IOP Conference Series: Materials

Science and Engineering

PAPER • OPEN ACCESS You may also like

- Large eddy simulation of turbulent mixing
Performance Simulation for Gas Flow through a by using 3D decomposition method
Alibek Issakhov
Porous Media in Packed bed Columns using CFD - CFD Simulation of Velocity Distribution in
a River with a Bend Cross Section and a
Cubic Bed Roughness Shape
To cite this article: Asaad Salim Bded 2019 IOP Conf. Ser.: Mater. Sci. Eng. 579 012001 Hayder Q. Majeed and Ali M. Ghazal

- Computational Fluid Dynamics Modelling

and Experimental Study on Pressure drop
through Vertical Packed Clay Composite
Pellets
View the article online for updates and enhancements. C R Hiremath, P V Goggal, SV Hiremath
et al.

This content was downloaded from IP address 183.177.126.153 on 09/09/2024 at 13:42

1st International Conference on Petroleum Technology and Petrochemicals IOP Publishing
IOP Conf. Series: Materials Science and Engineering 579 (2019) 012001 doi:10.1088/1757-899X/579/1/012001

Performance Simulation for Gas Flow through a Porous Media

in Packed bed Columns using CFD

Asaad Salim Bded

Petroleum Engineering Department, Kerbala University, Karbala, Iraq

Corresponding email: [email protected]

Abstract. Two dimensions of Computational Fluid Dynamics simulation for gas flow in the
column with filling material as bed at low particle Reynolds numbers and high particle
Reynolds numbers have been presented. The simulation test has been done with the column
length 0.9 m. Porous media model was used to simulate the gas flow inside porous granular
media. Model of standard k-e for the flow turbulent range was used; the difference on the mesh
size used to study effect it on the performance of gas flow through porous media. The result of
flow rate of the gas was an important variable influencing bed and there are good results
between the CFD simulation of pressure drop and the predicted values from Ergun model, and
Computational Fluid Dynamics gave us simulation with high details for porous flow in packed
bed equipment.

1. Introduction
The flow through column filled with granular particles is of frequent occurrence in petroleum industry
and chemical process, therefore an expression are needed to predict the pressure drop within material
of beds due to the resistance caused by the filling material inside of column [1]. The flow at low
velocity and high velocity through the spaces between the particles have been retained by porous
medium to allow us obtaining the influence of some of parameters on the performance of flow inside
bed columns. Therefore this study carried out a computational fluid dynamics (CFD) based on the
standard k-e model investigation of flow through porous media ranging from laminar to turbulent flow.
CFD is relatively lower in cost than the physical testing, ability to simulate real conditions, many flows
and transfer processes cannot be easily tested [2]. By using a Computational Fluid Dynamics (CFD)
the studies were extended to treat various geometries with different boundary conditions, more
applications for flow studies in porous media have produced extensive basis information in published
literature, of which R.S.Maier & R.S. Bernard.[3]used the lattice Boltzmann method to the velocity
distribution under the spatial resolution in packing without gives the details of the packing to simulate
viscous fluid flow through a column of glass beads.
J. Gunnar I. Hellstr, & P. Jonsson [4] analysis of a quadratic array of cylinders shown that a number
of different Reynolds number values can be defined for porous media, CFD simulations for flow
through a quadratic array of cylinders at three solid fractions with a turbulent, found that turbulence
flow needs be considered when Rep is above 300, since the laminar setup of equations failed to predict
the experimental results over this value F.Augier&J.Y.Delenne[5] verified the transport and transfer
properties inside packed beds of particles by CFD simulations at low to moderate Reynolds numbers.
They found a problem of contacting points between particles, which is inherent to finite-volume
numerical methods and solved by applying a shrinking of 2% along of material of the bed when transfer
properties been estimated of increasing in arranging packing complexity. Kamyar M. & Ali A.&

Content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence. Any further distribution
of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
Published under licence by IOP Publishing Ltd 1
1st International Conference on Petroleum Technology and Petrochemicals IOP Publishing
IOP Conf. Series: Materials Science and Engineering 579 (2019) 012001 doi:10.1088/1757-899X/579/1/012001

Eckehard S.[6] used the porous media model as a method for simulation to ability the prediction of
CFD using as a tools design for such packed columns to flow in the reaction media, The results of
computational fluid dynamics simulation was excepted with results of the experimental in all cases
studies.
The present research uses the software ANSYS-Fluent to develop the porous media in column with
filling materials as bed to simulate of performance the flow of gas from the laminar to turbulent gas
flow through the filling materials (porous media) within the range of Rep ( 7.87 – 1115.87)
respectively, throughout change in axial temperature distributions along materials column.

