Modelling Pressure Swing Adsorption

to Recover Hydrogen from Refinery

Off-Gases/Steam Reformer Feeds in
Comparison to Cryogenic Technique
(CT)
Presented by :
Dr. Sultan Al-Salem
Petroleum Research Center
Kuwait Institute for Scientific Research

Agenda
Introductory Note:
Program Development within the Petroleum Research Center

Motivation & Objectives

Overall Plan & Implementation

Problem Statement: How to devise a mathematical solution!!

Methods & Implementation
Overview & System Studied
Model Implementation
PSA System first and second trials
Cryogenic separation
Past experiences with MT separation

Recommendations

Motivation: Message of PRC-KISR

Within the Petroleum Research Center (PRC) of KISR, the RCEF program is set to meet the
following challenge in the petroleum industry:
To support the oil industry in developing new products, new technologies and extending
the throughput and flexibility of petroleum refining activities

Solution Areas
Improving the operability of selected refinery
processes

1. Utilizing ROGs

Database of the molecular composition of refinery

streams

2. Hydrogen Management
3. MT Implementation

4. CCS
5. Energy Efficiency Solns.
6. LESP

Utilization of refinery waste streams

Background

Refinery off-gases (ROGs) are by-product gases produced in an oil refinery

typically

used as a fuel gas or as a stand-alone product.

Table 1: Main Components in the ROGs

ROGs encompass somewhat high HVs.

Hydrogen gas enrichment and separation is

Typically undertaken via three main techniques:

Component

Mole%

Component

Mole%

Hydrogen

10 50

Ethane

5 20

Hydrogen Sulfide

0.1 3

Propane

16

Nitrogen

2 10

Butane

0.5 - 1

Methane

30 55

Pentane +

0.2 - 2

1) Pressure Swing Adsorption (PSA)

2) Membrane Separation; &
3)

Cryogenic (cryo-system) separation.

Hydrogen is extensively used in HDT processes for increasing the hydrogen to carbon (H/C)
ratio of products (hydrocracking) and for the hydropurification treatments, amongst other
processes crucial in Kuwait.

Problem Statement

Hydrogen gas is the main utility of refineries considered heart and sole, hence utilizing every
stream of it is essential for refining flexibility and expansion.

ROGs contain a hefty proportion of hydrogen, hence trying to separate this fraction is of
paramount importance.

Reliance on heavier crudes nowadays means optimizing the carbon emissions from
process, whilst using more hydrogen.

Hydrogen production processes intensify the carbon emissions. HENCE, need for optimal
refinery studies that will utilize every fraction of H2 gas available.

Aim of the Work

In this work, a ROG stream is mathematically modeled with the aim of recovering hydrogen
gas using a PSA unit and CT system.

Sensitivity analysis shows the strong points.

The Studied Case:

Focus on the H2 recovery unit
Unit objective:
Recover high purity H2 from the desulfurized high pressure H2 rich gases obtained from the
refinery.
The unit operates an H2S absorber designed to handle 92 MMSFCD of high pressure sour
gas with a pressure swing adsorption (PSA) unit capable to handle 89.4 MMSFCD of
desulfurized high pressure H2 rich gas.
H2 rich gas from a number of units in the refinery are passed through a ADIP solution
absorber, in order to, remove H2S. After which, the gas is sent to the PSA unit.

The Studied Case:

Focus on the H2 recovery unit
Tail gas was modeled via the separation techniques.
Components of the Modeled gaseous stream.

Composition

Mole%

Hydrogen

29.13

Ammonia

9.48

Methane

50.2

Ethane

5.46

Propane

3.84

Butane

1.57

Nitrogen

0.32

Schematic flow diagram of hydrogen recovery unit (i.e.

PSA) in MAA showing tail gas processing in the
refinerys configuration.

H2 To Users

Product
Hydrogen

H2 Rich Sour Gas

H2S
Absorption
(ADIP)

Sweet Gas
Compression

P
S
A

Tail Gas
Compression

Refinery
H2
Configuration.

Recovery

To Fuel Gas

Unit

D (m)

Technique #1
- Pressure Swing Adsorption

i1

i9

PSA is a cyclic process based on the effect of

i8
pressure changes in fixed beds for selective
adsorption.

i7

L (m)

PSA beds are loaded with an adsorbent to separate

i6
i5
i4

the desired gas

Zeolite was considered in this work.

i3
i2
i1

The separation consists of four consecutive steps:

recompression, adsorption, desorption and blow
down.

Feed

yi (mol)

Technique #1
- Pressure Swing Adsorption - Freundlich isotherm
The following assumptions were made to develop mathematical expression:

Due to high concentration of H2 gas and methane (CH4) gas, the ROG was assumed to be a binary mixture
of H2 and CH4.

The mixture behaves as an ideal gas.

Axial pressure gradients and dispersion are negligible.

