Dynamic Simulation of Breakthrough Curves On Multilayered Beds For Hydrogen Purification From Syngas Mixtures
Dynamic Simulation of Breakthrough Curves On Multilayered Beds For Hydrogen Purification From Syngas Mixtures
Dynamic Simulation of Breakthrough Curves On Multilayered Beds For Hydrogen Purification From Syngas Mixtures
A publication of
This paper modeled the behavior of a double-layer adsorption column to process a current with a Syngas-like
composition. The modeling was performed on Aspen Adsorption software to evaluate its performance by
comparison with experimental data found in the literature. The simulation aims is to evaluate the modeled
system for potential use in PSA systems for hydrogen purification from a Syngas stream, with successive
simulations on different plant scales.
1. Introduction
Pressure Swing Adsorption (PSA) is an excellent gas mixture separation and purification method. Currently,
PSA technologies are used for hydrogen purification in more than 85% of global hydrogen production systems
(Tao et al., 2019).
PSA systems have grown in popularity over the last few decades, owing to their ease of use and low operating
costs. These systems' main applications are the recovery of high-purity hydrogen, methane, and carbon dioxide
and the recovery of nitrogen and oxygen (Linde Engineering).
One of the primary benefits of using PSA systems is their extreme simplicity, which results in process reliability
with relatively low investment and maintenance costs. Furthermore, these systems do not require many
resources, making them suitable for various applications. Because the gas purification technology is dry, it does
not require water and thus does not cause aqueous effluent problems. Moreover, no heat is needed for the
process. As a result, operating costs are low, with energy costs limited to the need for pressurized gas (Bauer
et al., 2013).
The flexibility of PSA cycles has been studied in the context of energy savings and in areas other than hydrogen
production (Šulc et al., 2021).
Currently, no hydrogen-selective adsorbents are available (Liemberger et al., 2017). The use of multiple
adsorbents in adsorption beds as layers is an important modality for current gas purification technologies.
In the design of a PSA system, the logic is to have each species adsorbed onto the specific solid to achieve an
adequate trade-off between favorable adsorption and easy desorption.
Instead of using multiple layers, a series of columns with different adsorbent beds could be used to create an
adsorption gradient (Chlendi et al., 1995).
These performance enhancements can also be pursued by developing better-performing adsorbents and
changing PSA system operating schemes and varying process parameters (Ribeiro et al., 2008).
In chemical process design, complex mathematical models are frequently used to calculate process streams'
chemical and physical properties. The use of specific simulation software provides a significant advantage for
analyzing existing processes, designing new processes, implementing control strategies, and comparing
different processes to determine the best solution as needed (Cimini et al., 2005).
The Aspen Adsorption software was used in this study to replicate experimental results found in the literature
for potential use in PSA systems for hydrogen purification from a Syngas stream.
Paper Received: 10 December 2022; Revised: 27 April 2023; Accepted: 14 May 2023
Please cite this article as: Cecchini F., Innocenzi V., Corradetti V.M., Prisciandaro M., 2023, Dynamic Simulation of Breakthrough Curves on
Multilayered Beds for Hydrogen Purification from Syngas Mixtures, Chemical Engineering Transactions, 100, 283-288
DOI:10.3303/CET23100048
284
2. Model description
Aspen Adsorption is a comprehensive simulator for adsorption process analysis, design, simulation, and
optimization. For example, this software can simulate ion exchange, liquid phase adsorption, and gas phase
adsorption processes. The Gas Cyclic Steady State (gCSS) calculation mode is also presented to simulate
cyclic processes such as PSA. The simulator supports the use of Aspen Properties components' physical and
thermodynamic properties; there is also a wide range of material and energy transport models. For the
simulation of complex processes such as adsorption, finite element solvers such as Comsol Multiphysics (Xiao
et al. 2016) or Fortran-based packages such as ODEPACK (Chou et al. 2013) are frequently used. Process
simulators such as gPROMS (Ribeiro et al. 2008) or Aspen Adsorption (Zhang et al. 2021) are used less
frequently. The literature mostly reports on cases where these software were used to optimize 2-bed PSA or
VPSA processes. Cases of simulation of breakthrough curves on dedicated software, such as Aspen
Adsorption, are more uncommon. As a result, Aspen Adsorption was chosen for the mathematical
implementation of the models described, mostly because it can interact with all AspenTech packages.
