EP A-520/ 3-76-009
REACTOR SAFETY STUDY (WASH-14OO)
A REVIEW OF THE FINAL REPORT
LT S. ENVIRONMENTAL PROTECTION AGENCY
Office of Radiation Programs
m
-------
EPA-520/3-76-009
REACTOR SAFETY STUDY (WASH-1400): A REVIEW OF THE FINAL REPORT
JI'HF 1976
U.S. ENVIRONMENTAL PROTECTION AGENCY
OFFICE OF RADIATION PROGRAMS
WASHINGTON, D.C. 20460
-------
FOREWORD
The Environmental Protection Agency (EPA) considers the Reactor
Safety Study (WASH-1400) to be a critical document relative to the
potential environmental and public health Impact of nuclear power.
EPA's Office of Radiation Programs has committed significant
resources to provide 1n-depth review of the draft and final version
of that study. Our comments on the draft statement were made in two
stages — Initial comments by our staff on November 27, 1974, and
final comments on August 15, 1975. The final EPA conrnents Included a
detailed report from our contractor, Intermountain Technologies Inc.
(ITI).
On October 30, 1975, the Nuclear Regulatory Commission (NRC)
released the final version of WASH-1400 (NUREG-75/014). We have
reviewed the final Reactor Safety Study Report to determine the
extent that our comments on the draft report were resolved and to
provide a technical evaluation of any new material or significant
revisions to the draft report. The analysis of the consequences
of reactor accidents, Appendix VI, had been completely revised
and thus was material that had not been previously subject to review;
it received particular attention.
We present In this report the results of our review of final
WASH-1400, together with ITI's report of its review, so that they
will be available as a resource to the scientific community
and the general public. Because we consider the Reactor Safety Study
to be of great value 1n the development of reactor safety and in the
development of methodology for assessment of risk, the review comments
published 1n this report are Intended to provide constructive
criticism which may be helpful to the NRC and others who may undertake
further work In risk assessment.
We welcome comment on this report and would appreciate receiving
any corrections or critical comments on the information and conclusions
presented. Please send any such comments to the Environmental Protection
Agency, Office of Radiation Programs (AW-459), Washington, D.C. 20460.
W. D. Rowe, Ph.D.
Deputy Assistant Administrator
for Radiation Programs
-------
CONTENTS
Page
SECTION 1. SUMMARY AND CONCLUSIONS 1-1
INTRODUCTION 1-1
GENERAL CONCLUSIONS 1-2
RESULTS OF EPA REVIEW 1-4
RECOMMENDATIONS 1-7
SECTION 2. HF*LTH EFFECTS 2-1
A. Categorization.of Upper, Central and Lower Bound 2-2
B. Dose Rate Effectiveness Factors 2-5
C. Mitigation of Effects by Medical Treatment 2-7
D. Radiation Bioeffects of the Thyroid Gland 2-7
E. Risk Estimates for Genetic Disorders 2-11
F. Radiation Dose Causing Specific Clinical Effect 2-13
G. Other Biological Consequences 2-13
H. Underestimates of Risk 2-13
I. References 2-16
SECTION 3. OVERALL REVIEW 3-1
A. Comments on the Consequence Calculation Description 3-1
B. Corrections or Information Additions Needed (Appendix VI) 3-5
C. Actions to Mitigate Radiation Exposures 3-9
D. Corrections or Information Additions Needed (other than 3-14
1n Appendix VI)
E. Review of Response to EPA Comments on Draft WASH-1400 3-21
F. References 3-22
SECTION 4. REPORT BY INTERMOUNTAIN TECHNOLOGIES INC. 4-0
-------
SUMMARY AND CONCLUSIONS
INTRODUCTION
The U.S. Environmental Protection Agency (EPA) has always
considered as part of its concerns the risk of accidents that may lead
to contamination of the environment and subsequent injury to the
health of the public. In our reviews of the environmental impact
statements for light-water reactors (LWR's), which have been conducted
since 1971, we have emphasized the need for a thorough evaluation of
the environmental risk, including risks from accidents, associated
with LWR technology. In August 1974, the U.S. Atomic Energy
Commission (AEC) published for comment a draft report entitled
"Reactor Safety Study: An Assessment of Accident Risks in U.S.
Commercial Nuclear Power Plants," WASH-1400. This report presented
the product of a major.study, in total spanning three years and
costing four million dollars, which was directed by Professor Norman
C. Rasmussen of the Massachusetts Institute of Technology. The
Reactor Safety Study is the first comprehensive study of reactor
accident risks to utilize a systems analysis approach in order to
quantify the accident risks in terms of probabilities and
consequences, where historical and empirical data are inadequate.
Because the Reactor Safety Study (the Study) may be considered the
most definitive assessment to date of the risk from accidents, it is
imperative that the Reactor Safety Study report be reviewed in depth
and as impartially as possible so that the validity of the Study's
methodology and results may be determined.
The Office of Radiation Programs of EPA initially conducted a
comprehensive review of the draft report including the environmental
consequence models, the general study methodologies and the
conclusions, and issued formal comments to the Atomic Energy
Commission on November 27, 1974. EPA obtained additional technical
support for the review effort through a contract with Inter-mountain
Technologies Inc. (ITI) of Idaho Falls, Idaho. The ITI effort was
directed at examining the Study's evaluations of accidents and their
event sequences to determine whether any significant failures of
systems or equipment had been omitted or any major error or system
biases of data analysis had been incorporated. For any area
identified with possibly significant errors or omissions, the impact
on the risks of the necessary adjustments in the variables or
sequences of events was evaluated by ITI. The results of ITI's in-
depth review and EPA's further comments were sent on August 15, 1975.
to the U.S. Nuclear Regulatory Commission (NRC), which had assumed
sponsorship of the Reactor Safety Study upon dissolution of the AEC.
In October 1975, EPA published a report entitled "Reactor Safety Study
(WASH-1400): A Review of the Draft Report," EPA-520/3-75-012,
consisting of reprints of (1) EPA's conments of November 27, 1974,
1-1
-------
(2) EPA's comments of August 15, 1975. and (3) the ITI report of Its
review.
On October 30, 1975, the NRC released the final version of MASH-
1400 (NUREG-75/014). The principal changes 1n the final version from
the draft are a complete revision of the consequence calculations In
Appendix VI, the addition of an Addendum to the Main Report providing
further explanation of the event-tree, fault-tree methodology, and the
addition of Appendix XI providing responses to comments on the draft
report.
As Indicated In the Foreword to our report "Reactor Safety Study
(WASH-1400): A Review of the Draft Report," we have undertaken a
review of the final version of WASH-1400 using again the assistance of
ITI. We have reviewed the final Reactor Safety Study report to
determine the extent that our comments on the draft report were
resolved and to provide a technical evaluation of any new material or
significant revisions to the draft report. The analysis of the
consequences of reactor accidents, Appendix VI of the Study, had been
completely revised and thus was material that had not been previously
subject to review; It received particular attention.
We consider the Reactor Safety Study a major step forward In
understanding and estimating the risks from nuclear power plants. EPA
supports the Study, Its concept, and the need for continuing the
effort to eventually arrive at a reasonable consensus of the level of
risk associated with reactors. EPA's comments are primarily aimed at
Improving the quantitative risk estimates, and thereby Increasing
their value as an Input to the risk assessment process.
EPA's review of the final Reactor Safety Study has been
accomplished In two segments: the EPA staff has reviewed the new
accident consequences section and the health effects models in Appendix
VI of the Study, and the responses to previous comments; and ITI has
reviewed the responses 1n the final report to their previous comments.
GENERAL CONCLUSIONS
In reviewing the Reactor Safety Study, we have tried to keep in
perspective the original intent of the Study - to provide a
technically sound overall assessment of accident risks from commercial
nuclear power plants in the United States. We have also taken note of
the limits to the scope of the Study and limitations which the report
attaches to the applicability of the Study's results and methodology.
With respect to the intent to provide a technically sound overall
assessment of the accident risks, we have identified several
significant areas in which we have found the WASH-1400 report either
deficient or containing unjustified assumptions. These are
1-2
-------
(1) failure to address fully the health effects expected after an
accident and to consider adequately a technical basis, which includes
a broad range of perspectives,for estimating the incidence of the
associated bioeffects, (2) the assumptions made in regard to evacuation
as a remedial measure, (3) improperly or incompletely evaluated
parameters used in determining accident event-sequences and probabilities,
and (4) inadequate description of the analysis of the consequences
of the release of radioactive materials to the environment.
It would be desirable 1n any review of a document such as the
Reactor Safety Study to evaluate the sensitivity of the assessment to
the various assumptions made In determining the estimates of
population impact. Because the numbers of Individuals exposed at the
various doses of Interest are not specified In the Study, EPA was not
able to evaluate the assessment In terms of the range of
uncertainties. We have found, however, that what we consider more
reasonable assumptions in health effects, emergency actions, and
estimates of probabilities of releases, would cause modifications to
the overall risk analysis.
It appears that If late somatic health effects were adjusted 1n
accordance with EPA's assessment of the numerical health risks, the
estimates would Increase by a factor of from 2 to 10. The potential
change in the estimated early fatalities and Injuries could not be
determined by EPA from the Information provided.
There are two deficiencies In the assumptions for evacuation as a
protective action. The first Involves the use of a constant 25 mile
evacuation sector for all core melt accidents. This 1s at variance
with present and planned practice and 1n some cases overestimates the
risk and the need for evacuation, and in other cases underestimates
the risk when larger groups may have to be evacuated.
The second set of deficient assumptions Involves the amount of
time that persons would be exposed prior to and during evacuation and
the evacuation speeds. Because the details of handling of these
parameters are lost In the description of the Study's modeling, it 1s
difficult to assign quantitative values for the range of error In
consequence estimates. The report Itself, however. Indicates that the
early deaths and Injuries would not be more than seven times as great
If the evacuation were completely Ineffective.
The effect of using other values of certain parameters in
determining the outcome of specific accident sequences and their
probabilities 1s also difficult to assess. An upper bound may be
estimated for BWR reactor protection system failure using tightly
coupled common mode failure; this dominant effect would apparently
raise the overall risk (for the 100 reactors) by no more than a factor
of 10. Bounding estimates on changes in other parameters, e.q,, the
1-3
-------
PUR containment failure pressure, Indicate that Individually they
could not change the overall risk by more than about a factor of 3.
It Is not possible to combine these uncertainties directly to
obtain an overall corrective risk value because they address different
aspects of probabilities and consequences and because the most likely
values within these ranges cannot be derived from the limited
Information In the Reactor Safety Study report. However, we believe
the Study has understated the risk based on underestimated health
effects, evacuation doses, and probabilities of releases. The range
1s believed to be between a value of one and a value'of several
hundred.
The results of our review of the final report have not altered our
opinion that the Reactor Safety Study provides a major advance 1n risk
assessment of nuclear power reactors, and that the Study's general
methodology provides a systematized basis for obtaining useful
assessments of the accident risks where empirical or historical data
are presently unavailable. In any effort to determine the
acceptability of nuclear power 1n Its present form, consideration
should be given to the results of the Reactor Safety Study only as
suitably modified to correct for the underestimation of health effects
and other deficiencies noted. Such a determination should consider
that the risk of severe accidents, such as described 1n the Reactor
Safety Study, 1s only a portion of the overall risk and that the rest
of the fuel cycle, sabotage, and lesser accidents are also among the
contributors to the risk. We emphasize that the risk comparisons made
In the Reactor Safety Study are only one step toward a determination
of risk acceptability of nuclear power.
RESULTS OF EPA REVIEW
1. The most significant disagreement 1s that associated
with the health effects model In Appendix VI. We
particularly disagree with the WASH-1400 conversion of
radiation exposures to health effects, which tends toward
underestimation of the health effects. Such risk estimates
are highly uncertain on the basis of present knowledge. We
consider the adequacy with which these uncertainties are
addressed central to providing an Informative estimate of
reactor safety. In the environmental Impact statements
developed by the Nuclear Regulatory Commission (NRC),
estimates of radiation risk have been based on the
recommendations prepared for the Federal Radiation Council
(whose function now resides In EPA) by the National Academy
of Sciences1 BEIR Committee. The BEIR Committee's
recommendations were prepared 1n response to a request of
Congress that the National Academy of Sciences recommend
numerical risk estimates for Federal use. These
1-4
-------
recommendations were not followed by the NRC 1n the
development of the Reactor Safety Study. Instead, an ad hoc
advisory group was specifically established to provide risk
estimates for the final report, the BEIR report having been
used 1n preparation of the draft. While this ad hoc group
developed many Innovative and original views on assessing
radiation risk, 1t 1s this Agency's finding that based upon
current Interpretation of accepted data many of the risk
assumptions made In the Reactor Safety Study are premature
and that the advice of the National Academy of Sciences was
not used judiciously.
2. In examining the discussion In WASH-1400 of actions to
mitigate radiation exposures, we find that although we do not
agree with the selection of a few parameters and concepts,
the general approach appears to be suitable, developed as 1t
was In advance of EPA's Issuance of relevant guidance. It
appears, however, that Incomplete consideration had been
given to requirements that may Indicate the use of evacuation
as a protective action. Pending the Issuance of EPA guidance
dealing with protection from exposure to contaminated media,
the reliance by the Reactor Safety Study on recommendations
of the Federal Radiation Council and the Medical Research
Council of Great Britain regarding long-term exposure dose
criteria for external radiation and ingestion via milk and
other pathways also appears to be reasonable. He concur with
the statement 1n WASH-1400 that there 1s need for further
work In this area.
3. The report does not present sufficient Information
Illustrating the variations in consequences and risk to
nearby population groups caused by differences in site-
specific circumstances, which may be masked by the
restriction of the analyses of the six composite sites.
4. EPA commented on a number of areas that appeared to be
deficient In the draft report. The final WASH-1400 provides
satisfactory responses to a number of those comments, but not
to all. Of those previously-noted deficiencies which remain
in the final report, the ones for which an evaluation was
made are judged to have no significant effect Individually in
the major conclusions of the Study. By significant, we mean
the possibility of changing the overall estimate of risk 1n
WASH-1400 by as much as a factor of ten. However, the
deficiencies do tend to reflect on the care with which the
draft and final reports were prepared and therefore they are
worthy of mention.
1-5
-------
5. A significant deficiency found In the consequence
calculations was that the presentation lacks a description of
the specific analytical framework applied, I.e., a
description of the way the parts of the consequence
calculations are put together. The final WASH-1400 needs a
clear, concise description of the calculatlonal process,
which clearly distinguishes 1t from all the supportlnq
arguments, rationale, and background material. In addition,
Intermediate values from the calculatlonal steps for specific
cases should be provided. We find that Insufficient
Information has been provided for others who may wish to
utilize the developed techniques or even to trace through
them. This Information would facilitate an Improved and fair
assessment of the results and their uncertainties and would
allow the scientific community to utilize and Improve on the
developed techniques.
6. The final HASH-1400 provided additional Information and
clarifications 1n response to comments by our contractor on
the draft report. This resulted 1n many significant
Improvements 1n the report. ITI drew the following
conclusions from their review of the final WASH-1400; we have
examined them and their bases and we concur In these
conclusions:
(a) The final version of WASH-1400 has been Improved
compared to the draft version. The conclusions from ITI's
review of the draft version are, however, judged to be
applicable, with minor modifications, to the final version.
(b) The majority of the errors, omissions, Inconsistencies
and debatable assumptions In the final WASH-1400 do not have
a significant Impact on the overall risk assessments. (ITI
did not address the health effects models.)
(c) The summary presentations in the final WASH-1400
comparing nuclear power risks with other man-made risks are
sometimes misleading and Incomplete. This tends to undermine
the strength of the report's conclusions. Deficiencies in
the presentation Include the failure to Illustrate
calculatlonal uncertainties in reactor accident risks, the
comparison of calculated reactor accident risks with other
types of risks based on actuarial data without stressing the
differences, and the obscuring of latent deaths in reactor
accident risks.
(d) The risk assessment for the boiling water reactor
transient without scram appears to be the most significant
1-6
-------
analysis problem found 1n the final (and draft) WASH-1400,
which may underestimate the risks.
(e) The final WASH-1400 appears to have Improperly or
Incompletely evaluated human reliability, PVJR small break
analyses, common mode failures and some aspects of design
adequacy. Further, there 1s Insufficient Information In the
report to quantify the Impacts of these deficiencies on risk
evaluations.
(f) The validity of using the Surry reactor as a basis for
calculating the risks representative of all PWR's requires
additional justification, since only about 20 percent of all
PWR's scheduled for operation in 1980 are similar to the
Surry design.
(g) The assessment of the functionalllty of the Emergency
Core Cooling System (ECCS) appears incomplete and some of the
stated conclusions may be misleading.
(h) The assumed pressure which leads to failure of the PVIR
containment appears high and the basis provided for its
selection was found to be deficient.
(1) Of the 45 deficiencies identified in ITI's comments on
the draft WASH-1400. 34 remain in the final report, with 29
having the potential of altering the calculated risks.
Sixteen of these 29 are judged to have an insiqnificant
Impact on the results. One deficient area (BMR transient
without scram) may have a significant impact. The potential
of the remaining 12 for altering the risks was undetermined.
RECOMMENDATIONS
1. As a supplement to the Reactor Safety Study report,
further estimates should be made of the health effects from
accidents using different upper bound, central and lower
bound risk models.
The Agency suggests that (a) lower bound estimates
should be based on absolute risk, 30 year plateau, an-l dose
rate effectiveness factors not smaller than 0.5; (b) central
estimates should utilize an average of the absolute and
relative risk estimates stated 1n the BEIR report by :he
National Academy of Sciences; (c) while upper hound estimates
may require more speculation about additive and synenpstlc
health hazards than 1s warranted at this time, some
consideration of the sensitivity of the estimates to the use
of a convex relationship appears justified, or,
1-7
-------
alternatively, the relative risk estimates In the HEIR report
with a life risk plateau could be used as a surrogate.
2. The Reactor Safety Study efforts should be continued In
order (a) to provide verification of their applicability to a
broad spectrum of light-water reactors, Including those
beyond the first 100 (considered 1n WASH-1400); (b) update
the risk analyses as more operational Information and
Improved analytical techniques are developed; and (c) to
expand the applicability of the WASH-1400 techniques to the
process of licensing nuclear reactors.
3. As Incidents occur 1n the nuclear power Industry, their
significance relative to reactor safety should be evaluated
and placed Into meaningful perspective.
4. As recommended In the Study, the methodology of WASH-
1400 should be extended to evaluation of Floating Nuclear
Power Plants, LMFBR's, LWBR's and HTGR's as design and
operational data become available in sufficient detail to
make the effort worthwhile.
5. Further efforts 1n this area should assess the
variations Introduced by site-specific circumstances so as to
provide additional Insight Into the variability of th«j
consequences and the risk to nearby population groups.
6. A supplemental report or addendum to WASH-1400 siould be
prepared to detail the analytical techniques, including
Intermediate steps, some examples, and the computer programs
used to calculate the consequences given in Appendix -/I.
1-8
-------
SECTION 2. HEALTH EFFECTS
The number of expected health effects 1s the bottom line 1n any
assessment of reactor safety. Such risk estimates are highly
uncertain on the basis of present knowledge and the adequacy with
which these uncertainties are addressed Is central to providing an
Informative estimate of reactor safety. In the environmental Impact
statements developed by the Nuclear Regulatory Commission (NRC),
estimates of radiation risk have been based on the recommendations
prepared for the Federal Radiation Council (whose function now resides
1n EPA) by the National Academy of Sciences1 (NAS) BEIR Committee.
The BEIR Committee prepared the recommendations at the behest of
Congress that the MAS recommend numerical risk estimates for Federal
use. These recommendations were not followed by the NRC in the
development of the Reactor Safety Study. Instead, an ad hoc advisory
group was specifically established to provide risk estimates for the
final report, the BEIR report having been used 1n preparation of the
draft WASH-1400. While this ad hoc group developed many Innovative
and original views on assessing radiation risk, 1t is this Agency's
finding that many of the risk assumptions made 1n the Reactor Safety
Study are not generally accepted as yet within the scientific
community, and that the advice of the NAS was not used judiciously.*
An objective of the Reactor Safety Study (the Study) was to make a
realistic assessment of the health risks associated with potential
reactor accidents and to Indicate the uncertainties in such
assessments. In some cases, the Study failed to address fully the
health effects expected after an accident and to consider adequately a
technical basis for estimating the incidence of the associated
bioeffects,which Includes a broad range of perspectives. Therefore,
the Study did not provide an accurate and complete understanding of
the health risks associated with a reactor accident. At the least,
the Study should have provided a better analysis of the following
problems:
• proper categorization of upper, central and lower bound somatic
health risk estimates,
• justification of the selected dose rate effectiveness factors
(DREF) for chronic and acute bioeffects,
•£PA has recognized the continuing need for the National Acadei/ cf Sderces to update their review cf
radiation health e'fects. EP^-contrz"ed rersr'.s en Plutarlu- Hct Particle -ea't11 Effect: ard
Bereflts cf P.adlat on are to be completed rver.tarily !>./ 'AS. A co-tract, -"or an NAS review ef
Information of health effects since 1972. including dose-rate effects, has Mready been -.-.itiated by EPA.
2-1
-------
• minimal, supportive and heroic treatment for the mitigation of
acute effects as hypothesized In the report,
• radiation bloeffects of the thyroid gland,
• risk estimates for genetic disorders,
•estimates of radiation doses that result 1n specific clinical
health effects.
These defects 1n the report are discussed more specifically below.
A. Categorization of_ Upper, Central and Lower Bound Risk Estimates
The Reactor Safety Study categorized the long term health risk from
cancers Into three categories: (1) an upper bound estimate for which
the linear, non-threshold dose-effect model was assumed; (2) a central
bound estimate where the expected cancer deaths, based on the linear
non-threshold model, are reduced via a set of DREF, supposedly in
order to account for differences 1n health hazards following exposure
to low LET radiation at low doses and low dose rates; (3) a lower
bound estimate where a threshold dose for radlocarclnogenesls 1s
assumed.
While EPA does not find consideration of upper and lower bound
risk estimates Inappropriate for a study of this type, we believe a
more balanced consideration of the models used to describe each risk
category 1s appropriate. Contrary to the assumption 1n the report,
the non-threshold linear model does not lead to the most conservative
estimate of radiation risks.
A more accurate categorization of upper, central and lower bound
risk estimates would be to define them 1n terms that truly reflect the
several types of dose-response concepts that have been considered by
the scientific community, namely: convex upwards-response1"5, linear
response6'11, and concave upwards responses1*"22. Since the Reactor Safety
Study did not use the proper definition for each category, its
estimates of health risks are biased.
1. To establish an upper bound estimate of health effects, It
would have been more realistic to predict cancer Incidence on the
basis of a convex upwards dose response rather than a linear
response. Several Investigators, Including Baum1*2, Brown3,
Gibson, et al.23, Mole20, Morgan5, Casarett2!* and Lappenbusch25
have suggested that given certain defined conditions, a convex
dose response may occur. Baum ^reviewed and analyzed the
radiation dose-tumor (cancer) Incidence 1n mice, rats and man and
showed that carclnogenesls can be represented by a simple power
function of dose via the form E = CD where E = effect, C =
2-2
-------
constant, D = radiation dose and n = constant. Baum Indicated
that the exponent (n) 1s frequently less than 1.0 (0.19 for lung
cancer, 0.35 for stomach cancer and 0.5 for female breast cancer)
signifying that the dose-effect curve 1s convex upwards,
particularly at low doses. Baum's explanation is based on the
arguments that (1) humanity 1s a heterogenous population with
mixed predisposition to cancer due to genetic differences and
differential sensitivity to multiple stressors and (2) at low
doses 1n a heterogenous population, a fraction of the people
respond to one-target kinetics and, thus, cause a steeper slope
In the dose response curve at low doses as opposed to a reduced
slope at the higher doses where the less sensitive fraction of the
population requires additional Injury to show effects.
As stated 1n the Office of Radiation Programs' "Policy
Statement on Relationship Between Radiation Dose and Effect,"
March 10, 1975, EPA believes that the linear non-threshold model
as developed In the BEIR Report26 provides the best estimates of
human health risk. However, this model may not be conservative at
low doses. EPA recognizes2* that much of the available data for
the risk coefficients given 1n the NAS-BEIR Report are based on
high doses where cell killing may have perturbed the carcinogenic
dose response. This has been discussed by Baum2, Mole1* and
Morgan5. A convex dose response relationship for carclnogenesls
should have been considered 1n the Reactor Safety Study, since It
appears to be as scientifically tenable as the threshold
hypothesis used In the Study for lower bound estimates.
2. The National Academy of Sciences BEIR Committee recommended
the use of the linear non-threshold model for radlocarclnogenesls
to the Federal Radiation Council as a best estimate of radiation
risk, and 1n view of recent data6"11 Indicating its applicability at
low doses, It Is surprising that 1t was not used as the central
estimate In the Reactor Safety Study. No explanation Is given
In the Study for electing not to use the BEIR Committee relative
risk estimates. Such a central estimate of risk should have
weighted equally absolute and relative risk estimates or, as the
BEIR Committee did, list these two methods as upper and lower
bounds on the "central estimate."
