ML19351D309
| ML19351D309 | |
| Person / Time | |
|---|---|
| Site: | 05000447 |
| Issue date: | 09/10/1980 |
| From: | Henry J GENERAL ELECTRIC CO. |
| To: | |
| Shared Package | |
| ML19351D306 | List: |
| References | |
| NUDOCS 8010090504 | |
| Download: ML19351D309 (19) | |
Text
- ~
\\ _,
MARK III SRVA REVIEW J. B. HENRY TECHNICAL LEADER CONTAINMENT Et'GINEERING SEPTEMBER 10, 11, 1980 8 010 0 9080f[V
~
MARK III SRVA REVIEW QUESTION 1:
THE BASIS FOR USING 59 MONTE CARLO RUNS TO ARRIVE AT AN OVERALL FORCING FUNCTION WAS TO PROVIDE A 95%-95% CONFIDEtlCE LEVEL IN THE RESULTING LOADS.
THIS APPROACH ASSUMES THAT; (1)
THE PROBABILITY DISTRIBUTION FOR EACH RANDOM VARIABLE IS KNOWN WITH A 100% CONFIDE!!CE LEVEL, AND (2) USING THE FORCING FUNCTION TO CALCULATE DIFFERENT LOADS (E.G., THE LOADc
" THE VERTICAL DIRECTION AND THE OVERTURilING nUMENT) DOES NOT DECREASE THE FINAL CONFIDENCE LEVEL, WE WILL REQUIRE THAT THESE TWO ASSUMPTIONS BE JUSTIFIED AND THE ACTUAL CONFIDENCE LEVELS OF THE PROBABILITY DISTRI-BUTIONS BE USED Ill DETERMINING THE OVERALL 0NFIDEf!CE LEVEL.
I h0/80 1-1
r MARK III SRVA REVIEW RESPONSE #1
- PROBABILITY DISTRIBUTIONS ON RANDOM VARIABLES ARE NOT KNOWN TO 100% CONFIDENCE LEVEL, HOWEVER; INPUT DISTRIBUTIONS ARE GOOD REPRESENTATIONS OF THE DATA BASE, AND SENSITIVITY ANALYSIS SHOWS THAT BUBBLE FR500ENCY IS THE ONLY VARIABLE FOR WHICH SIGNIFICANT EFFECTS ARE OBSERVED.
THE OVERALL CONFIDENCE LEVEL IS UNAFFECTED BY CALCULATION OF FORCES AND MOMENTS; THESE ARE CALCULATED FOR EACH TRIAL AND USED AS THE BASIS FOR RUN SELECTION USE OF DESIGN PRESSURES INSTEAD OF HOMINAL INCREASES OVERALL CONFIDENCE LEVEL ABOVE 95%-95%
JBH 9/10/80 1-2 t
2 e
MARK III SRVA REVIEW BESPONSE #1 FORCING FUNCTION SPECIFICATION SELECTION FROM 59 TRIALS GIVES 95% - 95%
CONFIDENCE LEVEL IN THE CALCULATED SPECTRAL PEAKS SELECT RUNS WHICH PRODUCE PEAK SPECTRAL VALUES IN BASEMAT FORCE AND ROCKING MOMENTS.
THREE FREQUENCY INTERVALS ARE SELECTED BASED ON STRUCTURE CHARACTERISTICS:
4 - 12 Hz MAJOR STRUCTURES 12 - 20 Hz PIPING 20 - 28 Hz EQUIPMENT SELECTIONS ARE INDEPENDENT FOR EACH FREQUENCY INTERVAL AND EACH LOAD TYPE (FORCE AND MOMENTS)
L
)
JBH 9/10/80 1-3 s
MARK III SRVA REVIEW CRITICAL CASE SELECTION o
EXAMPLE FOR ONE LOAD CASE FREQUENCY VERTICAL ROCKING Mx MY 4-12 39' 57*
51 12 -20 54*
27*
57 20-28 29*
47 9'
)
' CRITICAL CASES GIVING HIGHEST SPECTRAL VALUE FOR EITHER VERTICAL FORCE OR ROCKING MOMENT.
FOR MOMENTS, THE CASES GIVING THE HIGHEST VALUES FOR EITHER Mx OR MY ARE SELECTED AND THE LOADS APPLIED IN BOTH DIRECTIONS, JEH t
o/10/80 1-4 l
\\
MARK III SRVA REV'EW RESPONSE #1 (CONT'D) e INPUT PROBABILITY DISTRIBUTI0flS ACCURATELY REPRESENT THE DATA BASE DATA BASE
- VALVE OPENING TIME
-408 TESTS ON 102 CROSBY VALVES
- 50 TESTS ON A SINGLE DIKKERS VALVE (SEE QUESTION 2B RESP 0tlSE FOR DATA)
- QUENCHER BUBBLE FREQUENCY
-132 IN PLANT TESTS I
e
- VALVE SETPOINT TOLERANCE
-2 r VALUE TAKEi! FROM TRIP SYSTEM REPEATABILITY ERROR C+.25%1
- PRESSURE RISE RATE
--RELATIVE FREQUE!!CY OF EVENTS WITH POTENTIAL TO LIFT 2/3 0F SRV'S.
