ML20155A685

Joint Affidavit of D Lurie & E Marinos Re Board Concerns on Statistical Inferences from Comanche Peak Review Team Sampling.Certificate of Svc Encl

ML20155A685
Person / Time
Site:	Comanche Peak
Issue date:	04/04/1986
From:	Lurie D, Marinos E Office of Nuclear Reactor Regulation, NRC OFFICE OF RESOURCE MANAGEMENT (ORM)
To:
Shared Package
ML20155A683	List: ML20155A679 ML20155A685
References
OL, NUDOCS 8604090221
Download: ML20155A685 (27)

Text

_ _ _ - . _ _ _ __

~

UNITED STATES OF AMERICA NUCLEAR REGULATORY COMMISSION BEFORE THE ATOMIC SAFETY AND LICENSING APPEAL BOARD In the Matter of )

)

TEXAS UTILITIES ELECTRIC ) Docket Nos. 50-445 COMPANY, et _al.

) 50-446

)

(Comanche Peak Steam Electric )

Station, Units 1 and 2) )

JOINT AFFIDAVIT OF DAN LURIE AND EVANGELOS MARINOS ADDRESSING THE BOARD'S CONCERNS ON STATISTICAL INFERENCES FROM CPRT SAMPLING We , Dan Lurie and Evangelos Marinos, being duly snorn, do depose and state as follows:

Q1. Dr. Lurie, by whom are you employed and what is the nature of I

your employment?

A1. My name is Dan Lurie. I am presently employed as a mathematical statistician in the Management Support Branch, Office of Resource Management, U.S. Nuclear Regulatory Commission. I am responsible for providing statistical expertise and advice as required by NRC Staff members.

l Q2. IIave you prepared a statement of your professional qualifications?

A2. Yes, a statement of my professional qualifications is attached to this joint affidavit.

• 8604090221 060404 5 DR ADOCK 05 i

I y...--- , . . . . . _ . . . , .w. _- ,

l Q3. Mr. Marinos, by whom are you employed and what is the nature of

! your employment?

A3. My name is Evangelos Marinos. I am a Senior Nuclear Engineer in i

the Division of Boiling Water Reactors, Office of Nuclea: Reactor Regulation, U.S. Nuclear Regulatory Commission. As a nuclear engineer, I am responsible for evaluation of the design and performance of reactor systems and components, from the standpoint of functional capability and integrity.

Q4. Ilave you prepared a statement of your professional qualifications?

A4. Yes, a statement of my professional qualifications is attached to this joint affidavit.

Q5. Gentlemen, what is the purpose of your joint affidavit?

AS. (Lurie and Marinos) Our joint affidavit addresses the technical concerns, as opposed to the legal concerns, II raised by the Licensing Boards 1# in their November 11, 1985 Memorandum (Statistical Inferences from CPRT Sampling) (" November 11, 1985 Memorandum").

-1/ In its Memorandum, the Board also raised questions regarding the

" level of safety" that must be assured by the CPRT sampling program, and the potential need for an exemption from 10 CFR Pas t 50, Appendix B. The Staff has addressed these legal concerns in a separate filing dated January 30, 1986.

~2/

On December 24, 1985, the two dockets in this proceeding were unified into one docket before the initial Licensing Board.

Therefore, all references in our joint affidavit will refer to the "B oard . "

Q6. Describe the statistical methodologies which may be used by the CPRT.

A6. (Lurie and Marinos) The statistical methodologies which may be used by the CPRT are given in Appendix D of the CPRT Program Plan. E Appendix D describes two different statistical method-ologies , each of which is applied to a different type of variable.

The first type of variable is a discrete variable, where the response i

is binomial in nature (i.e. , acceptable or not acceptable). The statistical methodology that may be employed by the CPRT for testing discrete variables is a " nonparametric" methodology and is described in Attachment 1 of Appendix D. Sampling for binomial attributes is performed on populations with the intention of providing a 95/5 statement of assurance, b The sampling scheme selects a

, 3/ It must be recognized that the CPRT utilizes, and Appendix D discusses two types of sampling: " biased sampling" and " random l sampling." Since biased sampling is not " statistical" in nature, it is i

not addressed here.

-4/ A 95/5 statement of assurance is a 95% level of assurance that no 4

more than 5% of the members of a population or stratum are

deficient. The Staff notes that because Appendix D, Attachment 4 j of the CPRT Program Plan provides for potential expansion of sampling, the assurance level for the Appendix D statistical sampling

methodology is necessarily somewhat below 95/5, contrary to

Applicants' discussion in Appendix D, Attachment 4, p. 12. This will be discussed further in the SSER evaluating the CPRT Program i Plan. However, other considerations and activities can potentially

, increase the assurance level above the 95/5 assurance level for some i applications of the Appendix D methodology in the various ISAPs.

