Enhanced method for the support of experiment safety analysis

Fairhurst Agosta, Roberto Enrique

Permalink

https://hdl.handle.net/2142/122024 Copy

Description

Title

Enhanced method for the support of experiment safety analysis

Author(s)

Fairhurst Agosta, Roberto Enrique

Issue Date

2023-11-29

Director of Research (if dissertation) or Advisor (if thesis)

Kozlowski, Tomasz

Doctoral Committee Chair(s)

Kozlowski, Tomasz

Committee Member(s)

• Fischer, Paul
• Munk, Madicken
• Uddin, Rizwan

Department of Study

Nuclear, Plasma, & Rad Engr

Discipline

Nuclear, Plasma, Radiolgc Engr

Degree Granting Institution

University of Illinois at Urbana-Champaign

Degree Name

Ph.D.

Degree Level

Dissertation

Keyword(s)

• Research reactors
• Experiments
• MCNP
• SCALE
• OpenFOAM
• Machine Learning

Abstract

Nuclear research reactors enable a wide range of applications and support the development of multiple fields, including energy production, isotope production, space discovery, among others. Research reactor safety analyses demonstrate the contribution of the operational procedures to the prevention of accidents, focusing mainly on the reactor core. These analyses rely on the assumption of heat being locally deposited in the core after shutdown. This work, however, is concerned with research reactor experiments, and it investigates more detailed methods targeting the specific reactor regions hosting the experiments. Additionally, research reactor experiments require the development of individual experiment safety analyses, which due to the lack of a streamlined workflow becomes effort and time-consuming as well as error-prone. This thesis supports the development of safety analyses for research reactor experiments with the objective of streamlining the calculation procedure as well as accelerating it. The methodology presented here calculates the delayed heating in experiments, generates the volumetric heat source term for thermal-fluids calculations, solves the experiment temperature evolution during a channel draining event, and determines if the event leads to a radioactive material release. Moreover, this work examines methods to accelerate the calculation, which include the creation of a generic material irradiation database and data-based modeling. This thesis starts with a literature survey summarizing previous work relevant to delayed heating calculations, transport/depletion solvers, and machine learning applied to nuclear engineering. The conclusions of the survey guide the choice of methods for the delayed heating calculation and the choice of machine learning algorithms for the data-based modeling. The following chapters focus on the calculation workflows. This thesis establishes a delayed heating calculation workflow based on the formal three-step method that builds on top of the MCNP-ORIGEN Activation Automation (MOAA) tool. This workflow is demonstrated and verified with several exercises, which also ease the workflow visualization. The workflow is also demonstrated in two full-size research reactors, for which the heat in the structures and an experiment are calculated. A simple modification of the delayed heating calculation workflow generates a shutdown dose rate calculation workflow as a value-added. This calculation workflow helps guide the post-irradiation examination (PIE) of experiments and helps prevent the unnecessary exposure of personnel and surrounding equipment. The calculation workflow is demonstrated for a Tristructural Isotropic (TRISO)-fueled experiment which highlights the applicability of this workflow to more complex geometries, such as High-Temperature Gas-Cooled Reactors (HTGRs). When focusing on the acceleration of the calculations, this thesis proposes the generic material irradiation database method. This method relies on the delayed heating calculation workflow to obtain the heat produced in experiments of individual chemical elements and through their combination calculate the heat produced in experiments of arbitrary material composition. This thesis verifies this method by studying a vast range of materials that could be used in a research reactor experiment. Moreover, this thesis investigates the applicability of data-based modeling to predict the outcome of the most thermally challenging accident in an experiment, a channel draining event. The data preparation process and several machine learning algorithms are described before studying a simplified model. Several useful conclusions are drawn with the simplified model before scaling it up, with some minor modifications, to a full-size reactor experiment. Finally, this thesis wraps up with the conclusions from all the previous calculations. Overall, the results presented in this thesis meet this thesis’ objectives and prove that the proposed methodology decreases the computational expense and enables quick calculations supporting research reactor experiments.

Graduation Semester

2023-12

Type of Resource

Thesis

Copyright and License Information

2023 Roberto E. Fairhurst Agosta

Owning Collections