Numerical Simulation of Biomass Gasi3ication in a Steam-‐Blown
Bubbling Fluidized Bed
Christos Altantzis, Addison Stark, Richard Bates, Akhilesh Bakshi, Rajesh Sridhar, Ahmed Ghoniem
Department of Mechanical Engineering Massachuse5s Ins7tute of Technology
NETL Workshop of Mul7phase Simula7ons, August 12-‐13, 2015
Project support by BP
Steam-‐blown 3luidized bed gasi3iers Fluidized bed: Favorable reactor technology for biomass gasifica7on with minimum pretreatment Advantages: • High levels of intermixing • Suitable for coarse par7cles with large residence 7mes Disadvantages: • Lower levels of carbon conversion with considerable tar content
Steam as a gasifica2on agent: Advantages: • Reduced cost, no air separa7on is needed • In the absence of oxida7on, hot zones are avoided in the bed Disadvantages: • Biomass devola7liza7on and char gasifica7on are endothermic • External hea7ng is needed for controlling the process
Key phenomena in a 3luidized bed gasi3ier Mul7-‐scale process: • Gas-‐phase chemistry • Surface chemistry • Single-‐par7cle modeling • Hydrodynamics
• Development of a mul7scale CFD methodology for the reac7ve mul7phase simula7ons to assist the design and op7miza7on of gasifica7on processes by reducing the cost, compared with experiments, and offering informa7on for integra7on in ROM
Goal
Modeling Challenges: Initial char loading • Challenge: Steady state char inventory takes hours to reach for
gasifica7on condi7ons – Ini7al transient too long for CFD simula7on
• Solu2on: Standalone MATLAB steady state char conversion model computes char inventory for CFD ini7al condi7on – Char gasifica7on and combus7on – Gasifica7on assisted a5ri7on due to hardness reduc7on
Steady state char inventory for steam blown gasi3ier
dreactor=10.2cmmchar=138g
Vchar =2029cm3
T=700 ○C
Hchar=24.8cm
Transient model:System of ordinary
differential equations
Steady state model: Iterative solver for
average residence time
Char conversion model
Outputs:
Inputs:
Average char inventory (kg)Average char resident time (sec)Average char conversion (-‐)
Gas composition (XCO2,XH2O…)Reactor temperature, pressure,Initial biomass particle size
��#$,&'( 𝑘𝑔 𝑠𝑒𝑐⁄ steady biomass feed rate𝑌12'&[𝑘𝑔 𝑘𝑔⁄ ] char devolatilization yield
Computational cost<1 sec
CFD ini2al condi2ons : -‐char inventory -‐par7cle size
Modeling Challenges: Chemistry description
• Challenge: Number of species/reac7ons prohibi7vely large for use in CFD • Solu7on: Use of Global models for pyrolysis and tar cracking [1]
• Devola7liza7on dynamics are strongly influenced by par7cle radius.
• Shrinking Core Model Implemented for Eulerian Modeling Framework:
• Trade-‐off of devola7liza7on 7me and mixing 7me important for well-‐s7rred assump7on.
Global devolatilization model
keff =1
1kkin
+ akcond
• Raw biomass injected as par7cles and removed as char • Biomass typically <2-‐3% of bed mass • Bed temperature 600-‐1000°C • Very rapid heat up • Mixing and par2cle residence 2mes are very important to product
composi7on and conversion
t2 t3 t4t1
Vol
atile
rele
ase
time
t4
t1 t3t2
t=t0
(b.1) Dapy,V>> 1
FUEL FUEL
t1
t=t0
t2
t3
t t4
(c.1) Dapy,V~ 1
FUEL
FUELPARTICLE
D
L
t1
