ML20038B996
| ML20038B996 | |
| Person / Time | |
|---|---|
| Issue date: | 11/06/1981 |
| From: | Bender M Advisory Committee on Reactor Safeguards |
| To: | Siess C ILLINOIS, UNIV. OF, URBANA, IL |
| Shared Package | |
| ML20038B995 | List: |
| References | |
| RTR-REGGD-01.023, RTR-REGGD-1.023 NUDOCS 8112090404 | |
| Download: ML20038B996 (7) | |
Text
__
fVh o
UNITED STATES
~g NUCLEAR REGULATORY COMMISSION e,
a e ADVISORY COMMITTEE ON REACTOR SAFEGUARDS O
g WASHINGTON, D. C. 205S5 s
l l
....+
j November 6, 1981 l
Dr. Chester P. Siess Professor Emeritus of Civil Engineering 3110 Civil Engineering Building University of Illinois Urbana, IL 61801
Dear Chet:
Coments on Regulatory Guide 1.23 As you know I won't be able to attend the December Regulatory Guide meeting when Regulatory Guide 1.23 will be discussed.
If it is to be released as a basis for regulatory requirements concerning emergency response, we ought to first discuss it with the Commissioners.
I am convinced that the NRC treatment of site meteorology has a badly distorted emphasis for all purposes, but it is at its worst for emergency planning and response applications.
Even though the monetary cost associated with the requirements is relatively small compared to total power plant costs, it is extremely high when related to the actual worth of the information.
Of more concern, however, is the waste of time and valuable manpower resources by diverting them from more important safety matters really needing attention.
In a fit of sheer exasperation I prepared the attached coments on Regulatory Guide 1.23, but there wasn't time to review my discussion in detail, and most of it came from mental recollection. Still, I don't think there are any serious errors in it. We should recomend to the Comissioners that Regulatory Guide 1.23 be abandoned because its purpose is no longer in keeping with current regulatory safety approaches.
When the guide was introduced, we were using the traditional arbitrary accident source term derived from small fuel melting experiments to establish containment leak tightness. The nuclide releases were based on a uniform mixture of iodine, noble gases and particulates within the contained atmos-phere.
We know then and now that this was a poor assumption but it was convenient for safety analysis purposes and it is very conservative. The meteorology was mainly used to show the dispersal of radionuclides from the containment at the prescribed leak rate.
It was completely artificial.
The use of such complicated meteorological analysis to show site boundary limits was not much better than using astrological principles to predict core melt.
It was always a lot of mystical hocos pocus.
8112090404 8111ie PDR REVGP NRCCRGR
_. __ __N
Dr. Chester P. Siess November 6, 1981 Since TMI-2 we have known that the approach was nonsense. We need to consider realistic source terms, but even if we keep the arbitrary source term basis we shouldn't try to use sophisticated meteorology based on very localized measurements to determine site boundary radiation exposure.
Nobody with even a microgram of common sense would think we could use such data to show radiation exposure of the bulk population following a major accident.
I suspect that much of what is proposed in Regulatory Guide 1.23 is the product of ACRS inquiries, improperly interpreted, and of licensee pencil sharpening to obtain approval of the less attractive sites, where stagnant air conditions and inversions caused concern for the dilution rates of nuclides dispersed to the environment. We now under-stand that we have not defined what is to be diluted. All we know is that leakage rates will be higher than assumed for containments but internal containment decontamination will be a great deal higher than-was credited. These are offsetting effects that have never been quantitatively related.
I, personally, believe that the decontamination effects are much more of a factor than leak tightness, so the situation will be better than expected, barring, of course, pressurized rupture of containment by some very low probability circumstance.
Such gross ruptures were never treated in the dose analysis anyhow.
I don't want to discredit the technological approach used in the Regulatory Guide.
