ML20056D993
| ML20056D993 | |
| Person / Time | |
|---|---|
| Site: | Yankee Rowe |
| Issue date: | 07/08/1993 |
| From: | Denise Edwards YANKEE ATOMIC ELECTRIC CO. |
| To: | Gillespie F NRC - REGULATORY REVIEW GROUP, NRC OFFICE OF THE EXECUTIVE DIRECTOR FOR OPERATIONS (EDO) |
| Shared Package | |
| ML20056D986 | List: |
| References | |
| SPS-93-057, SPS-93-57, NUDOCS 9308190152 | |
| Download: ML20056D993 (8) | |
Text
ig
^ p.
ENCLOSURE'2-
=
YANKEEATOMIC ELECTRIC COMPANY
'$y,@.s*]o'.;";"l" 580 Main Street, Bolton, Massachusetts 01740-1398
- YAMEE
~
July 8, 1993 SPS93-057 Dr. Frank T. Gillespie Chairman, Regulatory Review Group office of the Executive Director for Operations U.S.
Nuclear-Regulatory Commission Washington, DC 20555
Dear Dr. Gillespie:
I appreciate having had the opportunity to meet with you and discuss.the activities of your review group during my visit to Washington last May.
As a result of a review-of the draft report-
~
here at Yankee, we have been able to assemble what appear to be a.
meaningful-set of comments for your consideration..
We. intend to.
devote'some' time reviewing them in the near future to be sure their intent is clear prior to forwarding a. final' written response.
In anticipation of that submittal, however, I would like~to highlight a matter to you which may be.of particular significance to the report because it represents a large. potential savings to licensees if administered properly by the. NRC staff. : Additionally,.
it is not currently addressed in the report but will be. included in our comments on it.
This issue ~ concerns a very.significant' reduction in the perception of seismic hazard that has evolved'over:
the last decade but with no commensurate reduction in the NRC i
requirements for seismic " review" programs.
Over the last 15 years the Lawrence Livermore National'
- Laboratory (LLNL), under sponsorship ofLthe'.USNRC,.hasideveloped probabilistic models to assess the likelihood' of exceeding the -
seismic design basis at existing nuclear facilities in the~ eastern United States ~(EUS).
The earliest use:of the LLNL results was..in.
conjunction with the evaluation of the Systematic Evaluation'(SEP)
Plants (1979-1981).
/[-
]
-9309190152 930723 PDR ADOCK 05000027 I
.H PDR f-
Dr. Frank T.
Gillespie July 8, 1993 Page 2 Comparison of these early results to the seismic design basis 1
at older plants (pre - 10 CFR 100, Appendix A) indicated that the probability of exceeding the design basis was disturbingly high (possibly as high as 1 in 100 per year at some plants).
Comparison of these results to post Appendix A plants indicated that the probability of exceeding modern plant design spectra was on the order of 10'8 to 10" per year, which the NRC Staff determined would be acceptable providing no
" weak links" existed that could significantly reduce the capability of successfully coping with a large earthquake.
These early LLNL results, combined with the
' Charleston Issue',
spawned several years of heightened IGC sensitivity concerning seismic issues.
Arguably, it was this heightened NRC sensitivity that was the genesis of the two major seismic review programs currently in progress, namely, Unresolved Safety Issue (USI)
A-46,
" Seismic Qualification of Equipment in operating Plants" and the seismic Individual Plant Examination for External Events (IPE/EE).
During the eighties, with encouragement from the staff, the Electric Power Research Institute (EPRI) developed a state-of-the-art seismic hazard methodology.
Comparisons of the EPRI and LLNL results for 1985 at typical EUS sites showed dramatically different perceptions of seismic hazard, with the EPRI results much less distressing.
Comparison of the 1989 results showed some convergence but the differences were still apparent.
However, as a result of continued discussion between scientists, as well as evolutions in the state-of-the-art, today's LLNL results (1992) compare favorably to the EPRI results at most EUS sites.
Significantly, while the plant-specific harard associated with seismic events computed by LLNL and,
- thus, the perception of seismic harard that it connotes has been steadily evolving in a direction of reduced harard, the staff has continued to press for detailed seismic reviews of all plants (the seismic IPE), with the older vintage plants required to perform, not one, but two seismic review programs (IPE/EE and A-46).
