ML20094P695
| ML20094P695 | |
| Person / Time | |
|---|---|
| Site: | Byron  |
| Issue date: | 08/13/1984 |
| From: | Ericksen E BUSINESS & PROFESSIONAL PEOPLE FOR THE PUBLIC INTERES, TEMPLE UNIV., PHILADELPHIA, PA |
| To: | |
| Shared Package | |
| ML20094P677 | List: |
| References | |
| OL, NUDOCS 8408170326 | |
| Download: ML20094P695 (33) | |
Text
,
^
.?
- Business and Professional'
'h.S,..
pEDCO?MEpogC$
People For The Public; m.A =
Interest-1.].
-Datt:
Adp,ust 13,-1984"'
5 m.
00CKETED
~
UNITED STATES OF AMERICA 3
NUCLEAR REGULATORY COFB1ISSION 8L F016 F1220 BEFORE THE ATOMIC SAFETY AND LICENSING' BOARD In the Matter of
)
jCOMMONWEALTH EDISON COMPANY Docket Nos.
'(Byron Nuclear Power Station, Units 1.&'~2)
)
SUMMARY
OF THE TESTIMONY-0F DR. EUGENE P. ERICKSEN ON CONTENTION 1 (REINSPECTION PROGRAli - INSPECTOR QUALIFICATION AND UORK QUALITY)
I.
Dr. Eugene P. Ericksen is a senior sampling statistician at Mathematica Policy Research, Inc. and a1 professor at Temple University, i
II.
Dr. Erickson has reviewed the Byron Reinspection Report, the testimony of Anand K. Singh, and portions of the testimony of Louis 0. Del George, Robert V. Lancy, and John Hansel.
j Dr. Ericksen has analyzed the ways in which Edison used statistics and probability theory to support its conclusions concerning inspector qualifications and work quality.
III.
Dr. Ericksen concludes that Edison's sampling design and-statistical analysis suffer from four major flaws:
A.
Edison failed to distinguish elements based on their
~
safety significance when establishing its statistical criteria.
The company did not properly select confidence levels and acceptable reliabilities and failed to properly stratify its samples.
j '
B.
Edison over-generalized, offering conclusions about inspectors and elements that had no chance of being included in the reinspected sample.
i I
.T PDR
r-I s
I h
I C.
Edison used an inappropriate formula in calculating reliabilities.
Two assumptions of the formula were violated: inspections were not randomly selected and inspectors were not homogeneous.
D.
Edison did not account for the added uncertainty created by clustering of inspections by inspectors.
For these reasons, Dr. Ericksen concludes that the sampling design of the Reinspection Program and the statistical analysis of the Reinspection Report are inadequate to support Edison's general conclusions about work quality and inspector qualifica-tions.
2 m
x-
c UNITED STATES OF AMERICA NUCLEAR REGULATORY COMMISSION BEFORE THE ATOMIC SAFETY AND LICENSING BOARD In the Matter of:
)
)
Docket Nos. 50 454 OL COMMONWEALTH EDISON COMPANY
)
50-455 OL
)
(Byron Nuclear Power Station, )
Units 1 and 2)
)
TESTIMONY OF DR. EUGENE P. ERICKSEN Q1:
Please state your full name for the record.
A1:
Eugene P. Ericksen.
Q2:
Please provide your job titles and business addresses.
A2:
I am a Senior Sampling Statistician for Mathematica Policy Research, Incorporated, Box 2393, Princeton, New Jersey 08540.
I am also an Associate Professor at Temple Univer-sity, Philadephia, Pennsylvania 19122.
03:
Please describe your job responsibilities at Mathematica Policy Research, Incorporated and list some of your clients.
l A3:
I am responsible for sample design of surveys and statisti-cal evaluation projects.
My work includes construction and evaluation of samples, including the computation of sampling errors.
I have done work for many federal agencies including the Bureau of the Census, the Department of Labor, the Department of Justice, the Social Security Administration and the D~epartment of Health and Human Services.
1
r
-:A,
'l e
- e c
,t I have also worked-for.various corporate clients such as AT&T, GTE, _ Metromobile, !Inc., Blue Cross of Maine, Blue i
~
. Cross of* Massachusetts, and IMS America, and for private-organizations such.as the-American Medical Association.
In addition, I have done work for New. York City and for
~
. agencies of'the States of New' York, Pennsylvania and New-Jersey.
Q4:.Please describe your educational background and work experience.
A4:
I hold a Ph.D. in Sociology and an M.A.
in Mathematical-Statistics from the University of Michigan and a B.S.
in Mathematics from the University of Chicago.
These degrees were awarded in 1971, 1965 and 1963 respectively.
In 1970, I joined the Institute for Survey Research and worked as a sampling statistician.
From 1974 through 1981, I also worked as a Study Director at the Institute.
I left the Institute in 1981 to became a Senior Sampling.Statisti-clan for Mathematica Policy Research, Inc.
I have also taught courses in general statistics, survey sampling, and research methodology while working at Temple University as an Assistant Professor of Sociology from 1974 to 1978, and as an Associate Professor from 1978 through the present.
I.have been an active member in many professional organi-zations for a number of years.
Since 1975, I have served as a Proposal Evaluator for the National Science Foundation (NSF).
I have consulted with the Center for Measurement a
2
r Mcthods and Data Resources of NSF on the developmcnt of standard procedures to evaluate surveys.
I have served as the Chair of the Subcommittee to Review Proposed Internal Surveys of the American Statistical Association (ASA) since 1978, and was a member of the ASA Executive Committee Sub-section on Survey Research Methods from 1975 through 1977.
