Chemical Engineering Science 64 (2009) 2113 -- 2126

Contents lists available at ScienceDirect

Chemical Engineering Science

journal homepage: w w w . e l s e v i e r . c o m / l o c a t e / c e s

Analysis of flow in rotating packed beds via CFD simulations—Dry pressure drop and
gas flow maldistribution
Hugo Llerena-Chavez, Fa¨çal Larachi ∗
Chemical Engineering Department, Laval University, Québec, Canada G1V 0A6

A R T I C L E I N F O A B S T R A C T

Article history: Three-dimensional unsteady-state turbulent rotating single-phase flows were simulated in rotating packed
Received 14 November 2008 beds (RPB) and were validated using overall dry pressure drop measurements for three RPB designs
Received in revised form 7 January 2009 [Liu, H.-S., Lin, C.-C., Wu, S.-C., Hsu, H.-W., 1996. Characteristics of a rotating packed bed. Industrial and
Accepted 14 January 2009
Engineering Chemistry Research 35, 3590–3596; Sandilya, P., Rao, D.P., Sharma, A., Biswas, G., 2001b.
Available online 23 January 2009
Gas-phase mass transfer in a centrifugal contactor. Industrial and Engineering Chemistry Research 40,
Keywords:
384–392; Zheng, C., Guo, K., Feng, Y.D., Yung, C., 2000. Pressure drop of centripetal gas flow through rotat-
Rotating packed bed ing bed. Industrial and Engineering Chemistry Research 39, 829–834]. Analysis of the radial and tangential
Aerodynamics velocities highlighted the impact of gas feed entrance effects on the peripheral gas maldistribution in the
Tangential flow rotating packing module. Recommendations were formulated for an optimum design with the aim to re-
Pressure gradient duce gas flow maldistribution in RPBs. Breakdown of the overall pressure drop in its modular components
Maldistribution for the housing, the rotating packing module, the free inner rotational zone, and the gas disengagement
showed that the dissipation in the rotating packing could be a minor contributor to the overall pressure
drop which may be undesirable in terms of RPB mass transfer and reaction efficiencies. Analysis of the
simulated pressure drops allowed development of CFD-based Ergun-type semi-empirical relationships in
which the gas-slip and radial acceleration effects, the laminar and inertial drag effects, and the centrifugal
effect were aggregated additively to recompose the pressure drops in the rotating packing module.
© 2009 Elsevier Ltd. All rights reserved.

1. Introduction Compactness of RPBs makes them attractive for intensified reac-

Radial flows between two rotating parallel disks represent an
important class of basic flows owing to their tremendous indus-
trial applications, e.g., rotating machinery, lubrication, viscosime-
try, heat and mass exchangers, biomechanics, etc. (Batchelor, 1951;
Stewartson, 1953; Mellor et al., 1968; Bodonyi and Stewartson, 1977;
Szeri et al., 1983a,b; Jarre et al., 1996; Sandilya et al., 2001a). In
chemical engineering, these devices inspired the rotating packed bed
(RPB) in which a radial flow is forced through a rotating porous an-
nular layer placed inside an enclosure, see Fig. 1. RPB also referred
to as HiGee for high gravity matured on the basis of pioneering
works traced back to the 1960s (Vivian et al., 1965; Jackson and
Marchell, 1968; Podbielniak, 1966, 1967; Todd, 1969). When first in-
troduced at Imperial Chemical Industries, RPB was exploited for its
induced macro-gravitational field to enhance interfacial mass trans-
fers and to enlarge the loading zone in gas–liquid counter-current
flows (Ramshaw and Mallinson, 1981; Ramshaw, 1983).
∗ Corresponding author. Tel.: +1 418 656 3566; fax: +1 418 656 5993.
E-mail address: faical.larachi@gch.ulaval.ca (F. Larachi).
tion/separation purposes and their extension to various areas of
separation and materials' synthesis is growing steadily. Numerous
environmental applications of high gravity have been demon-
strated, such as VOC removal from groundwater (Singh et al., 1992;
Chen and Liu, 2002), CO2 scrubbing (Lin et al., 2008), phenol com-
plexation/extraction from wastewater towards immiscible organic
solvents (Yang et al., 2004), biosorption (Das et al., 2008), supercrit-
ical CO2 desorption of toluene from activated carbon (Tan and Lee,
2008), ozone scrubbing (Lin and Su, 2008), bio-oxidation, polymer
devolatilization, and hydrogen chloride stripping (Zheng et al., 2000;
Cummings et al., 1999; Chen et al., 2004a; Yang et al., 2005), gas
exhaust absorption (Lin et al., 2003; He et al., 2003), coal combus-
tion flue gas desulfurization (Pan and Deng, 2002), fly ash filtration
in power generation systems (Song et al., 2003), and distillation (Li
et al., 2008). Fossil fuel applications of RPB concern seawater deaer-
ation for re-injection into declining offshore oil fields for enhanced
oil recovery (Peel et al., 1998), or simultaneous H2 S and H2 O re-
moval from natural gas (Eimer, 2003). SINOPEC at Shengli oilfield
used 1.5 m diameter RPB in replacement of gigantic 30-m tall vac-
uum towers for water deaeration (Zheng et al., 1997). Quite recently,
Dow Chemical examined the production of hypochlorous acid using

