ML19291C049
| ML19291C049 | |
| Person / Time | |
|---|---|
| Issue date: | 11/21/1978 |
| From: | Hau S Office of Nuclear Reactor Regulation |
| To: | Knight J Office of Nuclear Reactor Regulation |
| References | |
| REF-GTECI-B-06, REF-GTECI-PI, REF-GTECI-RV, TASK-B-06, TASK-B-6, TASK-OR NUDOCS 8001110175 | |
| Download: ML19291C049 (8) | |
Text
/
UNITED STATES
(
- UcLcAn UcGULATORY CO*.**.*:cc:L't.
e jx
- Z (f WASHINGTON, D. C. 20000 ne e s o\\'TNf N0y 21 19T8 i
MEMORANDUM FOR:
J. P. Knight, Assistant Director for Engineering, DSS THRU:
f R. J. Bosnak, Chief, Mechanical Engineering Branch, DSS S. N. Hou, Mechanical Engineering Branch, DSS FROM:
SUBJECT:
SRSS EVALUATION & STAFF INTERIM POSITION The staff of MEB/ DSS, SEB/ DSS and EB/ DOR have completed their review of the criteria proposed by Dr. Newmark and Dr. Kennedy regarding application Our comments on the of SRSS method for combining dynamic responses.
proposed criteria and staff interim positior, on SRSS applications are enclosed.
1
<l, v
nou-nien Hou Mechanical Engineering Branch Division of Systems Safety s,
cc:
R. Mattson, DSS R. Bosnak, DSS F. Cherny, DSS H. Brammer, DSS R. Mattu, DSS D. Jeng, DSS F. Schauer, DSS D. Eisenhut, DOR j[}f jgg B. Grimes, D0R V. Noonan, D0R K. Wichman, DDR S. Hosford, D0R p
8 001110 ll[
Staff Evaluation of Newmark-Kennedy Criteria on Combination of Dynamic Responses
~
1.
For Criterion #1:
- e A.
Summary of the criterion:
It was proposed that SRSS can be used if each of the time i
functions, in either loading or response, has function character-istics similar to earthquake ground motions.
It means to meet the following:
1.
Limited number of high peaks:
- 5 or less exceeding 75% of the max., and
- 10 or less exceeding 60% of the max.
k 2.
Approximately zero mean 5
3.
'Short duration
- 10 seconds or less 4.
Time functions are relatively uncorrelated l
i
- from independent events, or I
l
- having random peak phasing i
B.
Staff connents:
1.
T.he proposed criterion represents a fresh approach by utilizing characteristics of time function, instead of attempting generic conclusions from a few case studies as done by others.
Generally speaking, the criterion is simple, easy to apply, and may have a 1734 187
good chance to achieve high non-exceedance probability in most cases.
However, the criteri n appears to be based on engineering judgement without the needed data basis to verify the claimed goal of achieving 84% non-exceedance probability.
2.
We are not certain the parameters used, such as number and level of high peaks, durations, etc., are sufficiently adequate to ensure justifiable SRSS combination in all cases meeting the criterion.
For instance, one important condition to viarrant the use of t
SRSS is that both time functions shall be rapidly varying.
Since " rapid varying" is a relative tenn, a comparison of frequency contents in time functions may be necessary.
The staff suggests that to conduct a search of controlling parameters to the characteristics of time functions in conjunction with sensitivity studies of those parameters that may.
be helpful to verify the adequacy of the criterion.
As an alternative, numerical examples associated with non-exceedance probability evaluation (i.e. use CDF) may be conducted to develop the data base for verifying the assurance level of the proposed criteria.
1734 188
The proposed criterion indicated that the use of SRSS can 3.
I be based on the characteristics of either the leading time
/
functions, or the response time functions. We feel that combination of responses should' be solely based on the characte7-l ihicsoftheresponsetimef ctions.
Since the characteristics of the responses are effected by the j
structural dynamic properti_es, such as natural frequencies, 3
ing values, etc. even the same loading can produce -esponses da of various nature. Thus a loading time function meeting the I
criteria may not necessarily mean that the response time functions
\\ it
)
will also meet the criteria.
