipple between a socket weld elbow and the first vent valve. The plant was in Mode 4 and the l l system was in operation to support steam generator blowdown testing.
On November 8,1987 a second failure occurred in a double valve instrument tap (AF0016 and AF0018) for the Train 0 flow element. The failure was between the sockolet on the main header and the first instrument root valve. Flow conditions of.the system were similar to
'those on November 5.
Based on failure analyses described in Section I.B the potential cause of the f ailures was determined to be f atigue cracking caused by cyclic forces during system operation. As a result, a vibration test program was developed (lTEP07-AF-0001). As described in Section 111.0 of this report, the test was intended to determine if steady i state vibration was sufficient to cause the failures and to identify the driving mechanism for the vibrations.
The steady state vibration test program was initiated on November 14,
.hd 1987. Shortly after the test was started, a cracked anchor (AF-1013-HL5002) was found downstream of the Train A crossover isolation valve (FV7517). A temporary support was installed near the cracked anchor and the system was inspected for any additional damage. No other damage was found and the vibration test'was
. continued. On November 15, 1987, while continuing the vibration-test, a water hammer occurred in the system. The water hammer occurred when the crossover isolation valves from Train A to Train D (FV7517. and FV7518) were being opened. The valves were being opened while the Train A pump was running but with no forward flow path established. An inspection of the system determined that no damage occurred as a result of the water hammer and the test was continued and concluded.
As a result of the November 15 water hammer which occurred when the crossover valves were being operated, the investigation then focused on the potential for air being trapped in the system. Venting procedures were reviewed and unvented high points in the crossover line were,1dentified. On November 18, it was decided to add five high point vents to the system. Proof tests were performed to quantify the air entrapped in the system prior to modification.
01481/00051 1
_ _ _ - - _ - - - -_- -. - - a
I Thgadditionalairventgdthroughthenewventswasapproximately1.5 ft , 0.5 ft3 and 1.0 ft for Trains A, C and D, respectively.
,\ (~T It was concluded that this volume of air in the system was sufficient
() to cause a water hammer. On November 19, a second water hammer was observed while opening the Train A and 0 crossover isolation valves .
(FV7517 and FV7518). As in the previous instance, the valves were l being opened under no flow conditions (i.e. the pump was running but no forward flow path was established).
l The venting procedures were then modified to add a dynamic sweep of the crossover line to assure that all air was removed from the system. Proof tests were again performed to verify that the dynamic sweep enhanced the overall venting process. An additional test (1TEP07-AF-0002) was successfully performed to demonstrate that a water solid system would not experience a water hamer due to the crossover isolation valves being opened under no flow conditions. At 3 this point the system was returned to service and the plant entered '
into Mode 3 on November 21, 1987.
On November 22, 1987, shortly after entering Mode 3, sustained piping vibration in Trains A and C was observed and reported by personnel in the Isolation Valve Cubicle (IVC). Subsequent walkdowns identified a cracked drain valve connection on the crossover line to Train A at a drain valve AF0238. In addition, pipe anchor AF-1013-HL5002 (the _
same anchor that was found cracked on November 14) had developed a '
crack in its embedment plate and a broken strut pin was found on pipe support AF-1047-HL5002 on the Train C crossover piping. ,
- (7 On Novenber 25, 1987 test ITEP07-AF-0004 was performed (as described u) in Section III.C) to verify that there were no substantial unvented pockets of air in the motor driven pump suction lines. In addition, test ITEP07-AF-0003 was performed to determine if the pump involved in the November 22 event had a characteristic curve which could cause surging. Also tested was the interaction between the pump, the auto recirculation control valve and the crossover isolation valve. A short-duration water hammer was experienced when closing the Train C crossover isolation valve at 650 gpm flow As described later in this report, this event was most likely caused by insufficient spring force in the air operator to maintain the valve in its closed i position. l Subsequent tests (ITEP07-AF-0005 and 1TEP07-AF-0007) were performed to reproduce the sustained vibration which occurred on November 22.
The multiple steam generator feed test induced sustained vibrations !
on December 6 and again on December 9. In both cases, the vibration occurred while one pump was running (Train B), the crossover isolation valves were open (on December 9 the C and 0 valves had been closed but were just opened) and the flow control valves were highly throttled at their neat-seat position. A review of test data showed that the system was being subjected to a sustained 24 hz vibration. j l
im
(
01481/0005i i
, ,, 1 q.
A series'of tests was performed on December 12 to determine which.
. . , , component was producing the 24 fiz" excitation. The test
((N '
systematically isolated the source of_ excitation to be the T' rain A
.d.-
. flow control valve ~ in a near-seat (highly throttled) position coupled -
with the Train D flow' control valve'in a near-seat position. The test also'showed that the Train B and C flow control valve throttling position, the pump being used and the crossover valves had-no effect-on.the resonant condition. The resonance was terminated when the Train A and D flow control valves were lifted off-their near-seated positions. i I
i d I 01481/00051 l
1
l' AFST
! s '~~ ' - 500,000 GAL. n
~
[h v >
FV7Sl8 JL p
AFP 14 W-i
'TRAI o
n l
D y k FV7517 3, e NOV
^
AFP ll JL
/
i
'TRAl p
0 u
Ch y k FV7516 3,
h JL '
1 y
AFP 12 W- 'TRA o
e >+4 cs g b FV7515 AFP 13 JL E'
'TRA!
y
bh X
F T .-
-X ,c NOV 8 M
V M
/
8 D' FV7526 X >W [
N : sE io o
($0M !=- M 4 iY N
L r [ NOV 14 & 22 i FT rh
' ' NOV-22 Y @ 5 6 A' FV7525 X >W
[
N = 18 m a
i; " 5 > 4 %
a X
FT E
y @ R 4 B' FV7524 r >W T N =18is ll " 5 >T 4 iW u
X
/ e NOV 22 FT 7J y s s pC' FV7523
= >W T N = s8 ic
- o t > < +
L LOCATION OF DAMAGE TO 3 PIPING AND SUPPORTS ..
h nn - b
l l
I.B. Fracture Analysis and NDE
(~T Fracture Analysis O
Several failed parts, three pipe nipples and one pipe anchor, were sent to Southwest Research Institute (SWRI) to perform f ailure analysis. The three pipe nipples were identified as samples No. 1, 2, and 3. Samples 1 and 3 were complete fractures, and sample 2 had a major crack approximately around 180 degrees of the nipple. The pipe support exhibited two major areas of cracking which were examined.
Sample No. 1, the complete fracture at vent valve assembly AF0187 (which was found on November 5), exhibited certain fine-scale features on one side that are consistent with fatigue propagation followed by overload fracture of the remaining ligament. (Figure I.B.1) The occurrence of fatigue on the opposite side is unable to be determined due to post-fracture damage. The cracking initiated on the outside surface of the nipple at the toe of the weld, in the heat affected zone, but then immediately propagated in the base metal.
Sample No. 2 from valve assembly AF0238 (which was found on November 22) exhibited identical distinct flat smooth zones as the other two samples.
(Figure I.B.1) The opposite side fracture zone was much smaller than was seen in sample 1 and 3, due to the pipe not being completely fractured.
The cracking was initiated on the outside of the pipe near the toe of the weld, in the heat affected zone, then immediately propagated through the base metal.
O Sample No. 3, also a complete fracture at instrument tap valve assembly Q AF0018 (which was found on November 8), exhibited a distinct flat smooth zone on one side with features characteristic of fatigue crack initiation and propagation. (Figure I.B.2) A second, smaller fatigue zone was present en the opposite side of the pipe, with the remaining areas exhibiting overload fracture characteristics.
A chemical analysis was performed on samples 1 and 2 to verify that the pipe nipple materials conform to the compositional requirements for ASTM A-106 Grade B (identical to ASME SA-106 Grade B). The results indicate that the nipples meet all the requirements of A-106 Grade B pipe.
The cracks on pipe anchor AF-1013-HL5002 (which was damaged on November
- 14) indicate that the primary cracking mechanism at both locations was fatigue. In one area, full penetration of the wall of the box beam had occurred over a substantial distance. (Figure I.B.3 and I.B.4) In the case of the stanchion to box beam connection, two distinct cracks were present, located approximately 180 degrees apart. All the cracking exhibited areas of flat smooth zones indicating fatigue crack initiation. In both areas, the cracks were located mainly at the toe of the weld.
o
-9 01481/00051
Based on all the examinations, the cracking mechanism in all the samples is f atigue crack propagation. The fatigue cracking has all O the indications of being caused by a cyclic force. Samples 1 and 2 U exhibited fatigue striations at a few isolated locations, and such striations may be considered to represent the extent of crack propagation in one load cycle.
Based on the spacing of striations and base metal thickness, a total number of cycles were calculated in the range of 7000 to 16000 cycles (for samples 1 and 2 respectively). This is a very approximate method of cyclic calculation but does indicate th6t the values of propagation data and cycles classify the fractures as low-cycle fatigue. Also, through all the examinations, the initiating point of the crack is near or at the toe of weld, in the heat affected zone.
However, no evidence was found in any of the samples that indicate welding technique of the welds, themselves, were defective.
Non-Destructive Examinations After the piping and hanger failures, additional non-destructive examination (NDE) testing was performed on similar type installations to assure further operation would not result in additional failures.
The NDE testing was performed in stages dependent on the assembly configuration which failed (double valve assembly, pipe support, or single valve assembly). The various methods of NDE testing were ultra-sonic (UT), liquid penetrant (PT), and magnetic particle (MT) dependent on base material.
O Seventy-six dcuble valve assemblies were tested using the ultra-sonic Q (UT) method. This examination was performed on all double valve assemblies af"er the second nipple failure was detected on November 8.
The basis for performing the UT was the preliminary hypothesis that the root cause of the failure was from induced vibration in the nipple due to the configuration being an unsupported double valve assembly. Ther* fore, single valve assemblies were not examined at that time. In order to ensure no additional cracking had occurred it was determined that UT testing would be performed.
The area of UT examination was all welds and base metal within 1" of the welds, between the branch connection and the first isolation valve (See Figure I L 5). Two double valve assemblies did not have sufficient clearance to allow the usage of the UT equipment and did not have UT examination pceformed. No other similar surface cracks were identified in the remainder of the assemblies, however, seven exhibited other rejectable indications. All nine of these assemblies (the seven with rejectable indications plus the two that could not be examined) were reworked and new nipples were installed, l
i O
01481/0005i i
(
[:
During.the above time frame, on. November'22 it was observed that a pipe anchor.(AF-1013-HL5002) on the Train A crossover line had visual N.
cracks'in the support base metal. -This visual observation was further confirmed by MT examination. Because of the damage to the anchor, it was decided to perform MT examination on two high stress points in Train A.(the tee and elbow on either side of valve.AF0041) and'on'eight supports. The NDE results indicated no surface cracking to be present.
The November 22 trant,1ent event also resulted in damage to a single-valve drain assembly (AF0238) in Train A. A. review of the system identified thirty-one additional single-valve assemblies.
These thirty-one were either PT examined or NT examined based on the L-base material, stainless steel (PT) or carbon. steel (MT). The results of this testing indicated three additional indications, two similar cracks and one weld defect. All three assemblies have been l reworked.
l l
To assure that no cracking occurred during the testing programs, South Texas Project has performed surf ace NDE on all assemblies which previously had been examined. In addition, all crossover pipe fittings in Trains A, B, C and D were surface examined. The scope of the examinations covered:the common crossover header back through each of..the four. flow control valves to the connection at the individual pump discharge header. No rejectable indications were found in any of the post-testing examinations except for a defect in a tee connecting the Train A crossover piping to the common crossover head. This tee is just downstream of the anchor (AF-1013-HL5002) that had developed unacceptable cracks. The tee was cut out and O- replaced. ~ Subsequent. examination by SWRI indicates the defect was a manufacturing defect (forging lap with some inclusion) and the tee could likely nave been repaired in' place.
J' 01481/00051 l
' POSSIBLE FATIGUE !
UNCERTA I N DUE TO I
' ' (') POST FRACTURE DAMAGE i
NJ l'
+
OVERLOAD OVERLOAD j L FATI GU E I
(A) SAMPLE NO.1
(
(^g 7FATIGU E V
O04(N N ^
LABORATORY FRACTURE I
N FATI GU E
' (B) SAMPLE NO. 2 FIGURE I.B.I FRACTURE SURFACE O- ,
^
O l SMOOTH FATI GUE CRACK PROPAGATION ZONE FINAL OVERLOAD FRACTURE j
' + r yHIGHER PROPAGATION-RATE FATIGUE ZONE SMOOTH FATIGUE CRACK PROPAGATION ZONE FRACTURE SURFACE OF SAMPLE NO.3 FIGURE I.B.2 O- 1 l
l__ _- - - _
. \
l WELD REMNANT
/c # ,A --
TUBE STEEL PLATE
'K WELD REM NANT
/" .