0009-2509/$ - see front matter © 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ces.2009.01.019
2114 H. Llerena-Chavez, F. Larachi / Chemical Engineering Science 64 (2009) 2113 -- 2126
Fig. 1. Schematic diagram of a rotating packed bed setup and corresponding segmentation (Sections 1–3).
an RPB pilot plant as a reactive stripper by contacting gaseous chlo-
rine with sodium hydroxide solutions (Trent and Tirtowidjojo, 2001,
2002). Other processes under exploration include high-gravity reac-
tive precipitation for nanoparticles production, e.g., calcium carbon-
ate, silica, titania, alumina, zinc sulfide, etc. (Chen and Shao, 2003;
Chen et al., 2003, 2004b; Zhu et al., 2003, 2004; Zhang et al., 2004; Li
et al., 2003), and for pharmaceuticals such as asthma treating salbu-
tamol, analgesic ibuprofen and other drugs (Ma et al., 2004; Zhou
et al., 2003; Chen et al., 2004c).
In spite of RPB superior mass transfer efficiency (Ramshaw, 1993),
understanding the partnership between its internals and the distri-
bution of the induced fluid flow is a challenging task (Guo et al.,
2000; Sandilya et al., 2001b; Zheng et al., 2000). In spite of a large
body of knowledge relating to RPBs, two interrelated issues received
partial coverage in the literature. (i) Scarcity of local measurements
to unfold the details of fluid hydrodynamics and distribution in-
side the RPB system and (ii) lack of space resolved three-dimensional
(3D) computational fluid dynamic (CFD) simulations to unveil de-
tails about the pressure field and velocity distribution, the gross flow
patterns, the maldistribution, etc.
The multizone character of the geometry of RPBs (Fig. 1) in addi-
tion to the rotation of the porous bed give rise, unlike radial or axial
flows across stationary porous beds, to numerous flow features, e.g.,
translational, rotational, sudden directional change, bend, swirling,
and abrupt contraction/expansion. In an effort to clarify the interac-
tions between fluids and RPB internals in terms of momentum and
mass, and energy transfers, knowledge of these features is of great
interest from the practical and fundamental viewpoints.
Procurement of local measurements regarding flow pattern,
phase holdups, pressure loss, fluid residence time, etc., inside RPBs
is somehow difficult. Keyvany and Gardner (1989) studied the fluids
residence time distribution but did not reveal much about the inter-
nal fluid dynamics. Burns and Ramshaw (1996) in their visual study
of liquid distribution across a rotating bed filled with a molded
foam packing concluded that the actual liquid flow does not reach
uniformity as assumed in several film models. Severe liquid mald-
istribution is observed at low rotational speed, whereas at higher
rotational speed, the flow pattern is shifted from maldistributed
rivulet flow to fine droplet flow. Guo et al. (2000) combining res-
idence time distribution and visual studies investigated the liquid
flow structure in a rotating bed. They concluded that the inner
bed is the region of most intense liquid deformation and mixing,
whereas, elsewhere, the liquid flows as a film covering the packing
surface. Flooding of RPB occurs at very high gas velocities (Lockett,
1995) producing thin draining films and droplets suggesting that
classification into different flow regimes is crucial. Burns et al.
(2000) proposed to distinguish between rivulet (pore) flow, droplet
flow, film flow, spray flow, mist flow, and flooding. However, these
flow regimes still need to be described using consistent models and
more elaborate experimentations.
Liquid hold-up is another factor closely related to liquid dis-
tribution. Bašic and Dudukovic (1995) published the first known
holdup measurements using an electrical conductivity technique.
They were able to assess the degree of anisotropy of liquid dis-
tribution as a function of operating variables and questioned the
physical justification of using film flow models (i.e., penetration the-
ory, convection–diffusion model) for estimating mass transfer coef-
ficients (Tung and Mah, 1985; Munjal et al., 1989a,b; Xinlin et al.,
2000). Pursuing similar objectives, Burns et al. (2000) investigated
the behavior of liquid hold-up in high-voidage RPBs. While main-
taining that three distinct flow regimes could exist (pore flow, film
flow and droplet flow), they concluded that liquid hold-up (i) is in-
versely proportional to the local packing radius, (ii) is independent of
gas flow, (iii) decreases with rotational speed, and (iv) and is weakly
affected by liquid viscosity.
Pressure drop analyses in RPBs were reported by Keyvani and
Gardner (1989) for air–water flows across high-voidage aluminum
foam metal beds. They found that: (i) both dry (gas-phase) and ir-
rigated (gas–liquid) pressure drops are proportional to the squared
rotational speed () and (ii) pressure drop increases with increased
gas flow rate (GFR). Liu et al. (1996) also examined the effect of op-
erating variables on pressure drop using lower voidage, rectangu-
lar and elliptical, random packings, and found that (i) dry pressure
drop and rotor speed are related somewhat linearly, (ii) at high ro-
tor speed, the pressure drop is strongly influenced by GFR and only
slightly by liquid flow rate, (iii) at low rotor speed, liquid flow rate
(via liquid hold-up) becomes influential on pressure drop, and (iv)
high rotor speed and low liquid flow rate can produce lower pres-
sure drop compared to the dry bed pressure drop, in accordance with
Keyvani and Gardner (1989), possibly because small amounts of well
dispersed liquid acts as a lubricant. Zheng et al. (2000) proposed
interesting arguments to explain part of the pressure drop behav-
ior using two-dimensional (2D) mass and momentum conservation
equations. They observed that pressure drop is mainly dependent
 H. Llerena-Chavez, F. Larachi / Chemical Engineering Science 64 (2009) 2113 -- 2126
Fig. 2. Gas feed details of the RPB according to (a) Liu et al. (1996), (b) Sandilya et al. (2001b), and (c) Zheng et al. (2000).
on rotational speed, liquid and GFR as well as on packing configu-
ration. It comes out that pressure drop behavior is partly related to
the gas phase conserving its angular momentum while flowing in-
wardly through the RPB from the housing section to the centrally
located gas exhaust sink.
A common conclusion that emerges from this literature survey is
that RPBs are still mostly looked at via their global parameters. For
instance, such global parameters are not helpful to identify or to as-
sess in a piecewise manner the relative contributions arising from
each zone in the RPB despite the multizone feature of RPBs (Zheng
et al., 2000). In addition, one encounters in the literature large dis-
crepancies and lack of standards in RPB designs, be it in terms of
arrangement and sizing of the equipment, or in terms of operating
parameters (Rao et al., 2004). This diversity in arrangements of RPBs
is also responsible for different types of flows that prevent reliable
comparisons among RPBs to be carried out based on empirical cor-
relations or based on global measurements.
We propose to analyze the behavior of single-phase (air) flows
through the entirety of the RPB setups with the aid of CFD. Specif-
ically, gas flow within RPB evolving in the absence of a second im-
miscible fluid is helpful in fingerprinting the role of each individual
physical section of the RPB on its aerodynamics. As such, single-phase
flow simulations are viewed as a compulsory step prior to any two-
phase flow hydrodynamics simulations. Turbulent 3D CFD simula-
tions were used and validations were made with experimental data
from three different RPB equipments for which overall dry pressure
drop measurements are available. This work discusses the influence
of the setup geometry on gas circulation and on the pressure drop
in each section of the RPB. Recommendations are formulated for an
optimum design with the aim to reduce gas flow maldistribution in
RPBs. Finally, the pressure drop in the packed bed rotating section
was analyzed with a goal to propose CFD-based Ergun-type relation-
ships useful for the design of RPBs.
2. RPB geometry and CFD implementation
An RPB setup can be divided into three generic sections (Fig. 1):
(i) a stationary housing (Section 1) from where gas flow is delivered,
(ii) a rotor section hosting an annular porous medium and eventually
including a free rotation zone (Section 2), and (iii) a (liquid) distrib-
utor section comprising also a gas discharging outlet (Section 3). We
will restrict ourselves in this study with an analysis of single-phase
gas inward flows in which the gas is fed radially, axially or tangen-
tially, from the stationary housing (Section 1) in its way through to
the distributor Section 3 (Fig. 1).
Three different feed designs considered in the literature have
been simulated. Gas feed usually consists of single or multiple
tube injectors installed on top or lateral side of housing (Section 1)
whereby the incoming gas engulfs into the casing cavity radially
(Fig. 2a), adjacently parallel to the rotor axis of revolution (Fig. 2b)
or tangentially (Fig. 2c). The gas arrives laterally on the outer face
2115
of the rotor at a tangential velocity that is either induced by the
rotor rotation itself or imparted by a fast incoming tangential flow
of gas as in Fig. 2c, or a combination thereof. The rotor section is
swept in a spiral-like manner by the gas flow before its discharge
in Section 3. When entering the distribution Section 3, the gas may
either expand or contract depending on magnitude of flow rate
and diameter of outlet pipe. Before to be vented, the gas leaves the
distribution section via a discharge chimney (Fig. 1).
The dimensions and design specifics of the RPB equipments stud-
ied by Liu et al. (1996) (Fig. 2a), Sandilya et al. (2001b) (Fig. 2b), and
Zheng et al. (2000) (Fig. 2c) are used as a basis for the present simu-
lations. A summary of the RPB characteristics is provided in Table 1.
When dimensions were not explicitly given, they were estimated
based on equipment layout, Table 1. The differences between the
three designs are quoted as follows:
• Location and incidence of gas inlet injector in housing compart-
ment: lateral/orthogonal to wall (Fig. 2a); atop/parallel to rotor
axis of revolution (Fig. 2b); lateral/tangential to wall (Fig. 2c).
• Partial or total filling of rotor section with porous medium: (i)
rotor section comprising packed bed zone and free rotation zones
(Figs. 2a and c); (ii) rotor section completely filled with packing
(Fig. 2b).
• Variation in diameter of gas outlet pipe and range in gas through-
puts: narrow discharge area (pipe opening 16 mm) and low flow
rate range (Fig. 2a), large discharge area and moderate to high flow
rate range (Figs. 2b and c; pipe opening: 50 mm).
The 1D and 2D simulations of free, i.e., no porous medium, ra-
dial flows between two rotating parallel disks have shown limits in
capturing the actual aerodynamics prevailing in these devices, there-
fore advocating the necessity of 3D simulations (Serre et al., 2001;
Raimundo et al., 2002). Furthermore, drawbacks of the standard k–
model for representing turbulent flows in rotating flows have been
reported in the literature (Shi and Ribando, 1992); besides, the Cori-
olis forces did not receive extensive coverage in turbulence model-
ing (Shi and Ribando, 1992; Yeung and Zhou, 1998). It is nonetheless
accepted that the computational overhead entailed by more com-
plex turbulence models is not always worthy in terms of precision
enhancement (Nallasamy, 1987) and that k– models still represent
an acceptable compromise (Karunakumari et al., 2005). Choosing the
most pertinent turbulence models for closure of the Reynolds stress
tensor for each RPB section is not obvious because of the various flow
types encountered in the RPB: expansion, contraction, free rotation,
sudden directional change (90◦ bend), etc. Moreover, assignment of
appropriate turbulence models to the pertinent RPB sub-domains
while simultaneously solving the whole flow field is not tractable us-
ing current commercial codes. In our simulations a single turbulence
model, namely, the renormalization group (RNG) k– model was re-
tained owing to its enhanced accuracy for swirling flows (Fluent,
2006). Fluent v.6 commercial package was used for solving the 3D
2116 H. Llerena-Chavez, F. Larachi / Chemical Engineering Science 64 (2009) 2113 -- 2126

Table 1
Summary of RPB dimensions.

