Centre for Research & Centre for Research & Technology Hellas (CERTH)Technology Hellas (CERTH)

Institute for Solid Fuels Technology & Applications (ISFTA)

64th ΙΕΑ

– FBC meeting. June 2012 Naples

Nikolopoulos A., Nikolopoulos N. Grammelis P., Kakaras Em.

Contact: +30 210 6501 510, Fax : +30

210 6501 598, E-mails:

a.nikolopoulos@certh.gr, grammelis@certh.gr

3-D CFD SIMULATION OF A CFB CARBONATOR COLD MODEL

Presentation overview

♦♦MotivationMotivation

♦♦CFD modelingCFD modeling

EMMSEMMS

Full loop Full loop simulationsimulation

ResultsResults

♦♦ConclusionsConclusions

Presentation overview

♦♦MotivationMotivation

♦♦CFD modelingCFD modeling

EMMSEMMS

Full loopFull loop

ResultsResults

♦♦ConclusionsConclusions

Calcium looping is an attractive post combustion CO2

capture technology especially for retrofitting power plants.

Motivation

CO2

650oC

900oC

ASUair

fuel

flue gasPower plantfuel

“clean” flue gas

O2

N2

carbon

ator

calciner

dryer

purge

F0

FR

FCO2

CaCO3

2 3CaO CO CaCO eat

3 2CaCO eat CaO CO

Carbonation

Calcination

The effectiveness of this process mainly relays on the design and operating parameters of the two interconnected Circulating Fluidized bed reactors (Carbonator -

Calciner). Especially,

carbonator is a novel reactor and there is no data available for

large scale. Proper comprehensive CFD modeling is important for design optimization.

Presentation overview

♦♦MotivationMotivation

♦♦CFD modelingCFD modeling

EMMSEMMS

Full loop simulationFull loop simulation

ResultsResults

♦♦ConclusionsConclusions

•

Isothermal CFD modeling of plexi-glass cold model (Carbonator) of USTUTT

•

Operating conditions received from USTUTT (PSD, TSI ≈

1.5 Kg, Superficial gas velocity 2.89 m/s).

•

3-D transient full full ––

loop loop (CFB loop) TFM CFD simulation of

the plexi –

glass CFB cold model.

•

Dense Grid applied (~287,000 cells287,000 cells, equivalent cell length to particle diameter ratio : ~ 21~ 21)

•

The EMMS scheme was applied for the hydrodynamic simulation of USTUTT’s CFB cold model isothermal flow.

•

For the returning system, a new stress model new stress model for the inter – particle friction forces was developed.

CFD modeling

Presentation overview

♦♦MotivationMotivation

♦♦CFD modelingCFD modeling

EMMSEMMS

Full loop simulationFull loop simulation

ResultsResults

♦♦ConclusionsConclusions

--

CERTH/ISFTA developed an advanced EMMS model for the CERTH/ISFTA developed an advanced EMMS model for the operating conditions of the plexi operating conditions of the plexi ––

glass carbonator cold model of glass carbonator cold model of

USTUTTUSTUTT

--

EMMS formulationCFD modeling

uf

upf

upfupf

upc

upc

upcuf

uc

usi

usi

g

Slip vel. definition

usf = uf ‐ upfusc = uc ‐ upc

upc

ucComputational domain

-

EMMS schemes. Dense

(clusters) and dilute

phase (dispersed particles) in each Control Volume (C.V.).

EMMS formulation

Particle Particle --

gas properties for USTUTT gas properties for USTUTT cold model testscold model tests

Particle Diameter 142 μm Particle Density 5700 kg/m3

Gas Density 1.225 kg/m3

Gas Viscosity 1.7894 10-5

kg/(ms)εmf 0.55

EMMS scheme is formulated incorporating:Mass and momentum conservation equations for (Dense

and dilute, C.V.)

Semi –

empirical equations (Clusters diameter and bulk density) Constraints

Objective function: Minimum energy

interexchange between gas and solids

- The governing equations(EMMS model) were obtained for the operating conditions of USTUTT cold

model, and solved for all possible combinations of voidagevoidage and uuslipslip prior to their CFD implementation

and numerical run.

-

The non-linear optimization EMMS problem was solved with GAMS software and the results were

integrated in Fluent package with C++ UDF (User Defined Functions) coding

CFD modeling

10

EMMS equations

1 11

min

st f f f c c c i i fg s

N m F U m F U m F U f

imum

= 1 1 1EMMS g c c f f s gF f g a f g a

,Wen Yud

Emms

FH

F

c g n 60.027 10 32cl p p s sd d d

13 3 14 4

cdc g sc sc di si si c s g c

p cl g

f fC U U C U U f g ad d

1 13 1 14

fdf g sf sf f s g f

p

fC U U f g a

d

1 11

f g g c gdf sf sf di si si dc sc sc

p cl p

fC U U C U U C U Ud f d d

1f c g gU f U f u 1 1pf pc s gU f U f u

1g c ff f

Semi –

empirical equations

Closure equations

Mass conservationMomentum conservation

Objective function

ResultsResults

CFD modeling

Dense phase Dilute Phase Inter-phase

Effective drag coef.

