CH11CH03_Dixon ARjats.

cls February 25, 2020 16:36

Annual Review of Chemical and Biomolecular

Engineering
Computational Fluid Dynamics
for Fixed Bed Reactor Design
Anthony G. Dixon and Behnam Partopour
Annu. Rev. Chem. Biomol. Eng. 2020.11. Downloaded from www.annualreviews.org

Department of Chemical Engineering, Worcester Polytechnic Institute, Worcester,

Massachusetts 01609, USA; email: [email protected], [email protected]
Access provided by 207.241.225.139 on 05/07/20. For personal use only.

Annu. Rev. Chem. Biomol. Eng. 2020. 11:3.1–3.22 Keywords

The Annual Review of Chemical and Biomolecular
computational fluid dynamics, CFD, fixed bed, reactor modeling,
Engineering is online at chembioeng.annualreviews.org
particle-resolved, computer-generated packing, microkinetics
https://doi.org/10.1146/annurev-chembioeng-
092319-075328 Abstract
Copyright © 2020 by Annual Reviews.
Flow, heat, and mass transfer in fixed beds of catalyst particles are complex
All rights reserved
phenomena and, when combined with catalytic reactions, are multiscale
in both time and space; therefore, advanced computational techniques
are being applied to fixed bed modeling to an ever-greater extent. The
fast-growing literature on the use of computational fluid dynamics (CFD)
in fixed-bed design reflects the rapid development of this subfield of reactor
modeling. We identify recent trends and research directions in which
successful methodology has been established, for example, in computer
generation of packings of complex particles, and where more work is needed,
for example, in the meshing of nonsphere packings and the simulation of
industrial-size packed tubes. Development of fixed-bed reactor models, by
either using CFD directly or obtaining insight, closures, and parameters
for engineering models from simulations, will increase confidence in using
these methods for design along with, or instead of, expensive pilot-scale

·.•�-
experiments.

3.1

, Review in Advance first posted

on March 9, 2020. (Changes may
still occur before final publication.)
 CH11CH03_Dixon ARjats.cls February 25, 2020 16:36

1. INTRODUCTION
A catalytic fixed-bed reactor is a tube, usually cylindrical, filled with active elements such as par-
ticles, open foams, or structured inserts. The particles are often spheres, cylinders, or rings, but
complex shapes, such as wagon wheels, multilobes, multichannel cylinders, and cylinders with ex-
ternal flutes or grooves, have been used for specific processes. If the reaction is accompanied by
strong heat effects that require heat transfer through the tube, then the tube–to–particle diameter
ratio (N) is quite low, typically 3–10, and wall effects play a major role in tube performance.
Models of fixed beds have been developed and used for many years. The conventional chemical
reaction engineering model approach treats the catalyst bed as a homogeneous continuum with
effective transport properties that reflect both the fluid and particulate phases. These models for
many years assumed constant axial velocity, v z , or plug flow. Recognizing the inadequacy of this
assumption, researchers looked to include the radial variation of the axial velocity, v z (r), in 2D
models. These types of treatment of the fixed bed velocity are not included in this review.
Annu. Rev. Chem. Biomol. Eng. 2020.11. Downloaded from www.annualreviews.org

Computational fluid dynamics (CFD) for fixed beds has two main forms: the porous medium
Access provided by 207.241.225.139 on 05/07/20. For personal use only.

model (PMM) approach and the particle-resolved CFD (PRCFD) approach. Both are discussed
in this article, with emphasis on the latter. There has been much interest in PRCFD (sometimes
called detailed modeling, interstitial modeling, or pore-scale modeling) in recent years. In this
approach, the fixed-bed model includes individual particles, solving equations for the fluid flow
around them coupled to pellet models for transport and reaction inside or on the surfaces of the
pellets. This approach allows much more detailed study of fixed bed phenomena at different length
and timescales than effective or continuous medium approaches, as Figure 1 illustrates.
The interest in PRCFD is reflected in two review papers. Dixon et al. (1) covered work up
to 2006, mainly studies of periodic bed segments and clusters of particles, and very low-N tubes
with few particles. The explosive growth of publications and methods since, up to 2017, has been
comprehensively documented and interpreted by Jurtz et al. (2).
There has also been discussion on the role of PRCFD. Dixon & Nijemeisland (3) suggested
that it could contribute to fixed bed design by improving the description of flow and by providing
detailed information to develop improved reduced models for design and analysis. Wehinger &
Kraume (4) asked if CFD was ready to be a design tool for low-N tubular reactors. They concluded
that despite heavy computational cost, PRCFD could replace classical approaches, advantages
being that correlations for effective parameters would not be needed and that PRCFD allows
algorithm-based design and optimization of new catalyst pellet shapes. Partopour & Dixon (5)
suggested some areas of focus for future research, including the application of machine-learning
methods to the large data sets produced by 3D PRCFD.
The present review attempts to identify recent trends over the past 10 years without giving a
comprehensive literature survey, including as-yet-unresolved questions, disagreements, and needs
for the future. To maintain focus, the scope of this article is limited. It does not present detailed
model equations or discuss experimental techniques for validation of PRCFD or structured
arrangements of particles. Only single-phase (usually gas) flow is covered; for CFD of multiphase
flow, Wang et al.’s (6) review is recommended. For more detail on fixed bed CFD than is presented
here, the reader is referred to References 1 and 2 and the specific references associated with each
topic.

2. FIXED BED COMPUTATIONAL FLUID DYNAMICS: THE POROUS

MEDIUM MODEL
We consider fixed bed CFD to consist of approaches that solve for the velocity field directly. In

·.•�-
this section, we briefly discuss the PMM, which is the standard model available for porous flows
in COMSOL® , ANSYS-Fluent/CFX, and STAR-CCM+.
3.2 Dixon • Partopour

, Review in Advance first posted

on March 9, 2020. (Changes may
still occur before final publication.)
 CH11CH03_Dixon ARjats.cls February 25, 2020 16:36

Catalyst
pellets

Pellet pores

Flow and
transport
interaction
Time

Surface kinetics
Annu. Rev. Chem. Biomol. Eng. 2020.11. Downloaded from www.annualreviews.org
Access provided by 207.241.225.139 on 05/07/20. For personal use only.

Effective diffusivity
Aldehyde and mass transfer
limitations
Epoxide

Surface sites, elementary

steps, and selectivity

Length
Figure 1
Multiscale nature of fixed-bed transport and reaction processes, from molecular scale on catalyst surface to
pore scale in catalyst pellets to bed scale in packed tubes of particles. Adapted in part with permission from
Reference 145; copyright 2005 American Chemical Society.

The PMM treats the fixed bed as a continuum; the presence of the particles is included not
explicitly but through source terms that account for momentum loss. The PMM adds the source
terms into the incompressible Newtonian fluid Navier–Stokes equations. The source terms in-
clude viscous loss resistance and inertia loss resistance. Although source terms can account for the
increase in resistance to flow owing to the particles, they cannot reproduce the radial variation in
axial flow that is important in reactor modeling. To do this, a measure of bed structure must be
used; the void fraction distribution ε(r,z) is usually chosen. Often the axial variation is neglected,
and a further simplification is to replace the oscillating voidage profile with an approximate expo-
nential profile that captures only the increase in voidage near the tube wall.
Jakobsen et al. (7) provided an extended treatment of the PMM termed the 2D CFD model.
They compared both 1D and 2D versions to classical 1D and 2D dispersion (effective parameter)
models. The variations in velocity, pressure, and mixture density allowed in the 1D CFD model
significantly affected chemical composition and conversion compared with dispersion models with
fixed values for synthesis gas and methanol production processes. Including nonuniform ε(r) in
the 2D CFD model affected axial pressure drop and fluid velocity but had little influence on
temperature and species mole fraction profiles.
Wu et al. (8) compared PRCFD (the realistic approach) to the porous medium approach. They
noted that the PMM could not reproduce secondary flows and flow anisotropies, such as separa-

·.•�-
tion, oscillation, and vortex formation, but could capture average pressure drop and temperature

www.annualreviews.org • CFD for Fixed Bed Reactor Design 3.3

, Review in Advance first posted

on March 9, 2020. (Changes may
still occur before final publication.)
 CH11CH03_Dixon ARjats.cls February 25, 2020 16:36

increase. The PRCFD approach was estimated at 3,000 times the computational cost of PMM.
Gao et al. (9) and Zhuang et al. (10) used the two-temperature 2D PMM to simulate reactive
processes in fixed beds with turbulent flow. They showed the two-temperature model gave better
results compared with experiments than did a single-temperature model but did not present any
results on the computed velocity field.
The 2D PMM solves for both velocity components v z and v r , but because v r is driven by
radial pressure gradients, which are zero across the entire bed diameter, v r is usually computed
as negligible. This is a problem of scales; over a single particle, the radially displaced flow gives a
significant v r , but when averaged over the bed radius, the positive and negative variations cancel
out. Recently, Hamzah et al. (11) used a form of the PMM to assist in the development of a new
segmented continuum approach to radial heat and mass transfer.
The PMM is frequently used to include improved treatment of the velocity field compared
with the simpler dispersion models, without incurring the high computational cost of full PRCFD
Annu. Rev. Chem. Biomol. Eng. 2020.11. Downloaded from www.annualreviews.org

models. There is a need to study whether the computed velocity field is realistic or, indeed, whether
it needs to be for good reactor simulation.
Access provided by 207.241.225.139 on 05/07/20. For personal use only.

3. COMPUTER GENERATION OF FIXED BEDS

Conducting accurate resolved particle CFD simulations of fixed beds without having proper pack-
ing geometry is impossible. Therefore, computer generation of fixed beds is essential for such
studies and has been the subject of many investigations during the past three decades. Spherical
particles are the simplest geometry. Soppe (12) showed that ballistic random deposition according
to the rain model coupled with a Monte Carlo–based compression process can accurately repro-
duce the packing of spherical particles and evaluate its properties. Freund et al. (13, 14) used a
similar algorithm to successfully generate and study fixed bed properties. Since then, there have
been significant advances in the field. The discrete element method (DEM) is a powerful tool to
generate packings of spherical and other particle shapes, such as cylinders. The method evaluates
the trajectory and velocity of the particles in the system by calculating the forces on each particle
owing to gravity and collision at each time step. Eppinger et al. (15) proposed an automated ap-
proach for numerical generation of packing spheres and cylinders based on the DEM algorithm.
The approach was also capable of generating packing of rings, however, only by replacing the
particles in the cylindrical packing. In conventional DEM methods, to have an accurate solution,
specific contact detection and force models must be developed for different geometries. A work-
flow around this approach is to represent (estimate) complex particles with a glued spheres or
clumped spheres model. In this approach, many spheres are glued together to represent cylinders,
rings, and other shapes. Caulkin et al. (16) showed that this approach can significantly affect the
accuracy of the models, especially owing to lack of true representation in the corners and edges.
Therefore, in recent years, there have been numerous efforts to develop flexible DEM algorithms
for arbitrary particle shapes.
Wachs and coworkers (17–19) conducted thorough investigations to develop generic DEM
algorithms for complex particle shapes. Wachs et al. (17) introduced a new algorithm called
Grains3D, which allowed simulation of flow dynamics of arbitrary particle shapes. The method-
ology is restricted to convex particle shapes and serial computing. However, it allows the user to
have a combination of shapes and sizes in the simulation. Rakotonirina & Wachs (18) extended the
algorithm to a parallel computing framework. Furthermore, Rakotonirina et al. (19) used glued
convex particles to represent concave objects in Grains3D, which made the algorithm a very versa-
tile methodology. Seelen et al. (20) also developed a unique methodology to simulate packings for