Liu et al. (1996) Sandilya et al. (2001b) Zheng et al. (2000)

Given Assumed Given Assumed Given Assumed

Housing
Shape Circular Polygon Circular Circular
Outer diameter 0.18 0.60 0.60
Axial thickness 0.04 0.04 0.078
Gas flow
Inlet inside housing Direct Top Tangential
Pipe diameter inlet 0.016 0.058 0.035
Pipe diameter outlet 0.016 0.05 0.05

Rotor
Internal diameter, dDa 0.016 0.06 0.045
External diameter, dDb 0.14 0.31 0.30
Axial thickness, h 0.02 0.022 0.048

Packing section
Shape Rectangular Wire mesh Foam
Elliptic
Internal diameter, dpa 0.09 0.05 0.15
External diameter, dpb 0.14 0.31 0.30
Specific area, as 524 1027 2196 1000
Porosity 0.53 0.39 0.91 0.94

Table 2
Experimental ranges.

Liu et al. (1996) Sandilya et al. (2001a,b) Zheng et al. (2000)

3
Gas flow rate (GFR), m /h 0–5 0–27 0–160
Rotating speed (RPM), rpm 0–2100 0–1420 0–1090

Eulerian forms of the continuity and momentum balance equations
for procuring the velocity and pressure fields across the different RPB
sections. The z-axis was chosen as the rotor rotational axis for all the
simulated RPB geometries. Furthermore, to capture rapid dynamic
features in the flow and to account for turbulence, the simulations
were carried out in non-steady state. Two types of momentum bal-
ance equations were solved, one in stationary non-rotating inertial
reference frames (e.g., housing section) and one in rotating reference
frame (rotor section,  being the reference frame local angular ve-
locity). A constant porosity value corresponding to the average bed
porosity value was assigned to the whole porous medium domain,
Table 1.
The closure equation for the turbulent Reynolds shear stress ten-
sor was handled via the Boussinesq approximation. An interpola-
tion scheme with PRESTO method was used for the pressure values,
and the pressure–velocity coupling was obtained using the SIMPLEC
algorithm. The finite-volume discretized equations were solved us-
ing a second-order upwind discretization scheme and ad-hoc over-
relaxation parameters. The time steps typically used were between
10−4 and 10−3 s and the number of discretization cells was approx-
imately 350,000. Slipless boundary conditions and near wall treat-
ment using enhanced wall treatment and pressure gradient effects
were applied.
3. Discussion of simulation results
experimental studies. The total pressure drop measured between
the gas inlet and exit (gas discharge chimney in distributor section)
is used as the observable variable for validation.
Figs. 3a–c compare simulated versus measured total pressure
drops as a function of RPM at constant GFR (Liu et al., 1996, Fig. 3a)
and as a function of GFR at numerous RPM values (Sandilya et al.,
2001b, Fig. 3b; Zheng et al., 2000, Fig. 3c).
The data of Liu et al. (1996) concerns two packing shapes: rect-
angular and elliptic (Table 1). For the latter, the measured pressure
drop appears to be constant, whereas the simulation predicts an in-
creasing trend as a function of RPM similarly to the rectangular pack-
ing (Figs. 3a). This suggests that the experimental data reflects the
existence of other factors inherent to the elliptic packing that have
not been described in the study of Liu et al. (1996) and therefore
were overlooked in the simulation.
Generally, good agreement between measured pressure drops
and CFD predicted ones is achieved for the much higher GFR ranges
covered by Sandilya et al. (2001b) and Zheng et al. (2000) where it
can be seen that the dependences with respect to RPM and GFR are
well captured (Figs. 3b and c). Another worth mentioning remark
regarding the simulated trend reflects on the occasional crossings
between simulated curves and experimental trends, vide supra,
Fig. 3a elliptic packing, Fig. 3b at RPM = 1420 or Fig. 3c at
RPM = 1090, suggesting that one turbulence model, irrespective of
Sections 1–3, for describing the aerodynamics over the whole RPB
domain may not be the best choice.

Lack of local experimental data, such as velocity and pres-

sure spatial distributions, reduces the possibilities to resort to the
overall pressure drop measurements for partial validation of the
CFD simulations for the different operation conditions, i.e., GFR
and rotor rotational speed (RPM), and different RPB arrangements.
Table 2 summarizes the ranges of GFR and RPM covered in the
3.1. Flow pattern in housing cavity
In addition to gas injection design, RPM and GFR, the gas dis-
tribution across the packed bed depends on the one developing in
the housing section. Enabling rotor motion prompts the gas to spin
 H. Llerena-Chavez, F. Larachi / Chemical Engineering Science 64 (2009) 2113 -- 2126 2117

distribute by splitting almost evenly clockwise and anticlockwise to

800 fill the housing space at RPM = 0 (Fig. 4b(1)). When clockwise rotor
motion is enabled, the gas is deviated in the housing (Fig. 4b(2)). The
700 collision of flow path lines against housing bottom, and then their
rebound, improves mixing and yields better gas distribution within
Pressure drop (Pa)

600 the housing section.

Owing to a tangential feed, the housing section is filled up spirally
500 G = 1.7 m3/hr by the gas flow for the last case (Fig. 4c). The only difference between
static and rotating rotors is the differing tangential velocity in the
Rectangular, meas.
400 Rectangular, sim. housing section.
Elliptic, meas.
300 Elliptic, sim.
3.2. Radial and tangential velocity components
200
800 1000 1200 1400 1600 1800 The design of the gas entrance determines the flow circulation
RPM in the RPB casing; hence, different velocity distributions arise de-
pending on RPB geometry. These differences are assessed from 2D
contour plots of radial and tangential velocities. Unless otherwise
600 specified, the sampled planes are taken at elevation z correspond-
ing to packed bed half thickness (Figs. 5 and 6). Three red circles
500 appear in the rotor section to distinguish the packed bed zone from
Pressure drop (Pa)

RPM = 1420 the innermost free rotation zone adjacent to the discharge tube, e.g.,
400
Fig. 5. Also, only two red circles appear in the Sandilya et al. (2001b)
design because the packed bed fills the entire rotor space (Fig. 6a
300
and b). In Liu et al. (1996) design, three black disks in the housing
RPM = 950
200 represent as many fixing stems (Fig. 5).
For static rotor (Fig. 2a), an inward jet (ur < 0) protrudes through
100 a strip of the packed bed extending from the feed pipe (Fig. 5a).
RPM = 0 The flow maldistribution across the packed bed is remarkable de-
0 spite an inwardly accelerating radial gross flow, Fig. 5a. Correspond-
5 7 9 11 13 15 17 19 21 23 25 27 ingly, the gas azimuthal velocity in the rotor section is progressively
Gas flow rate (m3/hr) attenuated in spite of tangential motions in the housing section,
Fig. 5b.
1400 After rotor (anticlockwise) rotation is primed, the gas flow ac-
quires well defined anticlockwise tangential velocities in the housing
RPM = 1090
1200 section that are much larger than at RPM = 0, Fig. 5b and d. The flow
maldistribution in the bed and housing persisted even in the pres-
Pressure drop (Pa)

