OPTIMIZATION OF DOSE ASSESSMENT MODELS INCLUDING THE INTERFA CE WITH ENVIRONMENTAL SURVEY, FOR USE IN CASE OF ACCIDENTAL RELEASES

Objectif

THE RESEARCH PROJECT AIMS TO DEVELOP A DOSE ASSESSMENT AND FORECASTING SYSTEM, OPTIMIZED AS AN AID FOR DECISION MAKING, SUBSEQUENT TO A SEVERE NUCLEAR ACCIDENT. THIS SYSTEM WILL BE DEMONSTRATED BY SIMULATION OF REFERENCE RELEASES AND APPLIED TO GENERATE SCENARIOS FOR EMERGENCY RESPONSE EXERCISES. THE FEASIBILITY OF ADJUSTMENT OF THE MATHEMATICAL SIMULATION OF THE ENVIRONMENTAL IMPACT BY FEEDBACK OF SURVEY RESULTS WILL BE EVALUATED. THE RESEARCH WORK IS COVERED BY THREE PROJECTS.

PROJECT 1 :

OPTIMISATION OF AN EMERGENCY DOSE ASSESSMENT AND FORECASTING SYSTEM.

PROJECT 2 :

FEASIBILITY STUDY OF FEEDBACK OF SURVEY RESULTS.

PROJECT 3 :

APPLICATION OF THE DOSE ASSESSMENT AND FORECASTING MODEL TO GENERATE REFERENCE SCENARIOS FOR EMERGENCY RESPONSE TRAINING.
Following the general evolution in the world of nuclear safety analysis and emergency preparedness at nuclear sites, real time emergency assessment software packages have been developed for use during accidental situations. Until 1984, those packages, known as computer aided emergency response systems (CAERS), mainly performed atmospheric dispersion calculations. Since then the system has been optimised and extended to define realistic source terms to be included in dose assessments covering all exposure pathways of importance for decision making during the early emergency response phase. This main package is known as CAERSDOS.

CAERS:
Basically the procedure CAERS, including an online version and an offline version, computes using a bigaussian dispersion model, half hour mean concentration values and deposition fluxes for a defined receptor configuration.
A meteorological mast provides all the necessary parameters to run the model. These parameters are stored in a monthly cycling database, which permits the use of any half hour of the last 31 days. The offline version allows other combinations of turbulence typing schemes and dispersion parameter sets.
The windvector puff trajectory model determines the centre of most of individually released puffs using 1 minute values for the windspeed and direction stored in the meteorological database.
The trigaussian puff dispersion model allows a variable source term combined with a variable meteorological input.

CAERSDOS can be considered as the natural extension of CAERS towards a dose assessment model for use during accidental situations, as well as a tool to provide emergency scenarios for training purposes. The backbone of CAERS has been retained.

The study of the detailed requirements of the system lead to the selection of radionuclides and exposure pathways to consider, based on a review of existing safety studies. The radionuclides are classified according their important for early effects, late effects co mmitted during the passage of the cloud, those due to external irradiation by deposited materials and those due to consumption of contaminated food. The criterion of 1% of the total dose due to a specific pathway, lead to the selection of 37 radionuclides to cover the 4 kinds of exposure.

The selected exposure pathways are in order of importance: irradiation by deposited materials, inhalation and external irradiation by the cloud. A subroutine, dealing with the dynamic behaviour of iodine-131 in the grass-milk pathway can be linked to the system. A more extensive and systematic handling of foodchain contamination will be included in the near future.

A relative simple semianalytical source term simulator model hasbeen developed to provide the core inventory for a large pressurised water reactor (PWR) in function of power level, burn up and cooling time. This simulator is developed to define source terms for the 44 isotopes for up to 21 time intervals of half an hour. The user has to define the release fraction of the inventory in function of time for each physicochemical group of radionuclides.

