BEGIN:VCALENDAR
VERSION:2.0
PRODID:IEEE vTools.Events//EN
CALSCALE:GREGORIAN
BEGIN:VTIMEZONE
TZID:America/New_York
BEGIN:DAYLIGHT
DTSTART:20260308T030000
TZOFFSETFROM:-0500
TZOFFSETTO:-0400
RRULE:FREQ=YEARLY;BYDAY=2SU;BYMONTH=3
TZNAME:EDT
END:DAYLIGHT
BEGIN:STANDARD
DTSTART:20261101T010000
TZOFFSETFROM:-0400
TZOFFSETTO:-0500
RRULE:FREQ=YEARLY;BYDAY=1SU;BYMONTH=11
TZNAME:EST
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20260504T234820Z
UID:F9820337-BE9C-4246-902A-FDF246245C1B
DTSTART;TZID=America/New_York:20260501T130000
DTEND;TZID=America/New_York:20260501T140000
DESCRIPTION:This presentation summarizes an integrated workflow to quantify
  and model how real char particle morphology influences gas–solid multip
 hase transport\, with a focus on improving the fidelity of reactor- scale 
 simulations for biomass–coal conversion systems. Char particles produced
  from bituminous coal and biomass (pine sawdust) are imaged using high-res
 olution micro-CT\, and their 3D geometries are reconstructed in ScanIP for
  particle-scale analysis. Using these realistic morphologies\, particle-sc
 ale CFD simulations resolve coupled conservation equations (mass\, momentu
 m\, species\, and energy) under combustion-relevant boundary conditions\, 
 enabling direct evaluation of morphology-driven flow and thermal fields an
 d their impact on aerodynamic drag. The study further assesses the accurac
 y of classical drag correlations developed for idealized shapes by compari
 ng them against 3D simulation results across multiple particle orientation
 s and Reynolds numbers (Re = 20–200)\, highlighting substantial model di
 screpancies when irregularity and orientation effects are present. Unlike 
 the conventional models\, to consider the full picture of the particle mor
 phology and reduce the computational burden of high-fidelity CFD\, a deep-
 learning algorithm has been developed in which a CNN ingests voxelized 3D 
 particle images along with Reynolds number/orientation encoding to predict
  drag coefficients\, achieving high accuracy and enabling rapid parameter 
 sweeps that are impractical with CFD alone.\n\nSpeaker(s): Dr. Dongyu Lian
 g\, \n\nRoom: E101\, Bldg: Engineering Building\, southfield\, Michigan\, 
 United States\, 48075
LOCATION:Room: E101\, Bldg: Engineering Building\, southfield\, Michigan\, 
 United States\, 48075
ORGANIZER:mguduri@ltu.edu
SEQUENCE:14
SUMMARY:Understanding the Effects of Particle Morphology in Multiphase Flow
  Using Micro-CT\, CFD\, and Deep Learning
URL;VALUE=URI:https://events.vtools.ieee.org/m/557494
X-ALT-DESC:Description: <br /><p class="p1">This presentation summarizes an
  integrated workflow to quantify and model how real char particle&nbsp\;mo
 rphology influences gas&ndash\;solid multiphase transport\, with a focus o
 n improving the fidelity of reactor-&nbsp\;scale simulations for biomass&n
 dash\;coal conversion systems. Char particles produced from bituminous coa
 l and&nbsp\;biomass (pine sawdust) are imaged using high-resolution micro-
 CT\, and their 3D geometries are&nbsp\;reconstructed in ScanIP for particl
 e-scale analysis. Using these realistic morphologies\, particle-scale CFD&
 nbsp\;simulations resolve coupled conservation equations (mass\, momentum\
 , species\, and energy) under&nbsp\;combustion-relevant boundary condition
 s\, enabling direct evaluation of morphology-driven flow and&nbsp\;thermal
  fields and their impact on aerodynamic drag. The study further assesses t
 he accuracy of classical&nbsp\;drag correlations developed for idealized s
 hapes by comparing them against 3D simulation results across&nbsp\;multipl
 e particle orientations and Reynolds numbers (Re = 20&ndash\;200)\, highli
 ghting substantial model&nbsp\;discrepancies when irregularity and orienta
 tion effects are present. Unlike the conventional models\, to&nbsp\;consid
 er the full picture of the particle morphology and reduce the computationa
 l burden of high-fidelity&nbsp\;CFD\, a deep-learning algorithm has been d
 eveloped in which a CNN ingests voxelized 3D particle images&nbsp\;along w
 ith Reynolds number/orientation encoding to predict drag coefficients\, ac
 hieving high accuracy and&nbsp\;enabling rapid parameter sweeps that are i
 mpractical with CFD alone.</p>
END:VEVENT
END:VCALENDAR