KOWALCZYK I INTRODUCTION
Department of Chemical Engineering
University of Detroit U. S. A. The designer of a chemical reactor must possess
a thorough knowledge of the subject and be equipped
with appropriate tools securing fast and reliable
results.
Basic knowledge of the reactor design problem
includes:
(1) Chemical Engineering Thermodynamics (heats
of reaction, chemical equilibria);
(2) Reaction kinetics for homogeneous and hetero-
geneous systems;
(3) Laws governing transport processes in the
presence of the chemical reaction;
(4) Fundamentals of optimization, stability and
control analysis; and
(5) Principles of process economics.
The mathematical model method seems to be the The preliminary study of the problem comprises
most popular method of today's reactor design. laboratory experimentation and retrival and eval-
It is fast, relatively inexpensive and accurate when uation of library data in the following areas:
reliable data on reaction kinetics, thermodynamics
and transport phenomena are available for the (1) Reaction kinetics;
process under consideration. The method begins (2) Thermodynamics;
with a detailed theoretical and laboratory study (3) Mass, heat and momentum transport coefficients;
and is followed by the formulation of the mathe- (4) Methods of reactant contacting for similar
matical model of the process, describing the model processes;
dynamic behavior by a set of differential equations. (5) Methods of heat removal or addition;
Analysis of these equations by advanced mathe- (6) Materials of construction; and
matical methods leads to the determination of the (7) Process economics.
optimal size of reactor operating at the lowest
cost for the required rate of production. Because The purpose of the kinetic studies is to establish
of the differential character of model equations, stoichiometry and mechanism of reaction, character
the model may be directly subjected to the optimiza- of elementary steps in complex reactions, orders
tion, stability and control analysis. The increasing and specific reaction rates of individual steps,
popularity of the application of the mathematical over-all rate expression, selectivity, effect of opera-
model to the reactor design is chiefly due to the tional parameters (temperature, pressure, catalyst,
advent of electronic digital and analog computers ratio of reactants, impurities, method of contacting)
which enable the designer to study and analyze and physical phenomena accompanying the reaction
the complex interactions of kinetic, thermodynamic on the rate of reaction and its yield.
and transport phenomena taking place in the The thermodynamic study leads to the values of
chemical reactor. C, and C v of reacting system, heat of reaction,
The equations describing the mathematical model heat of adsorption (gas-solid catalytic reactions),
are of semi-empirical nature, i.e., their form obtained energy of activation and equilibrium constant via
from the theoretical considerations of the process the free energy of reaction
or phenomena is well established and only the
constants and parameters and their changes asso- OF= RTIn K
ciated with the variation of environment of the (I)
reacting system must be determined by direct,
frequently independent experiment. In this way, for various operational conditions.
these equations have a physical and engineering The study of mass, heat and momentum transport
meaning and when applied to the mathematical coefficients (diffusivity, thermal conductivity and
model, permit for extrapolation into unexplored viscosity) in homogeneous and heterogeneous sys-
parameter regions and design the reactor, including tems is necessary when a multiple gradient (dispersed
its stability and control, at any feasible condition. parameter) reactor model is used. Reliable data
The mathematical model design involves the must be established for the whole range of opera-
following steps: tional conditions. If they cannot be retrieved from
the literature or obtained from existing pilot or
(1) Preliminary study of the problem; commercial reactors, they must be carefully evaluated
(2) Formulation of the model and design procedure; by inter and/or extrapolation of existing data or
and calculated from verified empirical formulae or
(3) Mathematical analysis of the model behavior theoretical considerations [2]. The thermodynamic
with regard to the reactor size, its optimization, and physical properties package data may also be
stability and control. available from big computer systems [3].
In the study of heat removal or addition methods, which the elementary volume properties (point
we are primarily concerned with design and operation properties) remain constant in time and space
of appropriate heat exchange equipment. Design (lumped parameter model). The distributed para-
of the heat transfer area for isothermal reactors meter model comprises the basic microscopic and
may represent quite a problem. Control of heat multiple gradient models, which are two-dimensional
removal and addition in isothermal and programmed for gradients in the axial and radial direction only.
reactors must also be studied. The simplified gradient models (multiple gradient
The study of constructional materials comprises models without radial gradients or without any
material availability, mechanical and thermal prop- dispersion) are unidimensional models. The lump
erties as well as chemical resistivity. Sometimes parameter model is equivalent to the macroscopic
search for new or less expensive materials may be model and is zero dimensional.
necessary.
