Powder Technology 228 (2012) 416423

Contents lists available at SciVerse ScienceDirect

Powder Technology
journal homepage: www.elsevier.com/locate/powtec

Review

A review of the Residence Time Distribution (RTD) applications in solid

unit operations
Yijie Gao, Fernando J. Muzzio, Marianthi G. Ierapetritou
Department of Chemical and Biochemical Engineering, Rutgers, The State University of New Jersey, 98 Brett Road, Piscataway, NJ 08854, USA

a r t i c l e

i n f o

Article history:
Received 30 January 2012
Received in revised form 21 May 2012
Accepted 26 May 2012
Available online 9 June 2012
Keywords:
Solids
RTD
Unit operation
Modeling
Measurement
Performance

a b s t r a c t
This review traces current applications of the residence time theory in various solid unit operations. Besides
reviewing recent experimental and simulation studies in the literature, some common modeling and tracer
detection techniques applied in continuous ow systems are also considered. We attempt to clarify and emphasize the inuence of the residence time prole on the unit performance, which is the key in system design
and performance improvement of practical unit operations. The development of predictive modeling is also
an important goal in the long-term development of the residence time theory.
2012 Elsevier B.V. All rights reserved.

Contents
1.
2.

Introduction . . . . . . . . . . . . . .
RTD modeling . . . . . . . . . . . . .
2.1.
CSTR and PFR series . . . . . . .
2.2.
Axial dispersion model . . . . . .
2.3.
Stochastic model and Markov chain
2.4.
Bimodal RTD . . . . . . . . . .
2.5.
Convolution . . . . . . . . . . .
2.6.
RTD constructed by velocity prole
3.
RTD measurement . . . . . . . . . . .
4.
RTD applications in solid process . . . .
4.1.
Continuous blender . . . . . . .
4.2.
Extruder . . . . . . . . . . . .
4.3.
Rotary drum . . . . . . . . . .
4.4.
Fluidized bed . . . . . . . . . .
5.
Conclusion . . . . . . . . . . . . . . .
Acknowledgement . . . . . . . . . . . . .
References . . . . . . . . . . . . . . . . .

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

1. Introduction
In chemical engineering and related elds, the Residence Time
Distribution (RTD) is dened as the probability distribution of time
that solid or uid materials stay inside one or more unit operations
Corresponding author. Tel.: + 1 732 445 2971; fax: + 1 732 445 2581.
E-mail address: [email protected] (M.G. Ierapetritou).
0032-5910/$ see front matter 2012 Elsevier B.V. All rights reserved.
doi:10.1016/j.powtec.2012.05.060

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

416
417
417
417
418
418
418
419
419
419
419
420
420
421
421
421
421

in a continuous ow system. It is a crucial index in understanding

the material ow prole, and is widely used in many industrial processes, such as the continuous manufacturing of chemicals, plastics,
polymers, food, catalysts, and pharmaceutical products. In order to
achieve satisfactory output from a specic unit operation, raw materials are designed to stay inside the unit under specic operating conditions for a specied period of time. This residence time information
is usually compared with the time necessary to complete the reaction

Y. Gao et al. / Powder Technology 228 (2012) 416423

or process within the same unit operation. For example, in continuous

powder mixing processes, powder is mixed in a continuous mixer.
The local mixing rate coupled with the time the powder stays inside
the mixer determines the unit performance. If the time required for
local mixing is longer than the actual residence time powder stays
in the system, the process cannot provide a complete mixture, and
it fails its designed purpose [1]. In other words, the performance of
any continuous unit operation is determined by the competition of
the two sub-processes: a batch process or reaction superimposed by
an axial ow. Therefore, the characterization of the RTD in different
continuous unit operations is the rst step in the design, improvement, and scale-up of many manufacturing processes in the chemical
engineering industry.
The research on the RTD in chemical engineering elds has focused on the inuence of operation conditions, materials, and the
unit geometry on the RTD prole, the improvement of measurement
methods, and the improvement of predictive modeling on different
processes and units. Most studies investigated continuous unit operations by using the RTD; few extended to the correlation between
the RTD and the reaction or process performance, which is usually
case-sensitive. For example, a continuous polymer foaming process
was studied in an extruder [2], where the thermal decomposition
rate of chemical blowing agent was compared with the RTD to investigate the optimization of the foam density; a chemical-looping combustor was investigated in both continuous and batch mode, in which
the RTD was used to develop a model for predicting the mass-based
reaction rate constant for char conversion [3]; the production of polypropylene was characterized in a horizontal stirred bed reaction by
considering the RTDs of catalyst and polymer separately, which
strongly depend on the temporal catalyst activity [4]; the emulsication process in polymer mixing was studied in a twin-screw continuous extruder, where the RTD and the morphology prole of the
mixture was examined simultaneously in one pulse test [5]; the
Cr(VI) reduction in wastewater treatment was investigated in an
electrochemical tubular reactor by applying CFD and velocity prole,
where the performance was coupled with the axial ow rate [6].
Due to the wide scope of the RTD issue, every year a large number
of papers have been published using this conception in a host of disciplines. A previous review by Nauman [7] summarized the theoretical development history of the RTD since the beginning of the last
century, especially for continuous uid systems. Some developments
have occurred since the previous review that will be covered in this
paper. Moreover, this review mainly proles the applications of the
RTD theory in characterizing solid chemical engineering unit operations. Also, this review differs from the previous one in that it emphasizes coupling the RTD with the unit performance of specic unit
operations. This paper is organized in the following manner.
Section 2 illustrates different modeling work of the RTD in these applications, followed by a discussion on general RTD measurement
methods in Section 3. In Section 4, recent applications of the RTD in
solid continuous ow and manufacturing systems are described in
details. Our goal in this review paper is to bring together the recent
applications of the RTD theory across a wide range of studies in the
chemical engineering elds, and contribute to the performance investigation of versatile continuous unit operations.
2. RTD modeling
2.1. CSTR and PFR series
Modern RTD theory originally developed from continuous uid
systems [8]. Early uid reactor models assumed plug ow in a
tubular-shape reactor (PFR), or perfect mixing in continuous stirred
tank reactors (CSTR). These conceptions represent two extreme RTD
proles in the reactor. In practical continuous ow systems, experimental RTD proles are usually between the two extremes. To

417

describe the non-idealness of the RTD prole, different combinations

of CSTR and PFR were introduced in modeling practical cases. CSTR in
a series model is one commonly used model [3,911]:
NV 0 =F
E

NNN1
expN
N1!

