Subscriber access provided by Nottingham Trent University
Separations
A novel adsorption process for co-production of hydrogen and CO from a multicomponent stream 2
Anne Streb, Max Hefti, Matteo Gazzani, and Marco Mazzotti Ind. Eng. Chem. Res., Just Accepted Manuscript • DOI: 10.1021/acs.iecr.9b02817 • Publication Date (Web): 22 Aug 2019 Downloaded from pubs.acs.org on August 29, 2019
Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.
Page 1 of 50 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Industrial & Engineering Chemistry Research
A Novel Adsorption Process for Co-production of Hydrogen and CO2 from a Multicomponent Stream †
Anne Streb,
1
†
Max Hefti,
∗,‡
Matteo Gazzani,
∗,†
and Marco Mazzotti
†ETH Zurich, Institute of Process Engineering, Zurich, Switzerland ‡Utrecht University, Copernicus Institute of Sustainable Development, Utrecht, the
Netherlands E-mail:
[email protected];
[email protected] 2
Acronyms
3
AC
activated carbon
4
Ads
adsorption step
5
BD1
6
BD-vac
7
C
8
CCS
carbon dioxide capture and storage
9
CSS
cyclic steady state
10
EOS
equation of state
blowdown to heavy purge pressure blowdown to subatmospheric pressure
compressor
1
ACS Paragon Plus Environment
Industrial & Engineering Chemistry Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
11
HP
12
LP1
purge with light product (here: H2 ), outow recycled to heavy purge
13
LP2
purge with light product (here: H2 ), outow wasted
14
MDEA
15
MO-MCS
16
MCS
17
PE
18
PE-BD
19
PE-Pr
20
Press
21
PSA
22
Rec-BD
23
Rec-Pr
24
SMR
25
VP
26
VPSA
27
VSA
heavy purge
methyl diethanolamine multi-objective multilevel coordinate search
multilevel coordinate search
pressure equalization pressure equalization - blowdown pressure equalization - pressurization pressurization, here: with feed pressure swing adsorption recycling blowdown step pressurization with recycle, here: H2 -rich stream
steam methane reforming
vacuum pump vacuum pressure swing adsorption
vacuum swing adsorption
2
ACS Paragon Plus Environment
Page 2 of 50
Page 3 of 50 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
28
Industrial & Engineering Chemistry Research
Symbols
29
Cs
heat capacity of adsorbent [J/kg/K]
30
di
internal column diameter [m]
31
dP
particle diameter [m]
32
∆HAds,i
isosteric heat of adsorption [J/mol]
33
e
specic energy consumption [kJ/kgProduct ]; equation 4
34
eReboiler
CO2 specic reboiler heat duty [kJ/kgCO2 ]
35
eel
CO2 specic electricity consumption [kJ/kgCO2 ]
36
ex
CO2 specic exergy consumption [kJ/kgCO2 ]; equation 6
37
EHP
energy required to compress the recycled part of the CO2 product from ambient pressure to PHP (in case the HP is carried out above ambient pressure) [kJ]
38
EH2
energy required for recompressing the hydrogen-rich stream [kJ]
39
Etot
total energy consumption [kJ]
40
EVP
energy required for evacuating the column and purging under vacuum [kJ]
41
i
component [-]
42
ki
mass transfer coecient, linear driving force approximation [1/s]
43
Lcol
column length [m]
44
Mw,i
molecular weight of component i [kg/mol]
45
Ncol
number of columns [-]
46
Ncol, min
minimum number of columns [-]
47
Ni,Prod
molar amount of component i in the product rich in i [mol]
48
Ntot,Prod
molar amount of product rich in i [mol]
49
Ni,tot
total molar amount of component i fed to one cycle [mol]
50
PAds
column pressure at the end of Press and during Ads [bar]
51
PAmb
ambient pressure [bar]
52
PBD-vac
column pressure at the end of BD-vac [bar] 3
ACS Paragon Plus Environment
Industrial & Engineering Chemistry Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
53
PFeed
feedstream pressure [bar]
54
PHP
column pressure at the end of BD1 and during HP [bar]
55
PRec-BD
column pressure at the end of Rec-BD [bar]
56
Ps
column pressure of step s [bar]
57
Plow
lowest column pressure reached at the end of innitely long BD or BD-vac step [bar]; equation 1
58
Phigh
column pressure at the beginning of BD or BD-vac step [bar]; equation 1
59
P rinf
ideal productivity for an innite number of columns [kgCO2 /tads /hcycle ]
60
P re
eective productivity [kgCO2 /tads /hcycle ]; equation 5; for comparison with state of the art as volumetric productivity [kgCO2 /m3 /h]
61
ri
recovery of component i [-]; equation 3
62
rr
recycle ratio: ratio between the recycled to the total molar outow of a specic step [-]
63
s
specic step in an adsorption cycle, i.e. Ads, Press, i.a. [-]
64
tAds
duration of Ads [s]
65
tAds,LP1/2 duration of Ads, during which all outow is used to purge the column [s]
66
tBD-vac
duration of BD-vac [s]
67
tcycle
cycle duration not including idle times [s]
68
tidle
duration of idle times [s]
69
tLP1/2
duration of LP1 / LP2 [s]
70
ts
duration of step s [s]
71
Tamb
ambient temperature, here: 298 K
72
TFeed
feed temperature [K]
73
TReboiler
reboiler temperature [K]
74
Vcol
column volume [m3 ]
75
V˙ Feed
feed volumetric owrate [m3 /s]
76
x
decision variables vector 4
ACS Paragon Plus Environment
Page 4 of 50
Page 5 of 50 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Industrial & Engineering Chemistry Research
77
yi,Feed
molar fraction of component i in feed [mol/mol]
78
ηis
isentropic eciency [-]
79
Φi
purity of product rich in component i [-], equation 2
80
ρb
bulk density [kg/m3 ]
81
ρM
density of adsorbent material [kg/m3 ]
82
ρP
density of adsorbent particles [kg/m3 ]
83
ξ
tting parameter describing the pressure decrease at the column outlet
84
85
during blowdown and evacuation [-]; equation 1
Abstract
86
The production of carbon neutral H2 is pivotal for reaching net-zero CO2 in 2050.
87
Undoubtedly, time and scale of this transition call for the decarbonization of H2 produc-
88
tion from natural gas, where the separation processes account for a large share of the
89
capital and operational expenditures. Energy and cost ecient processes are therefore
90
highly sought for. With this contribution, we have developed, modeled and optimized
91
new vacuum pressure swing adsorption (VPSA) cycles for co-production of high-purity,
92
high-recovery CO2 and H2 from a ternary feedstream with a signicant amount of an
93
impurity. We identied two cycles that can purify CO2 up to 95 % with recoveries
94
greater 90 % whilst co-producing hydrogen with the same specications. Key cycle
95
features include purge under vacuum with part of the hydrogen product, and recycle
96
of the hydrogen rich outow during the initial part of the blowdown. The latter should
97
be carried out via a compressor for very high hydrogen purities and recoveries, and
98
via a sequence of pressure equalization (PE) steps for the targeted separation, which
99
also drastically reduces the energy consumption. The volumetric productivity ranges
100
from 160 to 240 kgCO2 /m3 /h, which is signicantly larger than available open data for
101
absorption-based CO2 capture from hydrogen production plants (productivities in the
102
range of 60 to 90 kgCO2 /m3 /h). The energy consumption, when evaluated via exergy
103
to fairly compare heat and electricity, is in the range of state of the art processes (0.5 5
ACS Paragon Plus Environment
Industrial & Engineering Chemistry Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
104
MJ/kg CO2 ). Finally, the developed VPSA cycles reduce the separation steps from two
105
to one, which paves the way for further process intensication.
106
1
Introduction
107
Rapid and deep decarbonisation of the global energy system is of utmost importance: limit-
108
ing the global warming to 1.5 °C requires to reach net zero CO2 emissions by 2050. 1 Carbon
109
neutral hydrogen is regarded as instrumental for such an energy system to enable the de-
110
carbonization of industry, transportation and heating. However, it is very unlikely that the
111
production of carbon neutral hydrogen via electrolysis using renewable energy so called
112
green hydrogen will provide hydrogen at the required scale within the 2050 time framework.
113
At present, hydrogen is produced at rather large scale from fossil fuels, especially for use in
114
reneries and chemicals production, e.g. ammonia. Clearly this production route features
115
high CO2 emissions, and is not compliant with the 1.5 °C IPCC pathway. Accordingly, there
116
is an urgent need for a scalable hydrogen production process that can provide large quantities
117
of hydrogen already today with little associated CO2 emissions.
118
Coupling fossil-fuel based hydrogen production with carbon dioxide capture and storage
119
(CCS) is likely the only route that can provide carbon neutral hydrogen at the required
120
scale and time. Moreover, it will accelerate the transition to a H2 -based energy system,
121
with evident benets for electrolyzer development. Finally, when replacing fossil fuels with
122
biogenic sources, e.g. biogas, such technology will enable negative carbon emissions.
123
Dierent hydrogen production routes are shown in gure 1. The state of the art production
124
from fossil fuels is auto-thermal or steam reforming of natural gas followed by a water-gas-
125
shift reactor to convert CO to CO2 . Alternatives include coal or oil gasication, or partial
126
oxidation of either natural gas or biogas. The dierent processes produce high pressure syn-
127
gas containing hydrogen and CO2 , but also signicant amounts of impurities like N2 , CO
128
and CH4 . Therefore, the hydrogen production process requires a purication step, where
6
ACS Paragon Plus Environment
Page 6 of 50
Page 7 of 50
H2 is separated from CO2 and impurities. State of the art for this separation is a pressure Fuel feedstock
Syngas production
Utilization
Separation processes
1
Coal
1-3
Absorption
Gasification
- physical - chemi-physical
Oil Partial oxidation
4 H2/C
- Natural Gas - Biogas
Reforming
CO2
Syngas (H2, CO, CH4, N2)
Pressure swing adsorption H2 Impurities (CO, CH4, N2)
Water Gas Shift
PSA waste
specific to burners
Heating Fischer-Tropsch synthesis H2
CO2 stream
> 70 %
Refineries
> 99 %
- refinery streams - methanol offgas - coke oven gas - ammonia offgas
CO2 use CO2 purity > 95 %
b) New VPSA process CO2 H2 Impurities VPSA waste
> 60 %
Chemicals production
Transportation
- steam reforming - autothermal reforming
Residues of industrial process
H2 use
Gas Turbine combustion
a) State-of-the-art
H2 purity
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Industrial & Engineering Chemistry Research
> 99.99 %
Underground sequestration Enhanced oil recovery (EOR) Chemicals production Technological fluid
Figure 1: Fossil- and biofuel based hydrogen production coupled with carbon capture using (a) state of the art technology, or (b) a new VPSA process 129 130
swing adsorption (PSA) process, where a number of steps, including pressure equalizations,
131
a product purge and a product pressurization, are used. 2 The process runs continuously by
132
adopting several columns, often layered with dierent adsorbent materials. The nal prod-
133
ucts are (i) a H2 stream with purity sucient for reneries or transportation, and (ii) a waste
134
stream.
135
Notably, such a conguration requires an additional unit to capture CO2 , e.g. before or
136
after the PSA unit. Commercial processes for CO2 capture exist and are based on absorp-
137
tion using aqueous solutions, 3 e.g. physical solvents (Selexol, Rectisol or Purisol process),
138
or hybrid chemical/physical solvents (activated MDEA - aMDEA). Other options include
139
cryogenic separation, membranes or adsorption. 4 In gure 1, the state of the art is shown
140
with an absorption based capture system before the PSA unit, which is the preferred option
141
for coupling hydrogen production with CCS. 4 A promising alternative to these processes is
142
the integration of hydrogen purication and CO2 separation in a single adsorption cycle, i.e.