In 1978, I was appointed by the National Academy of Sciences to a committee evaluating the Census Bureau's method of estimating post-censal population size and per capita income of local areas.
I have published numerous technical papers relating to application of statistics and sampling methodology.
A selected list of these publications is included in my resume, Ericksen Attachment A.
05:
Are you familiar with the Byron Reinspection Program?
A5:
- Yes, I have reviewed the Report on the Byron QC Inspector Reinspection Program (Reinspection Report), the Report Sup-plement, all testimony of Mr. Singh, and portions of the testimony of Messrs. Tuetken, Del George, Hansel and Laney.
Q6:
What is the purpose of your testimony?
A6:
The purpose of my testimony is to evaluate Edison's use of statistics and probability theory in reaching conclusions concerning inspector qualifications and work quality.
I also identify the limits on conclusions which can be reached because not all work elements, work attributes and inspec-tors had a chance of being selected for reinspection.
3 a
i o
a.
Q7:
Is it useful to "7 ply statistics in this context?
A7:
Yes.
Where a 100 percent reinspection is not possible or practical but we wish to make a judgment about inspector qualifications and plant work quality, we can use statistics to draw inferences concerning many plant items and inspec-tors from inspections of selected items and inspectors.
We must be very careful, however, to properly choose the sample and properly determine the population about which inferences can be drawn.
08:
llave you formed an opinion on the adequacy of the samples chosen in the reinspection program and the statistical bases of Edison's determinations of inspector qualifications and work quality?
A8:
Yes.
The Reinspection Program's sampling design and statis-tical analysis is sufficiently flawed that it does not pro-vide adequate support for Edison's general conclusions and inferences about work quality and inspector qualifications.
Q9:
What are the major problems with the sampling design and statistical analysis?
A9:
First, in structuring the Reinspection Program and Report, Edison failed to distinguish elements which are most impor-tant to safety from elements which are less important, or to distinguish elements which are easy to inspect from elements which are difficult to inspect.
By lumping these elements together and failing to apply different criteria depending 4
l a
on the safety importance of the elements, Edison has not
.provided adequate assurance of work quality.
Second, in stating conclusions concerning all inspec-tions at Byron, Edison has seriously over-generalized, making inferences to inspections, work attributes and work elements that had no chance of being selected for reinspec-tion.
Edison lacks sufficient statistical basis for making such inferences.
Third, Edison's statistical methodology was faulty.
The Company used an inappropriate formula in reaching its stat-istical judgments.
Q10:
Why should Edison have distinguished elements based on their safety significance?
A10:
In order to assure that a plant can be operated safely, we are primarily concerned that proper inspections are made of those inspection elements which pose serious risks if not properly inspected, especially those which are hard to in-spect.
To give a simple analogy, it does us little good to know that 99.5 percent of the parts of an automobile were properly inspected if the 0.5 percent that were missed are the brakes and the steering.
To provide assurance that each type of element is properly inspected, Edison should have designed a stratified sample of elements.
The strata would be groups of elements cate-gorized by attribute, type of task, difficulty of inspec-tion, and safety significance.
In each stratum, we would 5
o want to be assured that sample sizes were sufficiently large to be confident of the results.
This would have enabled the Reinspection Program to establish acceptable confidence levels and reliabilities based on the importance of the element..
Ccnfidence levels indicate how certain a statistician is that his or her results are correct.
Reliabilities reflect the percentage of inspection which are correct. For inspection elements where the risks caused by a poor quality are great, we might want to be certain that all were correct and, therefore, reinspect all ele-monts.
For inspection elements where the risks are not as great, but still substantial, we might want to be quite sure that 09.9 percent were correct.
For other inspection elements which are less safety significant, we might be satisfied if we were reasonably certain that 99 percent were correct.
In order to determine the amount of certain-ty and perfection required for each element, choices should have been made using engineering judgments.
These judg-ments, along with their rationales, should have been deter-mined when establishing the program and clearly stated in the reinspection report.
A reasonable reinspection program might have required the following reliabilities and confi-dence levels for the following types of elements.
Type of Element Reliability Confidence Level Critical to safety 100%
100%
Very important to safety 99.9%
991 Somewhat important to safety 99%
95%
Least important to safety 90%
95%
6
1
- By aggregating data,- _ i.e.,. lumping / elements together,
~
Edison-failed _to1 provide adequate _ assurance;of safety..
'Even if we_ are 95 percent certain that 99 percent of all
- inspections that had.a chance of being included.in..our.
sample met design requirements, this does not allow us to state that we are 95 percent certain that.99. percent of the more safety significant-element.s met design requirements.
~
We, of course, want to be more than 95 percent certain that more than 99 percent of very important safety elements met design requirements.
In order to make such'a statement,_
-the, sampling plan should have incorporated special proce-dures for the more safety significant. elements and should have disaggregated cata, breaking it down by attributes and elements.
Q11:
Can you give us an example of a situation where a reli-ability was inflated because of aggregation?
All:
Yes.
In the Reinspection Program, Table VII E-3, Edison-lumped all Hunter " hardware" elements together and reported their reliability to be greater than 99.9% at a 95% confi-dence level.
However, the sample size for the " component inspections for piping and whip restraints", which Mr. Tuet-ken classified in his second most important safety category (Bleuel Attachment B) is too small to provide any meaning-ful basis for reporting a reliability.
Out of 4,321 original inspections of piping and whip restraints, only 4 reinspections were done.
(Ericksen Attachment B.) _This is 7-
s.
>far;below the _200 ' minimum number of1 inspections required by Military Standard 105D, the ' standard which Mr. Singh -
applied in assessing the adequacy offsample' size.
(See tr.