t=t0
t
(a.1) Dapy,V<< 1
FUEL t=t0
OXIDANT
FUEL
CHARCOMBUSTION
CHARGASIFICATION
BIOMASSPYROLYSIS
OXIDANT
FUEL
BIOMASSPYROLYSIS
CHARGASIFICATION
OXIDATIONOF CHAR AND
VOLATILES
TA
R C
RA
CK
ING
OXIDANT
FUELPYROLYSIS
CHARGASIFICATION
+OXIDATION
OF CHAR AND VOLATILES
+TAR
CRACKING
+
(b.2) Dapy,V>> 1(a.2) Dapy,V<< c.2) Dapy,V~ 1
Xb
Xb,avg
z
XbXb
(b.3) Dapy,V>> 1(a.3) Dapy,V<< c.3) Dapy,V~ 1
t2 t3 t4t1 t2 t3 t4t1
Vol
atile
rele
ase
time
t4
t1 t3t2
t=t0
(b.1) Dapy,V>> 1
FUEL
t4
t1 t3t2
t=t0
(b.1) Dapy,V>> 1
FUEL FUEL
t1
t=t0
t2
t3
t t4
(c.1) Dapy,V~ 1
FUEL
t1
t=t0
t2
t3
t t4
(c.1) Dapy,V~ 1
FUEL
FUELPARTICLE
D
L
t1
t=t0
t
(a.1) Dapy,V<< 1
FUEL t=t0
FUELFUEL
FUELPARTICLE
D
L
t1
t=t0
t
(a.1) Dapy,V<< 1
FUELFUEL t=t0
OXIDANT
FUEL
CHARCOMBUSTION
CHARCOMBUSTION
CHARGASIFICATION
BIOMASSPYROLYSIS
OXIDANT
FUEL
BIOMASSPYROLYSIS
CHARGASIFICATION
OXIDATIONOF CHAR AND
VOLATILES
TA
R C
RA
CK
ING
TA
R C
RA
CK
ING
OXIDANT
FUELPYROLYSIS
CHARGASIFICATION
+OXIDATION
OF CHAR AND VOLATILES
+TAR
CRACKING
+
(b.2) Dapy,V>> 1(a.2) Dapy,V<< c.2) Dapy,V~ 1(b.2) Dapy,V>> 1(a.2) Dapy,V<< c.2) Dapy,V~ 1
Xb
Xb,avg
z
XbXb
(b.3) Dapy,V>> 1(a.3) Dapy,V<< c.3) Dapy,V~ 1(b.3) Dapy,V>> 1(a.3) Dapy,V<<
1 (
1 (
1 (1 (
1 (1 (c.3) Dapy,V~ 1
OXIDANT OXIDANT OXIDANT
Fig. 6. (a.1)–(c.1): Possible situations according to Damkohler number Dapy,V in a bubbling FB. (a.2)–(c.2) Models of the limiting situations (a) (b) and general case (c). (a.3)–(c.3)Models of distribution of fuel in the three cases.
A. Gomez-Barea, B. Leckner / Progress in Energy and Combustion Science 36 (2010) 444–509458
Importance*of*accurate*predic$on*of*mixing*dynamics*
• Devola$liza$on* $me* scales* are* similar* with* mixing* $me* scales,* so* accurate*predic$on*of*the*posi$on*of*raw*biomass*is*important*
• Low* density* and* small* size* of* char* par$cles* will* result* in* segrega$on* during*gasifica$on.* Accumula$on* of* char* to* the* top* of* the* bed* causes* lower* heat*release*within*the*bed,*reducing*the*reac$vity**
t2 t3 t4t1
Vol
atile
rele
ase
time
t4
t1 t3t2
t=t0
(b.1) Dapy,V>> 1
FUEL FUEL
t1
t=t0
t2
t3
t t4
(c.1) Dapy,V~ 1
FUEL
FUELPARTICLE
D
L
t1
t=t0
t
(a.1) Dapy,V<< 1
FUEL t=t0
OXIDANT
FUEL
CHARCOMBUSTION
CHARGASIFICATION
BIOMASSPYROLYSIS
OXIDANT
FUEL
BIOMASSPYROLYSIS
CHARGASIFICATION
OXIDATIONOF CHAR AND
VOLATILES
TA
R C
RA
CK
ING
OXIDANT
FUELPYROLYSIS
CHARGASIFICATION
+OXIDATION
OF CHAR AND VOLATILES
+TAR
CRACKING
+
(b.2) Dapy,V>> 1(a.2) Dapy,V<< c.2) Dapy,V~ 1
Xb
Xb,avg
z
XbXb
(b.3) Dapy,V>> 1(a.3) Dapy,V<< c.3) Dapy,V~ 1
t2 t3 t4t1 t2 t3 t4t1
Vol
atile
rele
ase
time
t4
t1 t3t2
t=t0
(b.1) Dapy,V>> 1
FUEL
t4
t1 t3t2
t=t0
(b.1) Dapy,V>> 1
FUEL FUEL
t1
t=t0
t2
t3
t t4
(c.1) Dapy,V~ 1
FUEL
t1
t=t0
t2
t3
t t4
(c.1) Dapy,V~ 1
FUEL
FUELPARTICLE
D
L
t1
t=t0
t
(a.1) Dapy,V<< 1
FUEL t=t0
FUELFUEL
FUELPARTICLE
D
L
t1
t=t0
t
(a.1) Dapy,V<< 1
FUELFUEL t=t0
OXIDANT
FUEL
CHARCOMBUSTION
CHARCOMBUSTION
CHARGASIFICATION
BIOMASSPYROLYSIS
OXIDANT
FUEL
BIOMASSPYROLYSIS
CHARGASIFICATION
OXIDATIONOF CHAR AND
VOLATILES
TA
R C
RA
CK
ING
TA
R C
RA
CK
ING
OXIDANT
FUELPYROLYSIS
CHARGASIFICATION
+OXIDATION
OF CHAR AND VOLATILES
+TAR
CRACKING
+
