ML19343D210

Comments on Util Rept on Supplementary Seismologic Investigation,Ser & a Murphy Jan 1981 Memo.Supports Murphy Conclusion That Max Reservoir Induced Earthquake Estimates Are Not Sufficiently Conservative

ML19343D210
Person / Time
Site:	Summer
Issue date:	02/25/1981
From:	Simpson D COLUMBIA UNIV., NEW YORK, NY
To:	Boehnert P Advisory Committee on Reactor Safeguards
References
ACRS-CT-1319, NUDOCS 8104090930
Download: ML19343D210 (4)

Text

.

Lamont-Dohcrty G;ological Obs rvatory Pclis des, N.Y.10964 of Columbia University CT-/3/9 w= nsemans 25 February 1981

--n Mr. Paul Boehnert Staf f Engineer Nuclect Regulatory Cousnission Advisory Committee on Rasetor Safeguards Washington, D.C. 20555

Dear Mr. Boehnert,

At your request I have reviewed the following material concerning induced seismicity at Monticello Reservoir and the maximum likely earth-quake for the Virgil C. Summer nuclear station:

1. Report on Supplementary Seismologic Investigation: Virgil C. Sunaner Nuclear Station Unit 1 - prepared by South Carolina Electric and Gas Company, December 1980.

2. Safety Evaluation Report by the Office of Nuclear Reactor Regulation, U.S. Nuclear Regulatory Com=is,sion Related to Operation of Virgil C.

Su=mer Nuclear Station Unit 1.1, Reference number NURIG-0717. .

3. Memorandum from Andrew Murphy, Site Safety Research Branch, Division of Reactor Safety Research to Robert Jackson, Chief, Geosciences Branch, Division of Engineering, concerning Recoussendation of Maximum Reservoir - Induced Earthquake at the V. C. Sususer Nuclear Station -

dated 1/81.

In general, I support the conclusion by Dr. Murphy that the maximum resevoir induced earthquakes of M, 4.0 (applicant) or Mg 4.5 (1ASL) are not suf ficiently converserygtive ' based on the available data and tech-niques. My assessment would[that, while arguments presented by the appli-cant and LASL based on size of the active sone, magnitude-frequency rela-tionships and intensity-magnitude relationships indicate a maximum magni-tude of 4-4.5, the data and assumptions on which these arguments are based are not sufficiently well-established to provided constraints of better than 11/2 to 3 3/4 of a magnitude unit. Thus a reasonably conservative estimate vos;td have to be at least Mg ,5.0.

My main concern is with those inferences drawn from the spatial, temporal and statistical characteristics of the seisaicity recorded to date at Monticello. A ntsaber of studies including my own work at Nurek Reservoir have shown that induced seismicity is closely related to changes in water level at reservoirs. The filling of Monticello Reservoir occurred rapidly and relatively smoothly between November 1977 and January 1978. To my knowledge, the water level has since then remained at or very near full pond. The major burst of seismicity occurred during and iussediately fol-81040'00930 mwam -

s , 2, .

loving the rapid filling. The predominant mode of faulting, as indicated by fault plane solutions, appears to be high angle reverse faulting. (It should be noted that all of these characteristics are similar to the episode of induced seismicity at Manic 3 Reservoir in Quebec.)

It has been shown that in a region of thrust faulting, the combined

~

long-term ef fect of load and pore pressure related to an increase in water level is to produce increased stability beneath the reservoir. Within this everall increase in stability, smaller scale spatial and temporal vari-l ations in stress can produce mones of weakness and hence increased seismi-city such as that observed at Monticello. If the water level is maintained at a constant level, a new equilibrium between pore pressure and load is gradually reached, with a corresponding decrease in seismic activity such as that observed at Monticello and Manic 3. The theoretical work of Bell and Nur and my own observations at Nurek Reservoir show, however, that the most likely times for increased seismicity (and the occurrence of largest earthquakes) follows rapid , decreases in water level. No such rapid de-creases in water level have occurred at Monticello and I assume that it cannot be guaranteed that no such changes will take place in the future.

Thus, while the increase in seismicity during the first filling clearly indicates that the site is prone to induced activity, it is not clear that the parameters of the seismic regime observed during this episode can be considered typical of induced activity during other types of variations in water level.