A dose calculation module has been developed whereby for each half hour for which a source term has been provided, dose calculations are performed. The user can either introduce self defined meteorological situations or rely on the meteorological database.
The doses are assessed for the following exposure pathways: cloud shine (effective and skin), inhalation (effective, lungs, thyroid and red bone marrow), external exposure by deposits (effective and skin for integration intervals of 1 hour, 1 day, 1 year and 30 years). The cloud shine dose is performed using the numerically determined ratios of finite to semiinfinite cloud shine.
The user has to provide the fractions of elementary, organic and particulate iodine-131 to determine the effective deposition fluxes. The ingestion dose calculations, projected to be included very soon, will provide for the receptor with the greatest deposition the maximal concentration in milk, meat, green vegetables, grain and roots for the isotape iodine-131, caesium-137, strontium-90 and ruthenium-106.

The project evaluated the feasibility of updating some input parameters to the dose assessment model by comparison of environmental survey results to the corresponding values predicted by the model.
The study examines the possibility of feedback of early environmental survey data to the model to increase the confidence in its results.
Virtually all environmental survey data available up to now consist of concentration data gathered during tracer experiments. Therefore our attention has mainly been focussed towards the interaction of air concentration measurements on predictions. In simple terrain the main sources of uncertainty are source term, effective height, wind direction, dispersion parameters, and wind speed. The first optimisation scheme is based on the comparison, using a subjective acceptance criterion, of measured and calculated data. The main drawback of this procedure is the possible generation of solutions which fit well mathematically but without relations with the physical reality. Therefore a second optimisation scheme was considered which is based on a regression of measured data, assuming the validity of a general dispersion model. As will be shown in the tracer experiment examples, the best way to obtain a realistic set of model parameters is to sample on a few arcs centered around the ground level projection of the plume axis and located at both sides of the effective height dependent location of the ground level concentration (glc) maximum.

Relations between measurements and calculations of concentrations involved simulation of measurements with an exact bigaussian model and simulation of realistic measurements.

3 simple statistical measures or norms without arbitrary weights and with straightforward physical interpretation have been selected to establish conditions under which environmental data can be regressed under real time conditions namely chi square, relative Chi square and log norm.

The principles of feedback have been appli ed to a variety of controlled tracer experiments involving 1 hour mean air concentration. Feedback of environmental survey data can be applied in most noncomplex situations considering an adapted sampling scheme. An independent variation of model parameters using the relative chi square norm can fulfill the basic requirements. To reduce computer time a simple analytical (eg gaussian like) model is to be preferred. The sampling must be performed on several rings at several distances from the source. The samples must be within 2 to 3 delta y values from the ground level plume axis projection, rejecting a nongaussian background. A real time graphical output on video display can be very valuable. An inherent uncertainty factor, especially for the wind direction, has always to be considered. The effective plume height and the source term will have to be found from monitoring data and from the spatial variation of the glc. Finally, it is believed that as long as an accidental situation remains simple enough feed-back will be a very valuable tool, although without being a universal panacea to resolve all unknowns in the early investigation phase.

A generic methodology to define emergency response scenarios has been developed. This method should be implemented in an organised way on a personal computer (PC) and illustrated by a set of examples.

The definition of an emergency response scenario requires:
an analysis of the emergency preparedness organisation;
the description of the nature and severity of potential releases;
an analysis of the emergency decision logics.

The basic structure of the organisation for which scenarios will be prepared comprises 2 coordination units the implant coordination unit for which the nuclear facility operator is responsible and the offsite coordination unit for which offsite authorities are responsible. The scenario has to provide the necessary input information for the implant coordination unit.

The functions to be tested are as follows:
communications;
activation of emergency organisation (notification);
gathering of information;
evaluation of accidents;
implementation of actions (countermeasurements involving) corrective/mitigatory actions, protective actions, aid to affected persons, radiological survey in the environment and information to the public, authorities, neighbouring countries, international organisation.

The following elements defining the scenario, have to be assessed in accordance with the requirements for the testing of the functions concerned:
the type of the nuclear facility;
the type of incident;
the meteorological situation in function of time;
demographic characteristics;
results of the radiological survey in function of time at specific reference points in the environment.