·.•�-
different convex particle shapes. The algorithm advantage is its powerful method for identifying
the contact points for the faces in contact.
3.4 Dixon • Partopour

, Review in Advance first posted

on March 9, 2020. (Changes may
still occur before final publication.)
 CH11CH03_Dixon ARjats.cls February 25, 2020 16:36

Marigo & Stitt (21) conducted an extensive study on the accuracy and industrial applications
of DEM methods. The authors emphasized calibration and validation as crucial tools for realistic
DEM simulations. Furthermore, Yan et al. (22) studied the impact of DEM input parameters using
a statistical approach. The authors showed that prior simulations carried out along with sensitivity
analysis can help to identify important parameters that should be measured experimentally. Farsi
et al. (23) studied the potential applications of the combined finite and discrete element methods
(FEMDEM) in numerical packing generation. They suggested that FEMDEM can be used to
study the effects of particle–particle interactions during the packing and the impact of the loading
conditions on the final packing properties.
Marek (24) presented a numerical algorithm for filling a cylindrical column with equilateral
cylinders based on simplified mechanics. The algorithm used sequential deposition and the frozen
bed assumption—once a particle was at mechanical equilibrium, it could not be moved by later
particles. Collision detection was performed by covering the particle surface with markers and
Annu. Rev. Chem. Biomol. Eng. 2020.11. Downloaded from www.annualreviews.org

checking their location. Good agreement with radial voidage profiles was obtained, and the ori-
entation distributions were analyzed in a subsequent publication (25).
Access provided by 207.241.225.139 on 05/07/20. For personal use only.

Boccardo et al. (26) and Partopour & Dixon (27) investigated a promising alternative to DEM
for the numerical generation of packing of complex particle shapes. The methodology is based
on rigid body dynamics (RBD) embedded in the Bullet Physics library and implemented in the
open-source software BlenderTM . In the framework, particles are defined by surface meshes,
and therefore, they can have any arbitrary shapes. The contact detection is carried out based
on proximity queries for vertices in the objects. The computational space is usually divided into
subspaces using octrees to speed up the calculation. Both studies (26, 27) showed that geometric
properties of packings of complex particle shapes can be well reproduced using this methodology.
Moghaddam et al. (28) have used very similar RBD methods and implemented an alternative
method for transition between moving and resting particles. Fernengel at al. (29) have presented an
automated workflow for random packing generation, mesh generation, flow simulation, and reac-
tion simulation using bash and python scripts based on BlenderTM , OpenFOAM® , and ParaView® .
Fleischlen & Wehinger (30) conducted a comprehensive study to compare computer gener-
ation of fixed beds using RBD (in BlenderTM ) and DEM (embedded in STAR-CCM+). They
showed the two approaches are in very good agreement, although simulations carried out in
BlenderTM are less computationally expensive.
In most studies comparing experiments to computer-generated packings, the model generation
is based on imaging of the bed. Baker et al. (31) have presented an interesting change to that process
by creating the experimental bed using additive layer manufacturing or 3D printing from the
computer-generated model. The authors suggest that this could be a new technique for studying
flow in packed beds.

4. PARTICLE-RESOLVED COMPUTATIONAL FLUID DYNAMICS

METHODOLOGY
The main methods for solving the PRCFD models are continuum Navier–Stokes approaches,
such as finite volume (FV) and finite element (FE), and particle-based approaches, such as the
lattice Boltzmann method (LBM). A third group is the network models based usually on a
Delaunay tessellation of the bed structure (32), but this method is restricted to fluid flow at
present.
The LBM method has the advantages that no partial differential equations need to be solved
and it is computationally efficient and easy to parallelize. A related method, stochastic rotation dy-

·.•�-
namics, has recently been used (33) to determine the solvent hydrodynamics in a catalytic reactor.

www.annualreviews.org • CFD for Fixed Bed Reactor Design 3.5

, Review in Advance first posted

on March 9, 2020. (Changes may
still occur before final publication.)
 CH11CH03_Dixon ARjats.cls February 25, 2020 16:36

The LBM is inherently transient in nature, and early versions could simulate only flow problems
(with optional species) owing to the difficulty of including thermal relaxation times. Recent de-
velopments have led to the use of multiple–relaxation time collision operators to include thermal
collisions (34). Unresolved problems are the inability of the LBM approach to include conjugate
problems and high-Re turbulent flows owing to the wide variety of temporal and spatial scales
needed. Li et al. (35) reported an application of LBM to a fixed bed, but the reaction was confined
to the catalyst surface.
The continuum methods solve conservation equations of macroscopic quantities in the form
of the Navier–Stokes equations defined on a region of space, which is discretized (along with
time for unsteady problems) on a partition or mesh. This is the basis for most software packages;
for example, FV is the most popular and is used in ANSYS-Fluent, ANSYS-CFX, Star-CCM+,
and the open-source code OpenFOAM® . FE is preferred in the Multiphysics environment of
COMSOL® . There have been very few uses of PRCFD in COMSOL® ; for one example, see
Annu. Rev. Chem. Biomol. Eng. 2020.11. Downloaded from www.annualreviews.org

work by Touitou et al. (36). COMSOL® is a fine element–based solver that is more suited to
PMM models, which generally have smaller meshes. Most published work uses ANSYS-Fluent or
Access provided by 207.241.225.139 on 05/07/20. For personal use only.

Star-CCM+. We focus mainly on continuum methods in this review, because at the moment they
have the greatest potential for catalytic reactor simulations.

4.1. Meshing
The geometrical domain must be discretized or meshed into smaller regions for solution by one of
the CFD algorithms. Many ways to do this have been developed, reflecting different balances be-
tween fidelity to the original geometry, ease of mesh generation, and accuracy/cost of the ensuing
solution method. The two main classes of mesh are structured and unstructured. For a structured
mesh, the distribution of nodes throughout the domain is known once the surface nodes are set, as
the domain nodes are projected from the surfaces. For unstructured meshes, nodes are generated
as a space-filling algorithm sweeps through the domain. A mixture of these, hybrid mesh, can be
obtained by subdividing the domain.
For the LBM method, Cartesian structured meshes are used, and curved boundaries are ac-
commodated using a cut-cell approach. This can lead to a staircase representation of a curved
boundary, requiring refinement of the mesh. Structured meshes are easier to set up and require
fewer cells.
For FV and FE solvers, many types of mesh are used. Most common in complex geometries is
unstructured (usually tetrahedral) cells fitted as closely as possible to the original surfaces. Because
there are usually strong boundary layer gradients and accurate fluxes are desired at interfaces,
layers of thin hexahedral prism cells are frequently used in a body-fitted mesh. Recently, polyhedral
cells have been used to reduce mesh size and increase computational stability.
One attempt to combine the advantages of both main mesh types is the so-called chimera grid
(37), in which a hexahedral background mesh is overlaid by local body-fitted meshes. This requires
increased computational overhead for the interpolations necessary to exchange information be-
tween meshes.
Several alternatives to the use of Cartesian grids for fixed beds have been suggested. Finn &
Apte (38) compared body-conforming tetrahedral meshes to non-body-conforming fictitious do-
main approaches. Sulaiman et al. (39) coupled the sharp interface method to a fictitious domain
two-phase flow solver to treat convective mass transfer around reactive particles. Das et al. (40)
have used an immersed boundary method (IBM), originally applied to particulate flows, for sim-
ulations of foam and particle fixed beds. Most recently, Yuan et al. (41) put forward a method to

·.•�-
combine immersed boundary and adaptive meshing methods.

3.6 Dixon • Partopour

, Review in Advance first posted

on March 9, 2020. (Changes may
still occur before final publication.)
 CH11CH03_Dixon ARjats.cls February 25, 2020 16:36

The body-fitted grid combined with a FV approach is most common, as it can directly re-
solve the solid–fluid interface. Disadvantages are the resulting high cell count, more difficult mesh
refinement for error estimation, and problems with particle–particle and particle–wall contact
points, which can lead to highly skewed cells.

4.2. Contact Points

A meshing problem that is specific to PRCFD is that of contact points, either particle–particle or
particle–wall. Packings of ideal spheres have point contacts, but packings of cylinders and other
shapes can have point, line, and area contacts, which present many more difficulties. Mesh cells
can be distorted or very small, which leads to large, expensive meshes.
Early approaches for spheres (3) shrank all particles by a fixed amount, e.g., 0.5–2% of diameter,
the gap method. This is the easiest to implement and performs reasonably for flow simulations for
Annu. Rev. Chem. Biomol. Eng. 2020.11. Downloaded from www.annualreviews.org

1% reduction or lower. Guardo et al. (42) suggested expanding the spheres, the overlap method.
They reported no meshing problems, but this method does not eliminate a narrow fillet of fluid.
Access provided by 207.241.225.139 on 05/07/20. For personal use only.

The gap and overlap methods are termed global methods.

Local methods involve changing the geometry only in the vicinity of the contact points. The
bridges method (43) replaces the fluid and a small portion of the particles with a small cylinder at
the contact point. The caps method (44) introduced the idea of moving the surface nodes near a
contact point to locally flatten the curved surface—equivalent to removing a spherical cap.
Several authors have compared these methods for spheres. Dixon et al. (45) found that global
methods of gap and overlap at 1% changed the ε and P by too much. Local methods gave
much smaller changes in ε and P. For heat transfer, the caps method omits the finite area con-
tact conduction, and so the bridges method was preferred. To obtain the correct heat flow, the
correct bridge thermal conductivity must be used, and formulas were given for spheres. These
conclusions were supported (46) in an investigation of flow and heat transfer; both studies recom-
mended the bridge diameter should be smaller than 20% of the particle diameter. Rebughini et al.
(47) extended the bridge method to cases of isothermal external surface reaction on the particles,
at lower Re, and recommended bridge-to-particle ratios of 0.15 or higher to avoid skewness in the
cells. They found conversion was unaffected for ratios up to 0.3.
For nonspherical particles, the problem is much more difficult. The main task is finding the
contacts, which is easy for spheres. Contact algorithms have been developed for full cylinders (48),
which Wehinger et al. (49) used to develop methods based on bridges and caps and to introduce
a new united method.
For more complex particle shapes, it will be difficult to extend the bridges method, particularly
the bridge thermal conductivity formulas. Methods based on local mesh rearrangement, such as
Laplace smoothing, are easier to implement but may miss some heat-transfer pathways, especially
at the tube wall where gradients are steep. There is considerable room for development of this
topic.