This matter will also be discussed further in the SSER evaluating the 1 (FOOTNOTE CONTINUED ON NEXT PAGE) i j

l minimum of 60 items - at random from the population following the procedure described in Attachment 3 of Appendix D. Each item in the sample is then inspected for the attributes of interest.

If no deficiencies - are found in the items sampled, the entire population is accepted.

(FOOTNOTE CONTINUED FROM PREVIOUS PAGE)

CPRT Program Plan. As a separate matter, the Staff also points out that the CPRT may use a statement of assurance other than 05/5, according to Section 2.2 of Appendix D. Ilowever, that Section states that any exceptions to the use of a 95/5 assurance level will be reflected in the ISAP where the exception is used.

5_/ According to Appendix D, Section 2.1 and Attachment 4, the minimum faitial sample size for a 95/5 test is 60. Table 1 of Attach-ment 4 to Appendix D fndicates that larger initial samples (95, 126, 155, 183 and 210) may be selected. However, a sample of 45 may be drawn from populations with 100 or fewer items, according to the third note of Table 1, Attechment I to Appendix D.

6_/ The term, " deficiency", is defined by the CPRT in Appendix E in the context of both decfgn adequacy and construction adequacy.

A construction deficiency is defined in Appendix E, p.13, B.2(b) as:

"any identified construction deviation that has been determined to be safety-significant."

A construction deviation is defined as:

"any identified error related -to construction or installation of safety-related hardware that has been determined to constitute a verified failure to construct or install a safety-related structure, system or component in accordance with safety-significant attributes and criteria contained in design drawings and specifications or l installation procedures / requirements."

These definitions are essentially the same in the design area."

i i

. If one or more deficiencies are detected, each deficiency will first be analyzed by the CPRT to determine if a root cause can be identified.

Appendix D, Attachment 1, Paragraph 3: Attachment 4, Paragraph 2.

If for any attribute exactly one deficiency is identified in the initial sample and a root cause is not identified, sample expansion and a review of all attributes will continue in 2 hat population

) "until it is determined that either the deficiency is a random occurrence of very low frequency, or a trend or programmatic deficiency is identified ... (i.e., a potentially deficient stra t um) . " Appendix D, Attachment 4, Paragraphs 2, 4-5.

According to Appendix D, Attachment 4, Paragraphs 5 and 6, the sample will be expanded to include 35 additional items , starting where the initial sample ended. If one or more deficiencies continue to be detected in the expanded sample and cannot be associated with a specific stratum, 100 percent of the population is inspected. U Appendix D, Attachment 4, Paragraphs 2, 5-6. Appendix D does not specifically state what will occur if the deficiency (fes) in the expanded sample can be associated with a specified stratum; however, it appears that if these deficiency (fes) can be associated i

l l

-7/ The Staff's review of Appendix D did not identify a specific statement to this effect. However, the Staff's understanding is supported by the last sentence in Paragraph 2 of Attachment 4, and was confirmed by Applicants in a March 18, 1986 meeting.

4

with a new stratum, sampling of that new stratum can be instituted in the manner discussed below, while the original population will be augmented and accepted if no additional deficiencies are identified.

l On the other hand, if for any attribute one deficiency is

! identified in the initial sample and a root cause is identified, then the initial sample will be expanded along two parallel paths.

1 j First, a stratum containing those items with the suspect f

attribute be defined and that attributo (or a reduced set of

attributes in the case of ISAP VII,c) in that stratum will be reviewed. 8_/ According to Attachment 4 of Appendix D, items from the initial sample falling into the newly-defined strttum are

! removed from the initial sample and placed into a new sample, and j

j the new sample is exp9nded by randomly selecting items in the stratum until a total of 95 items are reached. U If no additional j deficiencies are detected in the new sample of 95, the pcpulation i

is accepted. On the other hand, if one or more deficiencies are detected in the new sample , and no different root cause is 1

-8/ Appendix D, Attachment 4, 2nd paragraph also discusses stratifica-tion for "certain characteristics."

-9/ Sample expansion into the newly-defined stratum can be done in two ways, as described in Attachment 4, Appendix D: (1) The stratum can i

be identified, itema in the stratum separated from the general population, the items numbered, and a random sample selected from the stratum, or (2) alternatively, the stratum is identified but left in the general population and sampling continues in the general j population until the number of items that belong to the. stratum

{ reaches the required stratum sample size.

l i

I l

1

~

identified for these deficiencies then 100 percent of the stratified population is reinspected. However, if a different root cause associated with the deficiency (les) is identified, second stratum may be established and sempling may continue in the second 40ratum, Second, in the remaining population without the suspect strath,

" sample augmentation" bi s used to verify that the deficiency is not asse>ciated with the remaining population. Appendix L, AttachrM 1.

Paragraph 3; Attachment 4, Paragraphs 4-5.

