Residence Time Distribution for Tubular Reactors

L. R. de Souza Jr.1, L. Lorenz1

Universidade Federal do Paran, Curitiba, Paran, Brazil

*Corresponding author: luiz_rsj@ymail.com; leonardolorenz@gmail.com

Abstract: In the core of Chemical Engineering is

the reactor design that includes most of all
scientific disciplines. The reactor, in general, are
treated ideally: mixed and plug-flow patterns.
Unfortunately, it is observed in the real world a
very different behavior from that expected. Thus,
to characterize nonideal reactors is used, among
others, residence time distribution function E(t),
mean residence time tm and cumulative
distribution function F(t). The aim of this present
work is to determine in the Comsol Multiphysics
a distribution of residence time of a tubular
reactor that is used, didactically, in the Chemical
Engineering Laboratory in Federal University of
Parana. The results were compared with
experimental dates that concluded the modeling
of the reactor showed good agreement according
to the correlation coefficient of 0.97. At last, it
was purposed a modification in the reactors
geometry to take note with its hydrodynamic
would get better.

It is nature to conclude how important is to

consider no-idealities in the reactor design
process.
Overall three interrelated factors make up the
contacting or flow pattern to account deviations
from idealities mentioned [2]:
1. The RTD or residence time distribution of
material which is flowing through the vessel.
2. The state of aggregation of the flowing
material, its tendency to clump and for a group
of molecules to move about together.
3. The earliness and lateness of mixing of
material in the vessel.
Light was placed just in the first statement.
Thus, the residence-time distribution (RTD) is a
characteristic of the mixing that occurs inside the
chemical reactor [1].
Deviation from ideal flow patterns can be
caused by channeling of fluid, by recycling of
fluid, or by creation of stagnant regions in the
vessel [2].

Keywords: Residence Time Distribution (RTD),

Simulation, COMSOL Multiphysics, Nonideal
Reactors.

1. Introduction
The Chemical Reaction Engineering (CRE)
that includes in its scope the design of reactors
uses information, knowledge and experience
from areas such as thermodynamics, mass
transfer, chemical kinetic, fluid mechanics, heat
transfer and economic analysis. In general, the
modeling processes involved in CRE establishes
idealized systems, with assumptions of perfect
mixing in the CSTR (Continuous Stirred Tank
Reactor), slug flow in PFR (Plug Flow Reactor)
and uniform composition in the case of Batch
Reactors. However, in spite of the simplified
mathematical treatment, many of cited
assumptions lead to real reactor behavior itself
far from ideal, mainly with the capacity and
products distribution with significant deviations
that may be caused by preferential flow path
formation, recirculation and dead zone [1, 2].

Figure 1 - Nonideal flow patterns [2].

Excerpt from the Proceedings of the 2014 COMSOL Conference in Curitiba

In all types of process equipments, such as

heat exchangers, packed columns, and reactors,
the type of flow showed, Figure 1, should be
avoided since they can lower the performance of
the unit [2].
1.1 Measurement of the RTD
The RTD is determined experimentally by
injecting an inert specie, called tracer, into the
reactor at some time and then measuring the
tracer concentration, C, in the effluent stream
over time [1,2,3]. The tracer must be a
nonreactive specie, easily detectable and should
have physical properties close to the reacting
mixture, in other words, it should be soluble in
the system. In addition, among others, the
tracers behavior must reflect the material
flowing through the reactor.
There are, mainly, two used methods of
tracers injection called pulse input and
step input [1].
In a pulse input, tracer is, suddenly, injected
in one shot into the feed stream, entering the
reactor in as short a time as possible. Since the
step input consider a constant rate of tracer
addition from an initial time, t = 0 t.u., before it,
there is no tracer been added to the feed. The
outlet concentration is measured over time, in
both methods showed [1,2,3].
In order to become more understandable
what was brought, there is a scheme below,
Figure 2.