JBH 9/10/80 1-5 L.
0, MARK III SRVA REVIEW FESPONSE #1 (CONT'D)
SENSITIVITY RESULTS SENSITIVITY OF PEAK SPECTRAL VALUES TO IMPUT DATA UNCERTAINTIES IS SMALL.
PEAK SPECTRAL VALUES VARIABLE AND VALUE i 2(
SENSITIVITY VALVE OPENING TIME MEAN 59 3 MSEC NONE STD. DEVIATION 9 t 3 Msec NEGLIGIBLE (-0.1% PER liSEC)
VALVE SETPOINT TOLERANCE 2.1 NEGLIGIBLE (-1% PER PSI)
PSI EUBBLE FREQUENCY MEAN 8.1.3 Hz FREQUENCY SHIFT STD. DEVIATION 1.7.3 Hz AREA UNDER PEAK IS PRESERVED PRESSURE RISE RATE RANGE 40-140 PSI /SEc NO CONSISTANT TRENDS (SEE RESPONSE TO QUESTI0tl 10)
CONCLUSION:
UNCERTAINTIES IN THE SHAPE OF THE INPUT PROBABILITY DISTRIBUTIONS ARE il0T IMPORTANT.
I JEH 9/10/80 1-6
MARK III SRVA REVIEW RESPONSE #1 (CONT'D)
. CONCLUSIONS
-95% - 95% CONFIDENCE L'iVEL IS FOR PE'4K SPECTRAL VALUES PRODUCED IN REPEATED MONTE CARLO TRIALS, NOT FOR THE LOADS.
-DESIGN LOADS B0UND MEASURED LOADS WITH LARGE MARGINS IN CAORS0:
~
- FACTOR OF 2 IN PEAK PRESSURE
- FACTOR OF 3 IN ARS (SEE QUESTION 9)
-INPUT PROBABILITY DISTRIBUTIONS ADEQUATELY REPRESENT THE DATA
--PEAK SPECTRAL VALUES ARE INSENSITIVE TO MOST INPUT DATA UNCERTAINTIES
. FORCING FUNCTION SELECTION IS BASED ON CALCULATED FORCES AND MOMENTS:
INDEPENDENT SELECTION FROM 59 TRIALS GIVES 95%-95%
CONFIDENCE FOR EACH SELECTION
- USE OF DESIGN PRESSURES INCREASES OVERALL CONFIDENCE LEVEL IN LOADS ABOVE 95%-95%
JEH 9/10/80 1-7
f a
MARK III SRVA REVIEW 1
QUESTION 4:
THE SENSITIVITY OF THE LOADS TO THE FORM OF THE RAtIDOM VARIABLE PROBABILITY DISTRIBUTIONS SHOULD BE SUPPLIED.
IN PROVIDING THIS INFORMATION THE STUDIES SHOULD BE EXPAt!DED TO INCLUDE THE EFFECT OF CHANGES Ifl THE t1EM OUENCHER BUBBLE FREQUENCY (QBF).
JBH 9/10/80 1
4-1 r17
+ - +
MARK III SRVA REVIEW RESPONSE #4
- RESULTS OF SENSITIVITY STUDIES ARE PROVIDED IN tlRITTEM RESPONSES AND SENSITIVITY RESULTS ARE SUMMARIZED IN QUESTION 1 RESPONSE e CONCLUSIONS
-PEAK SPECTRAL VALUES ARE INSENSITIVE TO INPUT DATA UNCERTAINTIES
- EFFECT OF CHANGING MEAN BUBBLE FREQUENCY IS TO SHIFT THE ENVELOPE OF SPECTRAL PEAKS, HOWEVEP. THIS IS ADEQUATELY COVERED BY THE GESSAR METHOD:
. LINE VOLUMES ACCOUNTED FOR e SUBMERGENCE EFFECTS NEGLIGIBLE e ADS EFFECT s1 H
- MARGINS NOT SIGNIFICANTLY AFFECTED
. PEAK BROAIENING APPLIED TO RESP 0flSE SPECTRA JBH 9/10/80 4-2
MARK III SRVA REVIEW QUESTION 5:
~
WE BELIEVE THAT ONLY THE PRESSURES ASSOCIATED WITH THE ENVELOPE OF THE FOURIER SPECTRA 0F THE 59 MONTE CARLO SIMULATIONS CAN BE CONSIDERED TO BE THE " BOUNDING" FORCING FUNCTION IN THE 95%-95% CONFIDENCE LEVEL SENSE.
