ML13323B140
| ML13323B140 | |
| Person / Time | |
|---|---|
| Site: | San Onofre  |
| Issue date: | 06/03/1986 |
| From: | Correia R, Gregg H, Milano P, Trottier E, Zech G NRC OFFICE OF INSPECTION & ENFORCEMENT (IE REGION I), NRC OFFICE OF INSPECTION & ENFORCEMENT (IE REGION V), NRC OFFICE OF INSPECTION & ENFORCEMENT (IE) |
| To: | |
| Shared Package | |
| ML13323B139 | List: |
| References | |
| 50-206-86-15, NUDOCS 8606110333 | |
| Download: ML13323B140 (35) | |
See also: IR 05000206/1986015
Text
U.S. NUCLEAR REGULATORY COMMISSION
REGION V
Report No.:
50-206/86-15
License No.:
Licensee:
Southern California Edison Company
Post Office Box 80.0
2244 Walnut Grove Avenue
Rosemead, California 91770
Facility Name:
San Onofre Nuclear Generating Station, Unit 1
Inspection at:
San Onofre, San Clemente, California
Inspection Conducted:
March 17-20, 1986
Inspectors:
.___
_8_
Patrick D. Milano, Inspector, Special Projects Inspection
Date
Section (SPIS)
Richard P. Correia, Inspector, SPIS
ate
Edouard H. Trottier, Inspector, Reactive Inspection
Date
Section (RIS)
(t.H. I. Gregg, I,1 ector, Region I
Date
Also participating in the inspection were:
0. 0. Rothberg, NRR
H. C. Rockhold, Consultant, EG&G, Idaho
Approved by:
5.3 -3
Gary G. Zech, Chief,
dor Program Branch
Date
8606110333 860606
PDR ADOCK'05000206
I. SUMMARY
Inspection on March 17-20, 1986 (Report No. 50-206/86-15)
Areas Inspected: An announced inspection by headquarters of activities
associated with the design, procurement, operation, and maintenance histories
of the MCC-Pacific main feedwater system check valves which had failed,
thus contributing to the water hammer event on November 21, 1985. Also,
a review was made of the design and procurement activities associated with
the Atwood-Morrill replacement check valves and the function of the inser
vice inspection and testing programs. The inspection involved 174 inspec
tor hours by five NRC inspectors and one consultant.
Results:
No violations or deviations were identified. This inspection
report, in conjunction with inspections of the associated check valve
vendors, is intended to serve as a portion of the NRC staff's review of
Southern California Edison's response to the check valve related aspects
of the November 21, 1985 water hammer event.
DETAILS
1. PERSONS CONTACTED
Southern California Edison Company
(e)
W. M. Lazear
(e)
H. E. Morgan
(e/b/x)
M. P. Short
(e)
B. Katz
(e/x)
D. E. Shull
(e/x)
W. G. Zint1
(e/b)
T. A. Mackey, Jr.
(e/b/x)
D. E. Nunn
(e)
A. E. Talley
(e)
C. A. Kergis
(e)
M. Macoundray
(e)
L. Bajada
(e/b/x)
L. Rafner
(e)
D. Bruce
(e)
B. Woods
(e)
P. Cray
(e)
T. Herring, III
(e/b/x)
C. Chiu
(e)
V. Salvator
(b)
S. D. Root
(b)
B. R. Doncil
(b)
J. Blanco
(b)
B. Watts
(b)
J. Yann
(b)
G. L. Stawniczy
(b)
H. L. Richter
(b)
V. A. Gow
(x)
J. T. Reilly
(x)
M. A. Wharton
(x)
R. L. Erickson
(x)
D. A. Herbst
(x)
N. Maringas
J. Hirsh
P. Croy
Bechtel Power Corporation, Western Power Division
(b) R. Gavankar
(b) F. McCluskey
(b) J. Statton
(b) P. Cruz
(b) R. L. Loos
(b) R. G. Allen
(b) R. Elder, Consultant
(b) P. Tulles, Consultant
(b) L. Wiedemann
(b) J. Hosmer
(b) A. Langmo
NRC Resident Inspectors
F. R. Huey, Senior Resident Inspector
A. D'Angelo, Resident Inspector
R. C. Tang, Resident Inspector
e -
Denotes attendees at entrance meeting on March 17, 1986
b -
Denotes attendees at Bechtel meeting on March 19, 1986
x -
Denotes attendees at exit meeting on March 20, 1986
.
2. LICENSEE ACTION ON PREVIOUSLY IDENTIFIED ITEMS
This subject was not addressed in the inspection.
3. UNRESOLVED ITEMS
No unresolved items were identified during this inspection.
4. BACKGROUND
During the San Onofre Generating Station, Unit 1 (SONGS 1) construction
in the 1960s, swing check valves manufactured by MCC-Pacific Valve Company,
purchased by the design agent Bechtel Corporation, were installed in the
feedwater system. The original main feedwater system in San Onofre Nuclear
Generating Station Unit 1 (SONGS-1) contained eight (8) check valves:
one
in each of the two (12") main feedwater pump discharge lines; one in each
of the three (10") steam generator feed lines and one in each of the three
associated (4") by-pass lines. Each steam generator feed line and its
associated by-pass line has the check valve located in close proximity to
the respective upstream flow control valves. After the water hammer
event of November 21, 1985, an inspection of the feedwater system was
conducted. Each of the 10" and 12" Pacific check valves were found to
have failed or degraded in-service prior to the event. The 4" bypass
line check valve in the B line failed as a result of the event.
The current system repair includes replacement of the Pacific feedwater
check valves with Atwood and Morrill swing check valves. The system repair
- also
includes relocation of the (3) 10" check valves farther downstream of
the flow control valves and the installation of drain and vent valves to
facilitate leak testing. Full scale flow testing of the 10" valves at
an offsite facility at representative flow conditions and piping arrangements
has been performed. Finally, the system repair includes the installation
of (3) additional 10" Atwood and Morrill swing check valves inside contain
ment, one on each steam generator feedwater line.
Specific details of the design, procurement, installation and failure
analysis reviewed during the inspection are presented in the following
subparagraphs.
