Optimising adsorption process design

and operation
Mayank Patel Applications Engineer

2014 Process Systems Enterprise Limited

 Agenda

Adsorption technologies
Types of systems
Considerations

Modelling of adsorption beds

Modelling of PSA processes

Explicitly
Computationally efficient approach

Conclusions

2014 Process Systems Enterprise Limited

 Adsorption technologies

2014 Process Systems Enterprise Limited

 Adsorption technologies
Overview

Gas-phase processes
(PSA, VSA, TSA)
Hydrogen purification
Air separation: O2 or N2 enrichment
CO2 capture
Recovery of gasoline vapours from air
Ethanol dehydration

Liquid-phase processes
(SMB, chromatography)
UOP Sorbex processes
e.g. separation of mixed aromatic C8
isomers (ParexTM /EbexTM/MX SorbexTM)

2014 Process Systems Enterprise Limited

 Modelling of adsorption processes
Main considerations/challenges

Spatially varying properties

Axial & radial variations in the adsorbent bed
Adsorbed species diffuse through pores in adsorbent
Inherently dynamic processes
Complex operating procedures (cycle schedules)
Major discontinuities (e.g. flow reversals)
Process improvement
Optimisation of equipment design & operating procedure
Usually meaningful only at cyclic steady state

gPROMS already a leading modelling tool for adsorption R&D

e.g. Google Scholar search for swing adsorption gPROMS ~300 hits

2014 Process Systems Enterprise Limited

 Modelling of adsorption beds

2014 Process Systems Enterprise Limited

 Modelling of adsorption beds

Adsorption
heat generation

Bed-to-wall &
wall-to-ambient
heat transfer

Intra-pore
Convective/dispersive
diffusion
flow in packed bed

Bulk-to-solid surface
Momentum losses
mass transfer
pressure drop

Non-equilibrium mass & heat

transfer between fluid & solid

2014 Process Systems Enterprise Limited

 Modelling of adsorption beds
in gPROMS ProcessBuilder

Adsorption
heat generation

Bed-to-wall & HTC based on flow

wall-to-ambient conditions
heat transfer
1D-flow; dispersion
Intra-pore
diffusion Convective/dispersive
flow in packed bed
coefficients via
correlations
Linear Driving
Bulk-to-solid surface
mass transfer
Force (LDF) Momentum losses
model pressure drop

Non-equilibrium mass & heat

transfer between fluid & solid
Adsorption isotherms Multi-layer bed
(T,P)-dependent
mass transfer coefficients
Extended Langmuir modelling
Langmuir-Freundlich
IAST
user-defined (any)
2014 Process Systems Enterprise Limited
 Pressure Swing Adsorption (PSA)

2014 Process Systems Enterprise Limited

 Example
Production of hydrogen from natural gas
Hydrogen produced from catalytic reforming of natural gas
CH4 + H2O 3H2 + CO
combined with water gas shift reaction: CO + H2O H2 + CO2

High-purity hydrogen
obtained via pressure
swing adsorption (PSA)

High temperature Purification unit

conversion
Heat exchanger Impurities:
carbon dioxide
Feed pre-treatment methane
carbon monoxide
nitrogen
Reforming and steam generation
water vapour
Source: http://www.global-hydrogen-bus-platform.com/Technology/HydrogenProduction/reforming

2014 Process Systems Enterprise Limited

 PSA process for hydrogen purification
A.M. Ribeiro, C.A. Grande, F.V.S. Lopes, J.M. Loureiro, A.E. Rodrigues Chem. Eng. Sci. 63 5258-5273 (2008)

Feed molar fractions

H2: 73.3 | CO2: 16.6 | CH4: 3.5 | CO: 2.9 | N2: 3.7
2-layer beds
50% activated carbon CO2 + CH4 + H2O
50% zeolite CO + N2
Adsorption isotherm: Multisite Langmuir model
4-bed, 12-step process
Steps: 1 2 3 4 5 6 7 8 9 10 11 12

Bed 1:
Bed 2:
Bed 3:
Bed 4:

2014 Process Systems Enterprise Limited

 PSA process for hydrogen purification
A.M. Ribeiro, C.A. Grande, F.V.S. Lopes, J.M. Loureiro, A.E. Rodrigues Chem. Eng. Sci. 63 5258-5273 (2008)

Bed 1 Bed 2 Bed 3 Bed 4

Steps: 1 2 3 4 5 6 7 8 9 10 11 12

Bed 1:
Bed 2:
Bed 3:
Bed 4:

2014 Process Systems Enterprise Limited

 Modelling of PSA processes
I. Single-bed modelling & simulation

Project tree

Palette of
library units

Workspace

2014 Process Systems Enterprise Limited

 Modelling of PSA processes
I. Single-bed modelling & simulation

Multisite Langmuir model

2014 Process Systems Enterprise Limited

 Modelling of PSA processes
I. Single-bed modelling & simulation

Breakthrough curves
100

Molar concentration [mol %]

80

60

40

20

0
1.00E+01 1.00E+02 1.00E+03 1.00E+04
Time [s]

Hydrogen Methane Carbon Dioxide Carbon Monoxide Nitrogen

Axial temperature Axial molar fraction

302.0 100

80

Molar fraction [mol %]

