ML20076G441
| ML20076G441 | |
| Person / Time | |
|---|---|
| Site: | LaSalle  |
| Issue date: | 08/26/1983 |
| From: | Schroeder C COMMONWEALTH EDISON CO. |
| To: | Harold Denton Office of Nuclear Reactor Regulation |
| References | |
| 7198N, NUDOCS 8308310292 | |
| Download: ML20076G441 (13) | |
Text
7 e
/-
Commonwealth Edison
) one First Natm.at Plazi Chiergo, Ilknois
(
1-~
l Address Reply to: Post Office Box 767
/ Chicago, filinois 60690 l
August 26, 1983 Mr. Harold R. Denton, Director Office of Nuclear Reactor Regulation U.S. Nuclear Regulatory Commission Washington, DC 20555
Subject:
LaSalle County Station Units 1 and 2 Methodology for Fatigue Evaluations (FSAR Question 111.87 Response)
NRC Docket Nos. 50-373 and 50-374
Dear Mr. Denton:
Attached is Commonwealth Edison Company's response to FSAR Question 111.87 on the methodology used by the Architect Engineer to assess fatigue in safety-related equipment.
This methodology was reviewed with Messrs. J. Singh and G.L. Thinnes of EG&G (Idaho) on August 1, 1983 at Chicago.
From earlier telecons with Mr. Singh, et al, it became obvious that all essential references had not been made available to EG&G for their independent review of this subject.
Missing information included the following:
1.
LaSalle " Design Assessment Report" which provides the NRC approved Mk 2 load combinations and the definition of the bounding load cases.
2.
" Mark 2 Dynamic Forcing Functions Report", (NEDO 21061, Rev. 2) which provides time-dependent generic contalment information on the events that generate cyclic pressure load parameters.
The design basis number of transient events and the number of pressure cycles per event are given in Section 5.3.
3.
The stress-reversal method for counting the unique number of cycles within each event is indicated in test documents such as
" Summary Report for Limitorque Valve Operator Testing" (Jan. 1982) which was provided to EG&G at the Chicago meeting.
This method defines the shaker table test specification to meet the event criteria.
The table discussion was helpful in clarifying the subtle differences in methodology as applied to analysis such as that included in the LaSalle SQRT record "Sur.Rary Report for Fatigue Analysis for the LPCS Pump" which had been sent to EG&G as a part of the 1980 LaSalle SQRT audit record.
No unresolved differences existed at the conclusion of the Chicago meeting.
$006
/ ik2 9308310292 830826 PDR ADOCK 05000373 P
1 H. R.
Denton August 26, 1983 To keep this Response 111.87 in proper perspective, it should be recognized that an extremely conservative methodology was utilized for the most adverse load combinations, and yet the calculated fatigue usage factors never exceeded 0.2 to 0.3 but were generally in the 0.0X range.
There is no safety significance based or fatigue considerations for any of the reviewed items; indeed, the representativeness of the analyzed equipment indicates that the design process was sufficiently conservative to totally preclude fatigue as a consideration.
Further refinements in the fatigue methodology were therefore not justified to establish more precision by more rigorous treatment.
The results indicate sufficiency for fatigue usage considerations.
To the best of my knowledge and belief the statements contained herein and in the attachment are true and correct.
In some respects these statements are not based on my personal knowledge but upon infor-mation furnished by other Commonwealth Edison and contractor employees.
Such information has been reviewed in accordance with Company practice and I believe it to be reliable.
Enclosed for your use, please find one signed original and forty (40) copies of this letter and the enclosure.
By copy of this letter, one copy of the enclosure is also provided to Mr. Jack Singh, EG&G.
If there are any further questions in this matter please contact this office.
Very truly yours, e/zcle C.
W.
Schroeder Nuclear Licensing Administrator 1m Enclosure cc:
NRC Resident Inspector - LSCS 1/0 Jack Singh - EG&G 1/1 7198N t
Response to Question 111.87 (a)
The first part of this question was revised slightly by EG&G to focus on the description of the AE's methodology, 1) for computing the usage factor for ANALYSIS purposes and 2) for assuring coverage of the fatigue issue in t_esting of equipment.