It is i=portant that you appreciate just how much the perception of hazard has been reduced.
Figure 1 shows the progressive reduction in hazard for an eastern U.
S.
site.
From these data it is readily apparent that the concern for seismic hazard in the late seventies and early eighties is now no longer substantiated by the research results from the staff's own consultants.
From a cost vs. risk viewpoint, we believe that both n
Dr. Frank T.
Gillespie July 8, 1993 Page 3 the A-46 and seismic IPE programs warrant some measure of review as a consequence of this latest result from the NRC funded research.
Both of these are discussed further below.
Regardless of these research results, several plants have recently received letters from the staff challenging the adequacy of.that plant's Design Basis Earthquake (DBE) as the appropriate review level earthquake for resolving USI A-46.
Some of these plants have com::titted to the use of a greater than design earthquake, others are considering a Backfit appeal.
I believe that it is imperative that the staff consult these latest LLNL hazard curve results for the insight they may provide regarding the adequacy of current DBEs.
It is almost a certainty that these new results will substantiate the seismic design adequacy of the plants that have been questioned by the staff.
At issue is a substantial investment, both in review and potential upgrade costs, if these plants are forced to increase their seismic designs based on earlier, but now clearly invalid, perceptions of hazard.
Further, with regard to Seismic IPE's, the level of effort for each plant intending to conduct a Seismic IPE has been established by a complex process that included some consideration of both LUIL and EPRI hazard results.
Simply put, the greater the perceived hazard at a site the larger the " screening" earthquake and the more effort would be required in the review process. There is a small savings (in relay reviews) afforded plants that exhibited significant margin in the probability of exceeding their design bases (reduced scope plants).
The respective screening earthquakes to be used in searching for low cost opportunities for significantly improved seismic ruggedness were not intended to be incredible; low probability perhaps, but not incredible.
Based upon the latest LUTL results, in the frequency range of interest for structures and equipment (2-l 10hz), at many sites, the screening earthquakes (eg. 0.3g NUREG/CR-0098) have such a low probability that were they " tornados", they would be diszissed by existing sta H guidance documents as incredible and would be considered out-of-scope for the plant's IPE/EE.
At such sites, the level of effort for the seismic IPE should be sharply curtailed.
It should be limited to a plant walkdown of critical equipment and structures by qualified seismic damage experts.
Seismic IPE/EEs '(or PPAs) are estimated by both the staf f and industry to cost on the order of $1M, less upgrades.
The
4 Dr. Frank T.
Gillespie July 8, 1993 Page 4 potential exists for sienificant savinas for many plants.
In sunnary, I believe that this topic deserves special note in the report.
As can be seen from the forgoing, there apparently has been a transition in beliefs regarding seismically induced design basis accidents.
It was perceived as a low probability event (pre-1980), then considered as perhaps highly likely (early 80's), and finally a consensus of LLNL and EPRI results that this design basis accident is, in fact, a very low probability event.
In the =ean time, significant suns of noney have been spent and are continuing to be spent without even a potential for a concomitant reduction of actual risk.
Indeed, some of the greatest excesses in regulation of nuclear power plants may have occurred in the seismic area.
The industry must be more diligent in seeking-out and reconnending relief fron unwarranted expenditures such as may be involved in this area.
It seems appropriate that the Review Group Report should focus on this issue.
Very truly yours, D.
W.
Edwards Director, Industry Affairs Attachment c:
S. Floyd (NUMARC)
T. Marsh (U.S. NRC)
R. Voll er (U.S. NRC)
?
I i
P i
i
1 s.
L Comparison of:1.0E-3 Uniform Hazard Spectra at a New England Site.
50 20' 0
1985 LLNL Results g)
Q ww w.. m.m
~
10
,""'~ %'- -------------% '-----
E
's O
SSE N
1989 LLNL Results y
.~ ~.,___---. %,%
n,s-.--____--________.
5
~ - -
~ ~ ~ ~ ~ ~ ~ ~ ~,,,,,,%g g
'.hw%
N u
.- - - _ 3 992. LLN L Besults.... _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ % '.~,,, _ _ _ _ N o
%c
__s O
2 q%
N
~~~~~~,,,
,,,,....N...g.,_ _ _, ~s
%q._.
3
}
Note: Original SEP/LLNL results were basis 9
G3 L._
for over react. ion to seismic issues.
