Lumped and
Distributed
parameter
System
Sec. 1.2 Models 5
Notice that both definitions stress that a model is a representation of a system or ohject. In
this textbook, when we usc the term model, we will be referring to a mathematical model.
We prefer to usc the following definition f~)r model (more specifically, a process model).
Working Definition: Pmccss Model
A process model is a set of equations (including the necessary input data to solve the equations) that
allows us to predict the behavior of a chemical process system.
The emphasis in this text is on the development and lise offundamental or first-principles
models. By fundamental, we mean models that are based on known physical-chemical rc~
lationships. This includes the conservation of mass and conservation of energy, I as well
as reaction kinetics, transport phenomena, and thermodynamic (phllse equilibrium, etc.)
relatiomdlips.
Another common model is the empirical model. An empirical model might be used
if the process is too complex for a fundamental model (either in the fonnulation of the
model, or the numerical solution of the model), or if the empirical model has satisfactory
predictive capability. An example of an empirical model is a simple least squares fit of an
equation lo experimental data.
Generally, we would prefer to use models based on fundamental knowledge of
chemical-physical relationships. Fundamental models will generally he accurate over .a
mueh larger range of conditions than empirical models. Empirical models may be useful
for "interpolation" but are generally not useful lilr "extrapolation"; that is, an empirical
model will only be useful over the range of conditions used for the "fit" of the data.
It should be nOled that it is rare for a single process model to exist. A model is only
an approximate representation of an actual process. The complexity of a process model
will depend on the final usc of the model. If only an approximate answer is needed, then a
simplified model can often be lIsed.
1.2.1 How Models Are Used
As we have noted, given a set of input data, a model is used to predict the output "re-
sponse." A model can be used to solve the following types of problems:
Marketing: If the price of a product is increased, how much will the demand de-
crease?
• Allocation: If we have several sources for raw materials, and several manufacturing
101' course, the real conservalioll law is that of mass-energy, but we will neglect the inter-
change of mass and energy due to nuclear reactions.
6 Introduction Chap. 1
plants, how do we distrihute the raw materials among the plants, and decide what
products each plant produces?
Synthesis: What process (sequence of reactors, separation devices, etc.) can he used
to manufacture a product?
• Design: What type and size of equipment is necessary to produce a product?
Operation: What operating conditions willmaxirnizc the yield of a product?
Control: How can a process input be manipulated to maintain a measured process
output at a desired value?
Safety: If an equipment failure occurs, what will he the impact on the operating per-
sonnel and other process equipment?
Environmental: IIow long will it take to "biodegrade" soil contaminated with haz-
ardolls waste?
Many or the rnodels cited above arc based on a steady-state analysis. This book will
cxtend the steady-state material and energy balance concepts, generally prescnted in an
introductory textbook on chemical engineering principles, to dynamic systems (systems
where the variables change with time). As an example of the increasing importance of
knowledge of dynamic behavior, consider process design. In the past, chemical process
design was based solely on steady-statc analysis. A problem with performing only a
steady-state design is that it is possible to design a process with desirable steady-state
characteristics (minimal energy consumption, etc.) but which is dynamically inoperable.
Hence, it is important to consider the dynamic operability characteristics of a process dur-
ing the design phase. Also, batch processes that arc commonly used in the pharmaceutical
or specialty chemicals industries arc inherently dynamic and cannot bc simulated with
steady-state models.
In the previous discussion we have characterized models as stcady-state or dy-
namic. Another characterization is in ter111s 01" lumped porarneter system.)· or distributed
parameter systems. A lumped parameter systcm assumes that a variable or interest (tem-
perature, for example) changes only with one independent variable (time, For cxamjJlc,
but not space). A typical example of a lumped parameter system is a perfectly mixed
(stirred) tank, where the temperature is uniform throughollt the tank. A distrihuted para-
meter system has more than one independent variable; for example, temperature may vary
with both spatial position and time.
EXAMPLE 1.1 A Lumped Parameter System
Consider a perfectly insulated, well-stirred tank where a hot liquid stream at 60 uC is mixed with
a cold liquid stream at lone (Figure 1.1). The wc11~mixed assumption means that the fluid tem-
perature in the tank is uniform and equal to the temperature at the exit from the t<mk. This is all
example of a lumped parameter system, since the temperature docs not vary with spatial pOSition.
Sec. 1.2 Models 7
Hot Cold
iii
Outlet
FIGURE 1.1 Stirred lank.