143
what is shown as option b) in gure 1. By removing one separation stage, this will likely 7
ACS Paragon Plus Environment
Industrial & Engineering Chemistry Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
144
imply a decrease in complexity, system cost and energy consumption. The development of
145
advanced sorbents together with a high exibility in terms of cycle design make adsorp-
146
tion processes suitable for a wide range of applications. However, adsorption processes for
147
co-production of both light and heavy component here H2 and CO2 , respectively have
148
not been studied and developed thoroughly, and only a few examples exist in the literature.
149
One such case is the Gemini process, which makes use of two interconnected PSA trains for
150
the production of high purity hydrogen and high purity CO2 ; dierent adsorbents can be
151
used in the dierent trains and a vacuum pump is needed for CO2 withdrawal. 5 The Gemini
152
process, however, still uses two trains undergoing dierent cycles, which are only connected
153
during specic steps. Notably, an industrial demonstration plant for CO2 capture from a
154
steam methane reforming (SMR) hydrogen production plant through adsorption exists at
155
the Port Arthur renery in Texas. 6 There, a vacuum swing adsorption (VSA) process is used
156
to separate CO2 from the syngas upstream of the PSA unit. 7
157
The co-production of two products at high purities and recoveries involves many addi-
158
tional challenges compared to the purication of either the light or the heavy product. A
159
suitable adsorbent selectively adsorbs CO2 over impurities and hydrogen to allow for high
160
purity CO2 product, while at the same time preferably adsorbing CO2 and impurities over
161
hydrogen to allow for high purity hydrogen product. In addition to the availability of an
162
appropriate adsorbent, performing this separation task within a single adsorption cycle re-
163
quires a precise management of multiple concentration fronts. Because of the cyclic nature
164
of adsorption processes, how fronts propagate in dierent steps is coupled in a complex man-
165
ner. As a consequence, it is not sucient to simply add steps typical for the purication of
166
the light component to a cycle designed for the purication of the heavy component, or vice
167
versa; instead new cycles have to be developed for dierent applications.
168
In this paper, we present an adsorption process for the co-production of hydrogen and CO2
169
from a multicomponent feed within a single cycle. In particular, it presents the development
170
and screening of dierent cycle congurations, and the rigorous optimization of promising 8
ACS Paragon Plus Environment
Page 8 of 50
Page 9 of 50 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Industrial & Engineering Chemistry Research
171
cycles to assess their separation performance, their energy requirement for a given separation
172
task, and their eective productivity taking into consideration the scheduling in a multi-
173
column setup. Our main process target is hydrogen production with CCS, where both CO2
174
and H2 products should reach high purities and recoveries. Therefore, a recovery ≥ 90 % at
175
purity ≥ 95 % for both products is pursued, which is in line with the recommendations of
176
the US DOE for CO2 capture rates 8 and CO2 purity for transportation and storage in saline
177
aquifers. 9,10 It is worth stressing that the process could be tuned for dierent specications,
178
e.g. higher purities with lower recoveries.
179
The structure of the paper is as follows. In the rst section, the cycle design is ex-
180
plained and four dierent cycles for the H2 -CO2 -impurity separation are introduced. In the
181
second section, the column model is described, the important parameters are provided and
182
the optimization procedure is explained. In the third section, the separation and process
183
performance of the dierent cycles is discussed explaining the inuence of important process
184
parameters, cycle conguration and scheduling constraints on the key performance indica-
185
tors, namely the hydrogen and CO2 purities and recoveries, the specic energy consumption
186
and the productivity. Finally, the most promising cycles are identied and compared to the
187
state of the art.
188
2
189
The cycles for the H2 -CO2 -impurity separation presented here have been developed by com-
190
bining in a new fashion the steps necessary to increase the purity or the recovery of either
191
hydrogen or CO2 or of both. These include high pressure adsorption, recycle of hydrogen and
192
CO2 -rich streams for increasing recovery, purge with hydrogen and CO2 to increase purity,
193
and the use of dierent pressure levels for hydrogen, recycle, waste and CO2 withdrawal,
194
including subatmospheric pressure. In addition to the puried hydrogen and CO2 streams,
195
an integrated adsorption process will produce a third stream containing the impurities to-
Cycle design
9
ACS Paragon Plus Environment
Industrial & Engineering Chemistry Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
196
gether with residual H2 and CO2 . The four most promising cycles are described below. 11 For
197
illustration, the evolution of the H2 molar fraction at the column top and of the CO2 molar
198
fraction at the column bottom over the course of one cycle at cyclic steady state (CSS) are
199
shown in gure 2 (a) for representative simulations of cycles A and B. The pressure proles
200
for the same simulations at the column top are shown in gure 2 (b). In addition, the frac-
201
tional CO2 uptake, dened as the ratio of the amount of CO2 adsorbed within the column to
202
the amount that would be adsorbed if the whole column would be in equilibrium with CO2
203
at the pressure of the heavy purge (HP), PHP , is shown in gure 2 (c). This illustrates the
204
sorbent saturation with respect to CO2 before the nal evacuation. For further illustration,
205
we refer to the internal gas phase column proles provided in the supplementary material.
206
Cycle A,
which is shown in gure 3, is a seven step VPSA cycle combining most of
207
the characteristic steps mentioned above. During a high pressure adsorption step (Ads),
208
hydrogen is produced at high purity, while impurity and CO2 adsorb (see gure 2). This is
209
followed by a recycling blowdown step (Rec-BD), during which the column is depressurized to
210
an intermediate pressure PRec-BD and the outow rich in hydrogen and impurity is recycled
211
to partially repressurize the column before adsorption. This reduces the loss of H2 while
212
increasing its recovery. The Rec-BD is followed by a blowdown to heavy purge pressure
213
(BD1), during which a waste stream lean in H2 and rich in impurities is produced (see
214
gure 2). The column is then purged at constant pressure PHP with part of the CO2 -rich
215
product. During HP, the impurities and the hydrogen - both in the adsorbed phase (mainly
216
impurities) and in the gas phase - are replaced with CO2 . This increases the achievable
217
CO2 product purity, while more waste is produced. At the end of this step, the column is
218
close to its maximum CO2 adsorption capacity at PHP , as shown in gure 2 (c). This step
219
is followed by a nal blowdown to subatmospheric pressure (BD-vac), which is driven by
220
a vacuum pump (VP). Here, the CO2 product is withdrawn at high purity, as it can be
221
seen in gure 2 (a), and part of it is recycled and used for the HP step. In the subsequent
222
two steps, the column is repressurized in a bottom up conguration, rst with the recycled 10
ACS Paragon Plus Environment
Page 10 of 50
Page 11 of 50
molar fraction in gas phase
(a)
cycle A - CO2 bottom
1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0
Ads
(b)
Rec-Bd
cycle A - H2 top
BD1
HP
cycle B - CO2 bottom
BD-vac
cycle A - pressure profile column top
LP
cycle B - H2 top
Rec-Pr
Press
cycle B - pressure profile column top
30
pressure [bar]
25 20 15 10 5 0
Ads
Rec-Bd
BD1
(c)
CO 2,ads / CO2,max
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Industrial & Engineering Chemistry Research
HP
BD-vac
cycle A
1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0
Ads
Rec-Bd
BD1
HP
LP
Rec-Pr
Press
LP
Rec-Pr
Press
cycle B
BD-vac
Figure 2: (a): Exemplary proles of the molar fraction of CO2 at the column bottom and H2 at the column top over the normalized step duration for cycle A and cycle B; (b): Exemplary pressure proles at the column top over the normalized step duration for cycle A and cycle B; (c): Ratio of CO2 adsorbed to maximum possible CO2 adsorbed at 1 bar (PHP ) and feed temperature over the normalized step duration for cycle A and cycle B; Note that cycle A does not have a LP step; cycles C and D are conceptually similar to cycles A and B and show the same trends, with the pressure decreasing and increasing in sequence over three PE steps replacing Rec-BD and Rec-Pr
11
ACS Paragon Plus Environment
Industrial & Engineering Chemistry Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 12 of 50
223
hydrogen-rich stream (Rec-Pr) using a compressor (C), and nally with the feed (Press), to
224
reach the adsorption pressurePAds . H2 product
PAds
PAds ↓ PRec-BD
waste 1
waste 2
PRec-BD ↓ PHP
PHP
PHP ↓ PBD-vac
C
PRec-Pr ↑ PBD-vac
PAds ↑ PRec-Pr
VP
Feed
Ads
Feed
CO2 product
Rec-BD
BD1
HP
BD-vac
Rec-Pr
Press
Figure 3: Cycle A: VPSA cycle for co-production of light and heavy product; compressor for recycle of hydrogen-rich stream 225
Cycle B:
A key drawback of cycle A is the large amount of CO2 present within the
226
column after the BD-vac step, as shown in gure 2 (c). This reduces the achievable hydrogen
227
purity and the CO2 cyclic capacity, unless a very high vacuum is applied for regeneration.
228
Therefore, in cycle B (gure 4) a purge step is added after the BD-vac step, where the column
229
is purged in a top down conguration with part of the light product (LP1), i.e. hydrogen.
230
During this step, the column is cleaned starting from the top from adsorbed impurities and
231
CO2 , while also displacing the CO2 in the void phase. The adsorbed CO2 within the column
232
decreases signicantly, as illustrated in gure 2 (c). A stream initially rich in CO2 with
233
increasing concentrations of impurities and hydrogen is produced, which is used as part of
234
the stream used to purge the column in the HP step. Therefore the CO2 molar fraction
235
of the HP inlet is lower compared to cycle A, as shown in gure 2 (a), but hopefully still
236
sucient to successfully purge the column and withdraw high purity CO2 in the following
237
step. The outow of the HP step consists mostly of impurity with increasing amounts of
238
hydrogen, which enters with the HP and is not adsorbed. This is dierent for cycle A, where
239
no new hydrogen enters during the HP and the outow consists mainly of impurity.
240
Cycle C, which is shown in gure 5, is similar to cycle A with one conceptual dierence 12
ACS Paragon Plus Environment
Page 13 of 50 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Industrial & Engineering Chemistry Research
H2 product
PAds ↓ PRec-BD
PAds
waste 1
waste 2
PRec-BD ↓ PHP
PHP
PHP ↓ PBD-vac VP
Feed
PBD-vac
C
PRec-Pr ↑ PBD-vac
PAds ↑ PRec-Pr
Rec-Pr
Press
VP
CO2 product
Ads
Rec-BD
BD1
HP
BD-vac
Feed
LP1
Figure 4: Cycle B: VPSA cycle for co-production of light and heavy product; compressor for recycle of hydrogen-rich stream, purge with hydrogen (LP), reduces to cycle A for tLP1 → 0 241
regarding the methodology for recycling the hydrogen rich intermediate product during the
242
column depressurization. While cycle A uses a compressor, which provides high exibility
243
in terms of the nal blowdown pressure but requires energy for recompressing the stream,
244
cycle C adopts a series of three pressure equalization (PE) steps. This results in an 11 step
245
cycle. A top down conguration is chosen for all PE steps, in which the top of the column
246
in the PE-BD step is connected with the bottom of the column in the PE-Pr step, to avoid
247
contamination of the column top end with impurities. PE steps are expected to lower the
248
energy consumption, while limiting the number of variables available for the cycle ne-tuning
H2 product
PAds
PAds ↓ PPE1
PPE1 ↓ PPE2
PPE2 ↓ PPE3
waste 1
waste 2
PPE3 ↓ PHP
PHP
PHP ↓ PBD-vac
PPE3 ↑ PBD-vac
PPE2 ↑ PPE3
PE-Pr3
PE-Pr2
PPE1 ↑ PPE2
PAds ↑ PPE1
VP
Feed
Ads
CO2 product
PE-BD1
PE-BD2
PE-BD3
BD1
HP
BD-vac
Feed
PE-Pr1
Press
Figure 5: Cycle C: VPSA cycle for co-production of light and heavy product; three pressure equalizations for recycle of hydrogen-rich stream 249
13
ACS Paragon Plus Environment
Industrial & Engineering Chemistry Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 14 of 50
250
Cycle D, which is shown in gure 6, is based on cycle C, but features an additional LP
251
step similar to cycle B, i.e. the possibility of venting the latter part of the outow during
252
the LP step (LP2). This could become necessary if the purge duration needed to reach the
253
target hydrogen purity is very long. Then, the hydrogen and impurity content in the outow of LP1 is too high for the HP step, hence it is wasted during LP2. H2 product
PAds
PAds ↓ PPE1
PPE1 ↓ PPE2
PPE2 ↓ PPE3
waste 1
waste 2
PPE3 ↓ PHP
PHP
PHP ↓ PBD-vac a
Feed
Ads
PBD-vac
PBD-vac
VP
VP
CO2 product
PE-BD1
PE-BD2
PE-BD3
BD1
HP
BD-vac
PPE3 ↑ PBD-vac
PPE2 ↑ PPE3
PPE1 ↑ PPE2
waste 3
LP1
LP2
PAds ↑ PPE1
Feed
PE-Pr3
PE-Pr2
PE-Pr1
Press
Figure 6: Cycle D: VPSA cycle for co-production of light and heavy product; three pressure equalizations for recycle of hydrogen-rich stream, purge with hydrogen, reduces to cycle C for tLP1/2 → 0 254 255
For adsorption processes, there is a trade-o between product purity and recovery. In
256
the case of a ternary H2 /CO2 /impurity separation, where a waste stream is produced in
257
addition to the two products, the purities and recoveries of CO2 -rich and H2 -rich product
258
are not directly linked. Therefore, all these four indicators have to be taken into consideration
259
for evaluating the separation performance. Notably, there are a few key design variables that
260
strongly aect the performance of the dierent cycles.
