ML20003F919
| ML20003F919 | |
| Person / Time | |
|---|---|
| Issue date: | 09/13/1980 |
| From: | Murley T NRC OFFICE OF NUCLEAR REGULATORY RESEARCH (RES) |
| To: | Harold Denton Office of Nuclear Reactor Regulation |
| References | |
| RIL-103, NUDOCS 8104230887 | |
| Download: ML20003F919 (6) | |
Text
'e Y
e-e v
/
[
o, UNITED STATES y
j NUCLEAR REGULATORY COMMisslON s
. a w AsHINGTON. D. C. 20555
't
(..../
- 4 SEP 13 Q MEMORANDUM FOR: Harold R. Denton, Director Office of Nuclear Reactor Regulation S
FROM:
Thomas E. Murley, Acting Director Office of Nuclear Regulatory Research i.
RESEARCH INFORMATION LETTER NO.103 =BEST ESTIMATE-2
SUBJECT:
EVALUATION METHOD (BE-EM) APPLIED CALCULATION OF THREE-
?.
DIMENSIONAL SEISMIC RESPONSE" c.:
This Research Information Letter (RIL) describes how two techniques, Best Estimate and Evaluation Method, can be applied to the traditional seismic analysis and design of a nuclear power plant. The seismic analysis and design methodology chain is cceprised of seismic input, soil-structure interaction, structural response and subsystem response.
e I
The objective of this study is to characterire the compounding effect of
{
conservative assumptions made at various links of the seismic design Future work should provide further insight on the overall chain.
conservatism in seismic analysis and design methodology as well as on relative conservatists by selectively combining desired design links.
~
An illustrative example is used in this study that links three-directional seismic excitation and structural response to compare the results of both techniques in terms of factors of comparison and probabilities of exceedance.
h INTRODUCTION NRC has established regulations, guides, and licensing review procedures k-that define seismic safety criteria for nuclear power plant design.
These criteria collectively constitute a seismic methodology chain (SMC). The seismic safety criteria for nuclear power plant design were J.
developed to ensure structural integrity and functional safety of buildings, equipment and components. They depart from the conventional S.
The overall earthquake engineering practice in detail and complexity.
l SMC is considered sufficiently conservative to ensure safety, however.
it is necessary to characterize the overall seismic safety and to Since this
~
improve it by estabitshing new criteria as may be required.
RIL summarizes results obtained from the Seismic Safety Margins Research Program (SSMRP), background inf'onnation on the SSMRP is included.
SEISMIC SAFETY MARGINS RESEARCH PROGRAM (SSMRP).
The SSMRP will provide the methodology to detennine safety margins in a nuclear power plant subjected to a large earthquake. The objectives of the SSMRP are to estimate the conservatisms (or lack of conservatisms) in the Standard Review Plan (SRP) seismic safety requirements and to y ~<
develop improved requirements.
8104280 7
E.
.' M
(
Harold R. Denton 2
There is a need to reexamine the traditional process of seismic analysis and design of nuclear power plants in an overall system context. This need comes principally from the widely held belief that a compounding of conservatisms ' occurs in the current process. That is, at each stage of the current process, conservatisms are introduced to account for un-certainties, and these conservatisms compound from one stage to the i
next. However, in each stage only minimal compensations are made for the compounding of conservatisms because they are not quantified.
For
',S example, the earthquake used in the seismic design represents the t
maximum earthquake potential (" safe shutdown earthquake" (SSE)) con-r.
sidering the geology and seismology, and specific characteristics of the y'
subsurface material. The earthquake motion is coupled to the bedrock and building foundation through the use of conservative soil properties to produce the highest responses (forces and stresses). Such responses
?
are compared to conservative estimates of the strength or capacity of
~-
each structure or component.
.j.
The SSMRP will develop an improved deterministic seismic safety design y
methodology and a methodology to perform earthquake risk assessments of i
nuclear facilities. Risk will be measured by various failure prob-abilities and by the probable release of radioactive materials. The 3
approach used integrates the elements of the seismic chain, including:
j Earthquake characterization Soil-structure coupling f.
j Structural building response t
}',
Subsystem structural response Local failure or loss-of-function l
g Systematics of i:ow local failures could combine and lead to a release.
l l
=
h These elements will be characterized realistically and probabilis-l y,
tically, rather than conservatively and deterministically.
.h DESCRIPTIVE
SUMMARY
5 1
In this concept, the systematic evaluation of the seismic analysis and design chain of a nuclear power plant can be simplified to encompass C
seismic input, soil-structure interaction, major structural response and L
subsystem response.
k The objectives of the present study are:
5A (1) to introduce the concept of BE-EM with respect to the seismic l
l J
analysis and design of nuclear facilities, i+.
~
..C p
l p
'P
?.T s:
L L.
r Harold R. Denton 3
(2) to demonstrate BE-EM through an illustrative example showing the coupling effects between seismic input and structural response, and (3) to show the sensitivity of response to three components of seismic motion.
i Details of the study can be found in reference 1.
'M OVERVIEW 0~ ANALYSIS II
~ (-
The present investigation is an extension and coupling of two previous 2
studies. The first addressed the topic of synthetic time-histories and their combinatior. versus recorded ground motions. The second addressed the practice of enveloping and broadening of in-structure response
-y spectra to account for uncertainties.
The key elements of the Best Estimate (BE) Method are:
.;v 7'
(1) the base excita-icns were the three components of recorded ground motion applied simultaneously and with their recorded
[
phasing;
)
(2) the variability in stiffness and structural damping was incorporated i.
in the analysis by randomly sampling on the assumed distributions,
- and,
- [
(3) the mean and mean-plus-one-standard-deviation (MSD) in structure response spectra, were generated.