1
2

where is the mean residence time (MRT), N the number of CSTR

tanks, V0 the volume of each tank, and F the volumetric ow rate.
E() represents the dimensionless RTD and = t/ the dimensionless
time. As a one-parameter model, the idealness of the RTD is represented by the number of CSTR tanks used (Fig. 1). Large number of
tanks indicates a PFR-like reactor (N ), and a small number leads
to a CSTR-like reactor (N = 1). Two modications were reported:
the tanks in series were followed by a PFR element in case of the
RTD rise part delay when axial dispersion is signicantly limited
[2,12]; backward ux was introduced among the CSTR tanks to capture the long tail in the RTD proles [4,13]. In the second modication, large backward ux indicates fast material exchange between
adjacent tanks. The long tail prole can also be modeled by a dead
zone volume cross-owing with the CSTR element [1416]. Notice
that the fraction of the dead zone element represents the degree of
non-idealness of the continuous ow system. Amador et al. [17]
reported a resistance network model that can be considered as parallel connections of a series of PFR elements [18], thus also belonging to
this kind of model.
2.2. Axial dispersion model
Despite the combination of CSTR and PFR elements, the axial dispersion model is an efcient alternative to generalize the conception
of the RTD to most non-ideal reactors. The differential equation representing the axial dispersion process of a tracer in the ow system,
or the FokkerPlanck equation (FPE), is expressed as a global 1D
equation:
c
2 c
c
E 2 vz
t
z
z

where c is the concentration of a component in the system, vz is the

global axial velocity, and E denotes the dispersion coefcient in
solid systems, or the diffusion coefcient in uid systems, respectively. The variables t and z represent time and axial distance from
the tracer injection point. The advantage of the FPE is the clear physical meaning of its parameters, where vz and E indicates the combination of the axial transport and the superimposed axial dispersion
in this model. There have been many industrial studies on the RTD
directly using the FPE [13,1923]. Based on previous literature
[24,25], Sherritt et al. [26] summarized various FPE solutions under
different boundary conditions in a rotary drum. For example, Vashisth and Nigam [27] applied the solution based on error function
to analyze single-phase laminar ow through a straight tube. The
most widely used solution of the RTD was developed by Taylor [28]
with openopen boundaries:
(
)
1
Pe12
E p exp
4
2 =Pe

where = t/ is the dimensionless time, is the mean residence time

(MRT); Pe = vzl/E represents the Peclet number, and l is the distance
between injection and detection points. This solution was applied in
the tting of experimental RTD data in various systems [1,9,2931].
Although some of the reported systems were not single phase or

418

Y. Gao et al. / Powder Technology 228 (2012) 416423

Fig. 1. E curves of some RTD models reviewed in this work.

with openopen boundary conditions as when this solution was derived, good agreement was obtained, especially for rise-delayed or
long-tailed RTD proles, showing the robustness of the Taylor's dispersion model. A more convenient way to estimate the Peclet number without RTD curve tting is using the following formulas on
the experimental RTD data:

0 tEt dt

s
2
8

Pe Pe2

where E(t) represents the RTD as a function of time, and is the second moment of the RTD:
2

0 t Et dt

Recent applications of Eq. (6) can be found in the work of Sudah et

al., Vikhansky, Fang et al., Kumar et al., and Waje et al. [23,3235]. In
granular systems, a unique method to estimate the dispersion coefcient was reported by using information of single particle movement
[22,36,37]:
x
E lim

t0

x2
2t

where x 2 is the square displacement of a single particle within the

time interval t. The symbol b > indicates the statistical expectation
of the parameter within. As these studies indicate, the prediction of
the axial dispersion coefcient is the key step for the development
of predictive RTD models.
2.3. Stochastic model and Markov chain
Another widely used RTD model is to consider the particle movement as a stochastic process. In particular the axial motion of any particle or small fraction of material inside a continuous system is
considered as a two dimensional probabilistic process. As a result,
the overall RTD curve can be calculated by the accumulation of the
random axial motions of these small fractions. It can be applied to a
process where the predictive RTD model is absent, for example, due
to complicate system geometry [38]. One stochastic model that is

similar to the CSTR and PFR series is the Markov chain model. This
model arbitrarily denes the ratio of ux exchange among assumed
elements connected in a network, and the next state of the system depends only on the current state and nothing else:
Sn 1 PSn