261
2.1
262
The duration of the
263
of both H2 and CO2 . Increasing it leads to impurity and CO2 fronts propagating further
264
into the column, eventually contaminating the H2 product and decreasing its purity. On
265
the other hand, because of the longer duration of this step, more H2 is produced during one
Important Parameters
adsorption step
tAds has a strong inuence on recovery and purity
14
ACS Paragon Plus Environment
Page 15 of 50 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Industrial & Engineering Chemistry Research
266
cycle, which increases the H2 recovery. For CO2 , the eect is opposite: longer adsorption
267
times are favourable for its purity, but unfavourable for its recovery.
268
The HP step, as explained before, is mainly used to increase the CO2 purity by replacing
269
the gas phase rich in H2 and impurities with a CO2 -rich gas phase. The variable inuencing
270
how much CO2 is used for the HP is the
271
ratio between the recycled to the total molar outow of this step. Increasing the recycle
272
ratio and using more of the CO2 product for the HP step therefore leads to an increase in
273
CO2 purity. However, with higher recycle ratios more CO2 is lost during the HP step and
274
the CO2 recovery decreases. Changing the recycle ratio has a strong inuence neither on H2
275
purity nor on its recovery.
276
The
pressure
recycle ratio
rr of the BD-vac step, dened as
PRec-BD is an important variable for cycle A and cycle B to nd the
277
optimal trade-o between H2 recovery and purity: lower pressure increases the H2 recovery
278
because more H2 -rich outow is recycled, but decreases the H2 purity because impurities and
279
CO2 build up at the end of this step, as shown in gure 2 (a). Decreasing PRec-BD therefore
280
also leads to a slight increase in CO2 recovery, but has no signicant eect on CO2 purity.
281
Increasing the
pressure
PHP leads to a decrease in CO2 purity. First, because more
282
hydrogen and impurities are contained within the column voids at the beginning of HP.
283
Second, with reference to ambient pressure, more CO2 -rich recycle is needed. For the same
284
purge eciency, stronger adsorption and higher gas density lead to slower propagation of
285
the CO2 front. For the same reason, less CO2 is lost during this step, thereby increasing the
286
CO2 recovery with higher PHP . The eect of the purge pressure on the hydrogen purity and
287
recovery is small.
288
When maximizing the recovery or purity for one product whilst constraining the recovery
289
and purity for the other, the eect of changing a variable on both products is relevant to
290
reach the maximum for one but still be within the constraints for the other. It is clear from
291
these considerations that the interplay among the dierent decision variables is not trivial
292
and a rigorous optimization approach is necessary to obtain the best separation and process 15
ACS Paragon Plus Environment
Industrial & Engineering Chemistry Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
293
performance for the dierent cycle congurations, as it will be explained in the next section.
294
3
295
3.1
296
A non-isothermal, one-dimensional model based on mass, energy and linear momentum bal-
297
ance equations is used for modeling the process. It has previously been validated for a variety
298
of conditions and cycles 1214 and has been extensively used for the design of new cycles. 15,16
299
The equations are provided in the supplementary material, the main simplications and
300
assumptions are summarized here.
Process modeling and optimization Mathematical model
301
no radial gradients in concentration, velocity and temperature
302
the Ergun equation is used to describe the pressure drop along the column
303
thermal equilibrium between the gas phase and the solid phase
304
the gas phase is described using the ideal gas law, 17 which is a well accepted simplica-
305
tion for VPSA simulation. As reference, the compressibility factor for the CO2 /N2 /H2
306
mixtures calculated with Refprop 9.0 is between 0.95 and 1.00 for the pressures and
307
temperatures of interest
308 309
mass transfer is described using a linear driving force approximation, and a temperature/concentration independent mass transfer coecient
310
axial dispersion and conductivity are neglected 12
311
constant heat of adsorption, molar heat capacities and viscosity of the gas phase
312
For modeling the cyclic nature of adsorption processes, a single column is simulated that
313
cyclically undergoes the sequence of steps by changing the boundary conditions until a CSS is
314
reached. At CSS, neither the internal composition and temperature proles, nor the product 16
ACS Paragon Plus Environment
Page 16 of 50
Page 17 of 50 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Industrial & Engineering Chemistry Research
315
composition and the total mass balance change between two consecutive cycles, as described
316
in detail in previous work. 15,16,18 The threshold for the mass balance is set to below 1 % for
317
cycles A and B and below 2 % for cycles C and D (due to the PE steps) and the threshold
318
for the internal proles to 10-4 . The pressure decrease during BD steps is described as
319
exponential decay in the following form:
P (ts ) = Plow + (Phigh − Plow ) exp(−ξts )
(1)
320
A value of ξ = 0.11 is used in this work. ξ has been tted to cyclic experiments under similar
321
conditions, as discussed by Marx et al. 19
322
An intermediate storage tank is assumed to be present for all recycling steps and adiabatic
323
operation is assumed for the columns. 18
324
Activated carbon (AC) was identied as promising adsorbent for the considered sepa-
325
ration and is used for all simulations. We do not expect this to be the optimal choice for
326
all applications, but want to keep our analysis general and leave the choice of the best ad-
327
sorbent or adsorbent combination to specic case studies. In addition, this allows us to
328
prot from a very good experimental database for AC in terms of single component and
329
binary isotherms for CO2 , N2 and H2 as well as mass transfer parameters from binary and
330
ternary breakthrough experiments. 17,19,20 The adsorption equilibria, which have previously
331
been characterized in our group, are described in the form of a temperature dependent mul-
332
ticomponent Sips isotherm. The equations and isotherm parameters are reported in the
333
supplementary material. The other relevant parameters are provided in table 1.
334
A generic inlet consisting of 50 % CO2 , 25 % H2 , and 25 % N2 as impurity at a pressure
335
of 30 bar is used for cycle screening and optimization. Here, the aim is to design and identify
336
cycles that can provide high purity - high recovery CO2 and H2 in the presence of signicant
337
amounts of impurities. Typically, syngas impurities, e.g. CO, N2 , CH4 , adsorb more than
338
H2 but less than CO2 . Moreover, they usually account for 5 to 10 vol.% of the feed gas. In
17
ACS Paragon Plus Environment
Industrial & Engineering Chemistry Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
339
this work, we have decided to represent all impurities with a large molar fraction of N2 , a
340
condition for which the model has been validated experimentally. 13
341
In addition to the impurities mentioned above, which are typically present in % or %
342
levels, several trace impurities like higher hydrocarbons or sulfur components can be present
343
in the feed in much lower concentrations down to ppm levels, depending on the upstream
344
process. A common feature of those trace impurities is that they adsorb stronger and often
345
irreversibly on adsorbents like AC or zeolites. Therefore, a guard layer is usually included
346
in the column to reversibly adsorb the trace impurities and protect the main part of the
347
bed. For this general analysis, we do not consider a guard layer, because it depends on the
348
nature and concentration of those trace impurities, which are process specic. However, a
349
few considerations as to how the design of the cycles presented above inuences such a guard
350
layer should be made. Several of the recycle streams (PE-Pr, HP and Rec-Pr) are directed
351
to the column top, thereby possibly leading to a further propagation of the impurities into
352
the column bed. Therefore, an external guard bed might be a more suitable solution for the
353
presented cycles. Alternatives include a longer guard bed or feeding the recycle streams on
354
top of the guard layer.
355
Note that also water, which on AC adsorbs more than CO2 if the relative humidity is large
356
enough (about 25 to 30 %), is usually present in traces in pressurized syngas at ambient
357
temperature, as relevant here. Its strong adsorption, hysteresis behavior, and competition
358
with CO2 make the quantitative description within simulation of CSS processes particularly
359
challenging. The H2 O adsorption behavior on AC - and related challenges - were extensively
360
discussed by Hefti and Mazzotti. 21,22 Examples in the literature where H2 O is considered
361
in cycle simulations are scarce, and typically refer to the use of layered beds in the context
362
of post-combustion CO2 capture, where water is a prominent impurity. 16,23,24 Moreover, H2
363
purication from wet syngas is current industrial practice in PSA. Accordingly, in this work
364
we don't consider H2 O in the feed to the VPSA, which is assumed to be perfectly dry.
18
ACS Paragon Plus Environment
Page 18 of 50
Page 19 of 50 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Industrial & Engineering Chemistry Research
Table 1: Material and column parameters for the simulations, details in Marx et al. 13
Parameter
Particle diameter Material density Particle density Bed density Heat capacity adsorbent Isotherm parameters Isosteric heat of adsorption mass transfer coecient Internal column diameter Column length
Symbol Unit
dP m ρM kg/m3 ρP kg/m3 ρb kg/m3 Cs J/kg/K See supplementary ∆HAds,i J/mol ki 1/s di m Lcol m
Value
0.003 1965 850 480 1000 material, as reported in 20 CO2 : 21000; N2 : 15600 ; H2 : 9800 CO2 : 0.11; N2 : 0.3; H2 : 1 a 0.025 b 1.2 b
a
tted to breakthrough experiments for a ternary mixture equivalent to the one used for this study: H2 :CO2 :N2 = 50:25:25 b based on lab-scale adsorption setup at ETH Zurich 365
3.2
Key performance indicators
366
The most important indicators to compare the separation performance of dierent adsorption
367
processes are the product purities and recoveries. The purity Φi is dened as the ratio
368
between the amount of moles of target component i in the product rich in this component,
369
Ni,Prod , to the total amount of moles of this product, Ntot,Prod :
Φi =
Ni,Prod Ntot,Prod
(2)
370
The recovery ri is dened as the ratio of Ni,Prod to the overall amount of component i fed to
371
the cycle, Ni,tot , as follows:
ri =
Ni,Prod Ni,tot
(3)
372 373
Two additional parameters are used to characterize the process performance for a given
374
separation, namely the specic energy consumption e and the eective productivity P re .