2. Material Properties

The details of system are a gas phase without change in its density and uniform of Glass balls size.
The properties, both of the (air) and material (Glass balls), are tabulated in Tables (1) to (3). Based on
an experimental setup [5] and the thermal properties of glass balls are taken from Peter, S.T. [8], and
the gas properties based on the ANSYS database.

Table 1. Properties of gas phase at the temp. 293 K

Parameter / unit value
(air) density (kg/ m3) 1.225
(air) viscosity (kg/ m.
1.7894
s)
Heat capacity (j/kg. k) 1006.43
Coefficient of thermal
0.0242
conductivity (w/m. k)

Table 2. the Physical properties of column (Steel) from ANSYS database

Parameter value
density (kg/ m3) 8030
specific heat (j/ kg. k) 502.48
heat conductivity
16.27
coefficient (w/ m. k)

Table 3. Material properties of simulated sphere [7]

Parameter value
Particle diameter (m) 0.005
Density (kg/ m3) 2590
Heat capacity (j/ kg.k) 780
Heat conductivity
0.9656
coefficient (W/m.k)

3. Models of Porous Media

2
1st International Conference on Petroleum Technology and Petrochemicals IOP Publishing
IOP Conf. Series: Materials Science and Engineering 579 (2019) 012001 doi:10.1088/1757-899X/579/1/012001

Forchheimer suggested the flow with in porous medium and described his suggestion when the flow
and the pressure gradient have a linear relationship (linear Darcy’s law) as shown in below form [9]:
∆𝑝 𝜇𝑢
𝐿
= 𝑘 + 𝜌 𝑘𝐹 𝑢2 = 𝛼𝑢 + 𝛽𝑢2 (1)
Where 𝑘𝐹 is the inertial resistance (Forchheimer coefficient) at low Reynolds number, the equation
(1) considered (Darcy law) that the viscous forces be controlled in porous media and ∆p is proportional
to the Darcy velocity (inertial effects can be neglected), while as the Reynolds number increases, the
inertial forces are considered and equation (1) gives non-linear relationship between ∆p and Darcy
velocity, and at high Reynolds number, the inertial forces became strong, ∆p is proportional to the
squared velocity [10]
In the ANSYS-FLUENT the superficial phase velocities can be calculated depending on volumetric
flow rate in a porous media of the regime [11, 16]. Porous Media Model for homogeneous medium
consists of two portions when the momentum term is added; inertial loss and viscous resistances
portions. Will be mixed with other factors of Darcy’s equation to predict the pressure drop through the
porous zone inside the gas cell zone (fluid volume) in the axial direction, this is called a Porous Media
Model by cell zone conditions [11].
The concept of permeability was developed by Kozeny and Carman the following equation (2). It
depends on the porosity and glass sphere diameter and medium’s permeability also depends on physical
and geometric model characteristics.

𝜌𝑔 𝜀 3 𝑑
𝐾= (
5𝜇 (1−𝜀)2 6
)2 (2)
Where 𝜀 is the porosity of medium’s, It is calculated from equation (3) Furnas [12] is proposed an
equation for the porosity in bed column as function of particle and bed diameter, as in the following
𝑑
equation: 𝜀 = 0.375 + 0.34 (3)
𝐷
The coefficient 𝛽 is the inertial resistance in the porous media along the flow direction per unit length
at high velocities [10]. Values of equation (1) are obtained from physical regime by solution of Navier-
Stokes equation relating to velocity and pressure subject to corresponding boundary condition in CFD
simulation. In this paper, air was passed through the bed to simulate the flow and to estimate the pressure
drop depending on properties of the fluid, and porosity of medium. Ergun equation [13], is defined as
follows:
∆p (1− ε) ² (1− ε)
= 150 𝜇 3 𝑢 + 1.75ρ 3 𝑢2 (4)
L ε d² ε d∅
The pressure drop predicted from CFD simulation of bed columns has been compared with values
from equation (4).

4. Modelling Fluid Flow

The basis for all CFD fluid flow modeling is the Navier-Stokes equations. The term Navier-Stokes
equations are used to describe three equations; the momentum equation, the continuity equation, and the
energy equation.