The method for computing fatigue usage factors analytically is described in the " Summary Report for Fatigue Analysis for LPCS Pump".
This report had previously been provded to EG&G's J. Singh and G. Thinnes.
This method is described again here.
A three dimensional finite element model of the pump-motor assembly and its support was initially developed.
Details of this finite element model are given in the qualification report for the Low Pressure Core Spray Pump (EMD File No. 028197) which EG&G has in their possession.
The analysis was performed in two distinct steps.
First, a modal analysis was performed for the LPCS pump assembly using a standard Subspace Iteration Technique to define the modal participation factors.
From this step all eigenvectors were saved in a permanent file for subsequent analysis with the various dynamic loading conditions.
Twenty-five modes were considered in this particular analysis, with the highest frequency being 140.5 cycles per second.
This frequency is high enough to cover the ZPA of all applicable response spectra for all the seismic and hydrodynamic events.
The second step of the evaluation consisted of a dynamic stress analysis based on these saved modal eigenvectors and the composite conservative response spectrum for each loading event.
The conservative response spectrum at the location of the pump was obtained as an output from the general finite element model of the reactor building as affected by the time-history inputs for the composite loading cases (combined events) as described in the NRC-approved LaSalle " Design Assessment Report".
Further conservatism was added in these building response spectra by the fact that the peaks were broadened 15 percent and these plant spectra for a particular location enveloped several plant elevations into a maximimum composite response spectrum (to limit the number of spectra to a tractable number).
In the dynamic analysis, a combined-stress by absolute sum was computed for each loading event.
Then, by use of the applicable ASME material fatigue curve (S-N curve), the corresponding fatigue cycle allowable limit, Na, was obtained.
The number of actual fatigue cycles, Ni, was determined as outlined in the response to Part b of this question.
The fatigue usage factor was then computed by the ratio Ni/Na for each loading event.
The total usage factor was obtained by summing the individual event usage factors (see pages 12 through 16 in the LPCS pump summary report).
The maximum usage factor thus obtained was 0.221 for the LPCS pump, thereby verifying that fatigue is not an issue in its overall qualification.
i
. Fatigue evaluation by testing is described in the report " Summary Report for Limitorque Valve Operator Testing Program", (CQD File No. 000731).
A copy of this report was given to EG&G at their request after release by G. Crane (CECO).
In this test situation, two representative motor-operated valve assemblies were tested for fatigue.
These valves were subjected to dynamic inputs which simulated very conservatively the amplitude, frequency content and duration for the most severe combination of seismic loads, SRV actuations and LOCA loads.
The procedure used to specify the test-table inputs and the test sequence is described in the previously indicated 000731 report.
It is briefly mentioned in the Part B response also.
(b)
The number of fatigue cycles expected to be experienced by a piece of equipment is aggregated from the frequency of occurrence of transient / accident / operational events over the life of the plant and the number of pressure (load) cycles ascribable to each type of event for its unique duration.
The design basis for LaSalle includes consideration of the following events with respect to cyclic loading of structures and equipment:
OBE, SSE, SRV synmetric and SRVasymmetric, and Chugging.
The transient events leading to stuck open relief valves (SORV) are included under the SRV asymmetric entry.
The annulus pressurization (AP) case was not considered because it is a single event, of very short duration, with effects limited to the area of the sacrificial shield wall.
A large LOCA accident, likewise, is a moot consideration for fatigue effects.
For LOCA's in general, only the small line breaks were considered because the number of pressure cycles associated with small breaks is much larger than the number of cycles associated with intermediate sized LOCA breaks.
During a small LOCA event there is no contribution from Condensation Oscillation loads (CO) hence they are not considered in evaluating fatigue effects for small break LOCA's.
The small break LOCA event also dominates
- the intermediate break LOCA combined with Condensation Oscillation loads as pertains to cyclic content and fatigue.
The frequency of occurrence for SRV discharge has been cataloged according to initiating causes, the number of valves discharging per event (for a particular type of plant), and the number of blowdown sequences to terminate the pressurization phase of the event.
These numbers are tabulated in the DFFR Section 5 for LaSalle.
For example, LaSalle has a design total of 5200 lifts of the SRV with the lowest pressure setpoint.
