HYDROCARBON PROCESSING / JULY 1999
General model--key features.
For
any given reactor, the general model
should predict the performance for
any known reaction system and cat-
alyst activity. Conversely, for a given
reaction system and catalyst activ-
ity, the model should predict the per-
formance for a variety of known reac-
tor configurations. Thus, the model
is applicable to any combination
between reaction and reactor sys-
tems. The user can use the model to
optimize the operating conditions
that best meet a given demand from
an existing reactor. Also, the user can
determine the best possible reactor
configuration by examining various
possible new design options or design
modifications of the existing unit. For
example, assuming the same catalyst
activity for a catalytic gas-solid reac-
tion system, the user can immedi-
ately determine the possible advan-
tages of switching from an existing
fixed-bed to a fluidized-bed reactor
for the next revamp or modernization
plan.
The model should be based on fun-
damental principles of reaction engi-
neering and reactor hydrodynamics.
Therefore, its application should not
be restricted to narrow operating
ranges or conditions of empirical
models. It should predict the reactor
performances, particularly the prod-
uct composition, for a wide variety of
designs and operating conditions,
including those used or are expected.
It should also cover design options
such as product recycle, multiple feed
injections, multiple reactor types in
series or parallel, and various inter-
nal flow circulation or mixing modes
within the reactor. The model should
cover conditions beyond normal oper-
ations and can predict upset, offspec
and unsafe situations. The model
should have a minimum number of
adjustable/experimental parameters
and can be solved by standard math-
ematical routines requiring mini-
mum execution time. It should be
easy to integrate into other commer-
cial or in-house simulation packages,
and into both off- and online control
algorithms. Finally, it should evolve
through extensive validation by
experiments in appropriate test
methods.
Is there such a model?
Models
are available; they apply to gas-solid
and homogeneous reaction and reac-
tor systems of virtually any kind and
Ten frequently asked questions
concerning reactor modeling
þ Q.1 Does the reactor currently have a deficiency to address?
It
is a difficult question. However, many old reactors are obvious candidates for close scrutiny
of their performance. A performance analysis with a modeling tool can determine the
improvement potentials for a given reactor. This is same as the performance diagnosis of
most other unit operations or segments of the process plants that are subject to revamps
and modernization. This diagnosis can be done with the help of commercial simulation
packages and other modeling tools developed in-house or available in the market.
þ Q.2 How much information is needed?
Any model is as good as the quality and quantity of data available. The model applicability
is also dependent on the ranges of key variables covered by the data on which the model
is built. The larger the ranges covered by the data, the better the applicability of the
model. A working model for an acceptable range of operations and design options can be
quickly built with the data and information that is available from the plant archive and open
literature. This model obviously gets refined as more data are gathered from operating the
reactor and fed back to the model.
þ Q.3 How should nonconventional designs, such as FCC, be handled?
An FCC can be viewed as a series combination of a completely mixed reactor (CSTR) near
the hydrocarbon feed point, followed by a circulating fluidized-bed reactor in the rest of
the riser section and a bubbling fluidized-bed reactor at the disengagement zone at the top.
A schematic representation of this configuration is shown in Fig. 4. Such a system can be
easily modeled by a general model package that includes all three reactor types and can
be executed as a combination in series. Most other reactors of a nonconventional design
can similarly be modeled by using the existing reactor modules within the package and
by making minor modifications or adjustments to fit the exact needs.
þ Q.4 What is the most critical step?
The most critical and often the most time-consuming step is determining the reaction mech-
anism and kinetics, especially the latter. A large number of data must be carefully screened
and analyzed for this purpose. Gaps in the data must be filled with new tests. Also, tests
must be done to replace bad data.
þ Q.5 What about data confidentiality?
Confidentiality of all information and data can be maintained by standard secrecy agree-
ments between the client, any external laboratory and third party involved in the process.
þ Q.6 How long/expensive are experimental test campaigns?
This depends on three factors: 1) quality and quantity of background information available
on the reaction system, 2) quality and quantity of lab or pilot plant data available, and 3)
complexity of the reaction system. Reliable data covering adequate ranges of the four major
variables--temperature, pressure, feed composition and flowrate--are needed for appli-
cation of the general reactor model. These test campaigns may run from a few weeks to a
few months depending on the situation. The cost can easily be justified from both short-
and long-term benefits derived from this effort.
þ Q.7 How do you assess the benefit?
The reactor is the heart of most process plants and refineries. Modeling and optimizing the
performance of almost all unit operations or segments of plants has become a routine mat-
ter. However, the reactors remained relatively untouched in most cases. We have lived with
these old reactors or old designs, particularly in refineries, for too long. Significant improve-
ment potentials of these reactors exist, both on a short- and a long-term basis. Using a good
modeling package could only assess the benefits from these improvements.
þ Q.8 What is the risk?
There is no risk in implementation modeling, since most of the effort is done on the side-
line. Total confidence is gained from a thorough model validation in the test facility or con-
trolled step tests in the operating plant. Only then can the modifications and improvements
be carried out in stages in the operating plant, and in revamp and modernization.
þ Q.9 Will the effort succeed?
A well-tested, practical and comprehensive general reactor model has a higher probabil-
ity of success and a return on investment than most trial-and-error procedures normally prac-
ticed today. Furthermore, similar to many simulation tools used today, the reactor model-
ing will also become refined and more powerful with time and its value and usefulness will
increase.
þ Q.10 What are the time and costs involved?
Time and cost involved in building such a tool are negligible compared to both short- and
long-term benefits expected from its applications. The expected improvements in yield and
productivity alone, ignoring many other benefits cited earlier, could justify these costs. For
revamp or modernization the justification is even stronger, since this tool could minimize,
if not eliminate, much more expensive pilot plant campaigns normally considered neces-
sary for such a job.
