ML20028C319
| ML20028C319 | |
| Person / Time | |
|---|---|
| Site: | Yankee Rowe |
| Issue date: | 12/31/1982 |
| From: | YANKEE ATOMIC ELECTRIC CO. |
| To: | |
| Shared Package | |
| ML20028C315 | List: |
| References | |
| NUDOCS 8301070328 | |
| Download: ML20028C319 (31) | |
Text
.
3 i
.E s.
,e j
l
, ~
y
- e EXECUTIVE
SUMMARY
Yankee Nuclear Power Station Probabilistic Safety Study December 1982 l
)
Yankee Atomic Electric Company 1671 Worcester Road Framingham, Massachusetts 01701 i
s-r 8301070328 830103 PDR ADOCK 05000029 P
\\.
w TABLE OF CONTENTS Page Section 1.
SUMMARY
AND CONCLUSIONS 1
II.
.3F.IEF DESCRIPTION OF YANKEE NUCLEAR POWER STATION t
3 Y
4 III. STUDY OBJECTIVES AND SCOPE APPROACH TAKEN AND IMPORTANT ASSUMPTIONS MADE 4
IV.
6
\\
V.
KEY RESULTS 7
VI.
UNCERTAINTY OF RESULTS 8
VII. KEY CONTRIBUTORS TO CORE MELT AND PUBLIC RISK VIII. APPLYING LESSONS-LEARNED FRO:t W ASH-1400 12 12 I X.
QUALITY ASSURANCE w
A W
k
+% l s
b i
LIST OF TABLES Page Table 1
YANKEE NUCLEAR POWER STATION GENERAL DESIGN FEATURES AND OPERATING PARAMETERS 13 2
PSS MODULES AND QUALITATIVE ASSUMPTIONS t
15 1-17 3
CORE MELT FREQUENCIES 4
RESULTS COMPARED TO NRC PROPOSED SAFETY GOALS 18 5
RESULTS COMPARED TO AIF AND NRC COST-BENEFIT CRITERIA 13 6
CORE MELT FREQUENCY COMPARISONS 20 7
RADIONUCLIDE RELEASE FREQUENCY 21 8
PUBLIC FATALITY RISKS 22 9
RISK-CONTRIBUTORS: ACUTE FATALITIES 23 10 RISK-CONTRIBUTORS: LATENT FATALITIES 24 LIST OF FIGURES Page Figure 1
AERIAL PERSPECTIVE SKETCH OF YANKEE NUCLEAR POWER STATION 25 2
RISK-PROFILE FOR ACUTE FATALITIES 26 3
RISK-PROFILE FOR LATENT FATALITIES 26 4
CORE MELT FREQUENCY COMPARISONS 27 5
RISK-PROFILE COMP ARISONS: ACUTE FATALITIES 28 6
RISK-PROFILE COMPARISONS: LATENT FATALITIES 28 i
4
-111-
1.
SUMMARY
AND CONCLUSIONS A Probabilistic Safety Study (PSS) of the Yankee Nuclear Power Station was performed by Energy Incorporated and Yankee Atomic Electric Company (YAEC) to provide additional insight into the design and operation of the plant and to utilize thec latest analytical tools in support of the decision making prohess.
A spectrum of events ranging from turbine trips to lar@ break los s-of-c oolant accidents was examined.
Plant-specific 8ata and modeling were used in assessing the likelihood of core melt and risk to the public.
A Technical Review Board (*)
consisting of acknowledged experts in this field was formed to provide its expert assistance and to impartially critique the effort.
The results indicate that the likelihood of core melt and subsequent adverse public health effects is substantially lower than assessed in the WASH-1400 study. The best-estimate mean core melt frequency is about one in 500,000 years [2(10)-6 per year], with 5% and 95%
confidenc e levels of one in 10 million years [1(10)-7 per year]
one in 100,000 years
[1(10)-5 per year].
Even using and conservative assumptions - e.g., Appendix K based success criteria, the mean core melt f requency is less than 2(10)-5 per year with 5%
and 95% confidence levels of 4(10)-6 per year and 5(10)-5 per year.
The best-estimate individual fatality risk is about a factor of 50,000 less than NRC safety goals; using conservative assumptions a bout a factor of 2,600 less.
Furthermore, Yankee results when compared to more recent probabilistic studies for other plants show that the Yankee Nuclear Power Station has a lower likelihood of core melt and lower overall risk to the public (see Figures 4 and 5).
Although this study did not address " external events" such as fires, floods and earthquakes, these events have previously been and are continuing to be evaluated by YAEC.
An important input in the assessment of these external events on the Yankee plant is the level of individual risk assessed in this study. The risk associated with YNPS operation is a factor of approximately 50,000 less than NRC g oal s.
Based on this finding, even if the likelihood of core melt plus vapor container (VC) f ailure due to seismic or flooding events 10-3 per year, NRC individual and societal risk was as high as All evaluations to date indicate that even if goals would be met.
l
(*) Technical Review Board consists of:
Professor Norman C. Rasmussen - Massachusetts Institute of Technology Dr. Salomon Levy - S. Levy. Incorporated Mr. Carry Thomas - Electric Power Research Institute Dr. Robert L. Ritzmann - Science Applications Incorporated
external events were included, thelikelihoodofcoremelt plus VC failure would be substantially less than 10-per year.
Final resolution of SEP topics involving " external event s" such as earthquakes and floods will be based on a value impact approach and the results of this present study.
These results demonstrate that operation of Yankee Nuclegr Power Station poses a very small risk to public health and saf&ty both according to the context of NRC and AIF proposed safety gdals, and comparisons with other probabilistic safety studies performed for other nuclear power plants.
The low core melt frequency re sult s from the conservative and diverse design of the Nuclear Steam Supply System and associated support systems.
In general, this plant has more systems available than contemporary plants.
and they are simpler in design Additionally, the f requency of of f-normal events at the plant has been shown to be small during its 22 years of operation.
An example of the redundancy and diversity of design is the number of ways to supply feedwater. There are ten pumping trains and eight flow paths available to supply feedwater to the four steam generators.
An example of the conservative design of the plant is that each of these systems is manually initiated.
Automatic initiation is not required because of the larger than typical inventory of the steam generators and the low operating temperature of the Main Coolant System.
An example of the low frequency of offnormal events is that the Emergency Feedwater System has never been required to operate in ' response to an offnormal event.
In other words, an extended loss of main feedwater has never occurred at this plant in 22 years of operation.