9079.)
d Ik is not"possible to give an example for Hatfield I
because Edison.did not disaggregate Hatfield data by inspection element.
Q12:
In_what way has Edison "over-generalized" in drawing con-clusions,about. work quality and inspector qualifications?
A12: -Statisticians are able to make generalizations.to all popu-lation elements having a known, nonzero chance of being selected into' the sample, and generalizations must be l
limited to this population.
In the Byron reinspeution I
program, numerous work elements 'and attributes had no chance of being included in the sample reinspected.1 Table 1,
i L
attached to-my testimony, lists these items.
In addition, in general, only inspections performed in the first' three months of an inspector's employment were eligible -for sam-l ple selection, and the sample provides an inadequate basis for statements concerning inspections in the second three-l month period or later.
Edison has not provided a statisti-l cal basis from which to draw inferences about the quality of work excluded from the sample.
E Certain inspectors also had no chance of being included in the sample.
Edison has not provided an adequate statis-tical basis from which to draw inferences about these inspectors.-
8
=
--- a
y 4
g Q13:
Is it possible to use inspactors' performance in reinspect-ing those elements and attributes which had a chance of being in the sample as a basis for generalizing to elements and attributes that had no chance of being in the sample?
A13:
Mr. Singh seemed to indicate during cross-examination (tr.
at 9105-9106) that such inferences could be drawn because inspectors were homogeneous.
However, actual data from the reinspection program show that inspectors were not homo-geneous.
Q14:
Why did you conclude that the Company's statistical method-ology was faulty?
A14:
Much of the important work in generating a statistical estimate should be done in advance.
Decisions must be made concerning the reliability sought, the confidence with which the reliability must be demonstrated, and the popula-tions and subpopulations for which generalizations are needed.
Once these decisions have been made, the cample can be planned and nelected.
The statistical planner should determine how large the sample must be to provide the desired confidence intervaln, and whether or not the sample should be stratified to provide estimates for impor-tant subgroups.
Contrary to the Company's assertions, Edison failed to take large enough samples to even assure 995 reliability at a 95% confidence level.
9 t
r-i Q15:
What was the major problem with the Company's application of statistics in estimating reliabilities for work quality?
A15:
Edison, in its analysis, applied a statistical methodology that assumes selection of a simple random sample of inspec-tions (Reinspection Report, page VII-4), but the Reinspec-tion P*rogram did not take such a sample.
Edison may have made this error because the Company designed its program to test initial qualifications of inspectors rather than qual-ity of work.
In calculating reliabilities, Edison used the formula R=
1 - 2.4055 n
where R = reliability at 95% confidence level n = number of inspections in the randomsample.
This formula was derived from page d'46 of Probability and Statistics for Engineers by I. Miller and J.E. Freund (Prentice flall, 1977).
According to Miller and Freund, the formula is an approximation that can be used, when no discrepancies are found, if the following assumptions are met:
"1.
There are only two possible outcomes for each trial 2.
The probability of a success is the same for each trial.
3.
There are n trials, where n is a constant.
4 The n trials are independent."
Id. at 54-55.
It was inappropriate for Edison to use this formula in calculating reliabilities in the Reinspection Report be-cause assumptions (2) and (4) were violated.
10
Assumption (2) was violated because' inspectors were not homogeneous; different inspectors had different probabili-ties of success.
Assumption (4) was violated because in-spectors were not randomly chosen; the selection of inspec-tors were not independent from each other.
Q16:
What is the basis for your conclusion that inspectors were not homogeneous?
A16:' Where inspectors are not homogeneous there will be simi-larities between inspections made by the same inspector.
This creates a commonality within the cluster which can be measured by the "intraclass correlation."
The intraclass correlation can range from a value slightly less than zero to + 1.0.
If the intraclass correlation is equal to zero, it means that inspectors are homogeneous and there is no l
increase in variance associated with cluster sampling.
If l
the interclass correlation is greater than zero, then l
inspectors are not homogeneous.
We can use data from Appendix B of the Reinspection Report to compute intraclass correlations.
The computa-tions show that for llatfield, Hunter and Pittsburgh Testing Laboratory, each contractor's overall intraclass correla-tion was greater than zero.
These positive intraclass correlations indicate that inspectors were not homogeneous.
Another indication of the lack ol' homogeneity. among inspectors is seen from the results of "F tests."
The F test is a common statistical tool that can be used to 11
I'.
.o W
_ a
.' determine whether.. observed variation in reliability:among.
nspect' ors for.a given-attribute is greater than one ex--
i pects by chance'alone.
For a sufficiently high F, Ewe'can
. conclude-.~ that 'inspe'etors are not homogeneous, at a particu-lar-level of~ significance.
. Applying.the F test to the data from Appendix B from the Reinspection Report, we reach the following conclusion:
~
For Hatfield, Hunter and Pittsburgh Testing Laboratory,.the-i F results for each contactor is sufficiently high to war-l rant rejection of the homogeneity hypothesis.
In fact, the l
F results are-so high that we are not only justified in rejecting the homogeneity hypothesis of the 10% level of j
significance and the commonly used 55 level of significance, but also at the particularly stringent 1% level of signifi-l
- cance, i
Q17:
What is the basis for your conclusion that the Program did not select a simple random sampic of inspections?
t i
A17:
When a simple random sample is taken, the selection of.each item is independent.
The inclusion of any one item in the sample should not affect the likelihood that any other item will be included.
In the Reinspection Program, the selec-i tions of inspections were not independent.
t A simple example will make this clear.
Assume Inspector A makes inspections numbered 1, 2, 3, 4 and 5 during his
-first three months of. Work.