The Reactor Safety Study neglected recent Information in
support of the linear non-threshold hypothesis:
(a) Modan, et al.6 recently completed a retrospective study
of 10,902 chTTdren who received a thyroid dose of
approximately 6.5 rads during x-ray treatment for tinea
capitis. Approximately 17.5 years thereafter, Modan found
that the Irradiated group had significantly higher absolute
risk estimates for malignant and benign head and neck tumors,
2-3
-------
especially of the brain, parotid and thyroid gland, than did
the controls. Calculating the absolute risk estimate for
thyroid tumors 1n Modan's observation (8.1 excess cases/106
/yr/rad) and comparing with the BEIR Report (2-9 excess
cases/106/yr/rem)t a linear, non-threshold relationship seems
most appropriate for the low LET radiation applied. It
should be noted that rather than considering these new data
to Indicate linearity at low doses, the authors of the
Reactor Safety Study preferred to set thyroid cancer 1n a
special category of Its own. Apparently, to do otherwise
would have negated use of a threshold carclnogenesls dose In
the "lower.bound" estimate (see below).
(b) Stewart and Kneale9 conducted an epidem1olog1ca1 study
on children dying from cancer within ten years after birth
who had experienced In-utero exposure to x-rays during
obstetric Investigations. Recent studies by Halford25 and
Newcombe and McGregor 29 Indicates that for these 1n-utero
exposures the leukemia rate was linear with doses as low as
0.25 - 0.5 rad. Mole 30 has shown that Stewart's results were
not biased by a predisposition to leukemia 1n the population
studied.
(c) Animal studies also Indicate linearity for
carclnogenesls at low doses. Shellabarger et al.8 subjected
40 day-old female Sprague-Dawley rats to 60 Co gamma rays at 40
R/m1n where exposures ranged from 15.6 - 250 R and found, as
1n prior studies, an apparent linear relationship, for both
acute and fractionated schemes for the Incidence of mammary
cancer. Borek and Hall10'11 , working with golden hamster embryo
cells, found that neoplastlc cell transformation shows a
convex dose response relationship below 1-10 rads, a linear
response above 10 rads and a response reflecting reproductive
cell death after 300 rads or more.
3. The Reactor Safety Study should have Included 1n the lower
bound category all of the concave dose-effect models, namely:
slgmoldal, quasl-threshold, quadratic and dose-squared. Such
response concave models could have taken Into account the proposed
DREF suggested In the Study and possibly a threshold dose if'
repair processes can possibly justify Its existence. Caution
should have been taken, however, not to confuse and translate well
known radiation Injury studies where cellular, organ depletion and
survival studies demonstrate clearly that biological repair
occurs, to the case of radiation carclnogenesls, because of the
lack of knowledge of whether the same mechanisms apply.
2-4
-------
B. Dose Rate Effectiveness Factors
The Reactor Safety Study does not provide an adequate scientific
basis for the DREF used, I.e., 0.2 for somatic bloeffects (cancers)
and 0.5 for acute Injury (radiation Injury to bone marrow). While the
Reactor Safety Study uses the DREF to obtain a five-fold reduction In
the estimated number of cancers due to an accident, 1t would appear
that several factors Indicate this choice 1s not prudent. These
factors are examined below.
1. The Reactor Safety Study leaned heavily on a study by Mays,
Lloyd and Marshall 17 to obtain the numerical values used. An
examination of the references given by Mays, Lloyd and Marshall17
shows that the DREF values could range from approximately 0.1 to
greater than 1.0. Furthermore, fractlonatlon studies, such as
those considered by Mays et al.17 have been shown to be greatly
Influenced by radiation conditioning31 , and that reducing the dose
rate results 1n wide variations 1n recovery capabilities among
animals32 . Finally, Mays et al.17 failed to differentiate between
results obtained from chronic Irradiation at low dose rates (the
case of Interest here) and studies where a fractionated dose was
delivered at high dose rates Intermittently over a relatively
long test period. For example, Hays et al.17 characterize dose
rates delivered at 8.5 R/m1n by Anderson and Rosenblatt33 as 0.006
to 0.06 R/m1n depending on the period of time over which the
several fractional exposures were given. In the discussion below,
this will be referred to as the average fractionated dose rate
(AFDR) rather than dose rate.
2. Mays et al.17 Indicate a chronic DREF of 0.0.'', for life
shortening 1n beagles based on an experiment hy Anderson and
Rosenblatt33 . However, 1n reviewing the data by Anderson and
Rosenblatt, it 1s evident that the data Indicate that 1n these
experiments life shortening 1s influenced only by total dose and
not dose rate, as shown In Figure 1. The data reproduced in
Figure 1 Indicates that 300 R is more effective than 100 R for a
given dose rate and that mortality following 100 R is greater than
that In controls for animals surviving longer than nine years
post-exposure. However, the mortality patterns are not dose rate
dependent, but rather show a DREF of 1.0 since there was no
significant difference In percentage cumulative survival of dogs
at any tine 1n the period 0-14 years post-exposure, whether they
experienced an average fractionated dose rate of 0.06 R/m1n or 8.5
R/m1n. Using Mays' method of analysis, it Is possible to use
these same data to show a DREF greater than 1.0; for example, a
DREF of 3.0 could be obtained by comparing animals exposed to 300
R at 8.5 tymln to those given 100 R at an AFDR of 0.0008 R/m1n.
Obviously, this is not a valid number, but rather shows the need
for care in Interpreting such data.
2-5
-------
3. It should be pointed out that In all of the studies cited by
Nays et al.17>where a relatively long life span animal was
Involved, the results of their analysis are rather unconvincing.
For example, a calculated DREF of 0.08 was obtained by Mays et
al.n using data reported by Casarett and Eddy3**. These animals
(dogs) were exposed at 0.06, 0.12 or 0.60 R/day. Yet, the average
age at death for each test group was essentially unchanged (12-13)
years, Indicating a DREF of 1.0. However, using Mays' formula,
DREF values ranging from 4.4 to 8.9 could be calculated.
4. The rodent data cited by Mays et al.17 is somewhat more
convincing. Using data by Shellabarger and Brown35 , Mays et al.
calculated a DREF of 0.68. However, upon plotting percentage
mammary cancer 1n rats following exposures at 0.03 R and 10 R/m1n
versus total radiation dose, It becomes obvious that 1n this study
DREF values range from 0.23-1.0, depending upon which dose levels
are compared.
Mays et al.17 also referenced papers by Mole20 ; Grahn, Fry and
Lea21 ; Upton, Randolph and Conklln36 ; and Donlach37 . These papers
provide chronic DREF values of approximately 0.14; 0.19; 0.07;
0.45; 0.14; 0.26; and 0.1 respectively. While these DREF values
do support Mays et al.17 and their suggestion that animal
experiments Indicate that In some cases radlocarclnogenesis 1s
less at low dose rates, the applicability of these data to humans
1s very tenuous. The paper by Mole20 1s a case 1n point. He shows
that the length of time over which the fractionated doses are
delivered 1s an Important factor 1n determining the resultant
carclnogenlclty and that long exposure periods can lead to higher
cancer Incidence. Obviously, only limited conclusions that are
applicable to radiocarcinogenesis can be gained from studies on
inbred mouse strains. EPA is aware that research in this area 1s
very active at present and that ongoing studies may result in a
much better foundation for the use of DREF risk evaluations.
However, their use in the Reactor Safety Study would appear to be,
at the least, premature.
5. The applicability of a dose rate effectiveness factor to
acute effects due to radiation is better established than for
cancer where such studies are just beginning. However, the
application of the 0.5 DREF for acute injury is not well
documented in the Reactor Safety Study. Clinical evidence showing
less acute effects for protracted (fractionated) exposures is
conflicting31. When fractionated exposure patterns are used, the
first dose in the fractionation may well condition the response to
the succpeding dose fractions. For example, Alnsworth et al.32
have shown that the LDso/eo for sheep at 4 R/min varied from 144 to 540
rads depending on the length of time following an earlier
conditioning dose.
2-6
-------
C. Mitigation of Effects by_ Medical Treatment
Death following acute exposure may be underestimated in the
Reactor Safety Study due to Incorrect assumptions concerning
availability of specialized medical treatment for acute radiation
Injury. In the report, lethal doses to the population near an
accident are estimated as 340, 510 or 1,050 rads depending upon
whether or not minimal, supportive or heroic medical treatment Is
provided. It appears that In the case of a severe accident some 5,000
people would receive a dose of 350-550 rads and, thus, need heroic as
opposed to minimal or supportive treatment. Since there are perhaps
only 12 hospitals 1n the U.S. which could provide heroic treatment to,
at most, 50-150 people, this leaves about 4,900 people without such
treatment. The heroic treatment scenario would appear to be
Impractical. Its use Implies that reactor safety assurance 1s
dependent upon the surrounding population's acceptance of such
treatment and the availability of medical care to a degree that Is
unrealistic.
D. Radiation Bloeffects of the Thyroid Gland
Thyroid Injury 1s an Important aspect of a reactor accident. The
associated risks may have been underestimated In the Reactor Safety
Study. The stated thresholds for radiation thyroldltls (25,000 rem)
and hypothyroldlsm post-exposure (20 rem) may be high by factors of
2.5 and 2, respectively.
1. In Section H2 of Appendix VI of the Study, concerned with
acute effects of Ionizing radiation on the thyroid, 1t was
reported that Belerwaltes and Wagner38 had found in 1956 that
radiation thyroldltls seemed to occur 3-7 days post-ingestion of
rad1o1od1ne In 4-5 percent of thyrotoxicosis patients.
Unfortunately, the Study failed to mention that Belerwaltes and
Wagner revised their position twelve years later (1968) suggesting
that thyroldltls would develop within 1-2 days post-radioiodine
treatment in approximately 5 percent of the patients. They noted
also that symptoms of thyroldltls seldom lasted over 24 hours but
might persist as long as 3-7 days. Beierwaltes and Wagner38 also
noted that radiation thyroldltls could lead to hemorrhage Into the
thyroid gland in patients on anticoagulant drugs.
2. Although presence or absence of clinical thyroldltls was not
reported, Miller et al.39 reported 6 to 8 patients receiving over
7 mC1 of iodine-131 (estimated dose - about 10,000 rads) showed
hlstologlc changes 1n the thyroid compatible with acute
thyroldltis or Hashimoto's thyroldltls depending on how long after
exposure they were examined. An additional 4 of 12 patients who
received between 0.9 and 5.5 mCl of 1od1ne-l3l (estimated dose
2-7
-------
1,200 to 7,200 rads) may have shown some hlstologlc evidence of
thyroldltls. Therefore, 1t would appear that the subjective
evaluation of the patient on the extent of thyroldltls 1s not a
good guide to actual Incidence and that the threshold value of
25,000 rem suggested In the report may not be appropriate.
3. In the discussion of thyrotoxicosis (thyroid storm) 1n
Section H2 of Appendix VI, reference Is made to a U.S Department
of Health, Education and Welfare (HEW) publication (HRA-74-1767)
suggesting that 29 cases of thyrotoxicosis occur per 100,000
persons (the HEW publication should be credited as HRA-75-1767).
The effects of this condition are likely to be more Important than
Indicated In Appendix VI. The assessment by Ingbar and Woeber1*0
used In the Reactor Safety Study that death may occur In 20
percent of thyroid storms should have been modified to reflect
that this 20 percent mortality occurs under an Intensive hospital
regimen to combat the thyroid storm. (What might happen 1n an
untreated group was not estimated.) Unless hospital care Is
available, mortality may run higher In cases of thyroid storm than
the report Indicates.
4. The threshold for hypothyroldlsm at low doses Is not well
established. The findings of Hamilton and Tompklns1*1 were
discussed but results for two patients were excluded, even though
hypothyroldlsm occurred In one case after a dose of about 10 rem,
because of a prior goiter condition. However, golterous persons
will be exposed In accident situations and all of the data should
have been Included In the risk estimates. The relatively high
Incidence of goiter 1n the U.S., particularly north of the 40th
parallel, suggests that 1 percent to 5 percent of the population
1s so affected and may be at risk for the low dose effects
Indicated In the study by Hamilton and Tonkins*1. The 20 rem
threshold for hypothyroldlsm is not well based. If there is a
threshold, a better estimate would be about 10 rem.
5. The discussion (Section H3.4.2 of Appendix VI) of
hypothyroldlsm after high dose (>2,500 rem) exposure to iodine-131
encompasses a rather interesting assumption, i.e., since surgical
treatment yields hypothyroldlsm at an incidence of about 0.7
percent per year, this same percentage should be accepted as the
spontaneous rate of conversion of Grave's disease to a hypothyroid
condition. This assumption is not fully justified, for the
following reasons:
(a) The relative Incidence of hypothyroldlsm depends partially on
the amount of thyroid tissue remalninq. In some surgical series a
portion of the gland 1s purposely left to prevent hypothyroldlsm;
2-8
-------
(b) delayed partial or complete loss of thyroid function may
result from either progressive restriction of blood supply or
autoimmune destruction of the thyroid remnant;
(c) defective organic binding of thyroldal Iodine follows
radlolodlne therapy so that the chances of hypothyroldlsm are
Increased. (See Beierwaltes and Wagner38 and Ingbar and Woeber1*0
for a more detailed discussion of why surgery and radlolodlne
therapy may lead to hypothyroldlsm.)
The phenomena Involved In postablatlon development of
hypothyroldlsm are quite unique to this treatment modality. The
associated changes In LATS and TSH response may also be applicable
only to stresses due to radioactivity. These factors were not
considered In the report.
6. Section H4.6.2 of Appendix VI of the Reactor Safety Study 1s
concerned with the effects of 1od1ne-l3l 1ngest1on on children.
It draws quite heavily from Marshallese exposure data to Imply
that the reported nodularlty rate observed there was Influenced by
a high degree of scrutiny and suspicion, leading to detection of
smaller nodules than 1n general clinical practice so that the
estimates used In the Study may be unduly conservative. However,
several factors Indicate that the opposite 1s more likely to be
the case. What Is not mentioned In the discussion of Marshallese
experience 1s that:
(a) In 1965, the more heavily exposed, Rongelap group was placed
on exogenous thyroxlne to nullify the stimulating effects of TSH
on the thyroid gland and Inhibit development of benign or
malignant nodules;
(b) In 1971, the A1l1ngnae group was placed on exogenous thyroid
hormone; and
(c) In 1974, all other exposed Marshallese, about 3/4 of the
total group, were Included In the hormone therapy program to
prevent the occurrence of more cancers'42. As a result, followup
of the untreated patients group Is limited. The results of the
followup at 20 years may have been confounded by the medical
treatment Initiated after the first 10 years and are compromised
as far as statistical risk estimates are concerned.
A possible Implication, or Inference, derivable from the data
considered In the Study Is that: 1f persons are exposed to radlolodlne
and their thyroid 1s not removed surgically, then they should be
placed on exogenous thyroxlne Immediately to reduce their chance of
developing thyroid cancer. This 1s the handling of the problem that
was developed 1n the Marshall Islands. The applicability of this
2-9
-------
solution to a U.S. accident situation was not developed In The Reactor
Safety Study, nor were consequences of surgical and hormonal
Intervention.
7. In summary, there are four basic areas of disagreement
concerning the risk to the thyroid due to radiation. First, In
the area of acute effects - radiation thyroldltls, there Is
evidence that the threshold for thyroldltls may be 10,000 rads or
lower. Perhaps It would be a subc11n1cal thyroldltls but It would
be present. A recommendation of 10,000 rads as the radiation
thyroldltls threshold seems more appropriate than the 20,000 rad
limit utilized In the Reactor Safety Study.
Second, In the area of continuing effects - hypo thy roidl sin,
the data presented 1n Appendix VI Indicates at least one case
occurred at 10 rem or less. In Table VI H-3 effects are shown In
the 30 to 80 rem exposure range but not 1n the 10 to 30 rem
exposure range. However, 1n view of the short followup (14 years)
and the small numbers Involved (146 persons), 1t may be premature
to say that the threshold for hypothyroldlsm 1s 20 rem. A
recommendation of 10 rem for the radiation hypothyroldlsm
threshold seems more appropriate.
Third, In the area of risk estimation, It 1s Important to
Indicate that although a risk estimate of 4.6 cases/106/rem/year
Is suggested, the curves shown 1n Figure VI H-l do not show much
linearity nor does data In Donlach1*3 nor Jackson1***. The data 1n
Figure VI H-l and Table VI H-3 can also, perhaps more grossly but
accurately, be fit by the following equation:
Risk = [40 x 10'6 + 12(n-l) x 10~6 yr] cases of hypothyroidism per rem
for n years after exposure.
8. Finally, relative to thyroid cancer, an upper bound estimate
of risk compared to external x-rays of 1/10, a central estimate of
1/20 and lower bound estimate of 1/60 is not adequately justified.
This 1s especially so If the latent period for cancers due to
1odine-13l Is similar to that for external radiation (up to 40
years) as Indicated In Appendix VI to the Reactor Safety Study.
At the present time the estimate made by Dolphin and Beach145
of iodlne-131 being about 1/10 as effective as x-rays In inducing
thyroid neoplasla, 1s a reasonable estimate. This same general
estimate is supported by Cole46.
2-10
-------
E. Risk Estimates for Genetic Disorders
Estimates of genetic risks due to radiation accidents are not as
clear cut as Indicated In the Reactor Safety Study.
1. In Section 14 of Appendix VI, concerned with the differential
sensitivity of males and females, the presentation suggests that
only slight numerical adjustments need be made to BEIR Report
estimates of genetic effects. However, the actual situation Is
the subject of active scientific debate and the question of
differential sensitivity of spermatogonla and oogonla 1s being
Investigated by numerous researchers. UNSCEAR1*7 suggested the
mutation rate In males was about 1/3 as great at low dose rates as
at high. Also, UNSCEAR*7 reported that, regardless of dose rate
effects, there Is a threshold dose rate below which the cellular
repair system 1s unaffected. This threshold 1s higher In
spermatogonla than In oogonla.
Abrahamson1*8 has Indicated that the mouse oocyte 1s so
radiosensitive that at the dose rates used 1n studies of mouse
oogonla, the reduced genetic effects observed are due to cell
death rather than to cellular repair mechanisms. Therefore, the
BEIR Report estimate of 0.25 x 10~7 per locus per rem may be a
factor of 2 In error. Since the male rate of 0.5 x 10~7 mutations
per locus per rem was divided by 2 to get the average rate for
males and females, 1f the estimate Is 1n error, 1t 1s low rather
than high.
2. Crow1*9 has suggested that the appropriate single-locus
mutation rate Is 2.6 x 10""7 per locus per rad for high dose rate
radiation and perhaps 1/3 to 1/4 as high for chronic radiation.
This would be 0.6 - 0.8 x 10~7 mutations per locus per rad, compared
to the original BEIR Report estimate of 0.25 x 10~7 mutations per
locus per rem. If Abrahamson and Crow are correct, the BEIR
Report estimate may be a factor of 140 percent to 220 percent low.
Indeed, Lyon et al.50 have suggested that the mutation rate
Increases with decreasing dose rate (below 0.001 R/min) so that
the 1972 BEIR estimate would be an order of magnitude low.
However, we believe It unlikely that the 1972 BEIR estimate 1s as
low as Lyon's suggestion.
3. In Section 15.1.1 regarding single-gene disorders, a
discussion of autosomal recessive mutants Is made where the
frequency 1s given as q and the post Irradiation frequency as
q + Aq. This implies that there is no loss of thp recessive genes
Setwepjt opneraHmn. A more accurate description $f Use
Irradiation frequency would be
2-11
-------
- s) + u(1 - q)/(l - sq)
rtiere: s Is the probability that the gene will be eliminated In
any one generation; u 1s the mutation rate for the dominant gene,
Q, to recessive gene, q, at that locus (perhaps the sum of
radiation Induced and "spontaneous" mutation rates)53.
Therefore, to show that there 1s little change In the
Incidence of a condition Instead of showing that recessive
q2 - (q + Aq)2, it must be shown that:
' sq)l2
In all of the discussions m botn :ne BEIR Report and
Appendix I of Appendix VI of the Reactor Safety Study, the
Questions of (1) Incomplete penetrance, (2) variable expression,
(3) multiple alleles and related questions have been avoided or
neglected52. These factors will, of course, affect the final risk
estimates.
4. In Section 15.1.2, concerned with multlfactorial diseases,
the statement that they depend on variation at more than one locus
may not be true. Thompson53 reviewed evidence on the nature of
polygenes and concluded the number of loci Involved Is about the
same as the number of "major mutant" loci. The polygenes may be
Isoalleles, "normal wild-type alleles," of major mutant alleles.
Polygenetlc conditions could, therefore, be single mutations
producing Isoalleles and expressed by changes 1n eplstasls caused
thereby.
5. In Section 16.2 regarding estimates of Increases in single-
gene disorders (point mutation), a discussion of doubling dose 1s
made and the article by Neel, Kato and Schull51* is used to support
the argument that the doubling dose in humans exceeds 140 rem In
males and 1,000 rem in females. This 1s unfortunate since this
estimate depends quite heavily on the assumptions made in the
Epllog of the paper by Kato et al.55, where some of the
assumptions (page 369) are not referenced but are only presented
as assertions. In view of the rapid developments in population
genetics, perhaps these assumptions thouftf be ffr-e*aTwrted<«&i
Adequately Sapporo Furtherwore, *feei et di.sl* use a factor to
compensate ror xne cfiange from high dose rate to low dose rate;
that Is, a reduction factor of 3 or 4 in males and 20 in females.
Use of these reduction factors will not be valid unless the
mutation rates were measured at relatively high doses1*8*50.
Perhaps move cemplete analysis or uie data on the dou-bling
dose and possible range of variation (3 R to 200 R) can be found
1n papers by Neel and Schull56 and Parker57. These analyses are
2-12
-------
possibly more pertinent since Neel et al.51*looked only at- gametes
which were lethal during the first 17 years of life. Estinioces of
doubling dose for other types of genetic doubling can be made fron,
"Long-Term Worldwide Effects of Multiple Nuclear Weapons
Detonation?"". Doubling of balanced chromosomal rearrangements
occurred at about 55 rem and sex-chromosome aneuploldy at about
175 rem. Both values are within the 20-200 rem doubling range
suggested by the BEIR Report. The disparity 1n values suggests
that mutation rates may have to be considered on the basis of
locus. This latter need 1s reinforced by the observation of Kohn
and Melvoid59 that the "7-locus," "6-locus" and H-locus tests In
mouse spermatogonlum had different x-ray Induced mutation rates -
relative values of 60, 20 and 1 respectively.
The whole area of radiation-Induced genetic effects Is In a
§tage of flux, and changes In rates of mutation at any loci should
have been carefully considered In the Reactor Safety Study.
F. Radiation Dose Causing Specific Clinical Effect
The Reactor Safety Study did not completely assess and assign
appropriate SO values (radiation dose causing specific clinical
effect). The Study should have specified SO values for respiratory
Impairment and hematopoletlc failure expected after a reactor
accident, If only in terms of ranges.
G. Other Biological Consequences
The Reactor Safety Study omitted some biological consequences of
possible significance. To the extent that data are available, the
following consequences should have been discussed:
(1) The synerglstlc and additive effects of other environmental
pollutants for specific Indicators of life shortening such as
cancer, and
(2) A review of the extent of bioeffects anticipated in the
terrestrial, aquatic and atmospheric ecosystems.
H. Underestimates of Risk
In regard to chronic-or latent cancer fatalities, EPA holds the
position that the linear, non-threshold hypothesis recommended by the
NAS BEIR Committee 1s most appropriate. Therefore, EPA considers this
model to represent the central estimate, as opposed to the definition
in WASH-1400 whereby the central estimate is one in which a OREF of
0.2 1s Inserted to calculate health risks following a reactor
accident. As a result, EPA suggests that WASH-1400 may underestimate
the risks from some accident scenarios by as much as a factor of 5. In
2-13
-------
addition, however, the Agency feels that the Reactor Safety Study
failed to recognize the Importance of relative risks, thereby possibly
Introducing another factor of 2 error. The middle estimate of late
somatic effects, then may be underestimated by a factor of 2-10.
2-14
-------
100
9-O* '
v '•& [
90'
80
< 70
cc
60
2 50
20
10
'-'.I
V
< >A
^ 40
UJ
u
c
£ 30
x CONTROL
:V
'\ \ \
\ . x
- \
\ •• x
V •«, \.
O 100 RAT 0.004 R/MIN (AFDR)
A 300 R AT 0.06 R/MIN (AFDR) "I
V 300 R AT 8.5 R/MIN \
(DATA SELECTED FROM ANDERSON \ <
AND ROSENBLATT, REFERENCE 33) frv
0 1 2 3 4 S 6 7 8 9 10 11 12 13 14
YEARS POST-IRRADIATION
FIGURE 1. INFLUENCE OF RADIATION EXPOSURE ON LIFE4HORTENING OF BEAGLES.
