ML20010F582
| ML20010F582 | |
| Person / Time | |
|---|---|
| Site: | Yankee Rowe |
| Issue date: | 08/05/1981 |
| From: | Bernreuter D LAWRENCE LIVERMORE NATIONAL LABORATORY |
| To: | Russell W Office of Nuclear Reactor Regulation |
| Shared Package | |
| ML20010F580 | List: |
| References | |
| EG-81-28, NUDOCS 8109100377 | |
| Download: ML20010F582 (4) | |
Text
EllCLOSURE 1 t=.
..c,
.~. >.
s
(
"b I
HUCLEAR SYSTEMS SAFETY PROGRAt4
. _..J EC-51-2d Augusb 56 1981 N1iilam Russeil *'
Mr.bematic Evaluation Program Branch Sys
' Division of Licensing U.S.ington, Nuclear Regulatory Commission Wash D.C.
20555
~~
SUBJECT!
Review of bhe Repork (YAEC-1263), "Seismi Response Spectra for the Yankee Nuclear P3we$5 eM Statinn, Rowe, Massachusetts," Yankee Atobicth-SG Company, June 1981.
h 5
]$
beir Nr. Russdi1) j Ab your requesk I have reviewed the reporb YAEC-1263 and 4
related reports,- My review has focused on methodology and opplication of,the methodology rather than on what represenks an acceptability conservative spectrum to be used for the analysis of the Yankee Nuclear Power Plant, Thus I have only cttempted to judge how well the 1000 year and 10,000 year-return period spoetra represent the seismic hazard at the Yankee site.
The key concius1ons In lhe Reporb YAEC-1263 seem ho me lo be,i I}
The compos 1he spectrum (based on a peak ground conservative estimate of the 10 gg, 6,1 is aprobabilishle acceleration of 0.lg) shown in F j
spectrum.
l 2)
There are no anomalous amplification ef fects ob hhe l
Yankee site and any potential spectral amplification 1s accounted for in both the deterministic and l
probabilistic spectra given in YAEC-1263, do noh agree with these conslusions.
In f act, my c dnion efter reviewing the study.YAEC-1263 is thah it provides udded confirmation of the results given in Vol. 4 o f NURE.G/CR-1582 (the LLNL/ TERA study).
The rest of hhis letter briefly, documents.my conclusica that the results given in Vol. 4 of NUREG/CR-1582 are a better representation of say the 1000 year 8109100377 010904 PDR ADOCK 00000029 p
~
i
7)
EG-81-28 Augush 56 1981 rekurn spechrum khan lhe speckra given 1n YAEC-1263.
Ocha11ed documentation of the conclusions will be provided by hhe end of August.
The key elements of dny methodoIogy ko essess khe seIsmkc bczard at a site are:
ih Sciection of a iiazard assessmenh approach, e.g.,
khe uso o f McGuire's computer program (probabilistic approach).
2)
Selection or an approach ko incorporate the uncertainty in all input parameters into the analysis.
i 3)
Selection of an overall earthquake occurrence model.
This includes zonation, rate of scismic activity in each zone, largest carthquake that can occur in each zone end the uncertainty in each parameter including zonation.
A ),
Selection of a ground motion model giving khe reishion between the earthquake size and its potential to generate ground shaking at a given site.
I have no problems with khe use of the McGuire computer program.
It is somewhat limited in its ability to treat uncertainhy as only the uncertainty in the ground motion model is formally incorporated into the computer program.
This limitation can be easily remedied by making additional calculations and analysis to explore the effect that Uncertainty in the various input parameters have on the ostimate of the seismic hazard ab the Yankee Site.
One of Ehe major differences between YAEC-1263 and llNL/ TERA study is how the bounds for uncertainty are obtaIned and uncertainty entered into the analysis.
In the LLNL/ TERA study ten.different experts were used to provide a range of input data.
This resulted in ten dif ferent overali earthquake -
occurrence models including a model very similar to the one used in YAEC-1263.
In my view one of the major shortcomings of YAEC-1263 is the failure ~ho include much of the uncertainty that exists into their models.. For example,.a rather detailed discussion is given to backup the uncertainties used for the ground motion model but only a very limited' study of the effect of zonation changes were made.
One logical choice to have.
studied would have been the Piedmont zone in conjunction with Boston-Ottowa zone as shown on Fig. 9.2 rather.than the_three choices used:
(1) WGC zones, (2) only Boston-Ottowa plus NGC zones to the south; and (3) Piedmont plus WGC zones to the.
north.