4.
It is rather difficult to determine whether;two time functions
}
~ are strongly correlated, or weakly correlated just by observation alone.
Furthermore, judgement by 6 pinion may vary from person
~
to person.
We have adequate confidence that the level of correlation among earthquake ground motion components is weak.
However, to determine the level of correlation among response time functions other than earthquake event may present a problem.
diditional work of calculating the coefficient of correlation
.may be needed. The work will be compounded when an increased number of time fu tions are needed to be combined at one time.
No 5.
Many response time functions may not have zero mean.
guidance is provided to resolve the case.
~
1734 189
. II.
For Criterion #2:
A.
Summary of the criteria:
A SRSS may be used if CDF of combined response 'ime functions meet the following:
1.
SRSS represents at least 50% (1) non-exceedance probability.
~
1.2 SRSS represents at least 85% (1) non-exceedance probability.
2.
B.
Staff coninents:
It was stated at the beginning of the report that the intent l,
of the proposed criteria of response combination is to achieve a non-exceedance probability of 84% (1). We feel i
that the proposed criterion #2 may be in error to use SRSS N
based on a 50% (1) non-exceedance probability, since it is N
inconsistent with the intent.
Furthennore, we are not clear of the purpose and the basis to add another requirement of 85% (1) non-exceedance probability at 1.2 SRSS, since the proposed value to be used is SRSS, not the 1.2 SRSS.
One of the many possible reasons for people to promote SRSS is because the method is simple and convenient.
If we go all the way to generate CDE, the value at the 84%-non-exceedance probability may be more justifiable than the SRSS values, then why bother to look back to two SRSS related values?
1734 190
. ~
2.
Since CDF curve is used in the criterion, guidance to ensure the validity of the curve may be necessary.
Notice that CDF curve may not be unique due to following circumstances:
a.
Response time function is not unique.
b.
Durations of the strong motion portions taken for response combinations are arbitrary.
c.
The probability density function assumed for defining phasing relationships among response functions is arbitrary.
Thus, how to define a bounding CDF curve to ensure adequate s
coverage of all mentioned uncertainties is essential to the application of the proposed criterion #2.
1734 191 e
Of D mD k d
.t:0 in: ri Accept:n:c Criteri M'
W
='
for Combination of Dynamic Responses Usir.g SRSS 1.
Each of the response time functions to be combined by SRSS should consist of the following characteristics:
A.
Function is rapidly varying with time.
B.
Duration of the strong motion portions is short.
C.
Function consists of a few distinct high peaks in random appearance.
D.
Response is not associated with r.ormal plant operation cccnt.
E.
Phasing relationship among functions to be combined are random.
F.
.Eesponse is calculated on linear elastic basis.
2.
SRSS may be used in combining responses to three perpendicular component loadings if the component loadings are relatively uncorrelated.
3.
SRSS may be used in combining modal responses provided that their frequencies are not closely spaced as delineated in Regulatory Guide 1.92.
4.
SRSS may be used in combining two responses if the SRSS level represents a non-exceedance probability equal or greater than 84%.
If the proba-bility is less than 84%, the level at 84% may be used.
1734 192
2
~
5.
For combining more than two responses, SRSS may be used in co...bining any two of the responses, provided all response time functions having characteristics as delineated in Item 1.
iiie remaining responses have to be combined in absolute sum.
6.
SRSS may be used to combine more than two responses if the non-exceedance probability equal or greater than 845 can be demonstrated.
7.
SRSS may be used if it can be verified that the dynamic margins at the specific structural location can adequately compensate for the uncertainty of SRSS as indicated in the exceedance probability.
8.
SRSS h.ay be used if it c'an be verified that the overall design rargins at the specific structural location can adequately compensate for the uncertainty of SRSS as indicated in the 'exceedance probability.
9.
SRSS may be used if it can be verified that the overall stress resultant is far greater than the dynamic portion of the stress and the uncertainty portion induced by using SRSS can be compensated by justifiable means.
SRSS may be used to combine responses of LOCA and SSE in the main l' oops 10.
of the lightwater plant.
1734 l93
_