2-15
-------
I. References
1. BAUM, J. W. Mutation theory of carclnogenesls and radiation
protection standards. Sixth Annual Health Physics Society
Symposium, Rich!and, Washington (1971) p.11.
2. BAUM, J. W. Population heterogeneslty hypothesis on radiation
Induced cancer. Health Phys. 25:97-104. (1971)
3. BROWN, J. M. Linearity versus non-linearity of dose response for
radiation carclnogenesls. Radiation Research (Abstract, in press:
paper submitted to Radiation Research, October 3, 1975).
4. MOLE, R. H. Ionizing radiation as a carcinogen: practicable
questions and academic pursuits. Brit. J. Radlol. 48:157-169.
(1975)
5. MORGAN, L. Z. Reducing medical exposure to Ionizing radiation.
Amer. Indust. Hyg. Assoc. J. p. 358-368. (May 1975)
6. MODAN, B., D. BAIDATZ, H. MART, R. STEINITZ, and S. G. LENIN.
Radiation-Induced head and neck tumors. Lancet 7852:279 (February
1975)
7. SELTSER, R. and P. E. SARTWELL. Amer. J. Epldemiol. 81:2 (1965)
8. SHELLABARGER, D. J., V. P. BOND, E. P. CRONKITE, and G. APONTE.
Relationship of dose of total body buCo radiation to Incidence of
mammary neoplasia 1n female rats. Radiation-induced cancer.
International Atomic Energy Agency, Vienna, p. 161-172. (1969)
9. STEWART, A. and G. W. KNEALE. Radiation dose effects In relation
to obstetric x-rays and childhood cancers. Lancet p. 1185-1188.
(June 1970)
10. BOREK, C. and E. J. HALL. Transformation of mammalian cell vitro
by low doses of x-rays. Nature 243:450-453. (1973)
11. BOREK, C. and E. J. HALL. Effect of split doses of x-rays
neoplastlc transformation of single cells. Nature 252:499-501.
(1974)
12. WARD, B. C., J. R. CHILDRESS, G. L. JESSUP, and W. L. LAPPENBUSCH.
Radiation sensitivity of the Chinese hamster, Crlcetulus griseus.
1n relation to age. Radiat. Res. 51:599-607. 0372)
2-16
-------
13. LAPPENBUSCH. W. L. and D. L. WILLIS. The effect of dimethyl
sulphoxlde on the radiation response of the rough-skinned newt
(Tarlcha granulosa). Int. J. Radlat. B1ol. 18(3):217-233. (1970)
14. ROSSI. H. H. and A. N. KELLERER. The validity of risk estimates
of leukemia Incidence based on Japanese data. Radlat. Res.
58:131-140. (1974)
15. KELLERER, A. M. and H. H. ROSSI. Theory of dual radiation action.
Curr. Top. Radlat. Res. 8:85-158. (1972)
16. ROSSI. H. H. and A. N. KELLERER. Radiation carclnogenesls low
doses. Science 175:200-202. (1972)
17. MAYS, C. W., R. D. LLOYD and J. H. MARSHALL. Late radiation
effects: malignancy risk to humans from total body gamma-ray
Irradiation. Preprint: p. 417-428. (1975)
18. UPTON. A. C., M. L. RANDOLPH and J. W. CONKLIN. Late effects of
fast neutrons and gamma-rays 1n mice as Influenced by the dose of
Irradiation: Induction of neoplasla. Radlat. Res. 41:467-491.
(1970)
19. YUHAS, J. M. Recovery from radiation-carcinogenic Injury to mouse
ovary. Radlat. Res. 60:321-332. (1974)
20. MOLE, R. H. Patterns of response to whole-body Irradiation: the
effect of dose Intensity and exposure time on duration of life and
tumor production. Brit. J. Radio!. 32:497-501. (1959)
21. GRAHN, D., R. J. M. FRY and R. A. LEA. Analysis of survival and
cause of death statistics for mice under single and duration of
life gamma Irradiation. Life Sciences and Space Research 10:175-
186. (1972)
22. FINKEL, M. P. and B. 0. BISKIS. Experimental Induction of
osteosarcomas. Progress In Tumor Research 10:72-111. (1968)
23. GIBSON, R., S. LILIENFELD, L. SCHUMAN, J. E. DOME and J. J. LEVIN.
Nat. Cancer Inst. 48:301. (1972)
24. CASARETT, G. W. Pathogenesls of rad1onuc11de-induced tumors In
RadlonucUde Carclnogenesls. C. L. Sanders, R. H. Busch, J. E.
Ballou and D. D. Mahluna. USAEC CONF-720505. (1973)
25. LAPPENBUSCH, W. L. and J. D. GILE. Effect of cadmium chloride on
the radiation response of the adult rat. Radlat. Res. 62:313-322.
(1975)
2-17
-------
26. NATIONAL ACADEMY OF SCIENCES ADVISORY COMMITTEE ON BIOLOGICAL
EFFECTS OF IONIZING RADIATION. The effects on populations of •
exposure to low levels of Ionizing radiation. Washington, D. C.
20418 (1972)
27. U. S. ENVIRONMENTAL PROTECTION AGENCY. Environmental analysis of
the uranium fuel cycle. Part III - nuclear fuel reprocessing.
EPA-520/9-73-003-D. Office of Radiation Programs, U.S.
Environmental Protection Agency, Washington, D.C. 20460 (1973)
28. HALFORD, R. M. The relation between juvenile cancer and obstetric
radiography. Health Physics 28:153-156. (1975)
29. NEHCOMBE, H. B. and J. F. McGREGOR. Childhood cancer following
obstetric radiography. Lancet p. 1151-1152. (November 20, 1971)
30. MOLE, R. H. Antenatal Irradiation and childhood cancer: causation
or coincidence? Brit. J. Cancer 30:199-208. (1974)
32. AINSWORTH, E. J., N. P. PAGE, J. F. TAYLOR, G. F. LEONG and E. T.
STILL. Proc. of Symposium on Dose Rate 1n Mammalian Radiation
Biology (CONF-680410). p. 4.1-4.21. University of Tennessee and
U.S. Atomic Energy Commission. Oak Ridge, Tennessee. (1968)
31. LUSHBAUGH, C. C., COMAS, F., EDWARDS, C. L. and G. A. ANDREWS.
Clinical evidence of dose rate effects 1n total-body Irradiation
In man. Proc. Of Symposium on Dose Rate in Man1mal1an Radiation
Biology (CONF-680410) p. 17.1-17.25. Univ. of Tenn. and
U.S.A.E.C. Oak Ridge. Tenn. (1968)
33. ANDERSON, A. C. and L. S. ROSENS L AH. The effect of whole-body x-
Irradlatlon on the median llfespan of female dogs (beagles).
Radlat. Res. 39: 177-200. (1969)
34. CASARETT, G. W. and H. E. EDDY. Fractlonatlon of dose In
radiation-Induced male sterility, 1n Dose Rate In Mammalian
Rad1t1on Biology, D. G. Brown, R. G. Cragle and T. R. Noonan, eds.
U.S. Atomic Energy Commission CONF-680410. National Technical
Information Service. Springfield, Virginia, p. 14.1 to 14.10.
(1968)
35. SHELLABARGER, C. J. and R. D. BROWN. Rat mammary neoplasia
following Co Irradiation at 0.03 R or 10 R per minute. Radlat.
Res. 51. Abstract ED-3. (1972)
36. UPTON, A. C., M. L. RANDOLPH, and J. W. CONKLIN. Late effects of
fast neutrons and gamma-rays 1n mice as influenced by the dose-
rate of irradiation: induction of neoplasia. Radlat. Res. 41:467-
491. (1970)
2-18
-------
37. DONIACH. I. Effects Including cardnogenesls of I and x-rays
on the thyroid of experimental animals: A review. Health Physts
9: 1357-1362. (1963)
38. BEIERWALTES , W. H. and H. N. WAGNER, JR. Therapy of thyroid
diseases with rad1o1od1ne. Principles of nuclear medicine. H. N.
Wagner, Jr., W. B. Saunders Co., Philadelphia, p. 343-369. (1968)
39. MILLER, E. R., S. LINDSAY and M. W. DAILEY. Studies with
rad1o1od1ne validity of hlstologlc determination of 1od1ne-l3l
radiation changes 1n the thyroid gland. Radlol. 65:384-393. (1955)
40. INGBAR, S. H. and K. A. WOEBER. The thyroid gland. Textbook of
endocrinology. R. H. Williams, ed. W. B. Saunders, Co.,
Philadelphia, p. 105-286. (1968)
41. HAMILTON, P. and E. TOMPKINS. Personal communication (September
4) cited in Appendix II to Appendix VI of WASH-1400. (1975)
42. CONARD, R. A. A twenty-year review of medical findings in a
Marshallese population accidentally exposed to radioactive
fallout. BNL 50424. Brookhaven National Laboratory. Upton, N.Y.
(1975)
43. DONIACH, I. Radiation biology. The thyroid. S. C. Werner and S.
H. Ingbar. eds. Harper and Row, Publishers, New York. (1975)
44. JACKSON, G. L. Rad1o1odine therapy of thyrotoxicosis. Amer. J.
Roentgenol. Radium Ther. Nucl. Med. 112:726-731. (1971)
45. DOLPHIN, G. W. and S. A. BEACH. The relationship between radiation
dose delivered to the thyroids of children and the subsequent
development of malignant tumors. Health Phys. 9:1385-1390. (1963)
46. COLE, R. Final report. Inhalation of Rad1oiod1ne from fallout:
hazards and counter-measures. ESA-TR-72-01 (Appendix F, Ionizing
Radiation: Induction of thyroid pathology) DCPA Contract DAHC 20-
70-C-0381, Defense Civil Preparedness Agency, Environmental
Science Associates, Burllngame, Calif. (1972)
47. UNITED NATIONS SCIENTIFIC COMMITTEE ON THE EFFECTS OF ATOMIC
RADIATION. Report of the United Nations Scientific Committee on
the Effects of Atomic Radiation Appendix C. The genetic risks of
Ionizing radiation. United Nations, New York. (1966)
48. ABRAHAMSON, S. IAEA International Symposium on Biological Effects
of Low Level Radiation Pertinent to Protection of Man and His
Environment. (1975)
2-19
-------
49. CROW J., Presentation of the USA National Academy of Sciences
report on the effects of Ionizing radiation (BEIR Report). 2.
Genetic Effects. Proceedings of the Third International Congress
of the International Radiation Protection Association CONF-730907.
W. S. Snyder, editor, U.S. Atomic Energy Commission, Oak Ridge.
p. 37-42. (1974)
50. LYON, M. P., D. G. PAPWORTH and R. J. S. PHILLIPS. Dose-rate and
mutation frequency after Irradiation of mouse spermatogonla.
Nature New B1ol. 238:101-104. (1972)
51. LI, C. C. Human genetics prlnldples and methods. (1962)
52. STERN, C. Principles of human genetics, 2nd edition. W. H. Freeman
and Company, San Francisco. (1960)
53. THOMPSON, J. M., Jr. Quantitative variation on gene number.
Nature 258:655-668. (1975)
54. NEEL, J. V., H. KATO, and W. J. SCHULL. Mortality 1n Children of
Atomic Bomb Survivors and Controls. Genetics 76_:311-326. (1974)
55. KATO, H., and W. J. SCHULL and H. V. NEEL. A cohort-type study of
survival In the children of parents exposed to atomic bombings.
Amer. J. Human Genetics 18:339-373. (1966)
56. NEEL, J. V.and W. J. SCHULL. The effect of exposure to the atomic
bombs on pregnancy terminations In Hiroshima and Nagasaki.
Publication No. 461. National Academy of Sciences - National
Research Council. Washington. O.C. 2041R (1956)
57. PARKER, D. R. Statement on the genetic effects of Ionizing
radiation. Proposed Appendix I Hearings, Docket Number RM50-2.
U.S. Atomic Energy Commission. Washington, D.C. 20545 (1972)
58. NATIONAL ACADEMY OF SCIENCES. Long-term world wide effects of
multiple nuclear-weapons detonations. Genetic effects on humans.
National Academy of Sciences - National Research Council.
Washington, D.C. 20418 p. 203-212. (1975)
59. KOHN, H. I. and R. W. MELVOLD. Divergent x-ray-induced mutation
rates in the mouse for H and "7-locus" groups of loci. Nature
259:209-210. (1976)
2-20
-------
SECTION 3. OVERALL REVIEW
A. COMMENTS on the CONSEQUENCE CALCULATION DESCRIPTION
As revised In the final WASH-1400 report, Appendix VI,
"Calculation of Reactor Accident Consequences," provides extensive
Information additions on some aspects of the calculatlonal process.
However, EPA's comment of November 27, 1974, Is still applicable; that
1s, "Furthermore, the description of certain critical portions of the
overall calculatlonal process should be significantly expanded to
permit a clear understanding of the relationships between the
radioactive material releases, Its dispersion, population
distributions, and the resulting health effects." In our review of the
draft WASH-1400 report, we noted the deficient description of the
.calculatlonal process, but chose to give the Reactor Safety Study some
benefit of doubt with regard to Us adequacy, where the adequacy could
not be discerned from the description. Since 1t has been necessary to
revise the consequences evaluation so extensively 1n the final report,
we believe that the revised analyses should be thoroughly documented.
Toward this end, we recommend that the Reactor Safety Study group
publish a technical report detailing the framework of the consequence
model calculations. The report should give all the Important
analytical relationships and their Interconnections as well as
complete documentation of the computer program details, so that any
compromises and limitations Introduced In the programming are also
made clear. Such a technical report Is needed to show that the
various parts of the calculations, some of which have been described
In detail 1n Appendix VI, have been used 1n a coherent and self-
consistent process to account comprehensively for the significant
modes of exposure, applicable exposure areas, and relevant time
periods. Without such a document, It Is difficult to assess the
overall realism and adequacy of the calculations. Such documentation
should be provided In the near future, In addition to the 5-year
update of the Reactor Safety Study suggested 1n the final report. The
following comments will Illustrate some of the deficiencies In the
description of the consequence calculatlonal process In the final
report.
1. Appendix VI does not give the relation of the various models
being used. The Introduction to Appendix VI states that the
consequence model Includes a "standard Gaussian dispersion model,"
and Section 13.2.1 Indicates that "...isopleths for airborne
radioactive material and for ground contamination...11 were
calculated, a process to which the Gaussian dispersion model lends
Itself. However, the description 1n Section 4, titled
"Atmospheric Dispersion," Indicates that the Gaussian dispersion
model 1s used to estimate a measure of the exposure at only the
centerpoint of each downwind sector-segment, which 1s then
considered to be uniformly exposed over its rectangular area. The
3-1
-------
relation of this calculation of exposures at a single row of
segment centerpolnts to the calculation of Isopleths 1s not given
1n the report.
2. ^ Appendix vi does not give a coherent picture of the way +>"•
eiiraMon of exposure for * population qrouo 1s calculated, wun
regard to trie duration or exposure, Section 4 state*, pInitially
the plume Is treated as a simple ground-level release of 0.5 hour
duration. It Is then corrected for ... buoyant rise, differences
1n release duration..." and, "Transport speed, stability, and
precipitation occurrence are updated by successive hourly weather
observation as Indicated." (It 1s believed the "as Indicated"
should be a reference to the statement 1n Section 5, "For all
meteorological data for a particular site, the consequence model
assumes that the condition that occurs at the site at a given hour
also occurs simultaneously at all downwind locations to whatever
distance the plume has traveled.") "Thus, the plume expands
continuously by vertical and horizontal Increments according to
spatial Increment length and location, hourly value of wind speed,
stability class, and mixing depth."
Assuming, as was done 1n the model, that there 1s no major
shift In wind direction, a release of 5 hours duration, for
example, will be covering lengthwise (downwind) more than one
spatial Interval (given In Table VI 4-1), over more than 5 hours
duration, as the plume also expands In the downwind'direction.
Plume expansion 1n the downwind direction 1s not considered,
however, as Indicated In Appendix A, "... neglects diffusion down-
wind compared to gross transport by the mean wind." .It Is not
explained how the duration of exposure 1s calculated at a
particular location, considering the hourly adjustments and the
downwind'extent of the plume. Perhaps from Sections 9 and
11.1.1.3, 1t may be Inferred that a fraction of the population 1n
each segment of an evacuation area Is postulated to move radially
away from the reactor at the selected effective evacuation speed
(corresponding to that fraction of the population) until the
radioactive plume catcties them. The duration of exposure begins
then; at the same time they turn to evacuate 1n the crosswlnd
direction, requiring 4 hours to escape the area which has been
contaminated (or 1s still being contaminated). They are
Irradiated by direct exposure to the plume for the duration of the
release, or for 4 hours, whichever 1s less. Their dose commltmert
from Inhaled radlonuclldes Is determined on a similar basis.
Their exposure to gamma radiation from the contaminated ground may
be based on the amount of radioactive contamination present, both
that which Is accumulating on the ground during plume passage and
that remaining after plume passage, but not for more than 4 hours
total. All these modes of exposure (except from the already
deposited and the already Inhaled contamination) presumably were
3-2
-------
updated hourly according to the weather conditions, and the total
exposure Is the sum of the hourly exposures for the duration of
plume passage, or 4 hours at most. Such a duration and summation
of exposure for a particular population group may be Inferred from
various parts of the text, but other Interpretations are possible.
For Instance, Section 9 may also be Interpreted to mean that the
exposed persons receive the "external dose from passing*cloud" for
the whole duration of cloud passage, regardless of whether 1t 1s
less than 4 hours or more.
3. Similarly, the treatment of the plume width and the
population exposed 1s not specified. Although 1t Is Indicated 1n
Section 4 that the plume width 1s three s1gma-y modified for the
release duration, It also appears that the meteorological data and
population data are based on 22.5 degree sectors. Perhaps a
certain number of persons are calculated to be exposed within the
modified three-slgma-y width, based on the average sector-segment
population density 1n terms of persons per unit area; then the
number of pers'ons exposed 1n a sector segment will change with the
hourly update In the meteorological data. If such a calculatlonal
scheme Is being used, It should be expHdty stated, rather than
being left for speculation.
Due to the evacuation being underway, the segment populations
will be different than originally, although how the changes are
accounted for 1s not explained.
Under some conditions, the model plume Is wider than a 22.5
degree sector; since the sector populations are not associated 1n
the model with real neighboring sector populations, 1t should be
stated that the sector population has been Increased prorata for
the additional width, If this has been done.
4. The description of the consequence model Is not clear even In
some of the aspects 1t covers, which 1s Illustrated by the
following:
(a) The modeling of the releases as puff or continuous releases:
Section 2 of Appendix VI Indicates: "These parameters, time
and duration of release, represent the temporal behavior of
the release 1n the dispersion model. They are used to model
a 'puff release from the calculations of release versus time
presented in Appendix V," but Section 4 says, "Initially the
plume Is treated as a simple ground-level continuous release
of 0.5 hour duration...," and gives an equation for a
continuous release. Since a 0.5 hour duration Is the
shortest release duration listed (Appendix VI, Table VI 2-1),
It appears that the calculation treats all releases using the
continuous model, rather than a puff model.
3-3
-------
(b) The treatment of the wind direction lacks clarity. The
report does not state directly that It used a uniform
distribution of wind "direction" to the composite site
populations; rather, 1t says, "Actually, for the six
composite sites representing the 68 actual reactor sites...,
It was assumed that the wind direction distributions for such
summations would approach uniformity."
(c) Similarly, It seems that the data In Tables VI 5-2A
through VI 5-2G were Included only as supporting background
Information, although they also could have been used to
establish the probability of occurrence of a specific set of
weather conditions at a site; their use was not specified.
(d) The treatment of precipitation duration In Section 6 of
Appendix VI Is confusing. In the case of continued rainfall
In 6 consecutive hours, for example, 1t 1s not clear whether
the model would employ 3 hours rainfall or 5 hours.
(e) Again, In Section 6, It Is stated that, "Dry deposition
Is assumed to proceed at all times, and variations 1n
deposition velocity because of precipitation, surface
wetting, vegetative cover, desorptlon, etc. are Ignored.
The consequence model uses a deposition velocity of 10~2m/sec,
with a possible range of in~3to 10~l m/sec...," but there Is no
explanation of how the possible range 1s used.
(f) Section 7 Introduces "scavenging coefficients" and
"removal times" but does not explain definitively what they
are or how they were used.
(g) Section 13.2.1 states "For each release category stated
1n Table VI 2-1, each of these 90 weather samples 1s used to
calculate Isopleths for airborne radioactive material and for
ground contamination." How these Isopleths are used In the
calculatlonal process 1s not Indicated, except for the
statement, "...the Interaction of each of these 90 Isopleths
with each of the 16 sectors 1s calculated." The use of
Isopleths does not seem to match the rest of the model.
(h) On page 8-4 there Is discussed a correction factor,
which 1s applied to account for the facts that the cloud is
finite and the receptor need not be on the centerline of the
cloud; 1t should be stated in addition that the receptor 1s
then assumed to be on the ground directly under the
centerline of the cloud, and the exposure for such a receptor
Is applied uniformly to all exposed persons In that segment
within the three sigma-y plume width.
3-4
-------
B. Corrections or Information Additions Needed (Appendix VI)
1. Certain of the accident consequences depend to a large extent
on the size of the radioactive particles. Particles 15
micrometers (urn) In diameter or more will tend to deposit more
quickly. Exposures from Inhalation are greatly dependent on
particle size. Appendix VI, Section 6.3.1 mentions, "... 1-
mlcron-dlameter aerosols (particles and vapors)...." It appears
(but 1s not stated) that this size was assumed for the calculation
of consequences. Appendix K to Appendix VI states "...whereas the
aerosols that would be expected to be released from a reactor core
meltdown would be at most a few microns In diameter." Appendix H
to Appendix VII refers to particles of sizes 0.02 to 0.1 urn of
density 5 grams/cubic centimeter (g/cc). Except for the
discussion of the steam-explosion case (Section 1.4), the rest of
Appendix VII discusses only particles 1n the range 10 ym to 15 vim.
Concerning the steam explosion case, Section 1.4 does not mention
sizes but does Indicate that fission products would be released
through vaporization. However, there 1s nothing 1n Appendix VII,
or elsewhere, which traces the transition between these particle
sizes and the "... 1-micron-diameter aerosols (particles and
vapors)...." There should be a discussion of the size of released
particles, of the factors which change the size distribution, and
of the size distributions to which the public might be exposed.
2. With regard to the resuspension discussed 1n Appendix VI,
Section 8 and Appendix E, It should be noted that the
"resuspenslon factor" discussed may be correctly applied to the
central portion of a contaminated area which 1s so large that by
the time air from outside the contaminated zone travels to the
center. It has acquired an equilibrium amount of contamination.
This will be of little Importance when health effects are being
estimated by use of a linear, non-threshold relation to exposure,
since It may be expected that people downwind from the
contaminated area will also be exposed by the resuspended
contamination and thus may compensate for those 1n the
contaminated area whose exposure Is diluted by Incoming
uncontamlnated air. This would not apply when health effects are
being estimated using a dose-rate or dose-magnitude dependence,
unless all exposures happen to be within the same range. It Is
not stated whether consideration was given to resuspenslon
exposures other than In the case where equilibrium Is reached and
all exposures are uniform.
3. It appears from the data presented In Appendix E of Appendix
VI, In Table VI E-3, that there 1s an Initial larger resuspenslon
rate, characterized by a resuspenslon factor of perhaps 10'2 to 10"*
with a half-life 1n the range of 30 to 90 days. It 1s not clear
3-5
-------
that this has been considered 1n assessing exposures and re-entry
or relocation requirements.
4. Section 8 of Appendix VI (begins:
"The release of radioactive materials to the
environment constitutes a potential hazard to man. The
word 'potential1 Is used to stress the point that, for all
practical purposes, man's concern about radioactive
materials In the environment exists primarily when the
material Is In sufficient concentrations to Impose a
radiological burden higher than some value which the
particular Individual finds acceptable In light of the
exposure he receives from natural sources."
The second sentence needs to be deleted. Hazard 1s quite
different from being concerned. Man can easily be unconcerned due
to Ignorance or misinformation. Since WASH-1400 (and Section 8)
deals with a range of radiation exposures which Includes exposures
lethal within a few months, this statement 1s Inappropriate.
5. Appendix VI, Table VI 8-4 needs revision 1n that columns for
shorter periods, e.g., 0 to 1 year, are needed, whereas the
columns for 0 to 20 years and longer periods are superfluous for
all the radlonuclldes listed, with the exception of stront1um-90.
6. With regard to the use of firehosing for decontamination,
there 1s an Inconsistency between Section 11.2.2.3 which states:
"As discussed in Appendix K, present experimental
evidence is not adequate to support any assumptions on
the effectiveness of wet decontamination (i.e., firehosing)
for the small aerosol particles released during the reactor
accident. Therefore, the removal of contaminated surfaces
is the only decontamination procedure postulated by the
study for hard surfaces."
and Section 12.4.1.2 which says:
"The costs of decontaminating developed property
are estimated on the assumption that two
alternative methods would be used, depending on the
degree of decontamination required to meet the
radiation exposure standards. If a decontamination
factor of 2 would suffice (50% reduction In
contamination), the method would consist of
replacing lawns and firehosing roofs and paving.
