A Simple Method For High Precision Evaluation Of Valve Flow Coefficient By Computational Fluid Dynamics Simulation

Author : Annette Volk
Publisher :
ISBN 13 :
Total Pages : 155 pages
Book Rating : 4.:/5 (955 download)

DOWNLOAD NOW!

Download or read book Quantification of Numerical and Modeling Errors in Simulation of Fluid Flow Through a Fixed Particle Bed written by Annette Volk and published by . This book was released on 2015 with total page 155 pages. Available in PDF, EPUB and Kindle. Book excerpt: Detailed description of flow through stationary particle beds is crucial for the design and implementation of municipal water filtration, material extraction systems for nuclear waste and industrial water purification systems. Knowledge of fluid-particle interactions and fluid flow properties through the bed is essential to design, but difficult to determine from experimental investigations. Combined granular-fluid simulation methods such as coupled Computational Fluid Dynamics and Discrete Element Method (CFD-DEM) have been used to bridge this gap in fundamental knowledge. Able to capture details of the small-scale and large-scale interactions that are difficult to study in physical beds, simulation findings have added great understanding to this field. Unfortunately, the reported results are occasionally flawed by a lack of understanding, specifically regarding the magnitude of numerical and modeling errors. Uniform reporting of error, investigations of simulation trend, and proof of mesh-independence have not been performed for granular-fluid simulations. A standard method of open-source granular-fluid flow simulation known as CFDEM is applied to the simulation of flow through a fixed bed. The Ergun equation is a validated empirical expression used to predict the drag force in fixed bed flow and this prediction is compared directly to simulation results. A grid-refinement procedure, standard for publication of CFD simulation results, is applied to the CFD-DEM simulations. The solution trend over the refinement range is investigated using the frequency of convergence, convergence types, and the proposed `offset' method; a comparison of the expected numerical error and actual extrapolated solution error. An optimal grid size resulting in the least amount of error is investigated by solution trend, drag profile comparison, and the grid-refinement study results. Error is seen to increase in the simulations at both large cell sizes and as the cell size approaches one particle diameter. A new grid-refinement study application that does not require analytical solution data is shown to be a good predictor of relative error in the grid solutions. Three drag correlations are applied to model fluid flow through a fixed bed. The Gidaspow drag correlation is an exact representation of the Ergun equation, and shows high accuracy. The Di Felice correlations is a continuous-function representation of expected drag force. The Koch-Hill correlation was developed from LB simulations for fluidization conditions, and is chosen as an example of a poor correlation choice. The grid-refinement study results are able to distinguish the poor performance of the Koch-Hill correlation from the highly accurate Gidaspow and Di-Felice correlations. The standard grid-refinement study is shown to be applicable to granular-fluid flows, and to produce results that are useful for common modeling choices. Relatively low convergence frequency of the grid-refinement studies is expected to hinder future application by requiring additional grid solutions. This procedure is recommended for all granular-flow simulations since it provides useful information which can prevent common modeling errors that have hampered fluidization research.