1(20)
2nd International Conference on Population Balance Modelling
May 5-7, Valencia, Spain
Validation of bubble breakage, coalescence
and mass transfer models for gas-liquid
dispersion in agitated vessel
Marko Laakkonen1, Ville Alopaeus1,2, Juhani Aittamaa1
1 HelsinkiUniversity of Technology, Laboratory of Chemical Engineering
2 Neste Engineering, Porvoo, Finland
2(20)
Motivation
Gas-liquid reactors used widely in industry but
their design is problematic:
- Mass transfer limitations
- Inhomogeneous conditions in large reactors
Local conditions can be considered by combining
mass transfer models to a Computational Fluid
Dynamic (CFD) code
3(20)
Objective
RATE = FLUX × AREA
A model for mass Population balance model
transfer fluxes for bubbles
Validation against mass Validation against measured
transfer experiments local bubble size distributions
Tool for the calculation of
gas-liquid mass transfer rates
4(20)
Air – tap water in 194 dm3 vessel
Local bubble size distributions
Overall gas holdup
Power requirements of gassed
agitation
Absorption and
desorption of dissolved
oxygen
Gas sparger
5(20)
Local bubble size distributions
Capillary suction probe
R1
R2
R3 ~2000 bubbles per distribution
R4
Operating conditions
R5 Stirring 155 – 250 rpm
Gassing 0.018 – 0.093 vvm
6(20)
Comparison of techniques
Digital Imaging (DI)
C
B
Phase Doppler
Anemometry (PDA)
14 dm3 vessel
Capillary Suction Probe (CSP)
7(20)
Comparison of techniques
Number distribution Volume distribution
min. of DI
min. of CSP
max. of PDA
n(di) [-]
bubble size [mm] v(di) [-] bubble size [mm]
PDA 5 µm – 1.4 mm PDA
Imaging 0.2 mm – Imaging
Capillary 1.2 mm – 6 mm Capillary
8(20)
Multiblock model for stirred vessel
Considers local mass
transfer conditions
Simple enough for
the fitting of unknown
model parameters
Population balance
for bubbles
Multicomponent gas-liquid
mass transfer
9(20)
Relative velocity of bubbles (slip)
Affects local bubble size distributions and gas holdup è
Mass transfer area
Solved from force balance on bubble motion
Turbulence dampening correction of Bakker & Van den
Akker (1994) with a slight modification
µ = µC + C11 ?C ε 1/ 3
a 4/3
ADJUSTABLE PARAMETER
10(20)
Population balance for bubbles
dY/dt = Transportation in/out balance region +
Birth/death by breaking +
Birth/death by coalescing +
Growth/size reduction by mass transfer
11(20)
ADJUSTABLE PARAMETER
Case A Case B
Breakage rate (Alopaeus et al. 2002) Breakage rate (Luo & Svendsen, 1996)
g (a j ) = C1e1 / 3erfc C2
s µC
+ C3 ε
1/ 3
(1 + ξ )2 exp − C6σ χ ( f )
g (a j ) = C5 ⋅ (1 − α ) 2
1 1
∫∫
2/3 5/3
?C e a j ?C ?D e1 / 3a 4j / 3 dξ df
a ξ 11 / 3 ρ ε 2 / 3a 5 / 3ξ 11/ 3
j 0 ξ min C j
Breakage kernel
2 C4 Breakage kernel
3 3 ai2 ai3 ai3
β (ai , a j ) = 9 + C4 + 9 ⋅ C4 + C4 3
33 1 − 3
2 C4
3 3 ai2 ai3 ai3
2
β (ai , a j ) = 9 + C4 + 9 ⋅ C4 + C4 3
33
2 2 a j a3j a
j
2 1 − 3
2 2 a j a3j a
j
Coalescence rate Coalescence rate
(Coulaloglou & Tavlarides, 1977) (Coulaloglou & Tavlarides, 1977)
