Modelling the Dynamics of Short-Term Exposure to Radiation

Principal investigator:

Prof. Dr. Juš Kocijan

Duration:

September 1, 2020 – August 31, 2023

Funding:

ARRS - Slovenian Research Agency, L2 - 2615 (B), co-financing from Krško Nuclear Power Plant

Abstract:

Nuclear power plants are an established method of generating electric power throughout the world. In the Middle and Far East, some 50 plants are being built for the needs of developing countries. We cannot influence the decision to build them, but we are worried about the potential consequences of nuclear accidents. The accidents in Chernobyl and Fukushima serve as reminders. In developed countries nuclear power plants are introducing the PCFVS (Passive Containment Filtered Venting Systems) safety filters, which would contain most radionuclides even during a major accident with atmospheric releases. The nuclear power plant in Krško (NEK) introduced these filters after the Fukushima accident.

If a release nevertheless occurs during an accident, the key information for taking proper action is the forecast of the spatial distribution of radiation doses received by the population as a radioactive cloud passes through. The received radiation dose is the measure of the damage caused to a person by radiation and depends on the radiation concentration, the exposure time, and on the initial emission of radionuclides. The received radiation dose denotes the weather impact on atmospheric dilution as a radioactive cloud passes through a particular area. We want to calculate the dose for the vicinity of the nuclear power plant where immediate action must be taken to protect the population. Due to the plants' extremely long service life, i.e. over 50 years in some cases, the assessment of the spatial distribution of received doses varies, because the spread of the radionuclide cloud and the doses received per inhabitant are affected by the meteorology of the atmosphere, which is subjected to long-term climate change.

How to effectively model and assess the long-term dynamic behaviour of the distribution of doses received, which greatly influences the magnitude of a potential nuclear accident, is the subject of preliminary research worldwide. Under the proposed project we will apply the systems theory methods to solving this problem. The answer to the posed question will be very useful for the non-nuclear industry with constant or risky emissions into the atmosphere which are regulated by the European directives IPPC, EWD and SEVESO.

We propose the development of a method that examines the dynamics of the behaviour of the doses received in the following way:

The spread of radiation will be treated as a dynamic nonlinear system, which is usually modelled with a complex system of partial differential equations.
This complex model will be simplified with the identification method. Through Gaussian-process modelling, we will identify a model that will illustrate the spatiotemporal short-term and long-term dynamics of the doses received due to the weather dynamics, taking into consideration the impact of climate change.
For modelling purposes, we will prepare a multi-decade system database. This large amount of data or big data will illustrate the diversity of potential radiation doses received by the population. To ensure transferability to similar problems, we will use the concept of the relative dose received to ensure independence of the specific amount of radionuclide emission.
We will prepare identification data using the method of fusing the dispersion data, the calculation of received radiation doses and the weather data.

We estimate that the problem can be solved with the systems theory methods and with the modelling of dynamic systems. The project consortium combines two research groups with the required knowledge of experimental modelling of dynamic systems and of modelling radiological-pollution dispersion

Worldwide legislation requires that all nuclear power plants periodically assess the variability of radiation doses received per inhabitant. The proposed method will greatly improve the precision and credibility of the long-term assessment and is of great importance to science, operational safety and general application.

Project workpackages:

Fusion of data from simulations of theoretical models and other data in an experimentation testbed. (Present level of realisation: 100 %)
Replacing complex models with an identified Gaussian-process models. (Present level of realisation: 100 %)
Experimentation in the Krško Testbed. (Present level of realisation: 100 %)

Project partners:

Selected publications:

PERNE, Matija, MLAKAR, Primož, GRAŠIČ, Boštjan, BOŽNAR, Marija, KOCIJAN, Juš. Fast numerical wind turbine candidate site evaluation. Applied sciences, ISSN 2076-3417, 2021, vol. 11, no. 7, str. 2953-1- 2953-18, doi: 10.3390/app11072953. [COBISS.SI-ID 57290499].

PERNE, Matija, BOŽNAR, Marija, GRAŠIČ, Boštjan, MLAKAR, Primož, KOCIJAN, Juš. Improving wind vector predictions for modelling of atmospheric dispersion during Seveso-type accidents. Atmospheric pollution research journal, ISSN 1309-1042, 2021, vol. 12, no 2, str. 76-83, doi: 10.1016/j.apr.2020.10.010. [COBISS.SI-ID 35712515].

KRIVEC, Tadej, KOCIJAN, Juš, PERNE, Matija, GRAŠIČ, Boštjan, BOŽNAR, Marija, MLAKAR, Primož. Data-driven method for the improving forecasts of local weather dynamics. Engineering applications of artificial intelligence, ISSN 0952-1976. [Print ed.], 2021, vol. 105, str. 104423-1-104423-14, doi: 10.1016/j.engappai.2021.104423. [COBISS.SI-ID 74917635].

BOŽNAR, Marija, MLAKAR, Primož, GRAŠIČ, Boštjan, GRŠIĆ, Zoran, HETTRICH, Sebastian, MANCINI, Francesco, PATRYL, Luc, THIESSEN, Kathleen M. Modelling air pollution around nuclear power plants : validation of dispersion models using tracer data. Journal of radiological protection, ISSN 0952-4746, 2022, vol. 42, no. 2, str. 020519-1-020519-13, doi: 10.1088/1361-6498/ac7a6f. [COBISS.SI-ID 114189571].

Department of Systems and Control

Jožef Stefan Institute