Processin cessing g Usin ing g Discre crete e Eleme ement nt - - PowerPoint PPT Presentation

Simulation mulation of Pa Particulat ticulate e Solids ids Processin cessing g Usin ing g Discre crete e Eleme ement nt Met ethod

• d

Oleh Baran an

DEM EM overvie view DEM EM capabi bilit ities ies in STAR AR-CC CCM+ M+

– Particle types and injectors – Contact physics – Coupling to fluid flow – Coupling with passive scalar

Perform

• rmanc

nce and scalabil bilit ity Simulat ation ion assist stant ant bene nefits ts Scaled led particle icle approa

• ach

ch Sum umma mary

Outli line

2

DEM EM is applicable able to solid id flows ws

– When part or whole solid phase is in dense shear flow regime – With particles of different shape and size distribution

DEM for parti ticle cle flow with th resolv solved ed collision isions

3

DEM exam amples ples

4

Moment mentum m conser nservat ation ion

𝑛𝑗 𝑒𝑤𝑗 𝑒𝑢 = 𝐺𝑗𝑘 + 𝐺

𝑕 + 𝑘

𝐺

𝑔𝑚𝑣𝑗𝑒

𝑛𝑗 and 𝑤𝑗 are mass and velocity of particle 𝑗, 𝐺

𝑕 = 𝑛𝑗𝑕 is gravity force, 𝐺𝑗𝑘 is

contact force between particle 𝑗 and element 𝑘

• DEM is a meshless method!
• DEM is computationally intensive method!

Conse nservati tion

• n of angul

ular r mome mentum ntum 𝑒 𝑒𝑢 𝐽𝑗𝜕𝑗 = 𝑈

𝑗𝑘 𝑘

– 𝑗, 𝐽𝑗 and 𝜕𝑗 are the momentum on inertia and rotational velocity of particle 𝑗. 𝑈𝑗𝑘 = 𝑠𝑗𝑘(𝐺𝑗𝑘 + 𝐺

𝑠) is the torque produced at the point of contact and it is the

function of the rolling friction force 𝐺

𝑠

DEM Governing erning Equatio ations ns

5

Base e model

• del is non-li

linear near Herz-Mi Mindli ndlin mode del

– Details in Di Renzo, A., & Di Maio, F. P. (2004). Comparison of contact-force models for the simulation of collisions in DEM-based granular flow codes. Chemical Engineering Science, 59, 525 –541 – Other models available (next slide)

The e normal rmal and tangen gentia tial l com

• mpone

ponents, nts, 𝑮𝒐 and nd 𝑮𝒖,of f conta tact ct force ce depen pends ds on overlap, rlap, particle cle proper erti ties es

– Young’s modulus – Density – Size – Poisson ratio, – and interaction properties, for example friction, rolling friction, restitution, etc

DEM in STAR-CC CCM+ + overvie view: : Contact tact Forces es

6

Basic c models

• dels

DEM Capab abil ilit ities ies: : Contact tact models els

Hertz tz-Mi Mindlin dlin Classical nonlinear contact force model for rigid bodies Friction, restitution (normal and tangential) Walt lton

• n-Br

Brawn Linear model for deformable particles Compression and tensile stiffnesses

• Optional models (for adding to basic model)

Roll lling ing Resi sistan stance Force proportional Set of rolling friction parameters Constant Torque Displacement Damping Linear ear Cohesio esion Constant attractive force matching either JKR force or DMT model for zero overlap Work of cohesion, model blending factor Artif tificia icial Visc scosit

• sity Additional velocity dependent damping model

Linear and Quadratic coefficients Paral alle lel l Bonds ds For modelling consolidated particles Max tensile and shear stress, Bond radius Condu duction tion Heat eat tran ansf sfer er For both particle-particle and particle-geometry contacts Ranz-Marshall or user set heat transfer 7

Partic icle le Type

Spherical Composite Rigid, unbreakable Clumps Flexible, breakable

Partic icle le Initi tializ alizati ation

• n

Volumetric injection

• Random Injector: on region or Part
• Random injector
• Lattice injector:
• Part injector: on volume cells

