ML20149F830

Submits Comments on 870608 Draft of Technical Basis for Graded Responsive Strategy.Substantial Reduction That Could Result from Proposed Provisions,Including Phased Response, Should Be Shown

ML20149F830
Person / Time
Issue date:	06/25/1987
From:	Martin J NRC OFFICE OF NUCLEAR REGULATORY RESEARCH (RES)
To:	Margulies T, Mcguire S, Mckenna T NRC
Shared Package
ML20149B718	List: ML20149F619 ML20149F626 ML20149F650 ML20149F663 ML20149F678 ML20149F687 ML20149F777 ML20149F792 ML20149F799 ML20149F812 ML20149F826 ML20149F830
References
FOIA-87-743 NUDOCS 8802170381
Download: ML20149F830 (2)

Text

- - _ _ _ _ _

[a.a wuco,b,",

UNITED STATES f EN n NUCLE AR REGULATORY COMMISSION

7. . WASW NG TON, D. C. 20555 tj

%...../

JUN 2 51987 i:0TE FCR: T. Marglies S. McGuire T. McKenna L'. Pratt A. Tingle FROM: Jim Martin, SAIB, RES

SUBJECT:

Comments on 6/C/87 Draft of "TECHNICAL BASIS FOR GRADED RESPONSIVE STRATEGY --- etc."

i. Change "graded" to "phased" - this better expresses the most important timing factors in an emergency response - which should be accorrodated in .

the plans. This should logically lead to EALS, which you hardly even nwntion, but they're the last step in the response logic - and the first step in the plans!.

2. The basic radiation protection criterfa for severe accidents relate to early fatality and injury (NRC, EPA, FEMA, ICRP, CCC, WHO, IAEA). The NRC Safety Goal is expressed in ter.T.s of health effects. You cite only some of these references. Cite them all. I suggest that McGuire cone get the citations from me. We've worked long and hard to get the current concensus - let's clearly state that we've got one, and what it is.

3. Moreover, all your results are in terms of doses (no health effects). For the severe accident postulates, perspectives on early fatality (esp) shculd be displayed. Since the first objective is provisions in the plans for substantial reductior in eerly severe health effects for severe 4

accidents, show the substantial reduction that could result from the proposed provisions, i.e. phased response. For smaller accidents, show dose savings. (The dose fetish in this agency is aweseme!)

4. The potential efficacy of your preferred phased emergency response strategy is not displayed explicitly - it must be inferred from your tables. It should be shown explicitly, i.e. how would the preferred response strategy meet the basic radiation protection objectives? For a spectrum of accidents,

r. Show the 1/r dose rate fall cff curve. It clearly, simply and obviously shows where the big risk areas would be. Further, plot some of the timing x fo evocuaticn information to clearly show the obvious - the earlier you Nicave, the lower vour risk (a picture, etc.). You should also show that I evacuation speed is of secondary import (ahove about 3 r.ph).

6. In the introduction your mention "previous supporting analyses (plural)",

cut you cite only one reference. lle have available an abundance of 8802170381 800204  %

PDR FOIA FDR d

SHOLLY87-743 1

J m _ _

.[.

2 published results er. which we based our current pr.siticns. The whole bunch should be citco (Aldrich, l'artin, McKenna, P.harya, Blond, Burke, Heising, Gant, Spector, CRNL, Sf!L, AIF, EPRI, IllPO, HUS). We were working in the full light of day - we weren't winging it. (Again, Steve should come and get the references frcm n'e.)

7. I agree with McKenna's pene earlier this tr.cnth that we haven't explained very well either our corrent position or our technical bascs. This paper should be heavily slanted to explain these two things. Once this is done satisfactorily I suggest 5t be published in Health Physics Jr. Nuclear Safety is an alternative, but H. P. Jr. will reach a wider, reare.gerrnane audience.

8. We've a way to go yet, however.- The current draft is a pot pourri. It's not pointed - even tho' it's all very interesting and informative. It doesn't ring - I don't hear the bells! Moreover, for a Journal article, it's too long, I think. State the objectives, get to the point (s), then sumarize the points. What are you trying to accomplish - especially in the light of all the previously published infortnation of the same ilk?

