Problem Statement

Why is Aspen Rate Based Distillation (RateSep) better than using equilibrium stages with efficiencies? There seem
to be a lot of fudge factors in rate-based distillation also?

Solution

Equilibrium stage efficiencies are a function of all column conditions (flow, temperature, pressure). This means that
they generally do not extrapolate well to other conditions. A rate-based distillation model has more basis in reality
using mass- and heat transfer correlations based on transfer properties and tray/packing geometry to predict column
performance, without the need of efficiency factors and the fudge factors have more physical meaning. Once the
column is calibrated to one set of conditions, it does generally extrapolate to other conditions. We have customers
who have had success using RateSep and being able to predict new scenarios after tuning the column to an initial set
of conditions. This added degree of rigor is especially critical for modeling gas scrubbers, sour water strippers,
azeotropic systems, reactive distillations, nitric acid absorption columns, narrow-boiling separations, and other
highly non-ideal separation processes.

The rate-based modeling approach has many advantages over the equilibrium-stage modeling approach. The rate-
based models represent a higher fidelity, more realistic modeling approach and the simulation results are more accurate
than those attainable from the equilibrium-stage models. The rate-based modeling approach can reduce the risk of
inadequate designs or off-spec operation because the rate-based models explicitly account for the actual column
configuration which affects column performance.

Designed to model reactive multistage separation problems rigorously and accurately, Aspen RateSep balances gas
and liquid phase separately and considers mass and heat transfer resistances according to the film theory by explicit
calculation of interfacial fluxes and film discretization. The film model equations are combined with relevant diffusion
and reaction kinetics and include the specific features of electrolyte solution chemistry, electrolyte thermodynamics,
and electroneutrality where appropriate. The hydrodynamics of the column is accounted for via correlations for
interfacial area, hold-up, pressure drop, and mass transfer coefficients.

Fast reactions for processes such as CO2 removal with MEA do need to use the discretization film model available
in the current rate-based distillation model not available previously in the older RateFrac.

RateSep uses a rate-based modeling approach, which is much more rigorous than the equilibrium stage modeling
approach employed by the original RadFrac module. The rate-based model assumes that the separation is caused by
mass transfer between the contacting phases, and uses the Maxwell-Stefan theory to calculate the mass transfer rates[1].
The equilibrium modeling approach assumes that the contacting phases are in equilibrium with each other, which is
inherently flawed because the contacting phases are never in equilibrium in a real column.

The rate-based (or nonequilibrium) modeling approach has a lot of advantages over the equilibrium modeling
approach, and is becoming more and more popular in both the academia and the industry. The rate-based model is
more reliable and accurate than the equilibrium model, and therefore can reduce the risk of inadequate designs or off-
spec operation. The rate-based model gives more realistic simulation results than the equilibrium model, because the
column configuration is included in the rate-based model and will affect the simulation results as it should. Engineers
can use the rate-based model to do more realistic what-if studies and improve the operations of existing columns. The
rate-based model can also predicts the behavior of a column more accurately, and can help the process control
engineers develop better control strategies.
It is generally accepted in the literature that the rate-based model is more accurate than the equilibrium model. In some
cases the rate-based model gives accurate results while the equilibrium model gives wrong results.

Taylor et al. [2] reviewed the modeling of distillation and concluded that the rate-based simulations "are not only
feasible, but in some circumstances they should be regarded as mandatory". For a bubble cap distillation column used
to separate water, ethanol and acetone, the composition trajectory predicted by the rate-based model agrees with the
experimental data very well, while the equilibrium model gives very different results. The rate-based model also
accurately predicts that the distillation boundary is crossed, while the equilibrium model wrongly predicts that the
distillation boundary can never be crossed. Taylor et al.[2] also used experimental data from Springer et al.[3] for the
water-cyclohexane-ethanol system to demonstrate that the rate-based model can predict the composition trajectory
accurately, while the equilibrium model cannot.

Kucka et al.[4] used the rate-based model to simulate the sour gas absorption by aqueous amine solutions. They
compared the simulation results with experimental data from a pilot plant column and an industrial scale column and
found that the rate-based model can accurately predict the temperature and composition profiles.

Springer et al.[5] used the rate-based model to simulate ternary azeotropic distillation and compare the simulation
results with experimental data from a bubble cap distillation column with the water-ethanol-methylacetate system.
They found that the rate-based model ?is in excellent agreement with a series of experiments carried out in different
composition regions. In sharp contrast, an equilibrium (EQ) stage model fails even at the qualitative level to model
the experiments.? They concluded that "for reliable simulation of distillation of azeotropic systems exhibiting a
distillation boundary, we must adopt a rigorous NEQ (or rate-based) stage model."

Eckert and Vanek[6] studied three-phase distillation columns with the rate-based modeling approach. Extensive
experimental work for the ethanol-water-cyclohexane system[7-9] was used to compare the rate-based approach and
the equilibrium approach. They found that the rate-based model can predict the composition profiles much better than
the equilibrium model, especially in the middle of the column. The rate-based model also predicts the disappearance
and reappearance of the second liquid phase very well.

Kreul et al.[10] studied the modeling of reactive separation processes in packed columns and found that the
concentration profiles calculated by the rate-based and equilibrium models can be very different, and the rate-based
model should be preferred for reactive separation processes.