Design activities must lead to the highest returns
or least expensive production. Accordingly, economic 10 DESIGN EQUATIONS
optimization of reactor size for the desired rate
of production is of primary importance. The The design equations for the chemical reactor
designer, therefore, must be well acquainted with involve:
cost of reactants, products and wastes; cost of (1) The rate expression;
reactor with auxiliary equipment and control (2) Mass balances (total or for individual reaction
instruments; and cost of reactor operation (energy, participants);
labor, cooling water, depreciation, overhead, etc.). (3) Heat balances (for the total reactor); and
(4) Momentum balances.
The rate equation for homogeneous reactions is For gas-solid fixed bed catalytic reactors
usually expressed by the law of mass action. For
reactant A in the batch system, it will be: (kinetic term)(potential term)
r=
(absorption term) (6)
dNA = n
k ( NA , V const. based on the Langmuir-Hinshelwood theory of
Vdt \ V
(2)
surface reactions.
For the gas-solid non-catalytic reactions, the rate
expression has a general form
dCA
kcCAn , V = const.
dt driving force
(3)
r=
resistance
(7)
where and depends upon the rate controlling step which
may be the gas film, diffusion through ash layer
N A moles of reactant A in the system
or chemical reaction. In certain gas-solid reactions,
V volume of the system the rate of reaction may be controlled by the rate
CA concentration of A of heat conduction through the ash layer.
t time In two immiscible fluid reactors, the chemical
k specific reaction rate reaction takes place at the fluid interface and the
n order of reaction rate depends upon the mass transfer rates in both
fluids and chemical reaction.
In complex, like free radical reactions, the rate
expression is based on the rate controlling step
and may have quite a complex form.
For the tubular flow reactor, the rate expression 12 MICROSCOPIC TRANSPORT
converts to: BALANCES
(4)
The base for all balance alterations and adjustments
for practical purposes is the microscopic balance.
It assumes that the system is a continuum and
ignores the detailed molecular interactions. The
where derivatives are based on mass, heat and momentum
changes in an elemental volume of the system.
FA molal flow rate of A The general form of microscopic transport balances
F v volumetric flow rate through the reactor is given below:
V reactor volume, and
FA Net _ Input through surfaces
CA=
Fv cumulation]
[
ac [of elemental volume
- Output through - Generation -
For the CSTFR: inside
surfaces
of elemental elemental
O CA = kcCA n
volume volume
t
(5) Consumption
inside
where t is the time of residence. elemental
The rate expression for heterogeneous reactions volume
may be more complex. (8)
A set of transport balance equations for reactant microscopic tubular flow reactor in cylindrical
A in a dilute binary system in a homogeneous coordinates is given in fig. 6.
aCAacA v e aCA vZ
aCA =
at + ( v ` at + r ao + az
Accumulation Transport through open surfaces by bulk flow
= DAB
( I a
(r aCA + 1
a2CA + a2CA
r ar \ ar ) r2 ae 2 aZ2
Transport through open surfaces by molecular
+ CA
Generation
(I)
mass diffusion
Heat balance
C13
/ aT + Vr al' + Vg a +V
Z al.
at ar r ae az )_
Accumulation Transport through open surfaces by bulk flow (convection)
1 a aT 1 a2T a2T
Cr
(2)
r ar ar ) + r2 ae2 + az 2
avZ
avZ + v e avZ + vZ avz
C at +
v`
Or r ao az
Accumulation Transport through open surfaces by viscous diffusion
_ Op
az
{ ^
a
L r 1 ar
Transport through open surfaces
( r
av 1
ar l
Z1
+ 2
r
a2vZ a2vZ
ao2 + az ]+^ g=
Generation
(3)
Notation:
Fig. 6
Microscopic transport balances in cylindrical coordinates for a tubular flow reactor
In deriving these equations the following assumptions G. F. FROMENT [4] divided fixed bed reactors into
were made: two classes: pseudo-homogeneous and heteroge-
neous. For the homogeneous class, it was assumed
(1) Newtonian fluid; that the concentrations of a selected reaction
(2) Negligible viscous dissipation; participant and temperatures in the fluid phase
(3) The same external force effect on each fluid and inside the solid particle are the same. In the
component; heterogeneous model, these parameters in the fluid
(4) Constant D, k and C p . phase and inside the particle are different. Conse-
quently, the balance equations for the first class
are similar to those for homogeneous systems, but
for the heterogeneous class the balance equations
are written separately for the fluid and solid
13 MULTIPLE GRADIENT TRANSPORT
phase.