If two deficiencies of the "same type" (nttribute) ara IGentified in the initial sample of the population which cannot he associated with a specific stratum , then 100 percent of the population will be inspected. Appendix D, Attachment 4, Paragraph 2. According to Applicants' representations at the March 18, 1986 meeting, if two deficiencies for the same attribute and the same root cause are idertided in the initial sample, then 100 percent of the population will be inspected for that attribute.

The second type of variable is that which is measured on a contin-uous scale.

Continuous variables tire addressed in Appendix D,

-10/ According to Applicants, the sample without the suspect strata is augmented "with additional items to br!ng the general population sample back to the minimum 95/5 sample size." Appendix D, Attach-,

(FOOTNOTE CONTINUED ON NEXT PAGE)

Attor:hment 2. " Sampling Guidelines for One-Sided Tolerance Limits."

" Tolerance limits" are numerical gz and 22 values constructed so that one has x assurance (expressed in percentage) that at least y percent of the measured values of the population lie between the constructed limits zy and z 2. A "one-sided upper (lower) tolerance limit" is a numerical value z g constructed so that one has x assurance (expressed in percentage) that at least y percent of the measured values of the population are below zy (above y z ). b Appil-cants indicated in the March 18, 1986 meeting that one-sided tolerance limits are not utilized in any ISAPs or DSAPs in Revi-sion 3 of the CPRT Pro < tram Plan, but that they may be used in the future. In their " Supplement to Memorandum in Response to Board's Memorandum" (April 1,19,5) (" Applicants' Supplement"), Applicants indicated that ISAP V.a (skewed welds) is currently utilizing one-sided tolerance limits. Applicants also stated in their Supple-ment, as well as at the March 18, 1986 meeting that the assurance level (denoted above by "x") is always 95 percent. The percentage of the population measures values (denoted by "y") which is assured to be bounded by the construction limits (z l # *2), however, is not (FOOTNOTE CONTINUED FROM PREVIOUS PAGE) ment 4, Paragraph 4. In the March 17, 1986 meeting, Applicants confirmed that this means that the sample will be augmented to bring the total items up to 60.

11/ An example of a criterion specifying a tolerance limit is in the ACI

~-

Concrete Code, which snecifies that at least 90% of the 28-day concrete cylinder strength samples fall above the required design strength.

l

(

._. - _ . _ . . . . . _ _ . _ _ ~. __ . . .

l l

i  !

predetermined, and wiIl be decided on a case by case basis. Appli-cants' Summary, pp. 3-4.

Q7. What is the Staff's understanding of the Board's concerns with regard to statistical sampling which were raised in the November 11, 1985 Memorandum?

A7. (Lurie and Marinos) The Board's concerns with statistical sampling are set forth in its November 11, 1985 Memorandum. Apparently, the Board has not seen "a clear statement [in the CPRT Program Plan] of how the applicants have designed their studies" (pp.1-2 of the November 11, 1985 Memorandum). More specifically, the Board suggested that any statistical program which the Board is asked to draw a statistical inference should include five elements described by Dixon and Massey. EI The five elements are:

1. Statement of the hypothesis and assumptions.

2. Statement of the level of significance chosen.

3. The test statistic and critical region.

4. Presentation of any computation. NI

5. A full statement of the conclusions.

-12/ Wilfred J. Dixon and Frank J. Massey, Jr. , Introduction to Statis-tical Analysis (Fourth Edition), McGraw-Hill, New York (1983), p.85.

13/ The Board indicated that in the alternative, there could be a reference to  ? verified code".

i e

The Board apparently is also concerned about how items to be evaluated by the CPRT are grouped to form " populations" from which samples are chosen (pp.2-4 of the November 11, 1985 Memorandum).

Q8. Does the Staff agree with the Board's understanding that all statistically-based sampling programs should address each of the five i elements described by Dixon and Massey?

A8. (Lurie) Statistical sampling is a procedure by which an inference about the population can be made by examining only a fraction of the population. Statistical inference may take two forms: b estimation of the magnitude of the population characteristics, and testing of hypothesis regarding population characteristics. Both forms are useful for making decision about specific characteristics of the population. These two forms are not necessarily mutually exclusive.

Indeed, hypothesis testing requires estimation of some parameters, and some tests of hypotheses can be shown to have a counterpart in interval estimation (confidence interval for a population mean). On the other hand, some estimation techniques, such as tolerance limits, 1

are not isomorphic (do not have a one to one correspondence) with the test of hypothesis.