If one is going to use a complicated plume analysis i
to determine mass transport of nuclides and if the leakage.is also small, making its behavior dependent on meteorological conditions, then a knowledge of air layer movement and diffusion will be needed. However, if the leakage is small enough for such phenomena to be controlling, then the internal decontamination effects will reduce the nuclide l
outleakage to levels where their radiation effects can be ignored.
What bothers me most about this analysis is that if public safety were really dependent on such sophisticated understanding of meteorology, we would have great difficulty defending the regulatory posture. We are not depending on such analysis but rather using it as procedural rote intended as a reminder of important phenomenological considerations that affect airborne nuclide transport. We could handle it much simpler if we recognized the decontamination effects of low leakage events, permitting us to ignore the meteorology.
We should remember that when nuclear power was initially being developed it was common practice to vent gaseous nuclides to the atmosphere.
ALARA had not been invented for effluent control. We used the meteorology to assure favorable conditions for venting.
For that application covered by 16 CFR 20 requirements the meteorological interest made more sense.
Our current ALARA limits make this venting issue a moot.
Dr. Chester P. Siess November 6, 1981 1.can't help but note the similarity between this situation and the air-controller strike circumstances.
In both cases the " overkill" regulatory practice wasn't recognized until the real need was tested.
It is obvious in both instances that safety practice was carried far beyond real need by the momentum of the system. Since we are pushing-risk evaluation principles in regulatory practice, this type of calcula-tional mysticism should lead the list of practices to be discarded.
Sin erely, M. Bender MB/cw Attachment cc:
S. Duraiswamy
' R. F. Fraley e
MBendsr Page 1 COMMENTS ON REGULATORY GUIDE 1.23
Background
Regulatory Guide 1.23 has been used for a number of years as the basis for evaluating site acceptability for nuclear power plants.
Its roots lie in the basic accident assurrptions used in site analysis.
In the early development of nuclear power plant safety principles it was presumed that certain forms of radionuclides would be released and the consequences of such releases would be minimized if sites were selected with favorable meteorology in terms of infrequent air inversions, quiescent air conditions or other factors that would reduce the dilution effects on the nuclides that might be released to the surrounding environment.
This was a logical position in that it avoided selection of less desirable plant sites when more suitable ones were available.
When 10 CFR 100 became the basis for site acceptability, the regulatory evaluation process, through its use of prescribed analytical methods using arbitrary accident assumptions, introduced an arbitrary but rigorous method of analyzing site boundary accident effects. The calculational methods became highly refined because some sites could only be shown to meet very conservative 10 CFR 100 limits by refining the interpretation of meteoro-logical conditions.
The regulatory staff, in order to apply sophisticated meteorological analysis required substantiating data. This in turn led to the requirement for meteorological towers instrumented to obtain data over an extensive time span.
Such data were intended to permit integration of radionuclide release effects over a long period of time.
The data were never applied to actual accident release conditions.
The data requirement became ingrained in the Construction Permit processing even though for good nuclear sites it was of little value.
The data was only needed because of the arbitrary calculational orocedures used for evaluating conformance with 10 CFR 100 limits. The data was not of use for the TMI-2 accident and it would not be of use for accidents of more serious nature since other factors dominate the public risk considentions.
MBender Page 2 Factors Affecting Airborne Radioactive Dispersions In core' damage accidents the physical characteristics of airborne radioactivity are understood only in a gross qualitative manner. The forms of the radionuclides when released from the fuel vary with time because of nuclide radio-decay. The chemical form of the nuclides depends upon reactions with other materials, the most interesting reaction being the cesium iodide combination which ties two important radioactive constituents together in a chemical compound.
The physical movement of nuclides from the fuel to the external containment environment depends upon the pathway of release (calculated by the Corral, March, Contain Codes or analagous techniques), the containment constituents of which moisture and dust particulates are probably most important, and the temperature and pressure in containment when the release is occurring. These factors represent two to three decades of uncertainty in the nuclide concentrations that could leave the contained atmosphere.