BALANCES
In the practical application to the reactor design,
the multiple gradient balance equations may be
Simplification of microscopic balances involving
considerably simplified.
modification of transport coefficients in balance
The momentum balance in batch, semi-batch, short
equations leads to the more realistic equations
tubular flow and CSTFR's has no significance and
which may be applied to reactor design purposes.
can be safely omitted in design.
In the microscopic approach, these coefficients are
The basic design equations are derived for tubular
calculated by quantum mechanics, equilibrium and
flow reactor. However, they may readily be adjusted
non-equilibrium statistical mechanics and classical
to batch and semi-batch reactors by deleting
mechanics. In the multiple gradient models, these
insignificant terms.
coefficients are empirical and must be determined
Furthermore, the heat balance equations should be
by direct experiment. Semi-empirical equation may
omitted for isothermal design. Finally, for adiabatic
be of great value, when available. In the multiple
design, the term related to the heat exchange
gradient model, one or more parameters (velocity,
with the environment should be dropped out.
temperature, concentration) may be dispersed in
the axial and/or radial direction. In the maximum
gradient model, only temperature and concentration
gradients in the axial direction are considered. All
radial gradients and dispersions are neglected. 14 BOUNDARY CONDITIONS
The transport balances for the simplified gradient
models are based on the following general equation: Differential equations are integrated within the
boundary conditions, which may be determined
from the problem statement or actual process.
Net ( 2 ) rate of They may be given or inferred from the state of
Rate of _ bulk flow physical transport phenomena at the system bounda-
[accumulation (') through open + ries. Written in the form of mathematical equations,
surfaces they indicate whether the dependent variables at
the boundaries are at equilibrium or whether some
Net rate of flow
transport of mass, heat and momentum takes
by molecular
place.
+ transport through
open surfaces
Net rate of flow Net
+ by molecular rate of (1) Accumulation = depletion.
transport through
generation (2) Net input = input-output
closed surfaces ( 3) in reactor ( 4 ) (3) Heat transport by conduction.
Mass transport by diffusion, osmosis, molecular sieves.
(9) (4) Generation = consumption.
22
DESIGN OF THE CHEMICAL REACTOR
Net rate of flow r Net (1) Algebraic equations for certain macroscopic
by molecular + rate of models like the steady state isothermal CSTFR
^
transport through generation (difference equations);
_ closed surfaces in reactor (2) Differential equations for batch and plug flow
(10) tubular reactors;
(3) Differential-difference equations for . transient (2) Economic (level of profit or investment).
operations of CSTFR's;
(4) Partial differential equations for semi-batch, It is easy to guess that the technical criteria are
transient and microscopic and multiple gradient nothing else but the disguised economic criteria.
reactors.
20 OPTIMIZATION BY COMPUTER
SIMULATION
17 MATHEMATICAL ANALYSIS OF
DESIGN EQUATIONS The computer approach to the study of dynamic
behavior of the reactor is called computer simula-
The purpose of mathematical analysis of design
tion. It is based on computer solution of design
equations is to better comprehend dynamic behavior
equations leading to time-varying curves of con-
of the model and to establish the effect of inter-
centrations, temperature, flow rates, initial con-
actions by process variables and parameters on centration ratios, etc. and their analysis with
the yield of reaction, product quality, process
respect to the objective function [3].
economics, reactor stability and control. Four
approaches to the analysis of design equations
will be discussed in the following sections. They are :
21 CALCULATION OF THE OPTIMAL
(1) Process optimization; REACTOR SIZE
(2) Reactor stability;
(3) Control; and The ultimate purpose of reactor design is to create
(4) Statistical sensitivity analysis. the best reactor for the desired production. Engi-
neering optimization criteria are satisfied with the
smallest reactor producing at the highest rate the
high quality product. However, the production of
18 PROCESS OPTIMIZATION chemicals is unquestionably an economic process
and cost of the reactor and of its operation
The modern reactor design should lead to the is directly or indirectly related to the overall
highest yield, highest product quality, least expensive economy of the plant. Consequently, the design
production or highest monetary returns. This may must also satisfy the economic criteria which may
be achieved by the application of the principles involve the smallest unit cost of product, smallest
of optimization in various stages of the design total cost of operation or highest gross profit of
project. the production unit.
