AALTO UNIVERSITY

School of Chemical Engineering

CHEM-E7160 Fluid Flow in Process Units

Seminar Report
Fluid flow in packed bed reactor – 3D model

Leo Kukkonen
Tyko Viertiö
Abstract
This report contains a computational study performed with COMSOL and a literature
review regarding flow behavior in the packed bed reactor. The used models were the
from COMSOL provided porous bed reactor and specially designed repetitive void cell
of body-centered cubic lattice.

The closely inspected phenomenon was the behavior of fluid closer to the particles.
The reason behind such inspection, was to determine the fluid velocities and pressure
losses. These properties may help to determine the velocity regulations for example
for enhancing external diffusion. Another important aspect is, that by determining the
pressure loss, the length of the fixed bed can be determined.

Simulations were carried out using isothermal porous packed bed reactor, 300 K and
gaseous phase, diluted reactants and nitrogen as carrier gas. In addition, gaseous
phase a reaction A+B  C was added.

Introduction
The demand on understanding of the fluid flow in porous media has been increasing
in the last years. Not only in chemical engineering, but also in agricultural sciences,
particularly the knowledge fluid flow behavior through soil or in wood fibers is on
demand. In chemical engineering, the packed porous bed can be found in reactors,
separation and purification units. (Freund, 2003). As the computation power is
increasing, the thermodynamic and heat transfer can be coupled to the computations,
although the complexity of the calculations increases tremendously. (Sullivan, 2005).

Another important aspect is, that with increased computation efficiency the behavior
of fluids at the boundary of the porous particle can be closer investigated. The
additional chemical reaction A + B C influences the viscosity of the fluid, which
affects the fluid velocities near the boundary layer. This provides another challenge
for modeling of the packed bed reactors. (Sullivan, 2006)
Model
Model was created according to instructions by Comsol (anon, 2014). In this model,
Comsol Reacting flow in porous media –physics is used. Reaction conditions are
isothermal, temperature 300 K, since for example viscosity, diffusivity and density
change as function of temperature, and are user defined constants in this model. The
mesh used in simulations is presented in figure 1.

Figure 1: Mesh for the reactor simulation, grid scale mm.

Free fluid flow in model area is described by the Navier-Stokes –equations (1). The
fluid flow in catalyst bed is described by the Brinkman equations (2), which is the
expansion of Navier-Stokes for porous media.

∇ ∙ [−𝜂(∇𝒖 + (∇𝒖)𝑻 )] + 𝑝𝑰 = −𝜌(𝒖 ∙ ∇)𝒖

∇∙𝒖=0

(1)
 𝜂 𝜂
∇ ∙ [− (∇𝒖 + (∇𝐮)𝑻 ) + 𝑝𝑰] = − 𝒖
𝜀𝑝 𝜅

∇∙𝒖=0

(2)

When comparing the amount of modeled species to amount of solvent gas nitrogen
𝑁2 , a low concentration of species is assumed and Fick’s law can be used as approach
in mass transport by diffusion. As a result, equation (3) presents the model for
diffusion.

∇ ∙ (−𝐷𝑖 ∇𝑐𝑖 + 𝑐𝑖 𝒖) = 𝑅𝑖

(3)

Reaction is assumed to occur only in porous part of the reactor. Reaction rates (4) are
calculated from concentrations as first order with respect to species A and B and
reaction constant is calculated from Arrhenius equation. Arrhenius equation variables
𝑚3
can be easily modified in variables window, now A of 106 and activation energy
𝑚𝑜𝑙 𝑠
𝐽
of 3 ∙ 105 were used.
𝑚𝑜𝑙

𝑅𝑖 = −𝑘𝑐𝐴 𝑐𝐵 , where i is A or B

𝑅𝐶 = 𝑘𝑐𝐴 𝑐𝐵 (4)

The boundary conditions used in model are the velocity and species concentration of
inlet flow in both inlets and the outlet pressure of 0.

To expand the model, our intention was to simulate

flow through a catalyst bed and especially flow
around small particles. The figure 2 presents the
basic unit that was created and could be multiplied
to create the catalyst bed model. Navier-Stokes
equations (A) are used to describe the flow through
the porous bed.