S(n) is a 1 m state vector of the system, describing the material distribution at time tn, where m is the total number of elements in the
system; P is a m m matrix of transition probabilities for the time interval between two adjacent states. Although the exchange of ux is
well modeled among different elements, mixing inside each element
is not considered in a Markov chain model, which is different from
the CSTR and PFR series model. A detailed description of the Markov
chain can be found in [39]. This model is very useful when the ow
regions within the system can be easily distinguished into different
elements, for instance, the granular ow with heterogeneous crosswise layers [16,40], the gas ow at different regions in an entrained
ow gasier [41], and in uidized beds [42,43].
2.4. Bimodal RTD
Bimodal RTD prole with dual peaks is not common in the literature. It results from two or more main ow components traveling differently, for instance, through two paths in reactor [9,44], or on
different layers in a granular ow system [40]. The dual peaks result
from the overlap of RTD components through different paths [27].
Due to case-sensitivity, no quantitative model has been developed
for bimodal RTD prole.
2.5. Convolution
In uid systems, the non-idealness of a tracer injection is common
due to a relatively short MRT compared to the injection period of a
uid pulse. To correct this non-ideal RTD, recent studies applied the
algorithm of convolution, introduced by Danckwerts [8], as the injection was usually not a perfect pulse [9,30,45,46]:
t

Eout t 0 Ein tEd

10

or
Eout t Ein t Et

11

where Eout(t) and Ein(t) represent the RTD curve before correction
and the imperfect pulse injection detected at the input of the system.

Y. Gao et al. / Powder Technology 228 (2012) 416423

The convolution integration corresponds to multiplication in the frequency domain, therefore:

FFT fEout t g FFT fEin t gFFT fEt g
Et FFT



FFT fEout t g
FFT fEin t g

12
13

where FFT and FFT 1 represent the fast Fourier transform and the inverse Fourier transform operations. In continuous blending systems,
this knowledge was also applied in characterizing the attenuation of
feeding uctuations [29]. Another application of convolution was
reported in the assemble of the overall RTD proles from the RTDs
of single elements in an extruder study [47]:
Et E1 t E2 t En t

14

Similar RTD assemble studies can be found in Gao et al. and

Essadki et al. [1,9].
2.6. RTD constructed by velocity prole
Various CFD software and velocity sensors were used to give velocity proles in uid systems, which were further applied to construct RTD proles [6,27,44,48,49]. In solid systems, the equivalent
tools are the Discrete Element Method (DEM) and positron emission
particle tracking (PEPT) [22,36,50]. However, due to the variability of
the velocity prole in most solid unit operations, the RTD evaluation
based on practical velocity prole is not common. In one specic
case, the convection model was specically used in describing the
RTD proles of laminar tubular ow. The axial dispersion is constructed from the well-known parabolic laminar velocity prole in
the tube by applying classic uid mechanics. The MRT is the spatial
time = Q/V where Q is the volumetric ow rate and V is the internal
volume of the tube. The application of a generalized form of the convection model can be found in both uid and solid extrusion systems
[4547].
3. RTD measurement
The stimulus response test is a common method used in most experimental RTD measurement. In this test, pulse injection or step
change of a tracer is performed at the inlet of a continuous system
where steady state of bulk ow is reached, and response of the tracer
prole at the outlet is recorded. The selected tracer is expected to
share similar properties with the bulk material, thus introducing as
few disturbances as possible on the bulk ow. For example, Harris
et al. [51] applied a phosphorescent pigment as the bulk ow in a circulating uidized bed (CFB), and the tracer was generated by activating the same kind of pigments. Thus, the tracer material is identical to
the bulk material, and hydrodynamic ow disturbances were
avoided. Another assumption of stimulus response test is a closed
inlet and outlet boundaries. This guarantees unidirectional ow and
no boundary dispersion. Exceptionally, in some systems with open
open boundaries, boundary dispersion can also be considered negligible when axial transport is much faster than the dispersion (for example, with large Peclet number). Therefore, the stimulus response
test developed under the assumption of closed boundaries can be applied in some open systems.
To our knowledge two types of tracer detection techniques have
been reported: inline and ofine detection. Inline detections directly
record the optical, thermal, or electrical signal of tracer concentration
from inline probes, and retain the data automatically for further analysis. It requires fast sample acquisition, signal conversion, and data
storage processes. NIR spectroscopy was reported for inline monitoring of drug concentration in a continuous powder blender [52]. When

419

one or more of these steps can not be nished immediately, ofine

detection is selected. For instance, digital imaging analysis can hardly
be inline due to the slow extraction of tracer concentration from digital photos [4,23,34,39,47,53]. Most tracer detections in solid processes were also ofine as the act of solid sampling is usually invasive, and
can not be very fast [54]. The treatment of solid samples, like dissolution of dye or salt tracers, also takes time, leading to ofine detections
[14,15,20,21,35].
The tracer detection can be performed optically or conductively.
Mainly two kinds of optical detection systems were reported. Digital
image processing and color meter were used to analyze the color of
photos or the signal from the sensor directly [11]. Fluorescence analysis also belonged to this kind of detection, which was applied more
in dark environment with accurate sensors [5,30,33]. Another optical
detection reported in many literatures utilizes different types of spectroscopy to analyze the light absorbance of the tracer, for example,
Parker blue dye dissolved in water [55]. Near infrared (NIR) and ultraviolet (UV) are two commonly used regions of the spectrum, reported
by many investigators [12,14,20,27,35,54,56,57]. On the other hand,
conductive detection systems are based on the difference of electrical
conductivity between tracer and bulk ow, for example, NaCl in deionized water [9,45,46,49,58]. Sensitive detector based on the difference of gas thermal conductivity was also reported in a gas uid
system [19]. Helium was used as the inert tracer in this study. Hot
or radioactive particles have been applied as tracer injection and
detected using thermistors or scintillation detectors in gassolid uidization systems [5961]. Tracer using organic gas such as propane
was also reported, the detection of which was using inline gas chromatography in a CFB system [13].
Besides the stimulus pulse test, particle tracking was reported as
an alternative RTD measurement method. Technique such as PEPT,
and the application of DEM or CFD simulations provide movement
and residence time information of a huge number of particles, and
the RTD can be derived from these particle tracking information.
Since no tracer is involved, this method does not introduce the problem of tracer disturbance. However, an additional difculty comes
from the amount of information that has to be recorded from intensive particle tracking experiments or computations, as well as the accuracy of experiments and simulations. The computational data
should be validated with experiments before the simulated RTD prole can be applied in practice. Details of efforts on this issue can be
found in [17,22,36,48,50].
4. RTD applications in solid process
4.1. Continuous blender
As much of the initial research on continuous powder blending
came from the extension of continuous liquid blending based on the
response-stimulus test, the attenuation effect of the RTD on feeding
uctuations was well studied in early theoretical development [8].
The axial mixing component of continuous powder mixing was claried by applying Eqs. (912) [29,62]. These studies on continuous
powder blending were summarized in the review by Pernenkil and
Cooney [63], which also placed emphasis on the similarity of blending
mechanisms between batch and continuous systems. This similarity,
described as the cross-sectional mixing component of continuous
blending, was veried by using the RTD or the mean residence time
(MRT) to link the continuous blending performance with the
corresponding batch system under similar operation and geometry
[1,64]:
2

