|
![](images/pix01.gif) |
|
|
|
Predictive Analysis of Radionuclide Source Term from Hypothetical Severe Nuclear Accident involving APR1400: Machine Learning Based Approach |
|
PP: 83-88 |
|
doi:10.18576/jrna/070311
|
|
Author(s) |
|
Soja, Reuben Joseph,
Muhammad Bello Gusau,
Usman Ismaila,
Samson Dauda Yusuf,
|
|
Abstract |
|
Nuclear accident results to significant released of discharged radionuclides to the environment, which endangers the public, and the environment. Decision-makers are always faced with challenges in such situation especially in determining the contributing variables associated with radionuclides source term released. In these situations, machine learning can be a crucial tool in helping decision-makers in determining the contribution of each variables associated with released radionuclides, which aids in early warning and helps to protect people and the environment. In this study, Source Term to Dose (STDose) model of Radiological Assessment System for Consequence Analysis computer Code (RASCAL) was used to generate radionuclides source term from Advanced Power Reactor (APR1400) by using two (2) hypothetical severe accident scenario for Long Term Station Blackout (LTSBO) and Loss of Coolant Accident (LOCA). Supervised Machine learning based approach-using Classification and Regression Tree (CART) Analysis was used to determine the most important predictors’ variable associated with radionuclides source term release for Cs-137, Cs-134, I-131, Te-132 and Xe-133 as well as their relative importance. The result of the prediction shows that containment leakage rate contribute 100% to source term released to the environment during nuclear accident with coolant duration contribute 0.1% of source term released. The containment leakage rate accounts for the wide spread and dispersion of radionuclide source terms to the environment. The LTSBO and LOCA scenarios are categorized as accident with wider consequences with significance released of radioactive materials to the environment by International Nuclear and Radiological Event Scale (INES). |
|
|
![](images/pix01.gif) |
|
|