If a decontamination factor of 20 were required
3-6
-------
(95% reduction 1n contamination), lawns, paving and
roofing would be replaced."
7€ Section 12 of Appendix VI appears to have omitted the costs
of decontamination and replacement of household goods.
8. Only In very rare Instances does WASH-1400 provide any
absolute dose values. One such Is the statement 1n Section
9.2.3.7 of Appendix VI: "Since dose rates 1n excess of 20 rads per
day could only be experienced within a mile or so of the reactor
In the event of the largest release....11 In another Instance, 1n
Section 9.2.2.1, It 1s stated, "In the event of the worst
calculated accident (corresponding to a probability of about 1(T9
per reactor year), the number of people receiving a dose 1n the
range of 350 to 550 rads would be about 5,000; none would receive
a dose above 550 rads." Nowhere 1n WASH-1400 are the absolute
exposures of this or any other specific case detailed, although
some relative doses are provided. Thus, 1t 1s not possible to
assess the accuracy of the consequence estimates by comparison to
calculations of radiation exposures.
It Is noteworthy that In no Instance does the WASH-1400
scenario call for heroic treatment using their health effects
evaluation, I.e., heroic treatment for those who receive more than
550 rads. A better evaluation of the health effects would
Indicate that the group receiving 350 to 550 rads should receive
such treatment, If available.
9. Section 13.3.2 of Appendix VI states, "Figure VI 13-26 shows
the conditional probability for an Individual of dying from latent
cancer as a function of distance from a reactor given the PWR-1A
or PWR-1B releases. The probability of latent cancer fatalities
1s relatively constant out to about 100 miles...." Although this
Figure VI 13-26 has been replaced by the recently Issued Errata
Sheets, the accompanying text In Section 13.3.2 has not been
corrected accordingly. The new Figure VI 13-26 shows more
realistically that the probability drops off rather rapidly from '
about one mile outward. However, it Indicates that the greatest
Individual probability of dying from latent cancer 1s 0.008; this
value appears to be quite low ant) should be explained.
10. Section 13.4 of Appendix VI states, In essence, that a 9
percent Increase 1n fatal cancers (from the largest calculated
accident) over a period of 30 years, "... would probably not be
statistically detectable because of the normally large variation
1n the rate." The projected Increase from the largest calculated
accident, even in the total population of the United States where
It would amount to about 0.4 percent Increase, might well be
detectable 1n the 30 year average. Furthermore, the statement
3-7
-------
ssutnes that no one 1s keeping track of the Irradiated population.
.Considering the attention that has been devoted to study of atomic
bomb victims. It Is Inconceivable that a study would not be
instituted to evaluate the effects of such a large accidental
radiation release. Furthermore, 1t Is not likely that the
Irradiated population would Ignore the health Implication to
themselves. On the contrary, there would more likely be massive
lass-action liability suits, supported by specific medical
investigations and group studies.
11. The estimation of the overall risk of severe accidents 1s the
Important result of the Reactor Safety Study; assessment of the
worst calculated accident 1s of little value If It 1s Isolated and
not put In the context of the spectrum of accidents and their
probabilities, such as fn Figure 5-3 of the Main Report. The
reason 1s that, as shown 1n the discussion 1n Section 13.2.1 of
Appendix VI, the probability of experiencing the largest
consequence on a per-reactor-year basis can be changed as desired,
over a large range. In the example given, It can be changed by
changing the number of population distributions ("sectors"), or 1t
can be changed by changing the number of weather samples.
However, the worst calculated accident attracts attention.
The Information 1n WASH-1400 about the worst calculated
accident does not seem to be consistent. Table 5-4 In the Main Report
Indicates, for an accident having a probability of one In a billion
per reactor year, that 3,300 early fatalities would be expected.
Section 9.2.2.1 of Appendix VI states "In the event of the worst
calculated accident (corresponding to a probability of about 10"9 per
reactor year), the number of people receiving a dose In the range of
350 to 550 rads would be about 5,000; none would receive a dose above
550 rads." It also states that the number of early fatalities 1s
estimated on the basis of Curve B 1n Figure VI 9-1. For this
Information to be consistent. It seems that all 5,000 must receive a
dose close to 550 rads, rather than a distribution over the range 350
to 550 rads. A better explanation appears to be needed.
12. There are a number of other Inconsistencies which, most
probably, can be clarified with a little additional explanation.
Figures VI 13-23 and VI 13-24 show curves of conditional
probabilities of early death which have shapes that appear
reasonable 1n view of other experience In calculating radiation
exposures. Figure VI 13-7 seems to be In conflict with these,
since 1t shows a mortality probability of 1.0 for some distance
from the reactor, as opposed to a maximum near 0.1 1n Figure VI
13-23 and 0.005 In Figure VI 13-24. It Is noted that different
accidents may be the subject of these figures; while Figures VI
13-23, VI 13-24, and VI 13-26 are specified as being for PWR-1A or
PWR-1B releases, the accident case for Figure VI 13-7 Is not
3-8
-------
Identified, although the discussion of Figures VI 13-5,6,-8,-9
Identifies them as the consequences of a large cold release, which
may or may not be PWR-1A. Whether any of these accident cases is
the same as the one 1n a billion event which 1s the subject of
Table 5-4 in the Main Report or the same as the "worst calculated
accident" rpferenceri In Section. 9.2.2,1 of Appendix VI, is not
r,le»r- 'The Use Of different population distribution, for
different cases should be made clear also; some are for 100
persons per square mile whereas the others may be for a "worst
case" population, which should be specified quantitatively.
13. The following are problems In Appendix VI that may be of an
editorial nature:
(a) Page 3-1, paragraph 3, line 7_ - 880 apparently should be
8,800, If the choice of burnup Is Intended to be somewhat
conservative.
(b) Page 4-2, The correction factor for air concentration should
have a minus sign 1n front of the exponent. (See equation 5.12,
Workbook of Atomspherlc Dispersion Estimates, D. B. Turner, PHS
No. 999-AP-26, 1967.)
(c) Page 9-34, Table VI 9-5_ needs additional explanation, in
footnotes or revision bT Tts title.
(d) Page 13-34. Table 13-6 - The manner of calculating the
average values of maximum Wealth consequences should be explained.
(e) Page C-l, equation VI C-2 - The constant term appears to be
missing from the equation.
(f) Page D-22, equation for R(t) - The first exponential term 1s
Incomplete.
(g) Page D-26 - The masses given for the three groups of organs
and tissues classified by Ostwald coefficients add up to 60 kg,
not 70 kg. The difference should be explained.
(h) APPENDIX K - This appendix should Include a table of limits
of surface contamination and similar "rules of thumb" such as
given 1n Section 3 of Technical Report Series No. 152, Evaluation
of Radiation Emergencies and Accidents, by E. TT~Valar1o,
published by IAEA In 1974.
C. Actions to Mitigate Radiation Exposures
In conformance with a Federal Register notice of Interagency
Responsibilities for Nuclear Reactor Incident Response Planning dated
3-9
-------
December 24, 1975, the U.S. Environmental Protection Aqency 1s
responsible for providing practical guidance to State, local, and
other officials on criteria to use 1n planning protective actions for
radiological emergencies that could present a hazard to the public.
Initial guidance, relative to exposure to airborne radioactive
material, was Issued to the public.in September 1975. Further
guidance dealing with protection from contaminated foodstuffs and
water and for protection from exposure to radioactive material
deposited on property or equipment Is under development and will be
Issued when 1t becomes available.
In view of our responsibilities and Interest In this matter, we
have taken a close look at those parts of the Reactor Safety Study
which deal with protective actions which would be taken to reduce the
health Impact of any radlonucllde release resulting from a nuclear
reactor accident. We have found that the discussion of actions which
would be taken to mitigate the consequences of long-term exposure due
to accidental radlonucllde releases Is generally In agreement with
studies being made toward development of EPA guidance, and that
pending the Issuance of EPA guidance dealing with protection from
exposure to contaminated media,^the reliance by the Reactor Safety
Study on recommendations of the Federal Radiation Council and the
Medical Research Council of Great Britain regarding long-term exposure
dose criteria for external radiation and Ingestion via milk and other
pathways appears to be reasonable.
However, we are not In such good agreement with the model which
the Study employed to simulate evacuation of people 1n the vicinity of
the reactor^during the emergency phase (within a short time) of the
accident. .In our view, the Implementation of a protective action,
such as evacuation, should be directly related to the severity of the
projected reactor accident health consequences. Considering the
projected offsite doses to be Indicative of the'potential health
Impact of a radlonucllde release, 1f evacuation Is chosen as a
protective action, then the dimensions of the evacuation area should
be determined by the projected offsite doses or a dose isopleth
corresponding to some predetermined acceptable level of risk. The
dose levels at which evacuation or shelter should be considered have
been recommended by the EPA in Its guidance Issued 1n September 1975,
dealing with exposure to airborne radioactive material.( However, it
appears that In the evacuation model chosen by the Reactor Safety
Study, the dimensions of the area which would be evacuated are fixed
and do not depend on the reactor accident severity, other than the
condition Indicated in Section 12.2.1 of Appendix VI that a core
meltdown Is assumed. Thus, 1t appears that people are assumed to be
evacuated from a large area even though evacuation of only the low
population zone might be necessary. On the other hand, it appears
that in the most severe accidents, people who might require protection
beyond the present evacuation area are not evacuated.
3-10
-------
The following are some specific comments pertaining to the
modeling of actions to mitigate radiation exposures In the Reactor
Safety Study.
1. In section 11.1.1.2, It 1s acknowledged that approximately 5
percent of the people might be expected to refuse to evacuate and
that this minority was not explicitly treated 1n the reactor
accident consequence assessment. Since the radiological Impact of
a radlonucllde release on these people might not be ameliorated by
evacuation, nor by medical treatment, their contribution to the
total population early health effects could be appreciably greater
than their relative proportion of the exposed population.
2. Apparently, the Reactor Safety Study considered only situa-
tions 1n which the population 1n something like the "keyhole" area
(Figure VI 11-2) would always be evacuated regardless of projected
reactor core melt consequences. If the evacuation decision were
based on a short-term exposure Protective Action Guide (PAG), such
as EPA's recommended PAG's, an Incentive would exist to first
evacuate the people who live closest to the reactor and to assign
a lower evacuation priority to residents further away from the
plant. Since this response could possibly result In fewer acute
health effects, future efforts should analyze the potential
effects of such an option on overall reactor accident
consequences.
3. It appears from WASH-1400 that evacuation more extensive than
the 25-mlle-radlus keyhole area, shown In Figure VI 11-2, was
never used In the consequence calculations, although dose factors
were developed for It, I.e., exposure to gamma rays from ground
contamination for 24 hours, as Indicated 1n Section 11.3.4. It
appears that discussion of such events 1s only background
material.
4. While we recognize the generic nature of the Reactor Safety
Study, we think that greater consideration should have been given
to the effect of site specific parameters on the accident
consequence model. For example, no attempt appears to have been
made to take Into account the actual status of development of
State radiological emergency response plans at Individual reactor
sites. It Is assumed 1n Section 12.2.1 of Appendix VI, that
emergency plans exist, but this may refer only to the plans of the
nuclear power plant operator. Likewise, site specific parameters
which might affect the speed of evacuation have not been
evaluated. An assessment of the range of consequences due to the
difference* 1n site specific circumstances would have made an
Informative contribution to the report. As 1t 1s, even the
variation in population distributions has been omitted; only an
overall average population density is given. Thus, it is
3-11
-------
difficult to judge even the difference In consequences between the
maximum population and the average. Some compensation for these
weaknesses Is made by the Inclusion of ranges of consequences, but
the parametric effect of various factors such as these 1s not
clearly presented.
5. The choice 1n Section 11.2.1.2 of lower dose criteria for
relocation of people 1n a rural area than for those who live 1n an
urban area 1s debatable. It 1s not clear that such an approach
would be workable. It Is likely that asking some population
groups to accept greater radiation exposure because of where they
live would not be well received.
6. Section 11 of Appendix VI discusses protective measures for
mitigation of radiation exposure. The distinction 1n Section 11
between evacuation and relocation appears to be unrealistic in
that, although an evacuation might be Initiated Immediately, 1t
may not be reasonable to return people to their homes if the homes
are so highly contaminated that long-term relocation 1s deemed
necessary. The people would only receive unnecessary additional
radiation exposure, and might tend to spread contamination by
bringing contaminated belongings with them 1n relocating.
7. In Appendix J of Appendix VI, data from the EPA report
Evacuation Risks - An Evaluation have been used to construct
predictive models for future evacuations, fhe analysis centers on
evacuation speed and has been done both foi single classes of
events and for a general evacuation. Two approaches have been
used. The first Is to determine an underlying probability
distribution for evacuation speeds. The second is to relate speed
to other factors through regression analysis. For either
approach, the choice of data is critical and the lack of
Information on data selection criteria 1s a major flaw in the
report.
The first step 1n the selection process is to determine
clashes of events with enough data for Individual analysis. The
three classes chosen (floods, hurricanes and transportation
accidents) are clearly the only ones with sufficient data for
analysis as Individual classes. However, the criteria for
selecting particular events within the chosen classes are not
clear. Using the data categories shown in Tables VI J-l, through
VI J-3 as a basis for Inclusion, there are five floods (No. 4a,
4b, 5, 29 and 36a) and one hurricane (No. 48) that should have
been Included. In the regression analysis, the total area
evacuated Is one of the independent variables used, and yet for
events No.30a and 30b (Table VI J-2) that information Is not
available and there 1s no Indication as to whether those events
were Included or excluded from the regression analysis. Thus, the
3-12
-------
conclusions drawn In the report for floods and hurricanes could be
Incorrect due to Improper data selection. It Is noted that
Section 11.1.1.3 Indicates that transportation accidents were used
as the basis for the model for evacuations due to reactor
accidents.
To do the modeling for the general case, the data set seems
to be made up of only the three largest classes of events. Unless
the general model Is only Intended for floods, hurricanes and
transportation accidents, other events with sufficient data should
have been Included. The restrictions Imposed through the
selection of data have thus made the general or "combined" model
not truly general.
While data selection Is the most critical problem, there are
also several points In the statistical analysis which need either
clarification or correction. The first model developed Is one
which assumes an underlying distribution of evacuation speeds.
The object of the anlysls Is to determine which distribution best
describes the data. As the analysis has been done here, the only
distribution used 1s the log-normal. A test of the data by means
of the "Lllllefors test" has shown consistency with the log-
normal hypothesis. It should be noted that, especially for the
single class analysis, the number of data points Is quite small,
and the ability of the Lllllefors test to reject log-normality
when the data are 1n reality from another distribution 1s quite
low for small sample sizes.
q:jie graphs 1n Figure VI J-l through VI 0-3 show that the
data, while generally following a linear trend, are not close
enough to the hypothesized line to warrant a claim of log-
normality from a graphical method. Therefore, any conclusions
based on the assumption of log-normality should be drawn with
caution 1n this case.
It should also be noted that the Lllllefors test assumes the
use of unbiased estimates for parameters. The maximum likelihood
estimate for the variance Is biased and, therefore, should be
modified. If this modification has been done, such was not stated
1n the report.
For the second model, an attempt has been made to relate
speed to other factors, such as distance traveled, size of area
evacuated and the number of people evacuated. The results of this
analysis are that speed Is related to distance, but not to the
other factors. This holds for both the single class analysis and
the combined data.
3-13
-------
The combining of data has been justified on two statistical
grounds. The first Is that the confidence Intervals for the
coefficients overlap. Although the large amount of overlap does
Indicate that the coefficients are probably not significantly
different, this procedure Is not a valid statistical test for
equality. The second test Involving an "F statistic" needs to be
clarified and referenced to determine validity. There are
standard procedures for determining the equality of regression
equations, In Brownlee3 (pp. 376-390) and in Draper and Smith4 (pp.
/*tm/ /)«
On page J-5 of Appendix J, to Appendix VI, the confidence
Interval given for predicting "v" Is Incorrect. There should be
at least one additional term Involving "d" 1n the exponent. For
discussions of confidence Intervals for predicted values, see
Brownlee5 or Draper and Smith6.
It should be emphasized that the combining of data does not
need to be justified on statistical grounds. The objective of
combining data 1n this Instance 1s to get a better indication of
the general situation. Therefore, the data for different classes
may follow different trends. The only valid test Is whether or
not the combined data can be described adequately by a single
regression line.
8. The discussion of reduction of early mortality on page 13-34
of Appendix VI 1s confusing. It 1s not clear whether the early
mortality probability at 1.2 miles per hour (mph) 1s a factor of
10 lower than at 0 mph or at no evacuation at all. For a given
population distribution around a reactor site, the average
probability of Individual early mortality should be proportional
to the number of early mortalities. Yet, according to Table VI
13-6, the ratio of early mortalities at 0 mph and 1.2 mph is
6,200/2,300 = 2.7. Thus, the factor of 10 would seem to be a
comparison between 1.2 mph evacuation and no evacuation. However,
the unqualified parenthetical Information Implies that, by
providing more shielding, the no-evacuation scenario provides
better protection overall than the 0 mph evacuation case.
D. Corrections or Information Additions Needed (other than In
AppendfFvT)
The following are comments on the parts of WASH-1400 other than
Appendix VI.
1. Even though the statement is made in the Executive Summary,
page 1, footnote 1, that data for long-term health effects are not
available for non-nuclear accidents, we believe that the long-term
3-14
-------
health effects for nuclear accidents should be indicated 1n the
footnote to present the risks frankly.
2. In Section 2 of the Executive Summary. "Questions and Answers
About the Study," a table presents the most likely consequences of
a coremelt accident, and shows that they are small. Vie believe
that this aspect of the results of the Study deserves more stress.
It should be emphasized that Industrial accidents will happen at
nuclear power plants, e.g., the Browns Ferry fire, but that in the
vast majority of cases, the consequences to the public will be
minor or nil, with the exception of Indirect expenses such as that
due to lost electricity production.
3. On page 49 of the Main Report, 1t 1s stated:
"The probability associated with a specific consequence
Is determined by combining the probabilities of the
Individual input parameters, i.e., by multiplying
Prelease x Pweather x Ppopulatlon. In determining the
consequence probability 1n this way. It 1s necessary to
assure the probabilities are reasonably independent. It
1s difficult to visualize that the accident and the
population densities can significantly affect one
another. It seems equally reasonable to assume that the
population and the weather have no strong dependency,
since the frequency distributions have been obtained by
combining actual meteorological and demographic data
applicable to a large number of sites."
The report fail5 to establish the validity of the
independence of meteorological conditions and population
distributions. It appears that any such Independence is an
artifact of the construction of composite sites- InJeed, ihe text
of Apfcfld/X «., Section b, argues* gainst £ucn, independence at any
specific site:
"Figures VI 5-la through g show wind transport
vector (direction towards which wind blows) roses
for each site for the hourly data averaged over the
year. Comparing the different wind-transport roses
points out the Influence of the Individual site
topography on the average wind flow. Site A shows
the Influence of the river valley, with a
predominant flow toward the south-southwest and a
3-15
-------
secondary maximum up-valley toward the north-
northwest. The valley curves at the site location,
which explains the nonallgnment of the up-and down-
valley maxima. Site D on the flat plain has the
most uniform (quas1-isotropic) direction
distributions. Site E has a remarkable maximum
direction frequency at southwest, due to the
presence of a small Mil to the northwest and a
gully cutting through the elevated coastal plain
adjacent to the hill, which effectively channels
the nighttime land breeze. The site G valley 1s
quite level near the site but rises to several
thousand feet on the southeast side at the Blue
Ridge-Smokey Mountains complex. Thus the up-and
down-valley flow 1s quite dominant. Actually, for
the six composite sites representing the 68 actual
reactor sites (described 1n section 10), it was
assumed that the wind direction distribution for
such summations would approach uniformity. This
assumption 1s based on the fact that local
topographic features which may strongly influence
surface winds are generally randomly oriented when
taken over such a large number of Individual
locations."
This quoted material, down through the word "Actually,"
was deleted by the Errata Sheet Issued in May 1976. The
referenced figures appeared only In the limited first
printing of WASH-1400. Although deleted, the quoted material
1s factual and serves to substantiate the point that there
are grounds for believing that the distribution of
meteorological conditions and the population distribution may
not be Independent as assumed, due tp their mutual dependence
at any Individual site on the local geographic and
topographic features.
Local geographic and topographic features which Influence
wind patterns In the vicinity of a power plant site are also
likely to have an Influence on the population distribution. For
example, neighboring towns and cities are likely to be in the
valley which channels the winds. Towns and cities are also strung
along the sea coasts which are subject to dally on-shore and off-
shore breezes. Such distributions argue against the suggested
Independence, especially with regard to nearby locations where
consequences may be sufficiently severe to produce acute health
effects. Perhaps the realism of associating the meteorological
conditions at a specific site with its own population distribution
Is unnecessary for the determination of average, overall risk; 1t
1s not obvious, however, and should be established.
3-16
-------
4. Section 5 of the Main Report of WASH-1400 states, "In the
largest accident predicted 1n the study, the 1500 latent cancer
fatalities would be distributed over approximately 10 million
people." Nowhere 1s 1t made clear how the figure of 10 million
people was determined; whether that 1s the population within a
22.5 degree sector In the downwind direction within 500 miles, or
whether It was determined 1n some other fashion. It would seem
that there must be sectors of that size which would contain
appreciably more than 10 million people, for example along the
Washlngton-to-Boston corridor. On the other hand, krypton-85 and
a few other radlonuclldes would contribute to world-wide exposures
1f released, Involving a much larger population. The criterion
for determining the affected population 1s of Interest, since the
consequences of reactor accidents are compared to the normal
Incidence of cancer fatalities, thyroid nodules, and genetic
effects, apparently In a population of 10 million (Tables 5-5 and
5-8). The tables should be corrected to acknowledge. If It Is
Indeed correct, that this Is the population for which the normal
Incidence figures are applicable, and that the more probable
accidents would affect a much smaller population to an equivalent
extent. I.e., . comparison to the normal Incidence from a
smaller population would be applicable.
5. In Section 5 of the Main Report, the figures which present
results, e.g., Figures 5-10 through 5-16, have footnotes stating
estimated uncertainties, as was done 1n draft WASH-1400. It is
difficult to give credence to the estimated uncertainties,
however, since 1n many segments of the figures the change from the
draft WASH-1400 to the final WASH-1400 Is greater than the
uncertainty estimated In either. For example, the probability per
year for 20 or fewer early fatalities (Figure 5-10, for 100
reactors) has been decreased by more than a factor of 10; this is
a large change compared to the estimated uncertainties, which are
factors of 1/3 (draft) and 5 (final). The new Appendix XI, on
pages XI 1-2 and 1-3, discusses changes made, most of them
Increasing the consequences or probabilities. There 1s no
explanation provided, however, for the large changes Indicated 1n
these figures. Instead, 1t asserts to the contrary:
"In general, the potential consequences predicted 1n the
final report have Increased over those predicted 1n the draft
report. All predicted consequences 1n the final report,
except one, were within the factors of 1/3 and 3 error bands
of the values predicted 1n the draft report. The predicted
average value of latent cancers Increased by a factor of
about 7, due principally to the error made 1n the weathering
half life that was assigned for cesium decay in the draft
report. This effect also Increased the land area needing
decontamination by 5 and that 1n which relocation is required
3-17
-------
by 10. Early Illnesses were calculated on an organ by organ
basis which Increased the magnitude by a factor of 6. The
rest of the changes were within the confidence bounds of the
predictions In the draft report."
and further:
"Although the probabilities predicted for the various
accident sequences have changed In some details, the overall
predicted probability of accidents did not change
significantly."
Comparison of Figure 5-12 with the corresponding figure 1n draft
WASH-1400 shows, rather than the Increase as In the above quotation, a
decrease 1n latent cancer fatalities by about one-half for any given
probability per year, over a large range of probabilities.
Similarly, 1n large parts of Figures 5-11 and 5-13, the change for
a given probability per year Is much greater than a factor of 3.
In Table 1-1 of the Executive Surmary, the "Individual chance per
year" from nuclear reactor accidents has also decreased by much more
than a factor of 3.
(i. In the Main Report, Section 6.4.9, It Is stated that "The
evacuation model used Is based upon one developed by the
Environmental Protection Agency (Ref.8)." To be correct, this
statement should be changed to read: "... based upon data
collected by ...," since the reference does not Include an
evacuation model.
7. We believe that the last sentence of Item a, In the last
paragraph of page 72 of the Main Report should be rewritten to
read, "This suggests that the likelihood of an accident which
affects the public Is less than 10"3 per reactor-year."
8. It Is not clear whether cost data and cost estimates, 1n the
Main Report, for property damage for the non-nuclear and nuclear
accidents are expressed on a consistent basis such as 1975
dollars. We believe that historic events (for example, the 1906
San Francisco earthquake and fire) should be costed using the same
calendar year dollar basis used to estimate the costs of predicted
future events (such as serious nuclear plant accidents). If a
conversion to the same calendar year dollar basis Is made, 1t
should also be explained.