c z 0 b t Et; zdt

15

where c2(z) indicates the variance decay in the continuous blender as

a function of location z; E(t, z) is the RTD measured at location z inside

420

Y. Gao et al. / Powder Technology 228 (2012) 416423

the continuous blender; b2(t) is the variance decay in an equivalent

batch blender as a function of time. Due to the exponential decay relationship of variance in most batch blenders:


2
2
2
2
b t ss 0 ss expkb t

16

Eq. (15) can be simplied as the following formula:



2
2
2
2
c z ss 0 ss expkc z

17

2
are the initial and steady state variance of mixture;
where 02 and ss
kc the variance decay rate in the continuous blender, and kb the decay
rate in the equivalent batch blender. The relationship between kc and
kb is

kc kb =vz

18

component ratios. Ziegler and Aguilar [11] studied the continuous

processing of chocolate in a twin-screw extruder, and the mean residence time was found inversely proportional to screw speed while
directly proportional to feed rate. The single screw study on rice
our by Yeh and Jaw [15] and the studies by Bi et al. [53] and
Kumar et al. [34] led to the same results. In the single screw dryer
studied by Waje et al. [35], which conserved the structure of screw
but was not an extruder, similar effects of feed rate and screw
speed were observed.
Polymer foam process was investigated based on the separate
consideration of the batch-like chemical blowing agent foam process and the RTD in the extrusion process [2]. A gear pump associated with the single screw extruder was used to adjust the RTD,
and thus the density of the foam in the polymer form process. In
another study of polymer form process, the injection of supercritical carbon dioxide was used as physical forming agent [14]. Results
showed that high screw speed or high temperature implies short
residence time while the form performance was not discussed in
that study.
One interesting RTD study using a nite element method in a reciprocating single-screw pin-barrel extruder was reported by Bi and
Jiang [47], where the RTD of the overall extruder was assembled
from the RTD simulations of single screw elements by the convolution technique. This work provides exibility based on the existing
geometry design method of continuous extrusion process.

Eq. (18) indicates the fundamental principle of continuous blending: fast batch-like mixing rate and slow axial motion (or long residence time) lead to good mixing performance.
Due to the importance of the RTD in the characterization of continuous solid blending, recent work applied it on both mixing components described above. Sherritt et al. [26] illustrated a method for
axial dispersion prediction in rotary drum reactors for both batch
and continuous mixers. Marikh et al. [39] examined the correlation
between operations and hold-up, MRT, and axial dispersion in a
convective mixer. The cross-sectional mixing efciency was studied
using the uctuation of the RTD measurement, and a correlation
between mixing performance and the tting error of the RTD
measurement was established [57]. In a previous paper from our
group [54], the effects of feed rate, blade angle, and rotary speed
was investigated in a convective continuous blender. The RTD was
used to correlate with the mixing performance under different conditions. One interesting point in this work is the introduction of the
number of blade passes, the product of the rotary speed and the
MRT. This term describes the total number of revolutions that
powder encounters in the continuous system [65]. Larger blade
passes usually indicates better mixing performance at constant
hold-up. Portillo et al. [22] also studied particle motion in a continuous mixer using PEPT. Based on the information of tracer trajectory, the MRT and axial dispersion coefcient was calculated, and a
linear correlation was found between trajectory length and the residence time of a single particle. The RTD of cohesive particles was
examined in one periodic slice of a convective mixer in a discrete
element method (DEM) environment [50]. The horizontal stirred
bed reactor for polymer production was investigated [3]. While
the polypropylene production was provided by blending between
polymer and catalyst, polymer was fed continuously along the reactor. This is equivalent to a batch blending process in which the
polymer was added continuously.

By solving Eq. (19), the volumetric hold up in the drum can be integrated from the bed depth prole

4.2. Extruder

V holdup 0 Rz z sinz coszdz

21



hz
z arccos 1
Rz

22

Recent extrusion studies mainly investigated the RTD in polymer

and food manufacturing processes. Much work focused on the operation of co-rotating twin-screw extruders or single screw extruders.
Polymer ow behavior was studied by Fang et al. [33] in a corotating twin-screw extruder, where polymers with higher viscosity
or under high pressure showed longer residence time. Baron et al.
[20] developed a model for RTD prediction in fully intermeshing extruders based on the extension of an axial dispersion model, including operations (screw speed and ow rate) and geometries (screw
prole and die design). In a twin-screw polymer mixing process,
Zhang et al. [5] applied the conception of the RTD into a simultaneous detection of polymer morphology of mixture with different