Surface injection Surface or Part Injector: on boundary cells, or on planar grid All injectors : - ability to set particle size distributions: constant, normal, log-normal, other

• ability to specify flow rate, initial velocities, orientation, etc

DEM- Capab abil ilit ities ies: : Pa Parti ticles cles

8

DEM Coupling pling to Fluid id Flow

Drag ag Force Di Felice Schiller-Naumann Gidaspow (for 2-way coupling only) User defined field function Drag ag Tor

• rque

With either Sommerfeld Rotational Drag or user defined rotational drag coefficient Lift Force Shear Lift: Choice of Saffman, Sommerfeld, user-set coefficients Spin Lift: Choice of Sommerfeld or user-set coefficients Press ssure Grad adient ient Force Buoyancy force Two-way coupling ling Fluid is affected by particles: Momentum source is applied to continuous phase Other er inter erac actions tions Gravity force, User-Defined Body Force, Particle Radiation, Energy model, Passive Scalar 9

New w in versi sion

• n 8.06

Ar Arbitr trary y num umbe ber r of new w particle cle proper perti ties es

– Color

• Tracing subset of particles
• Analyzing mixing efficiency

– Particle residence time or displacement

• ‘dead zone’ or ‘risk zone’ analysis of granular flow

– Coating amount

• Residence time in user-defined ‘spray zone’

– Wetness / dryness of particles

• Contribution from several processes

– Amount of chemically active component

Can intera eract ct with h Eu Eulerian ian pass ssive e sc scalar

DEM Pa Passi ssive e Scalar lar

10

Pa Passi ssive e scalar lar for binning ning and mixi ixing ng

11 Source term: $${ParcelCentroid}[0] * $ParticleDensity * $TimeStep * ($Time > 0 ? 0 : 1) Source term: ($${ParcelCentroid}[0]<0) ? 0 : 1)

Challe leng nge: e: Impr mprove e inter er-pa partic ticle e coatin ting g uni niform

• rmity

ity by using g optima mal l sprayin ying g equipm ipment ent set etting ngs

– Solution: using DEM passive scalar capability

Pa Passive e scalar lar source: ce:

($$Par arcelCe elCentr troid

• id("

("Cy Cyl")[2] 2] > 0.0 && $$Par arcelCe lCentr troid id(" ("Cy Cyl") ")[2] 2] < 0.22 22 && $$ $$Par arcelCe lCentr troid id(" ("Cyl") ")[0] 0] < 0.05+1.12*$$Par arcelCe lCentr troid id(" ("Cy Cyl")[ )[2] ] ) ? ? 0.1*$Par artic ticle leDens Densit ity : : 0.0

– Coating thickness is accumulated in ‘spray zone’ – Single simulation provides solution for two different spray methods

Pa Passi ssive e scalar lar for coati ting g applic lications ations

12

Pa Passi ssive e scalar: lar: Lagrangian rangian-Eu Eulerian lerian coupling ling

Example of particles in a pile ‘releasing ‘gas

– Left Inlet air flow 100 m/s, later 10 m/s – Particles initialized with non-zero ‘Particle Gas’ value of passive scalar 𝜚1 – Eulerian passive scalar 𝜚2 has diffusion and convection on, initial value zero in all cells – Volume weighted interaction model for flow rate between passive Lagrangian and Eulerian scalars:

• 𝐾 = 𝑙 𝜍1𝜚1 − 𝜍2𝜚2 𝐵𝑞 here 𝑙 = 0.01 is

user controlled interaction coefficient, 𝜍1 and 𝜍2 are densities of Lagrangian and Eulerian phases, 𝐵𝑞 is the surface area of the particle

13

Performanc

• rmance

e and Scalability ability

Perform

• rmanc

nce-im improving ving featu tures res:

– Load Balancing – Per-continuum parallel solver – Maximum Independent Set Algorithm in injectors – Skinning

Typical al simul ulat ation ion time e of Fluidize uidized d bed d in versi sion

• n 8.02:

02: 1 s s Physica cal l time me for 28 h / 118 18 CPU for 1. 1.3 million ions s of particles cles d=2 2 mm, , ρ=2440 kg/m3, E=10 MPa