5. Navier- Stokes Equation

The Navier Stokes equation is the equation of motion for a Newtonian fluid with constant viscosity
and density. The equation is greatly simplified when applied to 2D flow with the assumption that
velocity is only in the axial direction [11 14 17].

6. Standard k-e Turbulence Model

3
1st International Conference on Petroleum Technology and Petrochemicals IOP Publishing
IOP Conf. Series: Materials Science and Engineering 579 (2019) 012001 doi:10.1088/1757-899X/579/1/012001

At high Reynolds numbers, the viscous dissipation of kinetic energy (ԑ) will rise because of high
inertial forces, multiplied by the fluctuating vorticity and from Navier-Stokes equation it can be derived
the viscous dissipation at the exact transport fluctuating vorticity equation [10 14 17].

7. Preparation of media and design of column

The first step in CFD simulation of bed columns is Pre-processing, which has been done by ANSYS
Workbench 12.0 tool to form the geometric model with (hexahedral grids) of numerical density:
dividing 20 interval counts and 40000 quadrilateral cells (40k), 1600 2D wall faces, 40851 nodes, (40k)
computational cells as shown in Fig.(2).The geometric model was designed as two dimension (2D
geometry) on the (x, y) coordinates, with 0.041m in a diameter of column, 0.9 m length with four surface
boundaries: inlet (for inflow, t, u), outlet (for outflow, t, P,.etc), wall and porous medium. The bed
materials in this model of glass balls are packed into the column randomly as porous media. The model
assumption is shown in Fig. (1).

Figure 1. 2D schematic of the Bed column

Figure 2. 2D hexahedral meshing bed column.

8. Computational Fluid Dynamics (CFD) model using (ANSYS- FLUENT)

The method of (CFD) used to simulate gas flow in the assumed bed column by using the ANSYS-
FLUENT, the pre-processor, solver and post-processor are the basic parts of the ANSYS-FLUENT

4
1st International Conference on Petroleum Technology and Petrochemicals IOP Publishing
IOP Conf. Series: Materials Science and Engineering 579 (2019) 012001 doi:10.1088/1757-899X/579/1/012001

software to analyse the process. The pre-processor step is usually carried out through a graphical
interface and describe in section 3. The velocity of gas, and the data of the porous cell condition, density
of the gas and all coefficient values are entered to the solver software (user input values), to solve the
numerical equations based on the entered values and the results are obtained to provide the post-
processor step software [15]. In this work the single phase flow is initially in static condition inside the
bed column and the gas passes through bed from bottom of the column by velocity inlet at the central
of porous cell zone that is specified for the porous medium conditions, there are options related to
inertial loss and viscous resistances and axial direction vectors to be solved , void fraction has been
calculated from Equation (3), and the permeability has been calculated from Equation (2) for each of
the space coordinates. Flow Analysis by FLUENT through bed column using 2D model is solved in
steady state laminar and turbulent flow simulation. all models that described previously are available in
FLUENT software [15]. Select physical velocity under porous formulation, will take the volume
blockage into consideration and provides a more accurate representation of flow within bed column.
The boundary condition for laminar model and turbulent model flow will range from 0.023 to 3.23 as
inlet velocity and is used for inlet temperature equals 293ok (isothermal boundary condition), column
wall temperature be constant and equals to 373ok (heating wall), boundary condition for porous zone
are estimated in section 2. After the simulation is iterated, the final steps are post-processing and
displaying the output in contours or in vectors, the data that is obtained by the solver can be visualized
and displayed such as static pressure, velocity magnitude, static temperature as shown in next section.
The procedure analysis by CFD using model porous flow is shown in the Flowchart below:

Figure 3. Flowchart of (CFD) analysis process [15]

9. Results and Discussion

9.1. Effect of porous flow properties on pressure drop
The different velocity flow (porous flow) ranges have effected on pressure drop through the packed
column, the pressure difference between the bottom and top of the bed is (∆P), the profile of pressure
has been simulated over the Reynolds numbers range of (7.8 – 1115.876) as shown in Fig (4) CFD
contours simulation of static pressure, and as function to superficial gas velocity and length of bed