You've got a good start, but a rr.ajor rewrite is in order - and it'll be worth it!

} .

, lartin V

cc: J. Hulnen L. Soffer l

C

- - . 2

c-

, . s s A -A

' #g " ~ ' .

e Oak Ridge . . . .>. t e ho)

Associated 1019 Nineteenth Street, Northwest for Energy Universities Washington, D.C. 20036 Analysis NfW Jul Y, 9Yi Thomas E. Murley, Director Office of Nuclear Reactor Regulation

./ g U. S. Nuclear Regulatory Commission Washington, D. C. 20555

[u

. A g/((

ff/ !Wlus.<.e&- &

Dear Dr. Murley:

/vew M 'Ae. @&J "f0s. o.f $

oW and 4Y8d4 I am writing to inform you of possible errors in my others' testimony at the Indian Point hearing on risk in 1983 and N to highlight evidence that emergency planning may be increasing M ;

rather than decreasing the radiological risk posed by nuclear power plants. I learned of the possible errors quite recently,f4"%

and am writing so that you may take appropriate actions to explore the validity and implications of the evidence.

As you may recall, while serving as Assistant Director for (Safety) Technology in NRR, I headed the task force that prepared the staff testimony on risk at the Indian Point hearings, and was one of the principal witnesses for the staff in that ASLB Specia)

Proceeding.

The error in our testimony lay in assuming that a decision to evacuate in the more serious accident scenarios would be Imade well in advance of the beginnings of an offsite release. have interviewed a number of NRC staff members involved in evaluating emergency response drills, and they tell me that in the early 1980's, at the time the testimony was prepared, state or local authorities commonly preferred to wait for the occurrance of a release of radionuclides to the environment before initiating an l

evacuation. Anecdotes about recent drills in which such delays occurred suggest that even today, we cannot be certain that evacuation would always be initiated well ahead of a major release.

l Early fatality projections would have been higher -- for those i scenarios in which evacuation is r. viable option --

had we modeled the decision to evacuate as occuring only at the time of release or soon thereafter. See Enclosure 1 for the the technical basis for this conclusion.

l Let me emphasize that the bottom-line risk for Indian Point is I unlikely to be changed significantly, due to the insensitivity of l the low calculated risk to evacuation model assumptions.

( owever, the testimony on the effect of emergency planning upon thesradiological risk borne by the public would have been quite l

l different', so both the generic and plant specific conclusions on emergency preparedness could have been affected. I suspect that l

f S<

- m Aooress: P.O. Box 117 / Oak Ridge, Tennessee 378310117

T We would have found that the emergency planning instituted between the accident at Three Mile Island and the time of the hearings in 1983 might actually be tending to increase the risk to the public. See Enclosure 2 for an assessment of the significance of the new information for the Indian Point case.

Briefly, there are features of mass evacuations -- particularly if the evacuation is not begun well in advance of a severe atmospheric release of radionuclides --

that might actually increase the number of people subject to life-threatening doses, compared with alternative emergency response strategics such as:

1) sheltering followed by relocation --

af ter plume passage --

from hot spots of residual contamination, 2) a phased, selective evacuation designed to give first priority to moving people out of the path of the plume, or 3) anticipatory evacuation well ahead of the release.

Mass evacuations can increase the risk in the following ways:

• some people close to the plant over whom the plume passed may begin to evacuate along the path of the plume, thus receiving higher doses than they would have done if they had stayed in place pending relocation, because they lose the benefit of the higher chielding factors afforded by homes and buildings;

• some people whose homes, work sites, or schools were not in the path of the plume may evacuate along paths where the plume has just traveled;

• some of these groups of people may re-enter the plume, as both they and it move away from the plant, and/or they may travel inside the plume;

• traffic congestion due to the simultaneous evacuation of large areas may slow the departure of those nearest i the plant; since low wind speeds tend to yield the higher doses in any event, the congestion effect may selectively aggravate the more serious life-threatening accident scenarios; l

• traffic congestion may also cause the plume to catch i

up with and pass over evacuees who started off before the plume reached their starting point. Although i such people will be farther away when the plume reaches t

them than they would have been if they stayed put, the I

lower shielding factors of vehicles and longer emersion l times may neutralize or overwhelm this advantage;

• traffic congestion due to a mass evacuation may also i inhibit the movement of radiation monitoring personnel, l

I

a - / ,

police and other civil authorities who might otherwise be directing evacuees from the path of the plume;

• the heat released with the radionuclides may loft the plume over the terrain near the plant, causing areas at some distance to be more hazardous than areas in close: evacuees may be moving frcm less hazardous areas into more hazardous areas.