1000 ence of induced centrifugal forces. Nonetheless, the centrifugal force

RPM = 816 being greater in the packed bed zone due to rotor rotation, wider bed
800 portions are contacted by the inward gas flow, Fig. 5a and c. In the
packed bed section, the gas radial (respectively, azimuthal) velocity
600 RPM = 538
increases (respectively, decreases) inwards. The local effects shown
in the ur contours provide post-facto evidence of the necessity of 3D
400
RPM = 0 simulations in RPBs.
200 In the inner rotor free rotation zone, the gas radial flow decel-
erates, on the one hand, as a result of the sudden voidage increase
0 after the gas has left the packed bed, but accelerates as radius is
20 40 60 80 100 120 140 160 decreasing, on the other hand, because of compliance with the con-
Gas flow rate (m3/hr) tinuity equation. However, the gas tangential velocity in the free
rotation zone increases due to the inertial force upon leaving the
Fig. 3. Simulated versus measured total pressure drops. Data from (a) Liu et al. porous medium and also via the rotor radius reduction. Finally the
(1996) with rectangular and elliptic packings, (b) Sandilya et al. (2001b), and (c)
gas attains the distributor section with high tangential and radial
Zheng et al. (2000) as a function of rotational speed of rotor and volumetric GFR.
Dashed and solid lines are CFD computed. velocities. Due to the small pipe diameter, gas contraction occurs
causing high tangential velocity components.
For the geometry in Fig. 2b, the gas flow is less prone to mald-
istribution except near feed entrance. A rotating rotor induces in
around in the housing cavity. In the first case, the gas flow inlet the housing section a gas tangential velocity (Fig. 6b) which helps
discharges sidewise in the housing section perpendicularly to the improving azimuthal distribution of the gas radial flow in the ro-
wall as shown by the gas path lines of Figs. 4a(1,2). For static rotors, tor section, Fig. 6a. The gas radial velocity increases inwards in the
the gas flow smashes against the rotor portion along its radial ex- packed bed, whereas the gas tangential velocity decreases. Because
tension (Fig. 4a(1)). For rotating rotors, the gas trajectories undergo of no free rotation zone in this design, the gas attains the distributor
deviations that are commensurate with RPM and GFR amplitudes with large inward radial velocities and low to moderate tangential
(Fig. 4a(2)). An anticlockwise rotor rotation induced a gas flow cir- velocities.
cumferential motion in the RPB casing. In Fig. 2c geometry, gas is injected from a tangential leg bound
In the second example (Fig. 4b), the gas engulfs into the housing up with the housing wall bringing about an initial tangential gas
section atop at an off-rotor injection hole. The gas flow tends to velocity. This effect adds up with a bulk tangential flow induced in
2118 H. Llerena-Chavez, F. Larachi / Chemical Engineering Science 64 (2009) 2113 -- 2126
Fig. 4. Flow path lines of gas in housing cavity (a) Liu et al. (1996) design : GFR = 1.7 m3 /h; (1) stationary RPM = 0 and (2) moving (anticlockwise) rotor RPM = 870;
(b) Sandilya et al. (2001b) design: GFR = 13.1 m3 /h; (1) stationary RPM = 0 and (2) moving (clockwise) rotor RPM = 950; (c) Zheng et al. (2000) design with moving
(anticlockwise) rotor: GFR = 60 m3 /h; RPM = 816.
the housing by rotor rotation. The result is more or less improved
azimuthal uniformity of gas radial flow in the rotor section in spite of
strongly azimuthally biased radial velocities in the housing section,
Fig. 5c. The gas flow in the rotor section developed in much the same
way as in the Liu et al. (1996) design (Fig. 2a) attaining the distributor
section with high tangential and radial velocities, Fig. 6c and d.
3.3. Breakdown of pressure drop contributions
The relative importance of the drag, the centrifugal, the Coriolis
and the inertial forces in the studied configurations is reflected in the
zonal pressure drops, PI , PII , and PIII , corresponding to Sections
1–3 (Fig. 1) of the RPB, respectively. Due to the obstructing/rotating
internals, intrusion of pressure sensors in RPBs for measuring the
zonal pressure drops is uneasy; therefore, the overall pressure drop,
Po , determined for the whole device, is often reported. As proposed
by Zheng et al. (2000), it is reasonable to break down the overall
pressure drop, Po , into its zonal components as
Po = PI + [PPacked bed + PFree rotation ]II + PIII
The simulated pressure drops are shown in Figs. 7a–e as a function
of RPM and GFR for each section of the RPB configurations (bars
represent twice the value of the standard deviation, i.e., ± 15%). The
pressure drops are computed using the face-averaged pressures for
the cells located on the boundary surfaces demarcating Sections 1–3.
As Liu et al. (1996) used very low GFR (Fig. 7a, Table 2), the
contributions to the overall pressure drop are ascribed, in the order
of decreasing importance, to the free rotation zone, the distributor
section, the packed bed zone and finally the housing section. The
simulation reveals that the losses by the packed bed are marginal
for this configuration.
As illustrated in Fig. 7a, the slope of the pressure drop curve for
the free rotation zone turns steeper the higher the RPM, whereas
only a tenuous effect of rotor speed is observed on the rest of the
pressure drop curves. Coherent with Fig. 5d trend regarding the in-
creasing tangential velocity in the free rotation zone, the centrifugal
and Coriolis accelerations are the main drivers for such predomi-
nance in the free rotation zone. In addition, the gas discharge in the
distributor is the one that exhibits the least sensitivity to RPM. One
explanation could be ascribed to the stack (or chimney) effect which
(1) facilitates exit of the gas flow with increasing RPM.
 H. Llerena-Chavez, F. Larachi / Chemical Engineering Science 64 (2009) 2113 -- 2126
Fig. 5. Velocity contour plots of gas flow at mid-plane in RPB for Liu et al. (1996) design: GFR = 1.7 m3 /h (a) radial velocity static rotor, (b) azimuthal velocity static rotor,
(c) radial velocity moving rotor, RPM = 870, and (d) azimuthal velocity moving rotor, RPM = 870.
For the Sandilya et al. (2001b) design, the main contribution
to the overall pressure drop in both static (Fig. 7b) and rotating
(Fig. 7c) cases results from the rotor which consists solely of the
porous medium (no free rotation zone). Lesser dissipations stem
from the distribution and housing sections as compared to those of
Liu et al. (1996). As seen from Table 1, the wider packed bed in this
design is the main cause, i.e., 13 cm (Sandilya et al., 2001b) versus
3 cm (Liu et al., 1996).
Counter-intuitively, the pressure drop in the housing decreases as
GFR increases when rotor rotation is enabled (Fig. 7c). Greater GFR,
by virtue of the gas injection design, impinge on the opposite side
of the housing near the entrance (Fig. 4b) and attenuate the rotor-
induced rotational circulation of the gas in the housing thus lowering
dissipation. As shown in Fig. 7b and c, pressure drop in the rotor
section is monotonically increasing with GFR. However, for a given
GFR, the pressure drop increment contributed by rotor rotation is
huge suggesting that the centrifugal force largely outweighs the drag
force in the packed bed. A similar trend persists in the distributor
section as signature of the centrifugal force which under identical
GFR, yields PIII for a rotating rotor larger than at RPM = 0.
In Zheng et al. (2000) configuration (Fig. 7d and e), the major
contribution to the overall pressure drop arises from the distribu-
tor in both static and rotating cases. This is followed in the order of
decreasing importance by that of the free rotation, the packed bed
zone, and the housing section for rotating rotors (Fig. 7e). Compared
to the Liu et al. (1996) design, a reversal occurs between the pres-
sure drops stemming from free rotation and distribution zones. The
slope of the pressure drop curves in the distribution section is very
2119
sensitive to GFR. Very high GFR become determining because of the
losses associated with the sudden change in flow and contraction at
the exit of the distribution section.
For RPM = 0 and high GFR (Fig. 7d), sufficiently large tangential
velocities, induced by the tangential gas feed, cause the pressure
drop in the housing section to prevail on those in packed bed and
free rotation zones. However, the housing pressure drop becomes
marginal, regardless of GFR, if rotor rotation is enabled (Fig. 7e)
suggesting that the centrifugal force in packed bed outweighs the
drag.
3.4. Gas maldistribution in packed bed zone
To the best of the authors' knowledge, studies on gas maldis-
tribution in RPBs are virtually non-existent. In a quest of tractable
computations, flow axisymmetry and azimuthal invariance of some
hydrodynamic variables were assumed in most of the current RPB
hydrodynamic models (Liu et al., 1996; Zheng et al., 2000; Lin et al.,
2000; Sandilya et al., 2001b). These simplifications, least of all single-
phase gas flows, do not a priori seem justified in the light of above
simulation results. The rotating packing is the most crucial zone for
RPBs as the desired mass transfer and reaction processes are hosted
therein. A simple criterion to quantify gas flow maldistribution in
the packed bed section would be the deviation between the actual
azimuthal distribution of radial velocity components, as determined
by the 3D nature of the geometry, and an ideal azimuthally uniform
distribution of radial velocities. A bed gas maldistribution factor, rMfB
(  0), defined as a normalized variance, is calculated as a function
2120 H. Llerena-Chavez, F. Larachi / Chemical Engineering Science 64 (2009) 2113 -- 2126
Fig. 6. Velocity contour plots of gas flow at mid-plane in RPB: (a, c) radial velocity moving rotor, (b, d) azimuthal velocity moving rotor. Sandilya et al. (2001b) design (a,
b) RPM = 950 & GFR = 13.1 m3 /h; Zheng et al. (2000) design (c, d) RPM = 816 and GFR = 60 m3 /h.
of radial position, r, and averaged over selected angular intervals, ,
and over the whole bed height, h, of the rotating packing:
z=h/2
 n( ) 
 2
1 ur (r, i , z)
rMfB (r, h, ) = −1
n(, h) ũr
z=−h/2 i=1
where n(,h) is the number of collecting values, i.e., discretization,
over a selected angular interval, , and the whole height, ur (r,i ,z)
is the local radial velocity, and ũr is the azimuthally/axially aver-
aged radial velocity at rotor radius, r. If gas flow axisymmetry is ful-
filled at radius r, rMfB = 0. However, it increases as maldistribution in
the packed bed develops. Global maldistribution can be assessed for
 = 2 in Eq. (2), whereas quadrant-specific maldistribution can be
defined with  = /4, e.g., Fig. 9. A tangential gas maldistribution
function, Mf B , could be defined in a similar manner.
Radial profiles of global maldistribution factor, rMfB (r,h, = 2),
are shown in Fig. 8 as a function of RPM (Liu et al., 1996 design).
The simulations indicate that radial maldistribution of the gas is the
highest at the junction between the housing and the RPB sections.
Despite it is decreasing as gas progresses deeper inside the bed,
maldistribution remains unacceptably severe as has been foreseen
from Fig. 5c. Progressing towards bed interior implies a packing self-
improving distribution as a result of decreasing peripheral flow areas.
Also, increasing RPM improves distribution in the housing which
in turn translates into reduced maldistribution in the packed bed
(Fig. 8).
Radial profiles of global maldistribution factor, rMfB (r,h, = 2)
in the case of Sandilya et al. (2001b) design are illustrated in Fig. 9a
and b as a function of GFR and constant RPM. Except nearby the
packing-housing junction, the values for rMfB (r,h, = 2) are at
(2)
least an order of magnitude lesser than the above discussed case
(Fig. 8) pointing to the critical role of the design of gas injection.
Less severe azimuthal maldistribution occurs for the RPB design of
Sandilya et al. (2001b) which is further ameliorated by increased
GFR. Radial distribution of gas flow, unlike in Fig. 8, has a ten-
dency to slightly deteriorate as RPM increases regardless of GFR
(Fig. 9a and b). Such rotor-induced maldistribution develops with
increasing RPM because of the non-axisymmetrical gas injection
design and also because the gas experiences an increasing cen-
trifugal field in the housing section opposing to its entry into the
rotor. This is best illustrated by the quadrant-specific maldistri-
bution functions displayed for GFR = 6.5 m3 /h in Fig. 9c and d for
RPM = 950 and 1420, respectively. The rotor rotation is clockwise
so that quadrant IV is the most outlying with respect to feed. Devi-
ations between quadrants, especially I and II, at RPM = 950 are not
considerable, despite the best homogeneity attained in quadrant IV
(Fig. 9c). Higher RPM (Fig. 9d) deviates gas path lines in the housing
section thus worsening the specific maldistributions, especially in
quadrant III.
Zheng et al.'s (2000) design exhibits approximately the same de-
pendencies of rMfB (r,h, = /4) with respect to RPM and GFR as
the Sandilya et al. (2001b) one. Fig. 10 shows the quadrant-specific
 H. Llerena-Chavez, F. Larachi / Chemical Engineering Science 64 (2009) 2113 -- 2126 2121

