modelling of membrane based co capture processes€¦ · modelling of membrane based co 2 capture...
TRANSCRIPT
1
EU- Australian workshop, Oslo, Norway 13th- 14TH september 2017
Bouchra BELAISSAOUI Eric FAVRE
Modelling of membrane based CO2 capture processes
Structure/ thickness
Transport mechanism
Operating parameter
Local transmembrane flux
Driving force strategy
Flow pattern
DESIGN
Multistage/recycling Heat integration
Application
Available driving force Separation objectives
Cost minimization
From material to process design
2
MATERIAL MODULE
I II
III
3
Composition of emission points (Post-combustion)
Structure/ thickness
Transport mechanism
Operating parameter
Local transmembrane flux
Driving force strategy
Flow pattern
DESIGN
Multistage/recycling Heat integration
Application
Available driving force Separation objectives
Cost minimization
From material to process design
4
MATERIAL MODULE
I II
III
5
2
1
3 Diffusion Sorption
Desorption
3
2
1
Ji
Gas
Membrane
Gas Z
When permeability is constant, steady state flux prediction based on transmembrane pressure (P) and membrane dry thickness (z) is straightforward
ii"
i'i
ii p
Zpp
ZJ
iii SD ,
The simplest case: constant permeability (S and D constant)
Partial transmembrane flux prediction
Membrane type Material and/or carrier CO2/N2
selectivity
CO2 permeability
(Barrer) or permeance
(GPU)
Gas separation
membrane (dense
polymers)
PEO-PBT
PEG/Pebax©
PEG-DME/ Pebax©
PEGDA/PEGMEA
Polaris™
70
47
43
41
50
120 Barrer
151 Barrer
600 Barrer
570 Barrer
1000 GPU
Facilated transport
membrane
(FTM
PAAM-PVA / PS
PVAm/PS*
PEI / PVA
PDMA/PS
PDMAMA
80
145
230
53
80
24 GPU
212 GPU
1 GPU
30 GPU
5 GPU
Liquid Membrane
(LM)
PVAm-PVA/PS
PVAm/PVA
Amines/PVA
Carbonic anhydrase
Amines / PVA
90
90
500
250
493
22 GPU
15 GPU
250 GPU
80 GPU
693 Barrer
Some material performances for post-combustion carbone capture
MTR
NTNU
Tested at pilot scale
with real gas
*M.-B. Hägg’s group
Steady state flux prediction requires an explicit knowledge of
upstream and downstream pressure
S and/or D non constant, permeability is concentration(partial pressure) dependent
Variable permeability
7
Ji = f (pi’, pi”, T)
Extrapolation based on transmembrane pressure (P) can
lead to significant error*…. i
ii p
ZJ
*Mauviel, G. et al. J. Membr. Sci. (2005) 266, 62-67.
S and/or D non constant, permeability is concentration(partial pressure) dependent
Variable permeability
8
Rubbery (PDMS)
Glassy (PSf)
Superglassy (PTMSP) Permeability is not always constant…
Material, structure, transport
Material
Polymer
Inorganic
Hybrid
Carbone
Zeolite
Metallic
Ceramic
Structure
Rubbery
Glassy
Symmetric
Porous
Dense
Mixed matrix membrane (MMM)
Metal–organic (MOF)
Carbon molecular sieve (CMS)
Composite
Anisotropic
Asymmetric
Ionic liquid
Nature
Dense
Transport mechanism
Solution -diffusion Physical
Chemical
Knudsen diffusion
Molecular sieving
Facilated transport
[Solution- diffusion /reaction- diffusion]
Porous
Ionic liquid
Solution diffusion Physical
Chemical Facilated transport
[Solution -diffusion /reaction- diffusion]
Material landscape
Glassy polymers Dual mode
Facilated transport membrane
Rubbery membrane (vapor permeation)
N
k kk
iiiH
iiDiHiDi
pb
pbCpkCCC
1
'
,
,,,
1
N
k kk
i
N