Assume that Inspectors B, C, D and E make inspections numbered 6 through 25 during their l
12
first three months of work.
If a simple random sample of inspections is taken, the fact that inspection 1 is in-cluded in the sample will not affect the likelihood that inspection 2 will be included.
In the Reinspection Pro-gram, however, if inspection 1 was chosen to be included in the sample, there would be a 100 percent chance that inspec-tions 2, 3, 4 and 4 would be included in the sample.
Statisticians call this " clustering."
In the e xa m ple,
inspections are clustered by inspector.
Q18:
What is the effect of clustering?
A18:
Clustering almost always increases the uncertainty with which statistical estimates can be evaluated.
Let me illustrate with a simple example.
Let us assume that we have a population of four inspections with two inspectors, Mr. Short and Mr. Long, each making two inspec-tions of a pipe that is three inches long. Inspector Short's measurements are both 2 inches, while Inspector Long's measurements are both 4 inches.
The average of all inspections is 1/ 4 ( 2 + 2 + 4 + 4 ) = 3 inches.
Now let us consider all passible samples of size 2 (i.e.,
that include two different inspections), where no one inspection can be chosen more than once.
For clarity, we will call Short's first measurement 2 and his second measurement 29; like-A wise we will call Long's first measurement 4 and his A
second measurement 4 There are six possible ways in B.
which the inspections can be selected, disregarding the 13
i
- -y tv
,a:.
.. ~
- t x.
. order in which' selections are made -
-Sample-Sample Mean--
1, 2 '-2Bf 2.0 A
- 2, li:
~3 0 h
A 2
B-3.0 A
2N'A-.
3.0 B
2 4
3. 0 --
B' B
4.0 4 ' '4B A
In four out, of six cases one would expect to pick a sample that. yields the average inspection for the entire'popula-tion. #/.
Now let us consider a second type of sample, a clustered sample where the inspector is the unit of selection.
In other words, we take our sample of size 2 either by select-ing Inspector Short's work or Inspector Long's work.
Now there are two possible samples, namely:
Sample Sample Mean Shorts 2A,2B 2.0 Long:
4A' 4B 4.0
- /
In statistical terms, the sample mean is exactly equal to the
~
population mean in four of the six samples, but dif fers by one inch in two of the six samples.
Statisticians measure t,hese discrepancies by a concept known as the standard error, which is the square root of the average of squared deviations of sample meana from the population mean.
It is approximately:
Standarderror[Y)=
f(- kyY/g, where
- Ly = population r6ean TL = mean of sample i N = number of samples.
For the example 'just, described, the standard error ist (1 + 0 + 0 + 0 + 0 + 1)/ 6 = 0.577 35.
3 14 L
m
We have only two possible samples, and they happen to be the two whose values for the sample mean are farthest from the population mean.
In no cases could we pick a sample that yields the average inspection for the entire popula-tion.
The sample average would either be one inch too short or one inch too long.
1/
Hence, the uncertainty associated with the sample esti-mates generated from a clustered sample is greater than the uncertainty associated with the sample estimates generated from a simple random sample, in which all selections are independent from all other selections.
Edison should not have used a formula that assumes simple random sampling in determining the reliabilities of samples that were clus-tered by inspector.
Q19:
Can you give us an example from the Reinspection Program of a situation where a reliability was overstated because of the effect of clustering?
A19:
Yes.
A good example can be derived from data on the Hunter inspection element " Documentation on component inspections for piping and whip restraints."
There were 37,230 original inspections of this element and 1,476 reinspections.
(Erick-son Attachment B.)
The 1,476 reinspections, however, are clustered.
1/
The standard error is larger, namely:
{(1 + 1)/2 =
1.0.
15
To determine inspection reliability for a clustered sample, the statistician must first calculate the " design effect," the quantitative measure of the extent to which a reliability estimate is reduced by the effect of cluster-ing.
When the actual sample size is divided by the design effect, we obtain the ef fective sample size, which should be used in computing reliability.
In the case of " documentation on component inspections for piping and whip restraints," the design ef fect is 5.2257.
This yields an effective sample size of 282 rein-spections.
Correcting for the effect of clustering, the effootivo sample size of this inspection element falls from 1,476 to 282.
(see Appendix 1.)
282 reinspections out of 37,230 original inspections is far below the sample size of l
500 reinspections required by Military Standard 105D.
Edi-son, therefore, cannot assert a meaningful reliability for l
this element.
l 020:
Can you summarize the major problems, with the Botnspection i
Program?
A20:
Yes.
First, Edison did not establish adequato criteria for its statistical analysis.
The Company did not properly select confidence levels and acceptnble reliabilities, and I
failed to stratify the nample taking account of safety significance.
Second, Edison over-generalized, offering conclusions about inspectors and elements that had no chance of being included in the reinspected sample.
16
Third Edison used an inappropriate formula in calcu-lating reliabilities.
Two assumptions of the formula were violated inspectons were not randomly selected and inspec-tors were not homogeneous.
Fourth, Edison did not account for the added uncertainty created by clustering of inspections by inspector.
For these reasons, the sampling design of the Reinspec-tion Program and the statistical analysis of the Reinspec-tion Report are inadequate to support Edison's gee. oral conclusions about work quality and inspector qualifications.
l i
17
Pago 1 of 2 6
TABLE 1*
i i
ATTRIBUTES AND ELEMENTS TilAT llAD NO CllANCE OF BEING SELECTED FOR REINSPECTION l
IIATFIELD, l
Embedded conduit Underground duct runs llatorial and equipment receiving Cribic installation Non-neg bus duct l
Material handling l
Stud weldin; Limit switc1 gasket replacement Removal of heat shrink tubing on conax penetrations Ilousekeeping All welds for which the original inspector could not be identified **
l i
ilUNTER Visual inspection of valven Ferrite inspection l
Piping hydrostatic tent l
Piping wold interpass temperature innpoetion Joulos test inspection Codo name plate change Inapoction of weld defect removal cavity Whip restraint - fitup and tack weld Buried pipe coverin6 inspoetion Piping - pre-heat inn inction Whip runtraint - pro-acat innpection Pipe weld - Shield gan verificat. ion Component support - nnubber otroking l
Bolting - turn-of-nut.
