Professional Documents
Culture Documents
University of Technology
Mathematical
Modeling and
Numerical Analysis
1.1 Introduction
The objective of mathematical modeling is the development of sets of quantitative
(mathematical) expressions that capture the essentials aspects of an existing system.
A mathematical model can assist in understanding the complex physical interactions
in the system and the causes and effects between the system variables. Mathematical
models are valuable tools since they are abstract equations that can be solved and
analyzed using computer calculations. It is therefore safer and cheaper to perform test
on the model using computer simulations rather than to carry out repetitive
experimentations and observations on the real system. This becomes vital if the real
system is new, hazardous, or expensive to operate. Modeling thus prevails the field of
science, engineering and business. It is used to assist in the design of equipment, to
predict behavior, to interpret data, to optimize resources and to communicate
information.
1
strategies. Furthermore, to verify the reliability and safety of the process design
tests can be carried out even prior to plant commissioning.
• Plant operators training
Models can be used to train plant personnel to simulate startup and shutdown
procedures, to operate complex processes and to handle emergency situations
and procedures.
• Controller design and tuning
Models help in developing and evaluating better controller structure and
configuration. Dynamic simulation of models is usually employed for testing
and assessing the effectiveness of various controller algorithms. It is worthwhile
to mention that models play a vital part in designing advanced model-based
control algorithms such as model predictive and internal model controller.
Moreover it is a common practice of many control engineers to determine the
optimum values of the controller settings through dynamic simulation .
• Process optimization
It is desirable from economic standpoint to conduct process optimization before
plant operation to determine the optimum values of the process key parameters
or/and operating conditions that maximizes profit and reduces cost. Process
optimization is also performed during process operation to account for
variations in the feed-stock and utilities market and for changing environmental
regulations.
It is worth mentioning that despite all their usefulness, models at their best are no
more than approximation of the real process since they do not necessarily incorporate
all the features of the real system. Therefore modeling can not eliminate completely
the need for some plant tests, especially to validate developed models or when some
poorly known parameters in the process need to be experimentally evaluated.
Models can be classified in a number of ways. But since mathematical models are
developed from applying the fundamental physical and chemical laws on a specific
2
system, we review first the classification of systems since their nature affect the
modeling approach and the resulting model.
1.3 Systems
A system is a whole consisting of elements or subsystems. The system has
boundaries that distinguish it from the surrounding environment (external world) as
shown in Figure 1.1. The system may exchange matter and/or energy with the
surrounding through its boundary. Consequently, the state of a system can be defined
or understood via the interactions of its elements with the external world. A system
may be classified in different ways, some of which are as follows:
Boundary
System
Suroundings
3
1.3.2 Classification based on number of phases
• Homogeneous system
This is a system that involves only one phase such as gas-phase or liquid-phase
chemical reaction processes.
• Heterogeneous system
This is a system that involves more than one phase. This kind of systems exists
in multi-phase reaction processes and in phase-based separation processes.
4
processing systems a number fundamental quantities are used to describe the natural
state of the system. These quantities are the mass, energy and momentum. Most often
these fundamental quantities can not be measured directly thus they are usually
represented by other variables that can be measured directly and conveniently. The
most common variables are density, concentration, temperature, pressure and flow
rate. There are conveniently called 'state variables' since they characterize the state of
the processing system.
In order to describe the behavior of the system with time and position, the state
(dependent) variables should be linked to the independent variables (time, spatial
position) through sets of equations that are derived from writing mass, energy and
momentum balances. The set of equations describing these variables are called 'state
equations.'
7
S
Sin
System Sout
Figure 1.2
8
Rate of Flow of Rate of
Flow of
Generation of moles of (A) Accumulation of (1.3)
moles (A) in + = +
moles of (A) out moles of (A)
The component balance has the unit of moles A/s. It should be noted that unlike the
total mass, the number of moles of species A is not conserved. The species A can be
generated or consumed by chemical reaction.
Since the velocity v is a vector, the momentum, unlike the mass is also a vector. The
momentum balance equation using (Eq 1.1) is:
The momentum balance has the unit of kg.m/s2. The momentum balance equation is
P P
usually written using the Newton's second law. The law states that the time rate of
change of momentum of a system is equal to the sum of all forces F acting on the
system,
d (mv)
= ∑F
(1.6)
dt
9
Rate of Rate of Rate of Rate of ± Amount of
energy + Generation = energy + accumulation energy (1.7)
in of energy out of energy exchanged
with the
surrounding
The energy generated within a system includes the rate of heat and the rate of work.
The rate of heat includes the heat of reaction (if a reaction occurs in the system), and
the heat exchanged with the surroundings. For the rate of work we will distinguish
between the work done against pressure forces (flow work) and the other work such
as the work done against the gravity force, against viscous forces and shaft work.
For a given system the general conservation law (Eq. 1.1) can be carried out either on
microscopic scale or macroscopic scale.
10
z
y
∆z
∆y
∆x
y
r
θ
θ
r z
11
1.10 Macroscopic balance
In some cases the process state variables are uniform over the entire system, that is
each state variable does not depend on the spatial variables, i.e. x,y and z in cartesian
coordinates but only on time t. In this case the balance equation is written over the
whole system using macroscopic modeling. When modeling the process on
microscopic scale the resulting models consists usually of partial differential equation
(PDE) where time and one or more spatial position are the independent variables. At
steady state, the PDE becomes independent of t and the spatial positions are the only
independent variables. On the other hand, when the modeling is based on
macroscopic scale the resulting model consists of sets of ordinary differential
equations (ODE) .
The fundamental balance equations of mass, momentum and energy already
discussed are usually supplemented with a number of equations associated with
transport rates and thermodynamic relationships. In the following we present an
overview of some of these relations.