800 Rectangular Packing

100 Overall pressure drop
700 3 Overall pressure drop
G = 1.7 m /hr 90
600 80 RPM = 0
Pressure drop (Pa)

Pressure drop (Pa)

500 70
60 Rotor section
400 Free rotating zone
50
300 40

200 Distributor section 30

20 Distributor section
100 Housing section Packed bed zone 10
Housing section
0 0
800 1000 1200 1400 1600 1800 5 7 9 11 13 15 17 19 21 23 25 27
RPM Gas flow rate (m3/hr)

600

Overall pressure drop 450 Overall pressure drop

500
RPM = 1420 400
Pressure drop (Pa)

350
400
Pressure drop (Pa)

RPM = 0
300
Rotor section Distributor section
300 250
200
200
150
100 Housing section
100 Distributor section
Packed bed zone
50
Housing section Free rotation zone
0 0
5 7 9 11 13 15 17 19 21 23 25 27 20 40 60 80 100 120 140 160
Gas flow rate (m3/hr) Gas flow rate (m3/hr)

1600

1400 Overall pressure drop

1200 RPM = 1090

Pressure drop (Pa)

1000

800 Distributor section

600

400 Packed bed zone

Free rotation zone
200

0 Housing section
20 40 60 80 100 120 140 160
Gas flow rate (m3/hr)

Fig. 7. Breakdown of the overall pressure drop into its constitutive contributions and their variations as a function of RPM and GFR: (a) rectangular packing, Liu et al. (1996)
design; (b) and (c) Sandilya et al. (2001b) design; (d) and (e) Zheng et al. (2000) design.

maldistribution factor for RPM = 1090 and high a GFR ( = 150 m3 /h). 3.5. Alternative RPB designs
The memory effect of gas flow inlet in the housing is best exempli-
fied by the greater maldistributions in quadrants I and II, whereas the Key requirements to be fulfilled by RPBs are increased capac-
values of rMfB (r,h, = 2) remain comparable to those in Sandilya ity while maintaining compactness and also an ease in scale-up or
et al. (2001b) design, and much lower that in Liu et al. (1996) numbering-up (Wang et al., 2008). Greater capacity at lower cost
design. is on all fours with reduced pressure drops, whereas controlling
2122 H. Llerena-Chavez, F. Larachi / Chemical Engineering Science 64 (2009) 2113 -- 2126

maldistribution is compulsory for scale-up. On the basis of the pre- Maldistribution within the rotor section was seen to be due
vious interpretations, new designs can be imagined which would to position and type of gas inlet. To improve radial gas distribu-
allow reduction of maldistribution. tion in rotor section, five new designs were examined (Fig. 11a–e).

8 150 m3/h
0.1
PRM = 1090
7 0.09
Rectangular Packing Global
6 0.08
870 rpm I Quadrant
0.07
rMfB (r,h,Θ = 2π)

II Quadrant

rMfB (r,h,Θ = π/4)

5 1300 rpm
1700 rpm 0.06 III Quadrant
4 IV Quadrant
0.05
3 0.04
0.03
2
0.02
1
0.01
0 0
4 4.5 5 5.5 6 6.5 7 7 8 9 10 11 12 13 14 15
Radius (cm) Radius (cm)

Fig. 8. Radial profiles of global maldistribution factor, rMfB (r,h, = 2), as a function Fig. 10. Breakdown in quadrant specific maldistribution as a function of radial
of RPM, Liu et al., 1996 design. profiles: Zheng et al. (2000) design; RPM = 1090; GFR = 150 m3 /h.

0.1 0.1
0.09 0.09
0.08 0.08 RPM = 1420
RPM = 950
0.07 0.07 6.5 m3/h
rMfB (r,h,Θ = 2π)

rMfB (r,h,Θ = 2π)

13.1 m3/h
0.06 6.5 m3/h 0.06
13.1 m3/h 19.5 m3/h
0.05 19.5 m3/h 0.05 23.3 m3/h

0.04 23.3 m3/h 0.04

0.03 0.03
0.02 0.02
0.01 0.01
0 0
3 5 7 9 11 13 15 3 5 7 9 11 13 15
Radius (cm) Radius (cm)

0.1
0.1
0.09 6.5 m3/h 0.09
RPM = 950 6.5 m3/h
0.08 0.08 RPM = 1420
Total
0.07 I Quadrant 0.07
rMfB (r,h,Θ = π/4)

Global
rMfB (r,h,Θ = π/4)

0.06 II Quadrant I Quadrant

III Quadrant 0.06
II Quadrant
0.05 IV Quadrant 0.05 III Quadrant
IV Quadrant
0.04 0.04
0.03 0.03
0.02 0.02
0.01 0.01
0 0
3 5 7 9 11 13 15 3 5 7 9 11 13 15
Radius (cm) Radius (cm)

Fig. 9. Radial profiles of global maldistribution factor, rMfB (r,h, = 2), as a function of GFR and RPM: Sandilya et al. (2001b) design (a, b), breakdown in quadrant specific
maldistribution, rMfB (r,h, = /4), RPM = 950, GFR = 6.5 m3 /h (c); RPM = 1420, GFR = 6.5 m3 /h (d).
 H. Llerena-Chavez, F. Larachi / Chemical Engineering Science 64 (2009) 2113 -- 2126 2123

Fig. 11. RPB designs with modified feed options or funneling effects: (a) four inlet pipes placed laterally in housing at 90◦ , (b) four inlet pipes placed in housing parallel to
axis of revolution and extending on top disc of rotating packing, (c) four paddles at 90◦ installed on outer periphery of rotating packing, (d, e) static concentric packed bed
crown placed on housing with top or tangential gas feed.

