-

severe reactor accidents. We found that even a cursory review suggests that the path being pursued by PSNH and the NRC staff is ill-considered, and that many acknowledged experts including some within the Commission itself -- have offered comments that reflect poorly on the effort to reduce the size of the EPZ. We provide a few examples below.

On the need for planning for serious reactor accidents, from the report of the Kemeny Commission: ..We must not assume that an accident of this or greater seriousness cannot happen again, even if the changes we recommend are made. Therefore, in addition to doing everything to prevent such accidents, we must be fully prepared to minimize the potential impact of such an accident on public health and safety, should one occur in the future." Report of the President's Commission on the Accident at Three Mile Island, p. 15.

On the question of the accuracy of probabilistic risk assessments estimations of core meltdown probabilities: "I believe that the recent Davis-Besse event illustrates that, in the real world, system and component reliabilities can degrade below those we and the industry routinely assume in estimating core melt frequencies. Our regulatory process should require margins against such degradations and also to reflect the uncertainties in our PRA estimates." Harold Denton, NRC Director of Nuclear Reactor Regulation, in memorandum dated June 27, 1985, to William J. Dircks.

"There is a

On the possibility of severe reactor accidents: distinct possibility of one or more additional severe reactor accidents, beyond the one at Three Mile Island, in the remaining life of the plants now in operation or under construction, unless the estimated accident frequency declines sharply with modifications, or has been significantly overestimated in current PRAS and actuarial inferences. " -- NUREG-1070, "NRC Policy on Future Reactor Designs: Decisions on Severe Accident Issues in Nuclear Power Plant Regulations," August 1984, p. 108.

On the possibility of an accident in the U.S. as severe as the Chernobyl accident: "...given the present level of safety being achieved by the operating nuclear power plants in this country, we can expect to see a core meltdown accident within the next 20 years, and it is possible that such an accident could result in off-site releases of radiation which are as large as, or larger than, the releases estimated to have occurred at

.

Chernobyl. NRC Commissioner James Asselstine, in testimony before this Subcommittee, May 22, 1986.

Even these few examples suggest that both the NRC staff and PSNH are proceeding under a dubious assumption when they attempt to demonstrate an invincible technical basis premised on containment strength and severely limited accident scenarios in arguing that the Seabrook plant should be considered essentially immune to severe accidents.

21

PRIOR NRC COMMENTS ON CONTAINMENT PERFORMANCE

In recent times the Subcommittee has had numerous exchanges with the NRC regarding emergency planning, containment analysis, and source terms. Several statements by the NRC have either directly contradicted the positions now being taken by NRC staff and the PL&G consultants or have emphasized that the present state of the art is characterized by substantial uncertainties in analytical techniques. It appears evident the NRC staff has ignored such qualifications in its attempt to assist the utility to justify a reduction in the size of the EPZ.

The utility, its consultants, and PL&G have argued that the containment failure pressure is so high as to make a severe accident implausible. However, in a May, 1986 exchange between the Subcommittee and the NRC, the NRC clearly stated that uncertainties in containment analysis were too great to permit a precise estimate of containment failures. The exchange reads as follows:

QUESTION 5.

What degree of confidence does the NRC have in the ability of different containment buildings to prevent a major release of radiation during a core meltdown? For each type of containment building, what is the estimated probability of containment failure given a meltdown and state precisely what uncertainty bounds are assigned to this estimate and how it was calculated?

ANSWER. At present, the NRC staff cannot
specify with a high degree of preciseness, the
conditional probability (and uncertainty
bounds to be assigned) of containment failure
with a major release of radiation....There are
very low probability severe accident
conditions under which a containment may be
unable to prevent a major release. Although
containment structures are conservatively
designed to withstand the substantial

Such

temperatures and pressures associated with a
major pipe rupture...they are not designed to
withstand the additional challenges that might
be associated with a complete core melt.
challenges include phenomena such as increased
pressures from an uncontrolled hydrogen
combustion or release of large quantities of
noncondensible gases from core-concrete
interactions....

The exchange in Question 6 reiterates these inherent uncertainties.