k kk
iiiH
iHii
iD
iDi
pxb
p
pxb
p
Z
bCDpp
Z
kDJ
1
"
1
''
,
,
"',
,
"'1
dx
dcD
dx
dcDJ i
iHi
iDi ,, with :
dx
dcD
dx
dcDJ AB
ABA
AA
SD mechanism RD mechanism
Henry’s sorption Langmuir’s sorption
Instantaneous
reaction
"',
ii
iD
iAi ppl
kDJ
"AA,D
"A
'AA,D
'A
A,DTAB"A
'A
A,DAA
pKk1
p
pKk1
pkKCDpp
l
kDJ
2iiiii )1()1()ln()aln(
sati
ii
p
pa
iii VmC
mi
"i
'i
iiVZ
DJ
Pi
PiSi
iPiSii
K
KK
aKK
1))exp((
Transport model- partial transmembrane flux (illustration)
Florry Huggin :
ENSIC model :
10
Type de membrane Mechanism Transport model Model parameter
Rubbery (vapor
permeation) SD Florry Huggin/ENSIC 𝐾𝑃𝑖, 𝐾𝑆𝑖
Glassy SD Dual Mode 𝐷𝐷𝑖 , 𝐷𝐻𝑖, 𝑘𝐷𝑖 ,
𝐶𝐻𝑖′ 𝑒𝑡 𝑏𝑖
Fixed site carrier
membrane SD et RD Facilated transport 𝐷𝑖 , 𝑘𝐷𝑖, 𝐷𝐻𝐶𝑂3
− , 𝐾 𝑒𝑡 𝐶𝑇
Transport model parameters
11
D: diffusion S: sorption R : reaction
Need for experimental data for model parameter determination
Type de membrane Mechanism Transport model Model parameter
Rubbery (vapor
permeation) SD Florry Huggin/ENSIC 𝐾𝑃𝑖, 𝐾𝑆𝑖
Glassy SD Dual Mode 𝐷𝐷𝑖 , 𝐷𝐻𝑖, 𝑘𝐷𝑖 ,
𝐶𝐻𝑖′ 𝑒𝑡 𝑏𝑖
Fixed site carrier
membrane SD et RD Facilated transport 𝐷𝑖 , 𝑘𝐷𝑖, 𝐷𝐻𝐶𝑂3
− , 𝐾 𝑒𝑡 𝐶𝑇, Z, k-1
𝐷𝑎 =𝑘−1𝑧2
𝐷𝐴𝐵
Damköhler number
A=CO2
B = R − NH3+
AB = 𝐻𝐶𝑂3−
Kinetically controlled reaction
Transport model parameter
12
13
-10
10
30
50
70
90
110
130
150
0
100
200
300
400
500
600
700
800
1 2 3 4 5 6 7
F N2
(10
-6 c
m3
(STP
).cm
- ².s
ec-1
))
J CO
2 (1
0-6
cm
3 (S
TP).
cm- ²
.sec
-1))
p' (bar)
Points expérimentaux -PVAm/PVA (CO2)Points expérimentaux - BAM-1(CO2)Simulation - PVAm/PVA (CO2)
Simulation - BAM-1 (CO2)
Points expérimentaux -PVAm/PVA (N2)Points expérimentaux - BAM-1(N2)Simulation - PVAm/PVA (N2)
Simulation - BAM-1 (N2)
Facilated transport membrane - Model /experiment
Data from literature
Experimental results from :
- A. Mondal et al., Int. J. Greenh. Gas Control, 2015. - M. Wang et al., Energy Environ. Sci., 2011. - Y. Cai, J. Membr. Sci., 2008.
M. Pfister’s PhD
0
100
200
300
400
500
600
700
800
1 3 5 7
J CO
2 (
10
-6 c
m3
(STP
).cm
-
².se
c-1))
p' (bar)
14
Material-process design interaction
Need humidity (near saturation) (need to avoid gas drying)
Active site carrier saturation at high CO2 partial pressure
A=CO2
B = R − NH3+
AB = 𝐻𝐶𝑂3−
This impacts module driving force strategy : - Upstream pressure : limited to avoid active site saturation - Downstream pressure : limited to avoid membrane drying
PVAm/PVA Binary dry basis CO2/N2 (20%/80%) (Mondal et al.)
humidity control
Need to avoid gas drying +
Fixed site carrier membrane
15
Material-Module design interactions
Structure/ thickness
Transport mechanism
Operating parameter
Local transmembrane flux
Driving force strategy
Flow pattern
MATERIAL MODULE
I II
Feed
Membrane
Permeate
Retentate
Q Q-dQ
x x-dx
dQ
Pdownstream
Pupstream
ds
z
Qin, xin Qout, xout
Qp, yp
in
p
Q
QStage cut
Key variables
Pressure ratio
B
A
*
'
''
P
P
Selectivity
Membrane gas separation process simulation Baseline approach
Key assumptions
Binary mixture Well defined hydrodynamics (CPF, CC, Co) Constant P’, P’’, T Constant permeability Mass transfer resistance = membrane (no polarisation effect) No flux coupling
16
ΔP’’ ΔP’ Concentration
polarization
Joule
Thompson
cooling
Real gas
behavior P = f(T) P =f(P)
Alpers et al.