The Staff agrees that statistical sampling programs which lend themselves to a test of hypothesis should include implicitly, if not

-14/ Experimental Statistics by Mary G. Natrella, National Bdreau of

. Standard If andbook 91, 1966 reprint, p. 1-3.

explicitly, the five elements outlined by Dixon and Massey. The purpose of the five-element protocol is to make sure that the objectives of the sampling are clear and that the execution of the .

program is consistent with these objectives. However, the practice of explicitly articulating the five elements is not common in non-academic environments. ,

4 Q9. Does the CPRT statistical program address each of the five elements delineated by Dixon and Massey?

A9. (Lurie and Marinos) The CPRT's statistical methodology for binomial attributes does not explicitly address each of the five elements delineated by Dixon and Massey. Ilowever, the five elements listed by Dixon and Marsey can be derived from the CPRT's r.tatement of

their statistical sampling program as follows

1. Statement of the Hypothesis and Assumptions A formal statistical test of significance requires statements '

of both a null and an alternative hypothesis. Since the CPRT statistical sampling program has adopted a 95/5 level of assurance, see, eg, Appendix D. Section 2.1 and Attachment 1, the null hypothesis would be that the proportion of defectives in a population is 5 percent, and the alternative hypothesis is that tha.t proportion is less than 6 percent. In layman terms, this suggests that the true proportion of defectives is 5 percent, and this null hypothesis is rejected if there is enough evidence to l

)

l

disprove the null hypothesis in favor of the alternative hypothesis by showing that the sample proportion is unusually low (zero, in the case of the CPRT sampic cf fi0 items), b The C*>RT's assumptions for its statistical methodology for ,

binomial variables are net explicitly stated. However, as noted in my anawer to Question 8, it is not common to ,

articulate the elements of a statistical sampling program for ,

binomial variables in terms of the five elements identified by Dixon and Massey, In general, statistical sampling for binomial variables have the following assumptions:

(1) homogencity of items within a population.

(2) random selection of items within a population.

(3) ability to classify with certainty each item as defective or not defective. ,

(4) essentially infinite number of items in the population (a conservative assumption).

Many of these assumptions appear to have been recognized by the CPRT, For examplo , the assumption of (and 151 The Applicants' discussion of the null and alternative hypotheses on

~

pp. 2 3 of their April 1,1986 Supplement is the same as th,e Staff's discussion above. Applicants' January 31, 1986 Memorandum reverses the role of the null and the alternative hypotheses.

r consaquently the need for) homogeneity is recognized by -

the CPRT fr} Appendix D's (!!4cussion of stratificetion.

The nbed for random sampling manifesta itself in the CPRT procedure for generating random samples which is het ,

forth in Appendix D. Attachment 3 Recognition of the i i

assumption of infinite population is augge.Mted in the last sentence of the first paragraph of Attachdent 1 to Appendix D.

2. Statement of the tevel of_ Significance Chop The level of signlficance ja the probability that the test would determine that the proportion of defectives is legs ,

than 5%, when in reality that proportion is 5% or larger.

The level of algnificance essociated with a 95/5 statement .

of assutance is necessarily equal to alphas.05.

3. The Test Statistic and the Critical Region The test statistic is the actual count of defectiva items found in the sample. The critical region is the set of all ,

counts of defective items which lead to the rejection of the null hypothesis. For a sample of 60, the critical region is composed of the number zero. $ l

'-16/ As discussed in note 5 above, Appendix D sets the minimurr initial -

sample slae at 60. For initial sample sizes larger than 60, the l critical region will be a set of numbers other than zero. The critical i (FOOTNOTE CONTINUED ON NEXT PAGE) i

4. Presentation _of any_ Computation The computation of the statistic, is extremely simple; it is the count, or tally, of the deficient items in the sample.

5. Statement of the Conclusions The statement of conclusions essentially summarirca the results of the sampling and whether the hypothesis has been confirmed or not. The CPRT Program Plan,Section VI, indicates that the results of each ISAP (or DSAP, if applicable) will F ': ussed in Results Rcports. Thus, the Staff expec 'f

., .c statement of concluabns for any statistical sampl% , performed for an ISAP or DSAP will be contained in the Results Report for that particular ISAP or DSAP.

By contrast, the CPRT's statistical methodology for tolerance limits in Attachment 2 of Appendix D does not correspond to the five element protocol suggested by Dixon and Massey, because the tolerance limit, as applied to continuous variables, is strictly an (FOOTNOTE CONTINUED PROff PREVIOUS PAGE) region for larger sample sizcs can be derived from the column labelled " Detection Number" in Table 1, Attachment 1 to Appendix D; the detection number represents the upper limit to the critical region for the applicabic sample size. The Staff notes that in the context of the test of hypothesia, the column labelled " Critical Region" in Tablo 1 could more securately be labelled the " Hypothesis Ac~ceptance Region", or " Population Rejection Region."

estimation technique. As indicated in Answer 8 above, this estimation technique does not have a counterpart in the test of hypothesis and therefore cannot be made to correspond to the five element protocol.

P Q10. What are stratification and stratified sampling, and what are they used for?

A10. (Lurie) Stratification is the partitioning of a population into two or more sub-populations or strata in such a way inat all the members of the population in each strata have similar characteristics. Sampling carried on a stratified population is called stratified sampling.