5
1st International Conference on Petroleum Technology and Petrochemicals IOP Publishing
IOP Conf. Series: Materials Science and Engineering 579 (2019) 012001 doi:10.1088/1757-899X/579/1/012001

column (0.9) m, the pressure drop CFD results are compared with Ergun correlation as shown in Table
(4). The Fig (5) shows the results from both models and gives reasonable data with error (0.6% – 0.63%)
between data results from (CFD) analysis and values from Ergun equation, this error comes from the
equation (3) for the porosity of medium. for bed height with (0.9 m), maximum pressure drop is
(33381.57) (Pascal) under Reynolds numbers (turbulent flow), the reason is the effect of changes in
porous flow regime to turbulent flow (strong inertial forces) when the bed length increases and (∆P)
along column increases

Table.4. CFD results compared with Ergun equation

(∆P)
(∆P) (Pascal)
(Pascal)
CFD model
Rep Ergun
(0.9) m length
model (0.9) m
column
length column

7.872750643 2.7494 10.52869951

188.2614284 972.0385 946.9017
290.94948 2299.466 2261.608
328.60177 2927.186 2884.842
376.522857 3835.525 3787.607
492.90265 6551.711 6490.893
554.51548 8282.316 8215.037
835.196155 18729.81 18636.28
1115.8768 33381.57 33267.08

6
1st International Conference on Petroleum Technology and Petrochemicals IOP Publishing
IOP Conf. Series: Materials Science and Engineering 579 (2019) 012001 doi:10.1088/1757-899X/579/1/012001

Figure 4. CFD contours simulation of static pressure (Pascal) (a) laminar flow (b) Turbulent flow.
40000
35000 CFD (Pa) (0.9 m) length VS. U(m/s)
Ergun (pa) (0.9 m) length
30000
(∆𝑃) (pacal)

25000
20000
15000
10000
5000
0
0 1 2 3 4
Superficial gas velocity (m/s)

Figure 5. (∆P) (Pascal), CFD results compared with Ergun equation variations with superficial gas
velocity (m/s) along length of column 0.9 m.

9.2. Effect of Mesh size:

the concept of mesh density can be examined at different mesh sizes by temperature distribution in
the model as shown in Fig.(6).The comparison between steady-state temperature distributions is the plot
of temperature (K) vs. radial position (m) for a velocity of 0.023 (m/s) for four mesh sizes (12k, 30k,
40k, 64k ) at high that equals to 0.594 m from bottom of packed bed model, the profiles of change in
temperature inside column model comes from changes in mesh sizes on the flow gas for 0.9 m model
with the appearance of very little difference in temperature between the mesh (40k and 30k) in the
middle of the column and a great convergence in temperature at the wall of the column temperature.
The mesh 64k has a clear difference in temperature, indicated to mesh size and refinement created a

7
1st International Conference on Petroleum Technology and Petrochemicals IOP Publishing
IOP Conf. Series: Materials Science and Engineering 579 (2019) 012001 doi:10.1088/1757-899X/579/1/012001

major little difference in temperature. So the variation in the mesh density should be smooth to reduce
numerical diffusion.

374
mesh = 40k mesh = 30k
372 mesh= 12k mesh = 64k

370
Temperature K

368

366

364

362

360
-0.0205 -0.0105 -0.0005 0.0095 0.0195
Diameter m
Figure 6. Temperature distributions at four different mesh sizes from bed model.

Figure 7. Static temperature in (K) for axial position (m) at a velocity = 0.023 (m/ s) with 40K
mesh size.
9.3. Effect of gas flow rate:
The gas flow rates have an important variable in bed column system and have direct influence on the
process model. Fig.8 shows simulation Ansys-Fluent for Contour of static temperature values along bed
model and affects it by gas flow inside system model at laminar and turbulent flow. from Fig.8 the total
difference in temperature for laminar flow (Rep=7.8) is (79K- 80K) to reach steady – state condition,
but for turbulent flow Rep = 1115.87 the total difference in temperature is (59K) along bed column
model, and this difference is ranging by an increase in the gas velocity with (x/L) for model column,
The increase in temperature with increasing Reynolds number has the effect on amount of heat into the
bed at wall temperature is constant.(CFD) simulations give suitable prediction in amount of heat inside
geometry model column at change in gas flow regime. Table.(5) illustrates CFD results temperature (K)
profiles inside bed column model at Reynolds number ranged from laminar to turbulent flow for
different positions. Fig.9 illustrates porous medium model distribution of radial temperature by CFD
simulation at different positions with high Reynolds number that equals to 1115.87 variations with radial

8
1st International Conference on Petroleum Technology and Petrochemicals IOP Publishing
IOP Conf. Series: Materials Science and Engineering 579 (2019) 012001 doi:10.1088/1757-899X/579/1/012001

(r) m. so the prediction from CFD simulation gives high details for porous flow in packed bed
equipment.