FIGURE I.B.3 PI PE ANCHOR AF-1013-HL5002 0 1 l
SMOOTH (FATIGUE)
O
+ <
NOTE: FRESH 1.ABORATORY -
THIS REPRESENTATION
.~lS SCHEMATIC. ACTUAL ROUGH (PROBABLY
' FRACTURE IS THROUGH -
RAPID FATIGUE)
COLUM N WALL AT-ANGLE TO WALL SMOOTH (FATIGU E)
(A) STANCHION / BEAM JUNCTION F '
}
FRESH l LABORATORYS L
FRACTURE %
l VIEW AWAY FROM WALL g FATIGUE O
_s L co BEAM / WAu auNc-
.ag FIGURE I.B .4
1, I" ;-
e-L , ,, _ n -
t...
A VALVE d
u ,,
-U.T. EXAM. ZONE
[""}
\ l U Q" APPROX.--TY P.)
() - . )
( 0 O
- l' '
I C O
+ n.n VALVE i
MAIN HEADER FlGURE I. B.5 U .T. EXAMI N ATION SCOPE l
O I
O SECTION II SYSTEM DESIGN BASIS O
1 01481/00051
1 l
II.A. Description and Design Basis of System
()
V The Auxiliary Feedwater System provides feedwater for the removal of reactor core decay heat when the main feedwater supply is not 4 available. In addition, the system is designed to function during '
i plant startup to fill the steam generators and maintain the required water level. A non-safety related startup feedpump is normally used to maintain steam generator level using the main feedwater bypass control valves with a tie-in to the Auxiliary Feedwater System. The auxiliary feedwater pumps are used to maintain water level if the startup feedwater pump is not in service.
The system consists of four separate trains. Three of the trains (A, B and C) ute motor-driven pumps and valves powered from essential AC power sources. The fourth train (D) utilizes a steam turbine driven pump and valves powered from essential DC power sources. The only ,
exceptions to the complete separation of the four trains are the '
following:
Both Trains A and D receive automatic actuation signals from the same initiation Train (A).
All trains draw water from a common tank.
A crossover line allows any pump to provide water to any steam generator with remote manual alignment of crossover isolation valves.
OV A schematic of the AF system is shown in Figure II.A.I. The elevation of the various components is shown in Figure II.A.2 (5 sheets). As shown in these figures, three horicontal, centrifugal, multistage electric motor-driven pumps supply one steam generator each. Each pump is supplied power from a separate Engineered Safety Features bus and the power supply is separated throughout. A description of the characteristics for the three motor-driven pumps is provided in Table II.A.l.
The fourth pump is a horizontal, centrifugal, multistag;,
noncondensing steam turbine-driven unit which supplies water to the fourth steam generator. A description of the characteristics of the turbine driven pump is provided in Table II.A.l.
Each of the four pumps is provided with a minimum flow, automatic recirculation control (ARC) system. The ARC valve is provided by Yarway and consists of a flow balanced spring loaded check valve in the main flow stream and a combination piston valve and cascading piston to provide for the minimum flow required for the pump. When the flow rate requirements of the system exceed the minimum flow rate for the pump, the bypass line piston valve closes off entirely. The main check valve is concurrently positioned to allow flow around the v)
I 01481/00051
disc through the main valve outlet. Each valve has a 4" inlet and outlet with a 1 1/2" bypass line. The bypass is sized to pass O
C/
approximately 100 gpm. The bypass flow is routed back to the auxiliary feedwater storage tank.
Each AF supply line is provided with a flow control valve (valve numbers FV-7526, FV-7525, FV-7524, FV-7523). The operator can remote manually position the valve to obtain the required flow to the steam generator. In addition, this valve is controlled by the Qualified Display Processing System (QDPS) following an 51 signal, receipt of a two of four low-low steam generator water level signal from any steam generator or receipt of a low feedwater flow signal from AMSAC. QDPS :
will control the valves to provide approximately 600-625 gpm flow and '
maintain this flow within a 550 to 675 gpm band as long as the ESF signal is present. The valve logic contains a reset switch which allows the operator to bypass the ESF signal and remote manually position the valve.
Each train is also provided with a motor-operated stop check valve.
These valves serve the dual function of isolating their respective trains from the associated steam generator and preventing backflow into the AF process train should the valve be open. The valves can be remote manually opened and closed by the operator. In addition, the valve is automatically opened by a SI signal, a two out of four low-low level signal in any steam generator, or a low feedwater flow signal from AMSAC.
A crossover line is provided to allow the added flexibility to feed f) any of tne four steam generators from a single pump. The crossover !
V line is provided with crossover isolation valves associated with each train. The crossover isolation valves (FV-7518, FV-7517, FV-7516, FV-7515) are normally closed with a fail closed actuator and closa on an ESF signal. The valves in Trains A, B and C have a spring loaded actuator sized to operate against maximum differential pressures of 1560 psi from either direction. The valve in Train D is supplied by a different vendor. It has a spring-loaded piston actuator sized to operate against a maximum differential pressure of 2685 psig.
With the AF system in standby, which is the normal condition, the I status of each active component in the system is as follows:
AF Pump Not running Pump discharge ARC Valve Recirc j Flow control valve (MOV) Open i Train isolation stop check valve Closed Train crossover isolation valve Closed I
\
("%
01481/00051
l l
I i
The AFW system is capable of automatically initiating flow to the ,
steam generators upon any of the following: I
/ "
l (,)g 1. Low-low level signal in two of four level channels in any one steam generator
- 2. Safety injection signal i
- 3. Manual actuation
- 4. Low feedwater flow signal from AMSAC Actual AFW flow to the steam generators is dependent on one of the :
above signals. Remote operation capability for each AFW train is !
provided in the control room and the auxiliary shutdown control i panel. '
i n
( I v ;
1 D
(G j
~ ~
01481/00051
TABLE II.A.1 iv) Auxiliary Feedwater Pumps:
- a. Motor-driven Manufacturer Bingham-Willamette Number 3 Type Centrifugal, Horizontal, multistage Flow, gal / min 540 Design total dynamic head, ft 3,310 Motor, hp 800
- b. Turbine-Oriven Manufacturer Bingham-Willamette (Turbine driver provided by Terry Turbine.) ,
Number 1 Type Centrifugal, l Horizontal, multistage Flow, gal / min 540 (3 Design total dynamic head, ft 3,310
( ,/ Turbine hp 663 i
j i
(
01481/00051 l
l l
i.
^~-- .!
AFST wa l 500,000 GAL. n r, 7 .
't - / FC
- ' aj
! i hd FV7518' oooo TURBINE-DRIVEN AFP 14 JL N' -
'TRd y ARC
] VALVE ws JL v'm FC O'
hd FV7517!
TURBINE-DRIVEN AFP 11 JL i
y ARC 'TRh
] VALVE
.ws
! w_s a r, FC
- hN FV75f6i TURBINE-DRIVEN AFP 12 JL y ARC ' 'TRA[
] VALVE' w2 a v, FC O
FV7515 TURBINE-DRIVEN i
AFP 13 JL N' 'TRAq
-p ARC
' VALVE
- .]
v
~,
MiC@@. _ _ _ _ _ _ _ _
l l
f 71 l ){ CONTAINMENT l g g g (TYP.)
>< = >W N = I8 io
$O' .FV7526 [
- r s >i< +
n 2:
FT
)[
V @ M
>< = 1
>W N = sE a 3 A' FV7525 o
[
h b y n-FT
)(
V E E
>< Z- >W N
=18is e B- Fv7sz4 a
T hf $ NI a
):
FT
-)['
V E E
>< = >W N TO
= so ic
$ C' FV7523 I ,.
l f?M 5 > 4 W AUXILIARY FEEDWATER SYSTEM
) SCHEMATIC l
480145 00 SV -OD FIGURE IIM
_ - - - - - _ - - - - - - - - - - - - - - - - - - - - - - - - - - - J
OVERSIZE DOCUMENT PAGE PULLED SEE APERTURE CARDS NUMBER OF OVERSIZE PAGES FILMED ON APERTURE CARDS ?
APERTURE CARD /HARD COPY AVAILABLE FROM RECORD SERVICES BRANCH,TIDC FTS 492 = 8989 l
l
... s
& nm
@l[AF0359 7518 @EAFO358 7517
% # 'l4 AF-lO47 GA3 t EL 24'-O " % a.4%F 1 )l3-G t
_ 5'3" _ 5 ' 3" _
E A EL 22'- 0" p _ , _ _ y 4p._L_22'-6* y _ _, g T 'F" t PT%"
SEE SHT, 4 SEE SHT,1 AF02 AFO 108'O" CROSSOVER PIPE I
ff f
i 1
i l
b 7516
@ir s gAF0356 M
7515
@ir nAFO360
$. g g4%F-lOl4 GA3 g g4 AF.J047-GA3 4 '- 5 " _ 4 '. 8 _
A EL 2 2'- 6 " p .,,. _ g t pT P " AEL22'6"p__9
% P T K
SEE SHT,2 SEE SHT 3 A F02 i AFO
>t D : .:
g EL j6" 1
1 1
l AUX FEEDWATER SYSTEM TPAlN A,B,CD SYSTEM PROFILE SCHEMATIC SOUTH TEXAS PROJECT L z
SCALE N/A OWG N O REV SHT 5 OF 5 FIGURELA.2 o L
__ ~ _ _
7 O
SECTION III INVESTIGATIONS TO ESTABLISH ROOT CAUSE OF HYDRAULIC TRANSIENTS I
O 4
l 1
l l
}
i O l 01481/00051
/
l
y I
I I I . A'. Design' Review of Water Hammer Potential
. The auxiliary feedwater system was reviewed during design de'velopment-( for potential water hammer problems. This review included the suction and discharge piping. arrangements and valve opening / closing times. The suction piping was reviewed for possible void formation due-to excessively long runs of pipe and any high point loops. The discharge piping was reviewed for possible air entrapment and possible backleakage from the steam generator. Valves were reviewed to determine if there were any rapid strokes on openings or closings that could result in unacceptable transients. At the time the system was reviewed, the only transient analysis deemed necessary was for a check valve " slam" following a postulated feedwater pipe break.
upstream of the check valve. This analysis produced transient loads which were included in the pipe stress analysis-for the piping
. downstream of the break location. The piping analyzed is located primarily inside the Containment Building. Hydraulic loads
=
associated with normal system operation (i.e. pump starts and stops)-
are significantly less (i.e. typically less than 50%) than the loads analyzed due to seismic, thermal, and deadweight considerations.
The reason no other hydraulic-transient loads were addressed is that the layout had already been designed using the guidelines in SRP Sections 10.4.9 and 10.4.7 (Section 10.4.7 includes BTP-ASB 10-2).
Some of the major features implemented are as follows:
- 1. A separate nozzle is providedf'or the-introduction of AFW to the I steam generator. (The nozzle does not incorporate a feed ring or feed preheater design).
- 2. The length of. horizontal piping immediately upstream of the AFW nozzle is minimized (approximately 7 feet).
- 3. The AFW inlet piping within the steam generator is designed to be self venting.
- 4. The outlet of the AFW nozzle is designed to be below the normal steam generator water level.
In addition to the above layout considerations, there is also a temperature element used for monitoring any backleakage through the I stop check valves in the pump discharge piping. This element was ;
added to the design in light of NRC. concerns about potential steam binding of AFW pumps in the nuclear. industry. The temperature elements, however, provide an added benefit with respect to identifying backleakage of hot feedwater that might result in localized steam voids in the piping. Such an occurrence could result i in water hammer upon starting the pump.. By use of these temperature elements and by pyrometer measurement during the testing program, it was verified that no backleakage had occurred.
01481/00051
i
, Concerns about whether or not water hammer would exist in the system
, ,\
were also addressed during the pre-operational test program described in Section III.B. This testing verified that fluid induced (d transients did not occur during an ESF actuation of the system. ;
In addition, steady state piping vibration tests were performed >
during the pre-operational test program to comply with the recommendations of Regulatory Guide (RG) 1.68 and satisfy the requirements of ASME B&PV Code,Section III. Additional tests, including dynamic testing (as suggested in RG 1.68), are scheduled during power ascension testing. In addition to AF, the following systems were included in the vibration monitoring test program during the pre-operational testing.