In engineering work we distinguish two types of To solve this problem, we have to formulate an
optimization criteria: expression for the economic objective function
(reactor profitability) in terms of the capital invested,
(1) Technical (yield, quality of product); and operating costs, sales income, required production
rate, steady state mass and heat balances, and (4) Analysis via the transfer function (Laplace
values of input stream properties [1]. Optimal transform space);
solution of thus obtained equation leads to the (5) Use of the Lyapunov methods; and
optimal value of conversion, which, in turn, may (6) Special techniques.
be used to the optimal (economically) size of the
reactor. Application of these methods to stability of chemical
Now, the optimal reactor must pass the test of reactors can be found in many fine texts on
stability and control. stability. The matrix algebra method seems to be
quite popular [7]. If the optimum economic steady
state reactor design is found unstable even in
certain ranges of operation only, the change of
the operating conditions or application of an
22 STABILITY ANALYSIS appropriate controller is necessary.
(1) The trajectory of the system motion, after a A study of BiLous and AMUNDSON [8] revealed
small disturbance, tends to return to the original that even small changes in reactor parameters,
state of the system and remains at it or in its such as wall temperature, feed dilution and wall
vicinity an infinite time. This system is called a heat transfer coefficient may affect considerably
stable one. the reactor temperature profile and cannot be
(2) The system is asymptotically stable when the ignored in the control design of a reactor. Provisions
trajectory of the system motion after a small should be made to keep the parameter in question
disturbance returns asymptotically to the original close to its critical value to obtain the desired rate
state of the system. and quality of product.
(3) The system is asymptotically stable at large
when after any disturbance it returns asymptotically
to the origin.
(4) The unstable system, after disturbance, moves 24 MULTIPLICITY OF STEADY STATES
away from the original state and never returns.
Certain combinations of feed composition and
The stability of reactor is determined by the stability reactor temperature lead to more than one steady
analysis of the following phenomena related to the state solution. As a rule, the even numbered steady
design and operation of reactor: states are unstable. The Van Heerden study of
CSTFR (I & EC, 45, 1242, 1952) revealed existence
(1) Stability of design equations; of three steady states, whereas computations of
(2) Stability of numerical solutions if the design Copelovitz and Aris (CES, 25, 906, 1970) show
equations cannot be solved analytically; 15 steady states for a catalyst particle.
(3) Parametric sensitivity;
(4) Multiplicity of steady states.
6 MIRK 10 i regions
ment involves the following steps:
CA
CA o
0 6 .
1111EARIE (I) Selection of appropriate mode of control by
dynamic analysis of individual streams to be
0.4 controlled;
0. 2 EA& NM
mignams (2) Choice of control equipment: pneumatic or
electronic. Hence, the cost of equipment and its
0 maintenance might be a decisive factor;
22 24 26 2.8 30 32 34
C P To (3) Location of appropriate controllers for particular
Feed condition CHCA
a
jobs;
(4) Adjustment of controller constants for the best
Fig. 7
Instability regions for a CSTFR
responses of the control system.
The purpose of process control is to maintain the In the solution of design equations, we use numerical
desirable (optimal) producing conditions by means values of design parameters like rate constants,
of mechanical or electronic equipment. Two control reaction orders, energies of activation, temperature
systems are in general use: (in the case of isothermal reactors), initial conditions
and transport coefficients, whose reliability may
(1) The closed-loop system (feed-back control); frequently arise some doubt. The statistical sensi-
(2) The open-loop system (feed-forward control). tivity analysis [10] explores the question of how
26 Rev. Port. Quim., 17, 15 (1975)
DESIGN OF THE CHEMICAL REACTOR
the parmeter uncertainties propagate into output- BEVERIDGE, G. S. G. and SCHECHTER, R. S., Optimiza-
variable uncertainties. It is based upon Taylor tion: Theory and Practice, McGraw-Hill, New York,
1970, p. 27.
series approximations for the variance and expected
DENN, M. M., Optimization by Variational Methods,
value of the output variables of the model. The McGraw-Hill, New York, 1969.
first partial derivatives of the output variables DOUGLAS, J. M., Process Dynamics and Control,
with respect to each parameter are determined from Vol. I, Prentice-Hall, New York, 1972.
BILOUS, O. and AMUNDSON, R. N., A. I. Ch. E. (Am.
the model structure. Determination of these deriva-
Inst. Chem. Engrs.) J., 2, 117 (1956).
tives and formation of the sensitivity matrix is the
MCGOV]N, G. and PERLUMTTER, W., A. I. Ch. E. (Am.
essential part of the technique. The solution of the Inst. Chem. Engrs.) J., 17, 831 (1967) .
sensitivity matrix leads to a set of linear equations, ATHERTON, R. W., SCHAINKER, R. B. and DUCOT,
indicating relationships between time and individual E. R., A. I. Ch. E. (Am. Inst. Chem. Engrs .) J., 21
parameters, which provide necessary information 441 (1975).
RESUMO