}",
The corresponding elements of the Evaluation Method (EM) are:
- t..
(1) The base excitations were synthetic time-histories applied in
' f,'~
each of the horizontal and vertical directions independently.
'}f The resulting in-structure response spectra being combined
(,-
by the square-root-of-the-sum-of-the-squares rule (SRSS); i.e.,
,.i the spectral ordinate (S) at a point in direction 1 is computed by:
{
.s.
2 + 3]2 + 3]2)
[
Sj = (Sj e
tj where:
ff S
= response spectrum ordinate in direction 1 j
p; due to thr'ee components of motion (1, 2, 3) and response spectrum ordinate in direction 1 due to 5
33 = an excitation in direction j (j = 1, 2, 3).
[
.s t3 R
(2) The variability in stiffness and damping was incorporated by
((Q smoothing and peak broadening of in-structure rerponse spectra.
rh (3) The mean and MSD response spectra were generated.
..q
()E
I ~';
c 9.s.
t
~
Harold R. Denton 4
Two quantities were used in the comparison of BE and EM response.
Factors of Comparison (FOC) were computed as the quotient of the mean EM response spectra and the mean (or MSD) BE response spectra.
In addition, Probabilities of Exceedance (POE) were computed which estimated the probability of a BE response exceeding in corresponding EM i
response. Both F0C and POE vary over the frequency range of interest.
The results demonstrated the conservatism of the design criteria defined
..?
by the EM procedure (subject to assumptions of the study).
L 9
CONCLUSIONS AND RECOMMENDATIONS 1
,N This study represents the first attempts to systematically compare two 4
seismic methods.
Future studies should include as many links of the
?
SMC as accropriate to realistically analyze the ohenomencn of interest.
There are two key points to be empnasized in the BE-EM concept.
For y
example, in the BE-EM concept, it could be misleading to compare a soil-L.k structure interaction result such as base-mat response instead of a
'(
design parameter, including structural and subsystem response. Second, i
the basis of comparison will, in most cases, be statistical; that is, 7
mean vs mean, mean vs MSD, mean vs point estimate, etc. When calculating g
a BE response, this will always be the case since BE by definition includes a measure of uncertainty. The EM may or may not be statistical.
The lack of recorded data makes it impossible to calibrate the BE of seismic response against true behavior. Therefore, each link in the SMC 5
will require the formulation of a' ternative models.
Comparisons will be made between these models and the design methodology.
![g Conservatism
~
The results from this study showed that the mean FOC, calculated from
[ 7-the floor response spectra (called in-structure response spectra in i
reference 1), varied between 1.5 and 8.
The reason for the variation is jy that overestimation of response occurs at frequencies coincident with f.
the natural frequencies of the structure. The FOC is indicating that l(?;.
the seismic design loads for piping, for example, are 1.5 to 8 times the value that would be calculated using real earthquakes.
l '2 The POE is a probabilistic method to dgtermine the conservatisms in the structure. The POE varied between 10~ to 10-5 for the study. Stated I
another way, if a number of earthquakes occurs, each with a peak ac-
'1 celeration equal to the SSE, the probgbility of exceeding the design
- .3 seismic loads for piping would be 10~ to 10-5 The probability of f
earthquake occurrence was not considered in this study.
Q If.;
Floor Response Spectra lf' This study showed that the conservatism in floor response spectra at any N
g f
jh single point in the structure varied with frequency. By making the jg,
floor response spectrum smoother, a more uniform conservatism can be
' [.N 7
rL
~
h.,
.(
(
Harold R. Denton 5
obtained. This means that typical floor response spectra should have
?
sharp peaks broadened and lowered, and the valleys raised.
The study als'o pointed out that the conservatisms in the floor response spectra varied from point to point within the structure. This means there is no assurance that conservatism is applied where it is required
}
,G or where it would do the most good. The variations in conservatism can be removed or placed in a specific manner to satisfy safety objectives
. j' by specifying target values of conservatism. These values would be specified in a seismic design performance specification. Results from
. {*
the SSMRP, individual's judgement and other risk studies could be used i
to develop the target values.
Three Dimensional Response
?
The conservatisms described in the report resulted primarily from the methods used to account for the three components of motion. Present review criteria focus primarily on defining the peak horizontal ground-ig.';
acceleration for the SSE. Little requirement is placed on the relation between the amplitude and phasing of peak acceleration for the three
'I /
components, or the relationship between the three time-histories. This 4
study showed that changes in relationship between the three components of motion time-histories significantly varied the seismic response.
r ji Therefore, it may be as important to review various aspects of the three components of motion as it is to review the SSE peak acceleration.
t[
This study is the first attempt to systematically compare two seismic
- c methods. Further research on the concept is intended in the near future, however, we suggest that the results of this preliminary study
),
be reviewed and considered for future use in the regulatory review process.
t7 lt*E i.,
Thomas E. Murley, ting Director M
Office of Nuclear Regulatory Research UE.
I cc:
F. Schroeder, DST, NRR
- ?
G. Knighton, RCSB, NRR
'.tt
'.'i
,k p
e{
ll;
" I'.
IN Ih
.N
'8_
- p v
i e
i s.
(
REFERENCE S. E. Bumpus, J. J. Johnson and P. D. Smith, Lawrence Livemore National Laboratory, Livermore California, Best Estimate Method Versus Evaluation Method: A Comoarison of Two Technicues in Evaluating Seismic Analysis anc Design, NUREG/CR-1469, UCRL-52746, July 1980.
s, F g
es b
' O r, :,
.s
,d 4
g i
5 a
- [
F J
h
.P i
'.s
?..
.4, 2T
' f.T ik p
l
(,.
i?
f S..
gC.
~
- .E
.n
.h.
97 i
ih e
.w 9
m
.-g ys,--
-%w.