~
Source:
Written tuntimony of Richard B. Tuotken, I
Attachment B, tr at 8408.
Source:
Report on the Byron QC Inspector Reinnpoetion Program, at IV-a, discunning flatfield second audit.
i l
m.
..m.
Page 2 of 2
.A TABLE 1 (cont'd)
Documentation l
Forrite inapection Joulon tent Code name plate change Wold defect removal envity Component nupport - anubber stroking Bolting - turn-of-nut PITTSliURGli TESTING 1A110RATORY Robar detection i
Bolting - turn-of-nut (connectionn)
Calibrations (torque wrunchen, thermometers, feeler gauges, aculos, gauges)
Cadwolds (rebar coupling)
Soiln (buck fill)
Concreto field (pincoment)
Concret.e Lub (aggregato)
ATTRIBUTES AND El.121EllTS WilICII WERE RE!!!SPECTABIJ: BlJT Wl:Rh !;0T lli;IlltipKCTEl>
llATFIEl.!!
Cable pan coverri Cable pan identification l
li_l)NTER I
l Component uupnort finnt innpe d ion (typo 3)
Component nuppo r t.
final innpcetion (type 4)
Equipment installation Document a t lot),
Component nupport - final innpnetton (type 3)
Component nupport - final inapuetton (typo 4) l l
4 Al'PI;NDIX 1 Calculation of Design 1:ffect and Effective Samplo SLxo The design effect anacciated with a cluntered nample can be calculated by uning the following formular doff 1
+
roh (B-1),
a where doff design offect
=
roh the intraclass corrulation B
the averago cluster nino
=
Below, this Formula in applied to the llunter innpoetion element
" Documentation on component inspectionn for piping and whip restrafnts,"
Roh in and is equa,the encimated intraelnsa correlation for llunter inspectorn l to 0,0172477, B aquals the total number of reinnpoetions divided by the total number of c1untern (i.e., reinnpectorn),
In thin cano,11 equaln 1,476 divided by 6, which in 246.
Thorofore, doff a 1 4 0.0172477 (246-1) 5,2257 a
To calculate the effectivo namplo nize, and thereby adjunt the actual anmplo nize to reflect the offect of clunturing, wo une the following formula:
offectivo anmplo nine netual nampin_nin a
~dWf In thin case, tho effectivo namplo alzo in:
1,476/5.2257 282.45 a
or approximately 282 reliinpectionn.
r s
, y Page 1 of 3
.,, A L,.i ERICKSEN ATTACHMENT-A' 4
EU HC pef 4NELL ERICKSEN EDUCATION:
y
'19 71
!Ph.D., Sociology, University of H1ehigan 1965 H.A., Mathematleal Statistics, Unisersity of Michigan 196)
B.S., Mathemat !cs, Universi ty of Chicaqo POSITIONS:
1981 -
Senior kmpling Stat Istician, itsthematica Policy Hesearch, Inc.
1970 - 1931 Institute for Sursey Hesearch 197'+ - 19 61 Study Director 1970 - 1981 Sampling Stat tstician 1979 -
Department of Sociology, Temple lhlversity 1979 - 1981 Associate Professor 1974 - 1978 Assistant Professor 1969 - 1970 Student Fellow, University of Michigan 1967 - 1968, 1964 - 1966 Student Associate Institute for Social Research, LAilsersity of Michigan 1966 - 1967 Lecturer, Balham and Tooting College of Comerce, London EXPERIENCE:
At Mathematica Polley Research, Dr. Ericksen has had responsibility for the sample design of surveys on diverse populations including households in the thited States, industries using data comun1 cations equipment, physicians, social security recipient s, and mergency roms in tospi tals. He has also conducted statistical evaluation projects including several which were the basis for expert testimony in court room li tigat ion. He is currently the chief technical advisor for plaintif f s in several suits concerning the adjustment of the 1960 Census.
At the Institute for Sursey Researc5, Dr. Erteksen worked on 51rtually every major project as Sampling Stat ist ician. His duties included designing and constructing a national sample of households, adapting this sample to the sampling frm lists, constructing national samples, and evaluating the samples with respect to computing sampling errors. He also designed, constructed, and evaluated subnational surseys for particular states arsi loc.nl areas. As Study Director, Dr. Ericksen conducted studies mder three joint contracts with the Bureau of the Census. The objective of these studies was to develop a mettedology for using regression analysis with sample data to compute postcensal estimates for local populations, and they were conoucted from i972 through 1974 He was slso co-princ1;a' ir.vestigator on the studies "Etteilet ty and Community in a Metropolis." supported by the t&itional Institute of Mental Health, Center for Metropolitan Studles, 197 5 t hrough 1979, and " Fertility of an American Isolate Subculture (The Old Order Amish)," supported by the Hattonal Institutes of Health,1976 through 1978.
At Temple University, Dr. Ericksen has taught courses in general statistics, sursey sampling, research nettedology, fa:ntly sociology, ethnic groups, population, and human ecology. In the spring of 1980, as part of the Experimental Student In t e rt. Program of the Bureau of the Census, he taught a special course wherety mdergraduate student s were tralned to become eser,,erators in the 1960 Census.
