ML15237A335
| ML15237A335 | |
| Person / Time | |
|---|---|
| Site: | Diablo Canyon  |
| Issue date: | 08/21/2015 |
| From: | Alliance for Nuclear Responsibility |
| To: | Japan Lessons-Learned Division |
| DiFrancesco N, NRR/JLD, 415-115 | |
| Shared Package | |
| ML15237A311 | List: |
| References | |
| A.15-02-023 A4NR-3 | |
| Download: ML15237A335 (20) | |
Text
Case No: A.15 023 Exhibit No: A4NR-3 Witness: John Geesman
Application of Pacific Gas and Electric ) Company for Compliance Review of Utility ) Owned Generation Operations, Electric Energy ) Resource Recovery Account Entries, Contract ) Application 15 023 Administration, Economic Dispatch of Electric ) (Filed February 27, 2015)
Resources, Utility Retained Generation Fuel ) Procurement, and Other Activities for the Period ) January 1 through December 31, 2014. ) (U 39 E) ) __________________________________________)
SUPPLEMENTAL PREPARED TESTIMONY OF JOHN GEESMAN ON BEHALF OF THE ALLIANCE FOR NUCLEAR RESPONSIBILITY BEFORE THE PUBLIC UTILITIES COMMISSION OF THE STATE OF CALIFORNIA August 21, 2015
1 QO 1: What is the purpose of your Supplemental Testimony? A01: A4NR agreed with PG&E that it would provide a specific ratemaking recommendation in evidence after reviewing additional data responses from PG&E that were received after the July 14, 2015 submittal of its Prepared Testimony. This Supplemental Testimony provides that recommendation and reflects upon those data responses.
Q02: What is A4NR's ratemaking recommendation? A 02: A4NR recommends that the Commission disallow recover y of the $4.56 million recorded in the DCSSBA as costs incurred in 2014 for the AB 1632 Seismic Studies. PG&E' s refusal to interact with the IPRP as required by D.12-09-008 and D.10-08-003, as well as its failure to submit to IPRP review prior to publishing its final report, prevent the Commission from finding that these costs were reas onably incurred. A4NR also recommends disallowance of the $0.90 million recorded in the DCSSBA as costs incurred in 2014 for Project Management of the Long-Term Seismic Program. PG&E's failure to provide for timely IPRP review of the AB 1632 Seismic Studies, and obtain the IPRP's assurance that the AB 1632 Seismic Studies were properly incorporated into PG&E's SSHAC report as intended by D.12 008, prevents the Commission from finding that these Project Management costs were reasonably incurred.
Q03: What role did PG&E's post
-July 14, 2015 data responses play in A4NR's recommendation?
A03: PG&E's data responses strongly reinforce A4NR's conclusion that key analyses that deserved painstaking review were kept from the IPRP
. A4NR requested the deterministic 2 ground motion spectra plots (and associated 10
-6 and 10-7 plots) described in A4NR
-00660 for joint ruptures on the following linked faults: (i) Hosgri linked to faults up to Mendocino Triple Junction; (ii) Los Osos linked to Hosgri; (iii) San Luis Bay linked to Hosgri; and (iv) Shoreline linked to Hosgri.1 Despite the assurances in A4NR-00660, which was written by Geosciences Director Klimczak in the midst of PG&E's IPRP information blackout period, these deterministic joint rupture analyses never made it into the CCCSIP report. Q04: Since each of the deterministic joint rupture plots purports to show ground motion well below the 1977 Hosgri spectrum, what significance does A4NR attach to them?
A04: The methodologies used to calculate these plots should have been discussed with the IPRP. In explaining how it had derived an M8 assumption for each joint rupture, 2 PG&E went further and described the role "saturation" plays in its results:
The differences in the spectral accelerations at the DCPP site would be negligible between a M8.0 and a M8.5. This is because of the short
-distance large
-magnitude scaling known as "magnitude saturation" in ground motion studies. It is accepted within the scientific community, and both empirical observations and numerical simulations validate, that there is not an increase in high
-frequency (> 2 Hz) ground shaking levels for close in sites to shallow crustal earthquakes for magnitudes above about magnitude 6.5 (M6.5).