,~~
@ 0.5 - - - - currarituNCiesults snow th5t pr6babifities of - - - - - -
.f'------
~
~
G) exceeding the SSE compares favorably to O
^
the -1985 probabilities of exceeding a -
0.2
- - - -.0. 3 9 N URE G LCR: Q09 8.S p ect r u m,....
0.1
-1.
2
~3 5
- 10 :
20 30 Frequency (Hz) l J
i A Process tt, Expedite Grantina of Relief Reauests Generically i
Volume One, page 5 of the Review Group's draft report poses the question of why licensees are not taking advantage of
... existing opportunities offered by the Commission via the Technical Specification Improvement Program line-item improvements."
The Review group notes their agreement with the industry observation that "... amendment requests dealing with economic relief fall to the bottom of the priority list... " and
...it is the staff view that the need for each further notes "
licensee to customize requests eliminates the opportunity for rapid action."
We would note that there is another deterrent to licensee's exercising the line-item improvement option:
our experience has been that the simplest possible change to a plant's Technical Specifications costs at least $20,000.
The purpose of this discussion is to propose a cost-effective process to simultaneously remove an unnecessary requirement common to the licenses of a number of licensees.
At present, the process for implementing -generic issue license amendments functions as follows:
1.
The NRC issues ' for public comment a notice of a proposed generic communication which identifies the specific Technical Specification improvement opportunity.
Recent examples include:
i) relocation of the Technical Specification tables on instrument response time limits (58FR18118),11) line-item Technical Specification Improvements to reduce testing during power operations (58FR16881), and iii) modification of the Technical Specification administrative control requirements for Emergency and Security plans (58FR17293).
2.
Comments on the proposed generic communication are resolved and the resulting Generic Letter is formally issued to licensees.
The Generic Letter typically provides detailed guidance to assist licensees in preparing a license amendment request to implement the improvements offered in the letter.
3.
Licensees who choose to act on the opportunity afforded by the
. Generic Letter submit their plant specific license amendment request pursuant to 10 CFR 50.30 and 10 CFR 50.90.
4.
The NRC Project Manager for the plant is tasked with verifying that the license amendment request conforms to the guidance provided in the Generic Letter.
Technical assistance is typically sought from the appropriate NRC technical review Page 1
\\gla\\ cost jds
P branch, or the regional of fice. The additional required steps in the review and approval process are detailed in Section 3.3 of NUREG/BR-0073, Rev.
1, the NRC Project Manager's Handbook and 10 CFR 50.91 and 50.92.
Several details of the process described above merit additional comment because of the manner in which they impact the overall efficacy of the process.
Planning an effective, efficient work effort requires, among other factors, a reasonable estimate of the expected volume of required work and a knowledge of when information or data which is required to perform the work effort will be available.
The current process denies this information to the NRC staff and thereby creates a significant resource allocation problem.
When an improvement opportunity is formally noticed in a Generic Letter, the NRC has no way of knowing either how many licensees will choose to take advantage of the opportunity or when the license amendment requests will be filed.
Although most licensees attempt to keep their Program Managers apprised of anticipated submittals of license amendment requests, to the best of our knowledge, this information is not utilized for any centralized staff planning and resource allocation effort.
Once a license amendment request is filed, the subsequent review and approval schedule is determined primarily by the licensee's Project Manager and the availability of whatever technical support is required to complete the review.
What began as a generic treatment of a generic issue has at this point become a relatively fragmented activity.
Each Program Manager deals with the license amendment request from a
different perspective depending on factors such as his total workload, his personal.
knowledge of the issue, his assessment of what priority he believes the request merits, and his difficulty (real and perceived) in obtaining whatever additional support, (e.g.,
technical review resources) might be required to process the submittal.
An alternative approach for dealing with the
" generic" Technical Specification or license improvements could be readily constructed from the elements identified above.
The first recommended change in the existing process would be that when a draf t Generic Letter offering a specific improvement opportunity is issued for public
- comment, the FRC should also request a
preliminary expression of interest from those licensees who believe that they will act on the proposed improvement.
The draft G.L.
would necessarily contain a
generic safety evaluation which contains reasonable alternative solutions which would envelope expected submittals.