375
Whereas the specic energy consumption is an indicator of the operating cost of the plant,
376
the productivity determines the total plant volume, a major contributor to the capital cost,
19
ACS Paragon Plus Environment
Industrial & Engineering Chemistry Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 20 of 50
377
and the amount of adsorbent needed, which is part of the operating cost. The specic energy
378
consumption is given per mass of product i separated (CO2 or hydrogen) in kJ/kg, and the
379
eective productivity as mass of product separated per unit time and unit mass of adsorbent
380
in kg/(tads h):
e= 381
P re =
Etot Ni,Prod Mw,i
Ni,Prod Mw,i (tcycle + tidle )ρb Vcol
(4) (5)
382
Where Mw,i is the molecular weight of component i, ρb is the bulk density and Vcol the
383
column volume.
384 385
386 387
388 389
390
To compute the total energy consumption of the VPSA process Etot , three dierent contributions must be considered: the energy required by the vacuum pump (VP) for evacuating and purging the column at P < PAmb , EVP the energy required to compress the recycled part of the CO2 product from ambient pressure to PHP , in case HP is carried out above ambient pressure, EHP the energy required for recompressing the hydrogen-rich stream, EH2
391
Ultimately, all three energy needs are requested in the form of electricity to drive the three
392
corresponding machines. In addition, the separation process might aect the downstream
393
product compression, e.g. the CO2 compression for geological storage, or the H2 compression
394
for ammonia production or any other further use. This would be particularly relevant if (i)
395
the products pressure varies when changing the cycles variables, and/or (ii) H2 and/or CO2
396
are supplied at dierent pressures compared to the state of the art, e.g. CO2 above P Amb
397
or H2 at P < P Feed . However, for all four cycles considered here, the two products leave
398
the VPSA at constant pressure irrespective of the changes in the considered design variables
399
and in line with the state of the art processes, i.e. CO2 at P Amb , and H2 at about P Feed .
400
Accordingly, the products compression is not considered in our optimization as this would 20
ACS Paragon Plus Environment
Page 21 of 50 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Industrial & Engineering Chemistry Research
401
result in a constant demand, equal to the state of the art and dependent on the downstream
402
process, which is not included as part of this work.
403
The computation of the three energy contributions is based upon simple isoentropic
404
eciency ηis or more detailed models in Aspen Plus. Specically: (i) for EVP we assume
405
a linear stepwise decrease in the vacuum pump eciency for decreasing vacuum pressure,
406
which is required to account for lower performance at deeper vacuum conditions, as shown
407
in Krishnamurthy et al., 25 (ii) EHP is computed with a constant η is = 0.8, and (iii) EH2 is
408
computed using a polynomial regression of an intercooled compression simulated in Aspen
409
Plus, where the input variables are the total compression ratio and the H2 content in the
410
recycle. Notably, the latter allows to consider the changing composition and pressure ratio
411
in the H2 -rich recycle.
412
Because of the cyclic nature of adsorption processes, they are usually accommodated in a
413
series of columns executing the same sequence of steps shifted in time. To synchronize steps
414
like pressure equalizations which require the direct connection of two columns, or to full
415
additional process constraints, e.g. a continuous feed ow, or a continuous production, an
416
idle time tidle is necessary. The idle time depends on the cycle conguration, the step times
417
and the number of columns Ncol used. Notably, a minimum number of columns Ncol, min is
418
necessary to accommodate a specic cycle under specic constraints, yet the optimal number
419
of columns can be higher if this allows for a signicant reduction of idle times. The eective
420
productivity P re can dier signicantly from the productivity for an innite number of
421
columns resulting in zero idle times P rinf = Pr e (tidle = 0). Whereas the one column model
422
used in this paper computes the productivity for an innite number of columns, the eective
423
productivity used to compare dierent cycles P re as dened in eq. (5) takes the scheduling
424
and resulting idle times into account. The scheduling equations needed to calculate the
425
idle time for dierent cycle congurations are provided in the supplementary material. The
426
constraints are a synchronization of the PE steps and a continuous feed.
21
ACS Paragon Plus Environment
Industrial & Engineering Chemistry Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
427
3.3
Optimization and parametric analysis
428
3.3.1 Separation performance
429
For optimizing the separation performance, the code is combined with a multi-objective
430
optimization routine. The optimization is based on a multilevel coordinate search (MCS)
431
algorithm, that was adapted to handle multiple objectives (MO-MCS) as described in Capra
432
et al. 26 The variables considered in the optimization are pressures Ps and durations ts of
433
each of the dierent steps, s, whereas adsorbent and column related parameters are kept
434
constant. Because of the many steps and dierent pressure levels, only parameters with a
435
strong inuence on the cycle performance are included in the optimization. Those are listed
436
below for all considered cycles:
437
Cycle A:
438
Rec-BD step, PRec-BD and iii) the recycle ratio rr of the BD-vac step.
439
Cycle B: the same parameters as for cycle A are optimized, and in addition i) the duration
440
of the LP1 step and ii) the amount of hydrogen used for this step.
441
Cycle C: i) the duration of the adsorption step, and ii) the recycle ratio as for cycle A and
442
iii) the pressure at which the high pressure purge is carried out, PHP .
443
Cycle D: the decision variables are the same as for cycle C and in addition i) the duration
444
of both LP1 and LP2 and ii) the amount of hydrogen used for those steps as described for
445
cycle B.
446
To assess the separation performance, two dierent optimizations are carried out for all
447
cycles with the decision variables vector x as mentioned above:
i) the duration of the adsorption step, tads , ii) the pressure at the end of the
Optimization separation performance CO2 minimize (−ΦCO2 (x), −rCO2 (x)) x
s.t.
rH2 ≥ 0.90 ΦH2 ≥ 0.95
448 449
Optimization separation performance H2 22
ACS Paragon Plus Environment
Page 22 of 50
Page 23 of 50 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Industrial & Engineering Chemistry Research
minimize (−ΦH2 (x), −rH2 (x)) x
s.t.
rCO2 ≥ 0.85 ΦCO2 ≥ 0.95
450
A constraint of 85 % was chosen for the CO2 recovery for the optimization of the H2 separa-
451
tion performance, because not all cycles did reach 90 % whilst delivering a CO2 purity ≥ 95
452
%. This allows to assess the inuence of the dierent variables on both purities and recov-
453
eries and to compare the dierent cycle congurations. The decision variables, steptimes,
454
pressure levels and feedstream considered for all simulations are provided in table 2.
455
3.3.2 Energy consumption and productivity
456
To assess the energy consumption and productivity, an extensive parametric analysis was
457
carried out. With this approach, the constraints on hydrogen and CO2 recoveries and purities
458
do not need to be dened beforehand and the optimal performance of the dierent cycles
459
can be compared for dierent purity and recovery constraints. To obtain the same level of
460
renement for the Pareto front as with the optimizer, however, signicantly more function
461
evaluations would be necessary. Therefore, rst a coarse parametric analysis is carried out
462
which is subsequently rened for the promising ranges of dierent decision variables.
463
The decision variables are the same as for the optimization of the separation performance.
464
In addition, also the duration of the BD-vac step is set as variable thereby also varying the
465
lowest column pressure. The lowest pressure Plow , as dened in eq.1 is 0.01 bar. In accordance
466
with this, also for the Rec-BD step, the duration tRec-BD is set as variable instead of dening
467
the nal pressure PRec-BD . This allows to include the impact of the step duration on the
468
productivity. The decision variables, steptimes, pressure levels and feedstream condition for
469
all simulations are provided in table 2.
23
ACS Paragon Plus Environment
Industrial & Engineering Chemistry Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 24 of 50
Table 2: Cycle parameters and variables for optimization of CO2 and H2 separation performance and parametric analysis of energy consumption and productivity
Process conditions separation performance cycle
TFeed PAds V˙ Feed a PBD-vac yi,Feed d tAds e tRec-BD PRec-BD tBD1 tHP tBD-vac tLP1 tLP2 tRec-Pr tPE-Pr/PE-BD rr
A
B
C
D
energy/productivity A
B
C
K
298
298
bar
30
30
2 × 10-5
2 × 10-5
m3 /s bar
0.1 b
D
f(tBD-vac ) c
0.1
-
H2 :CO2 :N2 = 50:25:25
H2 :CO2 :N2 = 50:25:25
s
variable
variable
s bar
50
-
variable
-
variable
-
f(tRec-BD ) c
-
s
50
50
s
50
50
s
50
s
f
-
f(tAds,LP1/2 )
s
-
f(tAds,LP1/2 )
30
s
-
-
f(tAds,LP1/2 ) f
f(tAds,LP1/2 ) f
-
s -
variable f
5
f(tAds,LP1/2 ) f f(tAds,LP1/2 ) f
30 -
variable
-
5 variable
a
Dened at PFeed and TFeed Optimization of CO2 separation performance for dierent pressure levels PBD-vac from 0.01 bar to 0.3 bar, see gure 8 c Exponential pressure decrease according to equation 1 d Cycle A: Optimization of CO2 separation performance for two additional impurity contents: H2 :CO2 :N2 = 46.7:23.3:30 and H2 :CO2 :N2 = 53.3:26.7:20, , see gure 8 e Including pressurization and the time dedicated to provide the hydrogen for the purge LP1/2 f Depending on duration of adsorption step during which the outow hydrogen product is used to purge the column, tAds,LP1/2 : tLP1/2 = 50 × tAds,LP1/2 b
470
4
471
In this section we will show:
472 473
Results and Discussion
which cycles and which among their features are favourable for reaching a good CO2 separation performance; 24
ACS Paragon Plus Environment
Page 25 of 50 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
474 475
476 477
478 479
Industrial & Engineering Chemistry Research
which cycles and which among their features are favourable for reaching a good H2 separation performance; which cycles can achieve the desired coproduction of both products and at what energy penalty and productivity; how the most promising cycles perform compared to state of the art absorption processes for CO2 capture from H2 production facilities. The four cycles can be categorized as indicated in table 3. Cycle A as base cycle has no LP Table 3: Four dierent cycles developed for CO2 -H2 co-production w/o LP w/ LP
w/ Compressor Cycle A Cycle B
w/ PE steps Cycle C Cycle D
480 481
step and makes use of a compressor for recycling part of the hydrogen rich outow. Cycle
482
B has an additional LP step. Cycle C is based on cycle A, but makes use of PE steps for
483
recycling part of the hydrogen rich outow instead of a compressor, and cycle D has both a
484
LP step and PE steps. Therefore, the eect of the LP step and of exchanging a compressor
485
with a series of PE steps can be assessed separately.
486
Whereas we expect the LP step to have a positive inuence on the CO2 cyclic capacity, it
487
might result in a lower CO2 purity, because of the increasing amount of impurity in the
488
purge gas for the HP step. In section 4.1, we will show that the inuence on the CO2
489
separation performance is indeed positive and that CO2 can be produced at a sucient
490
purity. Exchanging a compressor with PE steps translates into a loss of exibility and
491
therefore possibly of separation performance, but also into savings in energy consumption.
492
We will show in section 4.1 that the inuence on the CO2 separation performance is marginal,
493
whereas the H2 separation performance decreases signicantly when making use of PE steps
494
instead of a compressor. For the desired product purities and recoveries, however, PE steps
495
are sucient and result in a reduced energy penalty compared to using a compressor, as will
496
be shown in section 4.2. 25
ACS Paragon Plus Environment
Industrial & Engineering Chemistry Research
497
The section starts by examining the eect of dierent factors on the separation perfor-
498
mance, including the cycle conguration, the feedstream composition and the evacuation
499
pressure. Subsequently, the energy consumption and productivity for the dierent cycles
500
will be compared including the eect of scheduling on productivity. Finally, the dierent
501
cycles will be compared to the state of the art based on their overall performance and their
502
suitability for dierent applications.
503
4.1
504
Figure 7 shows the Pareto fronts for the four dierent cycles when (a) maximizing CO2
505
recovery and purity whilst co-producing high purity and high recovery hydrogen and (b)
506
maximizing H2 recovery and purity whilst co-producing high purity and high recovery CO2
507
with parameters and variables as shown in table 2.