Stratification may be carried for administrative convenience. More importantly, however, stratification is used whenever one wishes to I make an inference about a population whose members are subject to one or more external factors which may have an effect on the I attribute under scrutiny. When the population is not homogeneous, the inference about the entire population may not be meaningful l because one cannot be sure that each of the external factors is properly represented in the sample. Accordingly, one should  !

l stratify the population and conduct stratified sampling, in order to control the effect of the external factors.  !

i Q11. Does the Staff agree with the Board's understanding that the CPRT sampling should be stratifbd to account for, inter alla, differences in the complexity of work or design processes, differences in the I l

i l

. . I l

l l

qualifications of QC inspectors or craftspersons, changes in supervisory and management personnel, and changes in the auditing of work or design processes?

All. (Lurie and Marinos) Ideally, every factor of the population which has the potential of altering the attribute under scrutiny should be included as a stratification factor. In practice, however, this is not always practical or achievable. For example, identification of all potentially-distinguishing factors and subsequent stratification may well require that each item in a population be in a stratum by itself.

In other circumstances, the factors to be used in stratifying the population are not obvious until some of the items of the population are actually inspected. Thus, from a practical perspective in developing a sampling program, one should initially identify those factors that could reasonably be expected to have such an effect, and then stratify the population accordingly. Once the sample is drawn and inspected, and factors affecting the attribute (if any) are identified, the original population should be restrati' led to account for the newly-identified ' actors affecting the attribute.

The Board has listed four factors that they believe affect the attributes to be inspected by the CPRT:

1. Differences in the complexity of work or design processes.

2. Differences in QC and craftperson qualifications.

3. Changes in superviscry and management personnel.

4. Changes in auditing of work or design processes.

- , n -

These factors may potentially affect the attribute under scrutiny.

As discussed in Answer 12 below, Applicants have elected to initially stratify their samples by work process attributes, including work process complexity, when utilizing statistically-based sampling in the self-initiated review of construction adequacy (ISAP VII.c). The Staff will address the adequacy of the Applicants' stratification process in its evaluation of the Applicants' Results Reports.

Q12. How does the CPRT Program Plan address stratification of samples?

A12. (Marinos and Lurie) The CPRT's review of construction adequacy (QOC) (App. B of the CPRT Program Plan, Revision 3), uttes issue Specific Action Plans (ISAPs) to address and resolve all external issues b on hardware and QA/QC adequacy (Category 1 ISAPs).

The CPRT has also developed ISAP VII.c. (Category 2 ISAP) to control the CPRT's self-initiated hardware reinspection program.

Statistically-based sampling is permitted in both Categories 1 and 2 ISAPs. See Appendix B, Sections II. A.1 and A.2. For those Category 1 ISAPs that employ sampling in accordance with Appen-dix D, any strafification of initial samples is described in those ISAPs. Stratification of subsequent samples will be done in accordance with Appendix D. As discussed earlier in Answer 6, if initial sampling discloses any deficiencies, each deficiency will be analy::ed to determine if a root cause can be identified. If a root

-17/ External issues are defined by the CPRT as those issues which have been identified by sources other than the CPRT and Applicants, eg. , the Staff's TRT, SIT, CAT and SRT, Intervenor CASE, and Cygna Energy Services.

cause is identified, a stratum containing items with the suspect attributes or characteristics will be identified, and a sample will be drawn from that stratum. EI The Category 2 ISAP (the self-initiated review of construction adequacy) requires that any sampling be initially stratified to account for differences in the complexity of construction work processes (activities). See Appendix B, Section II. A.2. A consistent set of technical attributes (eg. , cable tray installation attributes) will be identified which will define a homogeneous work activity (HWA) (eg, cable tray installation). The Staff understands from technical audits and meetings with Applicants that this process of defining HWAs will take into account work process complexity. A random sample consisting of 60 items or more will be selected from each HWA for inspection. Any items from this sample which the CPRT identifies as having more importance to safety (where possible, items selected from safe shutdown systems) will also be placed into a second sample. The CPRT will then randomly select other items from the HWA to bring the second sample up to 60 items, for populations with 101 items or more (45 items for populations of 100 items or less). Further expansion of sampling into newly-defined strata is dependant upon identification of either " safety-significant hardware deficiencies" or "potentially adverse trends of

-'-18/ The sample from the newly-defined stratum will consist of items from the initial sample falling into the newly-defined stratum, as well as additional items selected at random from that stratum.

l

non-safety significant deviations" which . are detected in the two initial samples. Appendix B, Section 2. A.2; Appendix C. ISAP VII.c. , Sections 4.3.2.1, 4.3.2.3.

The CPRT's review of design adequacy (DAP) (Appendix A to Revision 3 of the CPRT Program Plan) will be implemented by Discipline Specific Actions Plans (DSAPs) in three categories:

Category 1 DSAPs will address external source design issues, Category 2 DSAP implements the CPRT's self-initiated evaluation of design adequacy, and Category 3 DSAPs will address the special cases of piping and pipe support and cable tray support design adequacy. The DAP permits the use of statistically-based sampling.