The responding licensees should at the same time be requested to estimate how long it will take to prepare l
their plant specific submittals (e.g.,
confirmation that they are t
bounded by the generic analysis, plant-specific no significant hazard analysis, and Tech. Spec. markup, if different from that proposed in the G.L.
i Page 2 c Wrost jds i
e k
l When the Generic Letter is formally issued, licensees should be offered an opportunity for expedited processing of their 1
amendment request if they submit by a specific deadline specified in the letter.
It is expected that an appropriate deadline can be determined in large measure from the information provided by those licensees who have responded to the comment opportunity for the draft letter.
Licensees would not be prevented from submitting after the specified deadline but it would be understood that " late" submittals would not be included in the generic " batch".
All license amendment requests submitted by the specified deadline should be assigned to an " Issue Manager" who would bear the primary responsibility for planning and implementing the review and approval process, i.e.,
for this group of amendment requests the Issue Manager assumes the responsibilities normally assigned to the individual Project Managers (PMs). Requests after the deadline would still go to the Issue Manager but they probably would be batched for later processing.
The Issue Manager would be expected to take maximum advantage of the expected generic safety analysis and common elements of the submittals and minimize unnecessary duplication of effort.
This would include, as an example, the formulation of the staff's proposed "no significant hazards" determination which would envelope the submittals and be the basis for a single notice for public coraent.
The notice would specifically identify all of the
{
plants which had made timely filings.
At the conclusion of the public notice and comment period, the change would be issued to the plants identified.
Since these types of actions generally fall in the post notice category (i.e.,
the staff analysis has already determined a marginal or non-existent connection to safety),
requests for hearings could be satisfied without delaying the issuance of the change.
It is our belief that nothing in the "Sholly" amendment to Section 189a of the Atomic Energy Act precludes the treatment of more than one plant at a time in the steps required to satisfy the legal requirement of the Act.
We suggest that the staff consider applying the proposed process on a trial basis.
Specifically, we suggest that removal of the requirements for the Post-Accident Sampling System be utilized as the test case.
The NRC has already developed the necessary technical justification in NUREG/CR 4330 and we believe that relief on this issue will afford many licensees significant cost savings with respect to on-going training and maintenance requirements.
Page 3 ew. cost p.
s
~
L YANKEE ATOMIC ELECTRIC COMPANY "E, ";,isifl;"l"
~
Ab 580 Main Street, Bolton, Massachusetts 017401398 h us April 23, 1993 FYC 93-009 SPS93-035 Mr. David L.-Heyer Chief. Rules and Directives Review Branch Office of Administration U.S. Nuclear Regulatory Commission Washington, DC 20555
Dear Mr. Meyer:
These comments on NRC draft generic letter.
- Availability and Adequacy of Design Bases Information." (reference FR 15885. March 24,1993) are-submitted on behalf of Yankee Atomic Electric Company-(Yankee).
Yankee owns the nuclear power plant in Rowe, Massachusetts and Yankee's Nuclear Services Division provides services to other nuclear power plants in the northeast.
Yankee is a member of the Nuclear Management and Resources Council (NUMARC) and we fully endorse NUMARC's comments-concerning this draft generic letter. The NUMARC comment letter notes-that most, if not all of the requested information has already been provided to-the staff through a variety
~
of channels. We can confirm that this is.the case, for the plants with which we are most familiar.
Even more to the point. if the staff's own characterization of its knowledge.of licensees
- design document reconstitution-progrems as reported in SECY-91-364 is f actual, there-can be no reasonable justification for still another blanket survey.
Repetitive solicitations for-this information are clearly wasteful of both NRC and licensee resources and can'T. ave no credible safety benefit.
l We have reviewed in detail the relevant' supporting documentation for this proposal. We can find no evidence to substantiate a generic safety :
. concern hich' would give rise to. issuance of-this draft generic letter. The-w supporting information contains only. vague allusions to design balis.
' documentation maintenance deficiencies at some unspecified number of facilities.. This does not constitute..in our-opinion..the kind of rigorous.
analysis which should wa_rrant endorsement-of this proposal by CRGR.
In fact, g
other staff documents 'such as SECY 92-193 argue strongly to the contrary.
Likewise, we could find no evidence of an intended inspection plan which then would be 'prioritized" using the recuested information, f~
y,i U
omu a
N l
e_
a
e April 23.1993 Mr. David L. Meyer Chief, Rules and Directives Review Sranch Page 2 l
The draf t letter's assertion regarding the voluntary nature of a licensee's response appears to have been included for the sole purpose of mooting a regulatory analysis and backfit evaluation.