Component Cooling Containment Spray (except spray header)
Chemical and Volume Control Main Feedwater (safety-related portion only)
Main Steam (safety-related portion only)
Residual Heat Removal System Safety Injection System Essential Cooling Water System Diesel Generator Reactor Coolant Essential Chilled Water (safety-related portion only)
Fuel Pool Cooling and Cleanup Reactor Coolant Pressurizer System (PORV Discharge Lines)
Steam Generator Blowdown O
() Small bore piping was also included in the pre-operational test program. Inspection of piping systems included both large bore and small bore piping. Additionally, essential safety-related instrumentation lines up to the first rigid guide support were included in the vibration monitoring program during pre-operational testing.
Monitoring of vibration was by visual examination, supplemented by the use of portable instrumentation as needed. In locations in which visual observations identified significant vibrations, readings were taken to quantify vibration levels. Using the criteria defined in ANSI /ASME OM3 (Requirements for Pre-Operational and Initial Startup '
Vibration Testing of Nuclear Power Plant Piping Systems) the
, vibration levels were then documented as either being acceptable or requiring further evaluation. This evaluation could include additional testing or evaluation using a more rigorous calculation method (i.e. detailed vibrational stress calculations).
/~'%
1 01481/00051 1
-- -- - 2
i p
L i III.B. Review of Pre-operational and Hot Functional Testing i f As part of the root cause investigations, a review was made of l
. (]f- pre-operational and hot functional testing of-the auxiliary feedwater system to determine why hydraulic transient problems were not detected during these tests. This section summarizes the extent of >
these test programs. l
- 1. Pre-Operational Testing These tests and operations included crossover isolation vC /e l- closures with flew, crossover isolation valve opening under l
static conditions, pump performance tests, pump endurance tests,
.AF water hammer test, and blackout (loss of_AC power) test.
. Pre-Operational testing does not test single pump flow to more than one steam generator at a time since this is not a safety 1 design basis.
Closure of the crossover isolation valves with flow This was performed by operating each pump through its associated test line and adjacent train test line with flows in each test line of approximately 150 gpm and 450 gpm respectively. This test demonstrated the closure time of the crossover isolation' valves to be 10 seconds or less. The test closed each crossover .
isolation valve in the same train as the operating pump. This I is the maximum differential pressure that the valve would be p required to close against. The closure times measured were-t 1 Train A 7.3 sec, Train B 10.0 sec, )
Train C 7.7 sec, Train D 5.1 sec.
No hydraulic transients occurred during the valve closure tests.
i Crossover isolation valve opening time Crossover isolation valve opening was timed with system static !
pressure existing across the valve.
Pump Performance Test AFW pump performance testing was conducted by using the test line to return pump flow back to the AFWST. The test line was selected for use due to the limited volume available in the steam generator for accepting pump flow. The pump capacity was tested at three distinct flow points; 1) The miniflow point; 2)
Minimum flow point required by safety analysis (540 gpm); 3)
Pump runout point.
\
d l l
01481/0005i
The.miniflow point lwas tested by maintaining the flow control valve closed'and allowing the automatic recirculation valve to
[S s
bypass-flow back to the.AFWST. -The minimum flow required by-safety analysis ( .540 gpm at 3310 ft TDH) was tested by l maintaining the motor-operated stop check valve outside containment closed and regulating flow-to-the test line.
The pump runout point (1050 + 25 GPM) was tested by using test lines on two trains to handle the flow. This invclved setting
-up. flow in the " tested" pump train and'an adjacent train test
! line. The crossover isolation valves were used to setup this l- configuration.-
During Lthe " pump runout" testing for AFW Train C, a water hammer !
occurred when the Train C crossover isolation valve was opened, and the Train A flow control valve was still closed. An q inspection of the line revealed no physical damage however some-test ~ gauges were. damaged. . Venting of the crossover line was performed by sweeps using' Train C for; source water.and sweeping
-Trains B and A back to the AFWST. The. test was resumed and no-further, water hammers occurred. Several. maintenance / trouble shooting events occurred.which required that the AFW system be partially drained. One in particular involved the
. inspection / replacement of the Train.C pump rotor. After the-work was completed, the system was filled using AFWST as.the static head and vented using normal operating vents. The pump was run for testing and no waterhammer or other problems were noted.
Pump Endurance Testing AFW Pump Endurance Testing was conducted at 540 GPM using the pumps associated test line. The flow rate was maintained for a minimum of 48 hours5.555556e-4 days <br />0.0133 hours <br />7.936508e-5 weeks <br />1.8264e-5 months <br /> continuously then the pumps were shutdown.
The bearings were-allowed to cool to within 20*F of ambient temperature and the pump was restarted. A subsection of the endurance testing involved operation of the pump on miniflow for one hour, i
The Train D AFW Pump is steam turbine driven. This pump was i tested during Hot Functional Testing due to its requirement for steam. The pre-operational test (1-AF-P-02) was very similar to the motor driven pump testing outlined above. The test j consisted of valve testing; valve timing; performance testing; i and endurance testing. A unique portion of this pre-op test was the response of D Train to a total loss of AC Power.
l l
01481/00051
- __$-_-_______.__ _ _ - _ _.___E. _ _ _ _ _
4 l
1 Blackout Test i The total loss of AC Power Testing consisted of two parts: 1) y An endurance test of the Train D pump. 2) The ability of Train l l3 D to feed A, B, C and D steam generators.
The endurance test consisted of running Train D at 540 GPM with no ventilation available for 2 hours2.314815e-5 days <br />5.555556e-4 hours <br />3.306878e-6 weeks <br />7.61e-7 months <br />, using the test line.
l The test method for cross feeding each of the steam generators involved manually opening the Train D crossover isolation valve, manually opening the crossover, flow control, and outside !
containment isolation valves for.the generator to be fed and observing a positive level increase in that generator. During !
the cross feeding testing, a large number of station personnel ;
were located in the various compartments of the IVC. No unusual '
events were observed. ;
Auxiliary Feedwater Water Hammer Test This test initiates flow from each auxiliary feedwater pump to its associated steam generator. The steam generator water level is lowered.to below the auxiliary feedwater nozzle and the steam generator isolated. The piping is checked for steam backleakage
-using hand held pyrometers then an ESF AF Actuation Signal is jumpered in. The auxiliary feedwater pump starts and flow rate is controlled by QDPS. The system is monitored for piping movement and vibration while feeding. After feeding is secured ,
piping temperatures are monitored for four hours and piping and Q('N supports are visually inspected for damage. No unusual events were noted during the test.
- 2. Pre-Core HFT Operation !
During Hot Functional Testing (HFT), the AFW system supplied makeup to the steam generators. A typical lineup for this was:
- 1) One AFW pump running
- 2) Use of crossovers to allow feeding of steam generators The following sequence was performed numerous times during HFT for makeup to a steam generator.
- 1) Remote opening of running pumps crossover isolation .
valve !
- 2) Remote opening of corresponding steam generator crossover isolation valve
- 3) Remote opening of outside containment isolation valve (if not open)
O 1 01481/0005i
l 4) Feeding steam generator at 200-300 GPM l- 5) Stop feeding steam generator ,
l / i 6) Closing crossover isolation valves
- (.)
-7) Go on to next steam generator to be fed The running AFW oump was rotated to help equalize running time I on the pumps. Thus during HFT all four AFW pumps were used to feed different steam generators by use of the crossover piping and valves. No hydraulic transients were observed during this l testing.
- 3. Post-Core HFT Operation During post-core Hot Functional Testing a typical lineup for feeding steam generators has been to have one AFW pump running, all four crossover isolation valves open, and AFW flow control ,
valves open to maintain level in the steam generators. In this l manner one to four steam generators could be fed at any one time. ,
'3 The review of pre-operational and HFT testing of the auxiliary ,
feedwater system indicated that no sustained hydraulic transient l events occurred. In particular.none of the tests required the flow control valves to be opeated at their near-seat position which has subsequently been found to be the cause of the recent flow-induced transient events.
f~%
O .
i i
l l
i
/3 V
01481/00051
_ - _ _ _ - _ _ _ _ _ _ - _ _ _ - _ _ b
I III.C. Additional Testing Performed to Isolate Cause of Transients C
A series of staven system tests was performed to determine the cause of the fluid dynamic events experienced in the Auxiliary Feedwater system during the months of November and December 1987. The tests performed were as follows;
- 1. Auxiliary Feedwater Piping Vibration Test ITEP07-AF-000J
Purpose:
Determine if steady state vibration was sufficient to cause the vent instrument tap and pipe support failures experienced and to ;
identify the driving mechanism for the vibration. This test was '
begun on November 14, 1987. ,
i Method:
In the vibration test the Train A pump was run from minimum flow j condition to full ficw using the' test line back to the AFWST.
All three motor driven pumps were run separately at minimum flow and at low flow conditions to each one of the steam generators.
Only one steam generator was fed at'a time and only one pump was operated at a time during this test. Dynamic pressure changes were measured.on the pump suction and discharge and downstream of each of the four flow control valves. Vibration readings were taken at vent, drain, and instrument connections.
Results:
The results of the vibration test determined that steady state vibration levels are not sufficient in themselves to cause the valve assembly and pipe support failures experienced.
- 2. Auxiliary Feedwater Piping Venting Test ITEP07-AF-0002
Purpose:
Determine the quantity of air trapped in the system due to inadequate venting and if the water hammer associated with opening the crossover isolation valves under pressure with no flow conditions was eliminated by enhanced venting. This test was run on November 21, 1987.
Method:
In the venting test the system was first drained and filled without the use of five new intermediate high point. vents and using the revision of the venting procedure in place at the beginning of November.
I 01481/00051
_ _ _ - - . .- __________-_________-_____L
1 3 The system was then vented using the revised venting procedure and
~
l- the five'new intermediate high point vents. The additional air
' /G . vented was collected in plastic bags and measured. The Train A
'Q pump was then started and the Trains A and 0 crossover isolation valves opened. A waterhammer occurred as each crossover isolation -
valve was opened. The venting procedure was enhanced to require a sweeping through'the crossover lines. The test was then rerun.
. Train'A pump was started on miniflow and Trains A and 0 crossover isolation valves opened and closed without any incidence of water hammer. The Train A pump was secured and the Train B pump started on miniflow and Trains B and C crossover isolation valves opened.
and closed without any incidence of waterhammer. Dynamic pressure changes were recorded using the same pressure transducers installed
.for the vibration test.
Results:
The result of the venting test indicated that there was sufficient trapped air.(Train A 1.5 cu ft, Train C 0.5 cu ft, Train 0 1.9 cu ft) to potentially cause a waterhammer and that the air could be adequately purged from the system with the added vents and enhanced venting procedure.
- 3. Auxiliary Feedwater System Train "B" Pump Curve Determination 1TEP07-AF-0003
Purpose:
p Determine if the pump involved in the November 22, 1987 failure had-
\ a characteristic curve which would cause surging, if pump interaction with the Yarway ARC valve led to system vibrations and investigate the effect of closing a crossover isolation valve under flow. This test was run on November 26, 1987.
Method:
In the-pump test Train 8 pump was run on minimum flow, through its i own minimum flow line, to full flow through the Train C test line, i Flow was adjusted in approximately 10 gpm increments to try to achieve good resolution on the pump curve and to try to induce instability between the pump and Yarway ARC valve. The crossover isolation. valves were also closed against flow through them of ;
approximately 75 gpm, 200 gpm, and 650 gpm to try to initiate a j sustained waterhammer event. Pump suction and discharge pressure, Yarway ARC valve miniflow discharge pressure, system flow, minimum flow line flow, dynamic pump suction and discharge pressure changes, dynamic minimum flow line pressure changes, flow control j valve dynamic pressure changes, and dynamic pressure changes near i the crossover isolation valves were measured.
01481/00051
Results:
(3 The results of the pump test confirmed that the Train B pump curve V has very little slope below 200 gpm. The Yarway ARC valve was relatively stable in the crossover range and did not open and close rapidly due to minor changes in flow. The Train C and D crossover isolation valves caused waterhammer when closed against flow greater than 300 gpm. However, these were of short duration and were not severe enough to cause any damage.