By comparison, the most refined t
meteorological analysis car improve over the crudest practice by no more than a factor of three.
While we may be able to improve calculational 1
procedures within containment, we can never establish the in-containment accident environment well enough to make the external meteorological conditions a determining factor in evaluating human radiation exposures prior to, during, or subsequent to nuclear accidents. The meteorological data are of use mainly in timing controlled release of radionuclides over long time spans in the range of months and years.
The noble radioactive gases, mainly Krypton 85, will mix with the contained atmosphere and could be released in accidents. However, the quantity is very small even for a full core of spent fuel and when dispersed in the external air it will dilute regardless of meteorology to a concen-tration which will not cause measurable radiation injury at ground level.
Hence, for accident purposes it can be ignored with respect to human exposure during accidents where questions of evacuation, escape routes, and local radiation exposure might be involved.
Dispersal of Airborne Activity Released from Containment Air plume models are used to analyze the dispersion of particulate radionuclides assumed to be released from containment. The models always
MB:nder Page 3 assume a point source and are usually analyzed by assuming several release locations. Ground level (a few meters above the surface) and locations at the high point of containment or the containment stack usually bracket the effects.
The shape and energy content of the plume are the important considera-1 tions in bulk releases such as a containment rupture. Only gross variations in environmental conditions could be treated in the analysis and many of those would be of marginal importance.
The existence or nonexistence of rain, mist, or fog is a major factor. An air inversion might have an important effect for determining dispersal rates, but analysis probably can only discriminate between a stagnant air velocity, one of a few MPH or a few tens of MPH. This will indicate how fast the plume will move laterally, but higher velocity winds will also mean rapid plume mixing and dispersal. The combined effect cannot be calculated with meaningful accuracy.
Variations in air temperature and air layerine are of importance only for determining exposure if one were measuring integrated effects for long periods of time when the release is continuous and the physical nuclide form is known. The meteorological tower measurements have meaning only in the localized setting where the measurements are made.
They cannot be extrapolated for miles.
Integrated effects will be determined primarily by air and surface activity samples taken subsequent to the release.
The meteorology would hardly enter the evaluation process.
The CRAC code is the analytical technique currently in use for computing airborne accident effects.
It assumes an accident and then analyzes the release of the activity from a point source in containment. When last reviewed by the ACRS, it still did not account for gross meteorological variations and it did not have the capability to treat air layer diffusional properties.
Whether it has or could be refined for this purpose is unclear, but considering other uncertainties the value of such refinement is doubtful.
Nevertheless, if the data required in Regulatory Guide 1.23 is of any use it would have to be shown to be needed in a CRAC-code type of analysis.
Meteorological Factors of Importance When Nuclear Accidents Occur There are some meteorological considerations of importance when an accident of serious nature involving core melt occurs. These are:
9
~
(
MBender
-,.,1 Page 4 i
1.
Wind directions, if they can influence the time and route of evasuation.
2.
Variation in wind conditions as-indicated from weather service sources when a planned release such as venting is intended.
3.
Air stagnancy (inversion) circumstances that might influence dispersal and cause undesirable localized conditions during a planned release.
4.
Flood or tornado conditions that might jeopardize recovery operations. This type of information should come from local weather and emergency warning services.
Value of Regulatory Guide 1.23 With our present and prospective radionuclide dispersal knowledge, Regulatory Guide 1.23 is not of value for emergency evaluation purposes.
The expressed need based on requirements stemming from NUREG 0654, 0696, and especially NUREG 0737 only indicates that the reference documents are wrong and need correction.
This interest in refined meteorological analysis stems totally from the arbitrary analytical procedures used in determining site suitability to 10 CFR 100 conditions. More than likely, NRC could conserve its own and licensee resources by simplifying the computational procedures and putting more emphasis on such accident assumptions as containment leakage, nuclide dispersal within containment, and lapsed time under which accident conditions prevail.
4
- - -