In many cases the two transport mechanism occurs together. Therefore, the flux due
to the transport of any fundamental quantity is the sum of a flux due to convection
and a flux due to diffusion.
n Au = j Au + ρ A v u
R R R R R R R
(1.8)
12
total flux = diffusive flux + bulk flux (1.9)
The diffusive flux j Au (kg/m2s) for a binary mixture A-B is given by Fick's law:
R R P P
dw A (1.10)
j Au = − ρD AB
du
mixture and D AB (m2/s) is the diffusivity coefficient of A in the mixture. In molar unit
R R P P
dx A (1.11)
J Au = −CD AB
du
mole fraction of A in the mixture. For constant density the flux Eqs. ( 1.10 and Eq.
1.11) become:
dρ A (1.12)
j Au = − DAB
du
dC A (1.13)
J Au = − D AB
du
in the y-direction, consider a fluid flowing between two infinite parallel plates as
13
shown in Figure 1.6. At a certain time the lower plate is moved by applying a
constant force F x while the upper plate is maintained constant. The force F x is called
R R R R
a shear force since it is tangential to the area A y on which it is applied (Fig. 1.6). R R
y
∆y F, force
∆ Vx
is called a stress and denoted τyx (kg.m/s). It is also a shear stress since it is tangential.
R R
plate. Because of molecular transport, the layer above it has a slightly slower velocity
v x (y) and so on as shown in Figure 1.6. Therefore, there is a transport by diffusion of
R R
the x-component of the momentum in the y-direction. The flux of this diffusive
transport is in fact the shear stress τyx . R R
Therefore, the total flux πyx of the x-component in the y-direction is the sum of the
R R
π yx = τ yx + (ρv x )v y
R R R R R R R
(1.16)
momentum flux = diffusive flux + bulk flux (1.17)
For a Newtonian fluid the shear stress τyx is proportional to the velocity gradient: R R
∂v x (1.18)
τ yx = − µ
∂y
e u = q u + (ρC p T)v u
R R R R R R R
(1.19)
energy flux = diffusive flux + bulk flux (1.20)
The heat flux by molecular diffusion, i.e. conduction in the u-direction is given by
Fourier's law:
∂T (1.21)
qu = −k
∂u
The three relations (Eq. 1.10, 1.18, 1.21) show the analogy that exists between mass,
momentum and energy transport. The diffusive flux in each case is given by the
following form:
Flux = − transport property × potential difference (gradient) (1.22)
The flux represents the rate of transfer per area, the potential difference indicates the
driving force and the transport property is the proportionality constant. Table 1.1
summarizes the transport laws for molecular diffusion.
15
When modeling a process on macroscopic level we can also express the flux by a
relation equivalent to (Eq. 1.22). In this case the gradient is the difference between
the bulk properties, i.e. concentration or temperature in two medium in contact, while
the transport property represents an overall transfer coefficient. For example for mass
transfer problems, the molar flux can be expressed as follows:
J A = K × ∆C A (1.23)
where K an overall mass transfer coefficient.
As for the momentum balance, the macroscopic description generally uses the
pressure drop as the gradient while the friction coefficient is used instead of the flux.
The following relation is for instance commonly used to describe the momentum
laminar transport in a pipe.
f =
D
∆P
(1.25)
2v 2 Lρ
f is the Fanning friction factor, ∆P is the pressure drop due to friction, D is the
diameter of the pipe, L the length and v is the velocity of the fluid.
• Enthalpies :
Liquid and vapor enthalpies for pure component can be computed from simple
formulas, based on neglecting the pressure effect as follows:
~
h = C p (T − Tref ) (1.28)
~
H = C p (T − Tref ) + λ (1.29)
~
Where h is the liquid specific enthalpy, H~ is its vapor specific enthalpy, C p
the average liquid heat capacity and λ is the latent heat of vaporization
heat of mixing is negligible, the enthalpy of a mixture can be taken as the sum
of the specific enthalpies of the pure components multiplied by their
corresponding mole fractions Note that the above enthalpy functions are valid
for small temperature variation and/or when the heat capacities of fluids are
weak function of temperature. In general cases the heat capacity C p can be
R R
~
T (1.32)
H= ∫ C p dT + λ
Tref
• Internal energy :
17
The internal energy U is a fundamental quantity that appears in energy balance
equation. For liquids and solids the internal energy can be approximated by
enthalpy. This can be also a good approximation for gases if the pressure
change is small.
y Ai = F(x Ai )
R R R R (1.33)
The equilibrium relations are in general nonlinear. However, quite satisfactory results
can be obtained through the use of simpler relations that are derived form
assumptions of ideal behavior of the two phases:
yAG yAi
xAi
xAL
NA
Interface
18
• For vapor phases at low concentration, Henry's law provides the following
simple equilibrium relation:
p A = Hx A
R R R (1.34)
Where p A (atm) is the partial pressure of species A in the vapor, x A is the mole
R R R R
Dividing by the total Pressure (P) we get another form of Henry's law:
~
yA = H xA
R R R
(1.35)
~
The constant ( H ) depends on temperature and pressure.
ln( PAS ) = A −
B (1.37)
C +T
where A, B and C are characteristic parameters of the fluid.
• Raoult’s law can be modified to account for non-ideal liquid and vapor
behavior using the activity coefficients γ i and φi of component (i) for liquid R R R R
and vapor phase respectively. The following equilibrium relation, also known
as the Gamma/Phi formulation, can be used:
y iφi P = xi γ i Pi S (1.38)
19
The activity coefficients can be determined using correlations found in standard
thermodynamics text books. Equation (1.38) is reduced to Raoult law for ideal
mixture i.e. ( φi = γ i =1).
T.
• Dew point calculations :
For a given molar vapor compositions (y i ) and either T or P, the dew point
R R
or T.