WE ACKNOWLEDGE THAT SUCH A '?RCING FUNCTION CAN BE SYNTHESIZED ADEQUATELY WITH A SUBSET OF THE 59 TRIALS, PARTICULARLY IF SOME FREQUENCY SPREADING OF THE AMPLIFIED RESPONSE SPECTRA (ARS) IS EMPLOYED AT LATER STAGES OF THE ANALYSIS.
THE PROCEDURE DESCRIBED TO SELECT THAT FORCING FUNCTION (S) WHICH IS TO BE USED FOR DESIGN DOES NOT APPEAR TO SATISFY THESE REQUIREMENTS.
ACCORDINGLY, COMPARE THE SELECTED TRIALS WITH THE FOURIER SPECTRA 0F ALL 59 TO SHOW WHAT EXTENT THEY DO OR DO NOT REPRESENT AN ENVELOPE.
JBH 9/10/80 5-1
MARK III SRVA REVIEW RESPONSE #5
. THE REQUESTED COMPARISON HAS BEEN PROVIDED
. THE GESSAR RUN SELECTI0fl METHOD PROVIDES A GOOD APPR0XIMATION TO THE ENVELOPE OF SPECTRAL PEAKS FROM 59 RUNS e PEAK BROADENING OF THE RESPONSE SPECTRA WILL PROVIDE FURTHER BOUNDING OF THE ENVELOPE
. NOT APPROPRIATE TO USE PRESSURES ASSOCIATED WITH THE ENVELOPE
- NOT REPRESENTATIVE OF AN SRVA EVENT (WRONG TIME IIISTORY)
- ENERGY CONTENT IN THE ENVELOPE IS THREE TIF.ES THE AVERAGE JBH 9/10/80 5-2
- -- ~-
mmD
- D'3 w Jb.d i
m wo S
C
- a N
i
% e ee, m
e mee W
/
/*
i.,
- .m
=
4 te em.
m.
p A,
%Eur 4
I i
0 o
q
.y t-
.v) w
-1 J
~ -- ::
li w
W 3
l.
~T
..s.
u N
\\
O m,
t
.s n
~.h s.
.=
h d
as'
=e.
C V
.. ~. - - - -...,
n-.
..,k l
U
's _
~
M
)=;
q l-r 4
s,s
~
o s
w C.
a.
~
W 6 )
l--
4~ e"'
- c. 3 4
- =*
O c-
.e w,
6 M
U-
-4.
f*_.
-*=
4)
C
~'s h'
~
a.
1
- d 1 -.
s
\\r.,
4-i
\\
's M.-
C e^
~
i N
\\
t.
I i
1 h
t
,w b
o c
O m
3*
M b
" w N
d( -,
- .. ~
.4 e 7 ;, v..-c3
- t
~
b I
V en N
I.
e9 C
8 O
y
%=
9
~
- c
]
I C
0 e
0 v.-
r
-Y
+
.'s C
O w
N D
b b
+
- g
=
I C
h G
J R
[
~
+
Q
__ a L.
.i A
J P
C %
q M
q%
9 u
i-5 t
C 2
u I
c
- g
< -.
- w q
5CQ u
"a C.
W D
_~.
4..
~
)
w
)
d bd C
J
~
w u
s.t,s en J
e
,1m'*
q s
~
- ' c s
9
~%N t
[
_. A l
e A
}
l i
i
'l
=w
.-- =.
.f j
. c
^
m y
- w g
I
\\
...\\.
nF 9%
- f *E e
n.
A 4
D""
M Ua
.1.
y n
C 5
i s.
aeeme== + e ene +=wwhu.o + =* e -
4*
-=*'***em-O L '/
Y em.
--e-e e
s M,
Y
~
~
t.
2 b
~
C 6"
a r-w.
o M
=E.
~4
.N Q
m%
~
A V
e g
u T
4'aumm.
f 9Q s.;
.J -
,.a a '
s C
N N
e 7
~
w.a w 2
ar a=3 1CE
~
U C
a
%/
W m
~
h 4g
%'*,,5 4
4 w
(#
ese w
e i
O k
L
,,i 7
~
~
[ &--..... J
... C I
~7 i
i F
. +..
e
.,, v%
'D D
C C
C e
C
=-
N
),* C *, * '. /}
}ld
' ~
.J~b-v.aM j O.