4.3. Rotary drum

Rotary drum was widely used in blending, calcination, drying,
granulation, and many other solid unit operations. Current RTD studies on rotary drums have roots in the work by Saeman [66], where
the predictive model of mean residence time, bed depth distribution,
and other characteristics related to the axial motion of a solid was
given for a rotary drum without ights. According to this model,
the bed depth h at the axial position z can be described by the following differential equation:
i
dh 3 tan h 2
dR tan
2 3=2

F R hR

dz
2
dz
cos

19

where is the repose angle of the solid material, F the volumetric

ow rate; is the rotary speed (for example, rad/s), the incline
angle, and R the inner radius of the drum. The bed depth at the outlet
of the drum can be used as boundary condition of the differential
equation, which equals to the dam height or the average diameter
of the particles:
hz 0 hdam or hz 0 dparticle

20

Thus the MRT can be calculated as

V holdup =F

23

Notice that Saeman's model cannot provide prediction for the

axial dispersion inside the drum. McTait et al. [16] developed a
model to describe the particle motion, correlating the crosssectional rolling of particles with their axial motion. A theoretical estimation of the axial dispersion from the number of particle rotation

Y. Gao et al. / Powder Technology 228 (2012) 416423

was derived and validated. By using sand, a recent study by Liu and
Specht [67] veried Saeman's model under various feed rates and rotation speeds, and later Liu et al. [68] developed an analytical solution
for Saeman's model for cases with ll level lower than 25%. Experimental results indicated that the MRT is inversely proportional to
the rotation speed, and slightly increases with the feed rate. These results were the same as the work by Sudah et al. [23], where experiments were conducted to determine the effects of operations on the
overall hold-up and the RTD proles of cylindrical zeolite catalysts.
While the experimental values of hold-up and the MRT were close
to the predictions from the Saeman's model, the axial dispersion results were proportional to rotary speed and incline angle and was
limited when feed rate and ll level were large enough. A recent
study by Chen et al. [69] drew similar conclusions, which emphasized
more on the MRT and the mean ow rate (MFR) under various conditions. The MFR symbolized the productivity and the economic efciency of a rotary drum.
Scott et al. [70] carried out experiments in an inclined rotary drum
tted with various dams, focusing on the effects of the dam on the
bed depth distribution and particle ow pattern. Saeman's model
was still applicable in this case, which provided good predictions for
bed segments with different boundary conditions. The results are
useful for industrial kilns where conical brickwork dams at different
locations along the kiln are common. For rotary drums with ight,
the Saeman's model was not applicable anymore. The ow of particles
in this kind of geometry is similar to the convective mixer discussed
above. Cronin et al. [38] introduced stochastic modeling to account
for a larger axial dispersion due to the random effects of ight
sweep. Particle motions between ights and along the drum were
simulated as binomial or trinomial random walk, and were validated
by comparison with experiments and Monte Carlo simulations. Since
the ight geometry makes the axial and cross-sectional material
transport complicate, no predictive model has yet been developed.

4.4. Fluidized bed

Many recent publications on uidized beds focused on circulating
uidized beds (CFB), which is equivalent to a batch reactor whereas
the riser part can be considered as a continuous ow system. Chan
et al. [48] focused on the velocity prole and the RTD of solid phase
in the riser using PEPT technique. The existence of four operating
solid hold-up zones was validated in this work at prevailing conditions of supercial air velocity and solid circulating ux. Mahmoudi
et al. [13] investigated the back-mixing and the RTD of a gas phase
in the riser of a CFB. Results in this study showed that dilute riser
ow and dense riser upow led to plug ow, whereas the coreannulus operation enhanced back-mixing. Stochastic models were
developed for both four zones and core-annulus cases through the
use of a Markov chain [42]. Rodrguez-Rojo et al. [12] introduced
the application of uidized bed on particle coating with supercritical
carbon dioxide, where a well-mixed solid and supercritical uid is
the key point of good performance. The inuences of gas and solid
properties on the RTD were also analyzed. Andreux et al. [21] similarly studied the solid motion inside the riser of a CFB, focusing on the
pressure drop due to the solid ux, which divided the vertical riser
into the acceleration zone and the established zone. A decrease of
solid axial dispersion was observed with the increase of solid ux
rate, similar to the results in rotary drums. Other than the circulating
solid ux and the supercial gas velocity, the inuence of the riser
exit geometry was also investigated [31,71]. The geometry showed
a modest but consistent inuence upon the particle RTD. When reaction was involved, a continuous combustor similar to CFB was studied
and compared with similar batch tests [3]. From analysis of the gas
leaving the air reactor, the RTD of fuel particles and the circulating
solid ux in the air riser were determined in this reactor. The