• Flash calculation :
For known mixture compositions (z i ) and known (T) and (P), the flash
R R
R=
1 dni (1.40)
viV dt
20
where
ni
R R the number of moles
νi
R R the stochiometric coefficient of component i
V is the volume due to the chemical reaction.
The rate expression R is generally a complex relation of the concentrations (or partial
pressures) of the reactants and products in addition to pressure and temperature.
The law of mass action stipulates that the reaction rate is a power law function of
temperature and concentration of reactants, i.e.
R = kCRζ1 C Rζ 2 C Rζ r (1.42)
1 2 r
The powers ξ i are determined experimentally and their values are not necessarily
R R
integers. The temperature dependence comes from the reaction rate constant k given
by Arrhenius law
−E (1.43)
k = ko e RT
21
Depending on the value of f three cases can be distinguished:
• f = 0. The system is exactly determined (specified) system. Thus, the set of
balance equation has a finite number of solutions (one solution for linear
systems).
• f < 0. The system is over-determined (over-specified) by f equations. f
equations have to be removed for the system to have a solution.
• f > 0. The system is under-determined (under-specified) by f equations. The set
of equation, hence, has infinite number of solution.
22
In fact the only class of differential equations for which there is a well-developed
framework are linear ODE. However linearizing the original nonlinear model and
solving it is not always recommended (expect for control purposes) since the
behavior of the linearized model matches the original nonlinear model only around
the state chosen for linearization. For these reasons the solution of process models is
usually carried out numerically, most often through a computer programs.
23
Chapter 2: Examples of Mathematical Models for Chemical
Processes
Consider the perfectly mixed storage tank shown in figure 2.1. Liquid stream with
volumetric rate F f (m3/s) and density ρ f flow into the tank. The outlet stream has
R R P P R R
volumetric rate F o and density ρ ο . Our objective is to develop a model for the
R R R R
variations of the tank holdup, i.e. volume of the tank. The system is therefore the
liquid in the tank. We will assume that it is perfectly mixed and that the density of the
effluent is the same as that of tank content. We will also assume that the tank is
24
isothermal, i.e. no variations in the temperature. To model the tank we need only to
write a mass balance equation.
Ff
ρf V
Fo
ρo
Since the system is perfectly mixed, the system properties do not vary with position
inside the tank. The only variations are with time. The mass balance equation can be
written then on the whole system and not only on a differential element of it. This
leads to therefore to a macroscopic model.
We apply the general balance equation (Eq. 1.2), to the total mass m = ρV. This
yields:
Mass flow in:
ρf FfR R R
(2.1)
Mass flow out:
ρo Fo
R R R
(2.2)
Accumulation:
dm d (ρV )
=
dt dt (2.3)
The generation term is zero since the mass is conserved. The balance equation yields:
d (ρV )
ρ f F f = ρo Fo +
dt (2.4)
For consistency we can check that all the terms in the equation have the SI unit of
kg/s. The resulting model (Eq. 2.4) is an ordinary differential equation (ODE) of first
order where time (t) is the only independent variable. This is therefore a lumped
parameter model. To solve it we need one initial condition that gives the value of the
volume at initial time t i , i.e.
R R
25
V(t i ) = V i
R R R (2.5)
Under isothermal conditions we can further assume that the density of the liquid is
constant i.e. ρ f = ρ o =ρ. In this case Eq. 2.4 is reduced to:
R R R R
dV
= F f − Fo
dt (2.6)
The volume V is related to the height of the tank L and to the cross sectional area A
by:
V = AL (2.7)
Since (A) is constant then we obtain the equation in terms of the state variable L:
dL
A = F f − Fo
dt (2.8)
with initial condition:
L(t i ) = L i
R R R (2.9)
For the system to be exactly specified we need therefore one more equations. This
extra relation is obtained from practical engineering considerations. If the system is
operated without control (at open loop) then the outlet flow rate F o is a function of
R R
26
Note that at steady state, the accumulation term is zero (height does not change with
time), i.e., dL/dt = 0. The model of the tank is reduced to the simple algebraic
equation:
F0 = Ff
R R R R (2.11)
We consider the liquid tank of the last example but at non-isothermal conditions.
The liquid enters the tank with a flow rate F f (m3/s), density ρ f (kg/m3) and
R R P P R R P P
assumed constant. The effluent stream is of flow rate F o (m3/s), density ρ o (kg/m3) R R P P R R P P
and temperature T(K) (Fig. 2.2). Our objective is to model both the variation of liquid
level and its temperature. As in the previous example we carry out a macroscopic
model over the whole system. Assuming that the variations of temperature are not as
large as to affect the density then the mass balance of Eq. 2.8 remains valid.
To describe the variations of the temperature we need to write an energy balance
equation. In the following we develop the energy balance for any macroscopic system
(Fig. 2.3) and then we apply it to our example of stirred tank heater.
Consequently, the flow of energy into the system is:
~
ρf Ff h f R R R R
(2.12)
Q
L
Fo , To , ρo
Tst
Heat supply
27
Figure 2.3 General Macroscopic System
Substituting all these terms in the general balance equation (Eq. 1.7) yields:
(
d ρVh
~
) ~ ~
= ρ f F f h f − ρ o Fo ho + Qe + Qr (2.15)
dt
We can check that all terms of this equation have the SI unit of (J/s). We return now
to the liquid stirred tank heater. A simplifying assumption can be introduced, that
there is no reaction involved, i.e. Q r = 0.R R
28
~ ~
h = Cp(T − Tref ) (2.17)
~
Where Cp is the average heat capacity.