9. It appears that the last sentence of the first paragraph of
Section 6.4.3 of the Main Report, should read, "In the United
States single family dwellings represent about 40 percent of the
3-18
-------
value of all capital property and, therefore, the total capital
property damage could easily be a factor of 2 larger." If the
above does not accurately express the Intended meaning of that
sentence, then 1t appears that damage to commercial, Industrial,
governmental, etc., capital property due to earthquakes has not
been considered.
10. In Appendix XI, Section 3.2 and Its Attachment 1, "Analysis
of the Browns Ferry Fire," should have been written to Include a
short chronology of the events that actually occurred, starting
with the ignition by candle flame. From the reports of the fire,
It appears that a reasonable choice of phases would have selected
phase 1 as the period from the start of the fire until 5.5 hours
later when the last 4 relief valves became Inoperable; phase 2 as
the 4 hour period during which all relief valves were inoperable;
and phase 3 as the period beginning when some relief valves were
restored to service. Figure XI 3-8 (in the second printing of
final WASH-1400) clearly shows the boundary between phases 2 and 3
at some time roughly in the middle of the period during which all
relief valves were Inoperable. The selection of this time Is not
explained, but It appears to have been selected as the boundary
between periods of higher and lower probability of the incident
proceeding to core meltdown. The legend of Figure XI 3-8 and the
text describe the phases in a confusing manner. The legend of
Figure XI 3-8 identifies "Phase 2 of the fire, during which the
controls for the reactor vessel relief valves were or could have •
been failed...," when In fact there was, by various reports, a
period beginning at 5.5 hours during which all eleven relief
valves were inoperable. Similarly, on page XI 3-57, Attachment 1
states, "During phase 2, had the central system for all eleven
relief valves been Inoperable, the single operating control rod
drive (CRD) pump would have been Incapable of maintaining an
adequate level of water In the core at high pressures," which was
actually the condition while all relief valves were Inoperable, as
acknowledged on pages XI 3-59 and XI 3-60.
Perhaps Figure XI 3-8 incorrectly Indicates the time at which
phase 3 begins. In the "Evaluation of Phase 3 Logic Tree," the
text says: "The analysis 1s similar to that used for times shorter
than 5.5 hours. The event symbols are shown on Fig. 2, which
depicts the logic for times longer than 5.5 hours." Similarly,
Figures 1 and 2 showing the logic trees are clearly labeled "...
for Time Less than 5.5 Hours After Fire Start," and "... for Time
Greater Than 5.5 Hours After Fire Start." These all indicate that
the 5.5 hour point would have been a good time to end one phase
and begin another.
11. The discussion of "Other Possible High-Pressure Makeup
Sources" on page XI 3-57 needs additional explanation to the
3-19
-------
effect that an additional control rod drive (CRD) pump would not
double the existing CRD flow. Otherwise, 1t would seem that the
use of an additional CRD pump would be as effective as closure of
the bypass line In the CRD system, discussed In the subsequent
paragraph.
12. Some further qualification Is needed of the selected
probability of fire occurrence. On page XI 3-55, It Is stated:
"Since approximately 200 reactor-years of experience exist, the
probability of a fire occurrence 1s estimated to be 1/200, or 5 x
10"3 per reactor-year." Reports relating to the Browns Ferry
Incident 11st a number of previous fire occurrences; they were
simply of lesser safety significance. Noteworthy was the fire at
the Swiss Muehleberg nuclear power plant (July 28, 1971) which
also burned many cables and delayed operation of the plant about a
year. Admittedly, this fire was outside the United States data
base. However, there have been a number of other fires at nuclear
power plants which this selection of the probability of fire
occurrence appears to Ignore.
13. Although the Browns Ferry fire analysis 1s considered for the
fire as It occurred, WASH-1400 also looks at other possible
failures which could have made the results worse. One point not
discussed 1s whether further burning of other cables within the
two rooms that were the scene of the fire could, by Itself, have
led to the additional equipment failures necessary for core melt.
14. There are some discrepancies in the numbers quoted for the
likelihood of core melt due to a fire like the Browns Ferry fire.
On page XI 3-41, 1t 1s Indicated that the likelihood 1s 5xlO'6 per
reactor year, or just 10 percent of the 5xlO"5 per reactor year
total probability of coremelt given in the preceeding paragraph.
On page XI 3-52, however, a figure of 20 percent is given, which
1s In agreement with the probability of core melt of l.OxlO"5 per
reactor year given on page XI 3-59.
15. The statement on page XI 3-51 that "... the analysis for
plant No. 1 bounds the probability of core melt at Browns Ferry as
a result of the fire...," does not appear to be correct, because
the risk from damage to Unit No.2 (no matter how small) was 1n
addition to the risk from damage to Unit No. 1. Further, the
Browns Ferry fire demonstrated the possibility of simultaneous
damage at both units, In which case the sharing of redundant
equipment between units (e.g., spare control rod drive pumps)
means that it Is not available to both units simultaneously in
case it 1s needed and, thus, the risk when both units are Involved
1s greater than the sum of the risks for separate, non-
simultaneous Incidents at each unit.
3-20
-------
E. Review of Response to EPA Consents on_ Draft WASH-1400
The new Appendix XI, "Analysis of Comments on the Draft WASH-1400
Report" contains responses to the majority of EPA's comments,
Including complete coverage of EPA's comments of August 15, 1975.
However, not all of EPA's comments of November 27, 1974, were
responded to, nor was a response made to all the comments in the
report by EPA's contractor, Intermountaln Technologies Inc. (ITI).
For an analysis of the response to Ill's comments, refer to Ill's
supplementary report, "A Review of the Final Report Reactor Safety
Study (WASH-1400)," April 1976 (Section 4 of this report). Many of
EPA's comments of August 15, 1975, were based on Ill's report, and,
thus, a response that 1s Inadequate with regard to the comment In
ITI's report Is likewise Inadequate with regard to EPA's parallel
comment. However, the response to the remainder of EPA's comments of
August 15, 1975, Is considered adequate, where a response Is called
for.
A number of EPA's comments were directed at the draft Appendix VI.
Most of the specific comments are no longer of Interest, because the
revision of Appendix VI has substituted new material for that which
was the subject of the comment. However, the revised Appendix VI has
two of the principal deficiencies addressed 1n our comments of
November 1974 on draft WASH-1400. A substantial deficiency is the use
of health effects models which tend toward underestimation of the
number of health effects. The other deficiency is the inadequacy of
the description of the process used for calculating consequences.
3-21
-------
F. References
1. HANS, J. M. Jr., and T. C. SELL. Evacuation risks - An evaluation.
U.S. Environmental Protection Agency. NERC-LV EPA-520/6-74-002.
(1974)
2. LILLIEFORS, H. VJ. On the Kolmogorov-Smirnov test for normality
with mean and variance unknown. JASA 62: 318. pp. 399-402. (June
1967).
3. BROWNLEE, K. A. Statistical theory and methodology in science
and engineering. Second Edition, John Wiley and Sons Inc.,
New York. '(I960)
4. DRAPER, N. R. and H. SMITH. Applied regression analysis. John
Wiley and Sons Inc., New York. (1966)
5. BROWNLEE. p342.
6. DRAPER and SMITH, pp. 21-24.
3-22
-------
SECTION 4
REPORT BY INTERMOUNTAIN TECHNOLOGIES INC.
A REVIEW OF THE FINAL REPORT
REACTOR SAFETY STUDY
WASH-1400
The contract report reproduced as Section 4 of this report was
prepared as an account of work sponsored by the Environmental
Protection Agency. The contract report Is being published so that It
will be available as a resource to the scientific community and the
general public. It does not necessarily represent the views or
policies of the Environmental Protection Agency.
4-0
-------
A REVIEW OF THE FINAL REPORT
REACTOR SAFETY STUDY (WASH-1400)
APRIL 1976
By
P. R. DAVIS
INTERMOUNTAIN TECHNOLOGIES, INC.
505 Lomax P.O. Box 1604
Idaho Falls, Idaho 83401
Contract No. 68-01-2244
Project Officer
Dr. Jerry Swift
Office of Radiation Programs
Prepared for
Office of Radiation Programs
U. S. Environmental Protection Agency
Washington, D. C. 20460
4-1
-------
LEGAL NOTICE
This report was prepared by Intermountain Technologies, Inc. (ITI) as
an account of work sponsored by the Environmental Protection Agency (EPA),
Neither ITI, nor any person acting on behalf of ITI;
a. Makes any warranty or representation, express or implied,
with respect to the accuracy, completeness, or usefulness
of the information contained in this report, or that the
use of any information, apparatus, method or process dis-
closed in this report may not infringe privately owned
rights; or
b. Assumes any liabilities with respect to the use of, or
for damage resulting from the use of, any information,
apparatus, method or process disclosed in this report.
4-2
-------
TABLE OF CONTENTS
LEGAL NOTICE
TABLE OF CONTENTS
LIST OF TABLES
ABSTRACT
SECTIONS:
I. CONCLUSIONS
II. RECOMMENDATIONS
III. INTRODUCTION
IV. GENERAL RESULTS
V. ANALYSIS
VI. ADDITIONAL COMMENTS
VII. REFERENCES
VIII. GLOSSARY
Page
4-2
4-3
4-4
4-5
4-7
4-10
4-13
4-14
4-28
4-76
4-81
4-82
4-3
-------
LIST OF TABLES
Table Page
I. Summary Description of Apparent 4-17
Technical Deficiencies Found in
WASH-1400
II. Summary Description of Apparent 4-22
Deficiencies from General Observations
III. Summary Status of Apparent Deficiencies 4-24
in WASH-1400 (final)
IV. Summary of Risk Significance of 4-25
Apparent Deficiencies in WASH-1400
(final)
V. WASH-1400 (final) Unresolved Areas 4-26
with Potential Risk Changes
VI. Application of Surry Risk Analysis 4-69
to Trojan
4-4
-------
ABSTRACT
This report is an addendum to our report entitled, "A Review of the
Draft Report - Reactor Safety Study (WASH-1400)" which was included
in a publication of the U.S. Environmental Protection Agency entitled,
"Reactor Safety Study (WASH-1400): A Review of the Draft Report"
(EPA-520/3-75-012) issued in August 1975. The purpose of this report
Ls to present the results of a review of the final version of WASH-
1400 dated October 1975. The Review consisted of examining the same
areas of the final version of WASH-1400 which were considered in depth
during the review of the draft (reported on in EPA 520/3-75-012) and
to modify, if appropriate, the conclusions reached from the review of
the draft version.
The clarification and revised analyses contained in WASH-1400 (final)
resulted in some significant improvements over the draft document.
Improvements have been made in many of the areas found to be deficient
by the EPA review of the draft. In general, however, the ITI conclusions
reached from the review of the draft are found, in the present review,
to remain applicable to the final version of WASH-1400.
This report was submitted in fulfillment of Contract Number 68-01-2244
(Modification No. 4) by Intermountaln Technologies, Inc. under sponsor-
ship of the Environmental Protection Agency. Work was completed as of
April 1976.
4-5
-------
INTENTIONALLY LEFT BLANK
4-6
-------
I. CONCLUSIONS
The purpose of the effort described in this report was to review the
final version of WASH-1400 to determine whether the conclusions
resulting from the ITI review of the draft are still valid. The con-
clusions resulting from the draft review are contained in Reference (2)
The conclusions resulting from the review of the final version of
WASH-1400 are as follows:
A. The final version of WASH-1400 has been improved compared to
(3)
the draft version .
B. While some deficiencies in WASH-1400 Draft have been either
clarified, improved, or eliminated, the conclusions from the
review of the draft, with generally minor modifications, are
judged to still apply to the final version. The conclusions
are as follows:
(1) Although errors, omissions, Inconsistencies, and question-
able assumptions still exist in some areas of WASH-1400
(final), the majority of these deficiencies do not have a
significant effect on the overall risk assessments.
(2) The summary presentations in WASH-1400 (final) for compar-
ing the risks of nuclear power with other man-caused risks
are sometimes misleading and Incomplete. Factors such as
not illustrating calculational uncertainties in nuclear
power risks, making comparisons of calculated nuclear risks
4-7
-------
with actual risks from other sources without emphasizing
the distinction sufficiently, and particularly obscuring
latent deaths from nuclear power, all tend to undermine
the strength of the WASH-1400 (final) conclusions, both
expressed and implied.
(3) The WASH-1400 (final) risk assessment from transient with-
out scram accidents for boiling water reactors appears to
be the most significant analysis problem found in the
report. In this case, a preliminary sensitivity study
indicates that re-evaluation of the consequences of this
accident may increase the WASH-1400 calculated risks from
BWRs.
(4) Several areas were found which appear to be improperly or
incompletely considered but for which insufficient infor-
mation is available to determine quantitatively their risk
impact. These areas include
human reliability
PWR small break calculational techniques
common mode failure quantification
some aspects of design adequacy.
(5) The validity of applying the results of the risk assessment
using the Surry reactor, chosen to represent all PWRs in
WASH-1400 (final), to the 60 to 70 PWRs expected to be in
operation by 1980 needs additional consideration in WASH-
1400 (final). In several areas, design differences between
the Surry plant and a plant more representative of the 1980
plant population indicate that the Surry results may not
apply. Surry represents a PWR design similar to only about
20 percent of the anticipated 1980 PWR population.
4-8
-------
(6) The basis for selecting the PWR containment failure pres-
sure was found to be deficient, and the failure pressure
selected appears to be too high.
(7) The assessment of ECCS functionability appears to be in-
complete. Conclusions are stated which may be misleading.
4-9
-------
II. RECOMMENDATIONS
It Is recommended that three separate activities be undertaken by NRC,
as follows:
A. The 13 apparent deficiencies which have been found to still
exist in WASH-1400 (final) and which have the potential for
changing the risks should be resolved. These areas are dis-
cussed in Section IV and are tabulated in Table V. The assess-
ment and presentation of risks from latent fatalities should
also be resolved (see Section V, item 17c, and Section II,
item 2).
B. A revised study of the risks of nuclear power should be under-
taken no later than 1981. This revised study should include
the following:
(1) A resolution of the 21 deficiencies described in this
report which are not covered by recommendation A above.
Although these deficiencies do not appear to have a
significant effect on the risk assessments, they should
be repaired for the following reasons: (a) The existence
of errors, omissions, inconsistencies and questionable
assumptions In the report tends to undermine the confi-
dence gained by the reader in the results, especially
since in many cases the significance of such deficiencies
is not obvious, (b) As changes are made in the report,
the effect of some of the minor problems could be ampli-
fied. Changes in reactor design, as well as operating
and maintenance characteristics, could shift the emphasis
and accentuate the significance of deficiencies which
present]y appear to be minor.
4-10
-------
(2) Comparisons between nuclear and other man-made risks should
be revised to clearly and consistently Indicate:
(a) that the nuclear risks are calculated while other
risks are derived from actuarial data,
(b) the substantial uncertainty associated with the
nuclear risk calculations,
(c) the latent death risk from nuclear power plants.
(3) The next version of the Study should cover PWR designs
other than that represented by the Surry plant selected
for analysis in WASH-1400. Of the 60 to 70 PWRs expected
to be operating by 1980 (stated to be covered by the WASH-
1400 results), only about 20 percent are of the Surry
design. Some of the plants differ in design considerably
from the Surry plant, and further assurance is needed that
the Surry results apply to these plants as assumed by
WASH-1400. An analysis of risks from offshore plants
should also be considered in the next version if this
technology becomes a reality.
C. The Study should be continuously maintained. It is likely that
the power plants covered by the Study will undergo design, oper-
ational, and maintenance and testing changes, some of which may
be required by regulatory agencies. These changes should be
factored into the Study in a timely manner to determine the
effect of such changes on the risk evaluation. As the number
of operating reactor-hours Increases, the component failure rate
and accident frequencies should be monitored and periodically
4-11
-------
factored into the Study. This would improve the statistical
basis in the Study and could alter some of the results. Con-
tinuous maintenance of the Study would not only sharpen the
focus of the quantitative risk assessments, but would also have
the potential of promptly identifying problem areas as well as
improving the methods used in the Study.
4-12
-------
III. INTRODUCTION
In August 1974, the United States Atomic Energy Commission (AEC)
issued a draft document, entitled "Reactor Safety Study - An Assess-
ment of Accident Risks in U.S. Commercial Nuclear Power Plants" (WASH-
1400). The document concluded that risks to the general public from
power reactor accidents are substantially less than from other man-
made risks and most natural disasters.
In December 1975, the Nuclear Regulatory Commission issued a final
version of WASH-1400 . This version was modified from the draft
version, based, in part, on comments received from various reviewers.
The final version also concludes that risks to the general public from
power reactor accidents are substantially less than from other man-
made risks and most natural disasters. A new appendix (Appendix XI -
Responses to Comments on WASH-1400 Draft) has been added to the final
version. This appendix lists and discusses some of the comments re-
ceived from the major reviews of WASH-1400 Draft.
The United States Environmental Protection Agency (EPA) reviewed
WASH-1400 Draft and published the results of their review in August
1975. To provide assistance for the review, the EPA contracted with
Intermountain Technologies, Inc. (ITI). The results of the ITI review
are contained in Reference 2. Following the issuance of the final ver-
sion of WASH-1400, the EPA extended the contract with ITI to provide
for (1) an examination of the same aspects of WASH-1400 (final) and (2)
revisions, where appropriate, of the findings resulting from ITI's
review of WASH-1400 Draft. The purpose of this report is to describe
the results of the ITI effort.
4-13
-------
The ITI review of WASH-1400 (final) consisted mainly of three phases.
A search of Appendix XI was made initially to determine if the defi-
ciencies found in each of the areas analyzed during review of WASH-
1400 Draft were considered. If the deficiencies were considered and
responded to, an assessment was made of the response. The second phase
consisted of determining if modifications had been made to the area
analyzed and found to be deficient during review of WASH-1400 Draft.
If modifications were made, an assessment was completed to determine
if the deficiencies still existed. The third and final phase consisted of
evaluating whether the overall conclusions as stated in the review of
WASH-1400 Draft were still applicable and, if not, to make appropriate
modifications.
4-14
-------
IV. GENERAL RESULTS
The general results of the review of the areas in WASH-1400 (final)
which correspond to areas found deficient in the draft review are
presented in Tables I and II. Table I presents the results of the
review of the engineered safety systems, accident sequences, and
component failure assessments which are contained in WASH-1400
(final). The first column indicates the general area considered,
the second column describes the deficiency found in the review of
WASH-1400 Draft, and the next four columns summarize whether the
deficiency was considered.in WASH-1400 (final) and whether the defi-
ciency, if considered, was adequately resolved. The last column
indicates an assessment of the potential effect on the calculated
r:Lsk, if any, associated with the deficiency if it still exists in
WASH-1400 (final).
Table II summarizes the results of a review of the deficiencies iden-
tified in the General Observations section of the review of WASH-1400
(2)
Draft . This table has the same format as Table I.
Table III is a composite showing how many of the 45 deficiencies found
in WASH-1400 Draft (listed in Tables I and II) were considered and re-
vised in WASH-1400 (final).
Table IV illustrates the number of deficiencies remaining in WASH-1400
(Jfinal) which have the potential for altering the risk. The table also
shows whether each deficiency would tend to increase or decrease the
risk and, where determined, the significance of the change (the potential
change associated with five deficiencies was not determined).
4-15
-------
Table V lists the thirteen deficient areas remaining in WASH-1400
vihich appear to have either a significant (one area) or unknown
potential for changing the calculated risks.
Xn summary, the tables indicate that of the AS deficiencies found
in the review of WASH-1400 Draft (Tables I and II), 34 still remain
unresolved in WASH-1400 (final) as shown in Table III. Of these,
29 have the potential for changing the calculated risk. Sixteen of
the 29 were judged to have an Insignificant effect on the risk,
leaving 13 with either a significant (one item) or unknown potential
for changing the calculated risks (Table V).
It should be noted that the deficiencies represent areas where,
based on an independent judgment, a problem exists. It is possible
Chat in some cases the deficiency results from a misunderstanding of
Che information presented in WASH-1400.
4-16
-------
TABLt I - Surma rv Description of Apparent Technical Def ielcnci«"» Found In VASH-1400
Area
1. BUR Reactor
Protection
System
2. BWR Tran-
sient Acci-
dent
i
i
3. BWR Electric
Power Systen
Failure
4. PWR Electric
Power Systen
Failure
Deficiency
Identified in
Draft Reviev ^
a. Conservative assumption-
three adjacent rod fail-
ures cause scram failure
b. Single control rod
failure rate too low
c. Common mode failure con-
sideration Inadequate
d. Excessive credit taken fo
manual poison injection
a. Inadequate consideration
of the frequency and
severity of anticipated
transients
a. Loss of two dtesels rathe:
than four for LPCI failure
a. Insufficient consideratloi
for availability of
power at LOCA
b. Tine and environmental
effects on failure rate
not considered
c. Probability too low for
operator opening wrong
breakers
d. Evaluation of d.c. cower
unavailability not
adequate
(1) se«
Deficiency
Assessment of I Area Revised
Considered In' Appendix XI
In UASH-1600
Appendix XI Response j Final
Yes
Yes
Yes
Yes
Yes
No
No
Yes
Indirectly
No
Reference 2
Appears Inadequate
Adequate
Appears Inadequate
Adequate
Adequate
.
_
Partially adequate
Inadequate (for this
.specific area)
-
or a complete descript:
U - Amount of potential risk change undeterrai
No
No
No
Yes
No
No
No
No
No
Yes
on of the deflc
Assessment of
Revision
_
_
_
adequate
-
_
_
-
-
adequate
ency
Potential
Risk
Change
decreased (U)
none
increased (S)
none
none
Increased (I)
Increased (I)
increased (U)
increased (I)
none
ned
I - Amount of potential risk change judged to be insignificant
S - A:nount of potential risk change judged to be significant (more than a factor of 10)
I
H>
•vl
-------
TABLE I (continued)
Area
Deficiency
Identified ID
Draft Review (1)
e. Insufficient error con-
sideration of operator
opening wrong break
f. Maintenance errors result'
ing In bus unavailability
not considered
g. Possible incomplete con-
sideration of diesel
generator unavailability
h. Earthquake relay failure
crips in dieseli and load
breaker location not con-
| sidered
5. PUR High a. Operator reliability to
Pressure In-
jection
Systea
open water valves to lube
cil coolers ioo high
b. LPIS check valve failure
noc adequately considered
c. Additional double failure!
not included: failure of
both valves in 1) voluice
control tank drain, 2)
normal charging line, 3)
boric acid recirculation
system.
Deficiency Assessment of Area Revised
Consider,.,'. In Appendix XI j In WASH-1400
Appendix XI: Response j Final
No
Indirectly
No
Indirectly
Indirectly
No
No
(1) See Preference 2 fo
.
Probably adequate
_
Partially adequate
Inadequate (for this
specific area>
_
_
a complete dcscriptio;
No
Xo
No
No
No
No
No
of the deficie
Assessment of
Revision
_
_
_
_
_
_
_
cy
Potential
Risk
Change
Increased (I)
none
increased (I)
Increased (U)
increased (I)
increased (I)
increased (I)
co
U - Amount of potential risk change undetermined
I - Amount of potential risk change judged to be insignificant
S - Acount of potential risk change judged to be significant (more than a factor of 10)
-------
TA:.»L i
Area
6. PWR Snail
Bro iV. Li»ss
of Coolant
Accident
Analysis
7. rv\ -_o-. •• of
Porfer Trjn-
sicnc Acci-
dent Sequence
8. Component
Failure Modes
and Rates
9. Hunan Relia-
bility Analy-
ses
10. Condon Mode
Failures
11. Design
Adequacy
Deficiency
Identified in
Draft Review (i)
a. Assumption of adequate
core cooling for small
breaks not justified
a. Containnent loading froo
ste.ic when accumulator
water contacts core melt
a. Assessed failure proba-
bility not consistent
uitn data in several
are.is
a. Ir.cdequ.iCe and inconsis-
tent derivation of
specific human reliabil-
ity assessments
s. Insufficient description
of how cocsnon mode fail-
ures were quantified and
applied
a. Insufficient description
of the extent which de-
sign adequacies ware
assessed and applied
(1) See Ref
Deficiency
Considered I:
Appendix XI
Yes
Partially
Yes
Yes
Yes
Yes
erence 2 for
Assessment of
i Appendix XI
[| Response
Inadequate
Partially adequate
Partially adequate
Inadequate (more work
said to be useful)
Partially adequate
Inadequate
complete description <
i
Area Revised
In UASH-1400
Final
No
No
Partially
No
Yes
No
f the deficienc
Assessment of
Revision
—
~
Improved
Improved
Poccnclal
Risk
Change
increased (U)
increased (I)
unknown (I)
unknown
unknown
unknown
VO
U - Amount of potential risk change undetermined
I - Amount of potential risk change judged to be insignificant
S - Amount of potenrial risk change judged to be significant (more than a factor of 10)
-------
Area
12. PWR Low
Prer--i.rc
Injection
Systcn
Failure
13. p"p. 1 .••"
Pressure
Recircula-
tion Systeu
>
I
14. Core Melt-
down
Analysis
15. Containment
Failure
Pressure
16. Containment
Pressure
Response
Deficiency
Identified -In
Draft laviiv (1>
a. Incorrect val:.e assigned
to probability of the
operator closing second
wrong valve
a. Operator error-incorract
probability for failure
to open valves
b. Operator error-no consid-
eration of single fail-
ure valve closure.
c. Valve faults incorrectly
placed and counted twice
d. Operator switching to
hot leg recalculation
after hot leg break not
considered
a. Time to core melt exces-
sive
a. Surry containment failure
pressure not sufficiently
justified and appears
high
a. Containment failure time
should be reduced for
CSIS and CSRS failure
(1) See F
Deficiency i Assessment of
Considered In> Appendix XI
Appendix XI; Response
No
No
No
No
No
Yes
Yes
No
_
_
_
M_
Adequate
Inadequate
_
eferencp 2 for a complete descriptioi
Area Revised
In WASH-1400
Final
No
No
No
No
No
No
No
1
i Assessment of
Ravis ion
_
_
_
—
-
.