421

conversion rate and the limitation of this reaction were analyzed coupling with material transport.
5. Conclusion
The RTD method has been widely applied in industrial continuous
ow systems. It offers a convenient tool for understanding material
transport phenomena inside various unit operations, which is the
rst step for efcient operation design, troubleshooting, and system
improvement.
Focusing on solid systems, this paper summarizes the industrial
developments using the residence time theory since the last theoretical review of Nauman [7]. There remain many interesting aspects
that benet practical RTD applications and can be explored.
Efforts are required focusing on the connection between the RTD
prole and the performance of different continuous unit operations.
If the RTD of a continuous reaction system can be coupled with the reaction performance of an equivalent batch system, useful and efcient guidance can be provided on the design and the performance
improvement of the corresponding continuous system, which directly
benets industrial use.
Most current literatures focused on the RTD proles of various
continuous systems, or modeling selection based on RTD tting performance. Since no new modeling was actually developed in recent
years, the value of these studies mainly relies on the RTD characteristics explored for the specic systems.
Efforts are recommended that not only apply existing tting
modeling of the RTD, but also provide novel predictive modeling.
These efforts signicantly benet in the application of residence
time theory because time and material cost on the RTD experiments
can be saved. It is not very difcult to predict the MRT when a reasonable estimation of the material ll level or hold-up is achievable.
However, the critical step in these efforts is the prediction of axial dispersion, due to the complicate inuence of operations and material
properties in different unit operations. An alternative of the predictive modeling is the application of CFD or DEM simulations, where
validation and verication of the simulations with experimental results is always required. In this case, a valid correlation between practical parameters (operation, material properties, geometry etc.) and
simulation parameters should be the focus in order to achieve reliable
and applicable conclusions from the simulations.
The development of inline tracer detection method, especially for
solid systems, is also a promising direction that requires more work.
Acknowledgement
This work is supported by the National Science Foundation Engineering Research Center on Structured Organic Particulate Systems,
through grant NSF-ECC 0540855, and by grant NSF-0504497. The authors would also like to thank Douglas Hausner for editing this
manuscript.
References
[1] Y. Gao, M. Ierapetritou, F. Muzzio, Periodic section modeling of convective continuous powder mixing processes, AIChE Journal 58 (1) (2012) 6978.
[2] M. Larochette, D. Graebling, D. Nasri, Fdr Lonardi, Optimization of the polymer
foam process by the residence time distribution approach, Industrial and Engineering Chemistry Research 48 (10) (2009) 48844891 2009/05/20.
[3] P. Markstrm, N. Berguerand, A. Lyngfelt, The application of a multistage-bed
model for residence-time analysis in chemical-looping combustion of solid fuel,
Chemical Engineering Science 65 (18) (2010) 50555066.
[4] C.J. Dittrich, S.M.P. Mutsers, On the residence time distribution in reactors with
non-uniform velocity proles: the horizontal stirred bed reactor for polypropylene production, Chemical Engineering Science 62 (21) (2007) 57775793.
[5] C.-L. Zhang, L.-F. Feng, S. Hoppe, G.-H. Hu, Residence time distribution: an old
concept in chemical engineering and a new application in polymer processing,
AICHE Journal 55 (1) (2009) 279283.
[6] S.A. Martnez-Delgadillo, H. Mollinedo-Ponce, V. Mendoza-Escamilla, C. Barrera-Daz,
Residence time distribution and back-mixing in a tubular electrochemical reactor

422

[7]
[8]
[9]

[10]

[11]

[12]

[13]

[14]

[15]
[16]

[17]

[18]
[19]

[20]

[21]

[22]

[23]

[24]
[25]
[26]

[27]

[28]
[29]

[30]

[31]
[32]
[33]

[34]

[35]

