ML20009E083
| ML20009E083 | |
| Person / Time | |
|---|---|
| Site: | Marble Hill |
| Issue date: | 07/22/1981 |
| From: | James Keppler NRC OFFICE OF INSPECTION & ENFORCEMENT (IE REGION III) |
| To: | Cassaro M LOUISVILLE, UNIV. OF, LOUISVILLE, KY |
| References | |
| NUDOCS 8107270059 | |
| Download: ML20009E083 (7) | |
Text
r
' iC}t lbb5 \\
3069l o
RJUL 4 h d cd d,h S
\\
E Q gCSN< c Dr. Michael A. Cassaro ob Professor of Civil Engineering g k 3,
University of Louisville
.A p
s Louisville, KY 40208 b
M Q -gW I
Dear Dr. Cassaro:
f This is in reply to your letter of March 26, 1981 regarding concrete at the Marble Hill plant site. This letter is supplemental to our letter to you dated March 20, 1981, wherein we expressed our engi-neering judgements and technical basis for considering that Report No. SL-3753, Rev. 1, meets NRC requirements regarding the statistical determination of the homogeneity of in place concrete.
The NRC reviewed your concerns outlined in both your September 26, 1980, and March 4, 1981 letters, and provided a response to you dated March 20, 1981. We believe that this response fully described an acceptable basis for our decision to conclude that all NRC re-quirements relative to the reported statistical testing methodology and other emperical evaluations of concrete quality have been met.
Based on all of the investigations and evaluative actions taken by the licensee and its agents, and the results of the independent investigation you participated in, NRC has concluded that the quality of the in-place concrete at Marble Hill has been adequately assessed by the method described in the Sargent & Lundy report. To this extent, the NRC requirements have been met.
Our additional comments and elaborations in response to your letter of March 26, 1981, are outlined in attachment "A" to this letter.
Attachment "A" readdresses your major concerns as we understand them.
As you recall, subsequent o our receipt of your March 26, 1981 letter, we tried to arrarne a meeting to discuss these issues with you, your associate,
. Alexander, and the NRC staff here in the Glen Ellyn office of the NRC. However, your client, t
\\
8107270059 6
]
Dr. Michael A. Cassaro jut. 2 2 inal "Save the Valley", declined to have you participate. Even so, we would be pleased to arrange a meeting between you, Dr. Alexander and our technical staff to discuss any items of further concern on your part regarding this matter.
Sincerely, James G. Kepp er Director
Enclosure:
Attachment A cc w/ enc 1:
W. M. Petro, Executive Director Nuclear Project Management C. Kammerer, CA J. H. Sniczek, IE DMB/ Document Control Desk (RIDS)
Resident Inspector, RIII LeBoeuf, Lamb, Leiby & MacRae Dave Martin, Office of Attorney General John R. Galloway, Staff Director, Environment, Energy and Natural Resources Subcommittee E. P. Martin, Wabash Valley Power Association Save the Valley Mr. Dattilo, Esquire
/W 16t MQ U*
MfA lbJasn.[3kekwd urie. (%%3dkh)
RIII R.T RIII RIII RII RIII
- R1II, 7-ao P/ lC db
/
8 C%
~
Davip Kebpler Schweibinz/db Little. Williams pessard Norelius wIl'T 1lp)lU h c gI 1 P*
7/20/81 91 i
I'
'Ns ATTACHMENT A
SUMMARY
1.
Comment "Since the NRC has established the criteria, 95%
Pg. 1, Line 10 re'Jability with 95% confidence using equation 2, it must be recognized that this is a probabalistic equation implying no instrument or human error associated with the statistical approach."
Response
NRC was and is ful'_y cogni7 ant of the potential for human and instrument error.
In addition to our March 20, 1981 response (Item 3 of Attachment A), it was for this reason that (1) each sample was tested at numerous locations, all of which overlapped. A total of 1,444 separate microseismic tests were performed for the 60 samples reported. Moreover, each of the 1,444 separate microseismic test loca-tions were usually tested at least twice (approximately 2,888 separate tests) before they were accepted for record. No credit was taken statistically for multiple testing, as such credit does not fit the assumptions of the model. However, multiple testing in conjunction with other procedural requirements does add considerable additional engineering con-fidence in the accuracy of the testing. This was an extraordinarily cateful test.
(2) Each test that showed a reflector was independently analyzed to determine if it was caused by a planned asbuilt condition.
(3) If this analysis was not affirmative each area was cored or line drilled to destructively evaluate the subject test area.
(4) At least 4 areas reported to be homogeneous were also cored.
(5) At least three independent organizations partici-pated in the evaluations.
(6) The procedure was qualified and verified by destructive examination by the licensee with NRC participation prior to its impiementation.
Subsequent to issuance of the report (SL-3753), more than 22 additional cores and/or line drilling (destruc-tive tests) were performed (at the requast of the independent consultants) withot t identifying any errors in the test results prevl.ously established by the microseismic testing.
In other words, the documented procedure for the microseismic testing, approved and implemented at Marble Hill, provided more than adequate safeguards against human or instrument error in making deter-minations of the quality of the concrete. Not in
k one instance has extensive destructive test shown the microseismic testing and evaluation to be in-adequate or in error. For the 60 samples selected, over 1400 individual tests were taken for record and approximately 2800 such tests were done as a matter of procedure.
2.
Comment "For example, if the interpreter declared a " honeycomb" the same day he based the qualification test and the honeycomb _ turned out to have significant bubbles in the concrete, then_there appears to be reasonable room for error."