Without absolute assurance that there will be no retaliation should a licensee choose not to
" volunteer" information, there clearly is no voluntary aspect to this s oli cit ation.
Both the subsequent *prioritization" of the apparently, yet-to-be devised inspection program, and the more traditional
- adjustment" of SALP scores, serve as strong external compulsions. A voluntary program is characterized by the absence of external influences.
The Federal Register notice promises that *the NRC's final evaluation will include a review of the technical position and, when appropriate, an analysis of the value/ impact on licensees.*
However, nothing provided in the proposed letter nor in the supporting documents can reasonably be characterized as a " technical position." We are confident that a rigorous value/ impact analysis, such as should have accompanied this proposal in the first place, would demonstrate that the proposed generic letter is not j usti fi able.
^51ncereiy yours.
i D. W. Edwards Director, Industry Affairs DWE/dha
~
C76\\317 i
3-3, ENCLDSURE 5
.YA KEE ATOMIC ELECTRIC COMPANY:
"'y"'l's'g>l7;;"l" i
- - - g.
b A"
580 Main Street, Bolton, Massachusetts 01740-1399
[~AN KEE I
f June 21, 1993 FYC 93-015 i
SPS93-061 2q
)
Mr. Samuel J. Chilk 1
Secretary of the Commission U.S. Nuclear Regulatory Commission Wasnington, DC 20550 1
Attention:
Docketing and Services Branch
Subject:
Preposed Rule:
- Modifications to Titness-For-Duty. Program j
Recuirements (5SFR15810)
,I
Dear Mr. Cnilt:
Yankee Atomic Electric Co:rpany (YAEC) appreciates the opportunity to e
- omment on the subject proposed-rule.
YAEC is the owner of the Yankee Nuclear Power Station in Rowe, Massachusetts and provides engineering;end licensing.
services to nuclear power plants in New England.. These comments are filed on l
behalf of Maine Yankee Atomic Power Company and Vermont Yankee Nuclear Power
]
Corporation as well.
?
Secreround end Actual Experience In 1938, the. Regulatory Analysis supporting the then proposed Fitness
.for. Duty (FFD) rule; offered that the reason for initiating such a rulemaking.
was the need to assure that impaired performance of personnel-did.not decrease
... the effectiveness of. the response lto an accident." The analysis was grounded on-the presumption that *... given the pervasiveness of the problem in' our society, it seems reasonable to assume that alcohol' and drug l abuse. as well as other emotional and psychological factors, are also prevalent in the nuclear industry...." (2/9/BS. Regulatory Analysis". FFD rule).
Although.
l perhaps true in an' absolute sense.: data from the testing programs instituted es a result of the regulation (10 CFR 26) have. demonstrated that the assumed deleterious factors are not " prevalent
- in the nuclent industry.
In fact.;the three-year. average rate of random positives is 0.33% (see Enclosure 1).
bbW
- 40 u
tuun
~
+
I Mr. Samuel J. Chilk June 21,1993 Page 2 Indeed, data suggest that the behavior of personnel comprising the nuclear industry should never have been assumed to reflect the trends of the t
Such an assumption.
general population with regard to alcohol and drug abuse.
ignored the array of streening and ongoing monitoring measures to which each individual is subjected. These requirements applied to all nuclear plant workers (contractors as well as utility employees) and necessarily j
differentiated them from the general population. Specifically they included:
t Enckground Investigations (credit, education, court records, references)
Pre-erployment Physical. Drug / Alcohol Streening Psycholtgical Evaluation (which may incluce a clinical assessment)
In:iusion in Behavioral Observation Program Pre-Attess Drug and Aitohol Streening At this point, as the industry :0mpletes three plus years of 100t random testing. industry experience has ma:e it abundantly clear that the performance cojectives of 10 CFR 26.10 have been satisfied:
1 there is reasonable assurance that personnel are not under the influence cf any substen:e or physically impaired, i
reason!ble measures to assure early detections are in plBCe, and j
the elements necessary to achieve the goal of a drug free workplace are also in place.