4.650dc d cC C

4.650df d f fC C

4.65(1 )Di DoiC C f  

Standard drag coef. 0 0.313

24 3.6Re Red c

c c

C

0 0.313

24 3.6Re Red f

f f

C

0 0.313

24 3.6Re Red i

i i

C

Reynolds number Re g p

c scg

dU

Re g pf sf

g

dU

Re g cli si

g

dU

Slip velocity 1

c pcsc c

c

UU U

1

f pfsf f

f

UU U

11f pc

si fc

UU f U

Drag force 2

4 2p g

c dc sc sc

dF C U U

2

4 2p g

f df sf sf

dF C U U

2

4 2gcl

i di si sidF C U U

Numbers of particles or clusters

3

1

6

cc

p

fm

d

3

1 1

6

ff

p

fm

d

3

6

icl

fmd

EMMS formulation

020406080

100120140160180200

0.55 0.6 0.65 0.7 0.75 0.8 0.85 0.9 0.95 1

Hd (‐)

εg (‐)

Heterogeneity indexThe Hd

(εg

) function for uslip

=2 m/sec.

- 18

interpolation polynomials

for slip velocity: 0.25, 0.5, 0.75, 1, 1.25 1.5, 1.75, 2, 2.25, 2.5, 2.75, 3, 3.25, 3.5, 4, 5, 6, 8

| |,d slip gH f u

•

The results (Hd

index) of the optimization problem, were interpolated in order to be efficiently introduced in the Fluent

CFD

package (via UDF).FEMMS

= FWen,Yu / Hd

CFD modeling

Presentation overview

♦♦MotivationMotivation

♦♦CFD modelingCFD modeling

EMMSEMMS

Full loop simulationFull loop simulation

ResultsResults

♦♦ConclusionsConclusions

13

Inter-particle frictional forces

“Dilute”

flow (εs

< 0.5)

“Dilute”

flow Kinetic theory (Gidaspow)

“Dense”

flow (εs

> 0.5)

“Dense”

flow

Plastic theory

(Drucker -

Prager)

In the CFBs recirculation system the flow is dense and inter –

particle

friction forces prevail.

CFB flow is simulated with the Euler –

Euler (TFM) approach

Solids are considered as a “Pseudo”-

fluid

Full loop simulation

14

Yield criterion(Tresca, von Mises, Drucker-Prager, Gray –Stiles)

Flow ruleassociated flow rule Drucker-Prager

Plastic theory

Model development / State-of-the-art

The rate of energy loss during plastic deformation is zero

(W = Di σi = 0)

Εxtended von Mises, Drucker-Prager

Only dilatancy and NOT consolidation in a control volume is properly modeled

22sin 0dTY II

Full loop simulation

15

InterInter--particle frictional forcesparticle frictional forces

Yield criterion:

Pitman - Schaeffer - Gray – Stiles Critical point

The rate of energy loss rate of energy loss during plastic deformation during plastic deformation

is not zerois not zero(W = Di σi ≥

0)

Both dilatancy and consolidation in a control volume are properly modeled

Disadvantage: Numerical stiffness

sP u

Model development / new model

16

Constitutive equationsConstitutive equations

Conventional model

New model

TFM equations Stress ModelStress Model

InterInter--particle frictional forcesparticle frictional forces

210 4[1 (1 )]96 (1 ) 5

s s so s ss

s ss o

dg e

e g

s o ss sss ss

s o ss ss

g

ss o s s ss ss

5 2 1 3 124 1 3

451

6 1 3 1

g e edg e e

g d e e

s

s s o ss ss

4 15

king e d

sin

2fr

s dD

PII

2

2 2s

sin4sin ( )

s

dD s

Pu

2 24sin ( )s

s dD s

P

II u

kin

col ss s s o ss

4 (1 )5

d g e

fr

fr 0

μsshear

frs s kin col

frs s kin col fr

kin col fr

μsbulk

frs s kin

frs s fr

 

frs s kin

frs s fr

sP fr

s s kinfr

s s fr

P

P

frs s kin

frs s kin fr

P

P P

og11

3s

maxs

1

max2.5

smaxs

1s

17

InterInter--particle frictional forcesparticle frictional forces

Model validation through 2D CFD Model validation through 2D CFD simulation of a repose angle simulation of a repose angle ((φφexpexp