While some evacuees may escape the plume altogethef or incur smaller doses than they would' if mass evacuation was not the primary protective action, overall one cannot be sure whether the doses will be higher or lower.

Of course, none of these attendant problems with bulk evacuations would occur if one could be sure that the evacuation was always directed so that people generally move away from the path of the plume, even if that means initially traveling around the plant rather than away from it; However, the emphasis placed upon evacuating large areas, e.g. 360 degrees around the plant out to five miles and the down wind segment to ten miles, could interfere with and delay clearance of the plume path. This suggests that clearing the most hazardous parts of the path of the plume is not playing a sufficiently prominant role in

. emergency planning. The result may be that evacuation planning is unnecessarily increasing the radiological risk to the public.f At least one alternative form of emergency response is well-known -- the graded response conception -- that the staff has found to be much more effective in minimizing radiological risk. It avoids most of the attendant risks of mass evacuations suggested above. This concept of errergency response was first suggested by the Office of Nuclear Regulatory Research in 1983; see the attached references. A careful inquiry into the effect on radiological risk of other planning options is likely to reveal others. It appears that alternatives to large scale evacuation can be found that do a much better job of protecting public health and safety.

I urge the Commission to accelerate consideration of these important and fundamental issues. I would welcome an opportunity to assist in a generic or Indian Point-specific inquiry.

Sincerely, c _ _ _ _

Frank H. Rowsome ENCLOSURES: as stated.

r.

t 'O

REFERENCES:

1. Memorandum from James Martin (NRC/RES) to Distribution, "A Perspective on -Emergency Planning, Risk, and .the Source Term Issue," April 13, 1983.

2. Spector, Hershel, "Lessons from the Indian Point Hearing,"

Nulear Safety. Vol. 27, #83, July-Set .986, p. 305.

3. Surke, Richard P., and Carolyn D. Helsing,. "In-Plant Considerations for Off-Site Emergency Response to heactor Accidents," Health Physics Vol. 46, No. 4, April, 1984, pp. 763 - 773.

4. Spector, Hershel. Transcript of the June 22, 1987 meeting of the ACRS subcommittee on Occupational and Environmental Protection Systems. U. S. NRC.

I I

(

i e

I

ENCLOSURE 1 ASSESSMENT OF THE EFFECT OF EVACUATION ON RISK FOR INDIAN POINT r

In preparing for the ASLB Special Proceeding on Indian Point, the staff (in particular, the Reactor Systems Branch,.DSI, NRR) developed its own source term estimates, working with contractors at Brookhaven National Laboratory. Among the many parameters used to describe the source terms were ectimates of the-warning time

-- the interval The between thetimes warning decision to evacuate associated with theand thediscrete nine time of the release.

releases employed in the staff analyses and testimony are shown in Table III.B.3, taken from Dr. James F. Meyer's testimony on the release categories at the hearing (attached here as the first table). Note that in every case there was at least one hour --

in some cases much more -- between the decision to evacuate and the inception of the release. Such warning times represented the conventional wisdom of the PRA community at the time, originating in WASH-1400 source term conventions.

I have recently interviewed a number of NRC staff members

- involved in evaluating emergency response drills, and they tell me that in the early 1980's, and -- in some drills today -- FEMh and/or state and local authorities are inclined to wait for a release .of radionuclides to the environment to occur before initiating an evacuation.

Some attribute this hesitancy to the desire to be occur protectedif by an liability that might Price-Anderson from the The authorities may want evacuation. proved to be a fa3se alarm.

to verify that an "Extraordinary Nuclear Occurrance" is taking place before authorizing an official call for evacuation.

