摘要详情

ID / 提交时间

80 / 2024-07-25 01:10:49

标题

Development of Artificial Neural Network Based Neutron Diffusion Equation Solver

关键字

Neutron Diffusion Equation, Nodal Expansion Method, Artificial Neural Network

主题及专题

TRACK 5: NUCLEAR ENGINEERING

状态

摘要录用

作者

HASEEB UR REHMAN / Pakistan Institute of Engineering and Applied Sciences

Usman Khalid / PIEAS

Farhana Kausar / PAEC

Masroor Ahmad / PIEAS

Ali Waqas / PIEAS

Aman ur Rehman / PIEAS

摘要

This study presents Artificial Neural Network (ANN) based model to solve the 3-dimensional, multi-group Neutron Diffusion Equation (NDE). In this approach rigorous mathematical equations of Nodal Expansion Method (NEM) are used for developing the training dataset for the ANN model, which approximates the neutron flux through a 4th-order polynomial expansion. This approach differs from traditional methods, as ANN significantly reduces computational costs. It has been tested against 3D steady-state IAEA and 3D steady-state TWIGL benchmarks and the results show reasonable agreement with the reported results. However, the applicability of trained ANN models was also observed when the problem is significantly different from the trained data problems and hence limits its wider application for core calculations.