Y. Gao et al. / Powder Technology 228 (2012) 416423

operated with different inlet ow velocities, to remove Cr(VI) from wastewater, Chemical Engineering Journal 165 (3) (2010) 776783.
E.B. Nauman, Residence Time Theory, Industrial and Engineering Chemistry Research 47 (10) (2008) 37523766.
P.V. Danckwerts, Continuous ow systems: distribution of residence times,
Chemical Engineering Science 2 (1) (1953) 113.
A.H. Essadki, B. Gourich, C. Vial, H. Delmas, Residence time distribution measurements in an external-loop airlift reactor: study of the hydrodynamics of the liquid
circulation induced by the hydrogen bubbles, Chemical Engineering Science 66
(14) (2011) 31253132.
V. Shilapuram, D. Jaya Krishna, N. Ozalp, Residence time distribution and ow eld
study of aero-shielded solar cyclone reactor for emission-free generation of hydrogen, International Journal of Hydrogen Energy 36 (21) (2011) 1348813500.
G.R. Ziegler, C.A. Aguilar, Residence time distribution in a co-rotating, twin-screw
continuous mixer by the step change method, Journal of Food Engineering 59
(23) (2003) 161167.
S. Rodrguez-Rojo, N. Lpez-Valdezate, M.J. Cocero, Residence time distribution
studies of high pressure uidized bed of microparticles, Journal of Supercritical
Fluids 44 (3) (2008) 433440.
S. Mahmoudi, J.P.K. Seville, J. Baeyens, The residence time distribution and mixing
of the gas phase in the riser of a circulating uidized bed, Powder Technology 203
(2) (2010) 322330.
C. Nikitine, E. Rodier, M. Sauceau, J. Fages, Residence time distribution of a pharmaceutical grade polymer melt in a single screw extrusion process, Chemical Engineering Research and Design 87 (6) (2009) 809816.
A.-I. Yeh, Y.-M. Jaw, Modeling residence time distributions for single screw extrusion process, Journal of Food Engineering 35 (2) (1998) 211232.
G.E. McTait, D.M. Scott, J.F. Davidson, Residence time distribution of particles in
rotary kilns, in: L.-S. Fan, T.M. Knowlton (Eds.), Fluidization, 1998, pp. 397404,
1998 May 17-22; Durango, Colorado.
C. Amador, D. Wenn, J. Shaw, A. Gavriilidis, P. Angeli, Design of a mesh microreactor for even ow distribution and narrow residence time distribution, Chemical Engineering Journal 135 (Supplement 1(0)) (2008) S259S269.
T.N. Zwietering, The degree of mixing in continuous ow systems, Chemical Engineering Science 11 (1) (1959) 115.
T. Stief, U. Schygulla, H. Geider, O.-U. Langer, E. Anurjew, J. Brandner, Development of a fast sensor for the measurement of the residence time distribution of
gas ow through microstructured reactors, Chemical Engineering Journal 135
(Supplement 1(0)) (2008) S191S198.
R. Baron, P. Vauchel, R. Kaas, A. Arhaliass, J. Legrand, Dynamical modelling of a reactive extrusion process: focus on residence time distribution in a fully intermeshing co-rotating twin-screw extruder and application to an alginate
extraction process, Chemical Engineering Science 65 (10) (2010) 33133321.
R. Andreux, G. Petit, M. Hemati, O. Simonin, Hydrodynamic and solid residence
time distribution in a circulating uidized bed: experimental and 3D computational study, Chemical Engineering and Processing: Process Intensication 47
(3) (2008) 463473.
P.M. Portillo, A.U. Vanarase, A. Ingram, J.K. Seville, M.G. Ierapetritou, F.J. Muzzio, Investigation of the effect of impeller rotation rate, powder ow rate, and cohesion
on powder ow behavior in a continuous blender using PEPT, Chemical Engineering
Science 65 (21) (2010) 56585668.
O.S. Sudah, A.W. Chester, J.A. Kowalski, J.W. Beeckman, F.J. Muzzio, Quantitative
characterization of mixing processes in rotary calciners, Powder Technology
126 (2) (2002) 166173.
O. Levenspiel, W.K. Smith, Notes on the diffusion-type model for the longitudinal
mixing of uids in ow, Chemical Engineering Science 6 (1957) 227233.
E.T. van der Lann, Notes on the diffusion-type model for the longitudinal mixing
of uids in ow, Chemical Engineering Science 6 (1957) 187191.
R.G. Sherritt, J. Chaouki, A.K. Mehrotra, L.A. Behie, Axial dispersion in the threedimensional mixing of particles in a rotating drum reactor, Chemical Engineering
Science 58 (2) (2003) 401415.
S. Vashisth, K.D.P. Nigam, Liquid-Phase Residence Time Distribution for TwoPhase Flow in Coiled Flow Inverter, Industrial and Engineering Chemistry Research 47 (10) (2007) 36303638 2008/05/01.
G.I. Taylor, Dispersion of soluble matter in solvent owing slowly through a tube,
Proceedings of the Royal Society of London Series A 219 (1953) 186203.
Y. Gao, F. Muzzio, M. Ierapetritou, Characterization of feeder effects on continuous solid mixing using fourier series analysis, AICHE Journal 57 (5) (2011)
11441153.
F. Trachsel, A. Gnther, S. Khan, K.F. Jensen, Measurement of residence time distribution in microuidic systems, Chemical Engineering Science 60 (21) (2005)
57295737.
A.T. Harris, J.F. Davidson, R.B. Thorpe, Particle residence time distributions in circulating uidised beds, Chemical Engineering Science 58 (11) (2003) 21812202.
A. Vikhansky, Effect of diffusion on residence time distribution in chaotic channel
ow, Chemical Engineering Science 63 (7) (2008) 18661870.
H. Fang, F. Mighri, A. Ajji, P. Cassagnau, S. Elkoun, Flow behavior in a corotating
twin-screw extruder of pure polymers and blends: characterization by uorescence monitoring technique, Journal of Applied Polymer Science 120 (4) (2011)
23042312.
A. Kumar, G.M. Ganjyal, D.D. Jones, M.A. Hanna, Digital image processing for measurement of residence time distribution in a laboratory extruder, Journal of Food
Engineering 75 (2) (2006) 237244.
S.S. Waje, A.K. Patel, B.N. Thorat, A.S. Mujumdar, Study of Residence Time Distribution in a Pilot-Scale Screw Conveyor Dryer, Drying Technology 01 (25(1))
(2007) 249259.