Based upon the data and relative to the general population, the nuclear The industry reoresents a drug free pcpulation for all practical purposes.
first full year of data showed 0.23 positives for employees and 0.56%
pcsitives for contractors. The third year of testing yielded 0.20t and 0.45%,
respectively.
As a matter of interest, these rates of positives are 10 to 20 times LOWER THAN data presented for the U.S. Navy in the Statement of Considerations accompanying _ the 1938 proposed rule (54FR24474).
Given these results, it is clear that adjustments to the requirements and, particularly, reduction of the random testing rate is absolutely warranted.
NRC committed to this action in 1959 *... based en positive experiente in the industry" (54FR24474).
In a 1991 study, " Fitness for Duty in the Nuclear Power Industry" (NUREG/CR-5754), it was concluded:
... that results [of the study) indicate that the majority of continuous users will have been eliminated after two vears under.
the 100% Standard Approach.
Because licensees will have been testing on a random basis for at least two years prior to any change in testing rates that the Commission may consider, we would expect that all continuous users (except for a relatively few
- wn
( "
9' e
-m s
i Mr. Samuel J. Chilk June 21, 1993 Page 3 coming into the workforce or becoming continuous users after the 4
first two years of the program) will have been detected before the i
testing rates are changed.
If that is indeed the case, there would be little reason to consider efficiencv in the detection of centinuous users when reviewing the merits of alternative random r
testing rates * (p C-19). (emphasis added).
t Conclusions from the 1991 study are supported by the enclosed analysis of actual industry data provided in the Federal Register (5BFR15B12) and NUREG/CR-5758. To highlight these results. Some simple calculations are presented in Enclosure 1.
In our analysis, it is assumed that there are two populations of 1.000 individuals each.
In the first, there are assumed to be 2.3 chronic users (representing employees) and in the second 5.2 (representing contractors).
Application of the Binomial Distribution to these populations is justified because samples are returned to the population to potentially be chosen again in the next sample.
Based on these calculations, several conclusions can be drawn:
1.
Three years of 100% sempling has essentially eliminated chronic users from the two groups. The calculated undetected chronic users are less.
than 0.25 per thousand for contractors and 0.11 per thousand for-employees.
2.
The future sampling rate merely determines the rete at which the residual chronic user is detected. To effectively detect residual threnic users, sampling at 300% to 500% is required. Testing at these prohibitively high rates is neither cost-effective nor warranted.
3.
The predicted positive rate does not compare to the actual values observed. Thus, factors other than_ chronic addiction are involved in' this population. The Feoeral Register notice initiating this rulemaking identified the occasional user and the user of short term detectable drugs as additional elements in the population. This analysis supports their existence.
NUREG/CR-5764 documents the ineffectiveness of 100t or even 300% random sampling to detect these individuals.
Basically. the true value of random testing does not lie in detection r,early as much as it does in the creation of an environment of deterrence.
However, in terms of both detection and deterrente, the most effective tool is l
behavioral observation by supervisory personnel.
As shown by NUREG/CR-575S, the rate of positives for. observed behavior by supervisors is almost 26%.
This positive test result percentage far exceeds that for random testing.
+
1
- m sss
Mr. Sacuel J. Chilk June 21.1993 Page 4 Shortcomines Recardine the Proposal The proposal presented in the subject Federal Register Notice is:a renoom testing rate reduction to 50% for employees.
Random testing for contractors is to be retained at the 100% rate because of the " higher incidents" of positives. Although higher, the absolute value of the positive rate for contractors is very low and only looks "high* because of the extraordinarily low results for utility employees.
In both cases, the residual of the thronic substante abuser af ter three years of 100% sampling is very low and the occasional or event-driven user is not effectively detected by random testing (NUREG/CR-5784).
At issue is the r, erit of the relatively excessive random testing rates (50t and 100%) proposed in light of industry experience to date and the efficacy of random testing itself es a primary means of detection.
It is also in:trient to consider recent conclusions from the General Attounting Office
( G AD) in this regard.
In their recent assessment of random test programs atsinistered by 59 federal agencies, the GAO concluded:
... the percentage of positive test results identified through random drug testing does not very significantly among agencies, regardless of whether the agencies test at a lower level. such as ten percent or a higher level, such as fif ty percent.... [and)
... testing frequencies, whether lower or higher, do not have a cirect impact on the deterrent value of testing.* (GA0/GGD-93-13.