Figure 2: Basic unit used in small scale simulation

Results and Discussion
The figure 3 presents the velocity in model reactor. As can be seen, in porous parts
the velocity almost constant not depending on the distance to reactor walls. On free
flow area, the fluid acts newtonially as supposed.

Figure 3: Velocity diagram of the model. Legend on right side, magnitude m/s.

The concentrations of species A, B and C are presented in figure 4, 5 and 6. As stated

in Comsol instructions conclusions too, the reactor requires longer distance from the
injection tube to the catalyst bed or static mixer placed between the injection of
species A and the bed, because only part of the catalyst bed is utilized for the reaction
due to poor distribution of A in the mixture. (anon, 2014)

Interesting result is that when measuring the distance from the start of the bed, the
first transformations from A and B to C take place near the reactor walls. The volume
flow is smaller near the walls which may lead to this result.
Figure 4: Volume: Concentration of species A in system. Arrow surface: velocity

Figure 5: Volume: Concentration of species B in system. Arrow surface: velocity

Figure 6: Volume: Concentration of species C in system. Arrow surface: velocity

Smaller scale packed bed simulations were

performed to study the flow through the bed and
especially flow near the catalyst particles. The
diffusion and mass transfer of reactive species
first to the surface of the catalyst particle and
then to the catalytically active site is one of the
rate determining factors in industrial scale
catalysis. Figure 7 presents the result of small
scale flow around a catalyst particle.

Figure 7: Fluid flow through a catalyst bed in single “unit”. Arrow size describes
velocity.

In figure 8, the results of flow extended to multiple “units” of simple particles is

presented. The geometry of basic unit is not modeled to describe real catalyst bed,
but it is heavily simplified to model where one catalyst particle is located in middle of
four particles in previous layer. Also the model choice for particles is an important. In
this model, particles were modeled as no-slip walls. In real catalysis case, the particles
are porous and reactant and product flows through the surface of catalyst particle can
cause major modeling challenges. Model simulated in this case can be used for rough
estimation of pressure drop in catalyst bed, for example, but more complex one is
needed for realistic flow simulation.

Figure 8: Flow through multiple “units” catalyst bed. Scale m/s, particle diameter 3
cm.

Conclusions
Large scale simulations are used for industrial reactor design. The reported results for
this simulation describe the type of modeling needed: mixing and flow through the
reactor is needed. Modeling used in large scale method is capable to provide
information of the pressure drop and the bed “effectivity” or how the reactant flows
are spread to reactor.

Small scale flow simulations can be used to provide information and realistic
parameters for larger scale models, where the model needs to be simplified to be
effective. For small scale simulations, more complex modeling is needed for reliable
results.

Notations
η viscosity

ρ density

k reaction rate constant

CA concentration A

CB concentration B

Ri Reaction rate of reactants

Rc Reaction rate of products

u fluid velocity

I identity matrix

εp porosity

κ permeability

p pressure

D diffusion coefficient

T temperature
anon, 2014. Modeling Reacting Flow in Porous Media,
https://www.comsol.com/blogs/modeling-reacting-flow-porous-media/, 15.5.2015

Freund, H., Zeiser, T., Huber, F., Klemm, E., Brenner, G., Durst, F. and Emig, G., 2003.
Numerical simulations of single phase reacting flows in randomly packed fixed-bed
reactors and experimental validation. Chemical Engineering Science, 58(3), pp.903-
910. DOI: 10.1016/S0009-2509(02)00622-X

Sullivan, S.P., Sani, F.M., Johns, M.L. and Gladden, L.F., 2005. Simulation of packed bed
reactors using lattice Boltzmann methods. Chemical Engineering Science, 60(12),
pp.3405-3418. DOI: 10.1016/j.ces.2005.01.038

Sullivan, S.P., Gladden, L.F. and Johns, M.L., 2006. Simulation of power-law fluid flow
through porous media using lattice Boltzmann techniques. Journal of Non-Newtonian
Fluid Mechanics, 133(2), pp.91-98. DOI: 10.1016/j.jnnfm.2005.11.003