4.3. Microkinetics for Particle-Resolved Computational Fluid Dynamics

There are two main methods to couple microkinetics models with PRCFD simulations. The tradi-
tional method is to add microkinetics as boundary conditions for the flow field on the fluid–solid
interface. In this method, surface flux for each species is calculated using the detailed reaction
mechanism presented by the microkinetics model based on the temperature and partial pressures
of species at the solid–fluid interface. The method is reliable for very fast reactions where the

·.•�-
rate is controlled by diffusion. In these systems, reaction mostly happens on the surface of catalyst

www.annualreviews.org • CFD for Fixed Bed Reactor Design 3.7

, Review in Advance first posted

on March 9, 2020. (Changes may
still occur before final publication.)
 CH11CH03_Dixon ARjats.cls February 25, 2020 16:36

particles (e.g., methane partial oxidation on Rh). However, this framework could fail when reaction
also happens inside the particles. Therefore, a second methodology (with different approaches)
was introduced to include the reaction inside the solid catalyst particles. Maestri & Cuoci (50)
introduced a transient multiregion operator splitting CFD approach to couple CFD with full mi-
crokinetics models, whereas Dixon et al. (51) presented a steady-state solid particle approach to
couple microkinetics to transport and reaction inside the particle. Both methods are discussed in
more detail in the following section. Matera et al. (52) developed a numerical framework to inte-
grate first-principles kinetic Monte Carlo microkinetic models into stagnation flow and channel
flow reactors, but this has not yet been applied to fixed beds. Partopour et al. (53, 54) suggested
two different approaches based on full numerical solution of the microkinetics models learned by
either multivariate splines or random forest trees. Votsmeier has also shown the computational sav-
ings gained by using multivariate splines to interpolate precomputed reaction rates (55). Rebughini
et al. (56) proposed to use a cell agglomeration algorithm, whereas Blasi & Kee (57) used in situ
Annu. Rev. Chem. Biomol. Eng. 2020.11. Downloaded from www.annualreviews.org

adaptive tabulation for the task. All the approaches mentioned above successfully couple the reac-
tion inside the catalyst particles with flow and transport. The second methodology (reaction inside
Access provided by 207.241.225.139 on 05/07/20. For personal use only.

the particles) is more reliable than traditional methods (reaction as flow boundary condition) when
it comes to slow reactions. However, it is computationally more expensive than the traditional
method.
It should be noted that when reaction is added via boundary conditions, usually, the mass and
heat flux vectors only point outward into the fluid region. This is physically incorrect; in reality,
the heat flux could be directed into the solid catalyst particles too.

4.4. Pellet Effective Medium Models

To model the reaction and diffusion inside the solid catalyst particles for PRCFD, effective
medium models are necessary. The catalyst pores can range from nanometers to micrometers
and are several orders of magnitude different in scale to the dimensions of the void space that de-
termine transport and flow in the bed. Therefore, resolving pore geometry for fixed beds of several
hundreds of particles is computationally extremely difficult if not impossible. Several studies have
evaluated the accuracy of different effective medium models for species diffusion inside the parti-
cles. Dong et al. (58) showed that bimodal diffusion models with both macro and micro pore sizes
could be used to study the overall performance of the fixed bed and tune important parameters
such as selectivity and convergence. Wehinger et al. (59) compared three different approaches
for incorporating transport into catalyst particles in fixed bed CFD: the instantaneous diffusion,
effectiveness-factor, and 3D reaction–diffusion models. The authors suggested the instantaneous
diffusion model results in significant error for reactor simulations. Furthermore, although the
3D reaction–diffusion model has superior accuracy, its computational time could be 200 times
slower than the other approaches. The effectiveness-factor model, they believe, could perform
reasonably well while keeping the computational cost manageable. They emphasize studying the
catalyst structure (e.g., pore size distribution and bulk porosity) before choosing the right ap-
proach. Donaubauer & Hinrichsen (60) compared the three main effective multicomponent diffu-
sion models: the dusty gas, binary friction, and mean transport pore models. Their results suggest
that the discrepancy between these models is negligible. Furthermore, they identified porosity,
tortuosity, and pore diameter as the most important parameters for effective diffusivity, respec-
tively. Partopour et al. (61) studied the effects of flux dependency and anisotropic diffusion on the
conventional Fickian simplification of the dusty gas model. The results showed that species flux
changes inside the catalyst particles have considerable impact on the diffusion coefficient, whereas

·.•�-
anisotropic effects owing to temperature and species gradient in 3D are negligible.

3.8 Dixon • Partopour

, Review in Advance first posted

on March 9, 2020. (Changes may
still occur before final publication.)
 CH11CH03_Dixon ARjats.cls February 25, 2020 16:36

Two studies have examined resolved-pore simulations of thin catalyst layers: Bufe et al. (62)
simulated diffusion and reaction in liquid-filled transport pores for Fischer–Tropsch synthesis
to obtain optimal values of catalyst layer thickness and porosity, and Partopour & Dixon (63)
simulated CO oxidation in a catalyst layer of Rh/Al2 O3 where the macropore structure was ex-
plicitly modeled and the interaction between transport and reaction was shown. The coupling of
resolved-pore modeling in catalyst pellets with resolved-particle modeling of the fixed bed is a
level of simulation that is not yet achievable given present computational abilities.

4.5. Fluid-Particle Coupling

Early studies using PRCFD either ignored intraparticle transport or allowed only heat transfer
inside the pellet. For heterogeneous catalytic reactor simulations, it is desirable to include chemical
species inside the particle. A first approach to this treated the particles as porous regions (64, 65),
Annu. Rev. Chem. Biomol. Eng. 2020.11. Downloaded from www.annualreviews.org

with extra terms in the momentum balance to account for additional flow resistance. This method
has been shown (51) to allow too much convective heat and mass transfer into the particles and
Access provided by 207.241.225.139 on 05/07/20. For personal use only.

cannot be used unless a prohibitively fine mesh is placed at the particle surfaces.
A new method was put forward that models the particles as solid regions (51) and that employs
user-defined scalars to stand in for mass fractions of species inside the particles. Care must be taken
to properly couple the mass fractions and the species fluxes across the particle–fluid interface.
Several groups (66–70) have successfully used this method.
Maestri & Cuoci (50) put forward a different approach to the fluid–particle coupling problem.
Their priority was solving large sets of species balances arising from detailed microkinetics mod-
els. They coupled microkinetics reactions with CFD simulations using an operator-splitting tech-
nique that separates the fluid transport and particle reaction timescales. The authors developed
a new open source solver that works in the OpenFOAM® environment, called CatalyticFOAM.
They applied it to a 3D model of a narrow bed of eight Raschig rings. In this initial work, only
a surface layer of catalyst in contact with the flow was active. Maffei et al. (71) further developed
the code to allow internal heat and mass transfer in the pellets. These were successfully tested on
a packed bed of 50 spheres using a microkinetic model of H2 combustion on Rh. Bracconi et al.
(72) incorporated the in situ adaptive tabulation method into the operator-splitting algorithm to
improve the handling of large sets of microkinetics equations and tested it on a packed bed of 25
spheres using methane steam reforming.
The operator-splitting method has been implemented in commercial software such as STAR-
CCM+. Other approaches have been developed, such as the CAT-PP method for highly di-
luted heterogeneous catalytic systems (73) and DUO, which is an interface for OpenFOAM®
TM
and DETCHEM for elementary step kinetic modeling (74). DUO has been applied to coated
honeycomb monoliths but not yet to fixed beds.
The operator-splitting method is inherently transient, although steady-state simulation is pos-
sible by pseudotime steps; the iterative structure makes it computationally expensive, and it is
limited to surface kinetics, requiring a geometric correction factor for catalyst particles. The solid
particle method can be slow to converge, owing to the longer timescale needed for intrapellet
transport, and requires a nontrivial amount of user-defined coding. It can, however, represent re-
action throughout the pellet. At present, both methods are in use.

4.6. Turbulence Models in PRCFD

Flow regimes in packed beds (75) range from viscous flow (Re < 1) to steady laminar iner-

·.•�-
tial flow (5 < Re < 60), unsteady laminar inertial flow (60 < Re < 120), and turbulent flow

www.annualreviews.org • CFD for Fixed Bed Reactor Design 3.9

, Review in Advance first posted

on March 9, 2020. (Changes may
still occur before final publication.)
 CH11CH03_Dixon ARjats.cls February 25, 2020 16:36

(Re > 120). However, other studies (76) have suggested that turbulence does not appear until
Re > 900, although the transition to turbulence can take place at different locations in the bed
corresponding to different local flow rates.
Direct numerical simulation (DNS) in which all time- and length scales are resolved is not
practical for anything more than small clusters of particles, as fine meshes and small time steps are
needed. A compromise is large eddy simulation (LES), in which smaller length scales are mod-
eled and only larger ones are resolved. The most common method is to use Reynolds-averaged
Navier–Stokes (RANS) models in which all length scales are modeled, and which are based on the
turbulent viscosity concept.
Lee et al. (77) found that LES was preferred in their study of pebble beds, along with the k-ω
RANS model. Dixon et al. (78) compared turbulence models for drag and heat transfer on a single
sphere and also found that LES and the shear stress transport (SST) k-ω model gave superior
results. Shams et al. (79, 80) put forward their improved delayed detached eddy simulation method,
Annu. Rev. Chem. Biomol. Eng. 2020.11. Downloaded from www.annualreviews.org

which is a blend of LES and unsteady RANS. Note that LES and its derivatives are inherently
unsteady methods, requiring time integration up to a statistically stable steady state, which can be
Access provided by 207.241.225.139 on 05/07/20. For personal use only.

computationally expensive.
Guardo et al. (81) compared the Spalart–Almaras, k-ε, k-ε RNG, k-ε Realizable, and k-ω RANS
models and preferred the Spalart–Almaras model. Coussirat et al. (82) included Reynolds stress
models and concluded they were no better. k-ε models are the most common but are known to
be poor for swirling and rotating flows and flows with strong separation, such as are found in
PRCFD simulations, and do not perform as well as LES. More recently, Karthik et al. (69) con-
firmed these results for water flow in beds of spheres with different turbulence models compared
with their experimental measurements. Jiang et al. (83) studied the local flow structure and turbu-
lence quantities, such as the distribution of k and ε and the distribution of vortices with different
intensities.
As important a decision as the choice of turbulence model is the choice of wall treatment, given
the large amount of solid surface in fixed beds. The standard wall functions require y+ > 30, but
this is difficult to obtain in fixed beds, because flow over the particle surfaces varies strongly, and
y+ can range from one to several hundred over a single particle surface. A popular choice is some
form of interpolation between using a resolved-layer approach for low y+ and wall functions for
larger y+ , embodied in enhanced wall treatments in most of the available commercial codes.
To date, there is no clear recommendation regarding choice of turbulence model. Good turbu-
lence models (LES) are unsteady and expensive and are restricted to small beds of particles. More
realistic beds can be simulated using RANS models, but all have shortcomings. More insight is
needed as to what is gained in engineering predictions by using more refined turbulence models.