Coupling low core melt frequency with small core size, remote plant location, and an eff ective vapor container, the public risk profiles show exceedingly small overall risks of either acute or latent fatalities.
(The plant is about a factor of 5 smaller tha n contemporary plants and only 60 people live within 1 mile of the f
plant.)
In fact, only three events contribute at all (at the 50%
confidence level) to acute fatality risk. These are reactor vessel rupture and intermediate or large LOCAs with failure of the Emergency Core Cooling System. Even these rare events only result source terms in acute f atality risk if conservative WASH-1400 type are assumed and weather conditions are unfavorable.
if the source term were reduced by a factor Regardless of the event, less than 5_ f rom WASH-1400, there would be no acute fatalities even
.~.
I
with no evacuation.
Ongoing industry and government sponsored studies indicate reduction factors much greater than 5.
If the final result of these studies confirms reduction factors greater than 5_, then there would be no acute fatalities for even the most unlikely event.
Becausethelikelihoodofcoremeltandvaporcontainer((b) failure is low and the radionuclide inventory much smaller than cqmtemporary plants because of the small core size, the risk of plante operation indexed to annual man-rem exposure is extremely low.
The product (core melt plus VC failure probability) times (potential dose received by the public) is 0.22 man-rem per year. Man-rem exposure due to natural background radiation is in excess of 6 million per year. Thus, the risk of plant operation indexed to man-rem exposure is about a factor of 30 million less than that resulting from natural background radiation.
This small level of risk means that any changes to the current plant design and operational philosphies must be carefully evaluated.
Cost-benefit evaluations based on NRC and AIF criteria indicate that expenditures exceeding a few thousand dollars would not be justified even if they could reduce the risk to zero.
Since this plant poses a small risk, both by analysis and by the more than twenty years of excellent operation, well intentioned changes whether imposed by the regulator or proposed by the utility should be evaluated carefully to ensure that they do not in fact increase the risk of plant operation.
Beyond these conclusions, this study and the tools developed in support of this study can provide YAEC with an ongoing resource f or performing personnel
- training, evaluating plant modifications internally and in response to NRC inquiry, evaluating Technical Specification changes, and enhancing operational and emergency procedures.
YAEC is continuing to evaluate the current study findings and is formulating plans for integrating these techniques into the overall decision making process.
II.
BRIEF DESCRIPTION OF YANKEE NUCLEAR POWER STATION Yankee Nuclear Power Station was designed during the 1950's, and first licensed for operation in 1961 by the U.S.
Atomic Energy Commi ssion.
The plant is at a remote location in Western Ma s sa chu se t t s, in the town of Rowe.
It is relatively small, compare d to contemporary nuclear power plants, reliably producing about 185 MW electric power, with a lifetime capacity factor exceeding 70%.
The Nuclear Steam Supply System consists of a
pressurized j
light-water reactor, cooled via four main coolant loops each containing a U-tube steam generator.
During normal opeation, at 600 MW thermal power, the main coolant system circula'te s water between the reactor and steam generators at 100,000 gallons per
_3_
~
minute, which is maintained at 2,000 psig pressure and about 5400F average temperature.
Total volume of the main coolant system is 22,000 gallons.
A spherical vapor container surrounds the Nuclear Steam Supply System.
All equipment containing pressurized components within the vapor container is surrounded by a reinforced concrete
- cylinder, which is supported on eight steel-encased concrete columns penetrating the vapor container, as shown in Figure 1.7 Table 1 presents additional plant design and operational characteristics.
III. STUDY OBJECTIVES AND SCOPE The major objectives of this study were to 1) quantify the risk of plant operation - to the plant and to the public, 2) identify the specific aspects of plant design and operation contributing to these ri sk s, and 3) establish the in-hou se capability to conduct future i
studies and implement this model and associated methods into the decision making process.
The events considered in this study were based on a comprehensive review of possible plant perturbations to normal operation.
They included the following basic categories:
plant trips; power excursions; degradation of DC or AC power supplies; decrea se s in feedwater flow; decreases in steam flow; loss of component cooling, service water or control air; high energy line ruptures; steam generator tube ruptures; a
full spectrum of loss-of-coolant accidents; non-isolable LOCAs outside the vapor container; and reactor vessel rupture.
IV.
APPROACH TAKEN AND IMPORTANT ASSUMPTIONS MADE Approach State-of-the-art methodology and techniques in consonance with the PRA Procedures Guide, NUREG/CR-2300, were used in the performance of f
this study.
Plant-specific event and fault trees were developed.
trees developed were very detailed so as to model operator The event intervention and reflect current industry issues.
The resultant.
event trees represent over 500,000 sequences.
Fault trees were developed to a level of detail consistent witt. on going Interim Reliability Evaluation Program studies.
Human interaction was i
treated at both the f ault tree level and as top events in the event I
trees..
Information in NUREG/CR-1278 provided the bases for the human reliability evaluations performed in this study.
Quantification was based on the plant-specific models and plant-specific data for initiating event frequencie's and important system components such as diesel generators. This was performed by updatingrelevantgenericdatawithplantinformationobt[inedsince 1961 using Bayesian techniques.
~
The response of the containment and the assessment of posesible containment failure were determined using a plant-specific event l
I
tree, calculations with the MARCH, CORCON-MOD 1 and CORRAL 2 computer codes, and manual calculations of particle bed coolability and reactor cavity response. The impact of any postulated releases was assessed by performing plant-specific CRAC2 analyses using weather, site.
population and evacuation characteristics of the plant Key Assumptions probabilistic safety study performed in 1980i" indicated A scoping risks of YNPS operation, even if many conservative assumptions that were made, would be lower than risks identified in the WASH-1400 reactor safety study. Based upon these earlier indications, many of these same conservative assumptions were adopted for the present study.
This practice should facilitate review by the Nuclear Regulatory Commission, since most of these assumptions are required by its regulations.
Table 2
summarizes the qualitative a ssumptions adopted.
Cumulatively, these assumptions produce conservative values for risks of plant operation.
These values are characterized by the term " Conservative".
I The more important conservatisms present in the this evaluation are the following:
o Use of Appendix K criteria for defining LOCA mitigation success.
o Conservative system unavailabilities due to maintenance outages and human errors.
o No credit for recovery of systems initially f ailed because I
of operator / maintenance errors (e.g.,
improper positioning of emergency f eedwater system valving).