Kenig et al.[11] studied reactive absorption using models with different complexities. Both the steady state and dynamic
models were used. They found that for a H2S scrubber, the rate-based modeling approach can predict the concentration
profiles very well, while the equilibrium model may yield results completely inconsistent with the experimental
studies.
Sundmacher and Hoffmann[12] used a rate-based model to simulate reactive distillation of the MTBE system in a
packed column. The simulation results were validated with experimental data from a laboratory scale column with
excellent agreement.

Frey and Stichlmair[13] mentioned in their review paper on reactive distillation that rate-based model has been used to
accurately predict the performance of commercial reactive distillation processes in Germany and Texas, and that the
model has been validated for the design of TAME, ETBE and TAEE units with data from semi-commercial operations.

Pinjala et al.[14] used a rate-based model for reactive distillation of the MTBE and TAME processes in columns filled
with Koch Katamax? packing. The simulation results were compared to some pilot-plant data with good agreement.

Most recently, Kloker et al.[15] compared the rate-based model with the equilibrium model for two reactive separation
processes, the synthesis of ethyl acetate from ethanol and acetic acid via reactive distillation and the synthesis of octyle
hexanoate from octanol and hexanoic acid via reactive stripping. They found that the rate-based model with a fine
discretization is able to match the column temperature profile and the bottom concentration better than the equilibrium
stage model.

The rate-based modeling approach has been used successfully to model many industrial separation processes. It has
been adopted by many researchers to model reactive distillation of MTBE[12, 14, 16], TAME[13, 14], ETBE[13], TAEE[13],
ethyl acetate[15], propylene glycol[17], ethylene glycol[17], propyl acetate[18], diacetone alcohol[19, 20] etc. It is also used
to model other reactive separation processes such as reactive absorption[4, 11, 15, 21] and reactive stripping[15]. The rate-
based model has also found successful application in azeotropic distillation[5] and three-phase distillation[6, 22, 23].

The rate-based model is especially preferred over the equilibrium model in the following situations:

(1) Packed columns. The rate-based model does not need the HETP (Height Equivalent to a Theoretical
Plate) of the packing, while the equilibrium model does. The HETP can vary widely from component to
component and from packing to packing, and cannot be estimated accurately in most cases.

(2) Absorption and desorption. The absorption and desorption processes are normally operated in
conditions far away from phase equilibrium, and are mass transfer and heat transfer limited. The tray
efficiency for a absorption/desorption column can be as low as 5%[2]. Therefore, the rate-based model is more
appropriate than the equilibrium model.

(3) Reactive separation such as reactive distillation and reactive absorption. The rate-based model can
model reaction rates much more accurately than the equilibrium model. The efficiencies are not meaningful
for reactive separation processes.

(4) Strongly nonideal systems. The efficiencies are highly uncertain and can vary widely from tray to tray
and component to component. It is very difficult to determine the efficiencies for the equilibrium model.
Therefore, the rate-based model should be preferred.
 (5) Columns with a complicated configuration (pump-around, side streams). It is very difficult to
determine the theoretical stage locations for pump-around and side streams, which are necessary for the
equilibrium model. The rate-based model does not need any information about theoretical stages, and should
be preferred.

(6) Columns with mixed trays and packing. It is very difficult to determine the number of theoretical
stages for each tray/packing section. The rate-based model does not need any information about theoretical
stages, and should be preferred.

The major drawback of the rate-based model is that it is much more complicated than the equilibrium model.
Previously, the rate-based model was considered as impractical due to its complexity. Right now, the rate-based model
can be solved readily on any mainstream PC, thanks to the ever-increasing computing power.

There are some special advantages for RateSep:

(1) RateSep is seamlessly embedded in RadFrac. It is very easy to configuration a RateSep simulation for
someone who does not know very much about the rate-based model. The complexity of the rate-based model
is in the background and not noticeable to the user.

(2) RateSep has many built-in correlations for mass transfer and holdup. There are default correlations for
each tray/packing type. The user does not need to worry abut the correlation unless he/she wants to choose
other correlations.

(3) RateSep has detailed examples for user subroutines. The user can easily modify the examples to create
his/her own user subroutines for mass transfer, heat transfer and holdup.

(4) A continuation method has been implemented in RateSep to improve the convergence for problems
that are difficult to converge. The user only needs to choose two continuation parameters to use the
continuation method.

(5) Several tuning parameters have been defined to help the advanced users to model a column more
accurately. The user also has control on how he/she wants to calculate the fluxes, how to average the
diffusivities and mass transfer coefficients.

(6) It is very easy to switch between the equilibrium mode and the rate-based mode. For a large flowsheet,
if the block with RateSep fails to converge, the user can easily switch back to the equilibrium mode and
make the flowsheet converge. After that, he/she can switch to the rate-based mode and try to make the whole
flowsheet converge.

(7) RateSep supports film reactions and film discretization. Therefore, it is capable of solving reactive
separation problems accurately.

It is necessary to have accurate properties, kinetic and equilibrium constants if needed, and then tune the factors such
as the interfacial area factor and the heat transfer factor for the rate-based column. But, because the model has more
physical meaning, it can then be used to model new situations more accurately. A number of our customers have used
the Rate Based Distillation model successfully.