4.1 Review of the original MCC-Pacific (PAC) check valves
a. Valve Design
The inspection team reviewed the PAC catalog information, corre
spondence interchange between the licensee and the vendor, and
the licensee's report entitled "Failure Analysis of Swing Check
Valves," Rev. 1. Discussions were held with SONGS 1 and Bechtel
personnel. Additionally, several of the failed parts were
examined. The inspection team made the following observations
concerning the Pacific valve design.
2
The valve is a cast steel body, bolted bonnet design in
which the seat plane is approximately 50 from vertical in
the disc opening direction. The total disc movement from
closed to fully open against its stop would be from the 50
from vertical to a position approximately 50 below
horizontal for a total angular displacement of about 800.
When fully open the disc is nearly out of the flow stream.
This configuration tends to produce a low pressure drop
through the valve. However, there is the possibility of
the disc not seating tightly against its stop due to large
internal body bowl passage areas at the disc full open
position.
The internal parts making up the disc assembly consist of
the disc, a circular cast plate with the seat on one side
and a protrusion termed the disc pin on the other side; the
hinge which attaches to the disc at the disc pin and to the
body by means of the hinge pin; and a disc washer, disc nut
and disc nut pin to attach the disc to the hinge. The design
has built-in radial clearances between the disc pin and the
hinge and a longitudinal clearance between the disc, hinge
and nut. These clearances allow the disc to close against
the seat without an adjustment. However, with this built-in
play, the disc is free to rock and rotate when in the open
position. This can cause significant wear of the parts as
evident from the failed parts. Turbulence or specific flow
conditions also contribute to the disc to hinge movement.
Additionally, since the disc pin and nut are used as a stop,
the striking of the stop against the valve cover tends to
create additional clearances.
b. Purchase Specification
In a meeting attended by the NRC inspection team, Southern
California Edison and Bechtel Power Corporation, the original
architect/engineer for SONGS-1, the Bechtel technical staff
presented their check valve criteria and methodology.used
during the 1960's time period; the same period in which SONGS-1
was constructed. The Bechtel criteria based
the selection of
the main-feedwater system check valves on the general service
conditions required of the valves, the size of the pipe in which
the valves were to be placed, the system design pressure and
temperature, and engineering judgement and experience. Check
valves were preferably located in the piping system in a horizon
tal orientation, and in an area which would allow for maintenance
access. These parameters were utilized to produce a cost effec
tive layout and were derived from the designer's judgement and
experience. The specific types of valves to be used and the
manufacturer selected to supply them were accomplished via a
project master valve list and a general valve specification.
The master list was developed based on past experience and
industry practice at the time of development. The list contained
pressure ratings, material specifications and a mark number which
3
designated the type of valve (e.g., globe, gate, swing check, etc.).
In conjunction with the project master valve list, a general valve
specification formerly used on a fossil plant project, and each
bidder's published product descriptions, vendors were selected
for the types of valves required for procurement. No technical
information was exchanged between the vendor and the procurement
agent for the system in which the check valve would be placed,
the location of the check valve in the system relative to other
equipment, or flow conditions. No installation specification
was available at SONGS-1 which may have been part of the procure
ment package furnished by MCC-Pacific and there was no maintenance
or spare parts information.
The licensee provided the available documentation relating to
the purchase of the PAC swing check valves. This consisted of:
a general Bechtel specification No. BAL 560 for cast steel,
cast iron, and bronze valves for a fossil plant, the Alamitos
Steam Station Units 5 and 6, from which the valves were purchased;
the Bechtel piping specification Job No. 3246; a general valve
listing, part of No. BAL 560 and several pages of a valve
description; and one invoice page A-10335 dated 12/7/65 for a
10" Pacific swing check valve, Mark 222 Figure No. 680-7 WE80.
Specification No. BAL 650 was provided because the one actually
utilized was no longer available. Since Specification No. BAL 650
was from the same time period, it was believed to be similar to
that used for SONGS-1.
It was determined that the piping system including valves was
designed and purchased to ASA B-31.1. The piping specification
pressure - temperature service index and valve design rating
was 600 psig at 850 0F.
Bechtel and the licensee stated that the record retention require
ments of ASME Nuclear codes were not yet in place at the time of
construction of this plant. Thus, the licensee's information
was limited.
c. Failure Analysis Report
The inspection team reviewed the SONGS 1 report on "Failure
Analysis of Swing Check Valves, Revision 1," dated February 24,
1986. The inspection team concluded that the report provided a
general analytical approach for disc lift vs. velocity based on
an extrapolation of experimental information from another valve
vendor. However, the inspection team had concerns about the
small contribution the report assigned to turbulence. There was
no mention of the built-in clearances of the PAC disc arrangement
which would be further affected due to turbulence. Also, there
was no knowledge of the thru-porting areas to be able to
determine that the disc was against its stop under full flow
conditions.
4
Because of these factors, the inspection team disagreed with the
report's premise that the problem occurred solely due to the
licensee's recent operation at 92% rated thermal power (RTP) flow
conditions. In addition, specific aspects of this failure report
were reviewed by an NRC consultant, the results of which are
included as Appendix A to this inspection report.
The failure report has been revised and corrected in several
areas and resubmitted as revision 2 to the NRC. This revised
report properly recognizes turbulence as a contributing factor
to the degradation of the valve internals.
d. Observations of Failed PAC Valve Parts
The inspection team observed the discs from the (3) 10" swing
check valves FWS 345, 346, and 398. The seating surface of each
disc was in good condition. However, there was considerable
wear on the back of the discs, on the lower 1800 of the OD and
on the anti-rotation bars. There was considerable matching
wear on the hinge, and considerable wear of the disc pin hole
which resulted in a radial clearance with the disc pin of at
least 3/16 inch. There was also considerable wear on the back
of the hinge caused by the nut and on the end of the disc pin
where it strikes the open stop.
From the observed parts it was evident that the disc rocked,
fluttered and rotated. Initial clearances were greatly
increased by the wear, and the disc became free to rotate past
the anti-rotation bars.
This wobble of the disc to hinge arm
connection probably caused a changing angle of incidence
of the disc in the flow stream and compounded the problem
with stability at the disc stop.
e. Installed Location of PAC Check Valves
One of the 12" swing check valves (FWS-438 and 439) was
installed on each pump discharge and was located downstream
of a long radius elbow. The outlet side was relatively
unobstructed. The licensee's initial failure report portrayed
this arrangement as having no high upstream turbulence component.