- 4. Auxiliary Feedwater System Suction Piping Trapped Air Test 1TEP07-AF-0004
Purpose:
Determine if substantial unvented pockets of air existed in the motor driven pump suction lines. This test was run on November 25, 1987.
Method:
In the suction piping trapped air test the Train A, B and C pump suction lines were pressurized; then sufficient water was bled off to return the line to the initial pressure. The quantity of water bled off is related to the quantity of trapped air through the ideal gas law.
Results:
A Q No significant trapped air was found in the pump suction lines which further indicates that problems due to inadequate venting have been eliminated.
- 5. Auxiliary Feedwater Multifeed Vibration Test ITEP07-AF-0005 i
Purpose:
Try to induce the sustained flow induced vibrations which occurred on November 22, 1987 by duplicating the system conditions. This test was initially run on November 25, 1987; the same test '
procedure was used for subsequent multifeed testing in December.
Method:
In the multifeed vibration test flow was established to various combinations of steam generators to try to duplicate the conditions !
present when the sustained vibrations occurred on November 22, )
1987. The major difference was that the plant was in Mode 4 for !
the test and was in Mode 3 at the time of the event.
Instrumentation installed in test 3 was used as well as pressure transducers added in each train downstream of the flow control 4 valves to measure dynamic pressure changes, and hand held pyrometers to measure pipe temperature, g
01481/00051
Results:
p The Multiple Steam Generator Feed Test induced vibrations on v December 6. The vibrations were found to be at approximately 24 Hz. These results led to performing the Auxiliary Feedwater Pipe and Valve Vibration Test (lTEP07-AF-0007).
- 6. Auxiliary Feedwater System Performance Test ITEP07-AF-0006
Purpose:
The purpose of this test was to assure that the system would meet its safety design basis. This test was run on November 29, 1987.
Method:
In the Auxiliary Feedwater System Performance Test each motor driven pump was run to its associated steam generator by initiating an ESF signal. Af ter the pumps were started and QDPS had taken control of the flow rate the operator closed the associated flow control valve which put the pump back on minimum flow thus opening the Yarway ARC valve. The flow control valve was then reopened and a flow rate of 500 gpm established. The flow rate was then reduced in 100 gpm increments to a rate of 200 gpm at which time the flow control valve was closed concluding the test.
Results:
p)
(~
The Auxiliary Feedwater System Performance Test demonstrated that all three motor driven pumps successfully delivered the design rate of flow to their respective steam generators.
- 7. Auxiliary Feedwater Pipe and Valve Vibration Test 1TEP07-AF-0007
Purpose:
Try to induce sustained resonant vibration by duplicating the system condition of the November 22, 1987, and eliminate the resonant vibration by increasing the air pressure under the crosstie valves and/or increase flow through the flow control valves. Also examine whether resonant vibration is dependent upon operation of any particular pump or the crossover valves. This test was run on December 9 and 12, 1987.
Method:
In the Auxiliary Feedwater Pipe and Valve Vibration Test the system was run the same as in the multifeed vibration test (ITEP07-AF-0005). Each of the four flow control valves was individually tested by using the valve handwheel t6 open the valve (D
L) 0148i/00051
rc .,
I l
'l (from its. fully closed position) in small increments. The tes't-also included examining the effect of increasing the air pressure to the crossover valve operators.
- Both Train B and.C pumps'were used during the course of testing. "
Instrumentation was the same as in the multifeed vibration test. l except that' lanyards.were added to record flow control valve and }
crossover valve positions and piping movement. In addition j pressure gauges were added to measure the' outlet pressure of the !
-crossover valve air regulators. '
Results: 1 The Auxiliary. Feedwater Pipe and Valve Vibration Test induced 24 Hz 1 resonant vibrations en December 9 and 12. . Changes.in the air regulating pressure on the crossover isolation valves had no affect on the vibrations. The crossover valve and flow control valve stem measurements showed no movement had occurred indicating that. valve
" plug" motion was not the cause of the vibration. -The testing also l
showed that the vibrations were eliminated when the Train A and D
! flow control valves were opened beyond their near-seat, highly throttled position. The vibration event was not related to a-specific pump.
l l
O O 0148i/00051
III.D Review of Potential Causes of Hydraulic Tranjient Events (9 1. Data trom Other P1 ants L.J '
Using the Nuclear Plant Reliability Data System (NPRDS), a r.rint out of plant-specific data was obtained for "various failures of auxiliary feedwater system components" and for " failures on auxiliary feedwater system caused by water hammer, vibration, or rupture."
The first category included miscellaneous failures of valve parts, sensing lines, valve seats and pipe supports due to causes such as normal wear, system vibration, incorrect welding, cavitation and unknown (for the pipe support weld f ailure). The f ailures are not described as ha/ing been caused by water hammer in the specific plants. The one case of valve cavitation in a steam generator level control valve was discovered due to through-leakage.
The second category of failures included mainly failures caused by flow-induced vibr3 tion. These failures were generally found during routir.e maintenance or inspections during an outage, There were three descriptions of water hammer. In one case at Farley Unit 2 " slight w ter hammer" was caused by dirt under the seat of an AFW pump discnarge check valve. The second water hammer event, at San Onofre 1, damaged a pipe guide; however no cause was given for the water hammer. The third water hammer, also at San Onofre 1, damaged a snubber; the hamer was caused
[]
(,)
by misadjustment of the feedwater regulating valve control c irc uit. A review of these three events does not show any correlation with the South Texas Project prob'em nor did the corrective actions aid in resolving the STP problem.
- 2. Fast Valve Operation Hydraulic transient events are known to occur as a result of fast valve operation (closing or opening). Fast valve closure can result in column separation and rejoining that results in a load applied to the piping system that quickly dampens out.
Fast valve opening can also result in a positive pressure increase especially with the presence of air or vapor downstream of the valve. )
l The valves in the AF system are for the most part slow closing !
and slow opening. The flow control valve and stop check valve j are motor operated valves with an operating time of 40 seconds '
and 10 seconds respectively. Operation of these valves has not produced a hydraulic transient effect typical of fast valve j operation.
l 1
l O
l l
01481/0005i i
L_ _ ____ _ __
l l
The crossover isolation valves in Trains A, 8 and C are air operated valves with a 10 second closing time and a longer
/] opening time. Operation of one of these valves has on one l' V occasion caused a minor pressure oscillation while closing against full flow conditions when feeding the Af storage tank on the test line. This was not damage producing and not uner.pected due to relatively low back pressure on the valve.
The Train D crossover valve is a fast acting valve with 1-second opening and 3-second closing times. The 1-second opening produces a pressure oscillation which is not damage producing; however since the system requirements do not dictate a fast opening time this valve has been adjusted to slow down the opening time.
The Yarway auto recirculation valve has a fast switchover from bypass flow to main flow which is characteristic of the valve construction and not adjustable. This switchover does not produce significant pressure oscillations and ao structural interaction has been noted.
- 3. Entrainment of Non-Condensible Gases The potential for non-condensible gases being entrained in the system was extensively reviewed and tested (as described in Section III.C) to assure a water solid system. The tests performed prove the fact that the vent system modifications eliminated water hammer events caused by air voids, (q'
j In addition a review was performed to assure adequate nozzle submergence exists for the AF storage tank nozzle to the pump j
suction. The tank is maintained at a water level that keeps a positive head on the system and that eliminates the potential for vortexing.
The physical layout of the piping (as shown in Figure II.A.2) was reviewed, particularly the relative elevation of the tank water level, the check valve located near the steam generator, and the steam generator water level. The normal steam generator water level is higher (by about 20 feet) than the water level in the AF storage tank, thus keeping the system water solid once it has been vented.
- 4. Pump Instabilities Pump instability can take two forms. One is a flat or locally concaved head-discharge relationship and the other is flow recirculation in the pump suction when a pump operates at a small flow rate. The operating experience and testing program i
I o )
C 01481/00051 l
i
ha's indicated that the pump is not experiencing an instability as a result of the pump head-discharge curve.
, q d) When a pump operates.at a low discharge flow rele.tive to the design flow, recirculation can occur between the suction elbow and the first stage impeller. The recirculation will set up a l vortex which breaks down periodically. The frequency associated
!= with the vortex breakdown can become a forcing function which at times can excite a system. In the hydraulic tests conducted in December'.6, 9 and 12, a pressure transducer was installed at the suction of the Train 8 pump. The signals did not show a organized periodic pressure pulsation.. Furthermo N , the pressure traces clearly show that the oscillation at the pump suction' lagged behind, in time, relative to the oscillation at the pump discharge. Therefore, it can be concluded.that the pump.-suction does not produce periodic oscillations that excite the system.
- 5. Check Valve' Closure The auxiliary feedwater system includes a stop check valve and a swing check valve. Both valves are designed to prevent reverse flow migration. Severe water hammer events are known to occur when, upon a pump trip or.a power failure, check-valves close after a substantial reverse flow has established.. Because the-
- closure of:the check valve is assisted by the reverse flow, the rate of closure can be.very rapid. A rapid closure of the check valve can produce severe water column separation in the upstream piping-(viewed from the normal flow direction). The subsequent
.( rejoining of the separated water column can produce a water hammer.
The behavior of the check valves has been tested during the pre-operational. testing of the system. In such a test, the normal. operational pressure of the steam generator was simulated. No water hammer was noted. It is therefore concluded.that the check valve closure will not be a major cause of water hammer events.
- 6. Pump Trip As a result of a pump trip, a transient will be created by the flow interruption, with initial downsurge at the discharge side of the pump and upsurge at the suction side. The magnitudes of the surges.are dependent _upon the piping length, the flow velocity prior to the trip, the combined pump / motor inertia, etc.' For the South Texas Project auxiliary feedwater system, the length of the suction and discharge lines is about 150 ft and 200 ft, respectively. Based on past experience,-transients created by pump trip for such a short piping system are judged to be small, i
01481/00051 4
i- -
'7. Steam Void Collapse-
- l. If. steam was present-in the piping system when the auxiliary feedwater pump (s) were started there would be the possibility of steam void. collapse and momentum effects from water filling the R
piping. This is a specific condition which the system is L
designed to prevent by the use of a check valve at the steam generator, a normally closed stop check at the containment penetration and a check feature in the Yarway recirculation valve.to preclude intrusion due to steam back pressure from the steam generators. In addition, to keep the piping system water-solid, the inlet nozzle and tailpipe are sloped up into the steam generator and the'end of the pipe is submerged below ;
.the normal water levels. A temperature' element is provided in each train to monitor steam intrusion. During testing the pipes near the steam generators were checked and no steam intrusion was present.
- 8. Review of Valves in System The analysis included a thorough review of all the valves in the AFW system from the following two standpoints:
- a. Potential for initiating the water hammer event.
- b. Potential for providing sustained excitation following the water hammer.-
The system contains a number of valves which are (1) fluid flow
- p. activated, (2) spring loaded, air activated, (3) motor
.gi operated. These valves were reviewed and are coveref in this section in the following order.
8.1 Swing Check Valve (fluid actuated)'
8.2 Piston Stop Check Valve (motor operated) 8.3 Yarway Automatic Recirculation Control Valve (fluid Actuated / spring regulated) 8.4 Crossover Isolation Valves (FV 7515, 7516, 7517 and 7518) l-A. WKM valves (air diaphragm actuated / spring closed) 3 B. Valtek valves (air piston actuated / spring closed) q 1
8.5 Valtek Flow Control. Valves (motor operated) 1 A description of these valves and their potential contribution to the problem follows. Drawings of the crossover isolation valve and the flow control valve are included in the Appendix.
I 01481/00051'
8.1 Anchor Darling Swing Check Valves L
l O
. (,-/
There are four swing check valves in the AFW system: AF-0119, AF-0120, AF-0121'and AF-0122. These are located in each auxiliary feed line to the steam generators, and the function is to prevent reverse flow. These valves are of typical swing disc design which is activated by forward flow. The minimum velocity to fully open the disc is calculated to be in thc range of 9.5 fps to 12.0 fps. This corresponds to a flow rate of approximately 1400 gpm to 1700 gpm. Therefore, the valve disc will not be in the fully open position and firmly seated against
-the backstop at the design flow rate of approximately 600 gpm.
The flow velocity calculated at 600.gpm is 4.1 fps and the corresponding disc equilibrium angle is calculated to be 28 degrees from vertical. At lower flow rates the disc will be open at an angle even smaller, and it is predicted to be about 5 degrees from the seat at 200 gpm.