A linear driving force model was assumed for the fixed bed.

The process as a whole was assumed to be isothermal due to low adsorption heat.

A total cycle time (tend) of 600 s was assumed, and a bed length (L) and diameter of 9 m and 0.5 m were
selected, respectively, simulating industrial conditions.

Technique #1 - Pressure Swing Adsorption - Freundlich isotherm

Starting with the equilibrium equation for fixed beds:

qi

Where qi is amount adsorbed in the solid phase (kg H2/kg

adsorbent), t is dimensionless time (ti/tend), z is the
dimensionless axial distance (zj/L) and L is the length of
bed (m).
A, B and C are constants and are defined as follows:

ut
A end
L

The superficial gas velocity (u, m/s) and isotherm equation are as follows:

Two boundary conditions were defined to solve the model for

time (i) and position (j).
times

Dz is the difference in the axial distance corresponding to each of the

twenty isotherms and is equal to 0.5.
The successive concentration values are evaluated using explicit
method at new position as function of all known concentrations at
previous time step according to the following expression.

yi , j 1 yi , j

KDz y
A

m
i 1, j

m
i, j

Dz

B yi, j C
A

yi

z j

b tend kop(keq

avg AC

Byi C

kop yo tend

qi K y m

4
Adsorbed Hydrogen on Zeolite (wt%)

Dt is the difference between the two dimensionless

corresponding to each isotherm and is equal to 0.1;

y = 3.5962x0.2391
R = 0.9698

3.5
y = 3.4509x0.2404
R = 0.9883

3
2.5

y = 3.1238x0.2594
R = 0.9838
2
1.5
CHA 298K

CHA 350K

CHA 273K

1
0

0.2

0.4

0.6

H2 mol fraction

0.8

1.2

Technique #1 - Pressure Swing Adsorption - Freundlich isotherm

Laplace transform solution.

Design parameters effect.

0.3

tow = 0.2

0.25

tow = 0.3

0.2
0.15
0.1
0.05
0

tow = 0.4

0.05
0.045

tow = 0.6
tow = 0.7
tow = 0.8
tow = 0.9
tow = 1

0.04
0.035

Adsorption Bed Length (m)

14

No. Transfer Unit

-0.05

8.5

0.05

0.005

0.050.15

7.5

0.01

0.15

y = 0.0045x - 0.0015
R = 0.9999

6.5

0.015

0.150.25

0.25

0.02

0.35

5.5

0.025

0.250.35

0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5

0.03

H2 Conc. (mol)

Non-recovered H2 fraction (mol)

tow = 0.5

0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5
5.5
6
6.5
7
7.5

Second trial

tow = 0.1

tow = 0

0.35

8.5

Recovery issues w/model

-0.050.05

Technique #1
- Pressure Swing Adsorption Langmuir isotherm
u

Ci , j 1

(f

b )
2

(D 4 )

C
t

uC
z

1 q

Where Dt is the

Dz
Ci , j
(Ci 1, j Ci , j )
u Dt
Ci , j
qo bDz 1 Ci 1, j

u Dt 1 bCi 1, j 1 bCi , j

q
bP

qo 1 b P
cycle time used at

position z, and Dz is the difference in

the axial distance corresponding to

each of the ten theoretical equivalent

height trays and is equal to 0.9.

100

with %R.
The rate of flow
entering the column

80
60

1 bar

5 bar

10 bar

20 bar

30 bar

40 bar

50 bar

40
20
0
0

100
90
80
70
60
50
40
30
20
10
0
0

9 10

Effect of number of trays and feed pressure on

hydrogen recovery.

Highest recovery is obtained for a column length of 9

meters at 50 bar.
Adsorption temperature immensely affects the hydrogen

recovery due to isotherm dependency on the temperature at

which this surface phenomena occurs.

Gas temperature (K) effect on

hydrogen recovery (%) for
ROG flow rate of 5800 kg/hr:
Feed pressure 50-10 bar

20

40

60

Inlet Feed Pressure (bar)

No. trays

was found to be of no

significance

Hydrogen recovery as a function of inlet ROG feed

pressure [D=0.5 m, F=5800 kg/hr].

Hydrogen Recovery (%)

results in increase

Hydrogen Recovery (%)

Increase in Pf

Hydrogen Recovery (%)

Technique #1 - Pressure Swing Adsorption Langmuir isotherm

100
95
90
85
80
75
70
65
60
55

50 bar

10 bar

200

400

600

800

ROG Stream Temp. (K)

1000

increase in PSA, recovery

of

hydrogen

gas

is

increasing due to increase

in residence time with the

zeolite bed.

Increasing the number of

theoretical trays above ten

Adsorption cycle time (s)

As the number of trays

10 bar

30 bar

0.7

50 bar

600

0.6

500

0.5

400

0.4

300

0.3

200

0.2

100

0.1

0
0

20

40
60
Hydrogen recovery (%)

80

100

Final hydrogen concentration (mol/L)

700

doesnt affect the model

results significantly.