Each of the 10" swing check valves (FWS 345, 346 and 398) was
installed approximately two pipe diameter downstream of a double
ported top and bottom guided control valve. This arrangement was
portrayed in the licensee's revised failure report as a major
turbulence producing component immediately upstream of these
The inspection team agrees with the conclusions of the revised
failure analysis that stated there will be strong turbulence
effects within the 10" check valves because of the control
valve located immediately upstream. The inspection team also
agreed that the long radius elbow upstream of the 12" check
valves has less of an effect.
5
During a meeting with cognizant SONGS-1 personnel on 3/18/86 the
inspector expressed concerns about the close location of the 10"
check valves to the control valves. The inspection team pointed
out the general valve placement practice of providing a separation
between these components of 5 to 10 diameters upstream and 3 to 5
downstream of the check valve.
f. Licensee Check Valve Review Program
Because of the observed failures with MCC-Pacific check valves
in feedwater service, the licensee initiated a program that
entailed a visual inspection of all Pacific valves and a sample
inspection of other check valves in turbulent, lower port flow
velocity situations.
A total of 29 Pacific check valves were disassembled and .inspec
ted. In addition to those failures resulting from the event,
one check valve, FWH-437, feedwater heater drain pump check
valve, was found with the disc detached due to a disc pin failure.
A sample lot of check valves for inspection was also developed
from calculations of system performance which would have
subjected the valves to a condition of being less than fully
open during operation. The sample consisted of 15 valves
which were also found to be in potentially turbulent flow regimes
because of being located less than 10 pipe diameters downstream
of a turbulence generating device. This inspection resulted in
no other valves which exhibited disc detachment. One valve,
SDW-002 in the services and domestic water system, was found
with excessive corrosion.
Finally, the licensee reviewed the plant maintenance orders
for all SONGS-1 swing check valves. This review identified
five check valves which had required corrective maintenance
for valve internal deficiencies. These valves were also
disassembled and inspected and no deficiencies noted.
4.2 Review of Replacement Atwood and Morrill (AM) Check Valves
The proposed modified main feedwater system at SONGS-1 is to contain
eleven (11) check valves:
eight direct replacements of the originals
and three additional valves which are to be placed one in each of the
three steam generator .feed lines inside the reactor containment. These
replacement check valves were procured from Atwood-Morrill.
The three
check valves which are downstream of the flow control valves are to
be relocated further downstream from the control valves.
Bechtel
Power Corporation and Southern California Edison have concluded that
the turbulence created by the flow control valves was a contributing
factor in the failure of the original MCC-Pacific swing check valves.
6
Current Bechtel practices and criteria guidance used in the selection
and location of check valves includes using a tilting disc type of
valve and locating the valve at a distance of 5-10 pipe diameters from
a turbulence generating source. The tilting disc check valve design
incorporates a shorter swing arc to facilitate opening and closing of
the valve and to reduce the radial velocity when the valve closes.
The location criteria of 5-10 pipe diameters away from a turbulence
generating source is to allow a majority of the flow turbulence to
subside prior to entry into the check valve to prevent banging of the
disc into the valve body and/or seat. Bechtel used velocity criteria
taken from Crane Technical Paper No. 410, "Flow of Fluids Through Valves,
Fittings, and Pipe," to determine minimum required flows through the
check valves in order to maintain the disc in the full open position
to prevent "dangling" of the disc in the flow-stream.
At a meeting held at the SONGS-i site, Bechtel engineering managers
and staff presented to the NRC inspection team their current design
standards for the SONGS-1 check valve replacements. Bechtel sizing
and location criteria were compared to Atwood-Morrill sizing and loca
tion criteria. Atwood-Morrill recommends a port velocity of 10-20 ft./
sec., and to avoid velocities of less than 5 ft./sec. or greater than
40 ft./sec. Bechtel has calculated that the flow through the replace
ment check valves will be 22 ft./sec. at 92% rated thermal power.
This value is slightly higher than Atwood-Morrill's recommended value
but Bechtel concludes that the slight difference in the actual versus
recommended values is acceptable. Atwood-Morrill location criteria
for swing-type check valves is to optimally have straight pipe at a
distance of 10 pipe diameters upstream and 5 diameters downstream of
the valve.
a. Valve Design
Based on the review and evaluation of documentation provided
by the licensee and Bechtel and additional review of catalog
information and the valve drawings, the following was determined:
o
The valves purchased to replace the Pacific valves and
the three new valves are 900# class, cast steel swing check
valves with pressure seal bonnet (internal pressure
produces a large force on the bonnet which moves tighter
against a steep tapered seal ring).
o
The seat plane is 200 from vertical in the disc opening
direction. The disc movement from closed to fully open
against its stop is from the 200 from vertical position
to a position 150 below horizontal (a total of 550 angular
displacement with the disc partially in the flow stream
at full open position).
This configuration may produce a
slightly increased flow resistance but the disc should be
firmly seated against its stop under most flow conditions
as is the accepted practice for check valves.
7
The disc assembly is essentially one integral part. The
hinge arm is cast integrally with the disc and there are no
bolts, nuts or pins. The design utilized close clearance
bushings at the hinge pin resulting in little built-in play.
Also, with this type disc assembly there will be no rocking,
or rotation of the disc. Adjustment for seating is by means
of the rotatable external bearing covers.
o
Although not classified as a tilting-disc check valve by
Atwood-Morrill (AM), their swing-check valves have a shorter
disc-hinge swing arc than the MCC-Pacific valves. Bechtel
felt that this shorter swing arc is very similar to a
tilting disc design and will mitigate check-valve slamming
in reverse flow conditions.
b. Purchase Specification
The inspection team reviewed the licensee's documentation related
to the procurement of the AM swing check valves. This included
a review of: (a) portions of the SONGS 2 and 3 specification SO
23-408-1 that was used as the basis for quotations and purchase
of the valves; (b) Bechtel purchase memorandum 531 with Bill of
Material (BM) requirements that was sent to several vendors for
quotations; (c) Bechtel purchase memorandum 532, and subsequent
revisions, for purchase of the valves from AM; (d) the initial
SCE purchase order No. V 82000603 to AM for the valves, and
subsequent change orders; and (e) correspondence interchanges
between Bechtel and AM regarding details of the purchase require
ments and several technical issues related to installation loca
tion, flow conditions, expected differential pressure for the 10"
valves outside and inside containment, disc weights, seat angle,
port diameters, and the drilling of a hole in the 12" valve discs.