For the characteristics of the adsorbent materials and the experimental system to be simulated, the
modeling work refers to the article by Jee and coworkers (Jee et al., 2001). The characteristics of the adsorbents
and the adsorption bed are summarized in Table 1.
Table 2 displays the simulation's input stream characteristics and working data.
The Peng Robinson equation of state was chosen for the gas mixture under consideration. Aspen Properties
calculates the majority of gas phase properties using these equations of state.
For the material balance assumption, Eq. (1), the dispersion term is removed from the material balance, resulting
in plug flow with a zero dispersion coefficient (infinite Peclet number). The axial dispersion term can be ignored
to simplify the mathematical model, as demonstrated in other studies (Xiao et al. 2016).
Table 1 Characteristics of Adsorbents and Adsorption Beds (Jee et al. 2001; Xiao et al. 2016).
Adsorbent Activated carbon Zeolite 5A
The temperature of the process is greatly influenced by the adsorption process. The temperature, in turn, affects
the entire adsorption process. The general form of energy balance is described in Eq. (2):
∂2 𝑇𝑔 ∂𝑇 ∂𝑇𝑔 ∂𝑢𝑔 4ℎ𝑖
−𝑘𝑔 ϵ𝑖 + 𝐶𝑣𝑔 𝑢𝑧 ρ𝑔 + ϵ𝑏 𝐶𝑣𝑔 ρ𝑔 +𝑃 + ℎ𝑖 𝑎𝑝 (𝑇𝑔 − 𝑇𝑠 ) + (𝑇 − 𝑇𝑊 ) (2)
∂𝑧 2 ∂𝑧 ∂𝑡 ∂𝑧 𝐷𝐵 𝑔
285
𝑳𝑺𝑻𝑷
Gas flow rate, F ( ) 8.6
𝒎𝒊𝒏
𝑊
Gas thermal conductivity, 𝑘𝑠 ( ) 0.084
𝑚𝐾
The heat transfer to the environment, Eq. (3), was considered rigorous, with the following energy balances:
∂2 𝑇𝑊 ∂𝑇𝑤 4𝐷𝐵 4(𝐷𝐵 + 𝑊𝑇 )2
−𝑘𝑊 + ρ 𝑊 𝐶𝑝𝑤 − ℎ 𝑖 (𝑇𝑔 − 𝑇𝑊 ) + ℎ 𝑜 (𝑇 − 𝑇𝑎𝑚𝑏 ) = 0 (3)
∂𝑧 2 ∂𝑡 (𝐷𝐵 + 𝑊𝑇 )2 − 𝐷𝐵2 (𝐷𝐵 + 𝑊𝑇 )2 − 𝐷𝐵2 𝑊
Since it was assumed that the adsorption was non-isothermal, the temperature of the adsorbent is defined as
reported in Eq. (4) (Wood et al., 2018):
𝑛
𝜕𝑇𝑠 ∂𝑞𝑖
𝜌𝑠 𝐶𝑝𝑠 + 𝜌𝑠 ∑ 𝑄𝑖 + ℎ𝑖 𝑎𝑝 (𝑇𝑔 − 𝑇𝑠 ) = 0 (4)
𝜕𝑡 ∂𝑡
𝑖
For momentum balance, the Ergun equation was chosen, Eq. (5), which combines the descriptions of pressure
drops by the Karman-Kozeny equation for laminar flow and the Burke-Plummer equation for turbulent flow.
∂𝑃 1.5𝑥10−3 μ𝑔 (1 − ϵ𝑖 )2 (1 − ϵ𝑖 ) 2
= −( 2 + 1.75𝑥10−5 𝑀𝑤 ρ𝑔 𝑢𝑧 ) (5)
∂𝑧 (2𝑟𝑝 ψ) ϵ3 2𝑟𝑝 ψϵ3𝑖
𝑖
The linear driving force (LDF) model with a single lumped mass transfer parameter, Eq. (6), describes the
sorption rate into the adsorbent.
∂𝑞𝑖
= ω𝑖 (𝑞𝑖∗ − 𝑞𝑖 ) (6)
∂𝑡
According to the literature, the optimal activated carbon/zeolite ratio for the adsorption of the mixture under
evaluation is 0.65 (Yang & Lee 1998).
Given the large number of compounds to be removed from the stream, each layer will remove one or more
contaminants based on affinity. Generally, an initial alumina or silica gel layer will adsorb moisture in typical
syngas or catalytic reforming gas compositions. Then, along with CO 2, a layer of activated carbon will adsorb
methane and any heavy hydrocarbons. On the other hand, the final layer will be zeolite, due to its superior CO
and N2 adsorption capabilities. CO2 and H2O are strongly adsorbed onto zeolite, making desorption difficult. As
the work cycles progress, these components accumulate in the zeolite (Ribeiro et al., 2008).