No credit for recovery of operator errors or of f-site power.
o o
Conservative containment re sponse analyses (e.g.,
H2 was assumed to burn from 11% - equivalent to complete Zircaloy reaction of core material - to 0% instantaneously).
Realistic results were also determined by performing an analysis in which realistic assumptions were substituted for the conservative assumptions discussed above.
These results are characterized "Best Estimate", because they address the objectives of identifying, on a relative basis, which design features contribute most to risks of plant operation to establish a realistic basis for decision making.
Table 2 also provides the qualitative assumptions consistent with the Best-Estimate evaluation. The results of each of these analysis approaches are presented below.
g V.
KEY RESULTS Core Melt Frequency The mean core melt frequency is low with a Best-Estimate of about
[2(10)4 per year).
The most important reasons are the following: low frequency of of fnormal events; redundant sfid diverse simplicity and passive nature of desigIi; minimal mitigative systems; reliance of front-line systems such as diesel generators En support systems such as service water; and inherent margin to design Even using the conservative assumptions, the mean core melt limits.
frequency is less than 2(10)-5 per year.
Table 3 provides these a comparison of each of the key uncertainty (*and re sult s, their contributors to core melt.
)
As Table 3 shows, los s-of-coolan ;
accidents (LOCAs) are the key contributors to core
- melt, contributing about 70% to the total in both best-estimate and conservative cases.
the primary contributc,rs to core melt frequency, Because LOCAs are failure of those systems needed to mitigate a LOCA are the key systems relative to core melt frequency.
The two primary systems used to mitigate a IhCA are the Emergency Core Cooling System (ECCS) and Charging System.
Although these two systems were evaluated to have small failure probabilities, systems needed to respond to other non-LOCA transients were assessed to be significantly less likely to fail accentuating the LOCA contribution to core melt.
The lower probability of a LOCA-compared to most transients - did not offset this difference in failure probability.
Public Risk The public risk profile is low.
Figures 2 and 3 provide the public risk profiles for acute and latent fatalities.
Results for both conservative and best-estimate assumptions are provided.
The be st-e stimat e individual f atality risk level is about a factor of 50,000 less than NRC safety goals and 260,000 less than AIFs.
Even the conservative assumptions, individual fatality risk levels usingf actors of 2,600 and 13,000 less than NRC or AIF Saf ety Goals, are respectively at the 50% confidence level (factors of 100 and 900 less at the 95% confidence level).
This low public risk profile because of 1) the low core melt frequency, 2) the occurs effectiveness of the vapor container, 3) the small size of the Respectively, the associated 5% and 95% confidence limits f or
(*)
the conservative and best-estimate cases range from 4(10)-6 to 5.2(10)-5 per year and from 1(10)-7 to 1(10)-5 per year.
s
plant, and 4) a remote plant location.
Table 4 presents a comparison of these indices of the plant risk profile.
the VC Steam generator tube ruptures and non-isolable LOCAs outside are the dominant contributors to latent health ef fects.
These two events dominate because they are the only events that pegate the effectiveness of the vapor container.
Using realistic $vacuation assumptions, only three events contribute to acute fatal (3y risk -
reactor vessel rupture and large or intermediate LOCAs wi' h failure t
of ECCS. These are the only events that c uld result in core melt and VC failure prior to evacuation being effective in removing people out of the acute f atality area during certain adverse weather conditions. Even for these events, however, there would be no acute fatalities regardless of weather conditions or evacuation effectiveness if the source term were reduced by as little as a f actor of 5, f rom that calculated using WASH-1400 methods.
Cost Benefit Co st-bene f i t evaluations for risk reduction based on NRC and AIF criteria were performed.
These evaluations show that expenditures exceeding a few thousand dollars would not be justified to reduce risk of plant operation, even to zero.
The best-estimate analysis shows an annual statistical man-rem value of.22.
Results based on conservative assumptions show 4.4 man-rems per year at the 50%
confidence level.
These findings are significant especially in in the source light of the conservatisms which are believed to exist t erm s used in this study in both best-estimate and conservative tam e 5 presents this comparison.
cases.
Compa ri son s Comparisons of the YNPS core melt frequency to those published in similar studies perf ormed f or several other plants are provided in I
Table 6 and Figure 4.
Even the conservative value assessed in this i
study is lower than the results for most of the newer plants.
The "be s t-e st ima t e" value is about a factor of 10_ less than any value compared.
Similarly, the public risk profiles from other studies compared in Figures 5 and 6 with the public risk profiles for are the Yankee Nuclear Power Station.
Again, the overall public risk of YNPS is lower than assessed in studies for other plants.
VI.
UNCERTAINTY OF RESULTS This study addresses " rare" events; that is, events whose likelihood is extremely small.
For this reason, it is not possible to Precisely assess the core melt frequency and public risk.
There is uncertainty in this type of study.
Core melt has not occurred in 1500 reactor years of free world commercial operation, so that a firm, noninf erred statistical bases is unavailable.
1
_y_
-~
This uncertainty was formally and consistently taken into account in each of the major modules of this study - initiating event probabilities, important sy stem / component failure probabilities, containment perf ormance, and in consequence evaluations (source
- term, evacuation, prevailing weather conditions, and the radiological sensitivity of the public).
The outcomq. of this processisaformalrepresentationofcoremeltfrequencyjndpublic health impacts with a quantified range of values that exp3 esses the e
uncertainty in these results, Based on this comprehensive and consistent assessment of these unc ertaintie s, the following statements can be made about those events examined in this study.
o The likelihood of a core melt is between one in 20,000 to one in 10 million per year.
The best-estimate is about one in 1 million per year. There is some chance, 1% to 5%, that it is less than one in 10 million per year, and at least a 95% conf idenc e that it is no more than one in 20,000 per year.
The likelihood of an individual receiving an acute fatality o
dose is between one in 200 million per year to it "cannot happen".
The best-estimate is about one in 7 trillion per year. There is about a 5% chance that it cannot happen, and at least a 95% confidence that it is no more than one in 200 million per year.
Even at the 95% confidence level, the risk that an individual could receive an acute f atality dose is exceedingly low.
VII. KEY CONTRIBUTORS TO CORE MELT AND PUBLIC RISK Contributors to Core Melt j
Table 3 presents a comparison of key events contributing to the total core melt frequency f or both conservative and best-estimate cases.
Events initiated by a breach of the Main Coolant System loss-of-coolant accidents, reactor vessel rupture, non-isolable 14CAs outside the vapor container and steam generator tube ruptures - are the primary contributors to core melt, contributing over 80% to the total core melt frequency.