1999 - - + - + - -
Coker et al.
1999 +
Kaldis et al.
1998 and
2000
+
- - - - + -
Marriott et
al. 2001 + + +
Esteves et
Motta 2002 - - - - - - -
Makaruk et
Harasek
2009
- - - - - - -
Katoh et al.
2011
+
- - - - - -
M. Scholz, T. Harlacher, T. Melin, and M. Wessling, “Modeling Gas Permeation by Linking
Nonideal Effects,” Ind. Eng. Chem. Res., vol. 52, no. 3, pp. 1079–1088, Jan. 2013.
Process modeling studies : a complex landscape
17
Variable permeability behavior can impact module design…
18
50%
55%
60%
65%
70%
75%
80%
85%
90%
95%
10% 60%
y CO
2 (
%)
RCO2 (%)
Cas 1 (p'=10 bar/p''=1 bar)
Cas c (p'=1 bar/p''=0,1 bar)
Perméabilité constanteConstant permeability
Glassy membrane * Dual mode transport model
* 6FDA-TAPDO
CO2/ CH4 : 20%/80%
0.0%
10.0%
20.0%
30.0%
40.0%
50.0%
60.0%
70.0%
80.0%
90.0%
100.0%
0 100 200 300 400 500 600 700
Ace
ton
e r
eco
very
Surface (m²)
Variable permeability behavior can impact module design
Acetone recovery from air Elastomeric membrane (PDMS) Flory Huggins transport model
Natural gas treatment Glassy membrane (CA)
Dual mode transport model
CO
2 co
nte
nt
in r
ete
nta
te
Surface (m2) 19
Structure/ thickness
Transport mechanism
Operating parameter
Local transmembrane flux
Driving force strategy
Flow pattern
DESIGN
Multistage/recycling Heat integration
Application
Available driving force Separation objectives
Cost minimization
From material to process design
20
MATERIAL MODULE
I II
III
MEMSIC* : a user friendly CAPE OPEN simulation tool
Characteristics:
Multicomponent mixtures compatibility (> 50)
Constant or variable permeability behavior
Transport models (S-D mechanism): Flory-Huggins, Dual mode, ENSIC, faciltated transport membrane
Multistages with recycling loops
Sucessfully tested on PSE softwares
(Aspen, Hysis, ProII, G Prom’s…)
*: © R. Bounaceur (2015) 21
Tailor made multicomponent module simulation for PSE
Operating inlet data • Temperature, T • Pressure, Pin
• Composition, xin
• Flow rate, Qin
Specifications/ Objectives Purity, recovery ratio
Material data • Transport model + parameters Or Constant permeability model and active layer thickness Membrane surface area (design
option)
Energy consumption Required membrane surface area
Driving force • Feed compression /vacuum/
turbo-expander on the retentate.
Flow configuration • Co-current • Counter current • Cross-plug flow
Feed
Membrane
Permeate
Retentate
Q Q-dQ
x x-dx
dQP ’’
P ’
ds
z
Qin, xin Qout, xout
Qp, y
Overall mass balance
Local mass balance
i
iii "p,'p,y,xJdS
'dq
i
iii
iiii
yx
yxy ,
'q
"p,'p,y,xJ"p,'p,yi,xJx
dS
dx i
iiiiii
i
iin,i xxet 0'q ,0S à
MEMSIC : a user friendly CAPE OPEN simulation tool
23
MEMSIC* : a user friendly CAPE OPEN simulation tool
24
Post combustion carbon capture (CO2, H2 O, N2, O2) : multicomponent problem - Compression changes water content - Water decreases surface (internal sweep) Simulation based on the sole pressure difference dependency of permeability are not
able to correctly predict separation performance Rigorous steady state flux prediction requires an explicit knowledge of upstream and
downstream pressure
Taking into account variable permeability generates a large effort in terms of inputs data (such effort might be not necessary but it can be of major importance in some cases (glassy, reactive membranes, vapor permeation in rubbery)
25
Concluding remarks
26
Thank you for your attention
EU- Australian workshop, Oslo, Norway 13th- 14TH september 2017