The Extended Langmuir-Freundlich experimental isotherm, Eq.(7), predicted multicomponent adsorption
equilibrium:
𝑞𝑚𝑖 𝐵𝑖 𝑃𝑛𝑖
𝑖
𝑞𝑖 = 𝑛𝑗 (7)
1 + ∑𝑛𝑗=1 𝐵𝑗 𝑃𝑗
𝑘4 𝑘6
Where: 𝑞𝑚 = 𝑘1 + 𝑘2 𝑇, 𝐵 = 𝑘3 𝑒𝑥𝑝 e 𝑛𝑖 = 𝑘5 + .
𝑇 𝑇
Table 3 summarizes the isotherm parameters and the LDF coefficients (Jee et al., 2001).
Because the isotherm used in the simulations is not available in the Gas Dynamic simulation model, it was
created using the Aspen Custom Modeler code.
286
3. Results
The described model simulates the behavior of a laboratory-scale adsorption column using data taken from the
literature. The simulation input data concern the input stream's composition, flow rate, temperature, and
pressure. The outputs compared to experimental data reported by Jee and coworkers (Jee et al., 2001) concern
breakthrough curves, concentration trends, and temperature trends in the bed.
Table 3: Extended Langmuir-Freundlich Parameters and LDF Coefficient for Activated Carbon and Zeolite 5A
(Zhang et al. 2021).
𝑘1 𝑘2 𝑥102 𝑘3 𝑥104 𝑘4 𝑘5 𝑘6 𝐿𝐷𝐹 𝑐𝑜𝑒𝑓𝑓. 𝑄
ω𝑖
𝑚𝑚𝑜𝑙 𝑚𝑚𝑜𝑙 1 (𝐾) (−) (𝐾) 1 𝑐𝑎𝑙
( ) ( ) ( ) ( ) ( )
𝑔 𝑔𝐾 𝑎𝑡𝑚 𝑠 𝑚𝑜𝑙
Activated carbon
𝐻2 16.943 -2.100 0.625 1229 0.980 43.03 0.700 2880
𝐶𝐻4 23.860 -5.621 34.780 1159 1.618 -248.9 0.195 4290
𝐶𝑂 33.850 -9.072 2.311 1751 3.053 -654.4 0.150 4300
𝑁2 1.644 -0.073 545.0 326 0.908 0.991 0.261 1660
𝐶𝑂2 28.797 -7.000 100.0 1030 0.999 -37.04 0.036 5240
Zeolite 5A
𝐻2 4.314 -1.060 25.15 458 0.986 43.03 0.700 2800
𝐶𝐻4 5.833 -1.192 6.507 1731 0.820 53.15 0.147 5400
𝐶𝑂 11.845 -3.130 202.0 763 3.823 -931.3 0.063 5300
𝑁2 4.813 -0.668 5.695 1531 0.842 -7.467 0.099 5470
𝐶𝑂2 10.030 -1.858 1.578 207 -5.648 2098.0 0.014 9330
Figure 2 shows the breakthrough curves for the gas mixture in the bed. N2 is the first component to pass through
the bed, followed by CO and CH4. CO2 is the last component to pass through the bed. Several roll-ups can be
seen here, which go to distort the H2 curve, confirming the phenomenon observed experimentally. It is also
noted that the obtained results are comparable to the experimental results.
Figure 3 (a) shows the profiles of component concentrations in the bed mixture of 100 cm length after 180
seconds of simulation, while Figure 3 (b) shows the same profiles after 300 seconds. The simulation results
match the experimental results in terms of trends. The concentration plot for t=300 s confirms the upstream
reports by showing concentration peaks for CO and N2 in the zeolite section of the bed, while CH4 and CO2 are
adsorbed by activated carbon in the first layer of the bed.Figure 3 (c) illustrates the temperature profiles in the
bed, 180 and 300 seconds after the simulation began, respectively. The experimental data confirm the presence
287
of a second peak in the temperature profile at 180 s in the bed layer responsible for N 2 and CO adsorption. It
has also been observed that the temperature profile tends to become more uniform over time.
1
0.8 CO2
C [kmol/kmol]
0.6
CO
0.4
H2
0.2
0
10 100 1000
t [s]
Figure 3: concentration profiles in the bed at t=180 s (a) and t=300 s (b); temperature profiles in the bed at 180
and 300 s (c).