(b.2) Dapy,V>> 1(a.2) Dapy,V<< c.2) Dapy,V~ 1(b.2) Dapy,V>> 1(a.2) Dapy,V<< c.2) Dapy,V~ 1
Xb
Xb,avg
z
XbXb
(b.3) Dapy,V>> 1(a.3) Dapy,V<< c.3) Dapy,V~ 1(b.3) Dapy,V>> 1(a.3) Dapy,V<<
1 (
1 (
1 (1 (
1 (1 (c.3) Dapy,V~ 1
OXIDANT OXIDANT OXIDANT
Fig. 6. (a.1)–(c.1): Possible situations according to Damkohler number Dapy,V in a bubbling FB. (a.2)–(c.2) Models of the limiting situations (a) (b) and general case (c). (a.3)–(c.3)Models of distribution of fuel in the three cases.
A. Gomez-Barea, B. Leckner / Progress in Energy and Combustion Science 36 (2010) 444–509458
Devola7liza7on 7me scales are similar with mixing 7me scales, so accurate predic7on of the posi7on of raw biomass is important
Modeling Challenges: Hydrodynamics
CFD modeling strategy for gas-‐par7cle flows Two Fluid Model
Gas Phase à Eulerian Framework Par7cle Phase à Eulerian Framework
Both phases are considered as fully interpenetra7ng con7nua
ADVANTAGE: High computa7onal efficiency makes this method more a5rac7ve since parametric and design studies of large-‐scale systems are feasible DISADVANTAGE: Closures are required for the modeling of:
• Interphase momentum exchange • Par7cle-‐par7cle interac7on
Numerical tool: MFIX (DOE-‐NETL) [2,3]
Experimental Parameter Value
Bed material Size, dp, Density ρp
Olivine (Mg,Fe)2SiO4
270 μm 3300 kg/m3
Minimum Fluidiza2on umf ~0.037m/s
Superficial velocity, ,u0 ~0.12 m/s
Bed diameter, dbed Bed Height
4’’(0.106 m), 0.13m
Bed temperature, Tbed (heated walls)
750, 800, 850 ○C,
Inlet gas composi2on Xfeed 100% H2O(%vol) <2% Helium
Input biomass feed rate bio,0 800g/hour
Biomass mean diameter dbio,0 1 mm
Steam flow rate 800g/hour
NREL gasi3ier
130
cm
10.6 cm
• 2D simula7ons of a steam blown gasifier (Both reac7ng and non-‐reac7ng) • 3 solid phases considered
• Biomass : Rho=600kg/m3, d=1mm • Char : Rho=170kg/m3, d=0.38mm • Sand : Rho=3300kg/m3, d=0.27mm
• Drag model: Gidaspow • Inter-‐par7cle drag model: Gera et al. 2004 [4]
• Fric7on coefficient, Cf = 0.1 • Segrega7on slope coefficient, Cs = 0.1
• Par7al slip BC for solids • Specularity coefficient, φ = 0.05 [5,6] • Dirichlet BC for temperature (1023 K) along walls and inflow • Ini7al Condi7on
• Sta7c bed height: 24.4 cm • εs,sand = 0.4582 • Εs,char = 0.1218
Simulation setup
• Resolu7on: 40X400 cells
Chemical mechanism • Drying
• Devola7liza7on (Compe7ng pathways following Gronli 2000 [7]) bio -‐-‐> 7.7872H2 + 4.7274CO + 4.3016CO2 + 1.7109CH4 + 6.9712H2O bio -‐-‐> 6.2792tar1 bio -‐-‐> 40.8361FC1
• Tar cracking (Details in following slides)
• Water gas shix (Fast kine7cs Biba 1978 [8]) CO + H2O -‐-‐> H2 + CO2 H2 + CO2 -‐-‐> CO + H2O
• Char gasifica7on (Hobbs 1992 [9]) C + CO2 -‐-‐> 2CO C + H2O -‐-‐> CO + H2
Chemistry description
Tar1 à 1.5709CO+0.197CO2 + 0.4304CH4 + 0.6704H2+ 0.22Tari
Component mass frac2on
Seebauer 1999
Light gases 0.78
Tarinert 0.22
Tar cracking reaction
• Both tar1 and tari are considered to be benzene • The global reac7on is developed for biomass pyrolysis in inert
environment (N2) at moderate temperatures
Tar1 à light gases + inert tar
Comparison with NREL experiments, 1023K Steam blown bed
Comparison with NREL experiments, 1023K Steam blown bed