QUESTION 6: In a supplement to the record of
the Subcommittee's April 17, 1985 hearing, the
NRC staff wrote: "Analysis shows that there
are some kinds of accident sequences that
could cause failure of any containment
design.... For each type of containment
please enumerate each mode of potential
containment failure and the conditions that
can lead to each mode of failure. What is the
relative likelihood of different modes of
containment failure?

ANSWER. The potential containment failure
modes for severe accidents in all U.S. LWR
designs can be generally classified into six
groups. The definition of these groups
depends on the timing of the failure (relative
to core melting and major releases of
radioactive material into the containment) and
the failure location. These groups are:

2.

3.

1. Early failures directly to the atmosphere;
Early failures into other plant buildings;
Late failures directly to the atmosphere;
4. Late failures into other plant buildings;
5. Late failures into the ground;
6. No containment failure.

The exact failure modes and causes, and their
relative likelihoods, vary considerably among
plant types and even among plants of similar
containment design....The NRC staff and
supporting contractors are presently engaged
in a major reassessment of the risks of
current commercial reactors. More

23

quantitative information on the types of
containment failures and their relative
likelihood and importance will be provided in
the staff report NUREG-1150...

For PWR's with "large, dry" and

"subatmospheric" containment designs [like
Seabrook], early failures to the atmosphere
have been predicted to result from accident
phenomena such as direct heating, steam
explosions, and coincidence of hydrogen burns
and rapid steam generation. For most plants
with a large dry containment, the likelihood
of these events including an early containment
failure is quite small.

The above comments demonstrate that there are no ready answers to questions regarding containment failure. If the NRC is unable to supply any hard numbers and is reduced to citing analytical uncertainties and studies in progess, it is difficult to understand how in three short months PL&G could accomplish an analysis sufficiently well-grounded to warrant a change in emergency planning regulations for the Seabrook plant.

POTENTIAL TECHNICAL PROBLEMS WITH THE UTILITY'S CLAIMS

The utility's consultant has asserted that the pressure at which the containment would fail is significantly higher than for most other similar large dry containment pressurized water reactors. They attribute this greater strength to several factors, including a larger than average containment volume and more steel used in the construction of the containment. The Subcommittee staff questions whether PL&G has failed to consider adequately a number of significant phenomena and possible reactor accident scenarios, several of which could possibly result in outcomes very different from those cited by PL&G. A few examples of these technical questions may be summarized briefly as follows.

(a). The conclusions of the PL&G report depend crucially on the assumption that the containment failure pressure is roughly 225 psig. This value is much higher than the estimated containment failure pressures for other plants, and may not be adequately supported. In addition, containment failures generally may be assumed to occur at points of structural discontinuity

24

(e.g., where piping enters or exits the containment). Even if the shell of the containment is stronger than average, one does not expect a shell rupture to be the first failure point of the containment.

(b). PL&G examined only one containment bypass scenario; however, other bypass scenarios may well be relevant and could produce releases greater than those estimated in the PL&G study.

(c).

PL&G has argued that if noble gases are released from the containment after an accident, the containment will hold them long enough to permit decay to levels low enough to warrant a reduction of the EPZ. However, emergency planning as currently conceived assumes that off-site dose-response relationships must be examined without any protective actions. It appears that if noble gases are released without a significant delay, under various realistic worst-case weather scenarios, the acceptable off-site dose levels might well be exceeded at 10 miles.

(d). Assumptions about the radioactive inventory released under high pressure melt ejection scenarios may not be consistent with recent analyses conducted by Sandia National Laboratory on these phenomena.

(e). The report may not adequately consider the prospects that accident recovery actions could cause steam de-inerting by condensation, leading to hydrogen burns or detonation within the containment.

(f). In examining earthquakes as a contributor to radioactive release scenarios, the report may not adequately consider after-shocks while the containment is pressurized, which adds an impulse pressure load.

(g). The report's conclusions on radioactive releases as a consequence of steam generator tube ruptures may not be consistent with recent analyses conducted by Battelle Columbus Laboratory on steam generator tube rupture scenarios.

In summary, it appears that PL&G may have been selective about their assumptions; if one made other, equally reasonable assumptions, the outcome might be much different.