Furthermore since the tank is well mixed the effluent temperature T o is equal to R R
ρC p
(
~ d V (T − Tref ) ~ ) ~
= ρF f C p (T f − Tref ) − ρFoC p (T − Tref ) + Qe (2.18)
dt
Taking T ref = 0 for simplicity and since V = AL result in:
R R
~ d (LT ) ~ ~
ρC p A = ρF f C pT f − ρFoC pT + Qe (2.19)
dt
or equivalently:
d (LT ) Q
A = F f T f − FoT + ~e (2.20)
dt ρC p
Since
d (LT ) d (L ) d (T )
A = AT + AL (2.21)
dt dt dt
And using the mass balance (Eq. 2.8) we get:
dT Q
AL + T ( F f − Fo ) = F f T f − FoT + ~e (2.22)
dt ρC p
or equivalently:
dT Q
AL = F f (T f − T ) + ~e (2.23)
dt ρC p
The stirred tank heater is modeled, then by the following coupled ODE's:
A
dL
= F f − Fo
(2.24)
dt
AL
dT Q
= F f (T f − T ) + ~e
(2.25)
dt ρC p
29
Degree of freedoms analysis
For this system we can make the following simple analysis:
• Parameter of constant values: A, ρ and C p R
• Remaining variables: L, F o , T, Q e R R R
The degree of freedom is therefore, 4− 2 = 2. We s till need two relations for our
problem to be exactly specified. Similarly to the previous example, if the system is
operated without control then F o is related to L through (Eq. 2.10). One additional
R R
relation is obtained from the heat transfer relation that specifies the amount of heat
supplied:
Q e = UA H (T st −T )
R R R R R R (2.27)
U and A H are heat transfer coefficient and heat transfer area. The source temperature
R R
We revisit the perfectly mixed tank of the first example but where a liquid phase
chemical reactions taking place:
k
A
→ B (2.28)
The reaction is assumed to be irreversible and of first order. As shown in figure 2.4,
the feed enters the reactor with volumetric rate F f (m3/s), density ρ f (kg/m3) and R R P P R R P P
concentration C Af (mole/m3). The output comes out of the reactor at volumetric rate
R R P P
isothermal conditions.
Our objective is to develop a model for the variation of the volume of the reactor
and the concentration of species A and B. The assumptions of example 2.1.1 still hold
and the total mass balance equation (Eq. 2.6) is therefore unchanged
30
Ff
ρf
CAf V Fo
CBf ρo
CAo
CBo
Figure 2.4 Isothermal CSTR
We could write a similar component balance on species B but it is not needed since
it will not represent an independent equation. In fact, as a general rule, a system of n
species is exactly specified by n independent equations. We can write either the total
mass balance along with (n −1) component balance equations, or we can write n
component balance equations.
31
Using the differential principles, equation (2.32) can be written as follows:
d (VC A )
=V
d (C A )
+ CA
d (V )
= F f C Af − Fo C A − rV
(2.33)
dt dt dt
Substituting Equation (2.6) into (2.33) and with some algebraic manipulations we
obtain:
d (C A )
V + C A ( F f − Fo ) = F f C Af − Fo C A − rV (2.34)
dt
or equivalently:
d (C A )
V = F f ( C Af − C A ) − rV (2.35)
dt
In order to fully define the model, we need to define the reaction rate which is for a
first-order irreversible reaction:
r = k CA R R (2.36)
Equations 2.6 and 2.35 define the dynamic behavior of the reactor. They can be
solved if the system is exactly specified and if the initial conditions are given:
V(t i ) = V i and C A (t i ) = C Ai
R R R R R R R R R (2.37)
The degree of freedom is therefore 3− 2 =1. The extra relation is obtained by the
relation between the effluent flow F o and the level in open loop operation (Eq. 2.10).
R R
The steady state behavior can be simply obtained by setting the accumulation terms
to zero. Equation 2.6 and 2.35 become:
F0 = Ff
R R R (2.38)
F f ( C Af − C A ) = rV (2.39)
32
2.1.4 Isothermal CSTR of Two feed streams
More complex situations can also be modeled in the same fashion. Consider the
catalytic hydrogenation of ethylene:
A+BP (2.40)
Where A represents hydrogen, B represents ethylene and P is the product (ethane).
The reaction takes place in the CSTR shown in figure 2.5. Two streams are feeding
the reactor. One concentrated feed with flow rate F 1 (m3/s) and concentration C B1 R R P P R R
(mole/m3) and another dilute stream with flow rate F 2 (m3/s) and concentration C B2
P P R R P P R R
(mole/m3). The effluent has flow rate F o (m3/s) and concentration C B (mole/m3). The
P P R R P P R R P P
V
Fo, CB
r=
k1CB
( mole / m 3 .s )
(2.41)
(1 + k2CB ) 2
Where k 1 is the reaction rate constant and k 2 is the adsorption equilibrium constant.
R R R R
Assuming the operation to be isothermal and the density is constant, and following
the same procedure of the previous example we get the following model:
Total mass balance:
dL
A = F1 + F2 − Fo (2.42)
dt
Component B balance:
d (C B )
V = F1 ( C B1 − C B ) + F2 ( C B 2 − C B ) − rV (2.43)
dt
33
Degrees of freedom analysis
• Parameter of constant values: A, k 1 and k 2 R R R
We reconsider the previous CSTR example (Sec 2.1.3), but for non-isothermal
conditions. The reaction A B is exothermic and the heat generated in the reactor is
removed via a cooling system as shown in figure 2.6. The effluent temperature is
different from the inlet temperature due to heat generation by the exothermic
reaction.