SRV discharge pressure signatures for the Monticello plant were reported in NED0 21581.
Separately, now the SRV discharge pressure signature for LaSalle was determined to be markedly less than that from Monticello or the 4TCo experiments by GE.
u
Reference:
Mk 2 SRSS-LSCS Meeting Minutes of June 16-17, 1981 ceeting at GE (San Jose), attachment H page 4.
Copy provided to NRC.
. In preparing the test input profile, separate load-time histories representative of all pertinent dynamic events for a particular plant location (node) are input into the Response Spectrum Generator Program (RSG) which combines them with the corresponding widened response spectrum of that respective node.
The output of the RSG program is a forcing function-time history at the plant node - essentially a composite load profile containing all the frequencies and a shape and duration representing all the time history inputs.
Because it is not practical to test equipment to the varying, complex load-time profiles resulting from the time history input of events, an energy equivalent test profile is developed with the peak amplitude equal to the maximum load but with the total number of test cycles adjusted to an equivalent number per composite load profile.
This equivalence is accomplished by taking the peak distribution adjustment factor (as determined via the method approved by the ASME Special Working Group on Dynamic Analysis) and multiplying the total number of cycles in the composite profile by it.
This equivalent sinusoidal load-time sequence then has the characteristic of the maximum load amplitude and the number of uniform load cycles having the same total energy input to the test specimen as if it were exposed to the complex, time-varying composite load profile.
Equipment was then tested to 120 percent of the time duration.
For analysis purposes, these forcing function time histories at plant nodes are utilized at pipe anchor points, for example, to evaluate the piping system responses, valve responses, etc, and give displacement-time histories for the particular equipment mounted on the piping, valve, etc.
Each displacement-time history was then used as a ranking mechanism to evaluate fatigue damage potential in the equipment.
After choosing the displacement-time history which would give the maximum relative damage effect, the duration of each displacement time-history was increased by 50% by adding the first half of the time history profile to the end; this increased the potential fatigue damage by a factor of approximately 1.5.
At the test laboratory white noise traces were generated for every load case to assure that the desired fatigue damage value was actually tested.
To insure test coverage, the number of SRV and chugging excitations during testing were run to 1.5 times the expected SRV and chugging cycles during the plant lifetime.
According to AE calculations the fatigue damage on each Limitorque operator-valve combination caused by testing was at least five (5) times the expected fatigue damage assignable to the different plant dynamic loads over its 40-year lifetime.
. (c)
The justification for not iacluding events for which the equivalent cycles were "not defined" in the evaluation is that such events (intermediate line breaks, for example) are covered by other, more severe loading events from the fatigue point of view.
See (b) above.
(d)
The term " representative" components for fatigue testing refers to a specific set of LaSalle equipment wh0'e identity and s
categorization evolved during 1980 discussions with the NRC staff.
An outline of the origin and significance follows.
CECO-NRC staff discussions on SQRT during the spring months of 1980 resulted in the commitment to address fatigue by test and analysis.
At that point in time, most of the LaSalle SQRT testing was completed, however, for panels and instruments not yet tested, it was decided that fatigue testing would be included in future tests.
Where size of mechanical equipment precluded testing, detailed analyses would be performed for representative panels or assemblies to investigate fatigue on the worst case basis.
If either the future test results or the representative analyses indicated that fatigue was a problem, the entire SQRT program would undergo a fatigue assessment; however, if the future test results and representative analyses on large mechanical equipment established that fatigue was not a problem, then it could be logically concluded that standard design practices covered the fatigue consideration.
Fatigue Evaluation of LaSalle Equipment The ability of equipment to withstand cyclic loadings under combined seismic and hydrodynamic (LOCA) events was evaluated as follows:
1.
Fatigue evaluation by testing
- Electrical equipment such as control panels, motors, etc.
- Instruments such as indicators, sensors, relays, etc.
2.
Fatigue evaluation by analysis
- Mechanical equipment such as pumps and heat exchangers, etc.
3.
To investigate fatigue of large equipment under hydrodynamic and seismic loadings, two representative pieces of equipment were analyzed:
- LPCS Pump
- RHR Heat Exchanger The estimated fatigue life under these loads was determined by use of the procedure of ASME Section III, Subsection NB (NB-3222.4).