4.7. Newer Methodology in PRCFD

Despite many PRCFD simulations of flow and heat transfer in the literature, radiation has been
included only recently. Behnam et al. (84) used the surface-to-surface (S2S) model in a validation
study of a single-pellet-string reactor. Wehinger et al. (85) also used S2S radiation in a CFD model
of dry reforming to compare with experiments. More extensive studies (86, 87) compared the S2S
model, based on precomputing view factors, to solving the radiative transfer equation via the dis-
crete ordinates method for a steam reforming gas mixture. The S2S model was recommended
based on lower computational cost, with both models giving similar results. A strong argument
was made that radiation should not be neglected in simulating this process, based on steam re-
forming simulations in a short 0.5-m length of packing representing the reformer tube inlet, at

·.•�-
flow rates corresponding to Re = 755 and 7,554, with the tube wall temperature held constant

3.10 Dixon • Partopour

, Review in Advance first posted

on March 9, 2020. (Changes may
still occur before final publication.)
 CH11CH03_Dixon ARjats.cls February 25, 2020 16:36

220 K above the inlet gas temperature. Results showed significant differences in temperature for
the low reforming flow Re = 755 (up to 50 K), but less so for Re = 7554 (less than 20 K) and
decreasing with tube length. This is to be expected for the higher Re, for which radial thermal dis-
persion will be larger and the relative contribution of radiation will be smaller. This is in accord
with industrial experience on long reforming tubes at high flow rates (84).
Most PRCFD studies have been focused on steady-state operation. The dynamic behavior of
the reactor is an important issue, as understanding the transient effects of process changes and
disturbances is essential for safe and proper reactor operation. Simulations of an ethylene oxide
reactor tube under unsteady conditions were run (88) using steady-state kinetics assuming rapid
surface equilibration compared with transport processes, so a single timescale could be used. The
mesh was made up of 24 M cells, and transients of wrong-way behavior, reactor shut-off, and
reactor start-up were studied. These simulations were very demanding of computer time, taking
approximately a week to simulate one second of real time. Routine use of dynamic PRCFD will
Annu. Rev. Chem. Biomol. Eng. 2020.11. Downloaded from www.annualreviews.org

need bigger, faster computers than were available in that study.

Access provided by 207.241.225.139 on 05/07/20. For personal use only.

5. FIXED BED STRUCTURE AND HYDRODYNAMICS

To validate the computer generation of fixed beds, the radial void fraction profile caused by wall
effects (89) is most frequently compared with experiments or standard correlations. This is usu-
ally done for spheres, as experimental data are scarce and correlations virtually nonexistent for
nonspheres. Beyond voidage profiles, for cylinder-based shapes, the distribution of orientations
(25–27, 90) of the particles with respect to the tube axis can be assessed.
As computer generation methods continue to improve, confidence in them increases, and we
can consider using them for the development of new information. Correlations of ε(r) for, e.g.,
cylinders, rings, and multihole particles could be developed for use in formulas for effective pa-
rameter models. New features of packed bed structure, such as coordination number, can be in-
vestigated as well.
The influence of filling methods is being studied (91) in producing loose or tight packings.
Correlations of void fraction are needed for various packing states and different particle shapes.
Agrawal et al. (92) have used computer generation of sphere packs to investigate whether packing
defects observed at the top of the bed propagate through the bed, causing local flow variations.
They suggested that measures of packing nonisotropy should be considered, in addition to mea-
sures of void fraction.
Virtually all PRCFD simulations of packed beds report the axial pressure drop and the radial
profile of the axial velocity. These quantities are often compared with experiments for validation
of CFD methodology, as well as being highly relevant for the development of dispersion-type
reaction engineering models. Vollmari et al. (93) recently used PRCFD to assess pressure drop
correlations for packings of complex shapes.
Unfortunately, quality data inside a packed bed are hard to obtain. Some recent examples are
the magnetic resonance imaging studies of water flow of Robbins et al. (94) and Lovreglio et al.
(95). Using near-infrared tomography, Alzahrani et al. (96) visualized gas flow in beds of spheres.
Buwa and colleagues (69, 97) used particle image velocimetry, again for water as the working
fluid with spheres. All of the above authors carried out PRCFD simulations in parallel with the
experiments and obtained acceptable, if not perfect, agreement between them. There is a need for
more velocity data for the gas phase and nonspheres.
Most PRCFD simulations of fluid flows have been for beds of spheres (83, 95, 98). Dorai et al.
(99, 100) simulated flow through square ducts packed with spheres, cylinders, cubes, and tetrahedra

·.•�-
to obtain probability distribution functions of velocity and to study the influence of microstructure

www.annualreviews.org • CFD for Fixed Bed Reactor Design 3.11

, Review in Advance first posted

on March 9, 2020. (Changes may
still occur before final publication.)
 CH11CH03_Dixon ARjats.cls February 25, 2020 16:36

in the beds on pressure drop. Marek (101) simulated laminar flow in beds of Raschig rings by using
the IBM, whereas in later work (102) gas flow maldistribution was investigated for similar beds
of cylinders and rings. Pavlišič et al. (103) used PRCFD to develop pressure drop correlations
for beds of cylinders, trilobes, and quadrilobes. Finally, Moghaddam et al. (104) have examined
flow in beds of spheres and cylinders to contend that the usual azimuthal (angular) averaging for
deriving velocity profiles from PRCFD fails to capture the effects of local flow features such as
vortex regions.

6. FIXED BED HEAT, MASS, AND REACTION

Resolved particle CFD simulations are intended to study the interactions of transport phenomena
(heat and mass transfer and flow field) with reaction kinetics. These phenomena impact each other
simultaneously during the reactor operations. Often, chemical reactions are highly endothermic
Annu. Rev. Chem. Biomol. Eng. 2020.11. Downloaded from www.annualreviews.org

or exothermic. The heat generated or consumed affects the temperature of the flow field and
particles, and most transport parameters are also temperature dependent. The temperature profile
Access provided by 207.241.225.139 on 05/07/20. For personal use only.

within the reactor is also highly affected by the flow patterns. Clearly an accurate model of fixed
beds must consider all these variations and interactions. Therefore, many studies during the past
years (for example, 105) have addressed these issues.
Yang et al. (106) studied the effects of the fixed bed geometrical properties (e.g., particle size and
wall distance) on flow and temperature profile in packings of mono- and polydispersed spheres.
The authors showed that for packing of uniform spheres with fixed heat flux at the wall, the ve-
locity profile is higher and temperature is lower near the wall owing to high voidage. In packing
of polydispersed particles, with smaller particles near the wall, the voidage near the wall was de-
creased, and therefore the temperature gradient owing to wall effects was improved. Although this
method increased the radial heat transfer efficiency because of the smaller particles, pressure drop
was increased. This study shows applications of resolved particle CFD simulations for optimal
design. Zhang et al. (107) conducted a similar study for uniform randomly packed beds of cylin-
drical particles and showed similar flow and temperature gradients near the wall to the packing
of spheres. The authors also proposed an improved 2D homogeneous model by incorporating
the results of their 3D resolved particle model. Dong et al. (108) validated their resolved particle
CFD simulations using novel profile reactor measurements. They studied the radial heat transfer
for packings of spheres and rings. Although the simulations did not include the sampling capillary
used in the experimental measurements, very good agreement between the simulations and the
lab tests was found.
Das and coworkers (40) followed a different approach for studying the flow and heat transfer
in packing of spherical particles. The authors used fully resolved DNS with a sharp interface IBM.
The advantage of the proposed coupling approach is the ability to avoid the meshing difficulties
near the particle–particle and wall–particle contact areas. However, the mesh is not fully fitted
to the particles; despite that, good agreement for heat transfer and pressure drop with existing
correlations in the literature was found. Furthermore, the authors proposed new correlations for
wall-to-bed heat transfer.
Several recent studies have also used fixed bed CFD simulations to investigate large-scale im-
portant industrial reactions, such as ethylene epoxidation, maleic anhydride production, and steam
and dry reforming of methane. Karthik & Buwa (68) and Singhal et al. (109) studied fixed beds
under methane steam reforming conditions using resolved particle simulations. The latter group
studied spherical particles, whereas the former investigated several particle shapes, such as trilobe,
hollow cylinder, and 7-hole. Both studies included only a limited number of particles and a short

·.•�-
fixed bed length. Karthik & Buwa’s (68) study results show that shaped particles, such as trilobe

3.12 Dixon • Partopour

, Review in Advance first posted

on March 9, 2020. (Changes may
still occur before final publication.)
 CH11CH03_Dixon ARjats.cls February 25, 2020 16:36

and daisy, reduce the pressure drop and increase the dispersion in the methane steam reform-
ing fixed bed, whereas particles with hollow sections improve the conversion owing to reduction
in diffusion limitations. Singhal et al. (109) used the simulation to improve their 1D packed bed
model and applied it to packed bed chemical looping reforming.
Partopour & Dixon (110) studied ethylene partial oxidation to ethylene oxide using resolved
particle CFD. The study highlighted the failure of reduced microkinetics models under extreme
temperature and species gradients that form in 3D CFD simulations of fixed bed reactors.
Both Partopour & Dixon (111) and Dong et al. (112) conducted studies of n-butane partial oxi-
dation over iron molybdate catalyst. The former group focused on effects of temperature gradient
on selectivity and surface oxygen profiles along a long bed of 800 spherical particles and showed
that surface oxygen is highly correlated with temperature profile inside the fluid and solid particles
and controls the reaction selectivity. The latter group studied the fixed bed behavior filled with
Raschig rings and validated their study with profile reactor measurements. Furthermore, they de-
Annu. Rev. Chem. Biomol. Eng. 2020.11. Downloaded from www.annualreviews.org

veloped an optimum wall temperature profile using their fixed bed CFD model and incorporated
the temperature profile into their experimental setup whereby, as a result of the new temperature
Access provided by 207.241.225.139 on 05/07/20. For personal use only.

profile, selectivity was improved by 3% and the hot spot was removed from the bed.
Wehinger et al. (113, 114) studied dry reforming of methane using a similar methodology.
In their first study, they showed the feasibility of coupling detailed surface kinetics with fixed bed
CFD simulations, whereas in the second study they presented the effect of different particle shapes
in the system. They showed that packing of the cylindrical particles resulted in superior yield and
conversion compared with rings and spheres.
Behnam & Dixon (115) investigated local carbon formation in fixed beds under steam reform-
ing conditions, which is one of the main industrial issues in fixed bed reactors. They showed the
carbon formation is highly location dependent, where the maximum rate is on the outer surface
of the particles and correlated with temperature profile in the particles. Alzahrani et al. (116) used
a combination of PRCFD and near-infrared tomography to study the deactivation of CO in the
preferential oxidation of CO and also found that deactivation was highest at the pellet surface.