No uncertainty on the rates or seismicity were studied e
and only a very limited variation on upper magnitude cut-off
,/
was considered.
e 4
e,.-w
---.---,----w
7s-
'EG1 1-28 kuguck 5; 1981 8
Pulting aside khe queshlon 6f zonab[on which ks very
,difficult to resolve other than by expert opinion I have a
, number of problems with the seismicity parameters used in the analysis.
Given the zonation there are several different ways to arrive at the description of the seismicity in each zone.
Not enough information was provided to directly assess the choice of seismicity parameters used in the analysis dlscussed in YAEC-1263. In order to attempt to verify the values used I used the carthquake history data compiled as paft of the LLNL/ TERA study.
I found significant differences between khe scismicity parameters, rates and "b" values, that I would judge es reasonable, fits to the corrected data (corrected as discussed in Vol. 2 of NUREG/CR-1582).In YAEC-1263 My activity rates were generally much higher than those used These differences turned out to be significant.
At the 10-3 probcbility level _the estimated peak ground acceleration increased about 40%.
Dif ferent- (but reasonable) choices of scismicity rates and "b" values could have even greater effect as generally very steep negative slopes (greater than -1) were used.
Many of the experts in the LLNL/ TERA study felt that a slope of -0.9 should be used which results lo more larger events than for the b values generally used in YAEC-1263.
None of this uncertainty was included in YAEC-1263.
Three ground motion models were considered In YAEC-1263 (1) Bollinger attenuation model; (2) WGC modell (3) Nuttli's
~
theoretical model.
The Nuttli theoretical model was used as one of the ground motion models in Vol. 4 NUREG/CR-1582 but with a somewhat larger sigma of 0.7 as compared to 0.57 used in YAEC-1263..
Considering how uncertain Nuttli's theoretical model is the use of such a low value of sigma does not seem justified which is why a larger value was used in the LLNL/ TERA study.
The WGC model does not seem to.be correct in its scaling with magnitude because Ina scales only as 0.7 M whIch is much less magnitude dependence than typical.
I feel that the problem arises from the inadequate data base used to obtain -
tho model--particularly relative to magnitude scaling.
In order lo assess the impact that hhe Bollinger and WGC ground motion models might have on the results given in Vol. 4 NUREG/CR-1582, I re-ran several experts using's number of ground motion models.
For comparison I am using as the base case the no background case using the Ossippee ground motion model with a 2 sigma cut-off.
The Bollinger Model resulted: 1n about 15 to 25%_ decrease in peak ground acceleration.
The change depended on how many sigmas were used for the Bollinger model.
As used in YAEC-1263 the differences are only about 15%.
If we cut the model off at 2 sigmas this results in aboub a 25% reduction.
The change in peak ground acceleration using the WGC model is expert dependent and varies from a 40%
e A w-
.s.,_,.._m
-w,,,
,g.w,
,,g,,,
___,,,_y
f.
EG-Bi-28 d-Augusk $6 1981 Increase bo aboub 20% decrease depending upon the experb.and number of sigmas used in the analysis.
In conclusion, it Is my opinion khab YAEC-1263 does nok adequately incorporate the uncertainty in zooalion, seismicity percmeters and upper mcgnitude cut-off.
When the ground motion models proposed in YAEC-1263 are used with the overall earthquake occurrence models used to obtain the results given in Vr.l. 4 of NUREG/CR-1502 only modest reduction in the peak gradhd ecceleration occur.
The results are very sensitive to the choice of the ground mollon models but none of the proposed ground motion models suggest that the synthesis results presented in Vol. 4 of NUREG/CR-lS82 are in significant error.
Nor have arguments been developed or data presented which Vould oilow one to chooss one expert's model over another.
Thus a number of zonations, etc., must be used as was done in the LLNL/ TERA study to estimato-the se'ismic hazard at the Yankee site.
~
I cannob agree with the conclusions of Sechion 8 of YAEC-1263 that the Yankee site is a typical soil site and khat
- there are no anomalous emplification effects at the Yankee Site.
No basis is provided in the text to teach these conclusions.
The models studied only provided information obout the frequencies-at which amplification would occur.
No otudies were made comparing how much amplification would occur relative to typical western U.S. sites.
This point was discussed in some detail in my report to you dealing with possible site amplification studies at the Yankee site.
In od; enelysis we also found some emplification less than 1.5 kz.
This question need further study.
Sincerely, t
)4 Don.L. Bernreuter, Group leader Engineering Geosciences Group DLB/v]
O cc:
L. Reiter G.E.
Cummings
/
P.D.
Smith 89 e
4
- -.