Q05: Why is that statement significant?
A05: It effectively immunizes Diablo Canyon from high
-frequency ground shaking from earthquakes above M6.5 on close-in faults like Hosgri, Shoreline, San Luis Bay, and Los Osos, 1 PG&E's responseERRA-2014-PGE-Compliance_DR_A4NR_004
-Q01", along with the four spectral plots PG&E attached to it, is included in Appendix as Exhibit
- 1. 2PG&E's responseERRA-2014-PGE-Compliance_DR_A4NR_005
-Q01" is included in Appendix as Exhibit
- 2.
3 not to mention their joint ruptures, and materially underestimates the hazard from near
-source , long-period motions associated with large earthquakes. Q06: But if this "saturation" effect is real, what is the problem?
A06: After consulting with Dr. Blakeslee, A4NR sees several. The magnitude saturation assumption depends upon data extrapolation using far
-field earthquakes , because there simply is not much recorded data from large earthquakes in the near
-field (although there are some very large accelerations which have been observed in the extreme near
-field). The catalog of near-field data that measures the magnitude saturation observationally is small , with a large standard deviation, while using a numerical simulations approach to estimate the effect misses the influence of starting phases, stopping phases, heterogeneous stress drops, and asperities. And the question of shaking duration needs to be factored in. Put another way: shaking for 10 seconds above a prescribed level is very different from shaking for 60 seconds above the same level. In each case the peak acceleration may only be 0.4g but the damage is significantly greater when the duration of shaking is longer. A structure can literally shake itself to pieces over the longer duration of rupture produced by a long (e.g., 300 km) fault vs. a short er fault (e.g., 60 km). Q07: Can you elaborate on the hazard associated with near
-source, long
-period motions?
A07: Dr. Thomas Heaton, Director of the Earthquake Engineering Research Laboratory at the California Institute of Technology (with dual CalTech faculty appointments as a Professor of Geophysics and a Professor of Civil Engineering), expressed written concerns after PG&E's second SSHAC workshop on ground motion characterization:
4 If low frequency motions are a concern (sloshing of storage pools?), then whatever systems are affected are almost certainly not linear systems for very large motions. This means that modal analysis is not appropriate. It is important for the design engineers to directly communicate with scientists about what types of ground motion time histories are plausible. I would strongly discourage the use of "spectrum compatible motions" to simulate non
-linear long
-period dynamics.
The spatial distribution of slip is the key parameter that determines the nature of near
-
source long
-period ground motion. For example, two earthquakes of identical magnitude can have very different average slips. Furthermore the maximum slip can be much larger than the average slip. However, when considering low
-probability long
-period motion, it's critically important to characterize the statistical features of slip on segments of a fault that are close to the site. The current analysis does this problem by characterizing the source with moment magnitude, which is an averaging parameter for an earthquake. Variability is handled by assuming that long-period motions are log
-normally distributed about the mean appropriate for the magnitude. However, I am not aware of any evidence that shows that the slip at a point is log
-normally distributed about the mean. Instead, I would guess that we are looking at a power law distribution (Pareto). For a variety of reasons, I would argue that these slip distributions are approximately fractal in nature. Unfortunately, power law distributions are very hard to deal with when using standard statistical analysis. It may be more appropriate to simply say that PSHA is not well suited for this problem. The key issue is to design structures that are robust with respect to long
-period ground motions. (Please see Yamada, M., A. Olsen, and T. Heaton 2009, Statistical features of short- and long-period near-source ground motions, Bull. Seism. Soc. Am., 99: 3264 - 3274)3 Q08: Did Professor Heaton's post-Workshop #2 comments express any view about magnitude saturation in near
-source, short-period motions?
A08: Yes, as follows: When it comes to high
-frequency near
-source shaking, the evidence is good that observations are compatible with the hypothesis that pga's saturate with magnitude and that they are approximately log
-normally distributed about 1/2 g with a standard deviation of a factor of about 2. You argue that this variability can be decomposed into separate site and source variabilities; this seems to be convincing and I fully support this approach. However, I am concerned about using a log
-normal distribution to catch the tails of the distribution. In particular, I am concerned that we have now seen several
3 Southwestern United States Ground Motion Characterization SSHAC Level 3 Workshop #2 Proceedings, Appendix B, pp. B B-2.