7. PARTICLE-RESOLVED COMPUTATIONAL FLUID DYNAMICS

AND CHEMICAL REACTION ENGINEERING MODELS
Given the high cost of PRCFD simulations, their routine use as design models may be some
distance away, and their use for control or optimization is not likely to be feasible for a long time,
if ever. A very significant role for these methods is therefore to provide baseline data to improve
or develop parameters for use in simpler chemical reaction engineering models.
Particle-to-fluid heat and mass transfer coefficients are critical to models of catalytic reacting
systems. Atmakidis & Kenig (117, 118) simulated dissolution of benzoic acid spheres in water and
sublimation of naphthalene spheres in air. They showed that PRCFD gave good agreement with
literature correlations. Bale et al. (119) used DNS simulations at Re < 100 to study the influence
of the tube wall on the local mass transfer coefficient and found that it was higher for particles
near the tube wall, owing to the higher velocity there. In a second study (120), they showed that
N affects the influence of exit and entrance regions on the mass transfer coefficient.
Kruggel-Emden and coworkers (121, 122) used LBM to study heat transfer; they used favorable
comparisons of the heat transfer coefficient for benchmark simulations of single and small clusters
of spheres to confirm their thermal lattice approach. A new correlation was derived for low Re
(20–100) and for random packings of spheres with high void fractions (0.6–1). Buist et al. (123)
used IBM to obtain the heat transfer coefficient for some semistructured arrays of spheres to

·.•�-
compare with their experimental studies. Sun et al. (124) also used a variation of IBM for spheres at

www.annualreviews.org • CFD for Fixed Bed Reactor Design 3.13

, Review in Advance first posted

on March 9, 2020. (Changes may
still occur before final publication.)
 CH11CH03_Dixon ARjats.cls February 25, 2020 16:36

Re < 100 and ε = 0.5–0.9 to obtain a new correlation and develop a consistent two-fluid model.
Singhal et al. (125, 126) used PRCFD to derive new heat transfer coefficient correlations for
spheres and cylinders without wall effects at ε ≈ 0.4 (spheres) and 0.5 (cylinders of aspect ratio 4),
again at Re < 100. Based on the existing studies, there is a need for work at higher Re, for more
complex particle shapes, and for a focus on dense packing typical of fixed beds.
The use of PRCFD for bed-scale longitudinal and transverse dispersion of mass in fixed beds
has been less studied. This may be due to the need for longer beds of particles to obtain meaningful
statistics of the dispersion processes. Jourak et al. (127) built a network model based on thousands
of spherical particles. Their calculations simulated a breakthrough experiment for longitudinal
dispersion and a coated wall dissolution experiment for transverse dispersion at low Peclet number
values. Good agreement with experimental results reported in the literature was obtained.
Dixon & Medeiros (128) studied transverse dispersion with a focus on wall effects in narrow
tubes of spheres. Simulations of dispersion from a coated wall were run for 87 < Re < 870. The ob-
Annu. Rev. Chem. Biomol. Eng. 2020.11. Downloaded from www.annualreviews.org

jective was to see what effective parameter models were needed to account for the experimentally
observed decrease in transverse dispersion near the tube wall. A model with constant dispersion
Access provided by 207.241.225.139 on 05/07/20. For personal use only.

coefficient was inadequate to describe the radial concentration profile. Addition of a wall mass
transfer coefficient improved agreement somewhat, but the best results were given by a three-
parameter model of a radially varying dispersion coefficient. Extensions of these approaches to
nonspherical particles and to longer tubes could yield interesting results.
Determination of reliable correlations for the effective radial thermal conductivity and the
wall heat transfer coefficient in low-N fixed beds has been a long-standing problem. The effective
radial thermal conductivity is usually split into static and dynamic terms. Dixon et al. (129) used
PRCFD in an annular bed of spheres with N = 4 to test literature correlations for the radial
variation of stagnant thermal conductivity. Mathey (130) advocated the use of traditional models
with CFD input. An upscaling approach was used in which dispersive kinetic energy was obtained
from k-ε CFD and used to model the thermal dispersion term. Channeling effects near the walls
were also considered. The needed closures for the macroscopic equations were obtained from
microscopic CFD of 3D periodic arrays of spheres and cylinders. Thiagalingam et al. (131) studied
physical mechanisms contributing to wall heat transfer using 3D periodic packings of spheres and
rings. They used CFD to get turbulent and dispersive mixing contributions to the overall heat
transfer coefficient in a 1D model. In a subsequent contribution, Thiagalingam & Sagaut (132)
used a similar strategy to derive a four-region formula for effective radial thermal conductivity,
introducing the idea of a disrupted thermal boundary layer together with a channeling effect.
Dixon (133) used PRCFD simulation of steam methane reforming in a tube of N = 5.96
spheres to provide benchmark data to evaluate levels of model complexity, i.e., 1D and 2D ef-
fective medium models. The 1D model could not reproduce axial profiles without tuning the
particle–fluid heat and mass transfer coefficients. The 2D model gave good agreement if local
void fractions and axial velocity profiles were included.
PRCFD can also be crucial to the development of new model approaches to radial heat transfer.
Behnam et al. (134) have suggested a 2D pseudocontinuum model in which the effective thermal
dispersion contribution is replaced by direct inclusion of axial and radial velocity components
obtained from PRCFD, along with radially varying stagnant thermal conductivity (129). A differ-
ent new approach by Barreto and colleagues (135) develops a two-region, 2D model with a heat
transfer coefficient between wall and core channels and includes a true wall-to-fluid heat transfer
coefficient. They used PRCFD simulations on regular sphere arrays to evaluate these parameters
for convection (136) and gave a new model for conduction under stagnant conditions (137). Sim-
ilarly, Hamzah et al. (11) proposed porosity-dependent, segmented-continuum approaches, using

·.•�-
PRCFD to obtain dispersion parameters.

3.14 Dixon • Partopour

, Review in Advance first posted

on March 9, 2020. (Changes may
still occur before final publication.)
 CH11CH03_Dixon ARjats.cls February 25, 2020 16:36

8. NEW APPLICATIONS
Two relatively new types of packed beds in which PRCFD can play a role are micro packed bed
reactors and packed foam reactors. Micro packed bed reactors are a configuration of small catalyst
particles packed into the channels of metallic honeycomb matrices to enhance the heat transfer
properties and increase the catalyst load per reactor volume of structured reactors. In packed foam
reactors, either the empty pores of highly conductive metallic open-cell foams are filled with small
catalytic particles or catalytic particles are made out of the foam. The presence of the particles
increases radial and axial thermal dispersion to supplement the high thermal conductivity of the
foam, as well as to increase the catalyst inventory in the reactor.
Maestri and coworkers (138, 139) have used PRCFD as the most detailed simulation and high-
est level in their proposed hierarchical approach to reaction model development. In the first study
(138), detailed 3D simulations of a packed honeycomb channel provided suitable correlations for
gas-to-particle mass and heat transfer in micro packed beds. Good agreement was obtained with a
Annu. Rev. Chem. Biomol. Eng. 2020.11. Downloaded from www.annualreviews.org

1D heterogeneous effective parameter model showing the potential of hierarchical derivation of

Access provided by 207.241.225.139 on 05/07/20. For personal use only.

transport parameters using PRCFD. A follow-up study (139) treated the radial heat transfer prop-
erties, which are enhanced in a packed microchannel, and showed that a literature correlation for
the overall heat transfer coefficient U was applicable to a single packed channel of the honeycomb
matrix. This was then used in a 2D model of the matrix to simulate the partial oxidation of o-xylene
to phthalic anhydride, showing the use of the hierarchical approach when PRCFD would be too
expensive.
Kolaczkowski et al. (140) suggested using catalyst pellets made out of actual foams as supports.
Foam pellets have large open regions that allow reactants into the pellet, have lower pressure drop
owing to higher void fraction, and increase heat transfer by both conduction and radiation. Das
et al. (141) used the IBM to resolve flow around cubic porous particles, whereas flow through
the particle was represented using closure terms. Wehinger et al. (142) took a similar approach to
resolve fixed bed structure using FV CFD, whereas they used a PMM for flow and heat transfer
in the foam cubes.
CFD has been used to resolve the structure of tubes containing open-cell foams; Dong et al.
(143) provide an example with a comparison to spatial profiling data. Foams with particles filling
the pores are a newer concept, and the available studies are experimental (144). This would ap-
pear to be an excellent opportunity to use PRCFD in combination with resolved-foam CFD to
investigate the new configurations.

9. CONCLUDING REMARKS
The application of CFD to fixed bed modeling and design can be divided into two directions:
PMM and PRCFD. The former is well accepted, although more extensive tests of its predictions
would be useful. Most current research interest lies with PRCFD. Our review of the literature
developments in the last 10 years or so reveals some areas in which satisfactory progress has been
made and some in which more remains to be done.
One theme that emerges is the need for the field to move away from spheres to consider the
complex particle shapes used in practice. This has been done in computer generation of packings,
but problems remain in meshing and handling of contact points. Another area is the desire to
include more realistic microkinetics reaction schemes; promising methods have been put forward,
but often these lead to either restrictions in the reactor model or excessive computation time.
Effective medium models must be used for the catalyst pellets; resolved-pore calculations are out

·.•�-
of reach at present. Simulations of larger numbers of particles to approach industrial scales of tube

www.annualreviews.org • CFD for Fixed Bed Reactor Design 3.15

, Review in Advance first posted

on March 9, 2020. (Changes may
still occur before final publication.)
 CH11CH03_Dixon ARjats.cls February 25, 2020 16:36

diameter and length, and improved turbulence approaches for high flow rates, also remain goals
for the future.
Considerable activity and progress exist in using PRCFD to simulate reacting systems and to
provide data and correlations for effective parameter models. These efforts will continue, with
new applications such as foams and new methods such as radiation. It is not clear that PRCFD
can replace effective parameter models for design in the near future, but it can contribute to new
reactor development, new and better catalyst shapes, and improved understanding of fixed bed
reactors.

DISCLOSURE STATEMENT
The authors are not aware of any affiliation, memberships, funding, or financial holdings that
might be perceived as affecting the objectivity of this review.
Annu. Rev. Chem. Biomol. Eng. 2020.11. Downloaded from www.annualreviews.org
Access provided by 207.241.225.139 on 05/07/20. For personal use only.

ACKNOWLEDGMENTS
This material is based upon work supported by the American Chemical Society Petroleum Re-
search Fund.