At the Population Studies Center, University of Hichigan, Dr. Ericksen, under a joint contract with the Bureau of the Census, wrote a Ph.D. dissertation to develop the methodology for ustry regression analysis and sample data to compute posteensat population estimates for local 1 areas.
Dr. Ericksen is also a research associate for the Center for Ph!!adelphia Studies, University of Pennsylvania. He is also a member of the American Statistical Association, the Population Association of America, and the American Sociological Association.
1 m.
. m
f
' t.,
Page 2. of 3
?. '
' EUCENE PEtN4 ELL ERIICKSEN
.Page Two PROFESSIOML ACTIVITIES Atto 0FFICES:
Member, Executise Committee, Subsect ton on Sorsey Hesearch Hethods, American Statistical Association, 1975-1977.'
~
Member, Board of Hesiew, American Statist tcal Associat[or. Project on the Assessment of Sursey Research
}
Practices, 1976 and 1977. The coulttee esaluated the folluwing report: " Des eloping of Sursey Hethods to. Assess Survey Pract tees," by Barbara A. tbilar aid C. Michael Lanphler, and published by the American Statistical Associatic n, 1978.
Publications Liaison, Section on Sorsey Researdi Metrods, herican. Statistical Association,-1979.
Proposal Esaluator, t.at ional Science Foundat ion, l975 to prewnt.. He has also consulted on the desetopment of standard procedures to esaluate surseys nith the Center for Measurment Hettods ard Data Resources of the HSF.
Chair, Sut,comalttee to Rev tew Proposed laternal Surseys of the ASA ( American Statistical Association),
- 1978 to present.
Member, committee appointed by Hat tonal Wdemy of Sciences to evaluate Census bureau metted of estimating postcensal population size and per capita income of local areas.
SEL ECTED PAPERS AND PUBLICATIONS:
" voting Patterns in Pennsylsanta Judicial Primaries:
1983" report to Judiciary Committee of the Pennsylvanla State Senate. Presented Ibvember 30, 1963 (with Christena E. Hippert).
"Using Administratise Lists to Estimate Census Onissions: An Example," (with Joseph B. Kadane) 1983, presented at Hectings of American Statistical Associat ton.
"Using the 1980 Census as a Popululation Standard," (with 30seph 0. Kadane) 1963, presented at Heetings of American Statistical Association.
" Estimating the Population in a Census Year," presented to the Federal Court of the Sout he rn District of Nc= York, 1982, and to conference on " Data ik eds for America In Transitlon," sponsored by the Congressional Research Service, Library of Coryress,1983 (with Joseph B.
Kadane).
"Can Regression Be Used to Estimate. Local Unercount Adjustments?" Proceedings' of the 1960 Conference on Census thdercount, Aly 1960, pp. 55-61.
"The Cultisation of tre Soll as a Haral Direct ise: Popula tion Crowth, f amily Ties, and the Maintenance of Communi ty Arrung the-Old Order Amish." Rural Sociology, sol. 45, Spring 1960, pp. 49-66 (with Jutta A. Ericksen ard John Hostetler).
"Fert ility Pa tterns and Ti erds hcrs ti.e C'.d C.-der hish."
Populat tsn Studies. a !. 33, July M879, pp.
255-276 (with others).
"The Division of Family Roles." Journal of Harriage arw1 the Family, sol. 41, May 1979, pp.
301-313
(=1th 3ulla A. Ericksen and tit tlam 'rancey).
" Antecedents of Community: Eccnmic and Institutional Structure of Urban Heighborhoods."
American Sociological Review, sol. %, April 1779, pp. 253-262 (with *1L ilam L. Yancey).
"nork and Residence in Industrial Philadelphia." Journal of Urban History, sol. 5, March 1979, pp.
147-182 (alth William L. Yancey).
" Defining Criteria for Evaluating Local Estimates: Discussion of Papers by Conzalez ard Fay."
In Synthetic Estimates for Small Areas: Statistical workshop Papers and Discussio_n_,
t4tional Insti tute on Urug Abuse Hesearch Monograph Ho. 6, t ebruary v i /, pp. Id >- P/1.
"A Tale of Three C1tles: IBlacks and Immigrants in Philadelphla, 185G-1880,.1930, and 1970.
"The Arnals, sol. 441, 3anuary 1979, pp. 55-81 (with others).
i
~l i
. w.
Page 3 of 3
. '. ' S*
s.
- ' EUCENE PENNELL ERICKSEN
-Page Three SQ.ECTED PAPERS At40 PUBLICATI0ttS: (continued)
- " Immigrants ard their Opportunities: Ph!1adelphia, 1850-1936." Presentel at a sympostum on Invilgration held at the meetings of the %erican Association for the Advancement of Science, Houston, Texas, January 1979 (with willian L. Yancey).
"Repert of the Conference on Economic and Demographic Hethods for Projecting Population: Surrrnary and.
Recomunendat lons."
The American Statistical Association, Aprli 1978 (with. Richard Engels).
" Reply to Levine and Bergesen." Merican Socioloolcal Review, sol. 42, October 1977, pp. 825-827 (with William L. Yancey and Richard H. Juliant).
"Some Lessons Learned from Corutucting Federally Sponsored Surseys " Proceedings of the Sncial Statistics Section, American Statistical Association, August 1977, pp. 183-18).
" Sampling a Rare Population: A Case Study."
Journal of the Vierican Statistical Association, sol. 71, December 1976, pp. 816-822.
" Emergent Ethnicity: A Review and Reformulation." American Sociolontcal Review, so l. 41, 3une 1976, pp. 391-403 (with william L. Yancey and Richard Jullan1).