Response
We believe that this comment has reference to the mistaken belief that early on'during the testing,
" errors" of interpretation of microseismic test results were made and later changed without adequate basis. This was not the case at all. There were locations where Mr. Muenow changed deaignations of
" Honeycomb" to entrapped air. However, the change in terminology was not a change in interpretation of the test records, but only a change in designation.
After the first few tests, and in order to better quantify the test results, the original use of the term " Honeycomb" in the field data sheets was refined to address " entrapped air" and " honeycomb" separately.
The definition of these terms was presented in Report SL-3753, Rev. 1.
The NRC independent consultants requested further-justification' for this change of terms. They identified 28 test locations which were examined as follows:
Seventeen locations were line drilled and viewed with an optical fiberscope. Four concrete cores were drilled and examined, five additional areas underwent visual surface examination,'and two existing core holes were viewed with the fiber-scope.
The results of the fiberscopa examination supported Mr. Muenow's conclusion that no voids or honeycomb were present.
To the extent'of the above, there was no " reasonable room for error" identified, as the testing procedure and evaluation methodology precluded it.
3.
Comment "It appears that no written qualification exists in the record at Marble Hill. A qualification record must be performed to evaluate if we have 95% reliability with 95% confidence.". -
Response
A qualification test and record for the microseismic testing technique does exist at Marble Hill and it demonstrates the adequacy of the microseismic test method equipment and interpretations. This activity was documented in NRC Report No. 50-546/79-07; 50-547/79-07 dated September 18, 1979. Furthermore, in response to the NRC consultants request, on February 9,1981, a transducer performance test was made on a one foot cube of concrete. An internal discontinuity was indicated. The block was sawed and the discontinuity as indicated by the instrument and its interpretation, was found.
Dr. Cassaro and the NRC independent consultants witnessed this test.
The important question is: Has the test procedure failed to identify unacceptr.ble internal voids, honeycomb or significant cracks? The answer is no, as proved by the qualification test, the many cores and line drilled holes described above and destructive tests of the concrete patches witnessed by NRC.
Since adequately conservative procedural safeguards to preclude error during microseismic examination and evaluation were specified and implemented, it is inappropriate and unnecessary (in NRC judgement) to hypothetically introduce an additional factor regarding the potential for systematic error. The correctness of this judgement has been fully supported by the results of all destructive verification examinations to date.
4.
Comment "As it now stands there is certainly less than 95%
reliability as outlined in the March 4 letter, and we believe that no verbiage will erase that reality only, clear statistics."
Response
In general it is our conclusion that more than 95%
reliability with 95% confidence has been achieved.
It should be noted that neither the NRC nor the licensee and its agents relied on pure statistics
(" clear statistics") to comprehensively evaluate the concrete. A substantial amount of engineering judgement was used to conservatively bias the sample selection. No advantage was taken for this bias in establishing the statistical methodology. From the purely statistical perspective, this bias may be considered as an error. However, its effect is te emphasize examination of those areas of the concrete, which experience shows to have a higher probability for defects...
'N In specific response to Dr. Alexander's comments in your March 4,1981 letter, our judgements and opinions are as follows:
a.
The O.C. curve is used primarily for in process control of production to assess risks associated with various inspection plans. This curve can be constructed as Dr. Alexander has done from Equation (2) in the SL-3753 report.
This curve was not included in the report because it is not relevant to the discussion. Our goal was not to study variability of confidence. We intended to use a procedure by which 5% or more defective units in the lot would be detected by the tests with 95% confidence.
b.
The determination of quality of in-place concrete once the entire population is known can be achieved using Equation (1) As mentioned above, it is intended to test the hypothesis about quality of concrete using Equations (1) and (2).
The acceptance or rejection of this hypothesis at a given confidence level depends on the population size, sample size, and the number of defectives in the sample provided that sampling is random. Obviously, biasing the sample to favor congested areas in this situation is an engineering judgement to provide more conservatism for which no credit was taken.
Prior to performing tests, more than 60 test areas (units) had been identified.
Since the first 60 tests revealed no unacceptable unit, the hypothesis in question was tested and no reject was identified at a 95%
level of confidence.
The use of Equations (1) and (2) to deduce from observations an upper bound on the percent defective, p, is correct and appropriate. The Marble Hill case is an infinite population of which 60 random or conservatively biased areas were tested and no defectives were found. The conclusion is that we are 95% confident that the percentage of defectives in the population does not exceed 5%. This is statistically correct.
c.
Sargent & Lundy's definition of reliability is widely accepted in civil engineering practice; see reference cited p. 5 of Vol. I, SL-3753, Revision 1.
There is no ambiguity in consid-ering the reliability as pertaining to the in place concrete. ______-_________- _ ___ - __-_____ - _____ -
'\\
F d.
In regard to the effects of errors on the confidence level, operational and observational reasons have been presented to show that from an engineering standpoint, the pulse-echo technique and the data interpretation used at Marble Hill is extremely reliable at indicating unacceptable internal voids, honeycomb or significant cracks.
Thisimpliesthattheprobability,e$g,ligibly in Equation (b) of Dr. Alexander's addendum is n small or-very close to zero. Referring to Dr. Alexander's table, a 3% error (which is conservative for this testing procedure) results in a confidence level of 95%. The conservatism built into the program more than makes up for this " hypothetical" error. For example, an additional five areas where sleeves were going
-to be cored were tested by pulse-echo and cores taken and reported in SL-3753. The confidence level as: ziated with testieg 65 areas with no defects found in the sample is 96.4%.
This supports the conclusion of the SL-3753 report that Marble Hill concrete is acceptable with 95% reliability and 95% confidence level.
'1
} -.
-