Ff , CAf , Tf
Qe V Fo, CA, T
Assuming constant density, the macroscopic total mass balance (Eq. 2.6) and mass
component balance (Eq. 2.35) remain the same as before. However, one more ODE
will be produced from the applying the conservation law (equation 2.20) for total
energy balance. The dependence of the rate constant on the temperature:
k = k o e-E/RT R R P (2.44)
The general energy balance (Eq. 2.20) for macroscopic systems applied to the CSTR
yields, assuming constant density and average heat capacity:
34
ρC p
(
~ d V (T − Tref ) ~ ) ~
= ρF f C p (T f − Tref ) − ρFoC p (T − Tref ) + Qr − Qe (2.45)
dt
Where Q r (J/s) is the heat generated by the reaction, and Q e (J/s) the rate of heat
R R R R
removed by the cooling system. Assuming T ref = 0 for simplicity and using the R R
Q r = −(∆H r )Vr
R R R R
(2.48)
Where ∆H r (J/mole) is the heat of reaction (has negative value for exothermic
R R
reaction and positive value for endothermic reaction). The non-isothermal CSTR is
therefore modeled by three ODE's:
dV
= F f − Fo
(2.49)
dt
V
d (C A )
= F f ( C Af − C A ) − rV
(2.50)
dt
~ dT
ρC pV
~
= ρF f C p (T f − T ) + (−∆H r )Vr − Qe
(2.51)
dt
Where the rate (r) is given by:
r = k o e-E/RTC A R R P P R (2.52)
The system can be solved if the system is exactly specified and if the initial
conditions are given:
V(t i ) = V i
R R R R T(t i ) = T i R R R R and C A (t i ) = C Ai
R R R R R (2.53)
35
The degree of freedom is 5−3 = 2. Following the analysis of example 2.1.3, the two
extra relations are between the effluent stream (F o ) and the volume (V) on one hand
R R
and between the rate of heat exchanged (Q e ) and temperature (T) on the other hand,
R R
Consider the shell and tube heat exchanger shown in figure 2.7. Liquid A of
density ρ A is flowing through the inner tube and is being heated from temperature
R R
both liquids varies not only with time but also along the tubes (i.e. axial direction)
and possibly with the radial direction too. Tubular heat exchangers are therefore
typical examples of distributed parameters systems. A rigorous model would require
writing a microscopic balance around a differential element of the system. This
would lead to a set of partial differential equations. However, in many practical
situations we would like to model the tubular heat exchanger using simple ordinary
differential equations. This can be possible if we think about the heat exchanger
within the unit as being an exchanger between two perfect mixed tanks. Each one of
them contains a liquid.
Tw Liquid, B
TB1
Liquid, A
TA1 TA2
Liquid, B
TB2
36
For the time being we neglect the thermal capacity of the metal wall separating the
two liquids. This means that the dynamics of the metal wall are not included in the
model. We will also assume constant densities and constant average heat capacities.
One way to model the heat exchanger is to take as state variable the exit temperatures
T A 2 and T B 2 of each liquid. A better way would be to take as state variable not the
R RR R R RR R
exit temperature but the average temperature between the inlet and outlet:
TA1 + TA2
TA = (2.54)
2
TB1 + TB 2
TB = (2.55)
2
For liquid A, a macroscopic energy balance yields:
dTA
ρ AC p VA = ρ A FAC p A (TA1 − TA2 ) + Q (2.56)
A
dt
where Q (J/s) is the rate of heat gained by liquid A. Similarly for liquid B:
dTB
ρ BC p VB = ρ B FBC p B (TB1 − TB 2 ) − Q (2.57)
B
dt
The amount of heat Q exchanged is:
Q = UA H (T B – T A )
R R R R R R (2.58)
Or using the log mean temperature difference:
Q = UA H ∆T lm R R R
(2.59)
Where
(TA2 − TB1 ) − (TA1 − TB 2 )
∆Tlm =
(T − TB1 )
ln A2 (2.60)
(TA1 − TB 2 )
with U (J/m2s) and A H (m2) being respectively the overall heat transfer coefficient and
P P R R P P
heat transfer area. The heat exchanger is therefore describe by the two simple ODE's
(Eq. 2.56) and (Eq. 2.57) and the algebraic equation (Eq. 2.58).
37
• Remaining variables: T A2 , T B2 , Q R R R R
The degree of freedom is 5− 3 = 2. The two extra relations are obtained by noting
that the flows F A and F B are generally regulated through valves to avoid fluctuations
R R R R
in their values.
So far we have neglected the thermal capacity of the metal wall separating the two
liquids. A more elaborated model would include the energy balance on the metal wall
as well. We assume that the metal wall is of volume V w , density ρ w and constant heat R R R R
capacity Cp w . We also assume that the wall is at constant temperature T w , not a bad
R R R R
assumption if the metal is assumed to have large conductivity and if the metal is not
very thick. The heat transfer depends on the heat transfer coefficient h o,t on the R R
outside and on the heat transfer coefficient h i,t on the inside. Writing the energy R R
ρ BC p VB
dTB
= ρ B FBC p B (TB1 − TB 2 ) − ho , t Ao , t (TB − TW )
(2.61)
B
dt
Where A o,t is the outside heat transfer area. The energy balance for the metal yields:
R R
ρ wC p Vw
dTw
= ho , t Ao , t (TB − Tw ) − hi , t Ai , t (Tw − TA )
(2.62)
w
dt
Where A i,t is the inside heat transfer area. The energy balance for liquid A yields:
R R
ρ AC p VA
dTA
= ρ A FAC p A (TA1 − TA2 ) + hi , t Ai , t (Tw − TA )
(2.63)
A
dt
Note that the introduction of equation (Eq. 2.62) does not change the degree of
freedom of the system.