Consider now the ~leady-stale behavior of this process. If the only stream was the hot fluid,
thell the outlet temperature wOl/ld be eqllal to the hot fluid temperature if the tank were per-
fectly insulated. Similarly, if the only stream was the cold stream, then the outlet temperature
would be equal to the cold lluid temperature. A combination of the two streams yields an out-
let tClJlpcrutufc that is intermediate between the cold and hot temperatures, as shown in Fig-
ure ] .2.
60
o
f? 50
'0
@
.2 40
~
'"
Q.
~ 30
l-
ID
8 20
_~~------.l.- ------l~~~~~~~~
10
o 0.2 0.4 0.6 0.8
Hot Flow, fraction
FIGURE 1.2 Relationship between hot flow and outlet temperature.
We sec that there is a linear steady-state relationship between the hot flow (fraction) and the out-
let temperature.
Now we consider the dynamic response to a change in the fraction of hot flow. Figure 1.3
compares outlet temperature responses for various step changes in hot flow fraction at t :::: 2.5
8 Introduction Chap. 1
minutes. We sec that the changes afC symmetric, with the same speed of response. We will find
later that these responses afC indicative of a lincar system.
45 +20%
40
o
'" 35 f-~"-~~~~----Q~-o~---~-~~
~ . .- ----
ci:
E
230
-20%
25
,.--'-----
20 , ..
o 5 10 15 20 25 30
time, min
FIGlJRE 1.3 Response of temperature to various changes in the hoI llow fraction.
EXAMPLE 1.2 A Histributed Parameter System
A simplified representation of a counterflow heat exchanger is shown in Figure lA. A cold
water stream flows through one side of the excbanger and is heated by energy transfcrcd fro!ll a
condensing steam ,,,trcam. This is a distributed parameter systcnl because the tClnperature of the
water stream can change with lime and position.
~I-------F=:
Water in
Steam in
FIG-URE 1.4 Counterflow heat exchanger.
The steady~slate tell1perature profile (water temperature as a function of position) is shown in
Figure 1.5. Notice that rate of change of the water temperature \vith respect to distance decreases
as the water tcmpenlturc approaches the steam teinpcrature (I OOOe). This is because the tcmper-
Sec, 1,2 Models 9
ature gradient rOf heat transfer decreases as the water temperature increases. The outlet water
temperature as a function of inlet water temperature is shown in Figure 1.6.
100
80
0
'"
""ci. 60
E
JE
40
20
a 0.2 0.4 0.6 0.8
Z, distance
Flf;LJlU: 1.5 W::tter temperature as it function of position.
100
98
96
0
'" 94
"'"
J-'
E~
92
0
90
88
86
a 20 40 60 80 100
inlet temperature, deg C
FIGURE 1.6 Outlet water temperature;Is a function of inlet water temperature.
10 Introduction Chap. 1
Mathematical models consist of the following types of equations (including combina-
tions)
Algebraic equations
Ordinary differential equations
Partial differential equations
The emphasis in this textbook is on developing models that consist of ordinary dif-
ferential equations. These equations generally result from macroscopic balances around
processes, with an assumption of a perfectly mixed system. To find the steady-state solu-
tion of a set of ordinary differential equations, we must solve a set of algebraic equations.
Partial ditTerential equation models result from microscopic b<tlanccs and arc not covered
in this textbook. One of the main techniques for solving partial differential equations is
!:nlsed on converting a partial differential equation to a set of ordinary differential equa-
tions. Techniques deYeloped in this textbook can then be lIsed to solve these problems,
1.3 SYSTEMS
We have been using the term system yery loosely. Consider the following definition.
Dctinition: Svstem
A combination of several pieces of equipment integrated to perform a specific function; thus a fire
(arlillery) control system may include a tracking radar, computer, and gun (McGraw-Hill Dictio-
limy (?f5;ciellt!fic and Technical Tenns).
The example in the definition presellted is of interest to electrical, aeronautical, and rnili-
tary engineers. I'or our purposes, a system witl be composed of chemical unit operations,
such as chemical reactors, heat exchangers, and separation devices, which arc used to pro-
ducc a chemical product. Indeed, we will oftcn consider a single unit operation to be a
system composed of inputs, states (to be defined later) and outputs. A series of modules in
Section V of this textbook cover the behavior of a number of specific unit operations.
1.3.1 Simulation
One of the goals of this textbook is to develop numerical analysis techniques that allow liS
to "simulate" the behavior of a chemical process, Typically, steady-state simulation of a
lumped parameter system involves the solution of algebraic equations, while dynamic
simulation involves the solution of ordinary differential equations. We must be careful
when using computer simulation. First of all, we must he able to say, Do the results of this