Optimization: separation performance
CO 2 Recovery
1 0.95
(b)
0.9 1
0.85 0.8 0.75
0.99
H2 Purity
1 0.95 (a) 0.9 0.85 0.8 0.75 0.7 cycle A 0.65 cycle B rH2 ≥ 90 % 0.6 cycle C ФH2 ≥ 95 % 0.55 cycle D 0.5 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
H2 Purity
CO 2 Purity
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 26 of 50
0.7 1
0.3
0.98 0.97 0.96 0.95 0.9 0.92 0.94 0.96 0.98
H2 Recovery
0.4
0.5
0.6
1
0.7
rCO2 ≥ 85 % ФCO2 ≥ 95 %
0.8
H2 Recovery
0.9
1
Figure 7: Separation performance all cycles; (a): CO2 purity and recovery for dierent cycles, constraint: ≥ 90 % H2 recovery, ≥ 95 % H2 purity; (b): H2 purity and recovery for dierent cycles, constraint: ≥ 85 % CO2 recovery, ≥ 95 % CO2 purity;
508
4.1.1 Inuence of cycle conguration on CO2 purity and recovery
509
Figure 7 (a) shows that CO2 can be produced at very high purity (>99.7 %) or at very
510
high recovery (> 99 %) for all cycles whilst co-producing H2 at high purity and recovery 26
ACS Paragon Plus Environment
Page 27 of 50 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Industrial & Engineering Chemistry Research
511
specications. The target of ≥ 95 % CO2 purity and ≥ 90 % CO2 recovery, however, can
512
only be reached for two cycles, cycles B and D, whereas cycles A and C miss the target
513
region by a few percentage points. The gure shows clearly that cycles A and C on the one
514
hand and cycles B and D on the other hand have a very similar performance in terms of
515
CO2 purity and recovery.
516
Compressor vs. PE. The main dierence between cycle A and cycle C (or between cycle
517
B and cycle D, respectively) is the way the hydrogen is recycled: with a compressor in the
518
case of cycles A and B, while exploiting PE steps for cycles C and D. The dierence in the
519
CO2 purity/recovery is small for cycles that only dier in the way the hydrogen is recycled.
520
This shows that pressure at the end of the hydrogen recycle (PRec-BD or PPE-BD3 respectively)
521
has only a minor inuence on the CO2 purity and recovery. In addition, for a large part
522
of the Pareto front, the optimal pressure at the end of the Rec-BD reached when using a
523
compressor is close to the nal pressure reached after the third PE-BD step (4-5 bar). The
524
possibility of ne-tuning the Rec-BD pressure for cycles A and B therefore does not have a
525
signicant eect on the CO2 separation performance.
526
LP step.
527
the addition of a LP step to either cycle A or cycle C increases signicantly the separation
528
performance with respect to CO2 , which can now be obtained at high purity with high
529
recovery (approx. 97% CO2 recovery for 95 % CO2 purity possible). During the LP step,
530
hydrogen replaces CO2 within the gas phase. This leads to additional CO2 desorption,
531
which is withdrawn from the column at relatively high purity. Because the CO2 -rich outow
532
is recycled and used to purge the column in the HP step, the LP increases the eective cyclic
533
CO2 capacity. Less product CO2 has to be used for the same HP eciency, thus resulting
534
in a higher recovery at the same purity. For cycle D, also the option of wasting part of this
535
outow has been included in the optimization. This, however, has a negative eect on the
536
CO2 recovery because part of the CO2 is wasted, and only a minor eect on the CO2 purity.
537
Therefore, the duration of this purge is very small for all optimal points.
In contrast to the small inuence of exchanging the compressor for PE steps,
27
ACS Paragon Plus Environment
Industrial & Engineering Chemistry Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
538
4.1.2 Inuence of cycle conguration on H2 purity and recovery
539
Figure 7 (b) illustrates the optimization of the hydrogen purity and recovery. To be able to
540
compare all cycles, a constraint of ≥ 85 % CO2 recovery together with ≥ 95 % CO2 purity
541
was chosen. The Pareto sets show clearly that all cycles can easily produce high purity
542
hydrogen at high recovery, i.e. surpassing the target of ≥ 90 % H2 recovery and ≥ 95 %
543
H2 purity. A very high purity hydrogen product (> 99.97%) can be produced with cycle B
544
or cycle D including the LP step, whereas very high hydrogen recovery (> 99.96 %) can be
545
reached for cycle A and B, which make use of a compressor for recycling the hydrogen rich
546
blowdown outow. So both the LP step and the way the hydrogen rich outow is recycled
547
after the adsorption step strongly aect the hydrogen separation performance.
548
Compressor vs. PE. Whereas the nal pressure at the end of the Rec-BD step (relevant for
549
cycles A and B) has only a minor eect on both CO2 recovery and purity, it is an important
550
variable to nd the optimal trade-o between H2 recovery and purity: lower pressure leads
551
to an increase in hydrogen recovery, because more hydrogen-rich outow is recycled, and to
552
a decrease in hydrogen purity, because more impurities and more CO2 are recycled. For very
553
high H2 recoveries, the pressure PRec-BD approximates the minimum possible pressure at the
554
end of this step, PHP , thus making the intermediate BD1 unnecessary. When using PE steps
555
instead of a compressor, the nal pressure at the end of PE-BD3 is determined mainly by
556
the number of PE steps. In developing cycles C and D, three PE steps have been identied
557
as favourable for achieving the target CO2 purity and recovery while limiting the complexity.
558
The nal pressure after PE-BD3 is in the range of 4-5 bar, so signicantly higher than PHP ,
559
and therefore it limits the maximum hydrogen recovery for cycles C and D. Increasing the
560
number of PE steps would lead to an increase not only in performance (hydrogen recovery),
561
but also in the minimum number of columns and valves required.
562
For maximizing hydrogen purity, increasing PRec-BD (or reducing the number of PE steps)
563
is required, because less impurities are recycled. Therefore cycle A reaches higher purities
564
than cycle C. However, the addition of the LP step allows high purity hydrogen production 28
ACS Paragon Plus Environment
Page 28 of 50
Page 29 of 50 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Industrial & Engineering Chemistry Research
565
for both cycles B and D as discussed below. It should be noted that the inuence of PRec-BD
566
on purity is signicantly lower than on recovery.
567
LP step.
568
increases the maximum obtainable hydrogen purity signicantly, which is clear when com-
569
paring the H2 Pareto fronts of cycle A and cycle B (or of cycle C and cycle D). Not only does
570
the LP step result in the desorption of additional CO2 , but also in eective removal of CO2
571
and impurities from the column top, thus pushing the obtainable hydrogen purity to above
572
99.97 % for cycles B and D. This step, however, uses part of the H2 product, thus decreasing
573
its recovery. For high target hydrogen recovery, the duration of the purge approximates zero
574
and cycles B and D reduce to cycles A and C, respectively: the Pareto fronts overlap. It
575
should be noted, that cycle B, which combines the two favourable features for the production
576
of high purity hydrogen at high recovery, namely the Rec-BD with compressor and the LP,
577
achieves a hydrogen product purity of 99.9 % with a recovery of 90 % whilst co-producing
578
CO2 at recovery ≥ 85 % and purity ≥ 95 %.
579
4.1.3 Inuence of evacuation pressure and impurity content on CO2 purity and
580
The LP step is not only benecial for the CO2 separation performance, but it also
recovery
581
In addition to the cycle decision variables discussed above, there are two boundary conditions
582
that strongly aect the separation performance: i) the nal evacuation pressure PBD-vac , and
583
ii) the molar fraction of the impurity in the feed, yN2 . We have therefore investigated
584
how dierent values inuence the CO2 separation performance, by considering cycle A as
585
exemplary test case. It should be mentioned that the H2 separation performance follows
586
similar trends. The results are illustrated in gure 8, which shows the change of the Pareto
587
fronts when changing the evacuation pressure (a) and the N2 content in the feed (b).
588
Evacuation pressure.
589
because more CO2 desorbs and can be recovered as product during the evacuation. This
590
leads to an increase in CO2 recovery for the same purity. In fact, lowering the minimum
With lower evacuation pressure, the cyclic CO2 capacity increases
29
ACS Paragon Plus Environment
1 0.95 (a) 0.9 0.85 0.8 0.75 0.7 0.05 bar 0.65 0.07 bar rH2 ≥ 90 % 0.6 0.1 bar ФH2 ≥ 95 % 0.55 0.2 bar 0.5 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
CO 2 Recovery
CO 2 Purity
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
CO 2 Purity
Industrial & Engineering Chemistry Research
1
1 0.95 (b) 0.9 0.85 0.8 0.75 0.7 20 % N2 0.65 25 % N2 0.6 30 % N2 0.55 0.5 0.3 0.4 0.5 0.6
Page 30 of 50
rH2 ≥ 90 % ФH2 ≥ 95 %
0.7
0.8
CO 2 Recovery
0.9
1
Figure 8: Optimized separation performance for cycle A: maximize ΦCO2 and rCO2 for different cycles, constraints: rH2 ≥ 90 %, ΦH2 ≥ 95 %; (a): dierent evacuation pressures at yN2 = 25 %; (b): dierent impurity content in the feedstream at PBD-vac = 0.1 bar; when decreasing/increasing yN2 , the ratio of CO2 to H2 was kept constant. 591
evacuation pressure to 0.07 bar enables cycle A to reach the targeted CO2 purity and recovery
592
(95 % and 90 % respectively). It is anyhow worth stressing that a lower evacuation pressure
593
is favourable for the separation performance but leads to a higher energy consumption, as
594
more energy is required for the vacuum pump. Therefore the evacuation pressure is an
595
important decision variable for the optimization of energy consumption and productivity.
596
Notably, with the evacuation pressure as additional variable, all cycles can reach > 90 %
597
recovery for both products at > 95 % purity. It is worth mentioning that decreasing the
598
evacuation pressure also has a positive eect on the hydrogen separation performance: the
599
higher cyclic capacity for CO2 results in a longer duration of the adsorption step before the
600
nitrogen front breaks through, which leads to an increase in hydrogen recovery for the same
601
hydrogen purity.
602
Impurity content.
603
our calculations, the CO2 in the adsorbed phase and in the gas phase increases, which makes
604
the separation easier. Also for the hydrogen separation, reducing the impurity content is
605
favourable: under the operating conditions of interest, the convex isotherm shape allows for a
606
prompt N2 adsorption while limiting the N2 content in the gas phase signicantly. Therefore,
When reducing the N2 content, which plays the role of impurities in
30
ACS Paragon Plus Environment
Page 31 of 50 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Industrial & Engineering Chemistry Research
607
the propagation velocity of the impurity front decreases with a decreasing impurity content.
608
This allows for a longer adsorption step before the nitrogen front breaks through. This
609
leads to increasing hydrogen recovery whilst maintaining the same hydrogen purity. For an
610
increase in N2 concentration in the feed, the separation performance gets worse, based on
611
the same argument.
612
4.2
613
4.2.1 Cycle comparison
614
The results of the parametric analysis for minimizing energy consumption and maximizing
615
ideal productivity (i.e. considering zero idle time) by varying operating variables as shown
616
in table 2 are illustrated in gure 9 in the form of Pareto fronts, considering both CO2 (a)
617
and H2 (b) as products. Notably, the Pareto points might dier depending on the target
618
product, i.e. CO2 or H2 . This is because the recovery and purity of the CO2 and hydrogen
619
product can dier along the Pareto front, as long as they satisfy the minimum constraint.
620
Points representing a Pareto optimum for both products are indicated as lled symbols,
621
those representing an optimum for either CO2 or H2 only are shown as empty symbols.