LITERATURE CITED
1. Dixon AG, Nijemeisland M, Stitt EH. 2006. Packed tubular reactor modelling and catalyst design using
computational fluid dynamics. Adv. Chem. Eng. 31:307–89
2. Jurtz N, Kraume M, Wehinger GD. 2019. Advances in fixed-bed reactor modeling using particle-
resolved computational fluid dynamics (CFD). Rev. Chem. Eng. 35:139–90
3. Dixon AG, Nijemeisland M. 2001. CFD as a design tool for fixed-bed reactors. Ind. Eng. Chem. Res.
40:5246–54
4. Wehinger GD, Kraume M. 2017. CFD als Designtool für Festbettreaktoren mit kleinem Rohr-zu-
Pelletdurchmesser-Verhältnis: Heute oder in Zukunft? Chem. Ing. Tech. 89:447–53
5. Partopour B, Dixon AG. 2019. 110th anniversary: commentary: CFD as a modeling tool for fixed bed
reactors. Ind. Eng. Chem. Res. 58:5733–36
6. Wang Y, Chen J, Larachi F. 2013. Modelling and simulation of trickle-bed reactors using computational
fluid dynamics: a state-of-the-art review. Can. J. Chem. Eng. 91:136–80
7. Jakobsen HA, Lindborg H, Handeland V. 2002. A numerical study of the interactions between viscous
flow, transport and kinetics in fixed bed reactors. Comput. Chem. Eng. 26:333–57
8. Wu CY, Ferng YM, Chieng CC, Liu CC. 2010. Investigating the advantages and disadvantages of re-
alistic approach and porous approach for closely packed pebbles in CFD simulation. Nucl. Eng. Des.
240:1151–59
9. Gao X, Zhu Y-P, Luo Z-H. 2011. CFD modeling of gas flow in porous medium and catalytic coupling
reaction from carbon monoxide to diethyl oxalate in fixed-bed reactors. Chem. Eng. Sci. 66:6028–38
10. Zhuang Y-Q, Gao X, Zhu Y-P, Luo Z-H. 2012. CFD modeling of methanol to olefins process in a
fixed-bed reactor. Powder Technol. 221:419–30
11. Hamzah AB, Ookawara S, Yoshikawa S, Matsumoto H. 2019. CFD modelling of mass and heat dispersion
in sphere fixed bed with porosity-dependent segmented-continuum approaches. Chem. Eng. Res. Des.
141:93–114
12. Soppe W. 1990. Computer simulation of random packings of hard spheres. Powder Technol. 62:189–96
13. Freund H, Zeiser T, Huber F, Klemm E, Brenner G, et al. 2003. Numerical simulations of single
phase reacting flows in randomly packed fixed-bed reactors and experimental validation. Chem. Eng.
Sci. 58:903–10

·.•�-
14. Freund H, Bauer J, Zeiser T, Emig G. 2005. Detailed simulation of transport processes in fixed-beds.
Ind. Eng. Chem. Res. 44:6423–34

3.16 Dixon • Partopour

, Review in Advance first posted

on March 9, 2020. (Changes may
still occur before final publication.)
 CH11CH03_Dixon ARjats.cls February 25, 2020 16:36

15. Eppinger T, Jurtz N, Aglave R. 2014. Automated workflow for spatially resolved packed bed reactors
with spherical and non-spherical particles. In Proceedings of 10th International Conference on CFD in Oil
& Gas, Metallurgical and Process Industries, Trondheim, June 17–19. Trondheim, Nor.: SINTEF
16. Caulkin R, Tian W, Pasha M, Hassanpour A, Jia X. 2015. Impact of shape representation schemes used
in discrete element modelling of particle packing. Comput. Chem. Eng. 76:160–69
17. Wachs A, Girolami L, Vinay G, Ferrer G. 2012. Grains3D, a flexible DEM approach for particles of
arbitrary convex shape—part I: numerical model and validations. Powder Technol. 224:374–89
18. Rakotonirina AD, Wachs A. 2018. Grains3D, a flexible DEM approach for particles of arbitrary convex
shape—part II: parallel implementation and scalable performance. Powder Technol. 324:18–35
19. Rakotonirina AD, Delenne JY, Radjai F, Wachs A. 2018. Grains3D, a flexible DEM approach for par-
ticles of arbitrary convex shape—part III: extension to non-convex particles modelled as glued convex
particles. Comp. Part. Mech. 324:1–30
20. Seelen LJH, Padding JT, Kuipers JAM. 2018. A granular discrete element method for arbitrary convex
particle shapes: method and packing generation. Chem. Eng. Sci. 189:84–101
Annu. Rev. Chem. Biomol. Eng. 2020.11. Downloaded from www.annualreviews.org

21. Marigo M, Stitt EH. 2015. Discrete element method (DEM) for industrial applications: comments on
calibration and validation for the modelling of cylindrical pellets. Kona Powder Part. J. 32:236–52
Access provided by 207.241.225.139 on 05/07/20. For personal use only.

22. Yan Z, Wilkinson SK, Stitt EH, Marigo M. 2015. Discrete element modelling (DEM) input parameters:
understanding their impact on model predictions using statistical analysis. Comp. Part. Mech. 2:283–99
23. Farsi A, Xiang J, Latham JP, Carlsson M, Stitt EH, Marigo M. 2017. Simulation and characterisation
of packed columns for cylindrical catalyst supports and other complex-shaped bodies. In Proceedings
of the 7th International Conference on Discrete Element Methods, pp. 397–406. Springer Proc. Phys. 188.
Singapore: Springer
24. Marek M. 2013. Numerical generation of a fixed bed structure. Chem. Process Eng. 34:347–59
25. Niegodajew P, Marek M. 2016. Analysis of orientation distribution in numerically generated random
packings of Raschig rings in a cylindrical container. Powder Technol. 297:193–201
26. Boccardo G, Augier F, Haroun Y, Ferre D, Marchisio DL. 2015. Validation of a novel open-source
work-flow for the simulation of packed-bed reactors. Chem. Eng. J. 279:809–20
27. Partopour B, Dixon AG. 2017. An integrated workflow for resolved-particle packed bed models with
complex particle shapes. Powder Technol. 322:258–72
28. Moghaddam EM, Foumeny EA, Stankiewicz AI, Padding JT. 2018. Rigid body dynamics algorithm
for modeling random packing structures of nonspherical and nonconvex pellets. Ind. Eng. Chem. Res.
57:14988–5007
29. Fernengel J, Habla F, Hinrichsen O. 2018. Scripting as an approach to automated CFD simulation for
packed bed catalytic reactor modeling. Chem. Ing. Tech. 90:685–89
30. Fleischlen S, Wehinger GD. 2019. Synthetic packed-bed generation for CFD simulations: Blender ver-
sus STAR-CCM+. ChemEngineering 3:52
31. Baker MJ, Daniels S, Young PG, Tabor GR. 2014. Investigation of flow through a computationally
generated packed column using CFD and additive layer manufacturing. Comput. Chem. Eng. 67:159–65
32. Wu Y, Hou Q, Yu A. 2019. Pore-scale study of fluid flow and drag force in randomly packed beds of
different porosities. Ind. Eng. Chem. Res. 58:5041–53
33. Sengar A, Kuipers JAM, van Santen RA, Padding JT. 2019. Towards a particle based approach for mul-
tiscale modeling of heterogeneous catalytic reactors. Chem. Eng. Sci. 198:184–97
34. Kravets B, Rosemann T, Reinecke SR, Kruggel-Emden H. 2019. A new drag force and heat transfer
correlation derived from direct numerical LBM-simulations of flown through particle packings. Powder
Technol. 345:438–56
35. Li X, Cai J, Xin F, Huai X, Guo J. 2013. Lattice Boltzmann simulation of endothermal catalytic reaction
in catalyst porous media. Appl. Therm. Eng. 50:1194–200
36. Touitou J, Aiouache F, Burch R, Douglas R, Hardacre C et al. 2014. Evaluation of an in situ spatial reso-
lution instrument for fixed beds through the assessment of the invasiveness of probes and a comparison
with a micro-kinetic model. J. Catal. 319:239–46
37. Guo X, Dai R. 2010. Numerical simulation of flow and heat transfer in a random packed bed. Particuology

·.•�-
8:293–99

www.annualreviews.org • CFD for Fixed Bed Reactor Design 3.17

, Review in Advance first posted

on March 9, 2020. (Changes may
still occur before final publication.)
 CH11CH03_Dixon ARjats.cls February 25, 2020 16:36

38. Finn J, Apte SV. 2013. Relative performance of body fitted and fictitious domain simulations of flow
through fixed packed beds of spheres. Int. J. Multiphase Flow 56:54–71
39. Sulaiman M, Hammouti A, Climent E, Wachs A. 2019. Coupling the fictitious domain and sharp inter-
face methods for the simulation of convective mass transfer around reactive particles: towards a reactive
Sherwood number correlation for dilute systems. Chem. Eng. Sci. 198:334–51
40. Das S, Deen NG, Kuipers JAM. 2017. A DNS study of flow and heat transfer through slender fixed-bed
reactors randomly packed with spherical particles. Chem. Eng. Sci. 160:1–19
41. Yuan B, Xu J, Mao Z, Zhang Y, Yang C. 2019. Particle-resolved simulation of packed beds by non-body
conforming locally refined orthogonal hexahedral mesh. Chin. J. Chem. Eng. 27:2635–42
42. Guardo A, Coussirat M, Larrayoz MA, Recasens F, Egusquiza E. 2004. CFD flow and heat transfer in
nonregular packings for fixed bed equipment design. Ind. Eng. Chem. Res. 43:7049–56
43. Ookawara S, Kuroki M, Street D, Ogawa K. 2007. High-fidelity DEM-CFD modeling of packed bed
reactors for process intensification. In Proceedings of European Congress of Chemical Engineering (ECCE-6),
Copenhagen, Sept. 16–20, Pap. no. 1105. Kongens Lyngby, Den.: Tech. Univ. Den.
Annu. Rev. Chem. Biomol. Eng. 2020.11. Downloaded from www.annualreviews.org

44. Eppinger T, Seidler K, Kraume M. 2011. DEM-CFD simulations of fixed bed reactors with small tube
to particle diameter ratios. Chem. Eng. J. 166:324–31
Access provided by 207.241.225.139 on 05/07/20. For personal use only.

45. Dixon AG, Nijemeisland M, Stitt EH. 2013. Systematic mesh development for 3D CFD simulation of
fixed beds: contact points study. Comput. Chem. Eng. 48:135–53
46. Bu SS, Yang J, Zhou M, Li SY, Wang QW, Guo ZX. 2014. On contact point modifications for forced
convective heat transfer analysis in a structured packed bed of spheres. Nucl. Eng. Des. 270:21–33
47. Rebughini S, Cuoci A, Maestri M. 2016. Handling contact points in reactive CFD simulations of het-
erogeneous catalytic fixed bed reactors. Chem. Eng. Sci. 141:240–49
48. Kodam M, Bharadwaj R, Curtis J, Hancock B, Wassgren C. 2010. Cylindrical object contact detection
for use in discrete element method simulations. Part I—contact detection algorithms. Chem. Eng. Sci.
65:5852–62
49. Wehinger GD, Füterer C, Kraume M. 2017. Contact modifications for CFD simulations of fixed-bed
reactors: cylindrical particles. Ind. Eng. Chem. Res. 56:87–99
50. Maestri M, Cuoci A. 2013. Coupling CFD with detailed microkinetic modeling in heterogeneous catal-
ysis. Chem. Eng. Sci. 96:106–17
51. Dixon AG, Taskin ME, Nijemeisland M, Stitt EH. 2010. CFD method to couple three-dimensional
transport and reaction inside catalyst particles to the fixed bed flow field. Ind. Eng. Chem. Res. 49:9012–
25
52. Matera S, Maestri M, Cuoci A, Reuter K. 2014. Predictive-quality surface reaction chemistry in real
reactor models: integrating first-principles kinetic Monte Carlo simulations into computational fluid
dynamics. ACS Catal. 4:4081–92
53. Partopour B, Dixon AG. 2016. Computationally efficient incorporation of microkinetics into resolved-
particle CFD simulations of fixed-bed reactors. Comput. Chem. Eng. 88:126–34
54. Partopour B, Paffenroth RC, Dixon AG. 2018. Random forests for mapping and analysis of microkinetics
models. Comput. Chem. Eng. 115:286–94
55. Klingenberger M, Hirsch O, Votsmeier M. 2017. Efficient interpolation of precomputed kinetic data
employing reduced multivariate Hermite Splines. Comput. Chem. Eng. 98:21–30
56. Rebughini S, Cuoci A, Dixon AG, Maestri M. 2017. Cell agglomeration algorithm for coupling mi-
crokinetic modeling and steady-state CFD simulations of catalytic reactors. Comput. Chem. Eng. 97:175–
82
57. Blasi JM, Kee RJ. 2016. In situ adaptive tabulation (ISAT) to accelerate transient computational fluid
dynamics with complex heterogeneous chemical kinetics. Comput. Chem. Eng. 84:36–42
58. Dong Y, Keil FJ, Korup O, Rosowski F, Horn R. 2016. Effect of the catalyst pore structure on fixed-bed
reactor performance of partial oxidation of n-butane: a simulation study. Chem. Eng. Sci. 142:299–309
59. Wehinger GD, Klippel F, Kraume M. 2017. Modeling pore processes for particle-resolved CFD simu-
lations of catalytic fixed-bed reactors. Comput. Chem. Eng. 101:11–22
60. Donaubauer PJ, Hinrichsen O. 2018. A Monte-Carlo-based sensitivity analysis of multicomponent dif-

