A novel adsorption process for co-production of ... - ACS Publications

Industrial & Engineering Chemistry Research. 1. 2. 3. 4. 5. 6. 7. 8 ..... biogenic sources, e.g. biogas, such technology will enable negative carbon e...
7 downloads 0 Views 2MB Size
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

ˆ no radial gradients in concentration, velocity and temperature

302

ˆ the Ergun equation is used to describe the pressure drop along the column

303

ˆ thermal equilibrium between the gas phase and the solid phase

304

ˆ 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

ˆ mass transfer is described using a linear driving force approximation, and a temperature/concentration independent mass transfer coecient

310

ˆ axial dispersion and conductivity are neglected 12

311

ˆ 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: ˆ the energy required by the vacuum pump (VP) for evacuating and purging the column at P < PAmb , EVP ˆ 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 ˆ 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

ˆ 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

ˆ which cycles and which among their features are favourable for reaching a good H2 separation performance; ˆ which cycles can achieve the desired coproduction of both products and at what energy penalty and productivity; ˆ 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

ˆ model equations: mass, energy and momentum balance equations, EOS, correlation for mass transfer

972

ˆ isotherm parameters

973

ˆ calculation of the energy consumption for evacuating the column

974

ˆ scheduling equations

975

ˆ study of inuence of decision variables on separation performance

976

ˆ internal column proles for representative simulations of cycle A and cycle B

977

ˆ 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