Focusing on the test method step (injection

of tracer), illustrated in Figure 2, it can be seen
that at beginning the concentration of tracer is
low, however increases with time due to the flux
in the reactor.
At first, the concentration of tracer is low,
however increases with time due to the flux in
the reactor. The fact that the response step is not
equally the same is due to occurrence of
dispersions in the system, which affects the mean
residence time in the reactor and consequently,
alters its conversion, yield and capability.
If Co is the concentration of tracer added to
the reactor inlet, the F fraction of the tracer at the
outlet of the reactor will be:
F(t) =

C(t)
Co

(1)

The tracer concentration in the reactor outlet

is given by:
t

C(t) = C0 0 E(t)dt

(2)

Substituting (2) into (1), it follows that:

t

F(t) = 0 E(t)dt
() =

()

(3)
(4)

Thus, the curve of distribution of residence time

is obtained, Figure 3.

Figure 3 - RTD curve behavior [2]

After obtaining the curves of concentration

(curve C), the fraction of material in the reactor
outlet (curve F), and the residence time
distribution (curve E), all of them can be
analyzed qualitatively and the behavior of the
flow inside the reactor can be observed.
Figure 2 - RTD measurements [1]

Excerpt from the Proceedings of the 2014 COMSOL Conference in Curitiba

Figure 4 - F-diagrams: (a) piston flow; (b) piston flow with some longitudinal mixing; (c) complete mixing;
(d) dead water [4]

For example, the curve F can assume four

different aspects according to the flow tracer, as
Figure 4 shows.
Another important parameter to determine is
the residence time, . Conceptually, it is the time
that certain amount of molecules have remained
within a unit volume. For a flow rate Q into a
fixed volume reactor V, it is known that the
mean residence time ideal is given by:
=

V
Q

(5)

Strictly, the mean residence time for a real flow

is given by the following expression:

t = tE(t)dt

(6)

The closer are the results of equations (5) and

(6), closer the system will be of the ideality.

2. RTD experimental data acquisition

To make the acquisition of experimental data
and compare with the data from COMSOL
Multiphysics simulation, it was used a similar
system to the Figure 3. Thus, the reactor,
completely filled with NaCl (Brine), started to
receive water, step test. Over time, the solutions
concentration
was
measured,
with
a
conductivimeter, in the reactor outlet. The results
can be viewed in item 4.

3. Use of COMSOL Multiphysics

In order to obtain the behavior of the flow in
the reactor inlet due to the difference in
diameters of its inlet nozzle and its body, it was
selected a 3D modeling. The used physics were
laminar flow since the flow used in the
experiment was low - and transport of diluted

species in view of brines concentration used as

tracer.
The models geometry was constructed to
represent the real equipment. Furthermore, cut
up the geometry in half with a plane of
symmetry, in order to better visualize the flow
lines, the dispersion of the tracer as well as
reduce the computational effort.
In relation to the properties of the material
used, it approached the density and dynamic
viscosity of the fluid flowable - brine - for pure
water at the temperature at which the experiment
was conducted by interpolation physic-chemical
data Tables [5].
Diffusivities, at 25 C, o the chemical
species involved (Na+ and Cl-) were taken from
the literature [6] and corrected by the experiment
temperature, according to the following
expression [7]:
DT
25 C
=
D25 C
T

(7)

3.1. Reactor modeling

3.1.1 Laminar Flow
The equation that models the flow of fluid
throughout the reactor is the Navier-Stokes
equations.
(u. )u = . [pI + (u + (u)T ] + F (8)
Where, u is the velocity vector, is the density
of the fluid, is the viscosity, p is the pressure
and F is a body force term, such as gravity.
As the flow is incompressible,
. u = 0

Excerpt from the Proceedings of the 2014 COMSOL Conference in Curitiba

(9)

Figure 5 (a) Original Reactor; (b) Modified Reactor

Figure 6 - Tracer injection

The boundary conditions used to solve the

equation were the input, output and speed in the
reactor wall. Thus, it was used in the input
condition the flow rate that is equal to the output,
since the flow is stationary. In addition, the
output condition has been chosen as the
discharge pressure of the reactor, in this case, the
atmospheric pressure. Finally, it was selected the
condition of no slip on the walls of the reactor
(zero speed).
The mesh size element was calibrated for
fluid dynamics, and left to "normal" size - the
highest possible for that particular calibration.
The direct method was employed for the
solver as suggested for problems involving the
resolution of the Navier-Stokes equation [8].