·.•�-
fusion in porous catalysts. Chem. Eng. Sci. 185:282–91

3.18 Dixon • Partopour

, Review in Advance first posted

on March 9, 2020. (Changes may
still occur before final publication.)
 CH11CH03_Dixon ARjats.cls February 25, 2020 16:36

61. Partopour B, Troupel A, Dixon AG. 2019. Flux-dependent anisotropic pellet diffusivity in particle-
resolved CFD simulations of fixed beds. Chem. Eng. Sci. 198:224–34
62. Bufe A, Klee M, Wehinger GD, Turek T. 2017. 3D modeling of a catalyst layer with transport pore for
Fischer-Tropsch synthesis. Chem. Ing. Tech. 89:1385–90
63. Partopour B, Dixon AG. 2018. Resolved-pore simulation of CO oxidation on Rh/Al2 O3 in a catalyst
layer. ChemEngineering 2:2
64. Kolaczkowski ST, Chao R, Awdry S, Smith A. 2007. Application of a CFD code (Fluent) to formu-
late models of catalytic gas phase reactions in porous catalyst pellets. Chem. Eng. Res. Des. 85:1539–
52
65. Dixon AG, Taskin ME, Nijemeisland M, Stitt EH. 2007. 3D CFD simulations of steam reforming with
resolved intraparticle reaction and gradients. Chem. Eng. Sci. 62:4963–66
66. Taskin ME, Troupel A, Dixon AG, Nijemeisland M, Stitt EH. 2010. Flow, transport and reaction inter-
actions for cylindrical particles with strongly endothermic reactions. Ind. Eng. Chem. Res. 49:9026–37
67. Dixon AG, Boudreau J, Rocheleau A, Troupel A, Taskin ME et al. 2012. Flow, transport and reaction
Annu. Rev. Chem. Biomol. Eng. 2020.11. Downloaded from www.annualreviews.org

interactions in shaped cylindrical particles for methane steam reforming. Ind. Eng. Chem. Res. 51:15839–
54
Access provided by 207.241.225.139 on 05/07/20. For personal use only.

68. Karthik GM, Buwa VV. 2018. Particle-resolved simulations of methane steam reforming in multilayered
packed beds. AIChE J. 64:4162–76
69. Karthik GM, Thaker AH, Buwa VV. 2019. Particle-resolved simulations of catalytic fixed bed reactors:
comparison of turbulent models, LES and PIV measurements. Powder Technol. 361:474–89
70. Peng W, Xu M, Huai X, Liu Z. 2016. 3D CFD simulations of acetone hydrogenation in randomly packed
beds for an isopropanol-acetone-hydrogen chemical heat pump. Appl. Therm. Eng. 94:238–48
71. Maffei T, Gentile G, Rebughini S, Bracconi M, Manelli F, et al. 2016. A multiregion operator-splitting
CFD approach for coupling microkinetic modeling with internal porous transport in heterogeneous
catalytic reactors. Chem. Eng. J. 283:1392–404
72. Bracconi M, Maestri M, Cuoci A. 2017. In situ adaptive tabulation for the CFD simulation of heteroge-
neous reactors based on operator-splitting algorithm. AIChE J. 63:95–104
73. Corbetta M, Manenti F, Visconti CG. 2014. CATalytic—Post Processor (CAT-PP): a new methodology
for the CFD-based simulation of highly diluted reactive heterogeneous systems. Comput. Chem. Eng.
60:76–85
74. Hettel M, Diehm C, Deutschmann O. 2015. Numerical simulation of a structured catalytic methane
reformer by DUO: the new computational interface for OpenFOAM® and DETCHEMTM . Catal.
Today 258:230–40
75. Dybbs A, Edwards RV. 1984. A new look at porous media fluid mechanics—Darcy to turbulent. In Fun-
damentals of Transport Phenomena in Porous Media, ed. J Bear, MY Corapcioglu, pp. 201–56. Dordrecht,
Neth.: Martinus Nijhoff
76. Seguin D, Montillet A, Comiti J, Huet F. 1998. Experimental characterization of flow regimes in various
porous media—II: transition to turbulent regime. Chem. Eng. Sci. 53:3897–909
77. Lee JJ, Yoon SJ, Park GC, Lee WJ. 2007. Turbulence-induced heat transfer in PBMR core using LES
and RANS. J. Nucl. Sci. Technol. 44:985–96
78. Dixon AG, Taskin ME, Nijemeisland M, Stitt EH. 2011. Systematic mesh development for 3D CFD
simulation of fixed beds: single sphere study. Comput. Chem. Eng. 35:1171–85
79. Shams A, Roelofs F, Komen EMJ, Baglietto E. 2014. Large eddy simulation of a randomly stacked
nuclear pebble bed. Comput. Fluids 96:302–21
80. Shams A, Roelofs F, Komen EMJ, Baglietto E. 2015. Improved delayed detached eddy simulation of a
randomly stacked nuclear pebble bed. Comput. Fluids 122:12–25
81. Guardo A, Coussirat M, Larrayoz MA, Recasens F, Egusquiza E. 2005. Influence of the turbulence model
in CFD modeling of wall-to-fluid heat transfer in packed beds. Chem. Eng. Sci. 60:1733–42
82. Coussirat M, Guardo A, Mateos B, Egusquiza E. 2007. Performance of stress-transport models in the
prediction of particle-to-fluid heat transfer in packed beds. Chem. Eng. Sci. 62:6897–907
83. Jiang L, Liu H, Suo S, Xie M, Wu D, Bai M. 2018. Pore-scale simulation of flow and turbulence char-

·.•�-
acteristics in three-dimensional randomly packed beds. Powder Technol. 338:197–210

www.annualreviews.org • CFD for Fixed Bed Reactor Design 3.19

, Review in Advance first posted

on March 9, 2020. (Changes may
still occur before final publication.)
 CH11CH03_Dixon ARjats.cls February 25, 2020 16:36

84. Behnam M, Dixon AG, Wright PM, Nijemeisland M, Stitt EH. 2012. Comparison of CFD simulations
to experiment under methane steam reforming reacting conditions. Chem. Eng. J. 207–8:690–700
85. Wehinger GD, Kraume M, Berg V, Korup O, Mette K, et al. 2016. Investigating dry reforming of
methane with spatial reactor profiles and particle-resolved CFD simulations. AIChE J. 62:4436–52
86. Wehinger GD. 2019. Radiation matters in fixed-bed CFD simulations. Chem. Ing. Tech. 91:583–91
87. Wehinger GD, Flaischlen S. 2019. Computational fluid dynamics modeling of radiation in a steam
methane reforming fixed-bed reactor. Ind. Eng. Chem. Res. 58:14410–23
88. Partopour B, Dixon AG. 2019. Integrated multiscale modeling of fixed bed reactors: Studying the reactor
under dynamic reaction conditions. Chem. Eng. J. 377:119738
89. Zobel N, Eppinger T, Behrendt F, Kraume M. 2012. Influence of the wall structure on the void fraction
distribution in packed beds. Chem. Eng. Sci. 71:212–19
90. Caulkin R, Jia X, Xu C, Fairweather M, Williams RA, et al. 2009. Simulations of structures in packed
columns and validation by X-ray tomography. Ind. Eng. Chem. Res. 48:202–13
91. Fernengel J, von Seckendorff J, Hinrichsen O. 2018. Influence of cylinder-to-particle diameter ratio
Annu. Rev. Chem. Biomol. Eng. 2020.11. Downloaded from www.annualreviews.org

and filling speed on bed porosity of random packed beds of spheres. In Proceedings of the 28th European
Symposium on Computer Aided Process Engineering, pp. 97–102. Amsterdam: Elsevier. https://doi.org/10.
Access provided by 207.241.225.139 on 05/07/20. For personal use only.

1016/B978-0-444-64235-6.50019-X
92. Agrawal N, Nair P, Pöschel T, Roy S. 2019. Isotropy of sphere packings in a cylindrical confinement.
Chem. Eng. J. 377:119820
93. Vollmari K, Oschmann T, Wirtz S, Kruggel-Emden H. 2015. Pressure drop investigations in packings
of arbitrary shaped particles. Powder Technol. 271:109–24
94. Robbins DJ, El-Bachir MS, Gladden LF, Cant RS, von Harbou E. 2012. CFD modeling of single-phase
flow in a packed bed with MRI validation. AIChE J. 58:3904–15
95. Lovreglio P, Das S, Buist KA, Peters EAJF, Pel L, Kuipers JAM. 2018. Experimental and numerical
investigation of structure and hydrodynamics in packed beds of spherical particles. AIChE J. 64:1896–
907
96. Alzahrani F, Aldehani M, Rusi H, McMaster M, Abreu Fernandes DL, et al. 2015. Gas flow visualization
in low aspect ratio packed beds by three-dimensional modeling and near-infrared tomography. Ind. Eng.
Chem. Res. 54:12714–29
97. Thaker AH, Karthik GM, Buwa VV. 2019. PIV measurements and CFD simulations of the particle-scale
flow distribution in a packed bed. Chem. Eng. J. 374:189–200
98. Mohanty R, Mohanty S, Mishra BK. 2016. Study of flow through a packed bed using discrete element
method and computational fluid dynamics. J. Taiwan Inst. Chem. E 63:71–80
99. Dorai F, Rolland M, Wachs A, Marcoux M, Climent E. 2015. CFD simulations of flow in random packed
beds of spheres and cylinders: analysis of the velocity field. In Progress in Applied CFD: Selected Papers from
10th International Conference on Computational Fluid Dynamics in the Oil & Gas, Metallurgical and Process
Industries, Vol. 1, ed. JE Olsen, ST Johansen, pp. 143–48. Oslo, Nor.: SINTEF Acad. http://hdl.handle.
net/11250/2464782
100. Dorai F, Teixeira CM, Rolland M, Climent E, Marcoux M, Wachs A. 2015. Fully resolved simula-
tions of the flow through a packed bed of cylinders: effect of size distribution. Chem. Eng. Sci. 129:180–
92
101. Marek M. 2017. Numerical simulation of a gas flow in a real geometry of random packed bed of Raschig
rings. Chem. Eng. Sci. 161:382–93
102. Marek M. 2019. Gas flow maldistribution in random packed beds of non-spherical particles—a CFD
study. Chem. Eng. Sci. 197:296–305
103. Pavlišič A, Ceglar R, Poharand A, Likozar B. 2018. Comparison of computational fluid dynamics (CFD)
and pressure drop correlations in laminar flow regime for packed bed reactors and columns. Powder
Technol. 328:130–39
104. Moghaddam EM, Foumeny EA, Stankiewicz AI, Padding JT. 2019. Fixed bed reactors of non-spherical
pellets: importance of heterogeneities and inadequacy of azimuthal averaging. Chem. Eng. Sci. X 1:100006
105. Augier F, Idoux F, Delenne J. 2010. Numerical simulations of transfer and transport properties inside

