| |
By Terry Ring, Bin Wang, and Byung Sang-Choi, University of Utah, Salt Lake City, UT, and Kumar Dhanasekharan, Fluent Inc.
View the pdf of this article
In chemical reaction engineering,
the concept of residence time distribution
(RTD) is fundamental to
reactor design. RTD is the exit age
distribution of fluid molecules leaving
a reactor. The residence time is
the total time spent by the fluid molecules
within the reactor. The RTD
for process equipment is typically
measured using stimulus-response
tracer experiments to detect design
flaws such as bypass, channeling, and
dead zones, in addition to characterizing
a reactor’s mean residence
time and standard deviation. This measurement
has been done, historically,
when a complete velocity
distribution map for the fluid in the
vessel is not available. Because CFD
is capable of predicting the complete
velocity distribution in a vessel, it provides
an alternative, indeed, simpler
means of determining the RTD.
As part of a Department of Energy
Chemical Industry of the Future project
for modeling crystallization,
Fluent is working with the University
of Utah to predict RTD using FLUENT,
with the results being compared to
experimental measurements.
Fluid flow simulation of a 1.4 liter baffled stirred tank operating with a
Rushton impeller
Comparison of experimental and calculated residence time distribution function,
E(t), for a CSTR; operating conditions are 200 rpm, 40 ml/min
On the experimental side, a 1.4
liter stirred tank with complex multiple
feed tube geometry has been
used to measure the residence time
distribution as a function of flow rate.
The vessel is driven by a Rushton
impeller and is either baffled or not
baffled for the different cases studied.
At both ideal and non-ideal conditions,
experiments and simulations
of the RTD have been performed.
For the CFD work, there are multiple
approaches for predicting residence
time distribution. In one
approach, the tracer fluid is represented
by a large number of discrete
particles and Lagrangian particle tracking
analysis is done with the discrete
phase model (DPM). A histogram of
time at the outlet is the residence time
distribution. One drawback of this
method, however, is that a large number
of particles are required to ensure
proper statistics. Alternatively, the tracer
fluid can be treated as a continuum
by solving a transport equation
for the tracer species. It is the latter
continuum approach that has been
used for the crystallization project.
A single-species flow field was first
obtained using the k-e turbulence
model. A passive tracer was then introduced
with a step change in its concentration
in the feed. The tracer was
modeled using a user-defined scalar
transport equation. The surface-area
averaged tracer concentration was monitored
as a function of time at the outlet.
The results provided the exit age
distribution of the tracer in the reactor,
and therefore, represent the RTD.
The FLUENT predictions were found
to be in excellent agreement with the
experiment over a broad range of flow
rates and impeller RPM values, for both
baffled and non-baffled tanks.
Since residence time distributions
are routinely, and rather easily
measured for process equipment, the
RTD predictions allow a rather painless
way to validate a complex flow
simulation. In addition, with the aid
of FLUENT simulations, process
scale-up can be facilitated without
expensive work on an intermediate
scale in a pilot plant. The question
of whether to construct a pilot plant
or not depends on whether engineers
are in control of all the major variables
for the process. Using FLUENT,
a small-scale reactor can be modeled
to accurately predict its residence time
distribution. Various potential largescale
reactor designs can also be modeled
in the same manner, and the
residence times predicted to verify
that upon scale-up, the large-scale
reactor controls the major variables
for the process in the same way that
the small-scale reactor does.
|
|
|