Parametric analysis: energy consumption and productivity
622
The graphs show a trade-o between the energy consumption and the productivity for all
623
cycles. This is related to the change in evacuation pressure along the Pareto front, as shown
624
in gure 10: the evacuation pressure decreases for all cycles when moving from minimum to
625
maximum energy consumption (or minimum to maximum productivity). When decreasing
626
the evacuation pressure, the CO2 cyclic capacity increases, thus leading to a more ecient
627
use of the column and allowing for longer adsorption times without the CO2 and impurity
628
fronts breaking through. As a result, more CO2 is produced in a cycle, thus increasing both
629
productivities, Pr CO2 and Pr H2 . The energy consumption, however, increases because the
630
column is regenerated at lower pressures.
631
Moreover, it can be noted from gures 9 and 10 that there is a signicant dierence
632
between the four cycles. This indicates that both exchanging the compressor for the Rec-BD 31
ACS Paragon Plus Environment
2200
cycle C
2000 (a) 1800
cycle A
1600 1400 1200 cycle B
1000 800 600
ri ≥ 90 % Фi ≥ 95 %
cycle D
400 200 250 300 350 400 450 500 550 600
energy consumption [kJ/kg H2]
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
energy consumption [kJ/kgCO2 ]
Industrial & Engineering Chemistry Research
productivity [kgCO2/(t adshcycle )]
1.8 1.6
Page 32 of 50
104
cycle C
(b)
cycle A
1.4 1.2
cycle B
1 0.8 0.6 0.4 20
cycle D
ri ≥ 90 % Фi ≥ 95 %
25
30
35
40
45
50
55
productivity [kgH2 /(t ads h cycle )]
60
Figure 9: Optimized (parametric analysis) energy consumption and productivity for all cycles, constraints: rH2 ≥ 90 %, ΦH2 ≥ 95 %, rCO2 ≥ 90 %, ΦCO2 ≥ 95 %; (a): per CO2 separated; (b): per hydrogen produced; points representing an optimum for both products are represented as lled symbols and those only representing an optimum for either CO2 or H2 are represented as empty symbol 633
with a series of PE steps as well as the addition of a LP step play an important role in
634
determining both energy requirement and productivity along with the optimal evacuation
635
pressure.
636
Compressor vs. PE.
When comparing a cycle with compressor (cycle A or cycle
637
B) to the corresponding cycle with PE steps (cycle C or cycle D), the conguration with
638
the PE steps achieves always a lower minimum specic energy consumption, i.e. cycle C
639
has a lower minimum specic energy consumption than cycle A and cycle D has a lower
640
minimum specic energy consumption than cycle B. This is related to the additional energy
641
required for recompressing the Rec-BD stream in cycles A and B. Notably, cycle C also
642
shows higher productivities compared to cycle A, and cycle D compared to cycle B. This is
643
due to the shorter duration of the PE steps in comparison to the Rec-BD and Rec-Pr step.
644
The dierences in productivity are small compared to the dierences in energy consumption
645
and do not account for scheduling constraints, as is necessary to compare cycle times in a
646
meaningful way (see following section).
647
LP step.
When comparing a cycle without LP (cycle A or cycle C) to the corresponding
32
ACS Paragon Plus Environment
Page 33 of 50
0.09
0.1
(a)
ri ≥ 90 % Фi ≥ 95 %
0.08 0.07 0.06 0.05
0.09
cycle D cycle B
0.04
cycle A
0.03 0.02 0.01 min
(b)
0.07
cycle D
0.06 0.04
cycle A
0.03 0.01 min
max
cycle B
0.05
0.02
cycle C
CO2 energy consumption
ri ≥ 90 % Фi ≥ 95 %
0.08
PBD-vac [bar]
0.1
PBD-vac [bar]
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Industrial & Engineering Chemistry Research
cycle C
H2 energy consumption
max
Figure 10: Change of evacuation pressure PBD-vac moving along Pareto front from minimum to maximum CO2 specic energy consumption (a) and from minimum to maximum H2 specic energy consumption (b), constraints: rH2 ≥ 90 %, ΦH2 ≥ 95 %, rCO2 ≥ 90 %, ΦCO2 ≥ 95 % 648
cycle with LP (cycle B or cycle D), gure 9 shows that the addition of the LP step reduces the
649
minimum energy consumption. This is because cycles B and D w/ LP achieve the required
650
separation performance at a higher evacuation pressure, whereas cycles A and C w/o LP
651
require a lower evacuation pressure to reach the minimum target CO2 recovery of 90 %, as
652
shown in gure 10. Therefore, the minimum energy consumption is lower for cycles with a
653
LP step. The penalization of low evacuation pressure, as in real vacuum pumps, makes this
654
dierence even more pronounced. Besides the lower energy consumption, also the maximum
655
productivity is higher for cycles w/ LP (B or D) than for the corresponding cycles w/o LP
656
(A or C), because of the combined eect of lowering the evacuation pressure and of purging
657
the remaining CO2 and impurities out of the column. It is important to note that the Pareto
658
fronts for the cycles w/ LP (B or D) have to be at least as good as those for the corresponding
659
cycles w/o LP (A or C), because cycles B and D reduce to cycles A and C when the duration
660
of the LP approaches zero.
661
A better understanding of the energy consumption can be gained by analysing the in-
662
dividual contributions for all four cycles. As an interesting example, the Pareto point with
663
minimum CO2 specic energy consumption is considered. The individual contributions along 33
ACS Paragon Plus Environment
Industrial & Engineering Chemistry Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
664
the Pareto front are provided in the supplementary material.
665
The energy required for the recycle compressor in cycles A and B makes up a signicant share
666
of the overall energy consumption, which depends on the pressure PRec-BD , accounting for
667
approximately a fourth (cycle A) and over a half (cycle B) of the total energy requirement.
668
Nevertheless, for cycle A the minimum energy consumption is only slightly larger than for
669
cycle C w/o recycle compressor, as shown in gure 9. This is because cycle C has a worse
670
separation performance than cycle A and therefore requires a lower evacuation pressure, i.e.
671
larger vacuum pump consumption as shown in gure 10. For both, the nal evacuation
672
pressure is signicantly below 0.1 bar with 0.025 bar for cycle C and 0.032 bar for cycle A,
673
thereby reaching the limit of technically feasible pressures for industrial applications. How-
674
ever, when the LP step is included, the energy requirement for evacuation drops drastically:
675
it reduces by more than half for cycle B compared to cycle A and by approximately two
676
thirds for cycle D compared to cycle C. This is due to the better separation performance
677
when adding a LP step, which translates into a higher nal evacuation pressure close to 0.1
678
bar for both cycles C and D, as shown in gure 10.
679
In terms of minimizing the energy consumption, cycle D therefore is particularly promising
680
(as evident in gure 9) because it combines the positive eect of adding a LP step and of
681
replacing the compressor with a series of PE steps. It should be noted that for all the optimal
682
points, HP is carried out at ambient pressure.
683
4.2.2 Eect of scheduling
684
So far, we have only shown the ideal productivity, i.e. the productivity for the case of zero
685
idle times. Adsorption processes are usually accommodated in a train that consists of several
686
columns. The same cycle is repeated in every column shifted in time. Scheduling constraints,
687
e.g. deriving from the need of having continuous operation, lead to idle times depending on
688
the actual number of columns. The constraints considered here are: i) a continuous feed
689
for all cycles, and ii) a synchronization of the PE steps for cycles C and D. This requires a 34
ACS Paragon Plus Environment
Page 34 of 50
Page 35 of 50 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Industrial & Engineering Chemistry Research
690
product storage tank for hydrogen, a buer tank for the Rec-BD outow, and a buer tank
691
for the CO2 -rich recycle stream. These constraints lead to a minimum number of columns
692
needed to accommodate each specic cycle conguration and a reduced productivity com-
693
pared to the ideal productivity. To enable a fair comparison of the productivities for the
694
dierent cycles, the eect of scheduling therefore has to be taken into consideration.
695
In gures 11 and 12, two schedules for cycle C for the point with the minimum CO2 specic
696
energy consumption are shown. A minimum of 8 columns is needed to accommodate the
697
cycle and feed continuously, as shown in gure 11. Because of the constraint of a continu-
698
ous feed, the duration of the feed receiving steps (Ads and Press) is equal to the time shift
699
between two columns. At all times, exactly one column receives the feed. Because of the
700
required synchronization of the PE steps, there are large idle times after PE-Pr1, PE-Pr2
701
and PE-Pr3.
702
When increasing the number of columns, the time shift between two columns stays the
703
same to full the constraint of a continuous feed. Therefore, the scheduling remains the
704
same and an additional column will simply be idle for the duration of the time shift be-
705
tween two columns. When further increasing the number of columns, eventually the point is
706
reached where the feed can be split equally between two columns, which is shown in gure
707
12. This now reduces signicantly the idle time because of the shorter time shift between
708
two columns and therefore between two PE-Pr steps. It should be noted that when the feed
709
is split equally between two or more columns, a continuous hydrogen production is achieved
710
in addition to a continuous feed, as can be seen in gure 12: the duration of the adsorption
711
step is longer than the duration of the time shift between two columns. When adding more
712
columns, the eective productivity will decrease until the number of columns is high enough
713
to allow the feed to be split equally between 3 columns or 4 or 5 or more.
714
The trend described for cycle C (minimum number of columns required to accommodate
715
the cycle, decrease in productivity when increasing the number of columns until the next
716
favourable conguration is reached, continuous hydrogen production starting from this next 35
ACS Paragon Plus Environment
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Figure 12: Schedule for cycle C for the minimum CO2 -specic energy consumption and a favourable conguration with a high eective productivity using 13 columns
Figure 11: Schedule for cycle C for the minimum CO2 -specic energy consumption and the minimum number of columns (8)
Industrial & Engineering Chemistry Research
ACS Paragon Plus Environment
36
Page 36 of 50
Page 37 of 50
717
favourable conguration) is similar for the other cycles. This is shown in gure 13. The
718
gure shows the eect of the number of columns on the productivity Pr e for all cycles,
719
calculated as representative exemplary case for the Pareto point with the minimum CO2
720
specic energy consumption (as shown in gure 9a). In addition to the eective productiv-
721
ities, the productivity for an innite number of columns Pr inf is indicated as a vertical line
722
and the minimum number of columns as a horizontal line. The gure shows that there is a dierent minimum number of columns needed to accomcycle C
effective productivity in kgCO2/t ads /hcycle
effective productivity in kgCO2/t ads /hcycle
effective productivity in kgCO2/t ads /hcycle
20 18 16 14 12 10 8 Ncol = Ncol, min 6 4 2 0 200 250 300 350 400 450 500 550 600 Pr = Prinf
Pr = Prinf
cycle D
number of columns
cycle B
20 18 16 14 12 10 8 Ncol = Ncol, min 6 4 2 0 200 250 300 350 400 450 500 550 600
20 18 16 14 12 10 8 N =N col col, min 6 4 2 0 200 250 300 350 400 450 500 550 600 Pr = Prinf
number of columns
20 18 16 14 12 10 8 6 Ncol = Ncol, min 4 2 0 200 250 300 350 400 450 500 550 600 Pr = Prinf
number of columns
cycle A
number of columns
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Industrial & Engineering Chemistry Research
effective productivity in kgCO2/t ads /hcycle
Figure 13: eective productivity for all cycles depending on the number of columns for the Pareto point with the minimum CO2 -specic energy consumption; the maximum productivity for an innite number of columns and zero idle times is indicated as vertical line for all cycles; the minimum number of columns is indicated as horizontal line for all cycles 723 724
modate each cycle. The minimum number of columns increases when adding the LP step 37
ACS Paragon Plus Environment
Industrial & Engineering Chemistry Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 38 of 50
725
to cycles A or C, because of an overall longer duration of the cycle. It is shown clearly in
726
the gure that increasing the number of columns does not lead to a monotonic increase in
727
productivity, but that the productivity is maximized for only certain numbers of columns.