Appendix A,Section II. A.3. However, none of the DSAPs currently call for the use of statistically-based sampling as an evaluation methodology, b Cf. CPRT Program Plan, Appendix D, Section 2.3.

Should the CPRT subsequently decide to use statistically-based sampling in any DSAP, the sampling and stratification is to be conducted in accordance with the provisions of Appendix D. Id.

However, the DAP does intend to use sampling, albeit not statistically-based. The DAP will utilize engineering-biased sampling to assess the quality of the CPSES design. According to Appen-dix A, Attachment 4, Section 3.2, and Appendix D, Section 2.3,

-19/ The DAP also provides for the use of ISAPs to address external source issues on design matters. These ISAPs could potentially employ statistical sampling. However, there are no ISAPs for design issues in Revision 3 of the CPRT Program Plan.

I consistent sets of design activities (Homogeneous Design Activities, l

or HDAs) to account for differences in design criteria, design methodology, design control process, and design organization /

discipline will be identified. E According to Appendix D, the DAP will draw a " representative selection of items within each HDA" for review. E The selection of the specific designs for review and the number of designs within each HDA to be reviewed will be a matter of engineering judgment (i.e. , engineering-biased sampling).

Appendix D, Section 2.3.

The preceding statements are true and correct to the best of our knowledge and belief.

3A Dan Lurie Evangos Marinos Subscribed and sworn to before me i this f4 day of April,1986 kJNotary Publfdl My commission expires: /,// /g6 2_0 0 / The process and criteria for development of HDAs is discussed in greater detail in Attachment 4 to Appendix A. The actual HDAs are listed in Attachment 3 of Appendix A, and included in DSAPs VIII, X and XI.

-21/ The CPRT expects that there will be a large number of HDAs with relatively few design items within each individual HDA. .Appen-dix D, Section 2.3.

i i

1 STATEMENT OF PROFESSIONAL QUALIFICATIONS DAN LURIE My name is Dan Lurie. I am employed by the U.S. Nuclear Regulatory Commission (NRC) as a mathematical statistician in the Management Support Branch, Office of Resource Management. I joined the NRC in 1977. My responsibilities include: providing statistical assistance in experimental design, data collection, graphical representation, data analysis, interpretation of results and documentation of findings; reviewing technical reports and manuscripts for statistical validity; serving as statistical consultant on various working groups and committees; supervising data analysis and coordinating programming effort in data analysis; teaching in-house courses in statistical methodology. I have been consulted on sampling plans for inspection plans during the construction of nuclear reactors at Callaway, Marble Hill, Clinton, and Midland. I have also actively participated in development of a standard for containment leakage rate tests and in development and review of statistical methods applicable nuclear material accounting.

In 1971-1975 I was an Assistant Professor and in 1976-1977 I was an Associate Professor in Biometry at the Medical University of South Carolina, Charleston, South Carolina. As an Assistant / Associate Professor, I was responsible for teaching of courses in theoretical statistics, statistical methods, sampling, and nonparametrics statistics to graduate students; teaching of courses in biostatistics to medical students and to Doctor of Pharmacy candidates; rendering statistical services to various departments of the University; collaborating with faculty and students in quantitative research; serving on student advisory committees, research committees, and administrative committees; serving as graduate student advisor; directing student recruitment; and writing and reviewing research proposals.

Between 1964 and 1967 I was a mathematical statistician at the School of Aerospace Medicine at Brooks AFB, Texas. My responsibilities included design of experiments; analysis of data using parametric and nonparametric techniques; supervision and coordination of data collection and data analysis by data clerks; interpretation and documentation of statistical findings; review manuscripts prior to publications; computer programming for data editing and statistical analyses; teaching in-house courses in statistical applications. Additionally, between 1966 and 1976 I taught eleven courses in mathematics and statistics in San Antonio College, Texas A&M University and the College of Charleston, Charleston, South Carolina.

I received my Ph.D. in statistics from Texas A&M University in 1971. The entire course of study was sponsored under an NIH fellowship. Prior to that I j received my MS in mathematica) and experimental statistics in 1964 from Southern Methodist University, Dallas. Here, again , the entire course of study was sponsored by an NIH fellowship. I received my BS in mathematics from Southern Methodist University in 1961. In 1958 I received an AA in mathematics from Los Angeles City College.

. .__ ._ - . . ~ , . - -

A summary of my honors, awards and publications is set forth below.