The check valves were purchased as safety-related valves to
ANSI B31.1 (1980 Edition).
For the replacement.valves outside
containment, the valves purchased were three (3) 10" 900# valves;
two (2) 12" 900# valves; and three (3) 4" 900# valves. The design
condition for these valves was specified as 1350 psig and 420'F.
Each valve is to have butt weld ends prepared for schedule 80 pipe.
The (3) new 10" valves inside containment were also specified
as 900# but the design condition was 1210 psig and 420'F. These
valves are ordered with butt weld ends prepared for schedule
60 pipe.
The documentation was found to be acceptable. Several questions
were raised by the inspector for which a resolution was required
and later furnished by the licensee. These items are described
in the following paragraphs.
8
The replacement outside containment AM valves were purchased as
900# class for a design condition of 1350 psig, 420 0 F, whereas
the old PAC valves were 600# class. The new inside containment AM
valves are also 900# class at a slightly lower design condition
of 1210 psig, 420'F. The licensee was asked to provide the
rationale for the increased pressure class and justification
that existing piping and other valves are suitable for both the
old and new design conditions. Bechtel, by letter dated
March 27, 1986, provided the basis for the original check
valve 600# ASA rating as follows:
At the time of the SONGS-i
design, the applicable codes (1962 ASME Section I and 1955 ASA
B31.1) permitted either pressure or temperature to exceed
design for short periods of time provided the increase and
time was within the prescribed limits. Since (1) the plant
normal operating conditions of approximately 900 psig at 417oF
is well below the 600# ASA rating pressure at temperature
capability of 1320 psig at 420 0 F, and (2) the SONGS-i design
of 1350 psig at 420'F was based on main feedwater pump shutoff
conditions that would rarely occur and would be within the
permitted code variation limits, the code provisions were met.
The inspection team verified that the 1962 ASME Section I and
1955 ASA B31.1 permitted the above described variations.
During the initial purchase evaluation of the 12" AM check
valves a drilled hole was to be placed in the disc at a set
location. The final purchase document retained the hole
requirement but considerably more latitude was given for its
location. The licensee was asked to provide reasons for the
latitude in hole location. Bechtel by letter dated March 27,
1986, provided their response that: (1) there was no specific
process criteria involved in the hole location, (2) the hole
was to be located in a general area similar to the previously
installed valves, and (3) the manufacturer AM was given some
latitude to allow them to consider the particular disc design
and disc to seat configuration.
Based on an evaluation of the purpose of the hole, the above
response addressed the inspectors concerns.
During the Bechtel presentation concerning these valves, the
Bechtel representative stated there were numerous communications
with the vendor regarding the valve design and application
conditions. The inspection team reviewed the communication inter
changes which basically related to the 10" valve. The licensee
was asked for correspondence concerning the other valves (the
12" valve was mentioned). The licensee said they would review
this with Bechtel to see that all the correspondence was
presented and that the 10" valve may have been the valve with
the most bounding conditions. The Bechtel response of March 27,
1986, provided additional AM correspondence information. The
response also verified that the 10" valve was of particular
concern and that AM evaluated flows and the suitability of the
application.
9
The inspector determined that the information provided and the
commitment to have AM review the data for all check valves
satisfactorily addresses the inspectors concerns regarding
information interchanged between the licensee and vendor.
A question was raised concerning the flow coefficient, Cv for the
valves of both PAC and AM. The current PAC catalog provides
a Cv that appears extremely high. While the AM valves have a
somewhat lower Cv, in both cases the numbers appear to be higher
than the maximum theoretical value. Further, neither PAC nor AM
have a flow test facility to obtain or validate these values
emperically. This was not made an open item since the licensee's
evaluations did not utilize the vendor's Cv values, and actual
flow testing would be performed.
C. Installed Location of AM Check Valves
As mentioned previously, the inspection team's concerns about the
original close location of the 10" check valves to the control
valves was discussed with SONGS-1 personnel on 3/18/86. At
that time licensee personnel stated they were not relocating
the 10" AM check valves. However, on the following day, 3/19/86,
the Bechtel presentation included the subject of relocation of
the outside containment 10" check valves to a new location
approximately 8 diameters downstream of the control valves to
reduce turbulence in these check valves. The licensee said this
was the maximum distance available.
The new inside containment 10" AM check valves are to be installed
with essentially no upstream or downstream obstruction. However,
under conditions of auxiliary feed flow of only 50-200 gpm, there
is an uncertainty by the vendor of the 10" check valve ability
to satisfactorily function. In a letter to the licensee from
AM dated 2/20/86, the vendor stated that flows of 165 gpm or less
are considered severe service which could result in seat damage
from the disc oscillating closed. Consequently, the licensee
included low flow testing in his full scale testing program
discussed below. Based on the results of this testing and the
commitment by the licensee to open and inspect one of each set
of valves during the next refueling outage, the installation
location and valve capability are considered adequate.
d. Full Scale Testing
Southern California Edison (SCE) and Bechtel Power Corporation
have conducted and are continuing to conduct testing of the new
Atwood-Morrill check valves.
SCE feels that testing the valves
under conditions which simulate the feedwater system flow conditions
at SONGS-1 has verified proper operation of the valves, has
verified back-up calculations generated by SCE and Bechtel for
the feed-water flow conditions, and will provide data on wear
to facilitate anticipated maintenance of the replacement valves.
10
The first phase is complete and consisted of performance tests
through a 10" AM valve at the low flow ranges representative of
the auxiliary feedwater flows. Determinations were made regarding
the stability and ability to function of these large 10" valves
at low flows.
Flow rates were then increased to the higher normal
feedwater system ranges to determine where the disc seats against
its open stop. A discussion of these tests is provided as
Appendix B.
The second phase of this research is a long term wear study for
these valves under anticipated operational conditions.