728
Departing from there, it decreases when increasing the number of columns until a new
729
favourable conguration is reached that leads to a higher eective productivity. Once the
730
number of columns is high enough for the feed to be split between two columns, a continu-
731
ous hydrogen production is achieved in addition to a continuous feed, as illustrated for cycle
732
C above. The eective productivities for the additional constraint of a continuous hydro-
733
gen product withdrawal are the same as for a continuous feed but starting from a higher minimum number of columns. This is summarized in table 4. Table 4: Minimum number of columns needed for continuous feed and continuous H2 production for all cycles with the eective productivities, the ideal productivities in case of zero idle times and the decrease in % compared to the ideal productivities, given for the point with the minimum CO2 specic energy consumption, points shown with black outline in gure 13
Cycle
A B C D
continuous feed Pr CO2 NCol kgCO2 /(tads h) 438 7 391 9 384 8 362 9
∆P r % 5 11 30 31
continuous H2 Pr CO2 kgCO2 /(tads h) 438 414 472 434
production NCol ∆P r % 14 5 17 6 13 14 15 17
zero idle times Pr CO2 kgCO2 /(tads h) 463 439 551 523
734 735
When comparing the eective productivity with scheduling Pr e and the ideal produc-
736
tivity Pr inf for zero idle time for all cycles, it can be seen from gure 13 and table 4, that the
737
drop in productivity is signicantly lower for cycles A and B than for cycles C and D. This
738
is due to the additional constraint of synchronizing the PE steps in cycles C and D. This
739
constraint leads to large idle times, i.e. in the range of three times the time shift between
740
two columns, whereas for cycles A and B, the idle time is shorter, i.e. in the range of the
741
time shift between two columns. As a consequence, when using the minimum number of
742
columns, the eective productivities of cycles C and D drop below those of cycles A and B
743
respectively. This is in contrast to the ideal productivities with zero idle time, that are larger 38
ACS Paragon Plus Environment
Page 39 of 50 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Industrial & Engineering Chemistry Research
744
for cycles C and D than for cycles A and B. For cycles C and D, the gain when increasing
745
the number of columns to the next higher favourable conguration, e.g. 13 columns instead
746
of 8 columns for cycle C, is high (> 22 % increase in productivity), whereas for cycles A and
747
B, there is almost no improvement in productivity.
748
4.3
749
All four cycles presented above can co-produce hydrogen and CO2 at 90 % recovery and 95
750
% purity for both products. Cycle B (w/ LP), however, outperforms cycle A (w/o LP) in
751
all respects (better CO2 separation performance, better hydrogen separation performance,
752
lower specic energy consumption and higher eective productivity for the same energy con-
753
sumption and number of columns) and cycle D (w/ LP) outperforms cycle C (w/o LP) in
754
all respects. Therefore, only cycles B and D will be considered in the following.
755
The Pareto curves for minimizing the CO2 specic energy consumption and maximizing the
756
productivity for purities ≥ 95% and recoveries ≥ 90 % for both products are shown in g-
757
ure 14 for an innite number of columns (empty symbols) and for the minimum number of
758
columns required when accounting for the scheduling (lled symbols). Note that the min-
759
imum number of columns required decreases with increasing productivities. This is due to
760
an increase in the duration of the adsorption step.
Overall cycle assessment
761
Figure 14 shows that cycle D is clearly superior in terms of the minimum energy con-
762
sumption. Because of the LP step, cycle D achieves the separation with a relatively high
763
evacuation pressure (in contrast to cycles A and C) and without the need of a compressor
764
for recycling the hydrogen-rich Rec-BD outow as PE are used instead (in contrast to cycles
765
A and B). The maximum eective productivity for cycle D, however, is lower than for cycle
766
B due to the additional constraint of synchronizing the PE steps when scheduling cycle D.
767
In addition to the higher eective productivity and the lower minimum number of columns
768
for the target separation, cycle B is particularly promising when higher hydrogen purities
769
and/or recoveries are needed. This is due to a more precise control over the pressure until 39
ACS Paragon Plus Environment
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
energy consumption [kJ/kgCO2]
Industrial & Engineering Chemistry Research
1800 1600 1400 1200 1000 800
Cycle B w/ scheduling
NCol= 7 NCol= 8 w/o NCol= 9 scheduling NCol= 8
600
NCol= 9
400
w/ scheduling
scheduling
Cycle D w/o scheduling
200 200 250 300 350 400 450 500 550 600
productivity [kgCO2/(tads h cycle)]
Figure 14: CO2 specic energy consumption and productivity for cycles B and D w/ scheduling (lled symbols) and ideal productivity P rinf for an innite number of columns (empty symbols); eective productivity calculated for the minimum number of columns as indicated; constraints: rH2 ≥ 90 %, ΦH2 ≥ 95 %, rCO2 ≥ 90 %, ΦCO2 ≥ 95 % 770
which the hydrogen-rich outow after the adsorption step is recycled. The pressure at the
771
end of three PE steps is in the range of 4-5 bar, which is favourable for reaching the target
772
CO2 separation performance. For maximum ΦH2 and targeted rH2 = 90 %, however, a lower
773
pressure of around 3 bar is favourable; for maximizing rH2 for ΦH2 = 95 %, an even lower
774
pressure of just above 1 bar is required. Notably, neither is reached with three PE steps.
775
An option for maximizing either hydrogen purity or recovery whilst still fullling the three
776
other constraints would be the addition of more PE steps to cycle D. This, however, would
777
result in a more complicated schedule with an increase in the minimum number of columns
778
and a further decrease in P re . In addition, a cycle with PE steps will never cover the whole
779
range of hydrogen recoveries and purities, as it is possible with cycle B.
780
Thus summarizing, when considering all factors, namely the separation requirements, the
781
exibility requirements, the cost of adsorbent, columns and compressor, the use of the hy-
782
drogen product and the price of electricity, either cycle B or cycle D might result to be the
783
best choice. 40
ACS Paragon Plus Environment
Page 40 of 50
Page 41 of 50 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Industrial & Engineering Chemistry Research
784
785
4.4
Comparison of cycle performance to state of the art
786
To put the performance of the developed VPSA cycles into perspective, a comparison with
787
the absorption-based CO2 capture process is needed. To this end, we rst need to have a
788
fair energy comparison: Whereas the VPSA process requires electricity for evacuation and
789
compression, the energy required for absorption based processes is a combination of heat -
790
in the form of low pressure steam for regenerating the CO2 -rich solution - and electricity,
791
required for pumps, chilling, recycle compressors. To account for the dierent forms of the
792
required energy, we have computed the overall exergy ex for both processes. While for the
793
VPSA, the exergy consumption corresponds to the electricity consumption (i.e. electricity
794
is pure exergy), for the reference we need to convert the heat requirement into exergy.
795
Accordingly, the reboiler heat duty for solvent regeneration is converted to exergy by means
796
of the Carnot factor, which is computed between the reboiler temperature TReboiler and
797
ambient temperature Tamb , here taken as 298 K. The resulting exergy consumption of the
798
state of the art is as follows:
ex = eReboiler 1 −
Tamb TReboiler
+ eel
(6)
799
Concerning the productivity of the state of the art, dierent values are available in
800
literature, and most notably the detailed design of Shell Quest project (via Alberta's CCS
801
Knowledge Sharing Program). 4,2729 Here, for simplicity we calculate the productivity of the
802
reference by only considering the CO2 absorber and desorber.
803
The process performance of the VPSA in comparison to absorption based CO2 capture
804
is illustrated in gure 15. Note that the productivity is given per equipment volume and has
805
been calculated for the adsorption processes with the bed density as reported in table 1. The
806
energy required when using MDEA depends on the target capture rate and can be reduced
41
ACS Paragon Plus Environment
Industrial & Engineering Chemistry Research
1800 1600
exergy [kJ/kgCO2 ]
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 42 of 50
cycle B w/ scheduling
1400 1200 1000
cycle D w/ scheduling
800 600 400 200 40
aMDEA
80
Natural gas SMR: 300 - 600 kJ/kgCO2 Partial oxydation or coal gasification: 200 - 400 kJ/kgCO2
120
160
3
200
240
productivity [kgCO2/(m h)] Figure 15: Comparison of the process performance for VPSA cycles B and D w/ scheduling, and absorption based carbon capture, including physical and chemical solvents. The energy consumption for state of the art processes depends on the syngas production route, and is typically in the range of 300-600 kJ/kgCO2 for natural gas SMR, e.g. with MDEA, and 200400 kJ/kgCO2 for partial oxidation or coal gasication, e.g. with Selexol, Purisol, Rectisol. A more specic area is highlighted for aMDEA, the state of the art for SMR processes, using the following references: (i) Shell QUEST project via the Alberta's CCS Knowledge Sharing Program: 27 ex = 414 kJ/kgCO2 without heat integration, ex = 325 kJ/kgCO2 with heat integration, and P re = 71 kgCO2 /m3 /h; (ii) IEAGHG reports on hydrogen production 4,28 ex = 318 kJ/kgCO2 , and P re = 86 kgCO2 /m3 /h; (iii) Romano et al. 29 ex = 317 kJ/kgCO2 . For the exergy calculation, the reboiler duty has been multiplied with the Carnot factor using the reboiler steam temperature, which is provided in the references. For the productivity, only absorbers and stripper have been considered. Constraints for VPSA: rH2 ≥ 90 %, ΦH2 ≥ 95 %, rCO2 ≥ 90 %, ΦCO2 ≥ 95 %. 807
with advanced owsheet congurations. The typical exergy consumption is in the range of
808
300 - 600 kJ/kgCO2 , including the thermal energy required for the reboiler and the electricity
809
consumption for the pumps, with productivities below 100 kgCO2 /(m3 h). 4,2830 The energy
810
required when using physical solvents, which typically feature higher syngas pressure and/or
811
higher concentration of CO2 in the feed, is around 200 - 400 kJ/kg CO2 . 31
812
When comparing absorption based capture processes with the developed VPSA cycles,
813
the gure shows that only cycle D can reach the upper range of the reported energy con-
42
ACS Paragon Plus Environment
Page 43 of 50 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Industrial & Engineering Chemistry Research
814
sumptions for MDEA, whereas the energy consumption for cycle B is signicantly higher.
815
However, the productivity for both cycles is higher than the values reported for MDEA. In
816
addition, the use of electricity instead of steam reduces the process complexity signicantly.
817
Therefore, in our opinion cycle D presents a promising alternative to absorption based CO2
818
capture.
819
In addition to separating CO2 , the VPSA also puries hydrogen up to high purities, thereby
820
combining two separation tasks in one unit. When using absorption based CO2 capture sys-
821
tems, an additional PSA unit after absorption is required for hydrogen purication (compare
822
gure 1). Because the inlet stream is already available at high pressure, no additional energy
823
is required for the PSA.
824
By eliminating a whole separation stage while having a similar separation performance as
825
the reference system (MDEA capture + PSA), VPSA cycle D allows for i) a reduction in
826
complexity and ii) an increase in productivity compared to the state of the art while featuring
827
an energy consumption comparable to the upper end of values reported for MDEA capture.
828
Both features might lead to a signicant reduction in capital cost compared to MDEA +
829
PSA.
830
5
831
In this article, we have presented the development and assessment of new adsorption cycles
832
for co-production of CO2 and H2 from a ternary feed with signicant amount of impurity.
833
Four dierent VPSA cycles have been developed and optimized. The main dierences be-
834
tween the cycles are (i) how the hydrogen-rich outow is recycled either via a compressor
835
or a sequence of pressure equalization (PE) steps , and (ii) how the bed is purged after the
836
CO2 step, i.e. introducing a hydrogen light purge (LP) step or not. While the LP step is
837
favourable for the CO2 separation performance, using a compressor increases the H2 separa-
838
tion performance.