Professional Activities Membership in Professional Activities:

Kappa Mu Epsilon (mathematics honorary)

Sigma Xi (science research honorary)

American Statistical Association The Biometrics Society American Society for Testing Material - Committee on statistics Appointments and Membership:

American Statistical Association Council, 1976-1977 Biometrics Society, Committee on Training of Biostatisticians, 1976-1977 Graduate Council, Medical University of South Carolina, 1975-1977 Planning and Evaluation Committee, MUSC, 1975-1976 Curriculum Committee, College of Pharmacy, MUSC, 1973-1977 Committee on Cancer Chemotherapy, MUSC, 1974-1975 Graduate Student Advisor, Department of Biometry, MUSC, 1972-1977 Officer, South Carolina Chapter, American Statistical Association:

Vice President and Program chair, 1974-1975 President-Elect, 1975-1976 President, 1976-1977 Awards:

Recipient of NIH fellowship awards for over five years Sponsored Participant, National Science Foundation Summer Conferences Multivariate (U. of Alabama, 1973); Nonparametric Decision Making (Ohio State U. ,1974); Exploratory Data Analysis, (U. of Southern Massachusetts, 1977)

Sponsored Conference Director, National Science Foundation, Sampling (Medical U. of South Carolina,1975)

Editorial Services:

Associate Editor, Statistics, J. of Irreproducible Results Referee, The American Statistician Referee, J. of American Statistical Association Referee, Communications in Statistics

l Participation in Conferences, Meetings, and Symposia

" Simulation of Order Statistics" - Guest Epeaker at ASA Chapter, Columbia, SC,1972.

"A Goodness of Fit Test for Censored Da ta" - Joint ASA/Biometric meeting, Montreal, Canada,1972.

" Systematic Simulators of Joint Order Uniform Variates" - Computer Science and Statistics Interface, Iowa State University,1973.

" Anatomy of Analysis of Variance" - Joint ASA/Biometric Meeting Oregon State University,1975.

Directed:

National Science Foundation Research Conference on "Recent Developments in the Theory of Sampling and Its Applications", Medical University of South Carolina,1975.

Publications:

D. Lurie, " Statistical Analysis of the Effect of Radiation on Performance of a Learned Task," Technical Report SAM-TR-66-106, School of Aerospace Medicine, 12-13, 1966.

D. Lurie and H.O. Hartley, "A Goodness of Fit Test Based on the Spacing of Selected Order Statistics," THEMIS Report #32, Texas A&M University, 1971.

D. Lurie and H.O. Hartley, " Machine Generation of Order Statistics for Monte Carlo Computations," The American Statistician, 26-27, February, 1972.

D. Lurie and R.L. Mason, " Empirical Investigation of Several Techniques for Computer Generation of Order Statistics ," Communicationsin-Statistics, 2 (4) 363-371, 1975.

D. Lurie, H.O. Hartley, and M. R. Stroud , "A Goodness of Fit for Censored Data," Communications in Statistics, 3 (8) 745-753,1974.

R. L. Mason and D. Lurie, " Systematic Simulator of Joint Order Uniform Variates," Proceedings, Computer and Statistics: Seventh Annual Symposium on the Interface, 156-162, 1 % .

G.L. Awkerman, P.E. Teller, and D. Lurie, " Priorities in Ocean Science Study," Science Education, Wiley, 3 (4), 1976.

t S.D. Fritz, C.T. Fitts, and D. Lurie, "The Effect of Hypertonic Glucose Upon Survival in Hemorrhagic Shock Utilizing a Re-stress Model in i

Sheep," J. o_ff Trauma, 16 (4), 1976.

S.J. Levi, R.J. Grant, M.C. Westphal, and D. Lurie, " Development of I Decision Criteria Guide -

Optimal Discriminators as Determined by ,

Statistical Anelysis for Meningitis," Methods of Information in Medicine, .

15 (2) 89-70, 1976.

A.P. Stephans, V. Ward, and D. Lurie, " Relationship Between the Metacarpal Index and the Rate of n'andibular Ridge Resorption," J. of Oral Rehabilitation,1978.

D. Marcott, W. Dubin, and D. 1 urie , " Gender Attitude Toward Hospitalization During Short Confinement," J. of South Carolina Medical Association, 73 (8) 361-363, 1977.

A.J. Gross and D. Lurie, " Monte Carlo Comparisons of Parameter Estimators of the 2-Parameter Weibull Distribution, IEEE Transactions in-Reliability, R-26 (5) 356-358,1977.

P. Halushka, D. Lurie, and J. A. Collwell, " Increased Synthesis of Prosta-glandulin E-like Material by Platelets from Patients with Diabetes Mellitus," New England J. of Medicine, 297, 1306-1310, 1977.

, - - ., - _ , , , e - - , -p--

STATEMENT OF PROFESSIONAL QUALIFICATIONS EVANGELOS D. MARINOS I am presently employed by the U.S. Nuclear Regulatory Commission (NRC) as a Senior Nuclear Engineer in the Division of Boiling Water Reactors, Office of Nuclear Reactor Regulation. I am responsible for the evaluation of the design and performance of reactor systems and components , from the standpoint of functional capability, integrity and operation during normal, transient and accident plant conditiens.

Prior to joining the NRC in December of 1972, I was employed by the Detroit Edison Company as Senior Technical Specialist (from 1965 to 1972).

I was responsible for the design of reactor safety systems and balance of plant systems.

I graduated from Wayne State University with an MSEE and equivalent in Nuclear Engineering. I received a BSEE from Purdue University in 1965.