[36] A. Sarkar, C. Wassgren, Simulation of a continuous granular mixer: effect of operating conditions on ow and mixing, Chemical Engineering Science 64 (11)
(2009) 26722682.
[37] B.F.C. Laurent, J. Bridgwater, Inuence of agitator design on powder ow, Chemical Engineering Science 57 (18) (2002) 37813793.
[38] K. Cronin, M. Catak, J. Bour, A. Collins, J. Smee, Stochastic modelling of particle
motion along a rotary drum, Powder Technology 213 (13) (2011) 7991.
[39] K. Marikh, H. Berthiaux, V. Mizonov, E. Barantseva, D. Ponomarev, Flow analysis and
Markov chain modelling to quantify the agitation effect in a continuous powder
mixer, Chemical Engineering Research and Design 84 (11) (2006) 10591074.
[40] V. Mizonov, H. Berthiaux, C. Gatumel, E. Barantseva, Y. Khokhlova, Inuence of
crosswise non-homogeneity of particulate ow on residence time distribution
in a continuous mixer, Powder Technology 190 (12) (2009) 69.
[41] Q. Guo, Q. Liang, J. Ni, S. Xu, G. Yu, Z. Yu, Markov chain model of residence time
distribution in a new type entrained-ow gasier, Chemical Engineering and Processing: Process Intensication 47 (12) (2008) 20612065.
[42] A.T. Harris, R.B. Thorpe, J.F. Davidson, Stochastic modelling of the particle residence time distribution in circulating uidised bed risers, Chemical Engineering
Science 57 (22-23) (2002) 47794796 2002/12//.
[43] H. Gao, Y. Fu, Mean residence time of Markov processes for particle transport in
udized bed reactors, Journal of Mathematical Chemistry 49 (2) (2011)
444456.
[44] J. Song, G. Sun, Z. Chao, Y. Wei, M. Shi, Gas ow behavior and residence time distribution in a FCC disengager vessel with different coupling congurations between two-stage separators, Powder Technology 201 (3) (2010) 258265.
[45] C.G.C.C. Gutierrez, E.F.T.S. Dias, J.A.W. Gut, Investigation of the residence time distribution in a plate heat exchanger with series and parallel arrangements using a nonideal tracer detection technique, Applied Thermal Engineering 31 (10) (2011)
17251733.
[46] C.G.C.C. Gutierrez, E.F.T.S. Dias, J.A.W. Gut, Residence time distribution in holding
tubes using generalized convection model and numerical convolution for nonideal tracer detection, Journal of Food Engineering 98 (2) (2010) 248256.
[47] C. Bi, B. Jiang, Study of residence time distribution in a reciprocating single-screw
pin-barrel extruder, Journal of Materials Processing Technology 209 (8) (2009)
41474153.
[48] C.W. Chan, J.P.K. Seville, D.J. Parker, J. Baeyens, Particle velocities and their residence time distribution in the riser of a CFB, Powder Technology 203 (2)
(2010) 187197.
[49] Y. Le Moullec, O. Potier, C. Gentric, J. Pierre Leclerc, Flow eld and residence time
distribution simulation of a cross-ow gasliquid wastewater treatment reactor
using CFD, Chemical Engineering Science 63 (9) (2008) 24362449.
[50] A. Sarkar, C. Wassgren, Continuous blending of cohesive granular material, Chemical Engineering Science 65 (21) (2010) 56875698.
[51] A.T. Harris, J.F. Davidson, R.B. Thorpe, A novel method for measuring the residence time distribution in short time scale particulate systems, Chemical Engineering Journal 89 (13) (2002) 127142.
[52] A.U. Vanarase, M. Alcal, J.I. Jerez Rozo, F.J. Muzzio, R.J. Romaach, Real-time
monitoring of drug concentration in a continuous powder mixing process using
NIR spectroscopy, Chemical Engineering Science 65 (21) (2010) 57285733.
[53] C. Bi, B. Jiang, A. Li, Digital image processing method for measuring the residence
time distribution in a plasticating extruder, Polymer Engineering and Science 47
(7) (2007) 11081113.
[54] A.U. Vanarase, F.J. Muzzio, Effect of operating conditions and design parameters
in a continuous powder mixer, Powder Technology 208 (1) (2011) 2636.
[55] A. Cantu-Perez, S. Bi, S. Barrass, M. Wood, A. Gavriilidis, Residence time distribution studies in microstructured plate reactors, Applied Thermal Engineering 31
(5) (2011) 634639.
[56] J. Diep, D. Kiel, J. St-Pierre, A. Wong, Development of a residence time distribution
method for proton exchange membrane fuel cell evaluation, Chemical Engineering Science 62 (3) (2007) 846857.
[57] Y. Gao, A. Vanarase, F. Muzzio, M. Ierapetritou, Characterizing continuous powder
mixing using residence time distribution, Chemical Engineering Science 66 (3)
(2011) 417425.
[58] A.A. Kulkarni, V.S. Kalyani, Two-Phase Flow in Minichannels: Hydrodynamics,
Pressure Drop, and Residence Time Distribution , Industrial and Engineering
Chemistry Research 48 (17) (2009) 81938204 2009/09/02.
[59] F Larachi, B.P.A. Grandjean, J. Chaouki, Mixing and circulation of solids in
spouted beds: particle tracking and Monte Carlo emulation of the gross ow pattern, Chemical Engineering Science 58 (8) (2003) 14971507.
[60] D. Liu, X. Chen, Quantifying lateral solids mixing in a uidized bed by modeling
the thermal tracing method, AICHE Journal 58 (3) (2012) 745755.
[61] L. Glicksman, E. Carr, P. Noymer, Particle injection and mixing experiments in a
one-quarter scale model bubbling uidized bed, Powder Technology 180 (3)
(2008) 284288.
[62] R. Weinekter, L. Reh, Continuous mixing of ne particles, Particle and Particle
Systems Characterization 12 (1) (1995) 4653.
[63] L. Pernenkil, C.L. Cooney, A review on the continuous blending of powders, Chemical Engineering Science 61 (2) (2006) 720742.
[64] Y. Gao, M. Ierapetritou, F. Muzzio, Investigation on the effect of blade patterns on
continuous solid mixing performance, The Canadian Journal of Chemical Engineering 89 (5) (2011) 969984.
[65] P.M. Portillo, M.G. Ierapetritou, F.J. Muzzio, Effects of rotation rate, mixing angle,
and cohesion in two continuous powder mixersa statistical approach, Powder
Technology 194 (3) (2009) 217227.
[66] W.C. Saeman, Passage of solids through rotary kilns: factors affecting time of passage, Chemical Engineering Progress 47 (1951) 508514.

Y. Gao et al. / Powder Technology 228 (2012) 416423

[67] X.Y. Liu, E. Specht, Mean residence time and hold-up of solids in rotary kilns,
Chemical Engineering Science 61 (15) (2006) 51765181.
[68] X.Y. Liu, J. Zhang, E. Specht, Y.C. Shi, F. Herz, Analytical solution for the axial solid
transport in rotary kilns, Chemical Engineering Science 64 (2) (2009) 428431.
[69] W.Z. Chen, C.H. Wang, T. Liu, C.Y. Zuo, Y.H. Tian, T.T. Gao, Residence time and mass
ow rate of particles in carbon rotary kilns, Chemical Engineering and Processing:
Process Intensication 48 (4) (2009) 955960.

423

[70] D.M. Scott, J.F. Davidson, S.Y. Lim, R.J. Spurling, Flow of granular material through
an inclined, rotating cylinder tted with a dam, Powder Technology 182 (3)
(2008) 466473.
[71] A.T. Harris, J.F. Davidson, R.B. Thorpe, The inuence of the riser exit on the particle
residence time distribution in a circulating uidised bed riser, Chemical Engineering Science 58 (16) (2003) 36693680.