Of these events, loss-of-coolant accidents contribute as a class about 70% of the total core melt frequency.
As outlined earlier many reasons exist for the small core melt frequency assessed relative to results obtained in other studies.
Additional inf ormation is provided below:
l o
l Transient events have a low frequency of occurrence ( e.g.,
o only three loss of main feedwater events have occurred in 22 years; in each event main feedwater was recovered prior to the initiation of emergency feedwater).
o Secondary cooling means (i.e.,
feedwater) are more substantial than at most modern plants (ten pum;dng trains and eight flow paths are available to supply feldwater to
?"
the steam generators).
o Main coolant pumps are canned; a LOCA induced by coolant seal failures is not possible.
Additionally, makeup pump requirements to the Main Coolant System are minimal.
o Front-line sy stems rely only minimally on support systems (e.g., complete losses of service water, component cooling water, or control air are negligible contributors to core melt frequency).
o Three totally separate and self-contained air cooled diesel generators exist.
Their performance has proven to be substantially better than industry average.
Even for the station blackout event the plant can be safely o
brought to hot shutdown using the steam-driven emergency feedwater pump. This pump requires neither de nor ac power.
o Significant margins exist to design limits during normal Compared to more contemporary designs, Main operations.
System and Secondary System operating pressures and Coolant temperatures are lower, and the total secondary coolant inventory is larger compared to plant thermal output.
These favorable features reduce challenges requiring safety and relief valve operation, thereby reducing the likelihood of transient induced LOCAs. They also provide more opportunity f or successful operator diagnosis and intervention.
For these same reasons, loss-of-coolant accidents (LOCAs) are the dominate contributors to core melt frequency using either conservative or best-estimate assumptions.
Even from a
per spective,
core melt resulting from sequences probabilistic involving an extended loss of feedwater - main feedwat'er, emergency feedwater and other backups - is essentially an incredible event.
This results from the diversity and redundancy of the secondary plant design and the time available to initiate feedwater systems.
AWS ( Anticipated Transients Without Scram) sequences contribute less than 10% to the total core melt frequency using either conservative or best-estimate assumptions.
This is due to the low f requency of transients, the reliability of the reactor trip systems and the backup capability of the Charging System.
Ue of the 9_
1 Charging System as a backup to scram is well proceduralized.
Additionally, this system has a higher capacity - 100 gpm - relative to plant size than most modern designs.
Finally, station blackout is not a key contributor because of the three diesel generators and the steam-driven emergency feedwater pump, as discussed above.
a(It, LOCAs In comparison to the transient contribution to core contribute 70% because the systems avail.able to mitigate g LOCA are not as diverse or redundant.
Two systems are availabler for LOCA, ECCS and Charging; whereas there are ten pumping trains and eight flow paths available to supply feedwater.
Although the likelihood of a LOCA is smaller than most transients, it is not small enough to offset the dif ference in failure probability between those sy stems needed to respond to a LOCA and those needed during transients.
Contributors to Public Risk Vapor Container Failure Probability The most important events contributing to public health effects are those resulting in core melt and vapor container failure.
Table 7 presents the results for the likelihood of core melt plus vapor container failure.
The be st-e stima te mean frequency is 9.4(10)-8 per year; the conservatively assessed mean value is 1.9(10)-6 per Excluding those events which result in a direct bypass of the year.
i vapor container, on average about one in seventy-six core melts I
would result in vapor container failure in the best estimate analysis, about one in seventeen in the conservative assessment.
Only two events result in a direct bypass of the vapor container:
non-isolable LOCAs outside the vapor container and certain steam generator tube ruptures.
Non-isolable LOCAs outside the vapor container by their definition negate vapor container effectiveness.
Certain steam generator tube rupture sequences have a path through the ruptured steam generator tube (s) and secondary system to the outside environment. As shown in Table 7, these two events comprise j
the most important contributions to radionuclide
- release, contributing a total of about 74% in the best-estimate analysis and 52% in the conservative evaluation.
They are not however the key contributors to acute f atality risks f or the reasons discussed below.
Contributors to Public Health Ef fects Contributions of various initiating events to f atality risks dif f er significantly for acute and latent fatalities.
The acute fatality risk is most af fected by the magnitude and timing r. release.
This situation occurs because of the threshold dose for acute fatalities.
Importantly, using realistic assumptions for evacuation effectiveness, there are no acute f atalities for any sequ(nee which results in a release beyond 2 hours2.314815e-5 days <br />5.555556e-4 hours <br />3.306878e-6 weeks <br />7.61e-7 months <br /> of the initiation of 3he event.
, i
4 even assuming WASH-1400 equivalent source terms.
Evacuation would i
move people outside of the acute f atality area.
If assumptions are made that evacuation is ineffective, as was done for the 95% confidence level curve, then there is some likelihood of acute fatalities for certain weather conditions if,-WASH-1400 equivalent source terms are assumed for release s greatjr than 2 hours2.314815e-5 days <br />5.555556e-4 hours <br />3.306878e-6 weeks <br />7.61e-7 months <br />.
However, regardless of the release timing, a minor, reduction in source terms by a factor of 5 would result in no acute' fatalities regardless of assumption s on evacuation, weather conditions or release timing. Table 8 provides the assumptions made in this study and the resulting public acute and latent individual fatality risks and annual statistical man-rems.
30% delay their l
Using realistic assumptions for evacuation evacuation by 45 minutes, 40% by 75 minutes and the remaining 30% by 105 minutes, all traveling at 10 miles per hour - only three events contribute to acute fatality risk.
These events are reactor vessel f ailure and intermediate or large loss of-coolant accidents, coupled with failure of the Emergency Core Cooling System.
Even for these acute fatalities are possible only if the vapor container
- events, fails in a major structural mode due to excessive pressurization of the reactor cavity and adverse weather conditions axist.
As discussed above, a minor source term reduction - equivalent to a factor of about 5 below those determined using WASH-1400 methods -
would result in no acute fatalities even for these events regardless
)
of weather conditions or evacuation effectiveness. Table 9 provides l
the likelihood of those events leading to an early release. Results show that there is only about one chance in a billion per year of this type of release (about one chance in a ten million per year using conservative assumptions).
1 The latent cancer fatality risk is based on a statistical evaluation of the effects of individual doses.
Table 8
provides the a ssumptions made in this study.