Others attribute the reluctance to make quick unlike decisions to the police --

fact thatinsenior state of and localveryofficials are not the habit making quick official decisions. The They tend to want to hold meetings and develop a consensus.

social psychology of emergency decision-making, particularly in cases in which the severity or even the occurrence of a release is much in doubt, deserves expert study. Considerable confusion and contradictory information is likely to accompany a severe accident may be accident. The diagnosis of the reactor uncertain, and optimism about the possible success of recovery l actions may vary rapidly as the accident evolves.

It would appear that we were over-optimistic in assuming the

' decision to evacuate would always be made well ahead of the release in our analyses and testimony.

It would be a more i

accurate portrayal of the situation in 1983 to assume a warning i

i 1

l

time of zero hours for all release categories. Even for the

- conditions prevailing today, it would appear that in many cases the decision to evacuate may not reliably precede the release.

Ideally, one should employ careful human factors analyses of the potential for confusion and delays in implementing the offsite emergency response in PRA's; there may be a potential common-mode failure in the operators failing to diagnose correctly the accident in the plant --

or operators . distracted with attempts at recovery actions that they initially b'elieve can succeed -- and delays in initiating offsite emergency protective actions.

I see no reason to revise the estimated delay time (2 hours2.314815e-5 days <br />5.555556e-4 hours <br />3.306878e-6 weeks <br />7.61e-7 months <br />) or evacuation speed (1.5 mph), although a more realistic model would employ a correlated distribution of these parameters, to reflect the fact that some who are particularly quick to depart would encounter less traffic than those who set off after more typical delays.

Unfortunately, the staff did not perform sensitivity studies of the effect of variations on the warning time, or equivalently of delay time, upon consequences. We did evaluate three emergency response models, but only one of these entailed evacuation.

. Therefore, one cannot find in the testimony the data necessary to calculate the effect of revising the warning time. /

We can, however, estimate the effect of the change in warning times upon the most sensitive consequence, early fatalities. Let us consider all reactor accidents except those triggered by earthquakes and hurricanes, as these require a special model of emergency response. With the evacuation model of protective action, we found that virtually all of the early fatalities originated in Release Category B events, none from Release Category C --

due to the prior clearing of the EPZ --

and infinitesimally small contributions from the other release categories. See Table 2 Case 1 for the breakdown of the contributers to early fatalities for Indian Point Unit 2 in the anticipatory evacuation model of emergency response as used in the hearing testimony.

For the particularly important Release Category C -- by far the most probable of the severe releases for both Indian Point units we assumed a warning time of eight hours. Even with an assumed two hour delay time between the decision to evacuate and the beginning of population movement, and the assumed 1.5 mph evacuation speed, eight hours is enough to completely clear the Plume Exposure EPZ ahead of the plume. As one would expect, the early fatalities for Release Category C were calculated to be zero for this case when we prepared the testimony.

This assumption of complete clearance of the EPZ and no early

fatalities for Release Category C events appears to be particularly over-optimistic in light of the evidence that the decision to evacuate might be deferred until a release materializes. Our analysis indicated that in the damage state that gives rise to Release Category C, it is unclear whether the containment will hold or not. This damage state entails early core melt with containment heat removal failed. If containment heat removal can be restored, the containment will very probably hold. Therefore we cannot expect the operators to be able to predict reliably whether such damage states will yie18 a severe release or not. In any event, few utilities have operations personnel trained to do accurate prognoses of such severe accident scenarios. Therefore, we cannot expect accurate, confident, or consistent prognoses of such accidents before the release occurs.

Since this particular late overpressure failure event is the most likely of the potentially severe releases, a revision in the model to portray a post-release decision to evacuate could cause this release category to emerge as one of the dominant contributers to early fatalities. However, the long settling time for radionuclides in containment for this accident takes the lethal edge off the source terms. Even for the 8-hour relocation model of emergency response, the average value of early fatalities for this release category is only 15 deaths. (See Table 2, Case 2 for the early fatalities with the 8-hour relocation model of emergency response.) Therefore, I would expect conditional consequences of the order of ten early fatalities for the deferred evacuation case for this release category.

Now let us consider the effect of shrinking the advanced warning time upon the consequences of Release Category B. Several studies show the effect of stretching the delay time upon I consequences of an SST-1 release, which in many respects resembles Release Category B. Stretching the delay time has the same effect upon consequences as shortening the warning time.