4. Conclusions
In this paper, Aspen Adsorption software was used to simulate the behavior of breakthrough curves,
concentration, and temperature profiles within an adsorption bed. The results, which were confirmed by
experimental data, show how the mathematical models already included in the software can be used to predict
the behavior of experimental adsorption systems. The program's adaptability was also tested by including an
isotherm not found in the library of Aspen Adsorption's dynamic gas simulation mode. Future work will involve
the simulation of multibed PSA systems, first on a laboratory scale and then progressing to semi-industrial scale
simulation.
288
Nomenclature
L – bed length, m Cvg – Specific gas-phase heat capacity, cal/(mol
Lc – fin length, m K)
ki – isotherm parameters, - z – axial distance trough column, m
P – pressure, atm uz – superficial velocity of gas flow, m/s
Pi – partial pressure, atm DL – axial dispersion coefficient, m2/s
T – temperature, K g – gas density, g/cm3
Tg – gas temperature, K DB – bed diameter, m
Ts – solid temperature, K WT – wall thickness, m
Tw – wall temperature, K – spherical factor, -
qi – amount adsorbed, mol/g ap – surface area per volume unit, 1/m
qmi – equilibrium parameter for the Langmuir g – gas mixture viscosity, cP
Freundlich model, mol/g Mw – Molecular weight of the gaseous mixture,
Ci – molar concentration in the gas phase, kg/kmol
kmol/kmol
References
Bauer F., Hulteberg C., Persson T., Tamm D., 2013. Biogas upgrading-Review of commercial technologies
(Biogasuppgradering-Granskning av kommersiella tekniker) SGC Rapport 2013:270 ‘Catalyzing energygas
development for sustainable solutions’.
Chlendi M., Tondeur D., 1995. Dynamic behaviour of layered columns in pressure swing adsorption, vol. 9.
Chou C.T., Chen F.H., Huang Y.J., Yang H.S., 2013. Carbon Dioxide Capture and Hydrogen Purification from
Synthesis Gas by Pressure Swing Adsorption, Chemical Engineering Transactions, 32, 1855–1860.
Cimini S., Prisciandaro M., Barba D., 2005. Simulation of a waste incineration process with flue-gas cleaning
and heat recovery sections using Aspen Plus, Waste Management, 25(2), 171–175.
Jee J.G., Kim M.B., Lee C.H., 2001. Adsorption characteristics of hydrogen mixtures in a layered bed: Binary,
ternary, and five-component mixtures, Industrial and Engineering Chemistry Research, 40(3), 868–878.
Liemberger W., Groß M., Miltner M., Harasek M., 2017. Experimental analysis of membrane and pressure swing
adsorption (PSA) for the hydrogen separation from natural gas, Journal of Cleaner Production, 167, 896–
907.
Linde Engineering, no date, Hydrogen Recovery by Pressure Swing Adsorption.
Ribeiro A.M., Grande C.A., Lopes F.V.S., Loureiro J.M., Rodrigues A.E., 2008. A parametric study of layered
bed PSA for hydrogen purification, Chemical Engineering Science, 63(21), 5258–5273.
Šulc R., Ditl, P., 2021. The Potential of Energy Savings in Oxygen Production by Pressure Swing Adsorption,
Chemical Engineering Transactions, 86, 313–318.
Tao W., Ma S., Xiao J., Bénard P., Chahine R., 2019. Simulation and optimization for hydrogen purification
performance of vacuum pressure swing adsorption, Energy Procedia, vol. 158, 1917–1923, Elsevier Ltd.
Wood K.R., Yih A.L., Yueying Y., 2018. Design, simulation and optimization of adsorptive and chromatographic
separations: A hands-on approach.
Xiao J., Peng Y., Bénard P., Chahine R., 2016. Thermal effects on breakthrough curves of pressure swing
adsorption for hydrogen purification, International Journal of Hydrogen Energy, 41(19), 8236–8245.
Yang J., Lee C.H., 1998. Adsorption dynamics of a layered bed PSA for H2 recovery from coke oven gas, AIChE
Journal, 44(6), 1325–1334.
Zhang N., Bénard P., Chahine R., Yang T., Xiao J., 2021. Optimization of pressure swing adsorption for
hydrogen purification based on Box-Behnken design method, International Journal of Hydrogen Energy,
46(7), 5403–5417.