Computational modelling and uncertainty quantification (UQ) of turbulent flow and conjugate heat transfer within the t-junctions of nuclear power plants (NPPS)
File(s)
Author(s)
Lampunio, Lisa
Type
Thesis
Abstract
In T-junction pipes within nuclear power plants (NPPs) thermo-hydraulic circuits, turbulent mixing of hot and cold fluids can lead to significant temperature fluctuations. These temperature fluctuations can result in high-cycle thermo-mechanical stresses on the pipe walls, potentially leading to fatigue cracks initiation, and growth. Therefore, reliable predictions of the temperature fluctuations, and associated thermal loads, are an important part of plant safety analysis, assessment and risk management.
The aim of this PhD research is to develop high-fidelity modelling and simulations methods to enable reliable predictions of turbulent thermal mixing within T-junctions, and a mathematically rigorous approach to quantify the underlying uncertainties associated with the numerical models and simulations. This uncertainty quantification (UQ) approach utilises a data-driven Gaussian progress (GP) surrogate modelling methodology to estimate the predictive capabilities of the SST-IDDES turbulence model and propagate parametric uncertainties associated with the inlet velocity profile boundary conditions (BCs). These GP surrogate and UQ models can be combined with continuous health monitoring diagnostic sensor data, from nuclear systems, to produce data-driven digital-twin models of NPPs.
Power spectral density (PSD) and spectral proper orthogonal decomposition analysis (SPOD) are performed to assess the influence of different BCs on turbulent thermal mixing. The numerical results indicate that thermal loads due to thermal stratification, cycling and stripping are significantly higher for uniformly distributed BCs at the branch pipe of the T-junction compared to fully developed BCs. The PSD of the temperature signal shows peaks in the range of concern of high-cycle thermal fatigue, ≈5Hz, for uniformly distributed BCs at the branch pipe. The first SPOD mode of the velocity signal reveals that vortical structures, associated with this frequency, come from the shear-layer roll-ups and Kelvin-Helmholtz instabilities, originated at the upstream corner of the junction. These results are not observed for fully developed BCs at the branch pipe.
The aim of this PhD research is to develop high-fidelity modelling and simulations methods to enable reliable predictions of turbulent thermal mixing within T-junctions, and a mathematically rigorous approach to quantify the underlying uncertainties associated with the numerical models and simulations. This uncertainty quantification (UQ) approach utilises a data-driven Gaussian progress (GP) surrogate modelling methodology to estimate the predictive capabilities of the SST-IDDES turbulence model and propagate parametric uncertainties associated with the inlet velocity profile boundary conditions (BCs). These GP surrogate and UQ models can be combined with continuous health monitoring diagnostic sensor data, from nuclear systems, to produce data-driven digital-twin models of NPPs.
Power spectral density (PSD) and spectral proper orthogonal decomposition analysis (SPOD) are performed to assess the influence of different BCs on turbulent thermal mixing. The numerical results indicate that thermal loads due to thermal stratification, cycling and stripping are significantly higher for uniformly distributed BCs at the branch pipe of the T-junction compared to fully developed BCs. The PSD of the temperature signal shows peaks in the range of concern of high-cycle thermal fatigue, ≈5Hz, for uniformly distributed BCs at the branch pipe. The first SPOD mode of the velocity signal reveals that vortical structures, associated with this frequency, come from the shear-layer roll-ups and Kelvin-Helmholtz instabilities, originated at the upstream corner of the junction. These results are not observed for fully developed BCs at the branch pipe.
Version
Open Access
Date Issued
2023-03-15
Date Awarded
01/02/2024
License URL
Advisor
Eaton, Matthew
Duan, Yu
Grant Number
EP/L015900/1
Publisher Department
Mechanical Engineering
Publisher Institution
Imperial College London
Qualification Level
Doctoral
Qualification Name
Doctor of Philosophy (PhD)