Employee Drug Testing)
This GAD report has indicated that a reduction in random testing rates, where ap;ropriate, will result in significant cost savings without impacting the ceterrent aspect of the program.
A further deficiency in the subject proposti is the two-group testing-program it contains.
Ironically, the NRC, when re-examining their internal rattom sampling rate, rejected just that alternative because "... two distinct NRC random testing groups... result in administrative inefficiencies [and) limit cost saving/ avoidance * (SECY 92-176, *NRC's Drug Testing Program, p. 5).
The NRC has chosen a single reduced random sampling rate for their. Own program, but plans to impose this tumbersome administrative burden with its unnecessary costs on licensees.
Also contained within the subject proposal is a request for comments on the exclusion of *... certain positions critical to safe operation...such es retttor operators
- from the testing rate reduction. The assertion that continued random testing of this group at a 100% rate will be an effective detection mechanism is no more valid in this special case than for the more enan
i
- Mr. Samuel J. Chilk l
June 21. 1993
.Page 5 general ones examined above. The prcposed segmentation is completely.
r unwarranted and could negatively impact the morale, self image and motivation of this group of highly trained and dedicated specialists.
Alternetive Procesal Deterrence is assured through the synergism of ell elements of licensee.
Fitness for Duty Programs (i.e.. pre-access screening, random testing, for-cause testing f ollow-up testing and most importantly, behavioral observation).
It should be clear that the random testing rate is one of the least significant factors in the entire set of deterrent elements and two-
.t I
group random testing is an unnecessary and inappropriate complication.
As an alternative to the proposal presented in the subject Federal Register notice and, in recognition of the information presented herein, we w0uld propose the following:
i 1.
-Establish a sing 1'e random testing rate for all nuclear industry workers.
having unescorted access (employees and contractors).
I 2.
Monitor the test results from this group as a total population - i.e.,
co not base the regulation in any way on individual plant statistics.
i hemely, co not regulate to the lowest common denominator.
t 3.
Lower the required random testing rate to 10%'with the caveat that a l
statistical doubling of the population's-presently very low positive rate in any year from the then current industry average would be grounds l'
for immediate regulatory review. This rate will yield over.15.000
-samples per year which is more than ample to monitor the population.
Given this size sample, and given that 0.33 is the true population i
parameter for drug abusers, the likelihood of obtaining an estimate that is more than double the 0.33% estimate without being detected is essentially zero.
It should be noted that with a' smaller sample size.
the 95% confidence level limits ' change with a variance going from.06%
l to.18%.
These are still very low levels (see-Enclosure 1).
4 Continue to provide training for managers and supervisors in the aree-of i
behavioral observation.
T 1
C76t33)
I
.s Mr. Samuel J. Chilk June 21. 1993 Page 6 The annual cost for program irg}ementation is nearly 569M in 1990 The general breakdown is:I dollars.
(5 in 000's) i.
Industrv
@ Per Unit (3)
Direct Cost of Sampling 29,500 210 RetrainjggandSuitable 33.000 300 Inquiry lost Productivity 6,300 55 Total 65,500 565 If the random sampling is reduced to the appropriate level of 101, the potential annual savings for the industry without decradBtion in the overall orocret is in excess of $30M.
Summerv Steos should be taken, now, to implement a change to the regulation which takes into account the following:
1.
The proposed random testing rate reductions are inconsistent with actual experience - the rate shoulc be reduced to 10% for the entire populaticn of nuclear industry workers having unescorted access.
2.
Tne GAD has concluded from its study of 59 Federal Agencies that the percentage of positive test results does not change significantly whetner testing is conducted at 10% or 50%. They concluded, further, that testing frequency coes not have a direct effect on the deterrent value of testing.
3.
The value of random testing is the degree to which it contributes as a deterrent to drug / alcohol use as part of.the overall Fitness.for. Duty-Program.
II)
Eased upon'a NUMARC' survey of licensees reported to the NRC (T. E. Tipton to B. K. Grimes, September 20 -1991).
(2)
The cost' of retraining and suitable inquiry are not a function of sampling rate and are assumed to continue.
I3)
Average values may not be representative of individual plant costs.
C7is333
6
.a Mr. Samuel J. Chilk June 21. 1993 Page 7 4.