The Zheng et al. (2000) layout was used as a starting benchmark in 0.1
terms of dimensions of housing, rotor and distribution sections, and 0.09
of operating GFR and RPM. 0.08 GFR = 90 m3/h
First (Fig. 11a) and second (Fig. 11b) designs with quadrupled RPM = 816
0.07
rMBf (r,h,Θ = 2π)

inlets are evolved versions of, respectively, Liu et al. (1996) and
0.06 Layout Fig.11-c
Sandilya et al. (2001b) feed cases. In the latter case, these inlets were
moved towards the interior such that the gas flow impinges against 0.05
Layout Fig.11-a
the wall of the rotor flat disc. The advantage of these new layouts 0.04
vis-à-vis their counterparts of Fig. 2a and b is an improved radial 0.03
distribution of gas in the housing section, whereas the maldistribu- Layout Fig.2-c
0.02
tion descriptors in the packed bed rotating sections are shown in
0.01 Layout
Fig. 12. Fig. 11b layout achieves better distribution than the one of Fig.11-b
Fig. 11a as shown in Fig. 12 for GFR = 90 m3 /h and RPM = 816. The 0
7 8 9 10 11 12 13 14 15
third design which adjoins four paddles located on the rotor outer
Radius (cm)
periphery (Fig. 11c) causes the tangential gas distribution to deteri-
orate with respect to the other designs and especially towards the Fig. 12. Comparisons of radial profiles of global maldistribution factor,
housing side (Fig. 12). rMfB (r,h, = 2), for the different designs shown in Figs. 11a–c.
The fourth and fifth (Fig. 11d and e) layouts use concentric static
packed beds placed in the housing section to help funneling the
gas flow towards the rotor section in a more radial manner. The the rotor. The feed inlet is as shown in Fig. 11d and e. As shown in
concentric packed beds are 5 cm wide (radial-wise), fill the whole Fig. 13, the gas distributions obtained with this artifice were much
depth of the housing section (axial-wise), and are located 3 cm from better in comparison to Fig. 2c layouts without concentric static bed.
2124 H. Llerena-Chavez, F. Larachi / Chemical Engineering Science 64 (2009) 2113 -- 2126

0.3 gas velocities such that u < r signify that the gas flow is forced in
the  direction by the rotating bed. Conversely, when u > r, the
GFR = 30 m3/h bed is pushed tangentially by the gas flow in addition to the contri-
RPM = 1090
bution by the rotor own rotation. Negative relative tangential veloc-
Layout Fig.2-c
0.2 Normal ities arise at low GFR but a change in sign occurs beyond a threshold
rMfB (r,h,Θ = 2π)

Static bed crown GFR which depends on the rotor RPM. The simulations also revealed
a tendency towards virtually slipless tangential flows, i.e., u = r,
when decreasing bed porosity (results not shown). Expressing the
radial pressure gradient in a centrifugal field by means of an Ergun-
0.1 like drag force expression for the packed bed section and consider-
Layout Fig.11-e
ing in it the gas speed relative to rotor frame, one obtains (Sandilya
et al. (2001b):
2
dP (1 − )2 (1 − ) u u2
+  + r
0
= 150 2 3 Urel + 1.75 Urel
2
(4)
7 8 9 10 11 12 13 14 15 dr d  d 3 r r
Radius (cm)
The last two terms in Eq. (4) represent, respectively, the centrifu-
Fig. 13. Comparisons of radial profiles of global maldistribution factor, gal pressure gradient and momentum gain due to radial acceleration.
rMfB (r,h, = 2), for the designs shown in Figs. 2c and 11e. For circumferentially uniform radial flows across a stationary
porous medium, integration of Eq. (4) yields the pressure drop ex-
pression which is solely determined by the gas-packing drag effects:
2.5  
Q1
150(1 − )2  GFR ro
Q2
Rotor (ωr)
PPacked bed = ln
d2 3 2 h ri
2.0 Q3    
1.75(1 − ) GFR 2 1
Tangential velocity (m/s)

1
Q4 + − (5)
d 3 2 h ri ro
1.5
The problem to calculate the pressure drop in the rotating packing
Q5 module without recourse to CFD simulations is posed by the a priori
1.0 unknown u as this exhibits non-linear behavior as exemplified in
RPM = 300 Fig. 14. Rigorous radial velocity calculations would also be necessary
ε = 91% depending on the magnitude of the ratio 1.75(1−)r/d3 . It is worthy
0.5 Q1> Q2> …> Q5 of notice that the approximation of a constant relative tangential
slip velocity, u −r, is acceptable over some inner portions of the
bed (Fig. 14). It is suggested that u −r contribution in Eq. (4) can
0.0 be decoupled by silencing it in the laminar and inertial drag terms
0.035 0.04 0.045 0.05 0.055 0.06 0.065 0.07 0.075 0.08
and by defining an ad hoc empirical corrective function, Fc , which
Radius (m) internalizes the effects of relative tangential slip velocity and radial
acceleration of the gas. In these circumstances, the radial pressure
Fig. 14. Radial profiles of gas tangential velocity in rotating packed bed section for
increasing gas flow rates. Dotted line shows rigid body-like rotation by rotor. gradient in the packed bed zone can be expressed as

dP 150(1 − )2  1.75(1 − ) 2

= ur + ur
3.6. CFD-aided development of pressure drop relationships for rotating dr d2 3 d3
2
packing +  r + Fc (6)

In which Fc is a function that accounts for GFR, RPM, bed porosity,

CFD appears to be powerful for computing dry pressure drops
gas entrance and 3D effects. Note also that the centrifugal force term
whatever the geometrical specifications of RPB sections. In what fol-
was approximated in Eq. (6) as 2 r.
lows, its potential to develop semi-empirical pressure drop relation- Integration of Eq. (6) from ri to ro yields a semi-empirical dry
ships for the packed bed sections will be examined. pressure drop expression which accounts in an additive manner of
For a gas flowing through the rotor section, the gas speed relative the drag and centrifugal forces, the gas–solid slip and radial acceler-
to the rotor frame is expressed as ation effects:
 150(1−) 
2 
GFR

ro 1.75(1−)

GFR
2 
1 1

Urel (r, , z) = u2r (r, , z) + (u (r, , z) − r)2 (3) PPacked bed = ln + −
d2 3 2h ri d3 2h ri ro
 
Drag force
where ur and u are the absolute radial and tangential velocities
1
of the gas at position (r,,z) and where the motion along the axial +
2
2 (ri2 −ro2 ) + Fc

(7)
 