No
of the deficte
_
cy
Potential
Risk
Change
increase (I)
increase (I)
increase (I)
decrease (I)
increase (I)
none
increase (U)
increase (U)
I
N>
O
U - Amount of potential risk change undetemined
I - Amount of potential risk change judged to be insignificant
S - Amount of potential risk change judged to be significant (more than a factor cf 10)
-------
TABLE I (continued)
Area
I
ro
Deficiency
Identified in
Draft Review (D
b. Containment failure tlae
should be increased for
loss of electrical power
(1)
Deficiency
Considered IT
Appendix XI
Mo
See Referenc
Assessment of
i Appendix XI
Response
-
2 for a complete desci
Area Revised
In WASH-1400
Final
No
ption of the d
i
i
Assessment of
Revision
-
flciency
Potential
Risk
Change
decrease (U)
(U) - Amount of potential risk change undetermined
-------
lAbLE II - Summary Descr.ption of Apparent Deficiencies fron General Observations
(1)
Area
17.(3) Compara-
tive Risk
Curve
18. Nuclear
Plant Char-
acteristics
19. Realistic
vs Conser-
vative
Assumptions
Deficiency
Identified in
Draft Review (?)
a. Calculated vs actual
risks not distinguished
in comparative curves
b. Calculational uncertain-
ties not sufficiently
emphasized
c. Latent deaths not suf-
ficiently emphasized
d. Transition from solid
to dashdd lines in man-
caused rinks does not
correspond to data
a. Distribution of plant
types not considered
b. Risks computed based on
incorrect power level
c. Application of Surry
risk analysis to other
PWRs not Justified
a. Indicated realism not
consistently used
Deficiency Assessment of
Considered In! Appendix XI
Appendix XI Response
No(A)
Yes
Yes
No
No
No
Yes
No
B
Adequate
Inadequate
_
_
-
Probably adequate
_
Area Revised
j In WASH- 1400
Final
No
Yes
Yes
No
Yes
No
No
No
Assessment of
Revision
v
Adequate
Inadequate
_
Adequate
-
-
_
Potential
Risk
Change
N.A.
N.A.
N.A.
N.A.
None
decrease (I)
unknown (pos-
sibly de-
crease - see
Analysis
section)
decrease (U)
ho
tsj
(1) Sec Reference 2 for discussion of General Observations
(2) See Reference 2
(3) Areas continued from Table I
(4) Comment is listed but not answered
I - Insignificant
U - Undetermined
N.A. - Not applicable
-------
(1)
Aron
20. Conr .iri son
of Risks
!>i-i".i— •
Nuclear and
Other means
of Elec-
trical
eraticn
21. General In-
consisten-
cies
22. Conclusions
£»
1
Isi
U>
deficiency
Idi-ntififd in
Draft Kevicw *2)
a. Comparison between nuclea
and no.i-nui:lear electrical
tK-*cr generation risks is
required
b. Inadequate emphasis that
i.a'^s are fro.a in-plar.t
accidents only
a. Stjdy contains inconsis-
tencies in approach and
level of detail considerec
a. Continuous up-grading of
stJoy requircU
o. Expansion LO incorporate
different leactor con-
cepts reqa^red
Deficiency
Considered Ir
Appendix XI
No
Yes
No
Yes
Yes
1
i Assessment of
>' Apper.cix XI
Response
_
Inadequate
-
Adequate
Adequate
i
Ares Revised
1 In WASH-1400
Final
No
No
Yes
N.A.
N.A.
t
j
i
i
i Assessnenc of
Revision
—
-
Improved
-
-
i
i
i
Potential
Risk
| Change
N.A.
N.A.
N.A.
N.A.
N.A.
(1) See Reference 2 for discussion of Ganeral Observations
(2) See reference 2
-------
T.-3U III - Salary of Apparent Deficiencies :n U.iSH-l'iOO (final)
(1)
Total Number 1 Considered
oJ Deficiencies ir. WASH-1400 (final)
in WASH-1400 i (Appendix XI)
•Vai t
i
Kiiibdr
Considerec
"3 23
i.ssessren:
Adecuate^-) Inadequate
14
9
Revisions T.ade in
WASH-1400 (final)
Sunber
Revised
8
\ssessTent
Adrcuate
4
Inadequate
1
Improved
3
Total Nu-iber of
)eficiencies Re-
naining in WASH- 1400
(final)
34
! !
(1) does not include problems discussed in Section VI.
(2) includes those judged to be partially adequate.
-------
TABLE IV - Summary of Risk SigaificAr.ee of Apparent Deficiencies in WASH-UOO (final)
Total Number of
Deficiencies remaining
in WASH- 1400 (final) with
Potential for change
In Risk
29
Potential Change in Risk(1>
Increase
Significant
1
Insignificant
13
Undetermined
5
Decrease
Significant
0
Insignificant
2
Undetermined
3
Unknown
5
(1 insig-
nificant)
(1) Significant is defined as having the potential for changing the risks by more than a factor of 10.
N9
ui
-------
TABLE V - WASH-1400 (final) Unresolved Areas with Potential Risk Change
Deficient Area
1.
2.
3.
4.
5.
6.
7.
a.
9.
BWR Reactor Protection System: Conservative
Assumption - three adjacent rod failures
cause scram failure
BWR Reactor Protection System: Comnon mode
failure consideration inadequate
PWR Electric Power System Failure: Time and
environment effects on failure rate not con-
sidered
PWR Electric Power System Failure: Common
mode failures due to earthquakes, relay
failure, and breaker location not considered
PWR Snail Break Less of Coolant Accident
Analysis: Assumption of adequate core cooling
for small breaks r.ot justified
Human Reliability Analysis: Inadequate and
inconsistent derivation of specific human
reliability assessments
Common Mode Failures: Insufficient description
of hcvr common mode failures were quantified and
applied
Design Adequacy: Insufficient description of
the extent which design adequacies ware assessed
and applied
Containment Failure Pressure: Surry contain-
ment failure pressure not sufficiently justi-
fied and appeals f>o nign
Section V
Discussion
Page 4_29
Page 4-33
Page 4-39
Page 4-44
Page 4-47
Page 4-52
Page 4-53
Page 4-54
Page 4-58
Potential
Risk Change
decrease - extent unknown
increase - significant
increase - extent unknown
increase - extent unknown
increase - extent unknown
unknown
unknown
unknown
Increase - extent unknown
N>
ON
-------
TABLE V (Continued)
•ts
K)
Deficient Area
10. Containment Pressure Response: Containment
f.illura time should be reduced for CSIS and
CSRS failure
11. Conrnfnr.;ent Pressure Response: Containment
failure tine should be Increased for loss
of electrical power
12. Nuclear Plant Characteristics: Application
of Surry risk analysis to other PWEs not
justified
13. Realistic vs Conservative Assumptions:
Indicated realism not consistently used
Section V
Discussion
Page 4.59
Page 4_6Q
Page 4_67
Page 4-71
Potential
Risk Change
increase - extent unknown
decrease - extent unknown
unknown
decrease - extent unknown
-------
V. ANALYSIS
This section presents the analysis of each of the 45 deficiencies
(9)
found during the review of WASH-1400 Draftv . The analysis con-
sists of determining the extent to which the deficiency still exists
in WASH-1400 (final) . The analysis includes two distinct evalua-
tions for each deficiency. First, a determination was made if the
deficiency was considered in Appendix XI to WASH-1400 (final).
Appendix XI lists and responds to selected comments which were re-
ceived by the Nuclear Regulatory Commission (NRC) subsequent to the
publication of WASH-1400 Draft in August 1974. If the deficiency was
considered and responded to in Appendix XI, an assessment is given
of the adequacy of the response. Second, the area in WASH-1400 Draft
which contained the deficiency was found in WASH-1400 (final) and
reviewed to determine if an adequate revision was made. A discussion
is presented of both the response and revision, if they exist. The
deficiencies are separated into two groups. The first group consists
of deficiencies found during the review of the 16 technical areas
selected for assessment in WASH-1400 Draft. The detailed review of
these areas is contained in Section V, Part 3 of Reference 2. The
second group consists of deficiencies found in the general assumptions
and presentation of results. These deficiencies are contained in
Section VI (General Observations), Part 3 of Reference 2.
The deficiencies, and an analysis of them based on information contained
in WASH-1400 (final), are as follows:
1. BWR Reactor Protection System Failure
A total of four deliciencies were found during the review of the BWR
reactor protection system as described in WASH-1400 Draft. These
4-28
-------
deficiencies, and an evaluation of them based on WASH-1400 (final),
are as follows:
a., Conservative assumptions: three adjacent rod failures cause
scram failure - WASH-1400 Draft assumed that the failure of any
three adjacent control rods to insert in a BWR core resulted in
a scram failure. This assumption was judged to be excessively
conservative and not consistent with the WASH-1400 charter of
performing a realistic analysis.
Appendix XI Response - The deficiency described above (and in
more detail in reference 2) was considered in Appendix XI of
WASH-1400 (final) on pages XI 5-2 and 5-3. The response indi-
cated basically that "the major contributors to scram failure
are common mode failures of scram rods and common mode failures
due to test and maintenance. These common mode contributions
would give essentially the same probability of failure for not
only three rods but also for four or more rods. Within the data
accuracies, then, the total scram probability of 1.3xlO~ applies
to either three or four (or more) rod failures."
To determine the adequacy of the above response, Section 6.2.4.1
("Reactor Protection Control Rod System"), page 11-359 (Appen-
dix II) was reviewed. The contributors to RPS failure are (page
11-358):
"doubles - 3'2xl°~'
"triples = 5'8xl°
"test + main = 2.6xlO~7
Q j " 1.9x!0"6
^common mode
TOTAL 8.0xlO~6
4-29
-------
As can be seen, the Q . - contribution represents over 70%
triples
of the total. The Q. k , . (test and maintenance contri-
test + main
bution) is insignificant (VJ%) and Q , contributes
common mode
less than 25%, and has to do with roiscalibration of switches
that produce a trip signal. There is a common mode contribu-
tion included in the Q . .. value which apparently is the con-
tribution referred to in the Appendix XI response. To determine
how this contribution is affected by a change in the number of
rods that are required to fail to prevent scram, an analysis was
conducted usinr. the WASH-1400 method of incorporating common mode
failures for this case as described on page 11-362. Assuming,
for example, that it takes four adjacent rod failures to prevent
scram rather than three, the assumption of complete independence
yields (page 11-362).
(lxlO~V = lxl(f16
which would be the failure rate for any four rods assuming no
common mode contribution. Using the WASH-1400 (page 11-362) value
for the common mode contribution (1x10 ) and combining the two
values in a log-normal median fashion as was done in WASH-1400
yields,
PA ="\AlxlO~16) (lxlO~6) = lxlO~U,
which is a factor of 100 below the WASH-1400 assessment for P.
given on page 11-362, indicating that the common mode contribu-
tion is not a major contributor to control rods failing to in-
sert, contrary to the contention in Appendix XI. This reduction
will sij-nificantly reduce both cases (a) and (b) which were com-
bined to produi e the Q . .. contribution of 5.8x10
4-30
-------
Page 11-359 of Appendix II, WASH-1400 (final), describes the
assumption of three adjacent control rod failures as leading
to "an extremely conservative analysis" of the reactor protec-
tion systems. This statement seems to conflict with the assess-
ment in Appendix XI described previously. Also, testimony by
representatives from General Electric at a recent ACRS meeting
on ATWS(9)
assumption.
(9)
on ATWS ' tends to confirm the extreme conservatism of the
Analysis Change - No change was found in the WASH-1400 final
analysis compared to WASH-1400 draft relative to the assumption
of three adjacent control rod failures causing scram failure
(Appendix II, page 11-359 et seq.).
Conclusion - The WASH-1400 (final) response to the deficiency
identified from the review of WASH-1400 Draft relative to the
assumption of three adjacent control rod failures is assessed
to be Inadequate based on the foregoing analysis. However, it
does not appear that this assumption results in an overly sig-
nificant influence on the calculated BWR risks since, at most,
it could result in a factor of 4 reduction in the RPS failure
probability since other contributions become dominant. However,
further analysis is required to definitely quantify the effect.
It should be noted that the WAah-itUU method ot accounting £of
common mode failures in the manner described on page 11-362 has
been questioned and is discussed in area I-c, following. These
two areas are quite dependent and changes in one can affect the
results in the other.
4-31
-------
Miscellaneous Comments - Discussions with the General Electric
(4)
Co. tended to confirm the conservatism of assuming any three
adjacent control rod failures cause scram failure. In fact,
based on G.E.'s assessments, the loss of either subsystem A or
B control rods (each subsystem controls 1/2 of the control rods)
will not result in failure to scram for some anticipated tran-
sients. Thus, the common mode failure contribution for each of
these systems (described on page 11-363) may not be properly
assessed.
b. Single control rod failure rate too low - WASH-1400 Draft assessed
-4
the probability of failure for a single control rod to be 1x10
This was assessed to be not substantiated by data included in WASH-
1400, and possibly too low based on an independent assessment by
others. Also, WASH-1400 uses only two failures to compute the
control rod failure probability while Table III-7 in WASH-1400
Draft Indicates six failures occurred during the time period used.
Appendix XI Response - It is stated on page VI 5-3 that additional
control rod failure data for ]973 were analyzed which confirmed
the WASH-1400 value. Only two of six failures listed in WASH-1400
Draft resulted in failure to insert on demand.
Analysis Change - No change to the analysis of single control rod
failure probability was found in WASH-1400 (final) (page 11-362).
-4
Conclusions - It is concluded that the WASH-1400 value of 1x10
for single control rod failure probability is probably valid.
(4)
Discussions with G.E. tended to support the value.
4-32
-------
Miscellaneous Comments - References to Tables III 4-5 and III 5-3
on page XI 5-3 of WASH-1400 (final) are all incorrect.
Common mode failure consideration inadequate - The derivation of
1x10 for the "tight coupling" common mode contribution and the
subsequent log-normal combination of this value with the "complete
-12
independence" (1x10 ) value to obtain the probability of three
control rod failures was considered faulty during the WASH-1400
Draft review. The 1x10 value was derived by multiplying the
-4
basic single control rod failure rate (1x10 ) by 0.1 which, ac-
cording to WASH-1400 Draft, is the approximate common mode failure
frequency. A second factor of 0.1 was used to account for the
fact that only 10% of the common mode failures resulted in complete
failure to insert; the other "failures" resulted in partial degra-
dation. During the WASH-1400 Draft review, it was concluded that
(1) there was insufficient justification provided to support the
assumption that 10% of the control rod failures were common mode,
and (2) based on the derivation of the common mode contribution
(lxlO~ ) in WASH-1400 Draft, it appeared that 1x10 should be used
as the probability of three adjacent rods failing rather than a
_g
value (1x10 ) obtained by combining the common mode contribution
with the completely independent contribution. Using a value of
1x10 produced a factor of 30 increase in the BWR risks.
Appendix XI Response - WASH-1400 (final) (page XI 5-3) did con-
sider this apparent deficiency. The response states, "The 10
value was obtained from the analyses described in Appendix III in
which approximately 10% of all failures could be considered as ap-
proximating common mode behavior. Since all types of components
4-33
-------
were considered in obtaining this 10% value and since many of the
common modes did not cause failure but only minor degradations,
this 10 value was tr
being an upper bound."
this 10 value was treated as very conservative and, hence, as
Analysis Change - No change in the value of the probability of
three adjacent control rod failures was found in WASH-1400 (final)
(Section 6.2.5.1, page 11-362).
Conclusion - The Appendix XI response is judged to be inadequate,
and the assessment of the deficiency from the WASH-1AOO Draft
review is still considered valid. The response seems to indicate
that the 10 value does not include a consideration that many com-
mon mode "failures" result in only "minor degradations," and thus
the value represents an upper bound. However, the "minor degrada-
tion" contribution is said to be accounted for in WASH-1400 by the
additional factor of 0.1 in arriving at the 10 value so that the
upper bound argument is not valid. There is no additional justifi-
cation in the Appendix XI response for the validity of either the
10% value used for the common mode contribution or the additional
10% value used to account for the minor degradation contribution.
However, Table III 3-6 (pages 37 and 38) contained in Appendix III,
WASH-1400 (final), does list common mode effects and causes, and
does illustrate that a 10% contribution from common mode failures
may be reasonable.
Miscellaneous Comment - There appear to be major differences between
the WASH-1400 results of the BWR RPS failure probability and General
(9)
Electric1s assessment. G.E., in recent testimony , indicated that
the RPS failure probability, considering random failures only, was
4-34
-------
-9 -7
2x10 per demand. This compares to the WASH-1400 value of 3x10
obtained by taking out the common mode failure contributions
from the failure mechanisms listed on page 11-358 (Section 6.2.3).
d. Excessive credit taken for manual poison injection - It was con-
cluded from the review of WASH-1400 Draft that (1) the probability
of operator failure to initiate the liquid poison injection system
_2
during an anticipated transient was too low (3x10 ), and (2) for
at least some severe anticipated transients, the liquid poison in-
jection system was too slow to have any effect on the transient.
Appendix XI Response - According to Appendix XI (page XI 5-1), the
WASH-1400 Draft value of 3xlO~2 for the probability of operator
failure to activate the liquid poison injection system was too low.
An increased value of 10 has been assigned in WASH-1400 (final).
Relative to the effectiveness of the liquid poison injection system,
Appendix XI (page 5-8) argues that all anticipated transients are
slow enough to be adequately controlled by activation of the liquid
poison injection system assuming that fuel melting must be prevented.
Further, exceeding primary system pressure and fuel temperature
limits, which are very conservative, does not imply that an accident
has occurred or that a radiological consequence to the public will
result.
Analysis Change - The value for the probability that the reactor
operator unsuccessfully initiates the liquid poison injection system
has been increased to 1x10*" in WASH-1400 (final). The assumption
that the liquid poison injection system can successfully terminate
all anticipated transients has been retained in WASH-1400 (final).
4-35
-------
Conclusion - The WASH-1400 (final) assessment of the BWR liquid
poison injection system now appears adequate relative to the two
areas described in the preceding discussion. Discussions with
G.E. revealed that even for the most severe transients (ie,
main steam line isolation valve closure), the operator has suf-
ficient time (hours) to successfully initiate the liquid poison
injection system provided that recirculation pump trip occurs
(failure of recirculation pump trip was assessed to be negligible
in WASH-1400).
Miscellaneous Comment - It should be noted that, as a result of
Nuclear Regulatory Requirements , the need and possible mechanism
for improving the BWR reactor protection system has been reviewed
by General Electric Co. A report has been submitted to NRC
which contains a proposal to change the BWR Reactor Protection
System by, among other alterations, making the Liquid Poison In-
jection System an automatically controlled system thereby reducing
operator error. This proposal is currently under review by NRC and,
if accepted, will render the WASH-1400 (final) assessment of BWR
RPS failure invalid for those plants which are modified to meet the
proposal.
2. BWR Anticipated Transients
One deficiency was found in this area during the review of WASH-1400
Draft.
a. Inadequate consideration of the frequency and severity of BWR anti-
cipated transients - It was concluded from the review of WASH-1400
Draft that the assumed frequency of BWR Anticipated Transients
4-36
-------
(10/yr) was too high and that the actual frequency of those anti-
cipated transients which require core power shutdown should be 3/yr.
Appendix XI Response - On pages XI 5-8 and 5-9, WASH-1400 (final)
states that "Reactor experience clearly indicates that a frequency
of about 10 transient events per reactor-year must be considered."
Analysis Change - No change in the assessed frequency of BWR anti-
cipated transients was made in WASH-1400 (final) (Appendix V, page
V-36).
Conclusion - Based on a review of Reference 6 as well as discussions
(4)
with G.E. , it appears that the WASH-:
cipated transient frequency is correct.
(4)
with G.E. , it appears that the WASH-1400 assessment of BWR anti-
Miscellaneous Comments - Reference 6 describes nine anticipated
transients which are "...derived for ATWS consideration on the basis
of operational experience and have the potential of a frequency of
occurrence of at least once in four years of reactor operation at
power conditions such that a significant transient results and
scram is called upon to shut down the reactor." Table I 4-12
(page I 91/92, "BWR Transients") of WASH-1400 (final) lists 14 oc-
currences which are called "Likely Initiating Events" (also "likely
transient events" on page 1-64). Eight of the nine transient events
contained in the Reference 6 list are included in the WASH-1400
(final) list. (Inadvertent opening of the safety relief valves is
not included.) It is not clear why these lists are inconsistent.
3. BWR Electric Power System Failure
One deficiency was found in this area during the review of WASH-1400
Draft.
4-37
-------
a. Loss of two diesels rather than four for LPCI failure - It was
determined during the review of WASH-1400 Draft that an incor-
rect assumption was made regarding the number of diesels required
to permit continued operation of the Low Pressure Coolant Injec-
tion System. WASH-1400 Draft assumed that the LPCI system would
have insufficient capacity only if all four diesels were lost
coincident with loss of offslte power. A review of information
contained in WASH-1400 Draft (Appendix II, Vol. 3) as well as
the Peach Bottom SAR ' (Table 8.5.2b) led to the determination
that loss of two diesels would result in loss of LPCI capacity.
Appendix XI Response - None.
Analysis Change - None (Section 6.4.2 of WASH-1400 (final) un-
changed, in this regard, from corresponding section in Draft).
Conclusion - This discrepancy apparently remains in WASH-1400
(final).
4. PWR Electric Power System Failure
Eight discrepancies were found in this area during review of WASH-
1400 Draft as follows:
a. Insufficient consideration given to availability of power at in-
ception of LOCA - WASH-1400 Draft did not consider (1) that the
Technical Specifications, which require reactor shutdown if an
emergency bus becomes unavailable, could be violated, and (2) that
the detectors, indicators, or annunciators which tell the operat-
or that he is without a bus may fail.
4-38
-------
Appendix XI Response - None.
Analysis Change - None (PWR Electric Power System analysis as pre-
sented on page II-8.6 is unchanged from WASH-1400 Draft).
Conclusion - The apparent deficiency still exists in WASH-1400 (final)
b. Time and environmental effects on failure rate not considered -
WASH-1400 Draft did not appear to account for the fact that grounds
and faults on the PWR electric power systems may be affected by aging
and by extended exposure to the relatively harsh LOCA environment.
Appendix XI Response - A general response to the problem of aging
effects on component failures is given on page XI 14-3. The point
is made that the study results apply only to the 100 plants oper-
ating in the next five years, and suggests that the study be re-
peated in about five years. Also, it is stated that "Aging is a
separate question that perhaps could be analyzed when and if data
are available and, more importantly, if the need to do so clearly
existed." An Appendix XI response to another comment (comment 7,
page XI 2-4) also seems appropriate to this issue, although it was
not used as such. On page XI 2-4, it is stated that some of the
failure data examined was derived from actual field operation.
Thus, it is reasonable to assume that aging effects were included
in at least part of the data base as casually related failures.
Regarding LOCA environmental effects on the PWR electric power system
components, a general response is given in Appendix XI, page 2-5 to
the effect that environmental effects were considered in Appendix X
(Design Adequacy). A review of Appendix X reveals that "On-Site
Electric Power System" (item A6.3.6) is listed as an item in
4-39
-------
Table X 2-2 entitled "PWR Component Review - Environmental Quali-
fication Summary." However, for reasons not explained, none of
the columns which describe the extent of the review or the re-
sults are checked. A further investigation of Appendix X re-
vealed that on page X-58 a narration is provided relative to
item A6.3.6 which is renamed "Electric Power Distribution Systems."