The 5% confidence level values corre spond to optimistic beliefs and 95% confidence level to pessimistic views.
Because latent effects are not as strongly than 3
influenced by a dose threshold, total man-rem is more important for acute fatalities.
For this reason, the key contributors to latent canc er fatality risks differ significantly from those contributing to acute fatality risks.
Thus, as Tgble 10 shows, steam generator tube ruptures are the dominant contributors to l
latent cancer fatality risks, contributing about 41% and 53%,
f re spec tively for conservative and be st-estimate evaluation i
assumptions.
Non-i solable LOCAs outside the VC are the second biggest contributors, about 11%
or 21%,
re spec tively for Conservative or Best-Estimate assumptions.
Combined, these two events contribute about 52% in the conservative analysis and 74% in the best-e stimate analy sis because radionuclides are. released s
directly outside the vapor container.
j l,
j f
~
-.. ~
=
4 VIII. APPLYING LESSONS-LEARNED FROM WASH-1400 WASH-1400 (NUREG-75/014 )
was the first major application of probabilistic risk-based methods to determine risks of operating nuclear power plants.
Many reviews of that study have been conducted, and many constructive criticisms levied. As this summary shows, these lessons from WASH-1400 were incorporated
- into the present study.
Consequence modeling was site-specific rg,ther than generic.
Statistical data and its uncertainties werk treated consistently in assessing both the core melt frequency and public risk profiles. A plant-specific statistical data base was developed by using Bayes' Theorem to combine plant-specific experiences with industry generic data.
Finally, operator actions were incorporated into more complex event trees, to impart more human-factor realism into accident scenarios.
I X.
QUALITY ASSURANCE Yankee Atomic Electric Company and its consultant, Energy Incorporated, both individually and jointly followed a procedure designed to assure the quality of this study. The elements of this procedure included independent calculations, sensitivity studie s, and comparisons with other similar work.
In
- addition, an independent Technical Review Board was assembled at the outset to provide its expert assistance and to conduct an impartial review of the ongoing effort.
Review sessions were conducted at the beginning, intermediate, and final phases of the study.
A multi-disciplinary approach to quality assurance was the prevailing goal of all reviews.
Members of the Technical Review Board are:
Dr.
Salomon Levy, S. Le vy,
Inc.;
Professor Norman C.
Rasmussen, Ma ssachuse tt s Institute of Technology; Dr. Robert Ritzmann, Science Applications Inc.; and Mr. Garry Thomas, Electric Power Research Institute.
l I
l l
l t -. - _ __-
TABLE 1 YANKEE NUCLEAR POWER STATION GENERAL DESIGN AND_
OPERATIONAL CHARACTERISTICS PRIMARY SYSTEM 600 MWt Licensed Power Rating t
5150F Core Inlet Temperature Main Coolant System Design Pressure 2485 psig Main Coolant System Operating Pressure 2000 psig 5590F Vessel Exit Temperature 38.3 alba /hr Vessel Flow Rate 2
156,900 Btu /hr-ft Core Average Heat Flux 4.40 kW/ft Core Average Linear Best Rate 12.5 kW/ft Design Peak Linea-Heat Rate 6020F Maximus Design Hot Channel Exit Temp 3.2 Minimus Design DNB Ratio Reactor Protection System Overpressure Protection 1 PORV, Sprays 2 Code Safety Valves FUEL ASSEMBLIES 76 Number of Assemblies 16 x 16 Fuel Rod Array 3.5%
Fuel Enrichment Er-4 Cladding Type 91 Inches Active Fuel Length 24 Cruciform Control Rod Type SECONDARY SYSTEM 4
Number of Steam Generators U-Tube Type 1035 psig Design Pressure 530 psig Operating Pressure 3 Boiler Feedwater and Feedwater System 3 Condensate Pumps i
1 Stean-Driven Energency Feedwater System 2 Electric-Driven
- Manual Start -
Other Nonconventional Backups 600,000 lbs/hr/SC Full Power Steam Flow Passive and Automated Steam Line Isolation Automatic and Manual l
Feedwater Isolation Steam Dump to Condenser Overpressure Protection Atmospheric Dump Code Safety Valves i
Reactor Trip Fdnetions
TABLE 1 (Continued)
YANKEE NUCLEAR POWER STATION GENERAL DESIGN AND OPERATIONAL CHARACTERISTICS REACTOR PROTECTION SYSTEM p-1.
Power Range, Neutron Flux 3.
Intermediate Power Range, Neutron Flux 4.
Intermediate Range, Righ Startup Rate 5.
Source Range, Neutron Flux Low Main Coolant Flow (Steam Generator Delta P) 6.
Low Main Coolant Flow (Main Coolant Pump Current) 7.
8.
Lcv Main Coolant Pressure 9.
Righ Main Coolant Pressure
- 10. Righ Pressurizer Water Level
- 11. Low Steam Generator Water Level
- 12. Low Steam Generator Pressure
- 13. Turbine Trip
- 14. Generator Trip ENGINEERED SAFEGUARDS SYSTEMS Safety Injection System Containment Isolation System Containment Recirculation System Vapor Container f
Post-Accident Hydrogen Control System l
SAFETY INJECTION SYSTEM Three Pumping Trains Consisting of One Low Pressure and One High r
Pressure Pump One Accumulator Safety Injection Tank Associated Valves and Instrumentation Power from Off-Site AC or Diesel Generators l
I VAPOR CONTAINER 34.5 psig Design Pressure 2490F Design Temperature Double-Ended Guillotine LOCA Design Basis 125 Feet Diameter Spherical Shell
.875 to 3 Inch Thickness 3
860,000 ft Net Free Volume Totally Passive Heat Sink _ _ _ _ -
e' TABl.E 2 ess noost.as Ass quAs.Itative Asstserflous I
Impact of sabetiteeles coneereotive aset-setteste teet-setteste for Ceneervettoo pas stedete 9eelieetIwe Aosemptlece
$seIitatI** Aseenuptlese Assumptione 4
lettlettes avest aseed en anyeelen 9pdate of Fleet Specific Beeed en Beyeelen Update of Fleet Specific Itiner, aeduction le core melt Frequency Informaties. teojer conservatisme include:
Inforestion, including:
fregeoecy of approelmetely e
- 1) Aseming generic date en testege
- 1) piner leakage evente seeemed to be f ac tor of M.
j evento are lACAs regetring 2005 for successfully mitigated by eitisettee, morestly operatinut makeup systeme (CVCS), and 4
l
- 2) We credit for recovery of eyetse
- 2) Some credit-etill conservative-tekee J
fetteree contributing to laitiating event for recewery of failures leading te (e.g., the 3 lese of male feedveter events initiatina event.
et TWPS sete recovered before requiring l
Seersency Feedveter).