For example, the Siting Study shows effects of one, three, and i five hour delay times on early fatalities for an SST-1 release at Indian Point. Table 2.5-1 in the Siting Study 1 indicates that going from a one hour delay to a three hour delay increases the early fatalities by a factor of 5.11. We are actually interested in shortening the warning time by one hour, so we would expect a smaller multiplier -- perhaps a factor of two or so. The analogy is not exact because the Siting Study calculations assume a much faster evacuation speed --

10 mph instead of 1.5 mph, so the

1. D. C. Aldrich, et al. "Technical Guidance for Siting Criteria Development," NUREG/CR-2239, Sandia National Laboratory, December 1982.

l l

w

. actual casualty ratios may be different.

Another' clue can be found in the recent. draft of NUREG-1150.2 The tables in Section 10.4.1 "Emergency Preparedness", show the effect of delay or warning time changes on the early containment

. failure scenarios for 'Surry. These are- modern analyses of releases very similar- to Release Category.. B . at Indian .. Point. -

Here, again, we see dramatic changes -- roughly a factor'of five in the number of people exposed to' potentially life-threatening doses if the warning time -shrinks or. the delay time grows by 1.5 hours5.787037e-5 days <br />0.00139 hours <br />8.267196e-6 weeks <br />1.9025e-6 months <br />. This confirms the. evidence of the siting study with contemporary source term estimates.

In summary, then, the available results of sensitivity studies' on the consequences of severe reactor accidents supports the hypothesis that the projected early fatalities for Release Category B would have been considerably larger had we employed a-zero warning time instead of the one hour warning time assumed.in the testimony. The combined effect of increasing the early fatalities attributed to Release Category B by a factor of. two, and the introduction of early fatalities associated with Release Category C at an average of 10 per occurrance is shown in Table 2, Case 3. Note that this result yields a higher early fatality risk ' than the default or failed evacuation situation in Case 2.

This default model presumes no protective action until relocatiop takes place, eight hours after the release within ten miles, and.

twelve hours after the release at greater distances.

Of particular interest is the conclusion that the early fatality risk with evacuation begun af ter the release could be worse than that for the default case, i.e., the emergency response we'might have expected before the upgrading of emergency preparedness after TMI. In other words, the calculation suggests that it is plausible -- though by no means certain -- that the attendant risks of mass evacuations, particularly if begun after a release materializes, could make it worse than ad hoc emergency response.

i However, the precise quantitative results are less important than the qualitative policy implications. It is increasingly clear that there are emergency response strategies that are far superior to mass evacuation at saving lives, even if one can be assured that an evacuation will be called well in advance of a release. It is also clear that it is a mistake to be presumptuous about the benefits to be expected of emergency planning: the only alternative is careful analysis.

2. "Reactor Risk Reference Document." NUREG-1150 Vol 1, Main Report, Draft for Comment, February, 1987.

8-

Table 111.8.3 Radiological Releases fu ie Containment Building - CRAC Input ,

D E F G Hi It A B C Release Category NF**

y y (I c pa V, a 6, IR y Associated Failure Mode 9.4 12 3.0 2 72.0 2ft 3 1 13 Release Time (hours) 0.5 0.5 8.0 8.0 8.0 2.0 0.5 0. 5 0.5 Release Duration (hours) 180 0.3 0 0 0.5 98 137 180 Release 5.0 Energy 5

(BTU /hr x 10 )

1 1 67 1 ,,

Warning 1 1 8 1 1 Time 6 (hours)

Release Fraction (fractions of total core inv'entory) 8.5(-1) 1.4(-1)- 1.0(0) 7.0(-1) 5.0(-4)

Xe-Kr 1.0(0)## 1.0(0) 9.6(-1) 1. 3(- 1) 1.0(-1) 1.0(-1) 7.8(-2) 2.0(-3) 4.0(-4) 5.0(-6) 1-Br 8. 0(- 1) 7.0(-1) 9.8(-2) 8.1(-2) 6.2(-2) 9.0(-3) 1.0(-3) 1.0(-5)