38
Tw Steam
Ts(t)
Liquid, L
TL1 TL2
condensate, Ts
T s = T s (P)
R R R R (2.67)
Assuming ideal gas law, then the mass flow of steam is:
M s PsVs
ms = (2.68)
RTs
Where M s is the molecular weight and R is the ideal gas constant. The mass balance
R R
Where F c and ρ c are the condensate flow rate and density. The heat losses at the
R R R R
39
• (Forced variable): T L1 R
• Remaining variables: T L2 , F L , T s , F s , P s , Q s , F c
R R R R R R R R R R R R R R
The degrees of freedom is therefore 7 – 5 = 2. The extra relations are given by the
relation between the steam flow rate F s with the pressure P s either in open-loop or
R R R R
temperature and pressure passes through a throttling valve where its pressure is
reduced substantially. As a result, part of the liquid feed vaporizes. The two phases
are assumed to be in phase equilibrium. x i and y i represent the mole fraction of
R R R R
component i in the liquid and vapor phase respectively. The formed vapor is drawn
off the top of the vessel while the liquid comes off the bottom of the tank. Taking the
whole tank as our system of interest, a model of the system would consist in writing
separate balances for vapor and liquid phase. However since the vapor volume is
generally small we could neglect the dynamics of the vapor phase and concentrate
only on the liquid phase.
40
Fv
yi
P, T, Vv
Fo
zi
To
Po VL ρL FL
xi
• Liquid-vapor Equilibrium:
Raoult's law can be assumed for the phase equilibrium
xi Pi s
yi = (i=1,2,….,N c )
(2.74)
R R
P
Together with the consistency relationships:
Nc
∑y
i =1
i =1
(2.75)
Nc
∑x
i =1
i =1
(2.76)
41
• Physical Properties:
The densities and enthalpies are related to the mole fractions, temperature and
pressure through the following relations:
ρL R R = f(x i ,T,P)
R R (2.77)
ρv R R = f(y i ,T,P) ≈ M v aveP/R T
R R R RP P (2.78)
Nc
(2.79)
M v ave =
R RP P
∑y M
i =1
i i
Nc
(2.80)
h = f(x i ,T) ≈
R R
∑ x Cp (T − T
i =1
i i ref )
Nc
(2.81)
H = f(y i ,T) ≈
R R
∑ y Cp (T − T
i =1
i i ref ) + λm
Nc
(2.82)
λm R R = ∑yλ
i =1
i i
• Number of equations: 2N c +3: (Eq. 2.71, 2.72, 2.73, 2.74, 2.75, 2.76) R R
Note that physical properties are not included in the degrees of freedom since they
are specified through given relations. The degrees of freedom is therefore (2N c +5)- R R
(2N c +3)=2. Generally the liquid holdup (V L ) is controlled by the liquid outlet flow
R R R R
rate (F L ) while the pressure is controlled by F V . In this case, the problem becomes
R R R R
42
absorption towers, extraction towers and multi-stage flash evaporator where distillate
water is produced from brine by evaporation.
The same modeling approach used for single stage processes will be used for the
staged processes, where the conservation law will be written for one stage and then
repeated for the next stage and so on. This procedure will result in large number of
state equation depending on the number of stages and number of components.
The separation process generally takes place in plate, packed or spray-type towers.
In tray or spay-type columns the contact and the transfer between phases occur at the
plates. Generally, we can always assume good mixing of phases at the plates, and
therefore macroscopic balances can be carried out to model these type of towers.
Packed towers on the other hand are used for continuous contacting of the two phases
along the packing. The concentrations of the species in the phases vary obviously
along the tower. Packed towers are therefore typical examples of distributed
parameters systems that need to be modeled by microscopic balances.
In the following we present some examples of mass separation units that can be
modeled by simple ODE's, and we start with liquid-liquid extraction process.
Liquid-liquid extraction is used to move a solute from one liquid phase to another.
Consider the single stage countercurrent extractor, shown in Figure 2.10, where it is
desired to separate a solute (A) from a mixture (W) using a solvent (S). The stream
mixture with flow rate W (kg/s) enters the stage containing X Af weight fraction of
R R
solute (A). The solvent with a flow rate (S) (kg/s) enters the stage containing Y AfR R
weight fraction of species (A). As the solvent flows through the stage it retains more
of (A) thus extracting (A) from the stream (W). Our objective is to model the
variations of the concentration of the solute. A number of simplifying assumptions
can be used:
• The solvent is immiscible in the other phase.
• The concentration X A and Y A are so small that they do not affect the mass flow
R R R R
rates. Therefore, we can assume that the flow rates W and S are constant. A total
mass balance is therefore not needed.
43
• An equilibrium relationship exists between the weight fraction Y A of the solute R R
in the solvent (S) and its weight fraction X A in the mixture (W). The relationship R R
dX A
ρ1V1 = WX Af − WX A − N A (2.84)
dt
whereas N A (kg/s) is the flow rate due to transfer flow between the two phases. A
R R
similar component balance on the solvent phase of volume V 2 and density ρ 2 gives: R R R R
dYA
ρ2V2 = SYAf − SYA + N A (2.85)
dt
Since Y A = K X A and K is constant, the last equation is equivalent to:
R R R R
dX A
ρ 2V2 K = SY Af − SKX A + N A
dt (2.86)
Adding Eq. 2.84 and 2.86 yields:
dX A
( ρ1V1 + ρ 2V2 K ) = WX Af + SY Af − (W + KS ) X A
dt (2.87)
The latter is a simple linear ODE with unknown X A . With the volume V 1 , V 2 and flow R R R R R R
rates W, S known the system is exactly specified and it can be solved if the initial
concentration is known:
X A (t i ) = X Ai
R R R R R (2.88)
Note that we did not have to express explicitly the transferred flux N A . R R
S, YA S, YAf
W, XAf W, xA
The same analysis can be extended to the multistage liquid/liquid extraction units
as shown in Figure 2.11. The assumptions of the previous example are kept and we
44
also assume that all the units are identical, i.e. have the same volume. They are also
assumed to operate at the same temperature.
1 2 i N
A component balance in the ith stge (excluding the first and last stage) gives:
P P
ρ1iV1i
dX Ai
= WX Ai −1 − WX Ai − N Ai (i=2…,N-1)
(2.89)
dt
where N Ai is the flow rate due to transfer between the two phases at stage i.