Fig.8. Static temperature in (K) along geometry model (a) Rep=7.872, (b, c, d, e, f) Rep=turbulent
flow

Figure 9. Porous media model illustration radial temperature distributions by CFD simulation at
different positions.

Table.5.CFD results temperature (K) profiles inside bed column model at laminar and turbulent
flow for different position
Different positions along bed material column (m)
s number
Reynold

0.05 0.1 0.4 0.6 0.8 0.9,

outlet
7.872 370.00 372.82 372.89 373.00 373.00 373.00
188.26 335.55 352.34 368.61 371.04 372 372.6
290.94 326.98 341.20 363.28 371.15 371.06 371.21
328.60 324.81 338.14 361.37 369.28 370.49 370.78
376.52 322.51 334.87 359.05 367.32 368.94 369.49

9
1st International Conference on Petroleum Technology and Petrochemicals IOP Publishing
IOP Conf. Series: Materials Science and Engineering 579 (2019) 012001 doi:10.1088/1757-899X/579/1/012001

492.90 318.28 328.98 353.94 362.40 365.58 366.97

554.51 316.56 326.65 351.54 360.03 362.92 364.76
835.19 311.2 319.38 342.76 350.24 355.89 358.15
1115.8 308.02 314.96 336.48 344.55 348.64 352

10. Conclusion
The pressure drop simulation has been compared with the values calculated from Ergun equation,
the relative error is (0.6% – 0.63%) for our model and It was found that there is a good agreement
between CFD and Ergun model, and maximum (∆P) for model bed with (0.9 m) height is (33381.57)
(Pascal) under different in flow regimes and It illustrated the effect of changes in porous flow regime to
turbulent flow, It becomes (strong inertial forces), and gives increases in the (∆P) along column.
The profiles of change in temperature inside column model comes from changes in mesh size on the
flow gas, there is a very little difference in temperature between the meshes (40k and 30k) in the middle
of the column it is and very close in the wall.
influence on the process model. The total difference in temperature for laminar flow (Rep=7.8) is
(79K- 80K) to reach steady – state condition, but for turbulent flow Rep = 1115.87 the total difference
in temperature is (59K) along bed column model, and this difference is ranging by an increase in the gas
velocity with (x/L) for model column. The prediction from Computational Fluid Dynamics gives us
simulation with high details for porous flow in packed bed equipment, design its and performance flow
of gas through porous media in packed bed column.

References
[1] Burke S P and Plummer W P 1928 Gas flow through packed column Ind. Eng. Chem. 20 1196
[2] Bark H 2009 CFD Advantages and Practical Applications Glumac Inc.
[3] Maier R S, Kroll D M, Kutsovsky Y E, Davis H T and Bernard R S 1998 Phy. Fluid, 10 60
[4] Gunnar J, Hellstr I, Jonsson P and Staffan Lundstr 2010 Journal of Porous Media 13 1073
[5] Kamyar M, Alkhalaf A and Specht E 2018 Computational Particle Mechanics 6 157
[6] Augier F, Idoux F and Delenne J Y 2010 Chemical Eng. Science 65 1055
[7] Wen D and Ding Y 2006 Chem. Eng. Sci. 61 3532
[8] Peters S T 1998 Handbook of composites Springer-Verlag (Houston: Springer-Verlag) ed 2 p
1118
[9] G K Batchelor 1967 An Introduction to Fluid Dynamics (Cambridge: University Press
Cambridge) ed 1 p 615
[10] Hlushkou D and Tallarek U 2006 Journal of Chromatography 1126 70
[11] ANSYS Fluent Documentation ANSYS Inc. 2010 ANSYS Fluent 12.0/12.1 Documentation
[12] Furnas C C 1931 Ind. Eng. Chem 23 1052
[13] Stewart B R B 1960 Transport Phenomena (New York: John Wiley and Sons Inc.) ed 1 p 780
[14] Launder B E and Spalding D 1972 Lectures in Mathematical Models of Turbulence (London:
Academic Press) ed 1 p 24
[15] ANSYS Fluent Documentation ANSYS Inc. 2009 Release 12.0
[16] Gunnar J, Hellstrom I, Jonas P and Jonsson P 2010 Journal of Porous Media 13 1073
[17] Festuccia A and Bella G 2018 Journal of Porous Media 21 5

10