=36.03=36.03oo) measurement experiment ) measurement experiment (Geldart B)(Geldart B)

The new model through appropriate The new model through appropriate UDFs was implemented in Fluent 13 UDFs was implemented in Fluent 13

Transient simulation: Transient simulation: ΔΔT = 40T = 40μμss

φφexpexp

=3=366..0303οο

Batch of particles at T=0sBatch of particles at T=0s

Free fallFree fall

Two CFD models were applied: Two CFD models were applied: --

Conventional oneConventional one(Ext. von Mises)(Ext. von Mises)

--

New oneNew one(Pitman (Pitman --

Schaeffer Schaeffer --

Gray Gray ––

Stiles)Stiles)

Stress Model ValidationStress Model Validation

18

InterInter--particle frictional forcesparticle frictional forcesC

onve

ntio

nal m

odel

Con

vent

iona

l mod

elN

ew m

odel

New

mod

el 1.2 sec1.2 sec 2.2 sec2.2 sec1.8 sec1.8 sec

φφexpexp

=3=366..0303οοStress Model ValidationStress Model Validation

 

 

1.1 sec1.1 sec 1.4 sec1.4 sec 1.82 sec1.82 sec

t 1.20 sec

φφ==2121οο

φφ<4<4οο

Presentation overview

♦♦MotivationMotivation

♦♦CFD modelingCFD modeling

EMMSEMMS

Full loop simulationFull loop simulation

ResultsResults

♦♦ConclusionsConclusions

20

New stress model developed for the inter particle New stress model developed for the inter particle friction forces in the recirculation systemfriction forces in the recirculation system

CFB riser Loop - Seal

TFM equations Stress ModelStress Model

CFD simulation

continuity equations

momentum equations

Viscous Stress tensors

granular temperature

 

0

0

g g g g g

s s s s s

ut

ut

 

g g g g g g g

g g g g s g

s s s s s s s

s s s s s g s

u u ut

p g u u

u u ut

p p g u u

 

23

23

T

g g g g g g g g g

T

s s s s s s s s s

u u u I

u u u I

: 3 0ss s s sp I u

kin  210 4[1 (1 )]

96 (1 ) 5s s s

o s sss ss o

dg e

e g

 

 

s o ss sss ss

s o ss ss

g

ss o s s ss ss

5 2 1 3 124 1 3

451

6 1 3 1

kin

g e edg e e

g d e e

 

col   4 (1 )5

ss s s o ssd g e

 

4 (1 ) /5

ss s o ss kin s sg e d

 

fr  sin

2fr

s dD

P

II

2

22

sin

4sin

fr

s dD s

P

II u

 

fr   0   224sin

fr

s dD s

P

II u

 

og  11

3

max1 s

s

 max2.5

max1s

s

s

 

 

- The CFD model developed incorporates the EMMS scheme and the new stress model for the recirculation system.

21

-

The EMMS scheme developed increases the accuracy of the model especially in the

dense bottom region which is hard to model, and in which the majority of CO2

capture takes place.

CFD modeling of plexiCFD modeling of plexi--glass cold model (glass cold model (CarbonatorCarbonator) of USTUTT) of USTUTT

CFD simulation

22

The model incorporating the EMMS scheme efficiently captures the hydrodynamics of the CFB carbonator with high accuracy.

Regarding Pressure profile the mean error is less than 10%.

The error in the re-circulation flux is less than 2% depicting the sophistication of the developed models for the inter-particle friction forces.

CFD simulation

23

CFD simulationContours of time averaged volume fraction of solidsContours of time averaged volume fraction of solids

Riser exit -

cyclone

Bottom zone

Loop Seal

24

CFD simulationVectors of time averaged solids velocityVectors of time averaged solids velocity

Bottom zoneLoop Seal

Presentation overview

♦♦MotivationMotivation

♦♦CFD modelingCFD modeling

EMMSEMMS

Full loop simulationFull loop simulation

ResultsResults

♦♦ConclusionsConclusions

•

The developed EMMS scheme along with the implementation of a dense grid resulted in highly accurate results with respect to the governing hydrodynamics of the CFB cold model carbonator.

•

In full loop simulations of CFBs the inter –

particle friction forces should be accurately simulated. The stress tensor formulation based on the

von-

Mises

yield criterion severely under –

predicts the friction forces.

•

The developed stress tensor formulation based on Pitman - Schaeffer - Gray – Stiles Yield criterion efficiently captures the hydrodynamic behavior of the Loop –

Seal.

Conclusions

27

CERTH/ISFTA

Thank you for your attention!

Questions?

Acknowledgements: The present work was funded by the Research Programme of the Research Fund for Coal and Steel Coal RTD (Research Project CaL –

Mod

/ RFCS-CT-2010-0013