5 examples of near
-source peak accelerations whose time histories are asymmetric about their zero line (see Yamada, M., J. Mori, and T. Heaton, 2008, The slapdown phase in high acceleration records of large earthquakes, Seismological Research Letters; 80: 559 - 564). It has been hypothesized that this may be an example of 'slap down," a phenomenon that was well studied by the nuclear explosion test community. Slap dow n is clearly a nonlinear phenomenon and I would expect its statistics to be independent of the log-normal distribution that are [sic] used to characterize most of the data. It's very difficult to put an upper limit on slap down accelerations. There are many examples of objects that have been launched through the air in violent shaking from past earthquakes. Slap
-down seems to be a plausible phenomenon in the near source of earthquakes.
4 (emphases added) Q09: To what extent do Professor Heaton's concerns reflect a preference for using peak ground displacement ("PGD") rather than peak ground acceleration
("PGA") as a measure of intensity for long
-period ground motions associated with large earthquakes?
A09: The 2009 research paper referenced in Professor Heaton's po st-Workshop #2 comments observed, Generally speaking, the energy in ground motions from smaller, more frequen t earthquakes is mostly from the short
-period content, whereas the energy in ground motions from larger, less frequent earthquakes is primarily in the long
-period content.
5 The paper pointed out a fundamental divergence between the two methodologies:
As mentioned previously, PGA is known to saturate at magnitudes greater than 6, and the logarithm of PGD is known to increase linearly with respect to magnitude. Thus, there is essentially no correlation between these two intensity measures for near
-source ground motions from large events.
6 ***
4 Id. 5 Masumi Yamada, Anna H. Olsen, Thomas H. Heaton, "Statistical Fetures of Short
-Period and Long
-Period Near
-Source Ground Motions,"
Bulletin of the Seismological Society of America, Vol. 99, No. 6, December 2009, pp. 3264 - 3274, 3264.
6 Id., p. 3267.
6 Knowing the magnitude does not help predict the PGA because PGA saturates with magnitude, but knowing the magnitude helps to predict PGD because the logarithm of PGD is proportional to the magnitude.
7 Q10: Why does A4NR think it essential that PG&E discuss the deterministic joint rupture analyses with the IPRP before deciding to exclude them from the CCCSIP report?
A10: Because relegating consideration of joint ruptures to a probabilistic review greatly obscures catastrophic potential and tends to preclude the evaluation of mitigation options. I n PG&E's memorable phrase from the ongoing controversy about DCNPP's licensed seismic design basis, the "probability is so small that it would mask in PRA space any probability of an issue occurring."
8 Q11: Does that conclude your testimony?
A11: Yes, it does.
7 Id., p. 3271. 8 GRC2014-Ph-I_DR_A4NR_001
-Q02Supp01Atch20 unnumbered p. 2 is included in Appendix as Exhibit 3. PRA is an acronym for probabilistic risk analysis.
Appendix Exhibit 1 ERRA-2014-PGE Compliance_DR_A4NR_004-Q01 Exhibit 2 ERRA-2014-PGE-Compliance_DR_A4NR_005-Q01 Exhibit 3 GRC2014-Ph-I_DR_A4NR_001-Q02Supp01Atch20
Exhibit 1 ERRA-2014-PGE Compliance_DR_A4NR_004-Q01
Deterministic 84 th percentile spectra for the power block foundation level assuming a magnitude 8 earthquake on each source.
Deterministic 84 th percentile spectra for the turbine building foundation level assuming a magnitude 8 earthquake on each source.
Deterministic 84 th percentile spectra for the turbine building foundation level using the magnitude for a 1E-6/yr rate of occurrence.
Deterministic 84 th percentile spectra for the turbine building foundation level using the magnitude for a 1E-7/yr rate of occurrence.
Exhibit 2 ERRA-2014-PGE-Compliance_DR_A4NR_005-Q01
Exhibit 3 GRC2014-Ph-I_DR_A4NR_001-Q02Supp01Atch20