The maximum total usage factor for the LPCS pump was determined to be 0.221 (EMD File No. 028847).
The maximum total usage factor for the RHR Heat Exchanger was determined to be 0.373 (EMD File No.
. 028847).
Both of these usage factors are well below unity, hence fatigue damage is not critical to the structural integrity of these and similar other mechanical equipment when subjected to combined seismic and hydrodynamic loads.
Test results from eleven electrical panels and ten devices listed below show no significant fatigue damage and/or functional failure during or after testing.
For LaSalle, fatigue failure of electrical equipment and instruments is not expected under seismic and LOCA hydrodynamic loadings, a.
Local instrument panel (72") for LaSalle and Zimmer, SWRI Report #02-6056-001 (November 7, 1980).
b.
Local instrument panel (48") for LaSalle and Zimmer, SWRI Report #02-6056-001 (December 30, 1980).
c.
Local instrument panel (30") for LaSalle and Zimmer, SWRI Report #02-6056-001 (January 23, 1983).
- d.
Control room cabinets:
H13-P613, H13-P635, H13-P612, H13-P621, H13-P632, H13-P618, H13-P689 for LaSalle and Zimmer, Wyle Lab Report #58626 Volumes 1 through 3 (April 13, 1981).
- e.
Low Voltage Power Circuit Breaker Devices for laSalle, EMD File
- 029979.
f.
Instruments Barksdale pressure switch Model BIT M12SS ITT Barton level indicating switch Model 760 Yarway level indicating switch Model 4418C Barksdale vacuum switch Model DIT-H18SS
- Bailey meter flow summer Model 7524
- GE transducer power supply Model 9766 ASCO solenoid valve Model HT 832322 GE voltage preamplifier Model 828E309AA GE intermediate range monitor Model 368X102AA GE power range neutron monitor Model 368X105TD
- GE indicator and trip unit Model 12982802
- GE log radiation monitor Model 238X660
- GE sensor and converter Model 194X927 ITT Barton flow switch Model 289A Barksdale pressure switch Model PlH Pyco temperature element Model 102
- Eagle signal time indicating switch Model HP5 S&K flow transmitter Model 20 S&K flow. transmitter Model 91
- GE trip auxiliary unit Model 238X660
- Cutler Hammer switch Model 10250T
- GE Contractor Model CR 105/205
- GE time delay relay CR 2820
- GE IAC relay
- These equipments are located in the Auxiliary Building and do not Gxperience any pool hydrodynamic loads.
They were tested to five OBE's and one SSE event.
=_
. *GE IVC relay
- GE JVP-1 transformer
- GE JAH-0CT
- GE Ak-2A-50 circuit breaker
- GE Ak-2A-25 circuit breaker
- Bailey meter square root converter Model 136B3051 1
g.
Valves and operators 16-inch globe valve with Limitorque SMB-2 operator 4-inch globe valve with Limitorque SMB-000 operator 4
(e)
The load combinations are those previously agreed to by NRC as part-of the DAR for the appropriate station.
Normal loads are added to each load case.
4 LOAD COMBINATIONS ZIMMER 1 LASALLE 1 & 2 CASE OBE + ENV (SRV 1.
ALL-TQ) ADY-TQ +
ENV (0BE &.SSE) + ENV SSE + COLEVY-2 +
EMERGENCY (SRVASY-TQ & SRVADS-TQ)
SRV* ADS-TQ
+ CHG 3.
ENV-(0BE & SSE) + ENV Chugging + SRV* ADS-TQ (SRVASY-TQ &
+ SSE 0.6 SRVADS-TQ) + C01
~
4.
ENV (0BE + SSE) + ENV SSE + COLEVY-1
+
(SRVASY-T0 &
SRVADS-TQJ + C02 5.
- SRVALL-TQ and SRVADS are i
equal and are the envelopes of SRVALL-TQ and SRVASY-TQ-
- These equipments are: located in the-Auxiliary Building and do not experience any pool hydrodynamic loads.
They were tested.to five OBE's and.one SSE event.
. Notes:
Normal loads have to be added to each load case.