direction, i.e., uz , has been neglected. Centrifugal force
Gas–solid slip

Fig. 14 shows the deviations of the gas tangential velocity for var-
ious GFR as compared to the rigid body-like rotation of rotor drawn
as the dotted straight line u = r. A constant RPM = 300 through
a bed of porosity 0.91 was considered, and the outer (ro ) and inner
(ri ) radii of packed bed were 0.035 and 0.08 m, respectively. These
simulations were performed assuming 2D x–y geometry to illustrate
how Coriolis and inertial effects are reflected in the tangential slip
velocity. In a more general case, these deviations depend on GFR,
RPM, bed porosity, and 3D effects inherent to the RPB. Tangential
The corrective function Fc is obtained through correlating the
difference between the CFD simulated pressure drop in the packed
bed portion and the drag and centrifugal terms in Eq. (7) above.
Eq. (8) illustrates the dependences of this corrective function with
respect to GFR, RPM, bed porosity and 3D geometrical specifications
as accounted for via the parameters a, b and c:
c
Fc = (a − GFR + (b +  )GFR2 ) (8)
 H. Llerena-Chavez, F. Larachi / Chemical Engineering Science 64 (2009) 2113 -- 2126
300
Pressure drop - Calculated Eq.7
250
+/- 4 %
200
150
100
50
0
0 50 100 150 200 250
Pressure drop - CFD
Fig. 15. Parity plot of the pressure drop contributed by the rotating packed bed
comparing CFD simulated versus approximation by means of Eqs. (7) and (8) for
the designs of Liu et al. (1996) and Zheng et al. (2000).
For the cases of Liu et al. (1996) and Zheng et al. (2000), the fitted
parameter values were found to be a = −0.08 m3 /s, b = 2000 (rpm)c ,
whereas c = 2.34 (Liu et al., 1996 design) and c = 1.22 (Zheng et al.,
2000 design). Fig. 15 is a parity plot of the CFD predicted packed bed
pressure drops versus the ones given by Eq. (7). The ± 4% envelopes
indicate that the correlation Eq. (8) introduced in Eq. (7) achieves a
very good approximation for the pressure drop in the RPB. As Eq. (8)
involves easily measurable operating variables, it enables estimation
of the pressure drop in the RPB for a given RPB design and how it
compares to easily measurable overall pressure drop.
4. Conclusion
The 3D non-steady state turbulent rotating single-phase flows
were simulated with CFD in RPBs and were validated using overall
dry pressure drop measurements for three RPB designs. The gas cir-
culation was analyzed for the housing cavity, the RPB, the free inner
rotational zone, and the gas disengagement.
The salient features emerging from the analysis of maldistribution
can be summarized as follows:
• Liu et al. (1996) design equipment featured large pressure drop
in the free rotation zone, large maldistribution factor due to the
position of gas flow inlet, whereas the small diameter gas flow
outlet highlighted a stack effect.
• Sandilya et al. (2001b) layout exhibited low gas radial maldistri-
bution except throughout a bed thickness adjacent to the housing
section.
• Zheng et al. (2000) equipment showcased low bed maldistribution
especially the higher the GFR; however, the highest dissipation
lied in the free rotation and distributor zones.
Alternative designs were also analyzed with respect to different
feeding configurations. It was concluded that better tangential dis-
tributions in the rotating packing module could be achieved by using
four gas injectors placed on top of the rotor cladding or by inserting
a static concentric packed bed in the housing section to reduce the
gas phase tangential flow. Though this study conveyed a new and
important information that single-phase flow simulations unveiled
maldistribution of gas circulation in RPB devices, the dysfunctional
operation of RPBs must also be looked at closely in any future pub-
lished studies on two-phase flows in these reactors to assess the
extent of this phenomenon.
2125
Finally, the breakdown of the overall pressure drop in its mod-
ular components for the housing, the RPB, the free inner rotational
zone, and the gas disengagement, showed that the dissipation in the
rotating packing could be a minor contributor to the overall pres-
sure drop. Analysis of the simulated pressure drops allowed CFD-
based Ergun-type relationships to be developed wherein the gas-slip
and radial acceleration effects, the laminar and inertial drag effects,
and the centrifugal effect, were aggregated to reconstruct pressure
drops in the rotating packing module. It must be emphasized that
sometimes the pressure drops do not adhere to simple monotonic
behaviors as a function of rotational speed and GFR (Singh et al.,
1992) suggesting that the presence of a second immiscible phase
such as liquid flow can bring about important and atypical effects
on the pressure drop behavior. This means that validation by “dry
300 bed” (or gas-only) must only be viewed as a preliminary step before
two-phase flow simulations are undertaken.
Acknowledgments
H.L.C. would like to acknowledge financial support from the
MECESUP MAG9901 of the Chilean Government for his scholarship.
Financial support from the Natural Sciences & Engineering Research
Council of Canada is also acknowledged.
References
Bašic,
 A., Dudukovic,  M.P., 1995. Liquid holdup in rotating packed beds: examination
of the film flow assumption. A.I.Ch.E. Journal 41, 301–316.
Batchelor, G.K., 1951. Note on a class of solutions of the Navier–Stokes equations
representing steady rotationally symmetric flow. Quarterly Journal of Mechanics
and Applied Mathematics 4, 29–41.
Bodonyi, R.J., Stewartson, K., 1977. Unsteady laminar boundary-layer on a rotating
disk in a counter-rotating fluid. Journal of Fluid Mechanics 79, 669–688.
Burns, J.R., Ramshaw, C., 1996. Process intensification: visual study of liquid
maldistribution in rotating packed beds. Chemical Engineering Science 51,
1347–1352.
Burns, J.R., Jamil, J.N., Ramshaw, C., 2000. Process intensification: operating
characteristics of rotating packed beds—determination of liquid hold-up for a
high-voidage structured packing. Chemical Engineering Science 55, 2401–2415.
Chen, Y.S., Liu, H.S., 2002. Absorption of VOCs in a rotating packed bed. Industrial
and Engineering Chemistry Research 41, 1583–1588.
Chen, J., Shao, L., 2003. Mass production of nanoparticles by high gravity reactive
precipitation technology with low cost. China Particuology 1, 64–69.
Chen, J., Shao, L., Zhang, C.G., Chen, J.M., Chu, G.G., 2003. Preparation of TiO2
nanoparticles by a rotating packed bed reactor. Journal of Materials Science
Letters 22, 437–439.
Chen, Y.H., Chang, C.Y., Su, W.L., Chen, C.C., Chiu, C.Y., Yu, Y.H., Chiang, P.C., Chiang,
S.I.M., 2004a. Modeling ozone contacting process in a rotating packed bed.
Industrial and Engineering Chemistry Research 43, 228–236.
Chen, J., Li, Y.L., Wang, Y.H., Yun, J., Cao, D.P., 2004b. Preparation and characterization
of zinc sulfide nanoparticles under high-gravity environment. Materials Research
Bulletin 39, 185–194.
Chen, J.-F., Zhou, M.-Y., Shao, L., Wang, Y.-Y., Yun, J., Chew, N.Y.K., Chan, H.-K.,
2004c. Feasibility of preparing nanodrugs by high-gravity reactive precipitation.
International Journal of Pharmaceutics 269, 267–274.
Cummings, C.J., Quarderer, G., Tirtowidjojo, D., 1999. Polymer devolatilization and
pelletization in a rotating packed bed. BHR Group Conference Series Publication,
38, Process Intensification for the Chemical Industry, pp. 147–158.
Das, A., Bhowal, A., Datta, S., 2008. Continuous biosorption in rotating packed-bed
contactor. Industrial and Engineering Chemistry Research 47, 4230–4235.
Eimer, D., 2003. PI thinking applied to natural gas treating. In: Gough, M. (Ed.),
Fifth International Conference Process Intensification for the Chemical Industry,
Maastricht, Netherlands, 13–15 October. Better Processes for Bigger Profits. BHR
Group Ltd., Cranfield, UK, pp. 179–189.
Fluent User's Guide, 2006. Chap. 12.4.2 RNG k– model, Fluent Inc., Lebanon, NH,
USA.
Guo, K., Gou, F., Feng, Y.D., Chen, J.F., Zheng, C., Gardner, N.C., 2000. Synchronous
visual and RTD study on liquid flow in rotating packed bed contactor. Chemical
Engineering Science 55, 1699–1706.
He, G., Cheng, G., Deng, X., An, Y., 2003. Pilot test rig for mass-transfer performance
of absorbers in AIP and measurement of KL a. Huazhong Keji Daxue Xuebao,
Ziran Kexueban 31, 51–54.
Jackson, G.S., Marchell, J.M., 1968. Correlation of gravitational force for absorption
in packed columns. Industrial & Engineering Chemistry Process Design and
Development 7, 359–361.