301.0
Temperature [K]

60

40
300.0

20

299.0 0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

Axial [-] Axial [-]

Methane Carbon Dioxide Carbon Monoxide Nitrogen Hydrogen

2014 Process Systems Enterprise Limited
 Modelling of PSA processes
II. Multi-bed PSA process modelling & simulation

2014 Process Systems Enterprise Limited

 Modelling of PSA processes
II. Multi-bed PSA process modelling & simulation

2014 Process Systems Enterprise Limited

 Modelling of PSA processes
II. Multi-bed PSA process modelling & simulation

Auxiliary units, i.e.

gas headers, valves

Material
source
Calculate for
recovery

Scheduler

Key results
graphics

2014 Process Systems Enterprise Limited

 Modelling of PSA processes
II. Multi-bed PSA process modelling & simulation

12-steps per cycle

Steps: 1 2 3 4 5 6 7 8 9 10 11 12

Bed 1:
Bed 2:
Bed 3:
Bed 4:

Cycle time = 160 s Equates to each other

2014 Process Systems Enterprise Limited

 Modelling of PSA processes
II. Multi-bed PSA process modelling & simulation

Bed 1 pressure profile at product end Hydrogen purity at product end

Simulation of 100 cycles 13730 s (CPU time)

Getting to cyclic steady state is computationally expensive

Potential improvements?
2014 Process Systems Enterprise Limited
 Modelling of PSA processes
III. Self-interacting Bed (SiB) for CSS computation

All beds reach exactly the same cyclic steady state (CSS)
it should be possible to compute the CSS
by explicitly modelling only one of these beds

Bed-bed interactions: At CSS, material entering bed A

from bed B during a step in the cycle is identical to
material leaving bed A during a different step
Steps: 1 2 3 4 5 6 7 8 9 10 11 12

Bed 1:
Bed 2:
Bed 3:
Bed 4:

2014 Process Systems Enterprise Limited

 Modelling of PSA processes
III. Self-interacting Bed (SiB) for CSS computation

All beds reach exactly the same cyclic steady state (CSS)
it should be possible to compute the CSS
by explicitly modelling only one of these beds

Bed-bed interactions: At CSS, material entering bed A

from bed B during a step in the cycle is identical to
material leaving bed A during a different step
Steps: 1 2 3 4 5 6 7 8 9 10 11 12

Bed 1:
Bed 2:
Bed 3:
Bed 4:

similarly for downstream pressure seen by any

material leaving bed A
2014 Process Systems Enterprise Limited
 Modelling of PSA processes
III. Self-interacting Bed (SiB) for CSS computation

Product stream

Single bed Bed-to-bed

model connection

Calculate for
recovery
Scheduler

Feed stream

Purge stream

2014 Process Systems Enterprise Limited

 Modelling of PSA processes
III. Self-interacting Bed (SiB) for CSS computation

12-steps per cycle

Steps: 1 2 3 4 5 6 7 8 9 10 11 12

Bed 1:
Bed 2:
Bed 3:
Bed 4:

PSA cycle logic

Efficient mechanism for

storing key variable trajectories
during a cycle
retrieving necessary variable
values during the same or later cycles
2014 Process Systems Enterprise Limited
 Modelling of PSA processes
III. Self-interacting bed Results

Bed 1 pressure profile at product end Hydrogen purity at product end

4-bed explicit model Self-interacting bed

model

Mismatch of purity results < CSS

Simulation of 100 cycles 2326 s (CPU time)

Compared to 13730 s (CPU time) for 4-bed explicit model
2014 Process Systems Enterprise Limited
 Modelling of PSA processes
Explicit vs. implicit representations of multi-bed process

Multiple beds Single self-interacting bed

Identical CSS

Use for studying detailed

PSA process dynamics Use for efficiently
start-up
determining CSS
effects of disturbances
control system design & tuning
2014 Process Systems Enterprise Limited
 Conclusions

2014 Process Systems Enterprise Limited

 Conclusions

Detailed modelling of
physics of adsorption bed
cycle schedules of periodic adsorption process
Model implemented within gPROMS ProcessBuilder
general tool for steady-state and dynamic process modelling
user-friendly drag-and-drop flowsheeting environment
allows integration of adsorption units with other units

Catalytic reforming of methane + water gas shift + PSA-based hydrogen purification Hybrid membrane/PSA process
for hydrogen purification
2014 Process Systems Enterprise Limited
 Conclusions

Detailed modelling of
physics of adsorption bed
cycle schedules of periodic adsorption process
Model implemented within gPROMS ProcessBuilder
general tool for steady-state and dynamic process modelling
user-friendly drag-and-drop flowsheeting environment
allows integration of adsorption units with other units
Brings power of gPROMS platform to adsorption processes
Custom modelling
customisation of adsorption isotherms,
mass & heat transfer coefficient correlations, etc.
Parameter estimation
estimation of mass transfer characteristics from breakthrough experiments
Optimisation of bed design parameters, operating conditions, cycle schedule
dynamic optimisation problem
efficient handling of cyclic steady state poses special problems
2014 Process Systems Enterprise Limited
 Thank you

Contact PSE at info@psenterprise.com

2014 Process Systems Enterprise Limited