Load Abbreviations:
OBE = Operating Basis Earthquake SSE = Safe Shutdown Earthquake SRVADS = Safety Relief Valve Automatic Depressurization System SRVASY = Asymmetric MSRV Case SRVALL = All Valve MSRV Case ENV (
) = The Response Spectra in the Brackets are Enveloped to Obtain a Boundary Spectra C0
= Condensation Oscillation CHUG
= CHUGGING AP
= Annulus Pressurization TQ
= Tee Quencher (f)
This write up shows that the fatigue exponent of 4.3 used in the fatigue analysis / testing of valves for LaSalle County Station is conservative.
Background Information:
Fatigue is a progressive failure caused by repeated loadings.
Fatigue tests are usually reported in graphs of alternating stress amplitude versus number of cycles to failure (Fig. 1).
Occasionally these graphs are also represented in the logerithmic form.
The relationship between the stress amplitude S and the nunber of cycles N to failure is given by:
NS I = Constant The exponent Y in the above equation is called the " fatigue index".
As seen from Fig. 1, y' is the inverse slope of the S-N curve.
The value of 1" therefore depends on the material as well as on
'N' (or 'S').
In most ductile materials, around 10,000 cycles, Y' = 4. 3.
)
l FATIGUE-DAMAGE EVALUATION For fatigue' testing, a test time history must satisfy two conditions:
1.
The response spectrum generated from the test time history (TRS) must envelop the response spectrum generated from 4
analysis time history (RRS).
(It may be noted that all required time histories were also very conservatively computed and have a good margin of safety.)
2.
The fatigue damage potential of the test time history must be higher than the fatigue damage potential of the required (analysis) time history.
In satisfying the first condition, no use is made of the fatigue exponent.
For LaSalle, all test time histories generated TRS much higher than the RRS.
To satisfy the second condition, a certain fatigue exponent must be assumed to compute the fatigue damage potentials of the required and test time histories.
A most logical selection for this purpose is to take T = 4.3.
However, as is explained subsequently, the test input is conservative for all values of Y.
The fatigue damage potential, DR of the required time history is calculated by the following equation.
n DR=
ui i=1 Where ui = amplitude of each cycle written as a mean value.
n = number of cycles.
T = f atigue exponent.
Similarly the fatigue damage potential
'D ' for the test time T
history is also computed.
In Fig. 2, a plot of (D /D )_ versus "( ' is shown for thel 4" T
R valve.
Fig. 3.shows'similar plot for the 16" valve.
R As illustrated bynthese two' figures, the test input'is very conservative for all value's.of
'Y..
Hence~the 4.3 exponent used.
in the fatigue. damage analysis is justified.
3340L*-
u
]
l FI uu 1-TVPl CA L S-N CURVE S
A S
Ar s = t o,c oo STRESS 1 =
- 4. 3 Typical AMPLITU D C M ATE A.l A LS
~ _ _ _ _ _
e
- ~-
N
\\l
\\
/4 ATERIAL A h N N N N
/A ATERIAL B
\\
O N=l0000 N
1 LIFE C YCLE S N
t
{\\
R si
\\
t1 4
g
}n i
e
\\
d Jy
\\\\
$ ;N u
a
\\\\
5 m
E f
9 V
w
\\
e e
\\
e m
4
\\
A'N"
\\\\
w k
p\\
\\
a N
\\
4 N
t
\\\\
\\
{
N t
4 5.
\\
\\
Ne k
i
\\
\\
i(
s
!k
\\\\
M
\\
e
- u..
x\\\\
L
~
x \\x r,e ar i4
+
L ii
\\s ii
\\
\\\\
~~
+e 4T W
b n
OO
'k
\\
3
\\
s o
9 0
Y N
O N
e erw-WA w-Ak bs si e
\\
d
~
W
\\
i @l
~s a
ee
\\
NN
\\
s e
t
- \\
s
\\
\\
e
\\
a-R
\\
s
_a@
d ko O
a 1 N
\\
N&
ha N
- a x.
't ti
\\
m e
3 si h
a ff N
o W W 14 11
\\
s ii
\\
?e m
t a N
1.1 N
v
\\
a s
ad
\\
-e 2
1 y
y
..