Under a cover letter dated May 5, 1986, SCE provided the final
report of the performance tests conducted at the Utah Water
Research Laboratory. The tests indicated that the valves would
experience tapping during a range of flow rates. However, the
intensity of the tapping was not believed to be severe enough
to cause damage to the disc or body. The test conducted at low
AFW flow rated indicated that the valve would be stable.
e. Industry History with Atwood & Morrill Check Valves
During the inspection, the actions that SCE performed in review
ing the available history that the industry has encountered with
valves similar to those being installed were reviewed. This
information included that available from the Institute for Nuclear
Power Operations (INPO), USNRC IE Bulletins and Information
Notices, and direct contacts with utility maintenance representa
tives where the valves were being utilized.
The SCE Industry Surveillance Evaluation Group (ISEG) performs
the reviews of the received NRC and INPO information along with
support from the affected departments to determine the applica
bility to SONGS. If applicable, the information is transmitted
as an action request to the applicable department, and ISEG
tracks the request until a response is provided.
Because of the event, the SCE Reliability staff performed a
more detailed review of the INPO and Nuclear Plant Reliability
Data System (NPRDS) data on the Atwood-Morrill check valves.
From this review, SCE found that there were a total of 558
Atwood-Morrill check valves in the industry, with 179 reported
failures. The majority of the reported failures were in check
valves which had been in air and steam service. SCE therefore
concluded that they would expect no major problems with the
Atwood-Morrill check valves which they were placing on the
feedwater system at SONGS-1. The information obtained from the
NPRDS data on Atwood-Morrill valves was to be used as an
enhancement to SONGS-1 maintenance of the valves in question
in that SCE felt that they could better determine expected
problems rather than waiting for problems to arise.
0
11
As a check to NPRDS data, the licensee contacted a number of
utilities that utilized Atwood-Morrill check valves. These
contacts confirmed the NPRDS conclusions of no trends in
failure of the type of valves proposed for use.
5. Control of Purchased Material, Equipment and Services
As part of the review of San Onofre's control of purchased material, equipment
and services (Criterion VII of Appendix B to 10 CFR 50) receipt inspection
documents associated with the purchase of eight Atwood-Morrill check valves
(2 -
12 inch, 3 - 10 inch, and 2 - 4 inch) were examined.
All receipt inspection documents associated with the subject check valves were
found in Receiving Inspection Data Report (RIDR) No. RSO-1235-86, dated
February 13, 1986.
As reviewed by the inspection team, RIDR was found to
contain:
Certificate of Compliance attesting to conformance to B16.34
(1977), ASME Section IX, and Bill of Material Revision 9; Certified Material
Test Reports, hydrostatic seat leakage and operational test results; final
inspection data sheets; and an instruction manual. In addition, a Document
Discrepancy Notice was completed to substantiate that the Certified Material
Test Report for 12 inch check valve S/N 1-15487-02 arrived without the
required heat numbers.
(The CMTR had been torn in transit.)
All receipt
documents appeared to be in order.
6. Document Control
To verify proper control of the instruction manual for the new Atwood-Morrill
valves, the inspection team pursued the review and approval status of the
manual that was included in the RIDR package. According to the SCE Corporate
Document Management (CDM) System, the Atwood-Morrill instruction manual was
received (via Bechtel Power Corporation) on February 24, 1986. After initial
processing, the manual was sent to the Configuration Control Group. Following
review by Configuration Control, a copy of the manual and two configuration
control forms (183 and 184) were sent to the following SONGS departments on
March 25, 1986:
Operations, Maintenance Engineering, Maintenance Support
and System Descriptions. (Information on Form 183 is prepared by Configuration
Control and is intended as a "first cut" of procedures, drawings, lesson
plans and schedules that may need to be revised as a result of information
found in the instruction manual.) According to Configuration Document
Change Control for Proposed Facility Changes, 50123-XIV-4.2, Form'184 should
be returned by affected departments confirming their intention to make
appropriate revisions to applicable documents. When all revisions are
completed, the "open item" produced by receipt of the Atwood-Morrill instruc
tion manual will be closed on the San Onofre Commitment Register (SOCR).
Documents associated with the proposed facility change are to be sent to
Corporation Document Management for archiving when the facility change is
complete and associated SOCR items are "closed."
7. Check Valve Maintenance
The maintenance practices utilized for the failed feedwater check valves were
reviewed. It was found that the maintenance of check valves falls mostly in
the category of corrective maintenance with little preventive maintenance
being performed. For the failed valves, maintenance activities amounting
12
o opening the valve and inspecting the valve internals were conducted
during the refueling outages in 1979 and 1980. Since no unusual wear was
noted, the installation of anti-rotation lugs was believed to have solved
a previous disc detachment problem. No records of subsequent inspection
of these valves was noted.
Specific maintenance procedures are not utilized for .the work on check valves.
However, the detailed steps to be accomplished during the maintenance activity
are described on the Maintenance Order per the General Maintenance Procedure.
The steps are generally prepared in accordance with the requirements in the
Installation, Operation, and Maintenance (IOM) Manual.
In the case of the
MCC-Pacific valves, the licensee did not have a copy of the IOM prior to the
event. During the inspection, the recently received IOM was reviewed and
found not to contain any information that would have had a bearing on the
failures. The IOM did not contain requirements for dimensional measurements
or seat contact checks.
8. Inservice Inspection and Testing
Implementation of the San Onofre Unit 1 Inservice Inspection and Testing
Program was reviewed for compliance with the ASME Boiler and Pressure
Vessel Code,Section XI.
Implementation of the program was found to be, in general, consistent with
the requirements of the Code. The plant records of the testing performed
under the program were reviewed. Those records made available for review
indicated that the testing of the safety-related check valves had been
performed for the previous five to seven years. It was observed that five
check valves in the safety injection system are identified as never receiving
a full stroke exercise or a disassembly to verify condition and freedom of
movement of the internals. These valves are numbered SIS-003, -004, -010,
-303, and -304. Relief from the ASME XI requirements, however, had been
requested for these valves and will be further reviewed by the NRC staff.
Additionally, there are other check valves where relief had been granted
because of the inability to test them during plant operation. The failed
feedwater system check valves were in this category. This category is
generally tested during periods of cold shutdown and/or refueling outages.