h(ai , a j )= C7 ⋅ e 1/ 3
(a + a ) (a
2 2/3
+a ) ?(a , a )
2 / 3 1/ 2
2
(
h(ai , a j )= C7 ⋅ e1 / 3 (ai + a j ) ai2 / 3 + a 2j / 3 ) ?(a , a )
1/ 2
i j
i j i j i j
Coalescence efficiency
Coalescence efficiency (Prince & Blanch, 1990)
(Coulaloglou & Tavlarides, 1977)
rjk5 / 6 ρC1 / 2ε 1 / 3
λ (ai , a j ) = exp − C9
4
µ ρ ε ai a j
?(ai , a j ) = exp − C8 C 2C
σ 1/ 2
σ a +a
i j
12(20)
Fitting of bubble breakage and
coalescence models
70 bubble size categories needed to minimise
discretization error
Initial values of parameters by testing and from
literature
Insensitive and correlating parameters fixed on
reasonable values
All measured local BSDs from 14 and 194 dm3 vessels
included simultaneously to the fitting
13(20)
Volumetric BSDs in 194 dm3 vessel
Q=0.018 vvm Q=0.018 vvm
Q=0.041 vvm Q=0.041 vvm
v(di )
v(di )
N = 220 rpm N = 220 rpm
di di
Case A Case B
Markers (measured), Lines (simulated)
14(20)
Number BSDs in 14 dm3 vessel
n(di )
Case A
di
Case B
Digital Imaging
Stirring 340 rpm
Gassing 0.018 vvm
15(20)
Gas-liquid mass transfer fluxes
Two-film theory with simplified solution of Maxwell-
Stefan multicomponent diffusion
Validate the model by minimising the difference
between the simulated and measured concentrations
of oxygen vs. time
Mass transfer coefficients in liquid film (Kawase &
Moo-Young, 1990)
k L = C12 (ε v ) Sc −1 / 2
1/ 4
ADJUSTABLE PARAMETER
16(20)
Model vs. measured concentration of
dissolved oxygen in 194 dm3 vessel
Case A Case B
Markers (measured), Lines (validated model)
17(20)
0.28 (Prince & Blanch, 1990)
0.88 (Venneker et al. 2002)
0.02 (Bakker & Akker, 1994)
0.301 (Kawase & Moo-Young, 1990)
*not fitted at the final stage, # 95 % confidence limits
N = 250 rpm, Q = 0.093 vvm 18(20)
Bubble surface Mass transfer rates
Gas holdup [vol-%] area [m2/m3] [10-4 mol/(sm3)]
0 0 0
1.00 1.05 1.00 19 18 15 (-3) (-2) (-2)
[6] [5] [4]
0 0 0 = water
1.09 1.14 1.05 23 20 16 (-4) (-3) (-2) = nitrogen
[9] [6] [4] = oxygen
0 0 0 0 0
1.11 0.91 0.87 25 15 13 (-5) (-2) (-1) (-5) (-5)
0.85 17 [11] [4] [3] [10] [10]
0.86 17
0 0
0(-16)[33]*/
1.22 1.30 1.06*/2.14 33 35 30*/60 (-10) (-13) -0(-33)[69]
[21] [27]
8.98 187 -101
0.85 18 0 0 0 0 -1551 (-64)
30 24 109 (-6) (-4) (-3) (-4) (-22)
1.36 1.28 0.78 7.47 15
[130]
[13] [7] [5] [8] [45]
0.93 0.71 20 20 0(-3)[7] 0(-2)[4]
Simulated (Case A) local mass transfer conditions in the 194 dm3 vessel
at the time 100 s from the moment, when pure nitrogen feed was switched to air
19(20)
Conclusions
Volumetric bubble size distributions can be
measured from air – water system with a
capillary probe
Multiblock stirred tank model is a useful tool
for the fitting of model parameters against
local experiments
20(20)
Conclusions
Good agreement between the validated
models and local bubble size / mass transfer
experiments at various agitation conditions
The validated models can be included to a
CFD program and used for the design and
scaleup of agitated gas-liquid reactors