This system is said to consist of "transformers, switchgear and
motor control centers, emergency buses, containment penetrations,
and associated electric power and control cables and their support-
ing cable trays." The section next describes separately "Electrical
Cables and Terminations" (A6.3.7.1), "Electrical Containment Pene-
trations and Connectors" (A6.3.7.2), "Cable Trays" (A5.3.7.3), and
"A-C and D-C Switchgear" (A6.3.7.4). None of the other components
said to be Included in the PWR Electrical Power Distribution Systems
is considered further. Appendix X concludes that, for the compon-
ents described separately (Sections A6.3.7.1 through A6.3.7.4), the
electrical cables and terminations with "reasonable assurance" will
perform adequately during and following design basis events at the
plant. For "Electrical Containment Penetrations and Connectors,"
it is concluded that "...their design is adequate assuming that the
results of the LOCA environment tests are positive." Regarding Cable
Trays, it is concluded that "...their design is adequate... ." How-
ever, for "A-C and D-C Switchgear" an assessment of environmental
(as well as seismic) design adequacy could not be made since "No
seismic or environmental qualification information...(was) made
available for review and evaluation."
Analysis Change - No analysis change was made in WASH-1400 (final)
in this area (Section 5.1, Appendix II).
4-40
-------
Conclusion - The effects of aging on PWR electrical power system
failures are probably adequately considered in view of the Appen-
dix XI discussion on this and related matters, especially if the
study is repeated in five years, and further consideration is
given to this area as suggested in Appendix XI. It is judged
that the LOCA environmental effects are not adequately considered
in view of the Appendix X discussion, and especially due to the
(2)
poor experience in this area as described in the Draft review .
c. Probability too low for operator opening wrong breakers - It was
_A
judged during the review of WASH-1400 Draft that the value of 10
assigned to the probability that the operator, under stress, would
inadvertently open breaker 15H3 or 15H8 is too low.
Appendix XI Response - None specific to this area. However, the
general topic of human reliability is discussed on pages XI 2-40
and XI 14-1.
-4
Analysis Change - None. The value of 10 was used in WASH-1400
(final). (Ref: Fig. II 5-4, page 11-261 and 11-262, also Table II
5-3, page 11-203.)
-4
Conclusion - The value of 10 assigned to the probability of the
operator inadvertently opening breaker 15H3 or 15H8 in WASH-1400
(final) appears too low.
d. Evaluation of d.o. power unavailability inadequate - It was determined
from the review of WASH-1400 Draft that the probability of loss of
_3
the on-site d.c. power supply (assigned a value of 10 ) was not si
ficiently justified, particularly in view of the fact that loss of
4-41
-------
d.c. power would interrupt power to the safety systems requiring
power to operate independent of the availability of a.c. power.
Appendix XI Response - None.
Analysis Change - None (pages 11-81 through 11-95).
Conclusion - The assessed probability for d.c. power unavailability
(10~3) used in WASH-1400 (final) is not sufficiently justified.
Miscellaneous Comments - It is stated on page 11-92, WASH-1400
(final) that "For the d.c. buses, unavailability (of power) at LOCA
is represented by the coincident loss of offslte power and the
station battery." This statement is not entirely true based on a
review of the emergency power system which indicates that loss of
the d.c. power supply will Interrupt power to the safety systems
even if offsite power is available. Fault trees contained in
WASH-1400 (final) (Sheets 1 through 3 of Figure II 5-4, pages
II 255 et seq.) correctly show this relationship.
e. Insufficient consideration given to operator error (opening wrong
breakers) ~ It was concluded from the review of WASH-1400 Draft
that several breakers not considered could be inadvertently opened
by the operator failing critical portions of the PWR electric power
system during LOCA.
Appendix XI Response - None.
Analysis Change - None (pages 11-81 through 11-95).
Conclusion - The apparent deficiency still exists in WASH-1400 (final).
4-42
-------
f. Maintenance errors resulting in power bus unavailability not
considered - It was determined during review of WASH-1400 Draft
that maintenance errors, such as Improperly racking breakers in
or miscalibration of relays did not appear to have been consider-
ed in assessing the PWR electric power system failure.
Appendix XI Response - A general response applicable to this
comment, although not specifically addressed to it, appears on
page XI 2-4 and page XI 14-2. These responses generally indicate
that maintenance Induced errors have already been included in the
data base because it is derived for the most part from field ex-
perience.
Analysis Change - None (pages 11-81 through 11-95).
Conclusion - It is concluded that maintenance errors of the type
described have probably been accounted for In the manner described
in Appendix XI.
g. Possible incomplete consideration of diesel generator unavail-
ability - From the WASH-1400 Draft review, it could not be deter-
mined whether all of the concerns expressed in Appendix II, Vol. 2
(pages 35-38) regarding the inadequacies relative to diesel gener-
ator operation and reliability were appropriately considered in
the fault tree analysis of the system. In view of the importance
of the diesel generators as a source of emergency power during
LOCA, it was concluded that WASH-1400 should have included a more
complete description of how these inadequacies were considered.
Appendix XI Response - None.
4-43
-------
Analysis Change - None (pages 11-90 through 11-92).
Conclusion - This apparent deficiency still exists in WASH-1400
(final).
h. Common mode failures due to earthquakes, relay failure, and
breaker location not considered - The following three apparent
common mode deficiencies were identified as a result of the re-
view of WASH-1400 Draft: (1) In view of the fact that, accord-
ing to Appendix X, neither the a.c. nor d.c. switchgear could be
assessed as to seismic design adequacy, common mode failure of
the PWR electric power systems due to seismic events appears not
to have been adequately considered, (2) A relay failure which
could cause tripping of both the offsite and onsite power source
appears not to have been considered, (3) A fireball, generated
by a short circuit across a load breaker, could cause nearby cir-
cuit breakers to trip open. It does not appear that this poten-
tial has been evaluated.
Appendix XI Response - None specific to these areas. However,
the general method by which seismic caused common mode failures
were assessed is explained on page XI 15-2. Common mode failures
in general are also discussed on pages XI 2-4, XI 3-1 (et seq.)
and in Addendum I to the main report.
Analysis Change - None.
Conclusion - The consideration of electric power failure due to
seismic events appears to have been adequately considered in
WASH-1400 (final) based on the discussion contained on page XI 15-2.
4-44
-------
The other two potential common mode failures have apparently-not
been considered In WASH-1400 (final).
5. PWR High Pressure Injection System (HPIS)
Three discrepancies were found In this area during review of WASH-
1400 Draft as follows:
a. Operator reliability to open water valves to lube oil coolers
too high - It was concluded from the review of WASH-1400 Draft
that (1) the probability of operator failure to open the service
water valves to the standby HPIS pumps, If required, was too low
_3
(1x10 ) and (2) the probability that the operator falls to open
the service valve to the lube oil heat exchanger for the second
_3
standby pump given that he falls to open the first (10 ) should
be 1. Both of these apparent deficiencies were based In part on
Information contained In Appendix III, Section 6.1.
Appendix XI Response - None. However, the general area of human
reliability Is reviewed on page XI 14-1.
Analysis Change - None (Table II 5-21, page 11-230).
Conclusion - The deficiency appears to remain in WASH-1400 (final).
b. Lou Pressure Injection System (LPIS) check valve failure not ade-
quately considered - It was postulated from the review of WASH-
1400 Draft that a failure of any one of the check valves which are
contained in each of the three lines connecting the HPIS to the
LPIS would disable the HPIS during a small break LOCA. WASH-1400
4-45
-------
Draft assumed that such a check valve failure would disable only
one injection line, and the second line (to the remaining intact
loop) would still be operable. In view of the fact that all LPIS
fluid flows through a common header before branching out to the
three lines (one to each primary system loop) it appears that fail-
ure of any LPIS check valve will divert most of the HPIS flow to
the LPIS system.
Appendix XI Response - None.
Analysis Change - None (Section 5.6.4.4.3, page 11-146).
Conclusion - This deficiency appears to exist in WASH-1400 (final).
c. Additional double failures not included - Failure of both valves
in (1) volume control tank drain3 (2) normal charging line, (3)
boric acid recirculation system - It was determined from the review
of WASH-1400 Draft that the three double failure combinations listed
were not included in Section 5.10 which analyzes the HPIS.
Appendix XI Response - None.
Analysis Change - None (item b, page 11-146).
Conclusion - The deficiency remains in WASIPT400 (final).
6. PWR Small Break Loss of Coolant Accident Analysis
The following assessment resulted from review of WASH-1400 Draft:
4-46
-------
Assumption of adequate core cooling for small breaks not justi-
fied - It was concluded from the review of WASH-1400 Draft that
the (implied) assumption of adequate core cooling for all small
breaks assuming that emergency core cooling (ECC) systems do not
fail was not justified. This conclusion was based on a rather
extensive survey and comparison of analytical results from small
break analyses performed by the PWR vendors.
Appendix XI Response - On page XI 2-5, in response to the above
assessment, it is stated that "...the peak temperatures predicted
by different PWR reactor manufacturers vary between 1100 and 14OOF
...," and that "...(this) temperature range...is well below the
NRC peak clad temperature limit of 2200F. Thus, from the viewpoint
of reactor accident risk assessment, the study believes that the
expressed concern in this area is not germane."
Analysis Change - None (page 1-44 et seq.).
Conclusion - The Appendix XI response is incorrect in stating a
temperature range between 1100 and 1400F. The actual range found
during the review and quoted in Reference 2 was from 1075 to 1740°F.
A recent calculation shows a maximum temperature of 1657°F for
2
a break size of 0.5 ft , a size which was not selected to produce
the maximum temperature. In addition, it was concluded during the
WASH-1400 Draft review that "The fact that in none of the calcula-
tions do core temperatures become great enough to cause damage that
may lead to melting provides some confidence that adequate core
cooling may be achieved. However, the large differences in results
indicate that the processes occurring during the accident are not
well understood, and that significant effects may be overlooked or
improperly considered. In view of the fact that the small break
4-47
-------
LOCA is a dominating contributor to public risk from PWRs, it
is imperative that a substantial justification be provided to
establish that ECC systems are adequate for small break LOCAs*4*
It is concluded that the assessment of this area as applied to
WASH-1400 Draft applies to WASH-1400 (final). For a related
discussion concerning ECCS functlonability, see Section VI;
"Additional Comments."
7. PWR Loss of Power Transient Accident Sequence
The following area was found to be inadequately assessed during the
review of WASH-1400 Draft:
ft. Containment failure sequence - It was concluded during the review
of WASH-1400 Draft that the possibility of containment rupture
when accumulator water impacts the molten core mass in the reactor
vessel cavity following a loss-of-power transient was overlooked
in WASH-1400 Draft. The rupture could result from either over-
pressurization from steam generation or damage from vessel head
impact as a result of a steam explosion.
Appendix XI Response - Page XI 12-1 states that "When water is
introduced at the top of the melt...the potential for the coherent
interaction of a large quantity of molten material with water is
much smaller since the water cannot readily penetrate into or dis-
place the high-temperature melt." On page XI 3-58, a discussion
is given relating to the protection to the containment afforded
by the overhead crane from missiles such as the reactor vessel
head. It is concluded that "Since the crane is always present
4-48
-------
over the vessel centerline, the study concluded that the prob-
ability of such missiles (including the vessel head) leading to
a breach of the containment is negligibly small."
It is stated on page VIII-13 that "Steam explosions in the re-
actor cavity are not expected to threaten containment integrity
since the resulting pressures within the containment free volume
are predicted to be small compared to the design levels..." How-
ever, it appears that this assessment applies to the double-
ended pipe break LOCA and not to the loss of power accident dis-
cussed in the preceding discussion wherein two large containment
pressure loadings may be superimposed. In the double-ended pipe
break LOCA, the potential for reactor vessel steam explosion comes
long after the end of blowdown and the containment pressure will
have been substantially reduced. In the loss of power transient
accident, it was postulated in the review of WASH-1400 Draft that
the containment pressure loading from a steam explosion in the
reactor vessel cavity comes immediately after the blowdown loading
from vessel melt-through.
Analysis Change - None.
Conclusion - It is concluded that the WASH-1400 (final) response
is inadequate regarding the likelihood of a steam explosion from
the postulated molten core-accumulator water interaction. In the
first place, the rate of water discharge onto the core melt will
be quite high for the conditions postulated (several tons per
second), and displacement and penetration of the melt by the water
leading to coherent interaction (mixing) would appear likely.
Second, the East German Slag Incident described on page VIII-78
of WASH-1400 (final), wherein two separate explosions were caused
4-49
-------
by spraying water on top of molten slag, seems to indicate that
explosions under the conditions postulated cannot be dismissed.
Relative to the containment protection afforded by the overhead
crane, there is in all PWRs a polar crane bridge located above
the reactor vessel head. However, since the crane bridge is
relatively narrow, and is located a considerable distance above
the reactor vessel head, it is not clear how the bridge will be
able to stop all conceivable reactor vessel head trajectories.
In addition, there appear to be ambiguities on this matter within
the Study. On page V-30 (Example 1, stated to be a PWR accident
sequence on page V-30), it is stated "This (reactor vessel steam)
explosion may have sufficient energy to fail the vessel and have
the upper part of the vessel penetrate the containment structure."
Also, on pages VII1-13 and VIII-17 (et seq.), there appears to be
an implication that containment vessel damage by missiles from
reactor vessel steam explosions is possible. At least there is
no mention of the inhibiting influence of the polar crane bridge.
The overall conclusion to this problem is that the response in
UASH-1400 (final) is considered to be partially adequate.
8. BWR-PWR Component Failure Modes and Rates
The following deficiency was identified during the review of WASH-1400
Draft:
a. Assessed failure probability not consistent with data in several
areas - It was determined that the range and assessed median for
some component failure rates used in WASH-1400 Draft were not con-
sistent with the data. The components were pipes, pumps, and
diesels.
4-50
-------
Appendix XI Response - It is stated on page XI 14-2, in answer
to another comment, that
"As explained in Appendix III, the ranges assigned to
the data are not deterministic bounds and therefore do
not necessarily include all the source data. Thus, all
source data need not fall within the assigned ranges.
(It should be noted that in the calculations, the log-
normal distributions themselves were used, and not the
ranges.) Also, as explained In Appendix III, the ranges
and distributions were not derived from simple empirical
fits but involved some subjective judgments and decisions.
Sensitivity studies were performed to investigate possible
additional variations in the components mentioned, and
few significant effects were obtained."
Analysis Change - Some of the problems cited were eliminated in
Appendix III.
Conclusions - The explanation as given in Appendix XI and Appen-
dix III for why the assessed component failure rates do not, in
several cases, correspond to the rates from various data sources
appears reasonable and may be valid. However, since few details
were given relative to how the specific omponent failure rates
were derived from the data, and the justification for adjustments
which were made to «fehe data, it is not possible to assess the
validity of some of the failure rates used. Additional explana-
tion regarding the derivation of some of the failure rates (par-
ticularly pipe ruptures, pump failures, relief valve failures,
and diesel failures) appears to be needed.
4-51
-------
9. Human Reliability Analysis
During review of WASH-1400, the following general deficiency was found
in the treatment of human reliability:
a. Inadequate and inconsistent derivation of specific human relia-
bility assessments - it was determined that some of the specific
human reliability values used in parts of WASH-1400 Draft were not
adequately or consistently derived and insufficient explanation was
included to determine their validity.
Appendix XI Response - On page XI 2-4, it is stated that "There
is sufficient generalized information on human behavior to permit
a valid quantification of human error probabilities for use within
the accuracies needed for risk assessment." On page XI 14-1, it
is concluded that "The study also believes, however, that more
effort in the future devoted to a better understanding and model-
ing of human reliability factors would be useful.
Analysis Change - None (Appendix III and related areas).
Conclusion - Deficiency still exists since no additional information
was provided in support of the specific human error probabilities
used in WASH-1400 (final).
4-52
-------
10. Common Mode Failures
During review of WASH-1400 Draft, the following deficiency was noted:
a. Insufficient description of how common mode failures were quan-
tified and applied - It was concluded from the review of WASH-
1400 Draft that it was not possible to determine if common mode
failures were properly considered and accounted for. This was
due in part to a lack of a clearly defined, systematic approach
to identifying and evaluating common mode failures.
Appendix XI Response - In responding to other criticisms of WASH-
1400 Draft similar to that described above, Appendix XI includes
substantial discussion of how common mode failures were handled
and their general significance. These discussions are included
on page XI 3-22 (Section 3.1.2) and in various parts of Section
XI 3-1.
Analysis Change - None apparent (Appendix IV and related areas).
Conclusion - While some clarity has been provided in WASH-1400
(final) relative to the general treatment of common mode failures,
the deficiencies described above were not significantly improved.
Thus, by and large, the deficiencies still exist in WASH-1400
(final).
11. Design Adequacy
During review of WASH-1400 Draft, the following deficiency was noted:
4-53
-------
a. Insufficient description of the extent to which design adequacies
were assessed and applied - It was determined from the review of
WASH-1400 Draft that the completeness of the design adequacy study
could not be assessed. In addition, the application of the results
of the design adequacy study was very obscure.
Appendix XI Response - A general response relative to this issue
is presented on page XI 2-4, which indicated that appropriate
adjustments were made due to the inadequacies found in Appendix X.
Analysis Change - None (Appendix V; some minor changes made in
Table V 2-2).
Conclusion - Insufficient information exists in WASH-1400 (final)
to conclusively determine the extent to which design adequacies
were assessed and applied. Thus, the deficiency remains.
12. PWR Low Pressure Injection System (LPIS) Failure
Th€» following deficiency was found during review of WASH-1400 Draft:
a. Incorrect value assigned to probability of operator closing second
(wrong) valve - It was determined from the WASH-1400 Draft review
that the value (1x10 ) assigned to the probability of the oper-
ator closing the second incorrect valve (B01 or B03) should be 1.
Appendix XI Response - None.
Analysis Change - None (Appendix II, Figure II 5-32, page 11-293,
and Table II 5-16, pages 11-218 and 11-219).
4-54
-------
Conclusion - The deficiency remains in WASH-1400 (final).
Miscellaneous Comment - It is not clear why the probability
_3
of the operator incorrectly closing valve B01 (1x10 ) is dif-
-4
ferent than closing B03 (3x10 ) as shown on Table II 5-16,
page 11-218 (a similar difference exists for values A01 and A03).
13. PWR Low Pressure Recirculation System (LPRS)
During review of WASH-1400 Draft, the following four deficiencies were
found:
a. Operator error: incorrect probability for failure to open valves -
It appeared, based on information contained in WASH-1400 Draft,
that a value of 10 should have been used for the probability of
operator failure to open either valve 1860A or 1860B rather than
3xlO~3.
Appendix XI Response - None.
Analysis Change - None.
Conclusion - The deficiency still exists in WASH-1400 (final)
(page 11-182).
b. Operator error: no consideration of single failure valve closure -
It appeared from the review of WASH-1400 Draft that "operator error,
operator closes MOV 1890C" should have been included as a single
failure under "Single Failures which can cause Insufficient LPRS
Flow."
4-55
-------
Appendix XI Response - None.
Analysis Change - None (Fig. II 5-6, page 11-325).
Conclusion - This apparent deficiency exists in WASH-1400 (final).
c. Valve faults incorrectly placed and counted twice - The valve faults
(B03 and B02 failures) were determined during review of WASH-1400
Draft to be incorrectly placed under the top event "Pump B01 Fails
to Start and Pump A01 Fails to Continue Running." Further, it was
determined that these faults should appear under LPRS suction and
discharge line faults where they did not exist. Also, they incor-
rectly appeared again under "Pumps A01 and B01 Discontinue Running."
Appendix XI Response - None.
Analysis Change - None (Fig. II 5-64, page 11-325).
Conclusion - The discrepancy exists in WASH-1400 (final).
d. "Failure due to operator switching to hot leg recirculation after hot
leg break not considered - It was assumed in WASH-1400 Draft (page
490 App. II, Vol. 2) that "...although no written procedures were
found to be available, it was assumed that at some time during the
first day following a cold leg break the LPRS should be realigned
to inject into the hot legs..." It was determined that, if a hot
leg break occurs, and the operator, assuming a cold leg break,
switches to hot leg circulation, LPRS failure may ensue for the
same reasons as stated on page 490, App. II, Vol. I of WASH-1400
Draft. Such an occurrence was not considered.
4-56
-------
Appendix XI Response - None.
Analysis Change - None.
Conclusion - The omission of this consideration still exists in
WASH-1400 (final).
14. Core Meltdown Analysis
The following conclusion was derived from the review of WASH-1400
Draft:
a. Time to core melt excessive - Based on an independent analysis of
the time required for a PWR core to reach fuel melting temperatures
following a LOCA, it was determined that the WASH-1400 assessment
was some 11 minutes too long.
Appendix XI Response - On page XI 2-6, it is stated that the poten-
tial time differential determined above will not influence risks
because (1) isotopes which are large contributors to the accident
consequences have half-lives greater than one day (and thus radio-
active decay for the 11-minute change in release will be negligible)
and (2) the evacuation mode is insensitive to small variations in
available time.
Analysis Change - None.
Conclusion - The WASH-1400 assessment of the significance of this
time change appears to be correct. However, the time to melt in
4-57
-------
WASH-1400 (final) is unchanged, and the difference still exists.
It is also possible that the difference in core melt time could
alter the accident sequence and the mode of containment failure.
15. Containment Failure Pressure
It: was determined from reviewing WASH-1400 Draft that:
a.. The Surry containment failure pressure was not sufficiently justi-
fied and appears high - An independent analysis of the Surry con-
tainment failure pressure determined that a failure pressure of
67.5 psig was as high as could be justified. The value used in
WASH-1400 Draft is 85 psig.
Appendix XI Response - On page XI 2-7, WASH-1400 (final) states:
"The containment failure pressure of 100 psia determined by the
study represents a nominal failure pressure for the containment.
A containment failure probability of 0.5 was assigned for the cal-
culated pressure of 100 psia. The containment failure probability
was represented as a continuous variable with a normal distribu-
tion about this value.
"It should be added here that the ITI report (contained in Refer-
ence 2) recommended a value of 67.5 psia (sic) for the minimum
failure pressure. This is roughly equivalent to the 2o lower
bound of 70 psia used in the study. Appendix E to Appendix VIII
has been rewritten to better clarify the approach taken and the
rationale behind the nominal failure pressure selected." (The
value of 67.5 psia is incorrectly quoted from the ITI report.
It should be 67.5 psig.)
4-58
-------
-Appendix E (page VIII-133) has been rewritten in WASH-1400 (final)
to clarify the approach.
Analysis Change - None (Appendix E to Appendix VIII).
Concluaiona - The failure pressure selected in WASH-1400 (final)
(100 psia) is considered to be, based on arguments presented in
Reference 2, unjustified and too high.
The justification for the 2o lower bound discussed in Appendix XI
(and Appendix E to Appendix VIII) is not discussed. Usually, a
standard deviation is obtained from data, and since no containment
failure tests have been run, there are no data. The ultimate
strength referred to in Appendix E of 140 psia is based in part
on the ultimate strength of the steel. The concrete will un-
doubtedly fall at a pressure much lower than that required to
strain the steel enough to develop its ultimate strength.
16. Containment Pressure Response
Based on an independent analysis of the PWR containment pressure response
following a LOCA, the following conclusions were reached compared to
similar analyses contained in WASH-1400 Draft.
a. Containment failure time should be reduced for containment spray
injection system (CSIS) and containment spray recirculation system
(CSPS) failure - It was calculated, based on methods and assump-
tions described in Reference 2, that the PUR containment failure
time would be 63 minutes after the initiation of the LOCA with
failure of the CSIS and CSRS safety functions. The failure time
calculated in WASH-1400 Draft for this case was 230 minutes.
4-59
-------
Appendix XI Response - None.
Analysis Change - None (Fig. VIII 2-6, page VIII-23).
Conclusion - The containment failure times have not been changed
in WASH-1400 (final). The change could alter the calculated
sequence and mode of containment failure.
b.. Containment failure time should be increased for loss of electrical
power - It was calculated, based on methods and assumptions de-
scribed in Reference 2, that the PWR containment failure would
occur about 100 minutes later (about 160 minutes after inception
of LOCA) than that assessed in WASH-1400 Draft.
Appendix XI Response - None.
Analysis Change - None (Fig. VIII 2-7, page VIII-24).
Conclusions - The containment failure time has not been changed
in WASH-1400 (final). The change could alter the calculated
sequence and mode of containment failure.
The following areas consist of the second group of deficiencies described
in the introduction of this section. These deficiencies are mainly re-
lated to general assumptions made in WASH-1400 and also to how the results
are presented and evaluated. A detailed discussion of each area can be
found in Part 3, Section VI of Reference 2. An assessment of the extent
to which these deficiencies have been considered in WASH-1400 (final)
is as follows:
4-60
-------
17. Comparative Risk Curve
The following deficiencies were determined to exist as a result of the
review of WASH-1400 Draft in the area of presenting risk comparisons
between 100 nuclear power plants and man-caused accidents:
a. Calculated vs actual risks not distinguished in comparative curves
It was concluded from the review of WASH-1400 Draft that the com-
parative risk curves contained in the Summary Report and in the
Main Report should indicate that the nuclear risks were calculated
while the other man-made risks were based on actual data.
Appendix XI Response - None, ('ihis is listed, along with others,
as comment 4 on page XI 2-2, but is not discussed in the corres-
ponding response.)