A dual population test program such as has been proposed is unnecessarily burdensome, administratively cumbersome, and is unjustified based on actual experience. Also. a dual population test program is detrimental to morale, teamwork. and overall safe operation.
5.
Conclusions about the behavior of the population must be founded on data representing the entire population and not impacted by statistical variances at individual units. Should individual units not meet industry norms, appropriate action should be taken against them, not the entire nuclear industry.
6.
Perhaps the most powerful element of the Fitness for Duty program is benavioral observation and emphasis in this area will have the most value.
We urge the NRC to ag;ressively pursue this alternative to the changes currer;tly preposed to the Fitness for Duty regulation. The industry has spent tens of millions of collars to implement programs that actuti experience (within the industry itself and throughout the government) has shown to be excessive.
Failure to act expeditiously to fix this situation is contrary to the regulators
- Obligation not to impose any more Durcen on the regulated community than is absolutely necessary.
Sincerely yours.
I r
D. W. Edwards Director. Industry Aff airs DWE/dnm Encicsure i
nsun l
.,,4 q
Analysis ofIndustry Data i
f, '
1.
Positive Test Results (years 90, 91, and 92) h b
EMPLOYEES CONTRACTORS SHORT TERM LONG TERM ALL
! Year sour::
Cua nos 5
I Cau nos '
5 Cua nos 5
Cuu nas 5
l
$63 - l C.56 90491 TR 20.08 497 C.25 44.873 496 C.55 16.4:0 67 0 41 101.283 NL'EI.G 98.(11 199 C.20 5004:
233 0 46 7.f*7 29 0.37 53.119 24 C 45 lF2 l C. 3 135.125 729 c.54 24317 96 0.39 159,412 825 0.52 AU 299.189 (95 Tcna: Tua - 459.101 TM Pcninvu - 1.$21 5 (3 yr rwa 0 C.33 gg w,g,3 95% Confidence Level Limits for 10% and 100% random testing 2.
rates assuming a Current Populaticn Estinate of.33%.
Random
- of testa Lower current Upper Testing per year Confidence Population Ocnfidence-Rate Limit Estimate Limit (3 yr results) f 20%
15,000
.0025
.0033
.0043 i
~
200,000
.0030
.0033
.0036 l200%
i i
Page 1 l
- 4 e
Analysis of Industry Dr.a (continued) 3.
The following example is used to examine the effectiveness of randon testing.
Assumptions:
2 Populations of 1000 each representing a closed system Pop #1 - 2.3 Chronic Users Pop #2 - 5.2 Chronic Users 100% sample size for 3 years 10% sa:ple size thereafter Pop d2 - E plevees Pop #2 - Contractors Year P (not Expected User P (not Expected User detected)
Re=aining detected)
Remaining t
r1
.3677 0.846 1.912 Same as o
!2
.1352 0.311 fee pep p1 0.703
!3
.0497 0.114 0.258
.1 14 at 2004
.01E3 0.0420 0.0952 4 at 504
.0301 0.0692 0.156 4 a.t 104
.0449 0.103 0.233 10 at 104 0.0246 0.0319 0.128 4.
The binonial distribution is the appropriate one to use for sampling from a population with return of the sample to the population (i.e.,
it could be sampled again)
X A
N!
P Q P (not detected)
=
XI (N-X)
Where:
P = 1/N Q = 1-P N = Total # picks X = # times picked The specific solution of this general expression for a determination of the probability of not being detected is:
A or Q" P (not detected)
(1000)
Q
=
Ol[0-Op 1
Where: N, = y N; f or n sa=ples Page 2 i
!a
... 4 Analysis of Industry Data (continued) 5.
Considering Example Population 1
- Third year actual rate is 0.00276 Positives / Test Third year predicted rate is 0.000114 Positives / Test
- Actual experience is that the rate of positives is 0.00265 nore than predicted or 2.65 positives /1000
- This demonstrates that other factors are involved which initiate substance abuse in the population.
In other words, the problem appears to be the occasional or event driven user rather than the chronic user.
- Randen testing is not ef f ective in detecting this type of user.
It is effective only to the' degree that randon testing can support a different mechanism for user detection, or
- better, reinforcement of avoidance behavior.
- Similar observations apply to example Pcpulation 2 with equal validity
- There is no fundanental difference in the behavior of either exanple population with regard to chronic users.
Page 3 i