Validation of a thermal non-equilibrium Eulerian-Eulerian multiphase model of a 620 MWe pulverized fuel power boiler

Abstract

The use of a thermal non-equilibrium Eulerian-Eulerian model for the simulation of a 620 MWe power boiler is proposed for capturing the combustion and radiative heat transfer found in the pulverized fuel systems. The models eliminates the use of a Lagrangian reference frame in tracking solid fuel particles thereby reducing the computational expense and time. The model solves the scalar transport for the particle mass, energy and radiation interactions between the pseudo-particle and continuous phases. The goal is to apply the modelling approach to generate a simulation database for different load cases and firing conditions which in turn will be used to study flexible operation. The model is validated against both numerical and applicable site data measurements. It is shown that the model is able to adequately resolve the furnace and superheater wall heat fluxes. Additionally the resolution of the flow field, combustion dynamics and wall fluxes are demonstrated for both an 80% and 60% operational loads. Moreover, it is shown that the Eulerian-Eulerian model results in approximately a 30% computational resource reduction when compared to traditional modelling approaches.

An integrated data-driven surrogate model and thermofluid network-based model of a 620 MWe utility-scale boiler

“…Validation of this studies 620 [ MW e ] CFD simulation model has previously been presented by researchers Rawlins et al, 9 where the developed CFD modelling methodology was validated and verified for multiple MCR load cases against both experimental and numerical results. The interested reader is directed to the works of Rawlins et al 10,20 for a more detailed description of the developed CFD multiphase modelling approach.…”

“…This approach allows for the adequate resolution of the temperature-dependent combustion processes and accurate prediction of the particulate radiative effects. The model has been validated and verified for a 2.165 [ MW th ] lab scale pulverised fuel burner by Rawlins et al 10 Furthermore, the CFD model was subsequently validated for the full-scale utility boiler of the current study operating for a wide range of loads. 9 Thus, this section provides a summary of the CFD models features and governing equations.…”

“…The current study makes use of an existing steady-state multiphase CFD model. 10 The CFD model utilises a Eulerian-Eulerian approach to capture the solid and gaseous phase interactions. The model was implemented in the commercial CFD software package ANSYS Fluent® v19.5 to resolve the fluid flow, heat transfer and combustion processes.…”

“…28 Typically, MLPs are adapted for supervised learning problems where the input variables are mapped to labelled output variables. The relationship between the input and output variables are learned by optimizing the weights (w) and biases (b) to minimize a selected cost function, which for most cases is the mean square error (MSE) given in equation (9). A cost function usually calculates the difference between the estimated (b y i ) and the desired values and is reported as a single number.…”

“…This integrated model is proposed to investigate the thermal response of the 620 [ MW e ] utility scale boiler under investigation. An existing validated steady-state multiphase CFD model 9,10 was used to generate a CFD simulation dataset for 180 different operating states. To determine the required complexity of the data-driven surrogate model, two different ANN frameworks were considered, namely a standard multi-layer perceptron (MLP) and a mixture density network (MDN), where the latter allows the introduction of uncertainty into the model predictions that stems from information density in the generated dataset.…”