In the Supplemental Seismologic Investigation (SSI) Section-4, pages l

l 12-15, I note four arguments used to support the use of g 4.0 as the maximum magnitude induced earthquake:

1. Decline in activity - As noted above this is the pattern of induced seismicity expected for the type of filling at Monticello. It should l

not be assu=ed that this decay represents a conplete release of stored I

strain. It can be expected that any future variations in water level I (and especially any rapid draw-down) may be accompanied by further increases in seismicity.

2. Limit of source dimension to less than 1 km - Based on seismicity patterns alone, I note (SSI, Appendix II, Figure 10) clusters of activity at least 3 km and, if clusters 2 and 3 are combined, up to 5 km in maximum dimension. Based on my own experience with studies of shallow earthquakes using dense local networks, I as well-avart of the problems involved in determining reliable depths and fault-plane solutions from such data. I do not consider the absence of any consistent planar depth distribution in hypocenters or the apparent variability in fault plane solutions to be sufficient evidence for ruling out the presence of individual fault surfaces greater than 1 km in dimension. I am not convinced by the evidence from geolog-ical, geophysical or stress data that restricted zones less than 1 km in dimension exist at hypocentral depths; nor am I convinced that if they did exist they would necessarily indicate that fault dimensions would be limited by these zones. In agreement with Dr. Murphy, in estimating the maximum rupte e dimension, I would place more emphasis on the apparent size of cluscers up to 3 km in dimension. It should

(

be noted that, although the total active area at Monticello is smaller j than that at Lake Jocassee, the clusters of activity are similar in

' size and Lake Jocassee has already experienced an earthquake of I .

b

Mt 3.7, a value uncomfortably close the the maximum of 4.0 estimated fEr Monticello.

3. Fre quency-sagnitude relationship - As stated above, it cannot be assumed that the parameters determined from the activity related to filling of the reservoir will be typical of activity during other I would have prefered to see a types of variation in water level.

discussion of b-value in terms of the more standard technique of maximum-likelihood estimation. It should also be noted that the non-linear nature of the b-value curve (with an increase in slope at higher magnitudes) can also be attributed to an insuf ficient obser-vation period, and hence incomplete catalog for the larger, less frequent events.

4.

Area of intensity VI - In agreement with Dr. Murphy, I am skeptical In of the application of these techniques to the southeastern U.S.

addition, I note the following inconsistency in the applicant's analysis in Appendix VIII of the SSI:

The maximum earthquake at 1.ake Jocassee was g 3.7.

corresponds Using the equation on page VM-2 this to log M = 21.6.

From Figure VIII-1 equation 2, this corresponds to log A = 12.7' 3

• * * *1***d7 8'***** th*" Eh* "*

or 100Akm

=3 used as the upper bound as stated on page VIII-3.

I make no coac:ent as to whether the inconsistency lies in the magni-tude determination, the accent calculation or the area estimate, but simply use this as an example of the gross uncertainties in applying techniques of this kind with the limited amount of available data.

Appendix 9 of the SSI includes probability estimates In based on the a report on frequency-sagnitude relationships observed at Monticello.

Joc a s se e-Keovee seismicity submitted to the Division of Site Safety in January 1980, I raised the question of the applicability of standard recur- The rence techniques for risk estimates in cases of induced seismicity.

main point was that recurrence relationships as obtained from natural earthquakes should not be used in cases of induced seismicity where, by definition, the seismic regins has been modified by the reservoir and Thehence above previous recurrence relationships will no longer be applicable.there is also some qu discussion implies that recurrence relationships as observed during short-term bursts of seismi-city on initial filling of a reservoir. I feel that this question deserves further consideration by the NRC.

Although it is not directly related to the question of induced seismi-city, I note considerable discussion in the documents on the Sununer plant related to the Charleston earthquake and the question of decollement reactivation. The model of decollement reactivation is obviously of

'--N -- _- _

---

- . . ~ . .

l

4.-

considerable importance for seismic cvaluation ct sitcs is tha castern U.S. , and I would urge the NRC to encourage further studies of this ques-tion.

• Yours sincerely,

/, h #

David W. Simpson DWS/11e 9

L E

A '

J w

4 a 4 k

, , , , y 9 y. -w q w p-,--,,..,y.. 3 ,py

-... . _ _ , , ,m.,.-.% .- ..-.w -

--..w._ - . - . , . , . -