·.•�-
packed beds of spherical particles. Chem. Eng. Sci. 65:1055–64

3.20 Dixon • Partopour

, Review in Advance first posted

on March 9, 2020. (Changes may
still occur before final publication.)
 CH11CH03_Dixon ARjats.cls February 25, 2020 16:36

106. Yang J, Wu J, Zhou L, Wang Q. 2016. Computational study of fluid flow and heat transfer in composite
packed beds of spheres with low tube to particle diameter ratio. Nucl. Eng. Des. 300:85–96
107. Zhang M, Dong H, Geng Z. 2018. Computational study of flow and heat transfer in fixed beds with
cylindrical particles for low tube to particle diameter ratios. Chem. Eng. Res. Des. 132:149–61
108. Dong Y, Sosna B, Korup O, Rosowski F, Horn R. 2017. Investigation of radial heat transfer in a fixed-bed
reactor: CFD simulations and profile measurements. Chem. Eng. J. 317:204–14
109. Singhal A, Cloete S, Quinta-Ferreira R, Amini S. 2017. Multiscale modelling of packed bed chemical
looping reforming. Energy Procedia 136:349–55
110. Partopour B, Dixon AG. 2017. Resolved-particle fixed bed CFD with microkinetics for ethylene oxida-
tion. AIChE J. 63:85–94
111. Partopour B, Dixon AG. 2018. n-Butane partial oxidation in a fixed bed: a resolved particle computational
fluid dynamics simulation. Can. J. Chem. Eng. 96:1946–56
112. Dong Y, Geske M, Korup O, Ellenfeld N, Rosowski F et al. 2018. What happens in a catalytic fixed-
bed reactor for n-butane oxidation to maleic anhydride? Insights from spatial profile measurements and
Annu. Rev. Chem. Biomol. Eng. 2020.11. Downloaded from www.annualreviews.org

particle resolved CFD simulations. Chem. Eng. J. 350:799–811

113. Wehinger GD, Eppinger T, Kraume M. 2015. Evaluating catalytic fixed-bed reactors for dry reforming
Access provided by 207.241.225.139 on 05/07/20. For personal use only.

of methane with detailed CFD. Chem. Ing. Tech. 87:1–13

114. Wehinger GD, Eppinger T, Kraume M. 2015. Detailed numerical simulations of catalytic fixed-bed
reactors: heterogeneous dry reforming of methane. Chem. Eng. Sci. 122:197–209
115. Behnam M, Dixon AG. 2017. 3D CFD simulations of local carbon formation in steam methane reform-
ing catalyst particles. Int. J. Chem. React. Eng. 15:6
116. Alzahrani F, Rusi H, Assabumrungrat S, Fernandes DLA, Aiouache F. 2019. Deactivation of the pref-
erential oxidation of CO in packed bed reactor by 3D modelling and near-infrared tomography. Chem.
Eng. J. 378:122082
117. Atmakidis T, Kenig EY. 2012. Numerical analysis of mass transfer in packed-bed reactors with irregular
particle arrangements. Chem. Eng. Sci. 81:77–83
118. Atmakidis T, Kenig EY. 2014. Numerical investigations of packed bed reactors with irregular particle
arrangements. In 24th European Symposium on Computer Aided Process Engineering: Part A, ed. J Klemes,
P Varbanov, PY Liew, pp. 217–22. Amsterdam: Elsevier
119. Bale S, Sathe M, Ayeni O, Berrouk AS, Joshi J, Nandakumar K. 2017. Spatially resolved mass transfer
coefficient for moderate Reynolds number flows in packed beds: wall effects. Int. J. Heat Mass Transf.
110:406–15
120. Bale S, Tiwari S, Sathe M, Berrouk AS, Nandakumar K, Joshi J. 2018. Direct numerical simulation study
of end effects and D/d ratio on mass transfer in packed beds. Int. J. Heat Mass Transf. 127:234–44
121. Kruggel-Emden H, Kravets B, Suryanarayana MK, Jasevicius R. 2016. Direct numerical simulation of
coupled fluid flow and heat transfer for single particles and particle packings by a LBM-approach. Powder
Technol. 294:236–51
122. Kravets B, Kruggel-Emden H. 2017. Investigation of local heat transfer in random particle packings by
a fully resolved LBM-approach. Powder Technol. 318:293–305
123. Buist KA, Backx BJGH, Deen NG, Kuipers JAM. 2017. A combined experimental and simulation study
of fluid-particle heat transfer in dense arrays of stationary particles. Chem. Eng. Sci. 169:310–20
124. Sun B, Tenneti S, Subramaniam S. 2015. Modeling average gas-solid heat transfer using particle-resolved
direct numerical simulation. Int. J. Heat Mass Transf. 86:898–913
125. Singhal A, Cloete S, Radl S, Quinta-Ferreira R, Amini S. 2017. Heat transfer to a gas from densely
packed beds of monodisperse spherical particles. Chem. Eng. J. 314:27–37
126. Singhal A, Cloete S, Radl S, Quinta-Ferreira R, Amini S. 2017. Heat transfer to a gas from densely
packed beds of cylindrical particles. Chem. Eng. Sci. 172:1–12
127. Jourak A, Hellström JGI, Lundström TS, Frishfelds V. 2014. Numerical derivation of dispersion co-
efficients for flow through three-dimensional randomly packed beds of monodisperse spheres. AIChE
J. 60:749–61
128. Dixon AG, Medeiros NJ. 2017. Computational fluid dynamics simulations of gas-phase radial dispersion

·.•�-
in fixed beds with wall effects. Fluids 2:56

www.annualreviews.org • CFD for Fixed Bed Reactor Design 3.21

, Review in Advance first posted

on March 9, 2020. (Changes may
still occur before final publication.)
 CH11CH03_Dixon ARjats.cls February 25, 2020 16:36

129. Dixon AG, Gurnon AK, Nijemeisland M, Stitt EH. 2013. CFD testing of the pointwise use of the
Zehner-Schlünder formulas for fixed-bed stagnant thermal conductivity. Int. Commun. Heat Mass. 42:1–4
130. Mathey F. 2013. Numerical up-scaling approach for the simulation of heat-transfer in randomly packed
beds. Int. J. Heat Mass Transf. 61:451–63
131. Thiagalingam I, Dallet M, Bennaceur I, Cadalen S, Sagaut P. 2015. Exact non-local expression for the
wall heat transfer coefficient in tubular catalytic reactors. Int. J. Heat Fluid Flow 54:97–106
132. Thiagalingam I, Sagaut P. 2016. Pseudo-homogeneous 1D RANS radial model for heat transfer in tubu-
lar packed beds. Int. J. Heat Fluid Flow 62:258–72
133. Dixon AG. 2017. CFD study of local transport and reaction rates in a fixed bed reactor tube: endothermic
steam methane reforming. Chem. Eng. Sci. 168:156–77
134. Behnam M, Dixon AG, Nijemeisland M, Stitt EH. 2013. A new approach to fixed bed radial heat transfer
modeling using velocity fields from CFD simulations. Ind. Eng. Chem. Res. 52:15244–61
135. Asensio DA, Zambon MT, Mazza GD, Barreto GF. 2014. Heterogeneous two-region model for low-
aspect-ratio fixed-bed catalytic reactors: analysis of fluid-convective contributions. Ind. Eng. Chem. Res.
Annu. Rev. Chem. Biomol. Eng. 2020.11. Downloaded from www.annualreviews.org

53:3587–605
136. Zambon MT, Asensio DA, Barreto GF, Mazza GD. 2014. Application of computational fluid dynamics
Access provided by 207.241.225.139 on 05/07/20. For personal use only.

(CFD) for the evaluation of fluid convective radial heat transfer parameters in packed beds. Ind. Eng.
Chem. Res. 53:19052–61
137. Suárez F, Luzi CD, Mariani NJ, Barreto GF. 2017. Effective parameters for conductive contributions
to radial heat transfer in fixed beds under stagnant conditions. Chem. Eng. Res. Des. 119:245–62
138. Rebughini S, Cuoci A, Maestri M. 2016. Hierarchical analysis of the gas-to-particle heat and mass trans-
fer in micro packed bed reactors. Chem. Eng. J. 289:471–78
139. Rebughini S, Bracconi M, Dixon AG, Maestri M. 2018. A hierarchical approach to chemical reactor
engineering: an application to micro packed-bed reactors. React. Chem. Eng. 3:25–33
140. Kolaczkowski ST, Awdry S, Smith T, Thomas D, Torkuhl L, Kolvenbach R. 2016. Potential for metal
foams to act as structured catalyst supports in fixed-bed reactors. Catal. Today 273:221–33
141. Das S, Deen NG, Kuipers JAM. 2018. Multiscale modeling of fixed-bed reactors with porous (open-cell
foam) nonspherical particles: hydrodynamics. Chem. Eng. J. 334:741–59
142. Wehinger GD, Kolaczkowski ST, Schmalhorst L, Beton D, Torkuhl L. 2019. Modeling fixed-bed reac-
tors from metal-foam pellets with detailed CFD. Chem. Eng. J. 373:709–19
143. Dong Y, Korup O, Gerdts J, Roldán Cuenya B, Horn R. 2018. Microtomography-based CFD modeling
of a fixed-bed reactor with an open cell foam monolith and experimental verification by reactor profile
measurements. Chem. Eng. J. 353:176–88
144. Visconti CG, Groppi G, Tronconi E. 2016. Highly conductive “packed foams”: a new concept for the
intensification of strongly endo- and exo-thermic catalytic processes in compact tubular reactors. Catal.
Today 273:178–86
145. Torres D, Lopez N, Illas F, Lambert RM. 2005. Why copper is intrinsically more selective than silver
in alkene epoxidation: ethylene oxidation on Cu(111) versus Ag(111). J. Am. Chem. Soc. 127:10774–75

·.•�-
3.22 Dixon • Partopour

, Review in Advance first posted

on March 9, 2020. (Changes may
still occur before final publication.)