" Outliers in Regression Analysis anen Heasuranent Error is Large." Proceedinqs of the Social Statistics Section,' American Statistical Association, August 1975, pp. 412-417.
" Population Estimation in the 1970s: The Stakes are Higher." Report to Bureau of the Census Hay 1975.
"A Regression Method for Estimating Population Changes of Local Areas." Journal of the %erican Statistical-Associat ton, 501. 69, December 1974, pp. 667-975.
"Recent Developments in Estimation for Local Areas." Proceedings of the Social Statistics Section, American Statistical Association, December 1973, pp. 37-41.
"A Method for Combining Sample Sursey Data and Symptomatic Indicators to Obtain Population Estimates for Local Areas." Demograohy, sol. 10, May 1973, pp. 137-160.
" Test of a Statistical Proce-dure for Computing Estimates for Local Areas." Report to Dureau of the Census, 3anuary 15, 1973.
l
F1 ERICKSEN ATTACllMENT B Page 1 of 6
. g, e Edison's Amended Response to Interrogatories ll(c) and -12(c)- of In tervenors ' First Set of Interrogatories I.
II.
III.
EV.
V.
Ntnter of
'Ibtal Inspecticns Total Inspectors
. Inspection Perfocmd Reinspections InspectirrJ Insp rtors (by attribute) through 8/31/82 Performed Attribute Painspected class I cable 26,230 4,776 9
2 pan hangers class I cable 1,643 80 10 1
FM cable ter::tinations 78,548 7,784 16 5
equipaent.
628 27 4
3 modifications class I exposed 30,210 2,793 15 6
conduit A-325 bolt 14 8
3 1
installation conduit "as-built" 180,000 44,777 28 8
program vistal weld 312,000 27,844 17 8
inspection Notes: The numbers in Column II are esti: rated and exclude inspections perfonred af ter Scptenkr 1,1982. The ntsber of total inspections and total reinspecticns shown for attributes 4 ard 6 refer to the ntmber of itans en an inlividual ungection report ard inspection reports respectively.
All other ntrbers in cnit:ms II ard III refer to irdividual inspections of various oxponents. The nt:bers in Column IV are the ntuber of inspectors who, on their first date of certification, were cartified in the inspection attribute and actinlly perform 2d inspections of that attribute between the date of first certification ard Septa,ber 1,1982. The total' ntmber of Hatfield inspectors cstployed between 1976 and Septaber 1, 1982 is 86.
Rurf inspretors are certified in more than one inspecticn procedure. For the c6jective inspections, inspection attributes 1, 2, 5 and 7 require similar inspection skills as do insWon attributes 3 and 4.
w r
w
,-,-ee w
m-,
Pcga'2 of 6
]
2c..
Total' Total Number of Inspectors Inspection Inspections Reinspections Inspecting
. Inspectors (by inspection element)
Performed Performed Inspection Element.
Reinspected 1.
Documentation for piping 9,745 247 12 1
mechanical joint witness of torque-initial, inter-mediate and final 2.
Documentation of piping 430 120 3
1-hydrostatic test 3.
' Documentation on piping 5,896 321 13 4
inter pass inspection 4.
Documentation on name 25 5
2 1
plate inspection 3.
Documentation on finished 187,129 14,584 16 11 weld inspection of piping and whip restraints Documentation on finished 29,272 963 10 6
weld inspection for component supports 7.
Documentation on component 37,230 1,476 16 6
inspections for piping and whip restraints d.,
Documentation on fit up 98,861 3,609 16
- 9 and tack welds for piping and whip. restraints
-s.---
9.
Documentation on piping 2,434 41 10 i
field bonds-final visual, ovality and radius j
8
, :. ].
i
Page 3 of 6 Total Total Number of Inspectors l
Inspection Inspections Reinspections Inspecting-Inspectors I
(by inspection element)?
Performed Performed Inspection Element Reinspected.
i
.0.
Documentation on rev'iew"of 168,815 21,161 6
6 type a inspection for' type 3 inspection final reivew
.l.
Documentation on mechanical 5,929 92 11 1
l joint inspection for piping preassembly inspection (component)
.2.
Documentation on piping 4,355 29 8
1 mechanical joint inspections line up inspections l
(fit up) 3.
Documentation on location 3,219 86 5
2 acceptance between com-l ponent support and item i
being supported
!.4.
Documentation on component 9,230 158 4
1 support inspection checklist Documentation on location 5,707 353 8
4
-of field welds for piping
-inspections f.
Documentation on piping 60 10 2
1 holiday jeep test I
!.7.
Documentation on component 2,589 782 5
4 supports concrete expansion l
anchors i
a e
M 0
g i
r--
a Paga 4Lof,6 Total Total Number of Inspectors Inspection Inspections Reinspections Inspecting Inspectors Performed Performed Inspection Element' Reinspected:
(by' inspection element) 3.
Documentation on piping
'2,483 231 6
2 a
cnd whip restraints pre-heat inspection 9.
Documentation on piping 685 10 4
1 verification of shield gas O.
Documentation on piping 401 122 4
2 and component. supports temporary attachments inspection 1.
Small bore type 3 final 3,503 3,014 5
5 hardware inspection reports 2.
Small bore type 4 final 47 35 2
2 documentation inspection reports Whip restraints type 3 185 176 1
1 final documentation inspection reports 4.
Whip restraints type 4 12 6
1 1
final documentation
,/
inspection report 5.
Equipmen t _ typt.3._ fina L....
13,._ _.
7_.
I 1
documentation' ~ inspection,._.,..
report 2
2
- 6.