Conclusion
43
ACS Paragon Plus Environment
Industrial & Engineering Chemistry Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
839
For very high purities of both products, a cycle with compressor and LP step is therefore
840
the most promising, namely cycle B.
841
For the targeted co-production of high purity H2 and CO2 (minimum 95 % purity) with high
842
recoveries for both products (minimum 90 % recovery), however, PE steps are sucient.
843
When including the evacuation pressure in the optimization, all cycles can achieve this sep-
844
aration target. Adding a LP step, however, decreases the energy consumption signicantly
845
because the separation can be achieved at higher evacuation pressures. In addition, exchang-
846
ing the compressor with PE steps reduces the energy consumption even further. Therefore,
847
for the given separation target, a cycle with PE steps and a LP is the most promising option
848
to minimize the energy consumption, namely cycle D.
849
When scheduling into a continuous process, the dierent cycles require a minimum of 7 to 9
850
columns and have a similar productivity.
851
Comparing to state of the art CO2 capture processes in hydrogen production plants (con-
852
sidering also public data of existing processes) reveals two interesting aspects. First, the
853
volumetric productivity of the presented VPSA cycles is twofold that of the state of the
854
art. Second, the exergy consumption of cycle D is within the range of reported absorption
855
processes. Whereas the state of the art adopts two separation stages, one for CO2 capture
856
and another one for H2 purication, the VPSA cycles integrate hydrogen purication and
857
CO2 separation within a single separation process, which makes them promising for process
858
intensication. Overall, cycle D is especially interesting for further development due to the
859
competitive exergy consumption in addition to the high productivity. Tangible options exist
860
to further improve its performance.
861
In an eort to keep our results general, we used a commercial activated carbon as sorbent,
862
but we do not expect it to be the optimal choice. Layering of dierent sorbents or the use
863
of novel sorbent materials, e.g. metal-organic-frameworks, with a higher cyclic capacity for
864
CO2 is expected to further reduce exergy consumption while also increasing productivity.
865
Whereas a very generic inlet stream with N2 as impurity has been chosen for cycle design 44
ACS Paragon Plus Environment
Page 44 of 50
Page 45 of 50 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Industrial & Engineering Chemistry Research
866
and development, the cycles proposed here are promising also for separations with dierent
867
feedstream compositions, as long as H2 and CO2 are the light and heavy component, respec-
868
tively. This includes feedstreams with CH4 or CO as main impurity or a mixture of dierent
869
impurities, i.e. N2 , CH4 and CO. With reference to the application of this technology to
870
steam methane reforming, which is beyond the scope of this manuscript and will be discussed
871
in a dedicated follow up paper, our preliminary results indicate that cycle D can successfully
872
purify hydrogen up to greater 99.99 % purity at around 90 % recovery whilst also separating
873
CO2 with CCS specications at an exergy penalty below 600 kJ/kgCO2 . With the choice
874
of an appropriate adsorbent or with a layering of dierent adsorbent materials, these cycles
875
could therefore enable a variety of separations relevant for hydrogen production coupled with
876
CCS.
877
References
878
(1) IPCC,
Global warming of 1.5°C. An IPCC Special Report on the impacts of global
879
warming of 1.5°C above pre-industrial levels and related global greenhouse gas emission
880
pathways, in the context of strengthening the global response to the threat of climate
881
change, sustainable development, and eorts to eradicate poverty ; 2018; In Press.
882
(2) Sircar, S.; Golden, T. C. Purication of Hydrogen by Pressure Swing Adsorption.
883
884 885
886 887
888
Sep-
aration Science and Technology 2000, 35, 667687. (3) Bui, M. et al. Carbon capture and storage (CCS): the way forward.
Energy Environ.
Sci. 2018, 11, 10621176. (4) IEAGHG,
Techno - Economic Evaluation of SMR Based Standalone (Merchant) Hy-
drogen Plant with CCS ; 2017. (5) Sircar, S.; Kratz, W. C. Simultaneous Production of Hydrogen and Carbon Dioxide
45
ACS Paragon Plus Environment
Industrial & Engineering Chemistry Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
889
from Steam Reformer O-Gas by Pressure Swing Adsorption.
890
Technology 1988, 23, 23972415.
891 892
Page 46 of 50
Separation Science and
(6) Baade, W. F.; Farnand, S.; Hutchison, R.; Welch, K. CO2 capture from SMRs: A demonstration project.
Hydrocarbon Processing 2012, 91, 6368.
893
(7) Palamara, J.; Guvelioglu, G.; Carney, S. Air products: success in advanced separation
894
and CO2 processing for EOR. Presented at the 19th annual CO2 ooding conference.
895
2013.
896 897
898 899
900
(8) U.S. Department of Energy, Clean Coal Research Program, Carbon Capture Technology
Program Plan ; 2013. (9) NETL,
Quality Guidelines for Energy System Studies: CO2 Impurity Design Parame-
ter ; 2013; DOE/NETL-341/011212, Rev 3. (10) de Visser, E.; Hendriks, C.; Barrio, M.; Mølnvik, M. J.; de Koeijer, G.; Liljemark, S.;
901
Gallo, Y. L. Dynamis CO2 quality recommendations.
902
house Gas Control 2008, 2, 478 484.
903 904
International Journal of Green-
(11) Streb, A.; Hefti, M.; Gazzani, M.; Mazzotti, M. A pressure swing adsorption process. 2019; application-nr: EP191723840.0.
905
(12) Schell, J.; Casas, N.; Marx, D.; Mazzotti, M. Precombustion CO2 Capture by Pressure
906
Swing Adsorption (PSA): Comparison of Laboratory PSA Experiments and Simula-
907
tions.
Industrial & Engineering Chemistry Research 2013, 52, 83118322.
908
(13) Marx, D.; Joss, L.; Hefti, M.; Gazzani, M.; Mazzotti, M. CO2 Capture from a Binary
909
CO2/N2 and a Ternary CO2/N2/H2 Mixture by PSA: Experiments and Predictions.
910
Industrial & Engineering Chemistry Research 2015, 54, 60356045.
46
ACS Paragon Plus Environment
Page 47 of 50 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
911
Industrial & Engineering Chemistry Research
(14) Marx, D.; Joss, L.; Hefti, M.; Mazzotti, M. Temperature Swing Adsorption for Post-
912
combustion CO2 Capture: Single- and Multicolumn Experiments and Simulations.
913
dustrial & Engineering Chemistry Research 2016, 55, 14011412.
914
(15) Joss, L.; Gazzani, M.; Mazzotti, M. Rational design of temperature swing adsorption
915
cycles for post-combustion CO2 capture. Chemical
916
394.
917
perature Swing Adsorption.
919
1554215555.
921
922
Industrial & Engineering Chemistry Research 2018, 57,
(17) Casas, N.; Schell, J.; Pini, R.; Mazzotti, M. Fixed bed adsorption of CO2/H2 mixtures on activated carbon: experiments and modeling.
Adsorption 2012, 18, 143161.
(18) Casas, N.; Schell, J.; Joss, L.; Mazzotti, M. A parametric study of a PSA process for
923
pre-combustion CO2 capture.
924
192.
925
Engineering Science 2017, 158, 381
(16) Hefti, M.; Mazzotti, M. Postcombustion CO2 Capture from Wet Flue Gas by Tem-
918
920
Separation and Purication Technology 2013, 104, 183
(19) Marx, D.; Joss, L.; Casas, N.; Schell, J.; Mazzotti, M. Prediction of non-isothermal
926
ternary gas-phase breakthrough experiments based on binary data.
927
20, 493510.
928 929
930
and N2 on activated carbon.
Adsorption 2012, 18, 4965.
(21) Hefti, M.; Marx, D.; Joss, L.; Mazzotti, M. Adsorption equilibrium of binary mixtures of carbon dioxide and nitrogen on zeolites ZSM-5 and 13X.
932
Materials 2015, 215, 215228.
934
Adsorption 2014,
(20) Schell, J.; Casas, N.; Pini, R.; Mazzotti, M. Pure and binary adsorption of CO2, H2,
931
933
In-
Microporous and Mesoporous
(22) Hefti, M.; Mazzotti, M. Modeling water vapor adsorption/desorption cycles. Adsorption
2014, 20, 359371. 47
ACS Paragon Plus Environment
Industrial & Engineering Chemistry Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 48 of 50
935
(23) Li, G.; Xiao, P.; Zhang, J.; Webley, P. A.; Xu, D. The role of water on postcombustion
936
CO2 capture by vacuum swing adsorption: Bed layering and purge to feed ratio. AIChE
937
Journal 2014, 60, 673689.
938
(24) Krishnamurthy, S.; Haghpanah, R.; Rajendran, A.; Farooq, S. Simulation and Opti-
939
mization of a Dual-Adsorbent, Two-Bed Vacuum Swing Adsorption Process for CO2
940
Capture from Wet Flue Gas.
941
1446214473.
Industrial & Engineering Chemistry Research 2014, 53,
942
(25) Krishnamurthy, S.; Rao, V. R.; Guntuka, S.; Sharratt, P.; Haghpanah, R.; Rajen-
943
dran, A.; Amanullah, M.; Karimi, I. A.; Farooq, S. CO2 capture from dry ue gas by
944
vacuum swing adsorption: A pilot plant study.
945
AIChE Journal 2014, 60, 18301842.
(26) Capra, F.; Gazzani, M.; Joss, L.; Mazzotti, M.; Martelli, E. MO-MCS, a Derivative-Free
946
Algorithm for the Multiobjective Optimization of Adsorption Processes.
947
Engineering Chemistry Research 2018, 57, 99779993.
948 949
950 951
952
(27) Alberta Energy. https://www.energy.alberta.ca/AU/Publications/KnowledgeSP/
Pages/KnowledgeSR.aspx, (accessed Mar 04, 2019). (28) IEAGHG, Techno-Economic
Evaluation of HYCO Plant Integrated to Ammonia / Urea
or Methanol Production with CCS ; 2017. (29) Romano, M. C.; Chiesa, P.; Lozza, G. Pre-combustion CO2 capture from natural gas
953
power plants, with ATR and MDEA processes.
954
Gas Control 2010, 4, 785 797.
955 956
957 958
Industrial &
International Journal of Greenhouse
(30) Meissner, R. I.; Wagner, U. Low-energy process recovers CO2.
Oil Gas J. 1983, 81,
5558. (31) Burr, B. L.; Lyddon, L. A comparison of physical solvents for acid gas removal. 2008; 87th Annual GPA Convention. 48
ACS Paragon Plus Environment
Page 49 of 50 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Industrial & Engineering Chemistry Research
959
Acknowledgement
960
The authors thank Dr. Lisa Joss (School of Chemical Engineering & Analytical Science, The
961
University of Manchester) for the many fruitful discussions during the preliminary phase of
962
this work.
963
ACT ELEGANCY, Project No 271498, has received funding from DETEC (CH), BMWi
964
(DE), RVO (NL), Gassnova (NO), BEIS (UK), Gassco, Equinor and Total, and is cofunded
965
by the European Commission under the Horizon 2020 programme, ACT Grant Agreement
966
No 691712.
967
This project is supported by the pilot and demonstration programme of the Swiss Federal
968
Oce of Energy (SFOE).
969
Supporting Information Available
970 971
model equations: mass, energy and momentum balance equations, EOS, correlation for mass transfer
972
isotherm parameters
973
calculation of the energy consumption for evacuating the column
974
scheduling equations
975
study of inuence of decision variables on separation performance
976
internal column proles for representative simulations of cycle A and cycle B
977
contribution of vacuum pump and compressor to the total energy consumption
49
ACS Paragon Plus Environment
Industrial & Engineering Chemistry Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
978
Graphical TOC Entry
979
50
ACS Paragon Plus Environment
Page 50 of 50