R R
ρ 2iV2i
dY Ai
= SY Ai +1 − SY Ai + N Ai … (i = 2…,N-1)
(2.90)
dt
Writing the equilibrium equation (Eq. 2.83) for each component Y Ai = K X Ai R R R
(ρ1iV1i + ρ2iV2i K )
dX Ai
= WX Ai −1 + SYAi +1 − (W + KS ) X Ai (i=2…,N-1)
(2.91)
dt
Since the volume and densities are equal, i.e.:
V 1i = V 1 and V 2i = V
R R R R R R (2.92)
ρ 1i = ρ 1 and ρ 2i = ρ 2
R R R R R R R
(2.93)
Equation 2.91 is therefore equivalent to:
dX Ai
(ρ1V1 + ρ2V2 K ) = WX Ai −1 + SYAi +1 − (W + KS ) X Ai (i=2…,N-1)
dt (2.94)
The component balance in the first stage is:
dX A1
(ρ1V1 + ρ2V2 K ) = WX Af + SYA1 − (W + KS ) X A1
dt (2.95)
And that for the last stage is:
45
dX AN
(ρ1V1 + ρ2V2 K ) = WX AN −1 + SYAf − (W + KS ) X AN
dt (2.96)
The model is thus formed by a system of linear ODE's (Eq. 2.94, 2.95, 2.96) which
can be integrated if the initial conditions are known:
X A (t i ) = X Ai (i=1,2…,N)
R R R R R R (2.97)
stream flows upward with molar flow rate (G) and feed composition (y f ). We are R R
interested in deriving an unsteady state model for the absorber. A simple vapor-liquid
equilibrium relation of the form of:
yi = a xi + b
R R R R (2.98)
Can be used for each stage i (i=1,2,…,N).
Assumptions:
• Isothermal Operation
• Negligible vapor holdup
• Constant liquid holdup in each stage
• Perfect mixing in each stage
According to the second and third assumptions, the molar rates can be considered
constants, i.e. not changing from one stage to another, thus, total mass balance need
46
not be written. The last assumption allows us writing a macroscopic balance on each
stage as follows:
Component balance on stage i:
dxi
H = G ( yi −1 − yi ) + L( xi +1 − xi ) (i=2…,N-1) (2.99)
dt
where H is the liquid holdup, i.e., the mass of liquid in each stage. The last equation
is repeated for each stage with the following exceptions for the last and the first
stages:
L, xf G, yN
stage N
xN yN-1
xi+1 yi
stage i
xi yi-1
x3 y2
stage 2
x2 y1
stage 1
L, x1 G, yf
48
Cw
D
xd
F
z
steam
B
xb
Since the vapor-phase has negligible holdups, then conservation laws will only be
written for the liquid phase as follows:
Energy balance:
d (M f hf ) (2.111)
= V f −1H f −1 − (V f H f + (1 − q) Fh f ) + L f +1h f +1 − ( L f h f + qFh f )
dt
Energy balance:
d ( M 1h1 )
= VB H B − V1H1 + L2 h2 − L1h1 (2.114)
dt
51
Liquid holdup: M i = f (L i ) R R R R
Notation:
Li, Vi
R R R R Liquid and vapor molar rates
Hi, hi R R R R Vapor and liquid specific enthalpies
xi, yi
R R R R Liquid and vapor molar fractions
Mi R R Liquid holdup
Q Liquid fraction of the feed
Z Molar fractions of the feed
F Feed molar rate
Mi R R n
MB, MD R R R R 2
Li R R n
B,R,D 3
x i,j
R R n(nc − 1)
x B,j ,x D,j
R R R 2(nc − 1)
y i,j
R R n(nc − 1)
y B,j
R R nc − 1
hi R R n
hB, hDR R R R 2
Hi R R n
HB R R 1
Vi R R n
VB R R 1
Ti R R n
52
TD, TB R R R R 2
Total 11+6n+2n(nc−1)+3(nc−1)
Equations:
Total Mass n+2
Energy n+2
Component (n + 2)(nc − 1)
Equilibrium n(nc − 1)
Liquid holdup n
Enthalpies 2n+2
Vapor rate n
hB = h1 R R R 1
yB = xB
R R R (nc − 1)
Total 7+6n+2n(nc-1)+3(nc-1)
Constants: P, F, Z
To well define the model for solution we include four relations imported from
inclusion of four feedback control loops as follows:
• Use B, and D to control the liquid level in the condenser drum and in the re-
boiler.
• Use V B and R to control the end compositions i.e., x B , x D
R R R R R
53
R, xd Vn, yn
Vn, yn
stage n
(a)
Qc (b)
R, xd D, xd
Ln, xn Vn-1, yn-1
stage i stage f
(c) (d)
stage 1
(f)
B, xB
(e)
L1, x1 V B , yB
L1, x1
Section 3
Section 2
L2 Figure (2.15) Reboiler of batch distillation column
V1
Reboiler M1
54
Total mass balance.
dM B
= L1 − VB (2.118)
dt
Component balance
d ( M B ⋅ x B ,i )
= L1 x1,i − VB y B ,i (2.119)
dt
Energy balance
d ( M B hB )
= L1h1 − VB H B + Qr (2.120)
dt
Where Q r is heat supplied by reboiler
R R
55
Figure (2.16) continuous packed reactive distillation column.
56
Lm+1 Vm
Lm Vm-1
Consider a fluid flowing inside a pipe of constant cross sectional area (A) as shown
in Figure 2.18. We would like to develop a mathematical model for the change in the
fluid mass inside the pipe. Let v be the velocity of the fluid. Clearly the velocity
changes with time (t), along the pipe length (z) and also with the radial direction (r).
In order to simplify the problem, we assume that there are no changes in the radial
direction. We also assume isothermal conditions, so only the mass balance is needed.