The natural frequency of the disc and hinge arm assembly is calculated to be in the range of 1.2 to 1.3 bz in the disc positions corresponding to the low flow conditions and high. flow conditions. The disc fluctuations are estimated to be 3 degrees and the corresponding pressure fluctuations less than 1 psi with the flow in the forward direction. This natural frequency of disc oscillations'is far away from the system resonant frequency of 24 bz. Therefore, the swing check valves by themselves are not the exciters of the system resonance which results in high pressure fluctuations.-
't^'i-g' This is consistent with the observed test results in which high pressure fluctuations were found in the piping system upstream of the flow control valves whereas these swing check valves are located downstream of the flow control valves.
8.2 Piston Stop Check Valves (Motor Operated)
The system uses a globe body piston type stop check valves for each one of the four lines: MOV-0048, M0V-0065, M0V.0085, and MOV-0019. These valves are manufactured by Rockwell International and are activated by forward flow. The minimum flow velocity for this valve to fully open was found to be 7 fps, which corresponds to 250 gpm. The maximum differential .
pressure across-the valve disc at the design flow velocity condition of 600 gpm is calculated to be 9 psi, and at flow {
rates of less that 250 gpm the pressure fluctuations are to expected to be less than 5 psi. Lift and piston type check valves are the most suitable type of check valves for service in the partially open position. At flow rates less than 250 gpm, l the disc will be partially open and it can oscillate axially and laterally when operated below a certain minimum flow rate. When operated very close to the seat, these valves can exhibit 1
01481/00051 j l
l
instabilities of the plug, i.e., chattering, which can be heard-
% outside the. valve body. Laboratory tests run by the L( ) manufacturer several. years ago on'a similar valve showed the i
(/ plug to chatter or wobble at flow rates below 60 gpm. However, l no frequency data'are available. No chattering has been heard from these valves during any of the tests in the AFW system.
Since high pressure fluctuations during instrumented tests on I theAFWsystemweredeterminedtobeconfined{upstreamofthe flowcontrolvalves,whereas)hesestopcheckvalvesarelocated l downstream, these ;te d. - valves are ruled out as the liki,7%
source of ained ;c, ;..; TW ;,; L...
If necessary, these valves can be used instead of the flow control valves to completely shut off the flow in any one of the l AFW lines leading to the steam generator,by =cr W h; th;ic"_ ,
s+T f =;ti;n.+ '
8.3 Yarway Automatic Recirculation Control (ARC) Valves There are four Yarway ARC valves in the system (AF-0036, AF-0058, AF-0091 and AF-00ll) for the protection of the four pumps. These valves serve the dual function of preventing reverse flow to the pumps and providing automatic recirculation for cooling flow to the pump during low flow operation. F Mure 1 schematically depicts the operation of the ARC feature.
These valves are spring-loaded fluid-actuated type. They bypass 2 the flow under low flow conditions until a flow rate of
(]/ approximately 150 gpm is achieved. After this crossover point, the fluid forces overcome the spring preload and the flow in the bypass line is closed off and all the flow is fed to the main line. The natural frequency for this spring mass system is calculated to be 3.8 bz. At the switching point, the differential' pressure across the vt,1ve is approximately 5 psi and it increases to 15 psi at the design flow of 600 gpm. The ARC valve's characteristic is that the disc position switches from low flow bypass position to the full forward flow position over a flow range of 110 to 170 gpm. Yarway has advised that the valve could exhibit some unstable behavior if operated in !
this general region.
Even though instability may occur in this flow range, the valve is not the exciter of the high pressure oscillations observed at 24 bz because of its much lower natural frequency. Extensive ,
testing has been done in this region of operation to generate K sustained" oscillations from the Yarway valves but no transients were observed in either the recirculation piping or the main header piping. Neither small jogs nor larger changes of flow in !
several series of ascending and descending flows would produce 01481/00051 .
2
_ __ _ _ i
Jany: response in the downstream piping other than small single-rise or-fall pressure surges with no sustained-f-
V} oscillations.
8.4 Crossover Isolation Valves There are four crossover valves in the AFW system to permit any AFW pump to feed any steam generator. These valve tag numbers are FV 7515 7516 7517. and 7518. The first three valves are manufactured by WKM, and valve 7518 is supplied by Valtek. All:
the crossover valves are pressure balanced control valves being used for on/off service with fail close action. The WKM valves use a spring loaded diaphragm actuator, and the Valtek valve uses a spring loaded piston actuator. Since there'are significant differences in the actual design of these two types of valves, they are discussed in detail separately.
A. _W KM Crossover Valves (FV-7515, 7516, 7517)
The.WKM valves are 4" 900# Class pressure-balanced, ,
cage-guided plug design. The plug size is the same as the nominal valve size, i.e., four inches. Figure 2 shows an assembly drawing of this type of valve and actuator. A
-more complete drawing is included in the Appendix. The actuator _has 280 sq. in. diaphragm area (nominal) and has'a spring on top of the actuator to achieve fail-close action. An' air regulator is set to supply 31 psi pressure' to the actuator. . The diaphragm has a maximum working _
pressure rating of 50 psi.-
The force-vs-travel characteristics for this valve were obtained by actual tests on a spare actuator and are shown in Figure 3.' The approximate spring preload.on the valve plug at'the fully closed position is 2300 pounds, the-average spring stiffness is 1900 pounds / inch, and the required force to fully open the valve is' approximately 7,000 pounds. The air pressure required to initiate the plug opening is 7-10 psi, and the pressure to achieve full stroke is approximately 20 to 25 psi.
If the air pressure supplied by the regulator is 31 psi as !
recommended by the manufacturer, a margin of 6 to 10 psi is available.to hold the valve in the fully open position.
This corresponds to a force margin of approximately 1700 to 2800 pounds under correct settings. Leakage from the diaphragm, which has been observed in several cases, takes away from this margin. The net effect of this leakage is to reduce the margin from the above range of 1700 to 2800 pounds to a lower value. It is also possible to have no I force margin in the fully open position if the diaphragm pressure is 25 psi or lower. With no margin present, it is
- u
)
01481/00051 l
l
l possible to excite a plug movement by hydraulic pressure transients in the main' piping system if.their mdgnitude is
~j sufficient to overcome the steady state axial forces on the
- V] ' plug.
- It is very important to note that even though the plug valves under discussion are the so-called " balanced" type, there are significant unbalanced areas in such designs that play a definite role in determining the net axial forces on the plug. With the plug in the closed position, these forces are depenaent upon'tne flow direction and the upstream as well as downstream pressures. As shown in Figure 4, the plug has an annular area, A 1, between the 0-ring seal of the plug that defines the top balancing-chamber and the main closure seat.
Another unbalanced area is the stem area, A . For the WKMcrossovervajves,theseareasareA1 =~ .47 in ,
and A2 = 0.60 in . Depending upon pressures P1 and P2 ' and their relative magnitudes, the net axial force on the plug due to the fluid flow can be positive or negative with the plug in the closed position.
The spring preload in the valve closed position must exceed the net upward force resulting'from fluid pressures in- )
order to maintain the closed position. With the plug in the open position .however, the plug annular . area A1 becomes ineffective in causing any axial forces and only the stem area A2 plays a role in causing forces on the t' plug due_to fluid pressure changes. :It should be pointed-out that there are some " secondary" imbalances that_ occur dce to pressure lags _between the top balancing chamber and the bottom of the plug under sharp pressure transients (usually less than 10 millisecond rise / fall times); ,
,. however, for liquid systems these effects are negligible. '
These crossover valves can have flow over the plug (P) can be greater than P 2 ) or flow under the plug. With the plug in the closed position, it is calculated tnat the plug is' seated against the seat with a force margin of approximately 1,200 to 1,700 pounds depending upon whether higher pressure is below or on the side of the plug. A i positive pressure transient of 2500 to 2800 psi can cause l the plug to lift off its seat.
i In the open position the fluid system pressure acts directly against the area A2 of the stem in causing axial forces. The stem area for this valve being 0.60 sq. in., a pressure fluctuation of 1,000 psi will cause a net axial force change of 600 pounds. If the plug is in the open position but does not have sufficient force margin, it can 01481/00051 I
l b start to respond'to system pressure fluctuations.
Additionally, if an unfortunate combination exists .in terms N of valve plug natural frequency of axial vibration being
@d
. close to-any of the excitation frequencies in the fluid systems, serious oscillation's of the plug can occur.
)
I l
Maximum amplitudes will be encountered if the plug is in i
the intermediate position so that stem motion'can build up.
without being impeded by hitting either stop position.
The overal1 review of the WKM crossover valves shows that l there is insufficient force margin in'the fully open -
position to keep the plug firmly in position and avoid any I interaction with the hydraulic transients in the system.
As stated earlier, under normal setting conditions the ,
margin can be as low as 1700 pounds which can be upset by.a negative hydraulic pressure transient of 2800. psi. .] 3 Furthermore, for valves FV7515 and 7517 there have been )
reports of leaking diaphragm gasket-to-yoke connection and j
. malfunctioning regulator. Thus the actual pressure in the diaphragm can be below 31 psi normal setting. Both of these factors in combination can allow this valve plug to start responding to hydraulic transients at pressures even lower than stated above. I
- Natural' Frequency of Axial Plug Vibrations.of WKM Crossover Valves: Calculations have been performed for the natural frequency of all the movable parts connected to the plug 1 modeled as a single degree of freedom spring mass system.
rg The' spring stiffness is based upon actual measurement from Ij one of the actuators. The weights of most of the parts used in the calculations are based upon actual measurements of the parts taken from the spare actuator. Weights of the remaining parts were obtained from the valve manufacturer.
It should be noted that in addition to the mechanical
. spring, the air in the pressure chamber of the diaphragm actuator also acts as a nonlinear' spring, the stiffness of which depends upon the nominal diaphragm pressure as well i as the actual position of the stem.
The natural frequency of this spring mass system, based ,
upon the mechanical spring alone, is 15.2 hz. The added 1 stiffness of the air spring under ideal conditions (no I leaking diaphragms, etc.)' increases the natural frequency
-of this spring mass system to 26.6 hz with the plug in the-mid-travel position (Figures 5 and 6 show the variation of j the air spring stiffness and the valve plug natural !
frequency as a function of valve' position respectively).
1
.01481/0005i
As shown in Figure 6, the natural frequency varies from 25.4 bz to 29 bz over the full range of valve positions.
CT This natural frequency curve for the plug mass system is
() the applicable curve from the standpoint of considering its interaction with the hydraulic system acoustic frequencies. Also, depending upon whether the diaphragm actuator is leaking or not, the actual natural frequency of the plug may be lower than the bounding values. i Thus, it is cartainly possible for the WKM crosstie valves to get into rt snsnt vibrations if the hydraulic piping system acoustic frequency is close to the natural frequency of the plug mass system. Instrumented tests have shown that the piping system exhibited pressure fluctuations at approximately 24hz. This could be close enough to the axial natural frequency of the valve plug to create an amplified response to pressure fluctuations. Thus a coupling between the valve resonant frequency and the acoustic frequency due to the organ pipe effect of the piping system could cause high pressure fluctuations i following a sharp transient event if the valves are not seated with sufficient force margin in the fully open or closed position.
Once this fluid / structure interaction between the flow and the valve is initiated, causing the plug to oscillate at 24hz, the structural frequencies of the piping system which are close to this excitation frequency of 24 hz can be
',q excited causing a severe piping movement at certain
) locations.
However, from the results of test 1TEP07-AF-0007 (see Section III.C), it was found that even though these valves t have natural frequencies close to the 24 bz range, the valves exhibited no instability and did not act as a source of excitation responsible for sustained high pressure fluctuations observed during any of the controlled tests.
These tests also included lowering the diaphragm pressure i to 19 psi to simulate low force margins that might have possibly existed due to a leaky diaphragm case. Lowering the diaphragm pressure did not create any pressure fluctuations in the system.
It is concluded that, even though the valves were unable to excite sustained pressure fluctuations in the system, it is prudent to raise their natural frequency and plug holding fcrce margins in both the open and closed positions to avoid any possible interaction at 24 hz frequency, which might occur under conditions that were not tested.
rm 01481/0005i L _--
I This has been_done by putting a stiffer spring with higher
~; preload and raisin
- /3 allowableof48ps$thediaphragmpressuretothemaximum for these actuators. A comparison of i
.V the old vs. new actuator stiffness curves and the axial !
natural frequency-of the plug is shown in Figures 5 and 6. l As can be seen, the plug natural frequency varies from 29 l hz to 35.3 hz for the new actuator, which is sufficiently away from the 24 hz hydraulic acoustic resonance of the piping system to avoid any possible interaction.
