ML19345G938
| ML19345G938 | |
| Person / Time | |
|---|---|
| Issue date: | 03/09/1981 |
| From: | Long L Advisory Committee on Reactor Safeguards |
| To: | Advisory Committee on Reactor Safeguards |
| References | |
| ACRS-CT-1320, NUDOCS 8104220783 | |
| Download: ML19345G938 (6) | |
Text
-
= m y g m..iu m MACGR SAFEGUARDS.U1NAC-
~
C.T-13 A6 WAR M9@ g 4,9,1011,12,ht 3M t
}
e \\ '% i vh G[
,\\
Report to Advisory Comittee on Reactor Safeguards
{ s ry
,f b
by Leland Timothy Long
- y 3
l i {,y s.
ACRS Subcomittee Meeting - Sumer plant APR } 0198;u h OL Review, February 26-27, 1981
/
g Coments on estimate of maximum reservoir induced earthq%. kp
Subject:
?>
'\\
d' The applicant, South Carolina Electric and Gas, has attempted to g
apply an interesting concept to the siting of plants near reservoirs exhibiting induced seismicity. They have attempted to define the stress conditions near the plant and hence make an intelligent and definitive statement as to the occurrence of an earthquake in the next 50 to 100 years. They should be comended for their attempts as the techniques they are developing could be vital in future hazard analysis.
The applicant has shown effectively that the general character of the stress field near the plant consists of an inhomogeneous ambient stress field with wavelengths less than or about 1.0 $.' Based on this conclusion and other supplemental data theyJconclude that the largest induced earthquake would have a magnitude of 4.0.
They then propose to use a magnitude 4.5 for testing purposes.
I perceive two fundamental problems with their arguments:
1.
They use the description of a general or typical condition to argue (improperly) that the specific occurrence of a large event could not occur. The demonstration of a typical or general condition does not preclude the existence of nontypical conditions. Along this line of argument, their data while presenting a strong base for the general i
l conditions do not define all the specific structures within a reasonable I
8104220783 m
v c.,
.-s-.
e
. - - +,.
m.
o f
radius of the plant. Hence, they can not preclude existence of conditions conducive to an event of magnitude greater than 4.0.
Their argunents to limit the extent of rupture along a fault rely 2.
heavily on the existence of " stress barriers" which they claim to be impe netrable. I believe the tenn " stress barrier" is misleading in that (1) it represents a surface (or zone) of zero deviatoric stress, (2) must move in response to changes in the stress field, and (3) may not offer significant resistance to a larger event derived from a near by stress It is conceivable that more uniform stress fields exist below the f i el d.
1/2 to 1.0 km depths proposed for the stress barrier and that these stresses could be the controlling mechanism for a larger event.
Taken together, the evidence presented in support of their argunent strongly indicates the existence of an inhomogeneous stress field.
However, I was not convinced by any single piece of evidence and in many instances the data presentation was weak and unconvincing. Some particular example follow:
A_. The gravity data at 1/2 to 1.0 km spacing is insufficient to do anything other than detect some of the structures proposed to have maximun dimensions of about 1.0 km. To be convincing, the analysis should include modeling attempts and statistics on the precision of the data. The lack of point plots made it impossible to evaluate the quality of the data, particularly near two or three anomalies that appeared too sharp to be real and may represent data errors.
L The aeromagnetic data, like the gravity data, were of insufficient density to provide meaningful analysis.
..z.__._-
,n.
The use of recursion statistics as evidence for maximtrn magnitude
_C2 was not sufficiently qualified or tested. Potential alternate sources of a change in slope exist and include affects of attenuation on coda length and instrunent response versus corner frequency interaction. The ambiguity of conclusions based on recursion relations for limited data sets can easily be verified by examinations of other reservoir areas. Clark Hill prior to 1974 magnitude 4.4 event exhibited a prolonged sequence of magnitude 2.0 to 3.0 events and a high "b" value. Jocassee reservoir induced a magnitude 3.8 event after most other activity had decayed. A magnitude 4.4 event near Lake Sinclair in 1964 was also overlooked in their review.