Cs-Rb 7.7(-1) 5.0(-1) 3.4(-1) 9.3(-2) 6.4(-2) 4.9(-2) 7.0(-3) 1.0(-3) 1.0(-5)

Te 7.5(-1) 1.0(-1) 3.8(-1) 4.4(-2) 9.2(-3) 7.1(-3) 1.0(-3) 1. 0(-4 ) ~1.0(-6)

Ba-Sr 8.6(-2) 6.0(-2) 3.7(-2) 1.1(-2) 5.6(-3) 4.3(-3) 6.0(-4) 7.0(-5) 1.0(-6)

Ru 6.1(-2) 2.0(-2) 2.9(-2) 5.0(-3) 6.6(-4) 8.6(-4) 6.6(-4) 9.0(-5) 1.0(-5) 2.0(-7)

La 9.8(-3) 2.0(-3) 4.9(-3) 9

Table 2 s .

EARLY FATALITY-ESTIMATES FOR INDIAN POINT UNIT 2 ADAPTED FROM STAFF TESTIMONY BEFORE THE ASLB IN 1983 "AFTER FIX" PLANT ACCIDENT FREQUENCIES Case 1: Anticipatory Evacuation as assessed in hearing testimony i

RELEASE RELEASE CONDITIONAL EARLY FATALITY RISK CATEGORY FREQUENCY

• EARLY FATAL. (DEATHS PER UNIT YEAR)

A 0 2300. O B 4.3 E-7 1300. 5.59E-4 C 1.8 E-5 0 0 D 1.0 E-6 4.8E-2 4.8E-8 E 1.6 E-7 6.5E-2 1.0E-8 F 5.0 E-6 5.8E-2 2.9E-7 6.3 E-5 0 0 G

2.21E-4 0 0 H

3.09E-4 0 0 I

Unit Total: expected early fatality risk excluding earthquake and

. hurricane induced accidents: 5. 59 E-4 deaths per unit year with anticipatory evacuation.1 /

Case 2: Early Relocation Model of emergency response as assessed in hearing testimony RELEASE RELEASE CONDITIONAL EARLY FATALITY RISK CATEGORY FREQUENCY

• EARLY FATAL. (DEATHS PER UNIT YEAR)

A 0 3070. O B 4.3 E-7 1550. 6.6?E-4 C 1.8 E-5 15. 2.7 E-4 D 1.0 E-6 2.6 2.6 E-6 E 1.6 E-7 3.19 5.1 E-7 F 5.0 E-6 3.68 1.84E-5 G 6.3 E-5 1.0 E-3 6.3 E-8 H 2.21E-4 0 0 I 3.09E-4 0 0

1. Accident frequencies and risks exclude accidents triggered by earthquakes and hurricanes, as these require different emergency response: modeling.

(:~

Table 2 Unit' Total: expected early fatality risk excluding earthquake and hurricane induced accidents: 9.59 E-4 deaths per unit year with early relocation.

Case 3: Evacuation following release; estimated values RELEASE RELEASE CONDITIONAL EARLY FATALITY RISK CATEGORY FREQUENCY

• EARLY FATAL. (DEATHS PER UNIT YEAR)

A 0 4000. O B 4.3 E-7 2600. 1.12E-3 C 1.8 E-5 10. 1.8 E-4 D 1.0 E-6 2. 2.0 E-6 E 1.6 E-7 2.5 4.0 E-7 F 5.0 E-6 3.0 1.5 E-5 G 6.3 E-5 0 0 H 2.21E-4 0 0

. I 3.09E-4 0 0

/

Unit Total: expected early fatality risk excluding earthquake and hurricane induced accidents: 1.32 E-3 deaths per unit year with evacuation following release.

)

e ENCLOSURE 2 IMPLICATIONS FOR THE INDIAN POINT DECISION The ASLB, in its "Recommendations to the Commission," dated October 24, 1983 (ASLBP 81-466-03-SP) depended upon the staff's estimates of the bottom line risks for Indian Point Units 2 and 3, but took little note of the staff testimony on the significance of emergency preparedness upon risk. The Commission, in its ruling, likewise depended upon the risk assessment but did not depend upon the staff assessment of the relationships between emergency planning and risk.