As indicated in the licensee's Failure Report, the A and B feedwater line
check valves were last tested in February 1985 and the C line valve was
tested in October 1984. In May 1985 the test procedure was changed to
require the leak rate test to be performed after steam generator pressure
is established. This was prepared to compensate for leak test failures
that had been encountered when the tests were performed with the plant in
cold shutdown. Although the vibration in the B feedwater line had been
attributed to either the gate or check valve, SCE did not avail themselves
of the opportunities for testing between May 1985 and the event.
The Operations Surveillance Procedures for performance of routine testing
of valves identified in the IST program were reviewed to determine if the
testing directed by these procedures could satisfactorily verify the opera
bility of the safety-related valves. The procedures appear to adequately
13
verify the operability of the valves at the time of the test in the direction
of the intended safety function. However, check valves were not tested
in both the open and closed position when the valve had only one intended
safety function. Although this degree of testing is not required by the
ASME Code, good engineering practice would, in most instances, require test
ing in both directions to assure full operability of the valve.
The testing being performed as required by the Code does not detect the
degradation of check valves. Because of the failure of the feedwater check
valves, the licensee has committed to the development of a quantitative leak
rate criteria for these five valves. The previous procedural requirement
provided no direct limits on the operator performing the leak test to
verify that the valves are closed. The leak rate that was considered
acceptable for the test was left to the operator's discretion.
During the inspection, the licensee was asked to explain why the quantitative
criteria was restricted to the feed system check valves. A response to this
question was not obtained. In addition, this proposed leak test acceptance
criteria will amount to a normalized measured leak rate. The licensee
intends to perform the test at any pressure and, by use of a standard
scaling technique, obtain a normalized leak rate. This leak rate test
will be used to determine the operability of the valves. Subsequent to
the inspection, the licensee's proposed leak rate criteria and testing
procedures were reviewed by the staff and found to be acceptable. In
addition, the licensee agreed to, within the next six months, determine
whether additional check valves warrant a quantitative leak check.
Interviews with licensee personnel concerning the mechanism for tracking
the performance of the valve tests, especially those identified for conduct
during cold shutdowns, indicated that an adequate method exists for ensuring
the test is performed and failures are identified for additional testing and
maintenance. The records for calendar year 1985 were reviewed and found to
be satisfactory.
14
APPENDIX A
0
Room 7-044, MIT
Cambridge, Mass. 02139
1 April 1986
Mr. Charles L. Nalezney
EG&G Idaho, Inc.
P.O. Box 1625
Idaho Falls, ID 83415
Dear Chuck,
This is the letter report reviewing the "Failure Analysis of Swing
Check Valves," Reference 3, by Chiu and Kalsi.
The accident is described
in Reference 4. The question which you asked is as follows:
"Is the approach taken to model the local turbulence effects sufficient to
account for all the expected effects (e.g. cavitation and vortex shedding)
on the valve discs for the old valves and the replacement valves such that
is can be expected that the replacement valve discs will be held firmly
against the open stop for the postulated flow conditions? If not, provide
a discussion of the uncertainty resulting from this analytical approach."
In order to understand the problem better, I hired an undergraduate,
Jennifer Snopkowski, to build and run an apparatus to see how swing check
values performed. The work was performed during spring break. Her report
is included. I worked closely with her and have run the apparatus many
times myself and now feel I have a good understanding of how these valves
perform. The details of the apparatus and procedure are in the enclosed
report. The important results are as follows.
The velocity at which the check valve is pegged can be calculated very
well using the formula and constant (K = 2.22) given in Chiu and Kalsi (3)
on the top of page B-3. See Figure.
(There is a power (2) missing as an
exponent for V in the second equation. The formula was used correctly,
however.) We find that the value calculated with that formula does not
work if the number of pipe L/D's lying between the flow control valve and
the check valve is less than 6. That is for the region
0 < L/D < 6
-1-
the turbulence and vortex shedding due to the upstream disturbance
dominate the pipe turbulence, and the valve chatters.
I think no check
valve should be placed closer than 6 L/D from the nearest upstream
disturbance. For short L/D's the valve chatters at velocities which are
much greater than that given by the formula which is cited above.
See
Figure 6 of the Snopkowski report.
With a modest safety factor, say 1.2, I think the minimum velocity
values calculated by the formula cited above are acceptable as long as the
valve is more than 6 L/D's downstream of any disturbance.
In order to use the check valves in a region which is closer than 6
L/D's to the upstream disturbance, it is necessary either to run an
experiment or to do a pretty complete dynamic analysis.
I don't think we
know enough at this time to do an analysis. I'd like to describe how one
should be done citing the references which apply and pointing out the
weaknesses if one is done by South California Edison and if they choose to
present it. An analysis based on a random vibration model is presented in
Appendix E of Reference (3).
It is a good start but needs more work. Let
me begin with the exciting force.
A blunt object, like a cylinder, in cross flow sheds vortices,
Reference 7. If these are close to a natural frequency of a structure
downstream from the blunt object vibrations can be excited. We don't know
what values of Strouhol number apply to the flow control valves illustratd
in Figures 1, 2, 3, and 4 of Reference (3).
Random vibration theory might
apply, but we don't have the precise PSD's needed for the geometries of
interest. A convincing dynamic analysis must be based on data taken in the
appropriate geometry.
I tried, in the library, to find some measurements of turbulence in
jets or the wakes of blunt objects and found some in References (5)
and
(6).
The values are much greater than those we find in pipe flow, but
frequencies are not given. Fluctuating velocities are of the order of 50%
of pipe velocity, in some places, rather than the 10% we find in pipe flow,
Reference (8).
It isn't clear to me, either, whether this problem is
-2-
better modeled as random vibration or as a forced vibration described by
the appropriate Strouhol number.
That question can only be settled by
experiment.
In our experiments, the gate moved up and down, tapping the stop
lightly. It was certainly not periodic but rather would show a cluster of
four or five impacts then a delay of several seconds or so before the.next
cluster of taps.
I'm not sure how that should be modeled. This kind of
turbulence is characteristic of jets (5).
Let us now turn our attention to modeling the gate response. In order
to determine this, it is necessary to be able to calculate an effective
mass, the spring constant and the damping factor.