W l

-- . ,c.-,.-, -- . . , , + - - - - ~

l e

UNITED STATES OF AMERICA goggrige NUCLEAR REGULATORY COMMISSION U5NRC BEFORE THE ATOMIC SAFETY AN1) LICENSING BOARD

'86 APR -? P'> :55 In the Matter of ) 0FflCE Of . . t s,

) 00CMEllNG .. uNr..

TEXAS UTILITIES ELECTRIC ) Docket Nos. 50-445 BRANCH COMPANY, et _al.

)30-446 (Comanche Peak Steam Electric )

Station, Units 1 and 2) )

CERTIFICATE OF SERVICE I hereby certify that copies of "NRC STAFF'S FURT11ER COMMENTS ON TIIE STATISTICAL INFERENCE MEMORANDUM" in the above-captioned proceeding have been served on the foUowing by deposit in the United States mail, first class, or, as indicated by an asterisk, through deposit in the Nuclear Regulatory Commission's internal mall system, this 4th day of April,1986:

Peter B. Bloch, Esq. , Chairman

• Mrs. Juanita Ellis Administrative Judge President, CASE Atomic Safety and Licensing Board 1426 South Polk Street U.S. Nuclear Regulatory Commission Dallas, TX 75224 Washington, DC 20555 Renea Hicks, Esq.

Dr. Kenneth A. McCollom Assistant Attorney General Adelnistrative Judge Environmental Protection Division Dean, Division of Engineering P.O. Box 12548, Capital Station

. Architecture and Technology Austin, TX 78711 Oklahoma State University Stillwater, OK,' 74078 Nicholas S. Reynolds, Esq.

William A. Horin, Esq.

Elizabet*13. Johnson Bishop, Liberman, Cook, Administrative Judge Purcell & Reynolds Oak Ridge National Laboratory 1200 17th Street, N.W.

P.O. Box X, Building 3500 Washington, DC 20036 Oak Ridge, TN 37830 Joseph Gallo, Esq.

Dr. Walter H. Jordan Isham, Lincoln & Beale Administrative Judge Suite 840 881 W. Outer Drive 1120 Connecticut Avenue Oak Ridge, TN 37830 Washington, DC 20036 4

e

_ ___ _ , . , . ._ . . _ _ _ _ . . . _ . . . , _ . . _ . , . ~ - - _

Billie Pirner Garde Mr. W. G. Counsil Citizens Clinic Director Executive Vice President Government Accountability Project Texas Utilities Generating Company 1901 Que Street, N.W. 400 North Olive Street, L.B. 81 l Washington, DC 20009 Dallas, TX 75201 l Ellen Ginsberg, Esq.* William L. Brown, Esq.

Atomic Safety and Licensing Board U.S. Nuclear Regulatory Commission i U.S. Nuclear Regulatory Commission 611 Ryan Plaza Drive, Suite 1000 Washington, DC 20555 Arlington, TX 76011 Robert A. Wooldridge, Esq. Lanny Alan Sinkin Worsham, Forsythe, Samples Christic Institute

& Wooldridge 1324 North Capitol Street 2001 Bryan Tower, Suite 2500 Washington, DC 20002 Dallas, TX 75201 James T. McGaughy Southern Engineering Co. of Georgia Mr. James E. Cummins 1800 Peachtree Street, N.W.

Resident Inspector / Comanche Peak Atlanta, GA 30367-8301 Steam Electric Station c/o U.S. Nuclear Regulatory Commission Atomic Safety and Licensing Board P.O. Box 38 Panel

• Glen Rose, TX 76043 U.S. Nuclear Regulatory Commission Washington, DC 20555 William H. Burchette, Esq.

Mark D. Nozette, Esq. Atomic Safety and Licensing Appeal IIeron, Burchette, Ruckert Board Panel

• a Rothwell U.S. Nuclear Regulatory Commission Suite 700 Washington, DC 20555 1025 Thomas Jefferson Street, N.W.

Washington, DC 20007 Docketing and Service Section*

Office of the Secretary Robert D. Martin U.S. Nuclear Regulatory Commission U.S. Nuclear Regulatory Commission Washington, DC 20555 611 Ryan Plaza Drive, Suite 1000 Arlington, TX 76011 Roy P. Lessy, Jr. , Esq.

Morgan, Lewis a Bockius Robert A. Jablon, Esq. 1800 M Street, N.W.

Spiegel & McDiarmid Suite 700, North Tower 1350 New York Avenue, N.W. Washington, DC 20036 Washington, DC 20005-4798 Thomas G. Dignan, Esq.

Anthony Z. Roisman, Esq. Ropes a Gray Trial Lawyers for Public Justice 225 Franklin Street 2000 P Street, N.W., Suite 611 Boston, MA 02210 Washington, DC 20036 W -

Gfaryy. Mizuno V CounseT for NRC Staff

. -