Seceees Criteria e Treneiset event ouecese celteria realistic. o Some for tremeleet evoet sequences.
Pwrtber reduction le eere melt f
e IDCA success critsrie beoed on Appendia E e lACA soccese criterie beoed en best freguesey by a facter er M.
analyele seeseptione.
eetiente 9tASCII emelyses, IAFT resulte Comslative impact of best-estleste I
e migh emergy line roptere ehetesen-and hand calculatione.
eseampetene - initiating event U
ocree-requiremente beoed on licensing e Wish energy line roptere ehetdeen frequency and succese criterie le I
enetyeis.
requiremente beoed on best eetleste redwetion le core melt frequency onelysis.
by a facter of 3.9.
Feelt tree Development e IREP level of detail.
e Same.
lepect eveleeted below le and Omentification e conservative maintenance e meeltetic eslotenance seseellebility
" Event tree Development and a
onevellebility eseemptione.
soeumptione.
quantifiesel=".
o conservative maintenance errore e Seelistic eyetse enewallebility due to es se ssment.
maintenance errors.
]
Evoet free 9evelopment e Detailed includlag eweleettee of a sema.
Fwether reJoetten le core melt end Osentificet to--
preseerleed thersel eleck induced fregmeery by a facter of of 2.3.
Seector vessel Rupture for all events, Cumulative impact of best-ee F.ee.
escepting lASA.
eseumptione - laitiating eveMt e me credit fer recovery of operator e anelletic seeesoment of operator frequencies, succese criteria and event tf& i$s' sit, fication - is e 1 errore er eyotees failed due to performence and recovery of failed it f eseity remedied esir.tenance/startup systeme due to, for esemple, volve reduction in core melt frequency errore (e.g., velvee eiselianed in miselignment.
by approsteetely a f actor of 9.
emersency feedveter dischera* line.
e Iso credit for recovery of off-site e Realistic recovery rate of power in lose of AC.
of f-site power (803 within 3 hours3.472222e-5 days <br />8.333333e-4 hours <br />4.960317e-6 weeks <br />1.1415e-6 months <br />).
1 i
P 91 i
TABLE 2 (Coettened) 1-a i
PSS IUDOWLES AIS QUALITATIVE AS$tBIPTICIS i
Tapect of Sabetitetfog t.
h I
Commersettve teet-Estleste Beet-tetteste for resservative pg5 Medele p 1Itatise AmosmetIsee Qas1itative Asemotteos AssumptIaos Osee Nolt Soeussee e Resed so IIABCE code end heece subject e Seen beste enthodelegy.
Dedeettee la libeltheed of poector Aeolyste to 88A801 llettettoes.
e Better eetlestes se to-vessel veneel fellere afven "Oooet of
]
e Itere eigetitcent IIARG liettettoes coolehtitty of a porttele bed used.
Core 9telt" for these oogeencee sere etrcumvented er steleteed by in obich condittees estet for I
code modifiesttees er eeneel calce-estabtf ehlag e eeelebte portf ele lettoes se to the case of particle bed. Overall fopect en rolesee bed coolehtitty. er frequency see seattelble, homever, i
e Dee of alternettve emelyttral mothede fee to desteence of e,eete each such as alaaleHIDel testeed of IsrfER se SCT9e end neo-teelebte ImeAa.
for debrio concrete toteractices.
8 I H leet aseteemelide o Bened oe WASD-1400 type semlysee e Same Its lopect.
5 Bobester sales COBRAL-2.
iI e necertalety to Source Thre treated l
le probability of frequency forest.
e Opper bened (958 coef teence level) j i
to eeesteteet eith CORRAL-2 results.
e Fifty percent coefidence level resulte 1
are apprestastely a facter of 3 loser.
e Five percent esaf fdence level resulte are apprestestely a facter of 9 loser.
Centateneet Feilere o Rosed es Inetts and GIECDIMIODI e Some Overall depect of best eetleste Isodee end Fregeoectee emetyees.
e Adlebotte N2 hore free St to esemoption en core sett and e Adlebotte 32 bere free meelen to 42.
containment fellere sedes reduced e Best-esttente eveleottens of ecote and latest concer fetality coecentrattee to 0, percent.
e Ceeeervettre eve.nettese of reactor cavity resposee.
rieke by obeet a factor of 300-reacter cavity resposee.
and,70, respectively.
Ex-Plent e---
e eseed se pleet-opecific CRAC2 model.
e Ceae.
Its fopect.
Eveleottees e Evacuattee and health ef fecto sedels treated in probability of fregeency forest. 90edian is best eettente.
951 confidence level to worst caeeg 51 coefteence level to optistette case.
TABLE 3 CORE MELT FREQUENCIES I.
OVERALL VALUES t
Qualitative Assumption (P Conservative Beet Estimate Statistical Index 4.0(10)-6 1.0(10)-7 5% Confidence 1.7(10)-5 1,9(10)-6 Mean 5.2(10)-5 1,o(10)-5 95% Confidence II.
CONTRIBUTING EVENTS Contributor Rankings Mean Core Melt Frequency Event Conserva tive Best Estimate Conservative Best Estimate Intermediate 4.6(10)-6 6.1(10)-7 1
1 LOCA Large LOCA 2.6(10)-6 4.2(10)-7 3
2 Reactor Vessel 2.7(10)-7 2.7(10)-7 11 3
Rupture Very Small 3.7(10)-6 1.2(10)-7 2
4 LOCA Small 1.7(10)-6 1.2(10)-7 4
4 LOCA Turbine Trip 8.9(10)-7 8.7(10)-8 5
5 (ATWS)
Degradation of DC Fower
,8. 2 (10)-8 8.2(10)-8 13 6
7.7(10)-7 5.0(10)-8 6
7 SGTR Loss of AC 6.8(10)-7 2.6(10)-8 7
8 Mon-Isolable LOCA Outside 2.0(10)-7 2.0(10)-8 12 9
.,T Feedwater Flow Decrease 5.4(10)-7 9.4(20)-9 8
10 (ATWS)
Remainder are all less than.5% each..