I believe that Reference (3)
did not consider the induced or virtual
mass of the water. At least on Page D-1, the mass calculated for the valve
gate does not appear to include it. Reference (1) describes virtual mass,
and Reference (20) gives the values for many shapes of interest, though
not, of course, for our valve. The contribution to the mass could be
appreciable for a gate in a pipe. For a disc of radias r oscillating in an
infinite pool, the virtual mass is equal to the mass of the water contained
in the volume
(8/3) r3
This mass is about 20% of that of the disc itself.
The calculation for the spring constant given on page D-1 of Reference
(3) does not include the gravity force acting on the disc. Whether it
enters in an important way depends, I think, on how wide open the valve is.
It should be included, however. For a wide open valve, it would be an
almost constant force and probably does not play an important role.
Our experiments indicated that the gate on the 2" valve we tested was
overdamped as long as it was in water. A rough scaling of the damping
ratio to a larger valve does not tell use whether the valve is over or
under damped though one would expect the valve to be more heavily damped
that the damping ratio of 5% assumed in Reference (3).
It might be that
-3-
there is no resonance because the gate is overdamped for all positions.
I think calculating the damping of the valve is quite difficult.
Using drag on a pivoting disc might yield an answer good to a factor (2),
however. I think it is far more heavily damped than the value implicit in
a critical damping ratio of 5% which is mentioned on page E-1 of Reference
(3).
That is a very conservative assumption.
Let us assume that we can overcome all these problems; how do we
determine how long the valve will last? It is impossible to start or turn
off the flow without the valve hitting the stop a few times. The question
is how much is tolerable. In principal, I guess one could get the PSD of
the vibration amplitude, calculate the kinetic energy of the gate at each
impact and determine the damage done if all that kinetic energy went into
yielding the stop or stud or both. I think that is a very conservative
procedure.
Perhaps a factor, much less than 1, could be found in the literature
for the "efficiency" of the impact so that one could justify reducing the
damage per impact. I suspect that measurements of the right kind have been
made, but I didn't look for them.
In summary, I think you can confidently license the plant if the
following conditions are met:
1. The L/D from the nearest upstream fitting to the gate is greater
than 6.
2. The velocity is greater than 1.2 times the velocity calculated from
the formula at the top of page 8-3 of Reference 3.
If the check valve is closer than 6 L/D's to the nearest upstream
fitting, I don't think we know enough to predict whether the installation
is safe or not.
If cavitation occurs, the collapse of the resulting vapor bubbles
could be quite violent, and I don't think anyone could say whether the
-4-
valve was safe.
I don't see why either the check valve or the flow control
valve upstream should ever cavitate, during normal operatuion, however.
If an experiment to determine the flow velocity at which the valve is
pegged is run full scale with the appropriate upstream geometry but cold
and the valve performs satisfactorily, I think you can scale that
experiment and comfortably license the plant. For L/D's less than 6, I
find such a test the only convincing justification to license an
installation.
I don't think we can calculate whether a valve is safe when the L/D to
the nearest upstream disturbance is less than 6 with what we know now.
If you have any questions, feel free to call; I'd be glad to answer
them. If you come through Boston, stop by and I can demonstrate our
experiment for you.
I'm going to make this one of the experiments in our
senior undergraduate lab and hope to have it running for several years.
Sincerely,
Peter Griffith
Consultant
PG/jg
References
1) Milne-Thompson, L.M., Theoretical Hydrodynamics, 2nd edition, McMillan
Co., New York, 1950, p. 229.
2) Patton, Kirk T., "Tables for Hydrodynamic Mass Factors for
Translational Motion," ASME paper 65-WA/UNT-2, 1965.
3) Chiu, C. and Kalsi, M.S., "Failure Analysis of Swing Check Valves,"
Feb. 24, 1986 revision.
4)
"Loss of Power and Water Hammer Event at San Onofre, Unit 1, on
November 21, 1985, NUREG-1190, January 1986.
5) Rouse, H., Advanced Mechanics of Fluids, J. Wiley & Sons, 1959, p. 396.
6) Hinze, J.0., Turbulence, McGraw-Hill, 1959, p.404.
7) Bleyins, R.D., Flow-Induced Vibration," Van Nostrand Reinhold Co.,
1977, p. 15.
8) Schlicting, H., Boundary Layer Theory, Pergammon Press, 1955, p. 576.
A*(24)
Check Valve Dynamics
Jennifer Snopkowski
Peter Griffith
1 April 1986
Abstract
A two inch diameter check valve was tested in water to find the range of
velocities and L/D's from an upstream disturbance when the valve was always
(1) pegged, (2) swinging and hitting (tapping), and (3) swinging free.
The
formula for pegging from the Chiu-Kalsi report was found to predict the
pegged velocity quite well when the flow was fully developed (i.e. L/D > 6).
For L/D < 6, .the valves tapped at velocities well above that given by the
Chiu-Kalsi formula. Tapping was found to persist down to about half the
maximum tapping velocity for fully developed flow, independent of L/D from an
upstream disturbance.
I. Introduction
Check valves used in a reactor feed-water delivery system are designed to
prevent the backflow of steam. Failure of the check valves can result in a
water hammer, causing severe damage to the feed-water system. An actual
water hammer occurrence due to check valve failure occurred and is reported
in reference 4 of the letter.
A diagram of one of the check valve installations in the system which
experienced failure is shown in Figure 1. The failure is a result of
repeated tapping of the check valve against the stop. This tapping is caused
by disturbances in the flow over a certain range of velocities.
The purpose of this experiment was to find out what the range of velocities
was which caused an experimental check valve to tap against the stop.
This
report compares the maximum water velocity at which tapping occurred with an
analytical method developed by Chiu and Kalsi given in reference 3 of the
cover letter.
The following questiongpresented in the work statementswill be addressed:
"Is the approach taken to model the local turbulence effects sufficient to
account for all expected effects (e.g. cavitation and vortex shedding) on the
valve disks for the old valves and the replacement valves such that it can be
expected that the replacement valve disks will be held firmly against the
open stop for the postulated flow conditions? If not, provide a discussion
of the uncertainty resulting from this analytical approach."
The 'results
reported here are the basis for the recommendations in the cover letter.