B. Valtek Crossover Valve (FV 7518)
The Valtek crossover valve FV7518 is different from the WKM crossover valves in certain respects. .A drawing of this valve is included in the Appendix. The crossover valve was changed from the original specification to accommodate higher pressure generation by the AFW turbine in the D Train under possible turbine overspeed conditions. It is also a cage-guided, balanced plug design actuated by a 200 i sq.in. piston actuator. The actuator has a fail close spring arrangement using two nested springs. It is a 4" 1500# Class valve with a 3.5" seat diameter. Compared to the WKM balanced plug valve design, it has a larger unbalanced area, A1, of 1.423 sq.in. on the plug. The stem unbalanced area,2 A , is 0.994 sq. in. Also, instead.
of using lower regulated pressures in the range of 31 psi as in the case of WKM crosstie valves, this' piston actuator uses the full supply pressure which is app aximately 100 (9
U psi. The valve is installed with the flow direction over the plug.
The implicat*ons of the higher supply pressure, the larger unbalanced area on the plug, and flow direction (over the plug) are that the opening forces on this valve are much' higher than on the WKM valves. Also the system pressure assists in the piston opening action, thereby causing much faster opening times. This creates a very sharp hydraulic transient in the system which has caused water hammer events on several occasions.
Even though this valve / actuator combination can be the initiator of the sharp hydraulic transient which results in a water hammer, it has a much higher force margin to hold the plug in the fully open position. This opening force margin is calculated to be 16,955 pounds, and it will take a hydraulic pressure transient of 17,057 psi to cause this valve to be destabilized once'it is in the fully open position. The natural frequency of this valve and actuator combination for vibrat1ons in the axial direction is calculated to be 61.6 hz. This is much higher than the WKM crosstie valves' natural frequency of 26.6 hz.
1 01481/00051
It is concluded that, even though this valve has been the culprit in initiating sharp short-duration transient
[] events, it is not susceptible to continued vibrations under !
C/ the anticipated pressure transients. The maintenance work j history of this valve shows some damage to the internals I which is directly attributable to high impact forces caused i during the operation of this valve. Action has been taken to slow down the valve response times.
8.5 Valtek Flow Control Valves All four of the flow control valves (FV-7523, 7524, 7525 and :
7526) are motor operated, high differential pressure service '
valves designed and manufactured by Valtek. The first three ;
valves are 4" 900 # Class design, and the FV-7526 is a 4" 1500 #
l Class. The difference between them is only in the weld end preparation to accommodate different pressure class ratings; the internals of all four valves are of the same design. A drawing l representative of all four valves is included in the Appendix. 1 These valves use an unbalanced plug with a special trim to avoid ;
cavitation under high pressure drops. The cage assembly consists of four concentric sleeves with multiple radial hole !
patterns arranged in several rows along the length.
Circumferential channels machined on the outside diameter of the sleeves intersect these radial holes. Figure 7 shows the details of the valve internals.
1 The individual sleeves are held together in a cartridge assembly p)
(
in the desired rotational orientation with respect to each other !
by an assembly pin. In the assembled position, the radial holes l in the sleevec are circumferentially offset with respect to each other to create a torturous, high-resistance flow path consisting of a series of radial holes and circumferential channels. The fluid flows from the outside of the cage to the inside, i.e., flow is over the plug in all four valves. l Review from Classical Stability Standpoint: From the standpoint !
of stability of operation, unbalanced plug control valves with flow over the plug and having a spring assisted actuator for closure are known to be most susceptible to unstable axial vibration. The specific criterion which has been demonstrated to result in instability is that the slope of the negative axial plug force gradient curve (due to fluid forces on the plug) must exceed the positive spring stiffness when the plug is moved from the fully closed to the open position. This results in a point J of inflexion in the combined force curve which causes 4 oscillation around that point of operation whenever the spring {
opposed actuator tries to approach the point.
]
l )
01481/00051 l
E __
t However, the control-valves under investigation are motor l operated with a solid mechanical connection between the stem 'and l (7 the stem nut in the Limitorque actuator in which a set of thrust
! (,) bearings provides axial reaction both in the up and down i direction. The thrust bearings for the SMB type of Limitorque l actuators are rigidly mounted in the housing, unlike in the SB '
type of actuators in which the thrust bearings float on a preloaded Belleville spring assembly, to reduce overload magnitude due to inertia of the moving parts. Thus the plug in the SMB actuators is held in any control position by essentially an infinitely stiff spring as compared to the magnitude of l negative axial plug force gradient on the plug due to fluid '
flow, i
This condition results in an inherently stable control valve, which is immune to the classical axial instability problems encountered in the air or hydraulic actuators with spring close action.
Excitation Source: Even though the motor operated control valves were found to exhibit no signs of classical instability as evidenced by no perceptible external motion of the stem at any control position during testing, it was found that some of these valves do indeed cause small magnitude pressure fluctuation.s directly upstream of the valve when the valve is in the cracked open position and the bottom end of the plug is '
'still in the vicinity of the seat area below the first set of holes in the cage. Frequency spectrum of these pressure (N fluctuations showed that they had a magnitude of 5-6 psi at the i, ) frequency of concern, i.e., 24 hz.
The stem positions at which these pressure fluctuations appear to reach the maxima were also found by individually testing each valve with small increments in stem position by rotating the j handwheel. It was also observed that these pressure 4 fluctuations around the 24 hz frequency were present only in valves A and D and were found to be negligible or absent in i valves B and C. When both valves A and D were cracked open to I these positions of maxima, the initially small pressure ~
fluctuations quickly excited up to +/- 1,000 psi range fluctuation in the piping system due to coupling with the organ pipe acoustic resonant frequency of the system. It is concluded .
that the flow control valves operating near the cracked open !
position, when only small clearance flow around the plug is i present, are a definite source of excitation. Even though all four valves are designed and manufactured to the same dimensional drawings, manufacturing tolerances specified on the ;
various dimensions can create some subtle differences in the !
interrelationship of the geometries of these parts defining the ,
flow path near the plug closed position. All four control {
b v
l 01481/0005i j J
valves were disassembled, visually and dimensionally inspected.
The inspection results and their implications on valve plug / seat / cage t,ehavior from the standpoint of instability, are O discussed later in a separate section.
The valve manufacturer was contacted to find out if they had observed similar performance in any other application of these valves. The manufacturer stated that they have had some l vibration problems from these valves under low flow conditions, I but that the frequency of vibration has been much higher, typically in the range of 2 khz.
They have identified this type of problem by the associated noise created in the system. It should be pointed out here that without proper instrumentation it would not have been possible i to detect the small pressure pulsations in the 24 hz vicinity l generated by these valves. Thus the manufacturer's statement l does not contradict the observed behavior.
Excitation Mechanism: The flow path between the cage, plug and g
the seat area is essentially an axisymetric path under nominal conditions. Recent work in flow induced vibrations in axial or i annular flow geometries shows that annular flows can lead to l self-excited vibrations of structures. There are several basic mechanisms that are responsible for lateral excitation of the l cylindrical body confined in a closely guided cylinder bore. ,
One of the principal mechanisms that has been identified is '
negative fluid damping as the center body is displaced from its concentric position. This situation is present in smooth flow O passages with a high pressure recovery geometry. Parallel or gradually divergent flow passages characterize such high pressure recovery geometries.
Both theoretical and experimental work has also shown that low pressure recovery geometries, e.g., sudden expansions, increased surface roughness provided by serrations or concentric grooves on one of the surfaces, can effectively inhibit lateral vibrations of structures otherwise susceptible to them. Such geometries tend to create increased turbulence which causes positive damping when the center body is displaced laterally.
Extensive research in this recent but important field of fluid structure interaction pertaining to fuel rod assemblies, slip joints, annular diffusers, and conccatric cylinders shows that self-excited vibrations can occur in relatively low frequency ranges, e.g., O to 100 bz. These vibration phenomena are present in the case of both compressible and incompressible flow media.
9 01481/0005i
Although the basis of self-excitaticn mechanisms are simple enough and well understood, quantitative and analytical I
e predictions of instabilities are not within the state-of-the-art; and, almost invariably, experimental testing of suspect geometries is resorted to in properly determining the self-excitation frequencies and the magnitude of response. The main hindrance to analytical solutions is the accurate calculation of the fluid forces associated with the flow along the always narrow and often variable geometry flow paths.
1 It has also been reported that very small changes in the geometry or eccentricity of a previously stable design have been !
t found to result in an unstable design. This fluid elastic I I instability is believed to be a possible excitation mechanism that is present in the Valtek flow control valve geometries, especially with the plug near the closed position and up to the travel range where the first set of holes in the cage is exposed by the plug. Once these holes are exposed, a relatively high level of turbulent flow is established in this area of the plug which provides positive damping and suppresses self-excited
( vibrations.
Another possible mechanism that can cause small pressure fluctuations upstream of this valve is jet switching at the exit of the plug-to-seat geometry. Under low flow, high differential pressure conditions, the flow from this area is not axisymetrically uniform. Inspection of the valve internals, especially the inside diameter of the cage which is made out of a relatively soft material (316 stainless steel), shows that the 9 flow patterns from the bottom of the cage appears to concentrate in a series of equi-spaced jets exiting the seat area when the plug is near closed position. It is possible that some sort of instability exists in this exit area which can cause switching of a jet between two alternate positions at a defined frequency. Again, though no analytical means are available to predict whether such jet switching is occurring and the frequency of it, it is considered a plausible mechanism which can cause some upstream pressure fluctuation at a defined frequency.
Regardless of the actual self-excitation mechanism that is present, if its frequency coincides with that of the acoustic resonant frequency of the dominant organ pipe in the piping system, severe pressure fluctuations can result due to coupling effects. Instrumented tests conducted during t% root cause investigation have confirmed these flow control valves to be the source of excitation, and either one of the above two mechanisms can be responsible for it. As described below, additional insight has been gained by visually inspecting the parts and making dimensional comparisons to understand the local phenomena and devise means to eliminate it.
O 0148i/00051
Dimensional Inspection, Comparison and Observations: All four m valves were carefully disassembled and visually inspected for
[d\ any signs of clues and other information which could help explain the different behavior of these valves manufactured to conform to the same drawings. From an overall inspection l standpoint, it was found that all the parts were within the
{
manufacturing tolerances and no manufacturing anomalies were j present. Detailed dimensional inspection results were reviewed )
and are shown in Figure 8 in a greatly magnified scale to show the interrelationship of the plug, the seat, and the cage area for all four valves. The average annular width of this flow passage is approximately the same for all four valves (0.003" to {
0.004"). {
Valves A and D, which had exhibited pressure fluctuations in the 24 hz range during controlled tests, have a relatively smooth channeled flow passage at the interface area between the plug, the seat and the cage. On the other hand, the geometrical relationship between these three parts clearly defines a relatively large expansion chamber at the bottom of the plug just above the sealing band on the seat for Valves B and C.
This causes a sudden expansion of the annular flow stream that is channeled down between the plug and the cage below the first row of drilled holes. This expansion ratio is approximately 2 to 1 for Valve C and 4.5 to 1 for Valve B; and the expansion chamber spans a considerable length; about four to eight times the average annular width of the flow channel for these two valves. The implication of this observation is that Valves B
[] and C suffer a high pressure loss locally due to sudden V enlargement of the jet whereas valves A and D do not.
This makes valve geometries of B and C to be of low pressure ,
recovery and high positive damping type. As mentioned earlier, i these two features have been found to eliminate self-excited lateral vibrations of concentric structures with annular flow.
Thus, even though some other factors might also be involved, it is concluded that this is the most significant geometry detail that is responsible for the differences in the performance of these four valves.
Corrective Actions: In order to avoid the flow instability problems discussed above, two distinct modifications can be '
made. One is to avoid operation in this small but undesirable range of plug travel. This can be done by administrative control or by providing both mechanical and limit switch stops to prevent the plug from approaching the fully closed position.
The second approach is to make necessary geometry modifications I l in this localized area to avoid instability problems. The South !
Texas Project has incorporated both physical modifications into the design of the system. The mechanical valve stop and limit l 13 U !