The requirements for the functions concerned to be generated may be more of less stringent for the selection of the accident scenario. Some functions (eg communications) so not require a specific scenario in order to be generated. Other functions require only a threshold on released quantities or on individual doses to be exceeded or a specific type of inc ident to have occurred. Still other functions require a more detailed scenario (eg individual doses situated between upper and lower bounds at certain places and with a certain distribution in time). Finally the elements are determined for the implementation of protective measures for the public in the early phase of an accident. The ultimate criterion for the decision on those measures is the individual dose. The decision making can be schematised through a so called decision tree. Decision trees or protective measures for the public (early phase) have been developed for 2 cases (the release has not yet started or is already ended).
3 basic principles are applied:
for a certain countermeasure to be imposed, a specific reference level (threshold) of individual dose has to be exceeded;
with less drastic countermeasures the individual dose remains higher than this reference level;
the imposition of the countermeasure has to bring about a sufficiently high benefit (=decrease of individual dose).

In view of a decision logic tree, the elements of the scenario can be determined in accordance with the countermeasure intended. As will be observed, a certain degree of freedom remains, depending on the action envisaged. It is possible to make use of this when several functions have to be tested simultaneously. The methodology foresees a first approach with a limited number of standard release scenarios and meteorological conditions. In this way the order of magnitude and the kind of meteorological sequence needed to reach the objectives of the exercise can be defined. In a second phase the elements of the scenario are to be refined.

The incorporation of an environmental survey monitoring data feedback module in the comprehensive decision support system for nuclear emergencies in Europe has been discussed. The main conclusions were as follows.
An off line module will be developed, with an input output compatability with the main system. The module can be inserted in the system under the control of the user and has to be considered as a tool for the user to explain differences between predictions and observations. It remains up to a user's decision to correct the predictions in the way proposed by the feedback module and to associate weighting factors to the reliability of individual monitoring results.
Regarding the complexity of the exercise, a feedback exercise has to be based on a bigaussian atmospheric dispersion model. Eventually it can be organized to confirm, or correct in a first instance, the trajectory of the center of mass of the releases, followed by a correction of source term, height of release and dispersion around the trajectory.
Parameters, optimized by the application of an optimization loop to a simple model, can then be introduced in a new assessment applying a more complex dispersion model.

The previously developed numerical optimizatiion technique has been demonstrated on the measuring data collected by a set of gamma exposure rate monitors. Alternative methods for handling the differences between predicted and observed environmental data and fuzzy logic and analysis of spatial distribution of predicted to observed ratios. These are being considered in the decision on the approach to be used in the European decision support system under development.
PROJECT 1 : THIS DOSE ASSESSMENT MODEL HAS TO BE CONSIDERED AS AN EXTENSION OF THE EXISTING MODELS AT CEN/SCK. A SPECIFIC RESEARCH AREA IS THE COUPLING OF A DOSE ASSESSMENT MODULE TO A TRI-GAUSSIAN ATMOSPHERIC DISPERSION PUFF-MODEL.

THE DOSE ASSESSMENT WILL CONSIDER NUCLIDES AND EXPOSURE PATHWAYS, RELEVANT FOR DECISION MAKING DURING THE EARLY EMERGENCY RESPONSE PHASE, IN CASE OF A SEVERE ACCIDENT AS CONSIDERED BY THE CURRENT PROBABILISTIC RISK ASSESSMENT STUDIES. AS SELECTION CRITERIUM IS SUGGESTED TO CONSIDER EACH NUCLIDE WHICH CONTRIBUTES MORE THAN 1% TO EARLY DOSES DUE TO CLOUD EXPOSURE, INHALATION AND EXTERNAL IRRADIATION OF DEPOSITS.

THE SYSTEM WILL ALLOW TIME VARIANT SOURCE TERMS AND CONSIDER DRY AND WET DEPOSITION AND RADIOACTIVE DECAY. A SEPARATE MODULE WILL BE DEVELOPED TO ASSESS DOSES WHICH MIGHT ME RECEIVED BY INDIVIDUALS, ACCORDING DIFFERENT PROTECTIONS STRATEGIES. THE MODEL IS OPTIMIZED TO BE USED FOR THE OTHER PROJECTS OF THE CONTRACT.

