EXAMPLES OF MATHEMATICAL MODELS OF CHEMICAL ENGINEERING SYSTEMS 83

3.16. A semibatch reactor is run at constant temperature by varying the rate of addition of
one of the reactants, A. The irreversible, exothermic reaction is first order in reac-
tants A and B.
k
A + B - C
The tank is initially filled to its 40 percent level with pure reactant B at a
concentration CBO . Maximum cooling-water flow is begun, and reactant A is slowly
added to the perfectly stirred vessel.
Write the equations describing the system. Without solving the equations, try
to sketch the profiles of. F, A, CA, and C, with time during the batch cycle.

Cooling
w
water

FIGURE P3.16

3.17. Develop a mathematical model for the three-column train of distillation columns
sketched below. The feed to the first column is 400 kg * mol/h and contains four
components (1, 2, 3, and 4), each at 25 mol %. Most of the lightest component is
removed in the distillate of the first column, most of the next lightest in the second
column distillate and the final column separates the final two heavy components.
Assume constant relative volatilities throughout the system: ai, a2, and as. The
condensers are total condensers and the reboilers are partial. Trays, column bases,
and reflux drums are perfectly mixed. Distillate flow rates are set by reflux drum

D3

F-

B3

FIGURE P3.17
84 MATHEMATICAL MODELS OF CHEMICAL ENGINEERING SYSTEMS

level controllers. Reflux flows are fixed. Steam flows to the reboilers are set by
temperature controllers. Assume equimolal overflow, negligible vapor holdup, and
negligible condenser and reboiler dynamics. Use a linear liquid hydraulic relation-
ship

where L and fi, are the initial steadystate liquid rate and holdup and fl is a con-
stant with units of seconds.
3.18. The rate of pulp lay-down F on a paper machine is controlled by controlling both
the pressure P and the height of slurry h in a feeder drum with cross-sectional area
A. F is proportional to the square root of the pressure at the exit slit. The air vent
rate G is proportional to the square root of the air pressure in the box P. Feedback
controllers set the inflow rates of air G, and slurry F, to hold P and h. The system is
isothermal.
Derive a dynamic mathematical model describing the system.

Restriction

FIGURE P3.18

3.19. A wax filtration plant has six filters that operate in parallel, feeding from one
common feed tank. Each filter can handle 1000 gpm when running, but the filters
must be taken off-line every six hours for a cleaning procedure that takes ten
minutes. The operating schedule calls for one filter to be cleaned every hour.
How many gallons a day can the plant handle? If the flow rate into the feed
tank is held constant at this average flow rate, sketch how the liquid level in the feed
tank varies over a typical three-hour period.
3.20. Alkylation is used in many petroleum refineries to react unsaturated butylenes with
isobutane to form high octane iso-octani: (alkylate). The reaction is carried out in a
two liquid-phase system: sulfuric acid/hydrocarbon.
The butylene feed stream is split and fed into each of a series of perfectly
mixed tanks (usually in one large vessel). This stepwise addition of butylene and the
large excess of isobutane that is used both help to prevent undesirable reaction of
butylene molecules with each other to form high-boiling, low octane polymers. Low
temperature (40°F) also favors the desired iC,/Ch reaction.
The reaction is exothermic. One method of heat removal that is often used is
autorefrigeration: the heat of vaporization of the boiling hydrocarbon liquid soaks
up the heat of reaction.
 EXAMPLES OF MATHEMATICAL MODELS OF CHEMICAL ENGINEERING SYSTEMS 85

The two liquid phases are completely mixed in the agitated sections, but in the
last section the two phases are allowed to separate so that the acid can be recycled
and the hydrocarbon phase sent off to a distillation column for separation.
Derive a dynamic mathematical model of the reactor.
Vapor to compressor

Acid recycle

k,
c; - p o l y m e r

FIGURE P3.20

3.21. Benzene is nitrated in an isothermal CSTR in three sequential irreversible reactions:

Benzene + HNO, -
‘I nitrobenzene + H,O
P
Nitrobenzene + HNO, -
” dinitrobenzene + H,O

Dinitrobenzene + HNO, x trinitrobenzene + H,O _I

‘L
Assuming each reaction is linearily dependent on the concentrations of each reac-
tant, derive a dynamic mathematical model of the system. There are two feed
streams, one pure benzene and one concentrated nitric acid (98 wt %). Assume con-
stant densities and complete miscibility.
 PART

II
COMPUTER
SIMULATION

I n the next two chapters we will study computer simulation techniques for
solving some of the systems of equations we generated in the two preceding
chapters. A number of useful numerical methods are discussed in Chap. 4, includ-
ing numerical integration of ordinary differential equations. Several examples
are given in Chap. 5, starting with some simple systems and evolving into more
realistic and complex processes to illustrate how to handle large numbers of
equations.
Only digital simulation solutions for ordinary differential equations are pre-
sented. To present anything more than a very superficial treatment of simulation
techniques for partial differential equations would require more space than is
available in this book. This subject is covered in several texts. In many practical
problems, distributed systems are often broken up into a number of “lumps”
which can then be handled by ordinary differential equations.

87
88 COMPUTER SIMULATION

Our discussions will be limited to only the most important and useful
aspects of simulation. The techniques presented will be quite simple and unso-
phisticated, but I have found them to work just as well for most real systems as
those that are more mathematically elegant. They also have the added virtues of
being easy to understand and easy to program.
Some of the simple linear equations that we will simulate can, of course, be
solved analytically by the methods covered in Part III to obtain general solu-
tions. The nonlinear equations cannot, in general, be solved analytically, and
computer simulation is usually required to get a solution. Keep in mind,
however, that you must give the computer specific numerical values for param-
eters, initial conditions, and forcing functions. And you will get out of the com-
puter specific numerical values for the solution. You cannot get a general
solution in terms of arbitrary, unspecified inputs, parameters, etc., as you can
with an analytic solution.
A working knowledge of FORTRAN 77 digital programming language is
assumed and all programs are written in FORTRAN. Those who prefer other
languages will find the conversion fairly easy since the programs are simple trans-
lations of the equations into source code. All programs can be run on any type of
computer, personal computers or mainframes. The big multicomponent distilla-
tion column simulations require a lot of number crunching so are usually run on
a mainframe. Most of the other programs can be conveniently run on personal
computers.
 CHAPTER

NUMERICAL
METHODS

4.1 INTRODUCTION
Digital simulation is a powerful tool for solving the equations describing chemi-
cal engineering systems. The principal difficulties are two: (1) solution of simulta-
neous nonlinear algebraic equations (usually done by some iterative method), and
(2) numerical integration of ordinary differential equations (using discrete finite-
difference equations to approximate continuous differential equations).
The accuracy and the numerical stability of these approximating equations
must be kept in mind. Both accuracy and stability are affected by the finite-
difference equation (or integration algorithm) employed, Many algorithms have
been proposed in the literature. Some work better (i.e., faster and therefore
cheaper for a specified degree of accuracy) on some problems than others. Unfor-
tunately there is no one algorithm that works best for all problems. However, as
we will discuss in more detail later, the simple first-order explicit Euler algorithm
is the best for a large number of engineering applications.
Over the years a number of digital simulation packages have been devel-
oped. In theory, these simulation languages relieve the engineer of knowing any-
thing about numerical integration. They automatically monitor errors and
stability and adjust the integration interval or step size to stay within some accu-
racy criterion. In theory, these packages make it easier for the engineer to set up
and solve problems.
In practice, however, these simulation languages have limited utility. In
their push for generality, they usually have become inefficient. The computer
execution time for a realistic engineering problem when run on one of these simu-

89