The tar cracking mechanism of Seebauer 1999 over-‐predicts the produced inert tars by two orders of magnitude
• Levoglucosan iden7fied as a the major tar species present axer devola7liza7on [10] • Global tar cracking mechanism in a steam environment, based on leveglucosan cracking • Benzene assumed as the major inert tar species
Identi3ication of major tars after devolatilization
Tar cracking mechanism was modified using a reactor network model employing the Ranzi mechanism to represent the LVG decomposi7on pa5ern in the absence of oxygen Tar1(LVG) à 3.27CO + 0.2CO2 + 0.65CH4 + 1.1H2 + 2.68H2O + 0.01038Tari(Benzene)
Comparison with NREL experiments, 1023K Steam blown bed
Comparison with NREL experiments, 1023K Steam blown bed
• Tar produc7on is predicted accurately • Discrepancies for CO2, H2 and H2O
Reac7ng vs non-‐reac7ng hydrodynamics
Summary and future work
• Ongoing work towards valida7on of a reac7ng CFD methodology for biomass gasifica7on in a steam environment
• Use of tools employing detailed chemistry to extract informa7on about the global chemical mechanisms implemented in CFD
• Revisit the devola7liza7on mechanism by implemen7ng more detailed schemes
• Poten7al effect of the emulsion phase on the water gas shix reac7on
• Enhancement of reactor network models employing detailed chemistry by feeding back informa7on about gas-‐solids mixing obtained from CFD
References [1] C. Di Blassi, “Modeling chemical and physical processes of wood and biomass pyrolysis”, Progress in Energy and Combus>on Science, 2008, 34(1):47-‐90 [2] M. Syamlal, W. Rogers, T.J. O'Brien, “Mfix documenta7on theory guide”, Technical Report, U.S. Department of Energy, Na>onal Energy Technology Laboratory, 1993 [3] M. Syamlal, W. Rogers, T.J. O'Brien, “Mfix documenta7on numerical technique”, Technical Report, U.S. Department of Energy, Na>onal Energy Technology Laboratory, 1998 [4] D. Gera, M. Syamlal, T. O’Brien, “Hydrodynamics of par7cle segrega7on in fluidized beds”, Interna>onal Journal of Mul>phase Flow, 2004, 30:419-‐428 [5] C. Altantzis, R.B. Bates, A.F. Ghoniem, “3D Eulerian modeling of thin rectangular gas–solid fluidized beds: es7ma7on of the specularity coefficient and its effects on bubbling dynamics and circula7on 7mes”, Powder Technology, 2015, 270(A):256-‐270 [6] A. Bakshi, C. Altantzis, R.B. Bates, A.F. Ghoniem, “Eulerian–Eulerian simula7on of dense solid–gas cylindrical fluidized beds: Impact of wall boundary condi7on and drag model on fluidiza7on”, Powder Technology, 2015, 277:47-‐62 [7] M.G. Grønli, M.C. Melaaen, “Mathema7cal model for wood pyrolysis-‐comparison of experimental measurements with model predic7ons”, Energy Fuels, 2000, 14:791–800 [8] V. Biba, J. Macak, E. Klose, J. Malecha, “Mathema7cal model for the gasifica7on of coal under pressure”, Industrial & Engineering Chemical Process Design and Development, 1978, 17:92 [9] M.L. Hobbs, P.T. Radulovic, L.D. Smoot, “Modeling fixed-‐bed coal gasifiers”, AIChE Journal, 1992, 38(5):681–702 [10] Addison Stark, “Mul7-‐Scale Chemistry Modeling of the Thermochemical Conversion of Biomass in a Fluidized Bed Gasifier”, Massachuse5s Ins7tute of Technology, PhD Thesis, 2015