Jarre, S., LeGal, P., Chauve, M.P., 1996. Experimental study of rotating disk instability.
I. Natural flow. Physics of Fluids 8, 496–508.
2126 H. Llerena-Chavez, F. Larachi / Chemical Engineering Science 64 (2009) 2113 -- 2126
Keyvani, M., Gardner, N.C., 1989. Operating characteristics of rotating beds. Chemical
Engineering Progress 85, 48–52.
Karunakumari, L., Eswaraiah, C., Jayanti, S., Narayanan, S.S., 2005. Experimental and
numerical study of a rotating wheel air classifier. A.I.Ch.E. Journal 51, 776–790.
Li, Y.-L., Wang, Y.-H., Chen, J.-F., Yun, J., 2003. Preparation of zinc sulfide nano-
particles by high-gravity method. Wuji Cailiao Xuebao 18, 1362–1366.
Li, X.-P., Liu, Y.-Z., Li, Z.-Q., Wang, X.-L., 2008. Continuous distillation experiment
with rotating packed bed. Chinese Journal of Chemical Engineering 16, 656–662.
Lin, C.-C., Chen, Y.-S., Liu, H.-S., 2000. Prediction of liquid holdup in counter-
current flow rotating packed bed. Chemical Engineering Research and Design 78,
397–403.
Lin, C.-C., Liu, W.-T., Tan, C.-S., 2003. Removal of carbon dioxide by absorption
in a rotating packed bed. Industrial and Engineering Chemistry Research 42,
2381–2386.
Lin, C.-C., Su, Y.-R., 2008. Performance of rotating packed beds in removing ozone
from gaseous streams. Separation & Purification Technology 61, 311–316.
Lin, C.-C., Chen, B.-C., Chen, Y.-S., Hsu, S.-K., 2008. Feasibility of a cross-flow rotating
packed bed in removing carbon dioxide from gaseous streams. Separation &
Purification Technology 62, 507–512.
Liu, H.-S., Lin, C-C., Wu, S.-C., Hsu, H.-W., 1996. Characteristics of a rotating packed
bed. Industrial and Engineering Chemistry Research 35, 3590–3596.
Lockett, M.J., 1995. Flooding of rotating structured packing and its application to
conventional packed columns. Chemical Engineering Research and Design 73,
379–384.
Ma, J., Wang, Y.-H., Wang, D.-M., Chen, J.-F., Yun, J., 2004. Preparation of particles
of ultra-fine salbutamol sulfate by anti-solvent re-crystallization in a rotating
packed bed. Cailiao Kexue Yu Gongcheng Xuebao 22, 74–77.
Mellor, G.L., Chapple, P.J., Stoves, V.K., 1968. On the flow between a rotating and a
stationary disk. Journal of Fluid Mechanics 31, 95–112.
Munjal, S., Dudukovic,  M.P., Ramachandran, P., 1989a. Mass transfer in rotating
packed beds (I. Development of gas–liquid and liquid–solid mass transfer
correlations). Chemical Engineering Science 44, 2245–2256.
Munjal, S., Dudukovic,  M.P., Ramachandran, P., 1989b. Mass transfer in rotating
packed beds (II. Experimental results and comparison with theory and gravity
flow). Chemical Engineering Science 44, 2257–2268.
Nallasamy, M., 1987. Turbulence models and their applications to the prediction of
internal flows: a review. Computers and Fluids 15, 151–194.
Pan, C., Deng, X., 2002. Process of dual-alkali FGD in multi-stage-spraying rotating
packed bed. Huanjing Wuran Zhili Jishu Yu Shebei 3, 88–90.
Peel, J., Howarth, C.R., Ramshaw, C., 1998. Process intensification: HIGEE seawater
deaeration. Chemical Engineering Research and Design 76, 585–593.
Podbielniak, W.J., 1966. Continuous centrifugal vapour–liquid contactor. US Patent
3,233,880.
Podbielniak, W.J., 1967. Continuous centrifugal deodorizer. US Patent 3,336,015.
Raimundo, A.M., Oliveira, L.A., Figueiredo, A.R., 2002. Numerical simulation and
experimental validation of heat transfer within rotating flows for three-
dimensional non-axisymmetric, turbulent conditions. International Journal for
Numerical Methods in Fluids 40, 821–840.
Ramshaw, C., Mallinson, R.H., 1981. Mass transfer process. US Patent 4,283,255.
Ramshaw, C., 1983. HIGEE Distillation—an example of process intensification.
Chemical Engineering (London) 389, 13–14.
Ramshaw, C., 1993. The opportunities for exploiting centrifugal fields. Heat Recovery
Systems & CHP 13, 493–513.
Rao, D.P., Bhowal, A., Goswami, P.S., 2004. Process intensification in rotating packed
beds (HIGEE): an appraisal. Industrial and Engineering Chemistry Research 43,
1150–1162.
Sandilya, P., Biswas, G., Rao, D.P., Sharma, A., 2001a. Numerical simulation of the gas
flow and mass transfer between two coaxially rotating disks. Numerical Heat
Transfer, Part A 39, 285–305.
Sandilya, P., Rao, D.P., Sharma, A., Biswas, G., 2001b. Gas-phase mass transfer
in a centrifugal contactor. Industrial and Engineering Chemistry Research 40,
384–392.
Serre, E., Bontoux, P., Kotarba, R., 2001. Numerical simulation of the transition
in three-dimensional rotating flows with walls: boundary layer instability.
International Journal of Fluid Dynamics 5 (Article 2), 17–30.
Shi, Q.P., Ribando, R.J., 1992. Numerical simulations of viscous rotating flows using
a new pressure-based method. Computers and Fluids 21, 475–489.
Singh, S.P., Wilson, J.H., Counce, R.M., Villiers-Fisher, J.F., Jenning, H.L., 1992. Removal
of volatile organic compounds from groundwater using a rotary air stripper.
Industrial and Engineering Chemistry Research 31, 574–580.
Song, Y., Chen, J., Fu, J., Chen, J., 2003. Research on particle removal efficiency of
rotating packed bed. Huagong Jinzhan 22, 499–502.
Stewartson, K., 1953. On the flow between two rotating coaxial disks. Proceedings
of the Cambridge Philosophical Society 49, 333–341.
Szeri, A.Z., Schneider, S.J., Labbe, F., Kaufman, H.N., 1983a. Flow between rotating
disk. Part 1. Basic flow. Journal of Fluid Mechanics 134, 103–131.
Szeri, A.Z., Giron, A., Schneider, S.J., Kaufman, H.N., 1983b. Flow between rotating
disk. Part 2. Stability. Journal of Fluid Mechanics 134, 133–154.
Todd, D.B., 1969. Multistage vapour–liquid contactor. US Patent 3,486,743.
Tan, C.-S., Lee, P.-L., 2008. Supercritical CO2 desorption of toluene from activated
carbon in rotating packed bed. Journal of Supercritical Fluids 46, 99–104.
Trent, D., Tirtowidjojo, D., 2001. Commercial operation of a rotating packed bed (RPB)
and other applications of RPB technology. Better processes for better products.
In: Proceedings of the Fourth International Conference on Process Intensification
for the Chemical Industry, Brugge, Belgium, 10–12 September 2001, pp. 11–19.
Trent, D., Tirtowidjojo, D., 2002. Piloting and commercialization of rotating packed
bed technology for reactive distillation. In: A.I.Ch.E. Annual Meeting, Indianapolis,
IN, USA, 3–8 November 2002.
Tung, H.H., Mah, R.S.H., 1985. Modeling of liquid mass transfer in HIGEE separation
process. Chemical Engineering Communications 39, 147–153.
Vivian, J.E., Brian, P.L.T., Krukonis, V.J., 1965. The influence of gravitational force on
gas absorption in a packed column. A.I.Ch.E. Journal 11, 1088–1091.
Wang, G.Q., Xu, O.G., Xu, Z.C., Ji, J.B., 2008. New HIGEE-rotating zigzag bed and its
mass transfer performance. Industrial and Engineering Chemistry Research 47,
8840–8846.
Xinlin, D., Xiaoyong, H., Yigang, D., Yuanxin, W., Dinghuo, L., 2000. A model
for mass transfer coefficient in rotating packed bed. Chemical Engineering
Communications 178, 249–256.
Yang, L.-R., Liu, Y.-Z., Jiao, W.-Z., Guo, Y., 2004. Study on treating phenol wastewater
through impinging stream-rotating packed bed by using emulsion liquid
membrane. Huagong Keji 12, 40–43.
Yang, S., Lin, C.-C., Tseng, I.-M., Liu, W.-T., Yu, H.-T., 2005. Method for removing
volatile components from a high viscosity liquid by using rotation pack bed. US
Patent 6,884,401.
Yeung, P.K., Zhou, Y., 1998. Numerical study of rotating turbulence with external
forcing. Physics of Fluids 10, 2895–2909.
Zhang, L., Zhang, P., Chen, J., 2004. Preparation and characterization of nanofibrous
-Al2 O3 . Shiyou Huagong 33, 240–243.
Zheng, C., Guo, K., Song, Y., Zhou, X., Al, D., Gardner, N.C., 1997. Industrial practice
of HIGRAVITEC in water deaeration. In: Proceedings of the Second International
Conference Process Intensification, BHR Group Conference Series Publication,
vol. 28, London, UK, pp. 273–285.
Zheng, C., Guo, K., Feng, Y.D., Yung, C., 2000. Pressure drop of centripetal gas
flow through rotating bed. Industrial and Engineering Chemistry Research 39,
829–834.
Zhou, M., Chen, J., Liu, X., 2003. Preliminary research on recrystallization method to
produce micronized ibuprofen particles. Huagong Jinzhan 22, 524–528.
Zhu, W., Chen, J., Wang, Y., 2003. Crystallization process and product stability of the
ultrafine calcium carbonate whiskers in high-gravity field. Qufu Shifan Daxue
Xuebao, Ziran Kexueban 29, 77–81.
Zhu, W., Chen, J., Wang, Y., 2004. Synthesis and characterization of ultra-fine calcium
carbonate whiskers in high-gravity. Huaxue Wuli Xuebao 17, 175–178.