1 01481/0005i I
L - - - - - - _ - _ - -
l l
l l
switches have been added to all four flow control valves. In addition, geometry modifications have been made on the Train A l f] and D flow control valves by machining the valve seats to create V the expansion chamber present on Valves B and C.
1 i
O i
1 1
l ll T
01481/00051
_ _ _ . _ - _ _ - _ _ ___ _ _-___-_-_________ _ _- - _ _ A
NORMAL MAM FLOW CHECK VALVE--OPEN
(',) P'LCf VALVE 4LOSED CASCADE VALVE 4LOSEO
% .; 1; NORMAL Ft,0W AANGE' CHECn ; l'
~ ~ ' ~
I I low FULCAUM 8 TLC? VALVE CASCADE 8'STON MP *E[NAEaM WUM W i
LOW MAIN FLOW MAIN FLOW CHECK VALVE-At suYCM POINT Pit,OT VALVE PAAflALLY OPEN CASCADE VALVE 4 PEN'NG I l CwECK
'l yy + - g -- - +- l .
VALVE d - ,
_w gyp 433 l l
\ [ Q'*
~~~~
Y C CVs a b! FULCAUM p"O!VA VE CASCACE D'5*CN h L MAIN FLOW
- s;4e ogyo NO MAIN FLOW MAIN FLOW CHECK VALVE-CLOSED PILOT VALVE--OPEN p CASCADE VALVEMN V)
(
CHECK VALVE ;------
/ &k __
/. . . . . . . _
gyp 433 FULCRUM - -- - - - - - - - - - ---
. mtOT VALVE CAsCAce asrON u=N Ftow FROM PUMP FIGURE 1: FLOW / SWITCH OVER CHARACTERISTICS OF YARWAY ARC VALVES
/^
x h
,N Eh,-
, m
- iI '
) c5ldv '
%) ,
~
~
'2 W"M '
S' f ? .
'^ '
's 7
I g aM _
W
__ ."f -
- 4. S
c t
+
@ ,d QiW
, {)
u b
~
g Gig = =_
+ 4
.M -
s qr yg >
/. ~-i
' '~
f-
} [. M ?<t 5S (C' M- T g:
J v
-f \ -
s g -
' 9
- r. -
- _ l- c-
, L ,, *
/ .-y. - -,-5) .,_
I l
FIGURE 2: W-K-M CROSS 0VER VALVE ASSEMBLY ,
I 1
5 3
_ = _ _ _ _ _ _ _ _ - _ _ _ _ _ _ _ _ _ - _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
j s -
l V, ,
l 7000{l2.6 o)SPRWG RA7E TEC7 or4 .
PV- 7Sl7 A c.T V A, o R. "
f rJ foco fyg,4)" TH E MkWTEHnNcc. SHo P , y i
~~~~ M.6 A W CAEAs *J C A A PRE SSvfE.
(9NAlb $7 EM M rh4 4) d.
f f. .; 7 B E(.REA si tJ G Ait. Pres su AE.
500 J <{l7,9)
//, 8 (0ovNweA,3siCM u rn oN )
[ /
7
. , e"i e 4ooo ~(I4,5)E 9 o 0' g j )
__ th i- 9
' 4 4
%+
$* i
} iLS Y !
4 30 D {lo.8) [ l
, oo f (9.7 f St )
I
' ~
0 , I l
2000 -47.I4) 9.'- (7 M v' _.
l
/ i l 1 food #d,(,) .
i i t i i i ,
l m
S Pit wa. _ 1 m vat.vE STft. ORE m, 4 PREwAf' ' "'
(2 *)
, i i , i l i ,
O f. o clo@ "o ** 2 o 8 "* * s . o ' ***
\
STE'M POSITION, QHCHEL} > -
FIGURE 3: SPRING STIFFNESS CURVE FOR W-K-M CROSSOVER VALVE ACTUATORS i I
I l
l
- - _ _ ___ __ ._ l
.r
~
I
-T sN N( ik '% erra
^^>A 2 i
P2 : ; /
441 ttt I b i - PLU G S EA L UNBALANcEb PLUG. ALAW OLES j AgEA, A , J 4 N N Z *~
(A Mv LAA.) p s j .
1 l
e p e-s . .
f- g s i i "
ff l sN h FIGURE 4: UNBALANCED AREAS yA AND A2 IN PRESSURE BALANCED PLUG l
t t
h
I l
,x D.
14'co o W
31300 -
l Aka corn 4;ned spriny mre R l0000 -
%N 3
1 e.u u~ , ,,. ,, ,, n , -
oco -
rate
.D fgyy g% '.'
1 b 000 "
% cid co mb.,s e =1~~~~~ '
seeong itnre.
451*
g s4 t3 f% 4411 g Q
t
~ -
,~
4 ooo -
CIAarespr; g~r~& ~~ ~
_ ' - - 34 .
(New moel,. rpnne etzh , stro 'L/m .
2.aeo ::. _
o (aid o.rantest spr >ng ca+e, too!Mn l 2 " sML-e \
g s t e I e i i caso aV % 4e Vn */a % '/a coen f
VAL VE PLUG POS/7/oM i
FIGURE 5: MECHANICAL AND AIR SPRING RATES FOR W-K-M CROSS 0VER VALVE ACTUATOR !
i i
l l
.(".
~
y 1 i
.s 1
1 l
4
- - i i
l:
jm- .
(t, )L.
4e e
o
<- ss' -
2 4.
h -Ncw AavA7
(
N Lk t
- 's N .
'O,'
qg s s
$k q .2 oLo Acrussin? ' ~ ~. ~ ~ _.,
\
,, 1 u
} '
.15 "
qb 2 " m ar _
2o , . . . . . .
C /ditfEh k k Id I N k,h ~ d/gA/
VALVE PCUS P0stT/oH FIGURE 6: NATURAL FREQUENCIES FOR W-K-M CROSSOVER VALVE ACTUATOR !
o
'l u;.
\
v- 4 t .
I
_,.mmm_ _ _ _ _ _ _ _m_ __.-----=---------A--- - - - - - - - - ^
i j~%
l \
's J '
hmA h A
- - - - - - .J .. :. L -xat 3
, W4 1
i
! ! ! $g i
. 8 & '
I l l l t -
s e n i PLUG (ncLy *
[ opsw) b
.- e
_,S 4 j ii!!' ~
$ + +- .,
b .h dl skkb d.h Is\ b i 7. s \
ql y 7i hyb be,$ Isi7 _ VALVE BODY N 'TldFg '
31 d19 A l l l 4 f '. j b ' PLV G 2 .-- CATE z!Jibsro- A Ezr co-a s s t ..
z- -. .
'S - S E A-f R ING l
- l l
r m n ~
I FLOW i
FIGURE 7: VALTEK CHANNELSTREAM FLOW CONTROL VALVE PLUG / CAGE / SEAT RING DETAIL ]
(Y l L)
3.1 4.2. !
g'4 .
' s' '
.[G k y
,/
1' ,
N ,'/
, @ k\ 4 i
Q6Th l A E R f
PLUG 1
,l 0
4.0
\\\\\\\\
4
\\\\\\\NN\\\\\\ "
4.8 Hip (- l. C
/////////// @-I 9 19 g, w x.
5 EAT ! #'
RiN4 j j
/ t* g+
3.0 37, 1 /
g' @ .
j
's e, @ dj
/' .
- 's S S N\\N \\\\\\W ' s \ \ \\\\\
so .;
, , , , ,g gf n
,- is s e.o . =uM,' 6s i 4 Illll))llll .
E% PA N 5 6 0N Ex PAN St0N CHAM 6ES CHAN8EA /j j ,, II/'
35 w i l
{
I I
,. w DIME 45fodS N l THooSA)DTH5 OF AN INCH ,!
lh 16 DEC87
[\ FIGURE 8: RELATIVE DIMENSIONS OF FLOW PATH BETWEEN PLUG / CAGE / SEAT RIi4GS FOR VALTEK FLOW CONTROL VALVES IN A, B, C AND D TRAINS
'l o
i
O 1
i i
1 l
SECTION IV SAFETY SIGNIFICANCE i
i O
E i
l 4
O 01481/00051
IV.A. Impact of Fluid Transients on System Integrity
(]
K/
As discussed in this report, the auxiliary feedwater system has been subjected to two types of hydraulic transients: (1) short duration single event water hammers generally caused by fast valve operation and (2) sustained dynamic vibration caused by the system's 24 hz resonant response caused by the Train A and 0 flow control valves.
In order to correlate actual damage with analytical results, three basic analyses were performed. These were a simplified water hammer analysis using a trapezoidal time history forcing function; a static analysis using peak displacement at selected locations; and a harmonic analysis using displacement time histories obtained via use of lanyards. The results of these conservative analyses were found to be acceptable except at two locatiuns in Train A and at several vent / drain connections. The two fittings were non-destructively examined and no rejectable indications were found.
The testing done by Southwest Research Institute (SWRI) concluded that the vent / drain / instrument tap valve assembly and pipe anchor failures were caused by cyclic fatigue. Based on the system testing done subsequent to SWRl's work, it has been concluded that the 24 Hz system resonance led to the hardware failures.
In order to determine the integrity of the piping system, all single and double-valve assembly connections have been surf ace examined since the last observed vibration event on December 12 with no rejectable indications being found. Previously examined fittings
'N' (tees and elbows) have also been re-examined. In addition, as (Q described in Section I.B, all fittings in the crossover piping in all four trains have been examined. Except for the manufacturing defect on the tee in the Train A crossover piping, no other rejectable indications were found.
In assessing the implications of the three analyses discussed above as well as the NDE results, the original design calculations were reviewed. These calculations showed design stresses to be below the endurance limit with a fatigue usage factor in the range of zero to 0.0005. These calculation results plus the fact that NDE showed no surface indications allow one to conclude that the total fatigue usage factor is less than 1.0 and that neither the short duration '
water hammers nor the system resonance has degraded the future integrity of the system.
The effects of the dynamic vibration were reviewed relative to other components in the system. The permanent plant instrumentation, pressure transmitters and flow transmitters are in the process of being recalibrates. The crossover valves were subjec*. to the highest vibration effects and were subsequently disassembled and inspected.
1 1
(
w/
01481/0005i 1
l L - - I
The Train B crossover va'ive was found to have a bent shaft. Based on this and a review of the vibration loading conditions, the valves in Trains A, B and C are being rebuilt with new parts including' the 9 plug, steam, yoke and actuator. The Train D crossover valve internals are being machined to repair damage done to the plug cage area. The pump has operated without any degradation of performance l and the vendor was contacted to review the nature of the event on the I pump. His recommendation was to perform a surveillance test to I assure that the temporary loss of suction pressure during the transient had not effected the wear ring clearances and thus the pump performance.
Section V.C outlines confirmatory testing that will be performed in Modes 4 and 3 to verify that the system components operate as required.
Therefore, based on the NDE work, stress calculations, the modifications to the system (described in Section V.8) and the confirmatory test program, the South Texas Project auxiliary feedwater system is considered ready for operation.
l l
l l
l 9
l O
l 01481/00051
i
\
i O ,
I SECTION V CONCLUSIONS OF INVESTIGATION O
1 0 ~ ~
01481/00051 .
u_-__-______-.
V.A. Root Cause
['
X The.results of the testing. program described in~ this report have demonstrated that the Train A and D flow control valves have the capability.to' cause-the auxiliary feedwater fluid / piping system to go
'into sustained.24 Hz-resonance. The resonance phenomenon occurs because the physical piping lengths of the crossover piping create an
'". organ pipe" effect which magnifies the 24 Hz pressure pulsation
. created.in the Train A and D flow control valves. This resonance provides the cyclic driving force which caused the hardware failures
[>
' described-in.this' report. ,
inrough the various events, the Train A and D ' low control valves'had been in throttled positions. However, subsequent testing has shown that there is a single critical position that causes the 24 Hz ;
' pressure l oscillations to be caused by the Train A and D valves but
.not-by the Train B and C valves.
IJ Summary of Relevant Test Results'
.The 24.Hz resonant response of the piping / fluid system was recorded during the December 6 testing and is shown in Figure V.A.1 which is a spectrum analysis of the pressure transducer signal-immediately upstream of the Train A flow control valve. ,
The initiation of the resonant condition is depicted on Figure V.A.2 H which was obtained during the testing on December 9.- The resonance l was. developed to a maximum amplitude about 20 seconds after flow was A established through the Train A FCV in its near-seat position. The-Q highest peak-to-peak pressure oscillations occurred in the Train A'-
. discharge piping and crossover header. Of significance is.the fact j
that no water hammer or initial " kick" by the operation of a 'l crossover valve was required to trigger the event. !