0 I do not share their confidence in hypocenter depth estimates 2
which are critical in defining the " stress barrier". Less than 10 percent of the events actually fall under a station (f.e. within a distance less than their estimated depth) which is a criteria essential for reliable
~
depths. Also, if the area is as inhomogeneous as they b.eMeve it would be virtually impossible to locate depths except where the depth was less than the distance to the nearest station.
L Focal mechanism solutions which are also essential in defining the stress field are presented with no estimate of uncertainty. Without such an analysis, many are not sufficiently well defined to use to define the directions of the principle axis of the stress field. It is important here to point out that different focal mechanisms can be observed ir, a single active zone which is compatible in size with the dimensions of faulting of a large event. Thus, the existance of changes in stress direction do not necessarily preclude the existance of larger events.
~
j L Most data for reservoir induced events are exhibiting stress drops of a few bars and often significantly less (e.g. 0.05 bars). The high stress drop estimate (25 to 100 bars) for the single event, is a preliminary analysis for only one event and not at all typical (perhaps unrealistic) for stress drops on large faults.
L The use of intensity VI areas in magnitude detennination for Eastern United States earthquakes has not been fully verified. I think the data base is too sparse to allow VI areas to provide valid estimates.
L The statement that events occurred prior to impoundment has significant implications. Why were these events not located? They could imply a potential for seismic activity prior to impoundment and hence a potential for a larger event. The analysis of these preimpoundment events is totally inadequate.
Wateree Creek Fault: Data available and presented at the meeting did not convince me that the Wateree Creek Fault could prove to'b'e' a causative feature for earthquakes. However, more data should be obtained pertinent to the possible extension of the Wateree Creek Fault.
Estimates of site induced earthquakes in next 30 years with probability of 90".
99". and 99.9". (specific request of comittee chainnan)
Mean or typical value for maximurn earthquake. If one uses the largest earthquakes observed near reservoirs which have induced seismic activity in the southeastern United States, then a mean value of 4.1 is obtained (CHRA, 4.3 2.5; Jocassee 3.8 e.3; Kiowee 3.9 t.3; Sinclair 4.4
.5) where the
.w-.-
1
)
for the errors quoted are estimates of about two standard deviations Hence the 50%
assembly of techniques used to determine magnitude).
A probability is that a 4.1 magnitude event will be equaled or exceeded.
time limit proportional to the duration of available seismic data would be Extension beyond 30 years would be appropriate (hence 25 2 5 years).
questionable because reservoir loading in a transient incroachment or For longer time periods one should revert to the statistics of nature.
Determinations of ptobabilities for 90.0, 99.0 and regional earthquakes.
99.9 percent confidence levels require that one first accept the Poisson distribution as appropriate and that second the data are sufficient to I do not believe the first is correct a'nd define the standard deviation.
However, for the benefit of the data set is insufficient for the second.
those who require ntnbers, I would place the 90% confidence level at a The 99 and 99.9 percent confidence levels would be at 4.7 magnitude 4.5.
and 4.9 respectfully, but these values far exceed the range of the data.
I would prefer to stand by the analysis of a maximtn plausible event I The submitted to the Corps of Engineers for the Richard B. Russell Dam.
argtnents were based on physical conditions and a mechanism for Piedmont Hence, I would say that the maximtsu possible " induced" earthquakes.
.3) and would be earthquake would be in the range of 5.0 to 5.6 (i.e. 5.3 In compatible with the largest observed event in the Piedmont Province.
stenary, the largest event at the site should be the largest possible earthquake in the province unless the stress field near the site can be A second question concerns the probability of completely defined.
If one relates this directly to the reoccurrence of a exceeding 0.2 g.
A recent Charleston type earthquake then a rough answer can be provided.
1 study (M.S. thesis by M. Onnsby) gives a probability of recurrerr.e of an q
intensity X event in 200 years as 0.12.
If one assumes the radius of influence of sucit an event to be about 100 len (to intensity VII) of within an area of approximately 500,000 len (the area base used in the estimate of 0.12) then the probability of exceeding 0.2 9 would be about 0.001 (or equivalen:1y one chance in 200,000each year).
l l
l l
l by;g 25533K yt
- ' ' - ~ ~
-