Within the staffs' assessment of risk, it was found that the contributers to risk originated predominantly in earthquake- or hurricane-induced accidents. Only those earthquakes or hurricanes powerful enough to devastate the entire plume exposure EPZ have the potential to cause reactor accidents. In light of the problems such an initiator would cause for ground transportation and communication in the EPZ, these accident scenarios were pared with a model of offsite emergency response that did not credit evacuation. Rather, we assumed that thy population would remain r place, without shelter, for 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> after plume passage. These accident scensrios would thus be unaffected by changes in the models of evacuation.

Such scenarios so dominated early fatalities that only 10.2% of the early fatalities originate in all the other accident scenarios for which evacuation might be a viable option, and 89.8% originate in earthquake and hurricane induced accidents.

This result is displayed, for example, in Dr. Sarbes Acharia's testimony Table III.C.10, covering both Indian Point units. As a result, even if the early fatalities attributable to the "evacuation" scenarios were to increase by a factor of two (or, alternatively, ten), then the bottom line early fatality risk would be increased by 10 % (alternatively, a factor of 1.92).

Other consequences are less sensitive than early fatalities to changes in the evacuation model.

The absolute value of the early fatality risk, as assessed by the staff, was very low (less than .02 early fatalities per site year --

considering both plants together and all contributory accidents), and the early fatality risk would remain very small even with dramatic increases in the early fatality risks

, associated with evacuation scenarios. The plants would continue to satisfy the Commission's quantitative safety goals.

Therefore, I am quite certain that the ASLB and the Commission, in its decision on the case, would still have concluded that the

• plants need not be shut down and that the contended backfits bearing upon plant design need not be required.

However, had the staff been aware that the attendant risks of mass evacuations were as large as I now believe they were -- and mz y still be --

and had we known that a superior emergency reaponse strategy was available -- as RES documented later in the spring of 1983,1 the testimony on the effect of emergency preparedness upon risk would have been very different/ and could well have affected the outcome of the proceeding on emergency planning.

By way of background, in the summer of 1982 the preliminary results of the re-evaluation of the licensees' Indian Point Probabilistic Safety Study suggested that the core melt frequency might be nearly one in a thousand per year for Indian Point Unit

2. Harold Denton, then Director of NRR, took the position that

' NRR had a continuing responsibility to protect public health and safety, and so should not wait for the hearing to take corrective action should the staff turn up evidence of undue risk. Denton regarded so high a core melt frequency --

with its particular association with fairly severe releases -- as indicative of undue risk. Therefore, the staff was engaged in searching out

, corrective actions for the particularly prominent contributers to Denton was prepared to order backfits if that had provep risk.

! to be necessary. As it happened, it was not necessary.

The licensees were also preparing improvements to lower the l risk of the plant at this time, based upon their PRA. The staff and licensees' lists of improvements crystalized just before the 1

hearings on risk in the winter of 1982-83. In the hearing itself, the licensees and the staff compared the concepts for backfitting the plants each had prepared independently, and the licensees agreed to implement those improvements that were on the NRC list l

that they had not already planned to implement. Therefore, it was not necessary for the NRC either to order fixes or to introduce contentions into the proceeding that such improvements should be included in the resolution of the case.

l l Had we known then that the attendant radiological risks I

associated with the prevailing emergency plan were higher, we would have looked for alternative emergency planning bases to

lower the risk. Since RES was developing evidence at the time l that the graded response conception is very much more effective i

than mass evacuation in reducing early radiological casualties, we would have found a superior alternative. It seems likely that we would have recommended that the graded response be ordered to l __________

1. See Reference 1.

l 1 i

l -. ,,

be the planning basis for Indian Point emergency response. In any event, I would have recommended such a change to the Office Directors, the ASLB, and the Commission.

In short, it is quite possible that the new information -- had it been available at the time of the proceeding --

would have yielded a very different outcome with respect to emergency preparedness. The NRC might well have chosen to require a change in the planning basis --

replacing the conception of mass evacuation on notice as the ultimate emergency response strategy with something along the lines of the graded response strategy.

k.

F' /

S d

.mauuunr.::uummm:

4 i

)

/