- n order to further examine and isolate the resonance phenomena,. the-testing on December 12 centered on the' behavior of the four flow control valves. This was done by using the hand. wheel on the v'alve to unseat each valve and open it in small increments. This testing j showed that the Train B and C flow control valves did not possess a l dominant frequency that caused any system resonance. A typical' set ~
of traces for the Train C valve is included. Figure V.A.3 shows the ;
spectral response at near-zero flow while Figure V.A.4 ~shows the ;
spectral response at 100 gpm. As. stated, there is no dominant- !{
frequency for the flow range tested. -!
To test i,5e sensitivity of the resonance phenomenon to the relative poseion of +he Train.A and D flow control valves, the two valves j were.,, laced in the near-seat position that previously caused system j resonance. As c'in be seen from Figure V.A.5,- it was possible to repeat the event during the December 12 testing using the Train B pump. The sae res7 nance occurred when the Train C pump was used.
I o -
i 01481/00051 )
w.U - _.- w.a.-__--._.A---__ _ . - ~ _ _ - . _ ~
In subsequent testing on the same day, it was found that the resonance was unaffected by operation of the crossover valves, again O pointing to the Train A and 0 flow control valves as the problem. It was clearly demonstrated that these valves were generating the 24 Hz oscillations when testing showed that opening either valve off its near-seat position greatly reduced or eliminated the oscillations.
Figure V.A.6 shows the Train A development into the resonating condition. Figure V.A.7 shows the oscillation was completely eliminated after the Train A flow control valve was opened further to deliver approximately 50 gpm. A similar reduction occurred when the Train 0 flow control valve was tested.
In summary, the pressure pulsations in the flow control valves were observed to dissipate when the valves were opened to deliver approximately 50 gpm or more. Therefore, the root cause for the hydraulic resonance can be eliminated by avoiding operation at flows l less that 50 gpm through a flow control valve.. As described in the l following section, hardware modifications have been made to prevent the flow control valves from being placed in a near-seat, highly throttled position.
1 O
O 01481/0005i
O - - - - - - - - -
l Z
H L
C H V 0 0
1 l
l E l T 0 S 0 E
T
+ .
0 h D 0 0 E
E 5 I T
F A I
R O T L
U B
I U
M G
7 N 8 I .
- R 6 U
- D 2
1 A B E E !
T / /7 F U U3 e 2 7 6 6 -
94F p.
8".'-
1 40'
- 91l 0
2 N
F 55 -
H - - . - - - - - -
1 A C i X
4 G A M 9 T S I U A S 61 R E
A '
N A
O
,y 1
(< ,
O . :
E . :
3 8
+
8 8
2
. T F
3 4
8 1
O AB E E T
A
/ / 1
, {U U .
L s 66 8 9 .
E ;I e H T
- - H A
ec N 63 E 83 H C
c t
2 V
E 53 -
a 1 - _ _ ~ _ _ _ _
v ,
9
- _ _ ~ - _ _ _ _
4 2 A R 9 1 ,
1
- u T R ;
s6 1 I t '"
R -
M R
, jlI O
h I
' i/
f
O -
N N . - .
O E - - - 5 0
,' W 5
H C 4 L
U h
T ,' y 0
S .
4 E
T D
E ,' 5 E 3 F
e0 I
T L i U E ,'
s W 3
M 3 ) e T 7 7 8 F 3 NW 8 - 8 ) 0
- 7 .
8 4,'8 5 2 K 2 1 8 1 I
- 3 -
)
2 8 O 5 +
1 3 4( '8 (
W 0 2
1 ( C W
W N )
F A 8 A C ( 5
- (E W 1
X -
G K )
T A 8 A E (
E P,'A ,W 1 0
/
VH Z 8 )
1 M P
3 8 L G
. 8 84 E N ,I5
- D
.f i 6. 5 I 3 1. N A A
,.I f.1 ~
6 _ E AN T
f
{P T4 K g
} s m:'.P' s l, ~ .' .? . } -
O 9 '
l I 9 !
O H C d
5 L -
U T ,' 0 S 8 E
T D
E ,' 'S E ) 3 F T
- F I 8 I T 8 L 8,' S 0 U E 1 P $
M - 3 3 8 2 7 8 8 ) 0 8 - 7 8
- 8 ( 5 Z 2
1 2
- 2 1,'D
(
H )
Id l2 4 H 2 A 8 1 C 8 S,'( 1 0 3 ,$
1 K C
- A O W F
A E
P 8 4I E
)
( 51 X 4 G 2 )
T - 8 F (
A A,'A E .'O l
/ Z 8 )
U H 1 M P
3 8 L G 88 E K
- 8 ( f . 5 65 H l 1; 7 A A 9 53 H R
{ 2 C T '
1 ' - - - s e
l - '
- - . - 4 . q e - ~
9 A n 3 .
a '
6 H. 1 e '
w n
H
O E a. 31
~ 00 0 o 00 00 00 8 4O4 0B o 00 84o48 00 8
H L
C U
, ' R Y S N
'D IO T
R S D LA N L Al C
E E , '
1 S 'N O S T 3 3 N S 0 0 ) l E
E + + 0 AV T 00 D 50
(
D CRL ETAV O
E 22T E
') i S
F N 0 E 2
- V 1 5L I
E (
O T
L A f ')C T U
M -
O 0
i Y.R N 6
G N
3 EO 2 I
(
B RC 7 N U 1
I q 'O
)
8 GW ABE (
I O
7 EEB 8- / / A F L U U7 )
F 2 3 1 662 2
- 0 8 -
- - F
@M
(
8 1
.A H 6'. 0 I 5
3 911 A 4
1 R 1 W H J - - - _ _w - -
R E
~ - .
C D m
, F _ - - - - _ _ - - - ~ - - - - . -
A 9 -
A R 7 X B R 1 G - U T T S 61 I H
OMH ll Illlll l
~ **
l
' ' . . ~ , . -w';. ~ 4. ; ; -
....g..9,..
. . . .: . r.M :
iSd '
o o o o o o o o o o o o
O l o
m o
O o
N o
O o
V o
T o
m o
N o
W o
Q m
l e
Z I
z o O__
- - - ~ cn O I Q_ I i
ld I
. . . .w , >
- z. I W . . . .m e <(
>- m m F ld M e e M W+
8 o E c2 9 e .e wm<C <C w~
W e- e .
Dw O z e N
W 7
- w 0 mI
' w. e --) 1--
O e . .n -
e
- 3
- 1 z v (O T ~ m Z
~ 5
, .m . .~ O 42W v 6.-
N 00W g WCD s qg q, ,, J N 8 ::> ::> N ^w _j
" M E --
g O i 3 .e N. i u, .g3 . -
N
.e. g y cn
@4 2g O
. M e
-.w wW*
gp g " Ow l g 4' a
m 5 .
i i
i i i i
i i
.i .i i
? C h X ""
e -2 .= -
T H . Me M Oi m g
5o 0 0 o 0 0 0 0 0 0
~ 0 0 o g
0 0 0 0 0 6
1 2
1 4 4 8 2 6
_ 1 1 H
L C
_ U l
_ l
)
E t l
l5 0
T 3 1 S 0 (
E + O T 0 0 )
D O 0 E 2 E 0 l
1 F (
I C T T )
L R 0 U E 5 l M V 1 E C 4 B 1
F 1
O )
2 O Q '5 l2 A BH C T 7
8 / /EEE AR A 2
- U U7 ) T 3 MS 1- 6 02 P 2 -80 +F - dB lI 1 (
c 9A NR 68 I E W
3 9 1 AT 1 RF 2 H T 1 W C
, F A
4 -
A R 9 X 1 G ~ U T H T S 6 1 I
ONt H
H
, ,i
V.B. System Modifications
('~'} This section summarizes the modifications made to the auxiliary C/ feedwater system as a result of investigations into the single-event water hammers and the resonant vibration phenomena.
- 1. Additional high points vents As a result of the initial water hammer events, the layout of the discharge piping was reviewed to determine whether there was the potential for an air-induced transient. Although there was a vent at the highest portion of the piping in Train B, three vents were added in the crossover piping of Trains A, C and D.
Two additional vents were added in the discharge piping just downstream of pumps 11 and 12. Venting test 1TEP07-AF-0002 was performed to verify that the added vents provided additional air removal (compared to the original vent procedure). ,
- 2. Removal of second block valve As described in Section I.8, all 1" connections on the system were non-destructively examined. The NDE work revealed nine double-valve connections with rejectable indications. These connections were repaired by replacing the schedule 80 1" piping (C.S. ASME SA-106 Gr. B) with schedule 160 material and by deleting the second block valve for the repaired vent connections.
p 3. Replacement of Air Diaphram Operator on Crossover Valves A higher spring force and air pressure actuator has been installed on Trains A, B and C. This provides greater margin away from the 24 Hz system fluid natural frequency and a higher closing force.
4 Slowing Down of Train D Crossover Valve -
During testing the Train D crossover valve was found to have a fairly fast stroke time. The valve actuator has been fitted with a needle valve in the air line to increase the stroke time. l S. Pipe Support Additions and Modifications Although the sources of water hammer or other dynamic effects have been eliminated, additional pipe supports were added to ,
harden the system from any future effects that could result from i unforeseen events.
l l
L) 01481/00051
- _ - - - - J
l l
- 6. Installation of Travel Stops on Flow Control Valves
(~')
I,, ,/
Based on the test results and evaluation, the FCV's have been limited in their closure position to avoid the 24 Hz resonance condition. This was accomplished by use of a close limit switch backed up by a mechanical stop. This feature will be in place until completion of Item 7.
- 7. Flow Control Valve M, modification A modification of the flow control valve will be developed and tested on Unit 2 to determine if the need for the positive stop can be eliminated. This modification will then be installed on Unit ~l.
,m f
RJ l
7%-
\j 01481/00051
V.C Confirmatory Testing
[D
%.)
in order to verify the proper operation of the auxiliary feedwater systems, a series of confirmatory tests are scheduled to be performed as outlined below: 1
- 1. The crossover isolation valves will be operated (opened and 4 closed) against static conditions to establish baseline !
operating times. j
- 2. The crossover isolation valves will be operated under full-flow l conditions to confirm their operating times.
- 3. The motor-operated pumps will be tested to determine if there has been any degradation in the head-flow characterisites (as recommended by the pump manufacturer). l
- 4. The flow control valves will be individually tested to verify that the resonant frequency problem has been e :minated. The i test will cover flows from the minimum established by the mechanical travel stop up to approximately 150 gpm. .
- 5. The system will be tested to confirm it responds properly to an ESF actuation signal.
The tests will be performed in Mode 4 although item 4 could be run in Mode 3.
p
\,
i i
l l
l /m U)
I i 01481/0005i 1
- _ _ _ _ _ _ _ . _ _ _ _ _ _ _ ._a
1 l
AUXILIARY FEEDWATER SYSTEM REPORT '
f]
1 Q Appendix Bechtel Drawing No.
~ i' l .1. Piping &' Instrumentation-Diagram SS149F00024.#1 R. 11 for' Auxiliary Feedwater
[ 2. Piping Isometrics for AFW 2C369PAF402 Sh. 01 R.6 Sh. 02 R.9 4G369PAF602 Sh. 01 R.12
'Sh. 02 R.10 Sh. 03 R.10 Sh. - 04' R.8 2G369PAF602 Sh. 11 R.3 Sh.'12 R.6 3G369PAF602 Sh. 14 R.6 Sh. 15 R.5 Sh. 18 R.5 ,
Sh. A01.R.6 5G369PAF602 Sh. 16 R.6 Sh. 17 R.6 3 Vendor Drawing No.
- 3. Valve Drawings: ;
t W-k-M Crossover Valve .RS258088 !
( FV-7 515. - 7516, ' 7517 ) '
l Valtek' Crossover Valve (FV-7518) A34753-20 Valtek Flow Control Valve A34753-4 1 (Typical) R I
- q ..
i 01481/00051 I
i
'e OVERSIZE DOCUMENT
~
PAGE PULLED SEE APERTURE CARDS NUMBER OF OVERSIZE-PAGES FILMED ON APERTURE CARDS qt '
APERTURE CARD /HARD COPY AVAILABLE FROM RECORD SERVICES BRANCH,TIDC I FTS 492-8989 1
..