THE CONTRACTOR WILL DESCRIBE THE MODEL AND SHOW ITS OPERATIONALITY IN NON-COMPLEX TERRAIN FOR DISTANCES UP TO 20 KM BY APPLICATION TO SOME SCENARIOS, SELECTED REGARDING THE OUTCOMES OF THE PRESENT RESEARCH ON SOURCE TERMS AND CONSEQUENCE ASSESSMENT.

PROJECT 2 : THE STUDY WILL STUDY THE FEASIBILITY OF UPDATING SOME INPUT PARAMETERS TO THE DOSE ASSESSMENT MODEL BY COMPARISON OF ENVIRONMENTAL SURVEY RESULTS TO THE CORRESPONDING VALUES PREDICTED BY THE MODEL.

DURING THE EARLY PHASE OF EMERGENCY RESPONSE SUBSEQUENT OR DURING A SEVERE RELEASE, SOME IMPORTANT INPUT PARAMETERS TO THE ENVIRONMENTAL IMPACT MODEL CAN BE RATHER UNCERTAIN CONSIDERING THEIR SENSITIVITY TO THE MODEL OUTPUT. THOSE PARAMETERS CONCERN SOME SOURCE TERM PARAMETERS (E.G. RELEASE RATES, ISOTOPIC COMPOSITION, HEAT CONTENT OF THE PLUME,...) SOME METEOROLOGICAL DATA (E.G. WIND DIRECTION, WIND SPEED,... AT THE POINT OF RELEASE), AND SOME ENVIRONMENTAL TRANSFER PARAMETERS 'E.G. DRY DEPOSITION VELOCITY, WASH-OUT-RATE,...). THE PROJECT AIMS TO STUDY THE FEASIBILITY TO USE SOME ENVIRONMENTAL SURVEY RESULTS TO ADJUST THE FIRST ESTIMATE VALUE OF THOSE PARAMETERS.

THE FEASIBILITY OF THE CONTROL SYSTEM DEPENDS ON WHETHER A LIMITED NUMBER OF WELL SELECTED FIELD-OBSERVATIONS CAN SIGNIFICANTLY INCREASE THE RELIABILITY OF MODEL PREDICTIONS AVOIDING CONFUSING OSCILLATIONS OF THE SUCCESSIVELY CALCULATED DOSE.
THE CONTRACTOR WILL MORE SPECIALLY INVESTIGATE ON :

- THE REVIEW OF SURVEY PROCEDURES, FOR THE PURPOSE OF FEEDBACK TO THE CALCULATION SYSTEM.
- SELECTION OF INPUT PARAMETERS WHICH MIGHT BE ADJUSTED.
- THE OPTIMISATION TECHNIQUES.

PROJECT 3. THE MODEL,DEVELOPED BY PROJECT 1,WILL BE USED TO GENERATE A SET OF SCENARIOS,FOR THE PURPOSE OF EMERGENCY RESPONSE TRAINING.THE SCENARIO WILL COVE BROAD RANGE OF COMPLEXITY OF METEOROLOG.SOURCE TERM EPISODES.SCENARIOS ARE AIMED TO BE SELECTED BY AN EMERGENCY RESPONSE EXERCISE COORDINATOR,ACCORING PROTECTIVE ACTIONS,WHICH SHOULD BE CONSIDERED BY THE EMERGENCY RESPONSE EXECUTIVES.

Champ scientifique

Programme(s)

FP1-RADPROT 6C - Multiannual research and training programme (Euratom) in the field of radiation protection, 1985-1989

Thème(s)

Data not available

Appel à propositions

Data not available

Régime de financement

CSC - Cost-sharing contracts

Coordinateur

CENTRE D'ETUDES DE L'ENERGIE NUCLEAIRE

Contribution de l’UE

Aucune donnée

Adresse

PARC SENY RUE CHARLES LEMAIRE 1
1160 BRUSSELS
Belgique

Coût total