II. Apparatus and Experimental Procedure
The apparatus was set up as shown in Figure 2. A diagram of the experimental
system is given in Figure 3. The check valve was placed downstream of a.
plastic insert simulating a valve which was used to produce a disturbance in
the flow. A tracing of the insert is shown in Figure 3. The distance L
shown on the diagram was varied to test the effects of L/D change on the
range of critical flow velocities. In this apparatus, D is the inside
diameter of the pipe and is equal to 2.062 inches. The plastic insert was
removed to determine a range for an asympotoic value of L/D.
-2-
The check valve used in the experiment is shown in Figure 4.
The total
weight of the gate, including the arm, was .471 lb m. The details are given
on Figure 4.
To take this picture, the bonnet on the check valve was removed and replaced
with a plastic window which was screwed into the thread originally holding
the bonnet. It was easy to see where the gate was through this window and
whether it was pegged or not. The window is visible in the photographs of
Figure 2 and Figure 5. Figure 5 shows the gate pegged and is photographed
through the window. The windows at the ends of the horizontal run shown on
Figure 1 were not needed.
III. Results
A plot of flow velocity versus L/D is given in Figure 6. The area used in
the velocity calculations was based on pipe diameter and is equal to
3.34 in2.
The lower half of the graph shows points where the valve
oscillated in the flow without tapping the stop. It can be seen that the
transition to velocities where tapping begins is not dependent on L/D. Above
this transition velocity, the valve repeatedly tapped against the open stop.
The frequency and intensity of the tapping increased with increasing
velocity.
The transition where the valve no longer taps aginst the stop but is pegged
against it is an important one. If the velocity is kept above this
transition velocity, the valve will be held stationary against the stop by.
the flow, and there is little possibility for failure.
At L/D values above
6, the velocity at which the valve was pegged is fairly constant. The far
right of the graph of Figure 6 shows the minimum tapping velocity for a fully
developed flow.
Below an L/D of 6, however, the pegging velocity
dramatically increases and is off scale for L/D values less than about 3.
. The frequency of tapping is high and the severity of oscillations large for
the low L/D values.
The analytical method mentioned in the Introduction and found in Reference 2
was followed for.the equipment used in this experiment. The calculations are
given in Appendix A. The minimum velocity needed to keep the valve pegged
was determined using this method. It was 6.75 ft/sec and is shown in
-3-
Figure 6. This valve clearly behaves as predicted for L/D values greater
than 6. Below this value, however, this method fails.
An attempt was made to find the approximate damping ratio experf'mentally by
disturbing the gate and watching it respond. As far as could be seen, this
gate was more than critically damped. One could not see any overshoot at all
for any gate position. The measurement was entirely visual however and
hardly one of high precision. Nonetheless, this gate appears heavily damped
and one would not expect it to display any resonance due to vortices or
turbulence generated at the disturbance upstream.
When there was no disturbance upstream, the tapping was easy to hear. It
could be heard more easily through a screwdriver held against the body of the
check valve with the handle held against the ear. With the disturbance
upstream, however, the flow noise overwhelmed the tapping and I (PG) couldn't
hear the tapping at all even though I could see the valve swinging and
hitting.
IV. Conclusions
O
1. The velocity at which the valve gate is always pegged is well predicted
by the Chiu-Kalsi formula for a fully developed flow. This means the L/D
is greater than 6 for the distance from an upstream disturbance to the
gate.
2. When the L/D is less than 6 from the nearest upstream disturbance to the
gate, the valve taps for velocities much greater than that calculated
using that formula.
3. The valve swings free without tapping for velocities equal to about half
of that given by the Chiu-Kalsi formula.
4. The gate in the 2" ID valve used in these experiments appeared to be
heavily damped for all positions. No resonance was observed or expected.
5. The fact that a formula developed for a valve of a similar type but of a
different geometry works so well it leads one to think it would work on
other valves too.
-4-
Appendix A. Pegging Velocity Calculations
The Chiu-Kalsi formula from the top of PB-3 of reference (3) of-the
letter is:
S
(W-
.5
W
(b)(g)
KpA sin 2
In this formula
V = velocity, [ft/sec]
W = mass of gate plus arm = .471 lb
m
Wa = mass of arm = .164 lbm
b =
bouyancy factor = .883
g = acceleration of gravity = 32.2ft/sec 2
k = empirical constant = 2.21
p
= density of water = 62.3 lb /ft3
.2
Ib/t
A =
pipe area = 3.34 in = :0232 ft2
= angle of valve from the horizontal when pegged = 16.08
The resulting minimum pegging velocity is
V =
6.75 ft/sec
This formula is derived from a force balance on the gate where the jet
force holds the gate up against gravity which is tending to make it fall off
the stop. It is the maximum tapping velocity. The constant 2.21,given as K
above was determined from the minimum acceptable operating velocity provided
by one of the valve manufacturers.
-5
(Astar25)
8" x 6" OS CONT. VALVE
100 6000 GATE
100' 6000 MOV
10
600 CH. VALVE
E_
EXPANDER
(1050 psi)---
590 PSI
31 -
- ~7
-
26 -**-7
-
2"
-
-
31.'
31'
NOTES:
1. MAJOR TURBULENCE COMPONENTS ( PRESSURE
REDUCING CONTROL VALVE.
AND AN EXPANDER)
IMMEDIATELY UPSTREAM.
2. ONLY IxD STRAIGHT PIPE LENGTH AFTER
HIGH TURBULENCE COMPONENTS.
FIGURE 1.
10" 600# CHECK VALVE INSTALLATION
(F WS
-
345. 346. 398)
2A
4.7S4
Figure 4. Experimental check valve.
Weight of gate assembly = .471 lb
Weight of arm = .164 lb
m
m
-7-.
e2-061
2.Es*-
i/
C
er-t Outfine 4C fu4 1lze
Figure 3.
Key dimensions of experimental set-up and outline of actual size
of the plastic insert.
e
.
Figure 2. Experimental apparatus.
0
-9-
0C
VW /ve
Figure 5. Operating check valve with no tapping.
0
-10-
I.~AP SWJ.AJ .
Cp
t Imr*
aE'x
to
.iur
6. FoUeoiyvru./
k7
4
APPENDIX B
0