=
~
_r-
MA.'K III SRVA REVIEW QUESTION 8:
DATA AND ANALYSIS SUGGEST THAT THE LOCAL HYDRAULIC PRESSURE AT THE-QUENCHER CENTERLINE HAS A STRONG INFLUENCE ON QUENCHER BUBBLE FREQUENCY.
VARIATIONS IN THIS PARAMETER CAN RESULT DUE TO DIFFERENCES IN SUBMERGENCE AS WELL AS FROM DIFFERENCES IN WETWELL PRESSURE.
FOR THE ADS TRANSIENT THE LATTER IS APPROXI-MATELY A THREE-FOLD FACTOR HIGHER THAN THE CONDITIONS UNDER WHICH QBF DATA WAS OBTAINED.
THE PROPOSED METHODOLOGY DOES NOT APPEAR TO ADDRESS THOSE EFFE PROVIDE JUSTIFICATION FOR NEGLECTING THESE EFFECTS OR INDICATE HOW THEY WILL BE CONSIDERED.
QUESTION 2(a):
WE BELIEVE THAT THE DATA BASES FOR VALVE OPENING TIME (VOT), VALVE SETPOINT TOLERANCE (VST) AND QUENCHER BUBBLE FREQUENCY (QB YET TO BE PRESENTED IN A DOCUMENTED FORM SUITABLE FOR FINAL AND EVALUATION.
ACCORDINGLY, THE FOLLOWING INFORMATION SHOULD BE PROVIDED IN ACCORDANCE WITH THE FOLLOWING GUIDELINES AND FO a)
A COMPARISON OF THE BACKPRESSURE USED IN THE TEST VS. ACTUAL CONDITIONS SHOULD BE PROVIDED AND ANY DIFFERENCE JUSTIFIED.
JBH 9/10/80 8-1
MARK III SRVA REVIEW RESPONSE #8 o
EFFECT OF BACKPRESSURE o
LOCAL HYDROSTATIC PRESSURE AFFECTS BUBBLE FREQUENCY THROUGH ITS EFFECT ON BUBBLE RADIUS AND PV WORK.
o COMPARISON OF TEST VS. MARK III CONDITIONS o
TEST RANGE 1.33 - 1.52 ATM o
MARK III NON ADS 1.41 ATM ADS 1.75 ATM o
SUBMERGENCE EFFECT NEGLIGIBLE o
v.1 Hz PER FOOT OF SUBMERGENCE o
CONTAINMENT PRESSURilATION EFFECT o ADS COMBINED WITH SBA LOCA GIVES + 5 Psi IN MARK III o
a 1 Hz IN BUBBLE FREQUENCY o
NOT SIGNIFICANT P:R LOAD DEFINITION
/ 0/80
. MARK III SRVA REVIEW QUESTION 10; WE BELIEVE THAT THE USE OF THE REACTOR PRESSURE RISE RATE (PRR)
AS A RANDOM VARIABLE MAY NOT RESULT IN THE " WORST" CASE FOR SRV LOADS.
TO DEMONSTRATE THAT THE " WORST" CASE HAS BEEN CONSIDERED, COMPARE THE FORCING FUNCTION CALCULATED BY ASSUMING THE PRR AS A' RAND 0M VARIABLE WITH THE FOLLOWING:
(1)
THE FORCING FUNCTION CALCULATED BY USING THE MAXIMUM PRR; (2)
THE FORCING FUNCTION CALCULATED BY USING A PRR WHICH WILL RESULT IN THE MOST PROBABLE "IN-PHASE" BUBBLE OSCILLATION BETWEEN THE SECOND AND THIRD SRV's SETPOINT GROUPS.
TO MAXIMIZE THE POTENTIAL FOR BUBBLE OSCILLATION IN-PHASE, THZ MEAN VALUES OF BUBBLE FREQUENCIES AND SRV LINE VOLUME SHOULD BE USED TO DETERMINE THE PRR.
i
=
JBH 9/10/S0 10-1
MARK III SRVA REVIEW BESPONSE #10 o
SENSITIVITY STUDIES SHOW MAXIMUM LOADS D0 fl0T ALWAYS OCCUR WITH MAXIMUM P.1 ESSURE RISE RATE.
o BOTH CASES REQUESTED ARE IflCLUDED IN SEtlSITIVITY STUDIES o
MOST PROBABLE IN-PHASE OSCILLATION OCCURS WITH PRR = 80 PSI /SEC o
THE GESSAR METHOD IS MORE CONSERVATIVE THAN SELECTI0f! 0F ANY SINGLE VALUE FOR PRESSURE RISE RATE.
f$0/80 10-2