Analysis Change - None (Fig. 1-1, Main Report and Executive Sum-
mary) .
Conclusion - While it becomes immediately obvious upon reading
WASHr-1400 that the nuclear risks are calculated and the other man-
caused risks are based on data from experience, the curves which
present these risks do not make the distinction. As long as the
curves are kept in the context of the WASH-1400 (final) report,
this lack of distinction in the curves is not important. However,
these curves have been and will undoubtedly be reproduced in many
other documents and publications in which the accompanying text
often does not make the distinction. It is considered important
that these curves clearly contain sufficient information such that
4-61
-------
the possibility of misinterpretation is minimized. It is thus
concluded that the deficiency still exists in WASH-1400 (final).
b. Calculational uncertainties not sufficiently emphasized - it was
concluded that the calculational uncertainties for nuclear plant
risks quantified in WASH-1400 Draft should be clearly indicated
on the risk comparison curves. Such curves in the Summary and in
the Main Report did not contain any indication of calculational
uncertainties for nuclear risks.
Appendix XI Response - It is indicated on page XI 2-3 that the
curves in the final report have been modified to show risk un-
certainties for nuclear risks.
Analysis Change - Footnotes have been added to the curves in the
Executive Summary (which is reprinted in the Main Report) to in-
dicate the assessed uncertainties in the nuclear risk curves.
Conclusions - The addition of the footnotes is considered an ac-
ceptable resolution to this problem. However, a better resolution
would be to show the uncertainties on the risk curve itself as is
suggested and illustrated in Reference 2.
c. Latent deaths not sufficiently emphasised - it was determined from
the review of WASH-1400 Draft that the calculated number of latent
fatalities were not sufficiently emphasized in presentations and
discussions of nuclear risks. In particular, it was pointed out
that: (1) the abscissa for the risk curves given in the Main
Report (page 153) and in the Summary are labeled merely "fatalities."
However, the fatalities depicted in these curves are only acute
4-62
-------
fatalities, and they were derived directly from curves in Appen-
dix VI which are labeled "acute fatalities;" (2) the risk dis-
cussions in the Summary Report to WASH-1400 Draft exclude any
mention of latent fatalities even though they are calculated to
exceed the acute fatalities by large factors.
Appendix XI Response - On page XI 2-3, it is stated that the mean-
ingful comparison of latent and acute fatalities "...is, in fact,
a troublesome matter that was considered with some care by the
study." The response goes on to state:
"The problem of placing radiation-induced latent cancer fatalities
in perspective is especially difficult since it is well known that
there are latent cancer fatalities attributable to many causes
(air pollution, chemical agents, etc) in our society. Although
there are sufficient data available to create models that can pre-
dict, albeit with some uncertainty, latent effects due to irradia-
tion, there is not sufficient information to do so for other
carcinogenic agents. Thus the study chose, as indicated in sec-
tion 5.5.4 of the Main Report, to compare the various radiation-
induced latent effects with the normal incidence of similar effects.
For instance, in connection with latent cancer fatalities, it is
shown that the numbers predicted due to potential nuclear reactor
accident (sic) represent a small fraction of the normal incidence of
cancers due to other causes. While in this type of comparison
potential latent effects from nuclear accidents are contrasted
with those occurring principally from non-nuclear, environmental
causes, the comparison provides some degree of perspective and is
considered fair because some epidemiologists have estimated that
4-63
-------
the majority of normally occurring cancer fatalities are due to
environmental causes."
Analysis Change - Footnote 1 (page 1) of the Executive Summary has
been added to indicate that the deaths indicated in the figures
are for "those predicted to occur within a short period of time
after the potential accident."
Conclusion - The response in Appendix XI is considered to be a
valid assessment of the latent vs acute death situation. It is,
indeed, a "troublesome matter" with no obvious easy resolution.
However, It is still concluded that WASH-1400 is deficient in not
sufficiently emphasizing latent deaths, for the following reasons:
1. The comparative risk curve abscissas are still labeled merely
"fatalities," even though they include only acute (early)
fatalities.
2. The discussion of the overall risks contained in the Executive
Summary (repeated in the Main Report) includes very little dis-
cussion of latent fatalities.
3. A unique feature of reactor accidents compared to other acci-
dents used in the risk comparisons is the very large latent
deaths to acute death ratio. This ratio is calculated to be
even larger in WASH-1400 (final) than it was in the Draft.
For example, according to Table XI 4-1 (page XI 4-1), there
are calculated to be, for the worst reactor accident, more
than 10 latent fatalities for every acute. This is substan-
tially greater than computed in WASH-1400 Draft.
4-64
-------
Part of the difficulty may be in the apparent inconsistent and
vaguely defined use of fatalities, illnesses and other latent
effects. For example, Table XI 4-1 lists "early fatalities,"
"early illnesses," and "latent cancer fatalities" as categories.
The comparisons on page 9 of the Executive Summary have categories
of "Fatalities," "Injuries" and "Latent Fatalities." The Tables
on page 10 use categories of "Fatalities," and "Latent Cancers."
The Main Report (page 71 et seq.) uses "early fatalities (defined
as occurring within one year after the accident)," "early illness-
es," and "latent cancer fatalities;" Appendix VI uses these defi-
nitions also. It is not clear what the term "injuries" used by
itself on page 9 of the Executive Summary means (early illness?
cancers? genetic effects? etc).
Also, the use of "fatalities" (unmodified) appears to mean only
"early fatalities," but it is not clear that this is always the
case. Are "latent fatalities" and "latent cancer fatalities"
synonymous? How many "latent cancers" result in "latent cancer
fatalities?"
In conclusion, while WASH-1400 (final) does increase somewhat the
emphasis on latent fatalities from nuclear power plants, it appears
that more could and should be done. (See item 2, SectIon*VI for
a more extensive discussion of this matter.)
4-65
-------
d. Transit-ion from solid to dashed lines in man-caused risks does not
correspond to data - It was determined from the review of WASH-
1400 Draft that the dashed line portion of the man-caused risks
do not correspond to the extrapolated portion of the curve. It
was concluded that the curves should be revised to clearly in-
dicate the point at which extrapolation commences.
Appendix XI Response - None.
Analysis Change - None (Chapter 6 of Main Report).
Conclusion - Deficiency exists in WASH-1400 (final).
18. Nuclear Plant Characteristics
The following areas relative to the characteristics of the 100 plants
to be operating in 1980, and to which the WASH-1400 results are said
to apply, were found to be deficient as a result of the review of
WASH-1400 Draft:
a. Distribution of plant type not considered - WASH-1400 Draft used
a simple averaging method to determine the risks from the PWR and
BWR population (100 total plants) expected to exist in the year
1980. The method did not account for the fact that of the 1980
reactor plant population, PWRs will account for about 2/3 of the
total. It was concluded that the calculation of risks from both
plant types should be revised to reflect the preponderance of PWRs
in the 1980 population.
Appendix XI Response - None.
4-66
-------
Analysis Change - The risk curves have been revised to account
for the larger number of PWRs in the 1980 plant population (Page
72 of Main Report).
Conclusion - The deficiency has been corrected in WASH-1400
(final).
Miscellaneous Comment - On page XI 17-1 of Appendix XI (final),
it is stated that the 100 plants considered in the study consist
"...of approximately equal numbers of PWRs and BWRs." In view
of the preceding discussion, this statement is incorrect and is
Inconsistent with the Main Report.
b. Risks computed baaed on incorrect power level - WASH-1400 Draft
assumed, in order to establish a radioactive source term for the
risk calculations, that all of the 100 power reactors would be
operating at 3200 MW. The actual average design power level of
the plants was determined to be 2400 MW for the BWRs and 2650 for
the PWRs, and it was thus concluded from the review of WASH-1400
Draft that the actual design power levels should have been used
in the risk assessments.
Appendix XI Response - None.
Analysis Change - None (Appendix XI, page 11-1, response to Com-
ment 11.4).
Conclusion - This deficiency exists in WASH-1400 (final).
c. Application of Surry risk analysis to other PWEs not justified - The
Surry reactor is a 3-loop Westinghouse design PWR. It was determined
4-67
-------
during the review of WASH-1400 Draft that the 3-loop Surry plant
design used to assess PWR risks is typical of only 14 of the 67
PWRs projected to be operating by 1980. There were found to be
5 separate PUR designs of which the Westinghouse 4-loop design
was the most numerous (22 plants). It was concluded that the
Surry risk analysis could not be assumed, a priori, to apply
across the board to all PWR designs as was done in WASH-1400.
In order to obtain some indication of the extent to which the
Surry risk analysis may apply to the more numerous Westinghouse
4-loop designs, a comparison between the two plants was made as
part of the WASH-1400 Draft review. The differences between the
two plants were investigated in six areas as appropriate to the
risk assessments involved. Table VI shows the results of the
comparison. The first column indicates the area investigated.
The second column Indicates whether or not the differences between
the plants were extensive enough that the Surry analysis did not
appear applicable to the Westinghouse 4-loop design as represented
(8)
by the Trojan reactor. The third column shows which systems
were found to be similar enough such that the Surry analysis would
apply, with minor or readily identifiable changes, to the Trojan
plant. The last column gives a qualitative indication of whether
the difference between the two plants would result in an increased
or decreased risk from the 4-loop design. As indicated in Table VI,
of the six systems analyzed, three were found to be similar enough
between the two plants such that the Surry results were judged,
with some modification, to be applicable to Trojan. Three areas
were found in which the applicability was judged not to exist.
For three of the areas, enough analysis was done to indicate quali-
tatively what the risk change would be in considering the 4-loop
4-68
-------
TABLE VI - Application of Surry Risk Analysis to Trojan
*»
o\
VO
Area Composed
Electric Power System Failure
High Pressure Injection
System Failure
Low Pressure Injection
System Failure
Low Pressure Recirculation
System Failure
Containment Failure Pressure
Containment Pressure Response
Does Not Appear
Applicable to
Trojan
X
X
X
Appears Generally
Applicable to
Troj an
X
X
X
Potential Risk
Change for
Tro J an
Decreased
Increased
Decreased
Undetermined
Undetermined
Insignificant
-------
design. Two areas were judged to result in a decreased risk and
one to result in an increased risk.
An analysis to indicate the applicability of the Surry risk as-
sessment to plants manufactured by Combustion Engineering and
Babcock and Wilcox (34 percent of the 1980 plants) was not at-
tempted.
Appendix XI Response - On page XI 2-11, it is stated, in response
to an Advisory Committee on Reactor Safeguards comment, that
"While the study believes that the extrapolation of the results
of the analysis of two reactors to the first 100 large light-water-
cooled plants is generally valid and that the data base used in
WASH-1400 for estimating accident sequence probabilities is ade-
quate for the purpose intended, draft WASH-1400 made the following
suggestions, as indicated in Section 7.4.2 of the Main Report:
1. It would be useful in the future to pursue the variations
in design from plant to plant and from site to site that
could potentially affect the applicability of the WASH-
1400 results to 100 reactors.
2. It would be useful to collect more data on nuclear plant
operating experience for use in future reliability and risk
assessments."
A similar discussion is given on page XI 17-1 of WASH-1400 (final),
where it is also stated that "...Chapter 7 of the Main Report
discusses in some detail the validity of the extrapolation to 100
reactors and suggests that such extrapolation is likely to be con-
servative.. ."
4-70
-------
Analysis Change - None.
Conclusion - The WASH-1400 contention that the use of the Surry
risk analysis is probably conservative when applied to the 1980
population of 100 reactors appears to be valid. However, until
the other reactor designs are analyzed in detail, the validity
of the contention cannot be substantiated. It is considered
likely that, for the purposes of the study and within the assessed
error bounds, the application of the Surry risks to the 100 plants
is acceptable. On the other hand, as the study recommends, it is
considered necessary to investigate the other plants to determine
which risk assessment areas are different than Surry and make ap-
propriate analyses of these areas.
Miscellaneous Comment - The second recommendation in Appendix XI
(No. 2 above) is not in Section 7.4.2 of WASH-1AOO Draft, contrary
to the statement in Appendix XI.
19. Realistic vs Conservative Assumptions
It was concluded from the review of WASH-1400 Draft that the following
deficiency existed:
a. Indicated realism not consistently used - It was determined that,
although WASH-1400 Draft contended (page 15, Main Report) that
the objective was to perform a "realistic assessment," it did not
seem to be consistently followed.
Appendix XI Response - None.
4-71
-------
Conclusion - Conservative assumptions continue to exist in
WASH-1400.
20. Comparison of Risks Between Nuclear and Other Means of Electrical
Power Generation
The review of WASH-1400 Draft resulted in the following two assessments:
a. Comparison between nuclear and non-nuclear electrical power gen-
eration risks is required - A consideration of the comparison of
risks between nuclear and non-nuclear electrical power generation
is needed in order to judge the acceptability of nuclear power.
Such a comparison was not provided in WASH-1400 Draft.
Appendix XI Response - None.
Analysis Change - None.
Conclusion - In order to evaluate the acceptability of nuclear
power from the standpoint of risk, an analysis of risks from non-
nuclear sources of electrical power generation is required.
b. Inadequate emphasis that risks are only from in-plant accidents -
It was determined from the review of WASH-1400 Draft that inade-
quate emphasis was placed on the fact that the report only con-
sidered the risks from accidents in 100 nuclear plants, and did
not consider risks from other parts of the fuel cycle, sabotage,
Pu diversion, etc. In particular, risk curves are entitled "100
Nuclear Power Plants" which may give the erroneous impression that
4-72
-------
the curves depict all risks associated with the operation of
the plants.
Appendix XI Response - On page XI 2-3 and XI 2-4, it is stated
that "The study agrees...that the risks from nuclear power in-
volve not only those from potential reactor accidents but also
those due to normal reactor operations as well as considerations
pertinent to the rest of the fuel cycle. These matters are out-
side the scope of this study, as indicated in section 18 of this
appendix; however, the published literature contains a significant
body of analysis of many of those areas." It is reiterated on
page XI 17-1 that "Other types of nuclear facilities were outside
the scope of this analysis."
Analysis Change - None. (See Curves in Main Report.)
Conclusion - It is concluded that WASH-1400 (final) does not ade-
quately emphasize that the calculated risks from nuclear power
are from accidents only in operating nuclear plants.
Miscellaneous Comment - The reference to Section 18 on page XI
2-4 is incorrect. It appears that the reference should be to Sec-
tion 17, although the discussion does not correspond entirely to
the description on page XI 2-4.
21. General Inconsistencies
It was determined from the review of WASH-1400 Draft that:
4-73
-------
a. Study contains inconsistencies in approach and level of detail
considered - It was concluded chat these inconsistencies should
be minimized.
Appendix XI Response - None.
Analysis Change - Some of the inconsistencies, based on comments
received from various reviews, have been eliminated.
Conclusion - While WASH-1400 (final) has been improved over WASH-
1400 Draft, some inconsistencies, as described in the preceding
analysis, remain.
22. Conclusions
In concluding the General Observations section of Reference 2, the fol-
lowing points were made:
a. Continuous upgrading of study required- It was concluded that
continuous, diligent effort must be applied to keep the study up
to date by timely incorporation of reactor design changes, new de-
signs, changing failure rates, etc. Only in this manner could
undesirable trends be perceived and corrected. Also, the study
should be extended to apply beyond 1980.
Response - It is stated on page XI 2-12 of Appendix XI that "The
study further believes that a WASH-1400 type assessment of water
reactors should be repeated in approximately 5 years. The inter-
vening period should permit the collection of additional nuclear
power plant failure rate data and the further development of the
4-74
-------
methodology to permit more precise assessments to be performed.
It is important that the collection of data and the development
of methodology be pursued vigorously if these goals are to be
achieved." On page 138 of the Main Report, it is stated that
"It would also be useful to repeat an overall WASH-1400 type
risk assessment for water reactors in about 5-10 years."
Conclusion - The WASH-1400 responses to this comment are adequate,
although there is some disparity over the recommended time period
for a repeat of the study (5 years appears preferable). It is
presumed that the efforts described in the Appendix XI response
are, in fact, being pursued.
b. Expansion to incorporate different reactor concepts required -
It was concluded that expansion of the study to incorporate dif-
ferent power reactor concepts was necessary.
Response - It is stated on page XI 2-1 that "The study agrees
that it would be useful to pursue the areas outlined (different
versions of contemporary light water reactors, high temperature
gas cooled reactors, liquid metal fast breeder reactors, and
variations such as barge mounted power plants) in future NRC
work." Also, page XI 2-11 states "...efforts of the type report-
ed in WASH-1400 should be continued in the future and that risk
assessments of the same type should be performed in connection
with advanced reactor designs such as the liquid metal fast breeder
reactor and the high temperature gas reactor at an appropriate time."
Conclusion - The WASH-1400 response is adequate.
4-75
-------
VI. ADDITIONAL COMMENTS
During the course of reviewing WASH-1400 (final), two areas were found
which appear to be incompletely assessed which was not directly
considered as part of the review of WASH-1400 Draft:
1. ECCS Functionablllty
The study received several comments relative to the question of suc-
cess or failure of ECCS even if all components operate as designed.
On page XI 7-1, a response to these comments is given, as follows:
"The question of the success or failure of ECCS — as a matter
of functionability, as opposed to operability — does not readily
lend itself to analysis by the methods used in WASH-1400. Thus,
the study decided to examine what level of failure probability
would cause ECF to contribute to potential accident risks. As
noted in Appendix V, section 4.2, sensitivity studies reveal
that "...even if values as high as 10 for ECF failure (proba-
bility) were to be used, any contribution made would be within
the accuracy of the overall calculations."
"Thus, although there appears to be no current basis for making
a rigorous quantitative assessment of the probability of ECF
failure, the analysis referenced showed that even if ECF failure
probability were as high as 10 , it would not change the results
of the study significantly. It is the view of the study that the
probability that ECCS will fail to cool the core adequately is
less than 10 ."
Two aspects of the above response need to be explored and clarified.
These are:
4-76
-------
a. The discussion In Appendix V pertains only to the problem of mech-
anical damage causing degradation and nonfunctionability of the
ECCS. The question of whether, based on thermal-hydraulic con-
siderations alone, the ECCS may not function as designed is not
addressed. In view of the fact that ECCS systems have not been
tested under conditions representing a LOCA, there would appear
to exist some uncertainty in their ability to perform the intended
function. Indeed, the NRC is sponsoring extensive programs (e.g.,
LOFT) to provide assurance that ECCS perform adequately. It would
appear prudent for WASH-1400 to acknowledge the possibility of
ECCS nonfunctionability in this regard, and expand the sensitivity
investigations to determine the consequences of various degrees
of nonfunctionability.
b. The Appendix V discussion considers only ECCS functionability loss
from mechanical damage. It is thus implied that ECCS functionability
need only be considered for large break LOCAS where the potential
for mechanical damage is greater than for small break LOCAs. The
Appendix V assessment subsequently includes only large break LOCAs
in sensitivity determinations of nonfunctionability of ECCS. It is
readily apparent that, since large break LOCAs do not contribute
significantly to the risks from nuclear power, the risks are not
sensitive to assumptions regarding ECCS functionability. In this
regard, it Is somewhat misleading to contend, as was done in Ap-
pendix XI and Appendix V, that "These sensitivity results for both
the PUR and BWR...reveal that the overall results of the risk
assessment would not be particularly sensitive to a wide range of
ECF (emergency cooling functionability) failure probabilities..."
It appears obvious that, since small break LOCAs contribute much
more significantly to the risks, the assumption of ECF failure for
4-77
-------
these breaks would have a much larger effect on risks. In view
of the vide variations in results from various PWR vendor calcu-
lations of small break LOCAs (see Reference 2, Part 3, Section
V.A.6, and discussion in Section V, item 6 of this report), it
would appear reasonable to assume that some uncertainty exists
as to the thermal hydraulic interactions occurring during ECC
injection during a small break LOCA. Such an uncertainty should
be acknowledged and ECCS nonfunctionability for small break LOCAs
should be Investigated in WASH-1400.
2. Latent Cancer Fatalities
It was concluded during the review of WASH-1400 Draft that the
latent death risk from nuclear power accidents was not properly pre-
sented, and that the report tended to obscure this risk. Section V
(item 17. c) of this report discusses the changes made in the presenta-
tion of latent death risks in WASH-1400 (final), and concludes that
such presentations are still inadequate. The computation and presen-
tation of latent cancer fatality risks from nuclear accidents is a very
important and complex issue. In order to attempt to understand and
evaluate the numerous considerations involved in these efforts, a fur-
ther investigation of related areas in WASH-1400 (final) was under-
taken. Several apparent problems were found to exist in these areas
in WASH-1400 (final) as follows:
a. Population considered - WASH-1400 (final) assumes that for the worst
accident considered, latent cancer fatalities would be distributed
over a population of 10 million people (page 74, Main Report). It
is not clear how such an exposure population was determined, but
it appears to be excessive, especially in view of the fact that
4-78
-------
the entire population within a 25-mile radius of the 100 reactor
sites was determined to be only 15 million (page 73). At any
rate, using the normal cancer fatality Incidence for 10 million
people produces a large number of fatalities which is used as a
comparison to determine the impact of the radiation-induced cancer
fatalities from the accident.
b. Averaging risk over the exposed population - It is apparently
assumed that the radiation-induced cancer fatalities occur at
the same frequency over the entire population of 10 million
persons. There is no consideration given in the comparisons in
the Main Report to the fact that certain segments of this popu-
lation, particularly those living near the site, may, due to
higher doses received, experience a much greater incidence of
cancer deaths. Thus, the risk due to cancer death may be much
higher to some segments of this population than others. The use
of the same fatality rate over the same (large) population of 10
million persons does not permit any consideration of variations
in risk due to site (or radioactive plume) proximity. The latent
cancer fatality distribution for discrete segments of the exposed
population as a function of location needs to be explored in order
to adequately assess the risks.
c. Pate and risk from leukemia fatalities - The fatalities due to
leukemia, based on Table VI 9-5, account for over 23% of all
cancer fatalities committed from the first year after the acci-
dent while the normal incidence of leukemia is about 4% (Table
VI 9-9). Thus, the Incidence of leukemia fatalities from a
reactor accident would be accentuated. After the accident,
then, the leukemia incidence would appear to be greatly increased
over the normal rate since the rate for all cancers is higher after
4-79
-------
the accident, and the leukemia incidence rate represents a
significantly higher proportion of radiation-induced cancers
than normal. It is also Important to note that, according to
Table VI 9-1, reactor accident-induced leukemia affects par-
ticularly the young and unborn persons in the population, making
at least the perceived risk greater.
Conclusions - It appears that the risk tables and curves pre-
sented in the Main Report both obscure and minimize the risk
from latent cancer fatalities following an accident. It should
be noted that the quantitative evaluations in this discussion
are generally based on the worst accident postulated in WASH-
1400 (final). However, the arguments apply to other accidents
as well.
4-80
-------
VII. REFERENCES
1- Reactor Safety Study - An Assessment of Accident Risks in U.S. Commercial
Nuclear Power Plants. WASH-1400 (NUREG-75/014), United States Nuclear
Regulatory Commission (Oct. 1975).
2. Reactor Safety Study (WASH-1400); A Review of the Draft Report (Part 3,
.A Review of the Draft Report - Reactor Safety Study, by P. R. Davis,
Intermountain Technologies, Inc.), EPA-520/3-75-012, Environmental
Protection Agency (August 1975).
3. Reactor Safety Study - An Assessment of Accident Risks in U.S. Commercial
Nuclear Power Plants (Draft"). WASH-UOO, United States Atomic Energy
Commission (August 1974).
4. Meeting between P. R. Davis, Intermountain Technologies, Inc., and L. B.
Claassen, P. K. Gururaj, and W. Sutherland, General Electric Co. (Jan. 23,
1976).
5. Technical Report on Anticipated Transients Without Scram for Water-Cooled
Power RearforsT WASH-1270 (Sept. 1973).
6. Studies of BWR Designs for Mitigation of Anticipated Transients Without
Scram* NEDO-20626, L. B. Claassen, E. C. Eckert (Oct. 1974).
7. Exxon Nuclear Company WREM-Based Generic PWR ECCS Evaluation Model -
Volume III. Small Break Model. SN-75-41 (Aug. 20, 1975).
8. Final Safety Analysis Report - TROJAN Nuclear Plant, Portland General
Electric Co., Docket No. 50-344, as amended through February 1976.
9. Transcriptions from Advisory Committee on Reactor Safeguards, Meeting of
the Working Group on Anticipated Transients Without Scram (January 7,
1976).
4-81
-------
VIII. GLOSSARY
BUR Boiling Water Reactor
CSIS Containment Spray Injection System
CSRS Containment Spray Recirculation System
ECF Emergency Cooling Functionability
ECC(S) Emergency Core Cooling (System)
GE General Electric Company
HPIS High Pressure Injection System
LOCA Loss of Coolant Accident
LPCI Low Pressure Coolant Injection
LPIS Low Pressure Injection System
LPRS Low Pressure Recirculation System
NRC Nuclear Regulatory Commission
PWR Pressurized Water Reactor
RPS Reactor Protection System
4-82
------- |