Effect on hydrogen recovery (%) by cycle time (s) and final hydrogen concentration (Cf).
Note: Same symbols indicate similar pressure (bar) for both axis.

Inlet feed pressures between 40 and 50 bar show that a plateau of recovery is reached.

However, final concentration of hydrogen existing the column is inter-related with both the
pressure (due to dependency on isotherm type) and the length of the column (see Fig).

Technique #1
- Pressure Swing Adsorption Remarks

The ROG feed pressure, which one might argue as the main influencing parameter

in such units, showed highest recovery of hydrogen (95-96%) at pressure between

40 to 50 bar.

This agreeable with the conditions in this work since the ROG stream originates
from a compression stage to a fired heater that is of similar conditions.

Nonetheless, results obtained by a past study were validated with high accuracy in
this work.

Technique #2
- Cryogenic Separation First stage compression
in

u
g z in
H

m
2 gc gc

out

u
gz
out
H

m Q WEC WS 0
2 gc gc

V dP C

W DH

nRT
P dP n CP dT

dT

R 2 ln( P2 / P1 ) A / T B CT DT 3 dT
A ln(T2 / T1 ) ( B)(T2 T1 ) (C / 2)(T22 T12 ) D / 2(T22 T12 )
N

O.F . Min
i 1

yi

A ln(T

100

/ T1 ) ( B)(T2 T1 ) (C / 2)(T22 T12 ) D / 2(T22 T12 ) LHS

Technique #2 - Cryogenic Separation Second stage Heat Exchanger

Applying the Antoine equation

B
log( Pi ) A
C Ti
Third stage Separation Valve

B
OF Ppi Min A
C Ti

z F x L yV
Objective function of (1 -

x / i y )
N

A three cycle process was estimated that operate with the following temperatures after solving
the problems at hand (oC): 70 (10 bar), 25 (8 bar) and -30 (5 bar).

Comparative assessment
For PSA: 5800 kg/hr, U = 14 m/s, P = 1 bar: %R = 12%
Recovery in PSA stretches to 96% at normal OC.
Optimizing CT delivers a 24% recovery of hydrogen gas.
Individual recovery of each component by the three cycles in the cryogenic technique (100 mole basis).

10 bar
CH4
C2H6
C3H8
C4H10
NH3
N2
H2

L
8.35
1.51
1.49
0.80
3.57
0.04
4.15

V
41.84
3.94
2.34
0.76
5.90
0.27
24.97

8 bar
L
1.94
0.38
0.38
0.20
1.00
0
0.84

V
39.90
3.55
1.95
0.56
4.89
0.27
24.12

5 bar
L
0.86
0.21
0.23
0.13
0.71
0
0.29

V
39.03
3.34
1.72
0.43
4.18
0.26
23.83

Past experience with membrane separation

Membranes work on the principle of selective separation of the polymer material.
Smaller molecules (hydrogen gas) are separated from lager ones, due to their high permeability.

The partial differential pressure across the membrane module is the driving force that such
operation work on.
Permeate side (Lp)

The following assumption are made:

1.
2.
3.

The polymer of choice is polyvinyl (tri-methyl) silane.

A perfect mixing model is assumed.
A multi-component mixture is assumed.
PF is the feed pressure (Pa),

Feed (Lf)

Pp is the permeate pressure (Pa),

I is the membrane thickness (m),

Zi

PF

Ai is the membrane surface area (m2),

L f I

P
Q A l F p (l )
i i

Qi is the permeability coefficient of the ith component (kmol (m/hr) m2 Pa),

Lf is the feed flow rate (kmol/h) on Feed (Lf)
Permeate side (Lp); and
is the dimensionless stage cut (Lp/LF).

Past experience with membrane

separation
Membrane model applied resulted in a 67% recovery [1 bar,
5800 kg/h].

H2 mol Fraction in Permeate

0.3
y = 0.0045x + 0.0097
R = 0.9961

0.25

0.2
0.15
0.1

0.05
0
0

By comparison it is much more economical and better in

20

40

60

Perm. Coff. (Barrer)

performance than other two.

Continuation of the work: Validate the membrane model

for the refinery case at hand; SA on CT.

H2 mol Fraction in Permeate

But in practice can it perform under nominal conditions.

1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0

y = 0.0892x + 0.0006
R = 0.9999

Feed Press. (Bar)

10

15

Concluding remarks:

As far as work objectives, these were met to the full understanding and representation of the
case at hand.

PSA was modeled, as well as, CT to estimate amount of hydrogen recovery from a ROG
stream.

PSA presented a better choice for recovery, that could be manipulated with varying OCs.

Past experiences show that MT show strong point?!

The economics of all systems must be considered at later stages, incorporating the
economy of scale to understand fully benefits of each technique.