~-
TABLE 4 RESULTS COMPARED TO NRC PROPOSED SAFETY 00ALS Individual Risks Societal Risks Y
5(10)-7 per year NRC Proposal.
Acute y
Fatality Risk (1-mile radius)
(50-mile radius) 1.0(10)-6 per year 3 per year Latent Yankee Nuclear Power Station 1.4(10)-11 per year 50% Confidence Acute Level Results Fatality (Conservative Risk (1-mile radius)
(500-mile radius) 2.2(10)-4 per year Case) 7.6(10)-10 per year Latent Fatality 1.4(10)-13 per year 50% Confidence Acute Level Results Fatality (Best Estimate Risk (1-mile radius)
(500-mile radius)
Case) 1.l(10)-5 per year Latent 3.8(10)-11 per year Fatality 4.9(10)-9 per year 95% Confidence Acute Level Results Fatality (Conservative Risk (1-mile radius)
(500-mile radius) 5.3(10)-3 Case)
Latent 5.9(10)-9 per year Fatality 99
,zm._,,um&Wg.Aa a M,.
m.
4
-4Maa 4
2*
.h-.
J.4mA-.-
mai-4 hw Laae a.2a.4et..
AJ u
4eu44-4J.4*m a
h=.4 4----i.+-
d<
l.w.a-.4
-m.--
T r
t
{
?
E
'i r
4 i
O i
1 8
1 f
e 1
4
.(
t l
i
+
h 4
h 1
4 4
a I
I 4'
l.
I i.
i i
1 b
T 1
i J
e v
t k
P e
e.
P I
A I
s L
f d-I i
t I
~ -. - -.
TABLE 5 RESULTS COMPARED TO AIF AND NRC COST-BENEFIT CRITERI A Maximum Justifiable Value-Impact Based Yearly Expenditsres 1
i T
TNPS I
Risk Man-Rem 9 $100 per man-ren*
9 $1000 per man-rem
- Profile Per Year (AIF Criterion)
(NRC Criterion) 5%
Confidence 0.14 14 8
140 Level (Conservative Case) 50%
Confidence 4.4
$ 440
$ 4,400 Level l
(Conservative Case) 95%
Confidence 36
$3,600
$36,000 Level (Conservative Case) 50%
Confidence
.22 22 220 Level (Best Estimate Case)
- Based upon reducing man-rem amounts to zero.
F e
m 9. _ _ _ _ -
e=
g TABLE 6 CORE MELT FREQUENCY COMPARISONS Transient Plant Core Melt
- Shutdown (Study)_
Frequency State Cometets 6(10)-5 Cold From Page-135 of Main Surry Report.
Point (WASH-1400)
Estimate. October 1975.
3(10)-5 Cold From Page 135 of Main Peach Bottom Repo rt.
Point (WASH-1400)
Estimate. October 1975.
Limerick 1.5(10)-5 Hot From Pages 3-100 of Main Report.
Mean Value.
(FRA)
March 1981.
9(10)-5 Hot From Pages 6-43.
Point Oconee Estimate. January 1981.
(RSSMAP)
Grand Gulf 3.7(10)-5 Cold Sum of Dominant Contri-butors from Figure (RSSMAP) 6-1.
Point Estimate.
Arkansas Nuclear 9.2(10)-5 Hot Sum of Dominant Contri-butors from Table 8-4, One - Unit I Page 8-60.
Mean (IREP)
Value. June 1982.
5.7(10)-5 Cold From Page 8.10-1.
Mean Zior Value. 1982.
(PRA)
Indian Point 2 9.0(10)-5 Cold From Table 8.3-2, Page 8.3-14.
Mean Value.
(PRA) 1982.
Indian Point 3 1.3(10)-4 Cold From Table 8.3-3, Page 8.3-15.
Mean Value.
(PRA) 1982.*
1.7(10)-5 Hot YNPS PSS. Conservative.
TNPS (FRA) 1.9(10)-6 Hot YNPS PSS.
YNPS Best Estimate.
(FRA)
.1
- Internal Event Contribution.
External Event Contributions either excluded, not considered in study, or not significant contributors. -_._
B TABLE 7 RADIONUCLIDE RELEASE FREQUENCY I.
OVERALL VALUES Qualitative Assumptions 5tatistical Index Conservative Best Esttmate d
5.0(10)-7 5% Confide ce 1.9(10)-6 9,4(10)-8 Mean 4.5(10)-6 95% Confidence II.
Q)NTRIBUTING EVENTS Qualitative Assumptions Contributor Rankings Event Conservative Best Estimate Conservative Best Estimate SGTR 7.7(10)~7 5.0(10)-8 1
i Non Isolable LOCA Outside VC 2.0(10)-7 2.0(10)-8 2
2 Reactor Vessel Rupture 5.7(10)-8 1.4(10)-8 10 3
Turbine Trip (ATWS) 1.5(10) 7 4.4(10)~9 3
4
~
Loss of AC 9.4(10)-8 1.3(10)~9 6
5 i
Intermediate LOCA 1.4(10)~7 9.0(10)-10 4
6 Large LOCA 8.0(10)-8 6.5(10)-10 9
7 Excessive Cooldown 8.5(10)-8 2.5(10)-10 7
8 Steam line Break 8.2(10)-8 2.0(10)-10 8
9 1Baall LOCA 5.2(10)-8 1.8(10)-10 11 10
.i Very Small
~
LOCA 1.1(10)~7 1.8(10)-10 5
10 l
Degradation 2.5(10)~9 1.3(10)-10 14 13 of DC Power All Remaining Events less than 0.1% each.,-
TABLE 8 FUBLIC FATALITY RISKS I.
OVERALL VALUES Individual Risk
- Societal **
l~
Confidence Risk Per Year Per Year Level Latent Latent Acute Cancer Cancer Man-Rem **
Fatality Fatality Fatality Per Year 5% Conservative 0
3.6(10)-17 3.8(10)-5 0.14 50% Conservative 1.4(10)-11 7.6(10)-10 2.2(10) 'I 4.4 95% Conservative 4.9(10)-9 5.9(10)-9 5.3(10)-3 36 1.4(10)-13 3.8(10)-11 1.1(10)-5
.22 50% Best-Estimate t
- Risk to individual within 1 mile of the plant boundary.
- Within 500 miles of the plant.
l ASSUMPTIONS FOR EVACUATION AND PUBLIC HEALTH CONSEQUENCES II.