14

DEM timestep ~𝑒

𝜍 𝐹

Material density of least dense phase 𝜍 Diameter of smallest sphere 𝑒 Young’s Modulus of hardest particle 𝐹 DEM Solver Timescale Number of elements Total Number of spheres Number of CPU Number of faces in mesh Max Physical time

Skin inning ning

15

Conta ntact ct det etecti ection

• n optimized

imized

– New skin parameter in DEM solver – the larger the skin distance, the less often neighbor lists need to be re-built,

• but more pairs must be checked for

possible force interactions inside

• ne neighborhood

Simulation mulation Assistant istant Benef efits its

16

Can we use ‘larger particles’ to reduce particle count without significant change nge in accur uracy acy of the e model?

• del?

– In particular for fluidized bed application? fine particles of size 𝑒0 scaled-up particle of size 𝑒

Sug ugges gested d correcti rection

• n to Gidasp

spow drag coef efficient icient

𝐷𝑒 ⇒ 𝑒 𝑒0

𝑜

𝐷𝑒

Scale led d Pa Particle ticle Approac

• ach

17

Fluidi idized zed bed set et up

18

• J. X. Bouillard, R. W. Lyczkowski and D. Gidaspow, "Porosity Distributions in a Fluidized

Bed with an Immersed Obstacle," AIChE Journal, vol. 35, no. 6, pp. 908-921, 1989

• Size of parti

ticles les – 0.503 3 mm

Model l parame ameter ers s and studie udied d configu gurati ations

• ns

Configuration Particle size (mm) Number of Particles Values of drag scaling exponent n Time to simulate 1 second No coupling DEM+CFD 1 2 1,271,020 1.576 18 h / 106 CPU 28 h / 118 CPU 2 3 374,746 1.858 6.5 h / 46 CPU 14 h / 46 CPU 3-9 4 156,506 1.889, 1.935, 1.977, 2.016, 2.052, 2.085, 2.116 5.1 h / 34 CPU 7.6 h / 34 CPU 10-12 5 79,410 1.997, 2.077, 2.299 3.4 h / 10 CPU 7.9 h / 10 CPU Material E (MPa) Poisson ρ (kg/m3) Particles polyester 10 0.34 2440.0 Geometry aluminum 68000 0.33 2702.0 Contacts Friction Restitution Particle-Particle 0.8 0.01 Particle-Geometry 0.01 19

Prep epar aring ing initi tial al configurations igurations

1. 1. Particles ‘poured down’ 2. 2. Bul ulk oscilla illati tion

• n damped

ed 3. 3. User r Body dy force ce applied ed to remo move e particles cles above e 28 cm Same me process cess for r each h consi side dered red partic icle le size e

20

Pressure essure drop calculations ulations

Two

• way-cou

couplin pling g activat ated ed

– All three inlets set to have same inlet velocity – Superficial velocity 0.0989 m/s

Simulat ated ed at least st 0.5 5 s in stea eady dy state Pressur ssure e drop

• p recor

ecorded ded Same me process cess for r all 12 conf nfigurat gurations ions

21

Pressure essure Drop p (d=4mm) mm)

22

Pressure essure drop analysis lysis

Com

• mpar

ared ed to Er Ergun un est stima imate e for 𝑒0 = 0.503 𝑛𝑛 parti ticles cles Ex Expon

• nen

ents ts > 2.1 1 corre respon pond d to fluidized uidized state

23

𝐷𝑒 ⇒ 𝑒 𝑒0

𝑜

𝐷𝑒

Simple le power-la law w correcti rrection

• n to drag force

ce was us used d for scaled led particles icles Results sults for pressu ssure re drop

• p collapse

apsed d on si single le cur urve

– Perhaps limited for the particular choice of model parameters

• Low coefficient of restitution
• Further investigations are underway

– More accurate scaling can be set from force balance

• Sakai et. al., Study of a large scale discrete element model for fine particles in a

fluidized bed, Adv. Powder Technology 23 (2012) 673-681

Scaled led particles icles can provi vide de same me press ssure ure drop

• p for less

s computat putation ion time me

Summa mmary y of fluidi idized zed bed stud udy

24

ありがとう

25