Where, ci is the tracers concentration, Di is

diffusivity and Ni is the diffusion flux.
Thus, to simulate the mass transport of the
tracer inside the reactor was selected as boundary
conditions the input, output and initial
concentration values. The velocity field used was
imported from the laminar flow study, described
above. The concentration of the input was
chosen to be zero (since water from the city
supply was injected into the reactor, free of
brine), while the initial value of concentration
inside the machine was 445 mol/m (first
measured in the reactor outlet).
The concentration provided by the model is a
measured defined in the outlet contour.

3.2 Proposed modification of the Reactor

3.1.2 Transport of Diluted Species
The phenomena of tracers diffusion and
convection, Figure 6, are modeled by the
continuity equation together with the equation of
the overall flow:

In order to reduce the non-idealities of flow,

evaluated a modification of the inlet nozzle of
the reactor geometry, as shown in Figure 5b, to
monitor the expansion of the flow lines.

4. Results and Discussion

ci
+ . (Di ci ) + u. ci = R i
t

(8)

Ni = Di ci + uci

(9)

To validate the model used in the simulation

was compared concentration data obtained
experimentally with those provided by the

Excerpt from the Proceedings of the 2014 COMSOL Conference in Curitiba

software, the correlation coefficient equal to

0.97.

So, there is an increase in the residence time that

can be verified through the curves E, Figure 9.
To compare with the value of the mean
residence time, space-time was calculated, ,
ideal flow, which is 6.10 min.

Figure 7 - Reactor outlet concentration

From equations (1) and (4) it is possible to

obtain the residence time distribution, Figure 8.
Furthermore, it was calculated the residence
time average for the model and for the
experiment.
The values obtained was 3.05 and 3.38 min,
respectively. A deviation of 9.8%.
With the validated model, was verified with
the simulation, the hydrodynamic behavior of the
flow - preferred paths, areas of recirculation and
stagnant zones - allowing for the non-ideality of
the reactor. In Figure 5, the flux lines of the
original and modified reactors are represented.

Figure 9 - RTD curve of original and modified reactor

Thus, it can be established that as the spacetime is twice greater than the mean residence
time, there is a strong trend that is occurring
preferential flow paths. Furthermore, by
modifying the reactor there is an increase in the
RTD, which leads to a decrease in e recirculation
zones, Figure 6, favoring a more homogeneous
flow.

5. Conclusions

Figure 8 - RTD curve

It can be seen that the implemented

modification generates less recirculating fluid,
indicating that the flow is more homogeneous;
consequently, preferred ways are not favored.

According to the results, it is observed that

the modeling of the reactor showed good
agreement with the experimental results, the
correlation coefficient of 0.97. Furthermore, with
simulation it was possible to verify the
hydrodynamic behavior of the flow - preferred
paths, areas of recirculation and stagnant zones allowing to establish the non-ideality of the
reactor. The studies showed a space-time of 6.10
min. and an average residence time of 3.05 min
(original reactor) and 3.38 min (modified
reactor). Importantly that closer are values of
space-time and residence time average, there is
an indication that the reactor will operate more
adequately. However, the hydrodynamic
problems are not obvious, that is why the
importance of computational fluid dynamics in
the analysis, design and operation of reactors.

Excerpt from the Proceedings of the 2014 COMSOL Conference in Curitiba

6. References
1. Levenspiel, O. Chemical reaction engineering.
John Wiley & Sons. 3rd ed., 1999
2. Fogler. H. S. Elements of chemical reaction
engineering. Pearson Education. 4th ed., 2006
3.Hill, Charles. An introduction to chemical
engineering kinetics and reactor design. John
Wiley & Sons, 1977.
4. Danckwerts, P. V. Continuous Flow Systems.
Distribution of Residence Times. Cambridge,
Department of Chemical Engineering, Tennis
Court Road, 1952.
5. Lide, D. R. CRC Handbook of Chemistry and
Physics.
6. BASMADJIAN, D., Mass Transfer:
Principles and Applications, pg. 97, CRC Press
LLC, Florida (2003)
7. CUSSLER, E. L., Diffusion Mass Transfer in
Fluid Systems, pg. 127, Cambridge Press,
(2009)
8. Dynamic Help do COMSOL

Excerpt from the Proceedings of the 2014 COMSOL Conference in Curitiba