Documentation on large bore 401-3.95 i
piping types final inspection 1.
'f
., "
- e. [....,,.1 :
iI q ;. ^
V
v Pagn.5 of 6'
^
s Total Total Number of Inspectors Inspection Inspections Reinspections Inspecting Inspectors (by inspection element)
Performed Performed Inspection Element Reinspected F.
Piping mechanical joints 2,714 606 12 10 witness of torque initial, intermediate, and final J.
Component supports torques 405 150 5
2 1
Finished weld inspection 4,395 2,291 17 17' for piping and whip restraints 1
Finished weld inspection for 3,282 1,437 11 9
component supports Piping and component supports 27 13 4
2 temporary attachments inspection 2
Component inspections for 4,321 4
16 1
piping and whip restraints 2,
Fit up and tack weld for 9,395 5
16 2
piping and whip restraints 4
Piping field bends inspection 729 417 10 9
final, visual ovality and radius 5.
Verified location acceptable 472 254 5
4 between component support.and item being supported Component support inspection 18,378 13,932 4
4 checklist t
2. ',
,,)
J e
s-
,,r g
l
)
.Pagi 6?of 61
.1 1
Total Total-Number of Inspectors:
Inspection Inspections Reinspections Inspecting Inspectors:
(by inspection element)
Performed Performed Inspection Element Reinspected!
7.' Dimensional on location of 976 567 8
8 field welds for piping inspections j
Component support concrete
.1,154 772 5
4 expansion anchors inspection 1
Small bore type 3 final 22,762 10,515 5
5 hardware inspection reports Small bore type 4 final 155 75 2
1 hardware inspection reports Large bore type 3 final 1,535 195 2
1 inspection report 2.
Whip restraints type 3 final 4,684 876 1
'l 3
hardware inspection report 3.
Whip restraints type 4 final 134 22 1
1 hardware inspection reports NOTES:
The Total Inspections Performed are those performed by the inspectors l
whose work wac reinspected in the reinspection program.
The total number of inspections is unknown.
Inspections conducted after August 31, 1982, are c.m. eKcluded..The. Number.,of Inspectors Inspecting _ Inspection Element is the number of inspectors wh~o, were certified to perform inspections for the inspection element and whose inspections were reinspected.
Information on the total h
number of inspectors inspecting'each inspection element is not available.
The total number of Hunter inspectors employed at Byron between 1976 and September
.)
l','1982 is 84.7 The certifications.of.these' inspectors permit them.to conduct J
4 inspections of moretthan'one. inspection element. ' Inspection. attributes 1-26,
~
27-28, 29-31, 32.38,. 39-41. and 42-43Lrequire simil'ar ;inspecti,on skills.
i 4
s.
.c.
n r -
3
.i
.. g..., :
g'
~ ', ' '
,' ;y.
e'
+
.)
P p.
e 4
I'
,, g; g,,,
4 6.
. t {';
c
,7
. s,
.L
m e s e
RELATED COMESPONDENCE UNITED STATES OF AMERICA NUCLEAR REGULATORY COMMISSION BEFORE THE ATOMIC FAFETY AND LICENSING BOARD ggggg U3nRC In the Matter of:
)
)
Docket No. 50-%d4 MD 16 R2:38 COMMONWEALTH EDISON COMPANY
)
50-455 OL
)
'60hb$gf3 fly ~,
f (Byron Nuclear Power Station,
)
Units 1 and 2)
)
SRANcy CERTIFICATE OF SERVICE I hereby certify I served copies of Intervenors' Motion for Leave to File Testimony of Dr. William H. Bleuel; Direct Testi-mony of Dr. William H. Bleuel on Contention 1 (The Reinspection Program); and Testimony of Dr. Eugene P. Ericksen on the fol-lowing persons by having said copies placed in envelopes, proper-ly addressed and postaged (first class) and Laving them deposited in the U.S.
mail at 109 North
Dearborn (or,
as indicated by an asterisk, sent by Purolator Courier or Federal Express), except that Mr. Miller's copy was hand-delivered.
Ivan W. Smith, Chairman Stephen Lewis, Esq.
Administrative Judge Office of Executive Legal Atomic Safety and Licensing Dircctor Board U.S. Nuclear Regulatory U.S. Nuclear Regulatory Commission Commission Washington, D.C.
20555 Washington, D.C.
20555 Dr. A. Dixon Callihan Office of the Secretary of Administrative Judge the Commission Union Carbide Corporation ATTN: Docketing & Service P.O. Box Y Section Oak Ridge, TN 37830 U.S. Nuclear Regulatory Commission Dr. Richard F. Cole Washington, D.C.
20555 Administrative Judge Atomic Safety & Licensing Alan S. Rosenthal, Chairman Board Administrative Judge U.S. Nuclear Regulatory Atomic Safety & Licensing Commission Appeal Board Washington, D.C. 20555 U.S. Nuclear Regulatory Commission Washington, D.C.
20555 1
i'
- v Dr. Reginald L. Gotchy Joseph Gallo, Esq.
Administrative Judge Isham Lincoln & Beale Atomic Safety ~& Licensing 1120 Connecticut Ave., N.W.
Appeal Board Washington, D.C.
20036 U.S. Nuclear Regulatory Commission Michael I. Miller, Esq.
Washington, D.C.
20555 Michael R. Goldfein, Esq.
Isham Lincoln & Beale U.S. Nuclear Regulatory Three First National Plaza Commission, Region III Chicago, IL 60603 ATTN: JOHN STREETER 799 Roosevelt Road Glen Ellyn, IL 60137 DATED:
August 13, 1984 h
V AttornWy ji 2
L