For Latent Confidence Cancer Level Evacuation Source Term 5%
100% 0.75 Hour Delay 5% Frequency. Non-10 Rem Threshold 10 MPH noble gas releases 10-Mile EPZ about a factor of 9 less than WASH-1465 method based results.
50%
30% 0.75 Hour Delay 50% Frequency. Non-Central Estimate 40% 1.25 Hour Delay noble gas releases 30% 1.75 Hour Delay about 3 times less than WISH-1400 method 10 MPH 10-Mile EPZ based results.
Linear Hypothesis 95%
95% 1.75 Hour Delay 95% Frequency.
5% Fail to Evacuate Releases equivalent 10 MPH to WASH-1400.
10-Mile EPZ 1
, l
TABLE 9 RISK CONTRIBUTORS: ACUTE FATALITIES Mean Release Frequency Qualitative Assumptions Contribetor Ranking Contributina Event Conservative Best Estimate Conservative-Best Estimate I
Reactor Vessel Rupture 2.2(10)-9 4.2(10)-10 3
1 Intermediate LOCA with 4.0(10)-8 1.2(10)-10 1
2 ECCS Failure Large LOCA with ECCS 2.6(10)-8 9.9(10)~11 2
3 Failure 6
.* ?
TABLE 10 RISK CONTRIBUTORS: LATENT FATALITIES Mean Release Frequency Qualitative Assumptions Contributor Ranking Contributina Event Conservative Best-Estimate Conservattve Be s t-Es timat e 7.7(10)-7 5.0(10)-8 y
y SGTR Non-Isolable LOCA 2.0(10)-7 2.0(10)-8 2
2 Outside VC Reactor Vessel Rupture 5.7(10)-6 1.4(10)-8 go 3
Turbine Trip (ATWS) 1.5(10)-7 4.4(10)-8 3
4 Loss of AC 9.4(10)-8 1.3(10)-9 6
5 Intermediate LOCA 1.4(10)-7 9.0(10)-10 4
6 Large LOCA 8.0(10)-8 6.5(10)-10 9
7 Feedwater Flow Decrease (ATWS) 4.5(10)-8 4.7(10)-10 12 8
l All other initiating events contribute less than 0.2% each.
1
.;i f l l
l I
I N;., N0.,
j y (s). ~
\\
y
- p
~.c
.. +,
w a.
T %f'n
' Mi
' ii i '. !
h
's.
(
- ' A (I. '+t.
t.
'% ' y',f 'i $
\\'
5
- t. c -
.,s. -
5
\\
' ' '%_.'{'{
- y
',[.f?
?
6.-
" lK.
r.fM. 'n'
?
."wy' 1s v p;.., - -
'a p- :- ' s
- 6..,
- ev
. " ?. e g a ' 3,.,#.
% It,f.Y-Q ',ff,
e
~
. i 4~ '. 2 4
.k..'
w,
- ?o4
- c. w.
u
.- (~, _..-
R.,\\ g' :.Q A,.
(
t:
- ?
. h
. j %..7n 4
3 2';
3
,, ; c' -
s
,5
- 4 d
N-
, ;T.
t
- .~ s f.; * [
,.... '. Y
- f,'%!
T: -
y k;
- 3.
. y sf.,..
n,
' ' p '.,.
4
- > l, &. ;
,j
)
a
~
y.f r
-i!
[
D.
(
k !f
_,4j j', l Q..,'
/
~
. f h..
4 /,7 % y h..
%. i :';
- s. % -
,#,./;/4 N. 4 ;. A.h,,
1 t
.y c.;_.'?'.. -
,8
- i r
M[b-:
//)gNf.[j.:is$9i}hV,
/ '> '
yI
.3 '.
~f j,.
s gJr
./
<u.
c.
^
(
Y-g'
~
.~-*
,'3 FIGURE I AERIAL PERSPECTIVE SKETCH Q:+.. g.*
,jt I.
+
@5 C
OF YANKEE NUCLEAR POWER 4,.,,
u..
4 f b..-..*O 3
d.
STATION -
I i
..=m iy..... '
.u.,,,
,,,,ssy Best Bettaste Baseedence Probability less then
.y 3.8E-13 et Acute Caesequence 10 --- -
&*'*1 of 1.85*00 and S0 Percent Confidence
]
=
r
.g a
10 j
9 ereestile Conservative FIGURE 2 E
10 h
g RISK-PROFILE FOR ACUTE
.a I
.3, FATALITIES 30 Percestile Consenstive I
f:
11 10 m
109 10
'30 2
3 4
10 10 10 ACITTE FATALITits, z 20 "
p
=
E
[ 5 10*5 a
i es i
at
=
95 Percentile Comeervative l
10-4 l
FIGURE 3 N
RISK-PROFILE FOR LATEN ~.
30 Fercentile, E
a-"* t i'*
FATALITIES 10.,
E lI 30 Peteestale Boot Estiaste
=0 o
10 a
T
=
, S Forcent11e a
Ceaservettve s
10 10' 10 10 10' 10 10' i
I LATDrf C WCta FATAL 171ts PER TEAA. I l ;
l
1.3:1.
- 5
-5 10
,,% 3,-s e.saio e.o.2o
-5 s
3.7 2.
e
-5 g
i.,., >
~
10 E
,0
.1iIll'!!iill:l i;
i
<m FIGURE 4 CORE MELT FREQUENCY COMPARISONS __
l d
..w-Tankee Best Estlante....l w
kneeedence Probability l
1.5 -11 et 1
=
Acute Fatality esseenvence favel l
.y g
88 JJ5 Bift
, " I'00 10 L
FIGURE 5 risk
'. RISK-FROFILE COMPARISO l
C 30-9 s
3 ACUTE FATALITIES
- ,m Omeerwdw
_g,gg,,y,g,,
=
10*10 2 and :
P fien j
A B
a R l A
f, lla E
I fGA 0
10
'10 10
,LO' 10' 10' ACUTE FATALITIES. I 10*'
5
....,y Ais anck Point K
10.,
, u.m
, idASH-1400 at l
N 19 ee,eio : 1 10 0 m
\\\\
FIGURE 6 N-
\\\\ m Yanke.
c meervett,
- RISK-PROFILE COMPARISON Redtin Petet 3 C
10
\\
[
LATENT FATALITIES s
i w
y
=\\
l sea se
' Beet Eatina
~
te 10-e l..
I 10*'
l 0
2 I
3 10 3,1 10
,LO 10' 10 LAftWT CANCER FATAL 171tl PER TEAR. 1 -
.