Since the velocity changes with both time and space, the mass balance is to be carried
57
out on microscopic scale. We consider therefore a shell element of width ∆z and
constant cross section area (A) as shown in Fig. 2.18.
ρ(t,z)
∆z
vo v(t,z)
We can check for consistency that the units in each term are in (kg). Dividing Eq.
2.127 by ∆t ∆z and rearranging yields:
(ρA) − (ρA) (ρvA) − (ρvA) (2.128)
t + ∆t t
= z z + ∆z
∆t ∆z
58
∂ρ ∂ ( ρv ) (2.131)
+ =0
∂t ∂z
The equation (2.131) is a partial differential equation (PDE) that defines the variation
of ρ and v with the two independent variables t and z. This equation is known as the
one-dimensional continuity equation. For incompressible fluids for which the density
is constant, the last equation can also be written as:
∂v (2.132)
=0
∂z
This indicates that the velocity is independent of axial direction for one dimensional
incompressible flow.
We reconsider the flow inside the pipe of the previous example. Our objective is to
find the velocity profile in the pipe at steady state. For this purpose a momentum
balance is needed. To simplify the problem we also assume that the fluid is
incompressible. We will carry out a microscopic momentum balance on a shell with
radius r, thickness ∆r and length ∆z as shown in Fig 2.19.
r
∆r
R
r
∆r z
∆z
P|z P|z+∆z
Momentum out:
59
(τrz 2πr∆z)| r+∆r
R R R
(2.134)
As for the momentum generation we have mentioned earlier in section 1.8.3 that the
generation term corresponds to the sum of forces acting on the volume which in this
example are the pressure forces, i.e.
(PA)| z – (PA)| z+∆z = P(2πr∆r)| z − P(2πr∆r)| z+∆z
R R R R R R R
(2.135)
There is no accumulation term since the system is assumed at steady state.
Substituting this term in the balance equation (Eq. 1.6) and rearranging yields,
rτrz |r + ∆r −rτrz |r r ( P |z − P |z + ∆z ) (2.136)
=
∆r ∆z
We can check for consistency that all the terms in this equation have the SI unit of
(N/m2). Taking the limit of (Eq. 2.136) as ∆z and ∆r go to zero yields:
P P
d (rτrz ) dP (2.137)
= −r
dr dz
Here we will make the assumption that the flow is fully developed, i.e. it is not
influenced by the entrance effects. In this case the term dP/dz is constant and we
have:
dP P2 − P1 ∆P (2.138)
= =
dz L L
where L is the length of the tube. Note that equation (2.138) is a function of the shear
stress τrz , but shear stress is a function of velocity. We make here the assumption that
R R
the fluid is Newtonian, that is the shear stress is proportional to the velocity gradient:
τ rz = − µ
dv z (2.139)
dr
Substituting this relation in Eq. 2.137 yields:
d rdv z ∆P (2.140)
µ ( )=r
dr dr L
or by expanding the derivative:
d 2 v z 1 dv z ∆P (2.141)
µ( 2
+ )=
dr r dr L
The system is described by the second order ODE (Eq. 2.141). This ODE can be
integrated with the following conditions:
• The velocity is zero at the wall of the tube
60
vz = 0
R R at r = R (2.142)
• Due to symmetry, the velocity profile reaches a maximum at the center of the
tube:
dvz
=0 at r = 0 (2.143)
dr
Note that the one-dimensional distributed parameter system has been reduced to a
lumped parameter system at steady state.
accumulation term since the system is assumed at steady state. The mass balance
equation is therefore,
(SN A )| z − (SN A )| z+∆z − (S∆z)kC A =0
R R R R R R R R R R
(2.147)
61
Porous
catalyst
particle
Exterior
surface
∆z
z=0 z=L
Figure 2.20 diffusion with chemical reaction inside a slab catalyst
slab. The second condition implies that the concentration is finite at the center of the
slab.
62
2.2.4 Temperature profile in a heated cylindrical Rod
fluid of temperature T m while the surface area of the rod is exposed to ambient
R R
describes the variation of the rod temperature with the position. The metal has high
thermal conductivity that makes the heat transfer by conduction significant. In
addition, the rod diameter is assumed to be large enough such that thermal
distribution in radial direction is not to be neglected. The system is depicted by figure
2.21. For modeling we take an annular ring of width ∆z and radius ∆r as shown in
the figure. The following transport equation can be written:
R
∆r
r
∆r
∆z
63
q r+∆r (2π (r+∆r)∆z)∆t
R R
(2.157)
Heat accumulation:
~ ~
ρ(2π r ∆r ∆z)( h t+∆t − h t ) R R R R
(2.158)
~
Where h is the specific enthalpy. Summing the above equation according to the
conservation law and dividing by (2π ∆r ∆z ∆t), considering constant density, gives:
~ ~
ht + ∆t − ht q − qz + ∆z rqr − rqr + ∆r (2.159)
ρr =r z +
∆t ∆z ∆r
64
Consequently, the above energy balance equation (Eq. 2.165) is reduced to the
following ODE
d 2T (2.166)
0 = kz − U ( 2πRL)(Ta − T )
dz 2
v(t,z)
CAo CA
z z+∆z z=L
ρ(t,z), CA(t,z)
In the following we derive the microscopic component balance for species (A) around
differential slice of width ∆z and constant cross-section area (S).
65
(C A S ∆z) | t+∆t − (C A S ∆z) | t
R R R R R R R
(2.169)
Generation due to reaction inside the shell:
− r(S∆z∆t) (2.170)
Where r = k C A is the rate of reaction.
R R
Substituting all the terms in the mass balance equation (Eq. 1.3) and dividing by ∆t
and ∆z gives:
(C A S ) |t + ∆t −(C A S ) |t (vC A S + N A S ) |z −(vC A S + N A S ) |z + ∆z (2.171)
= − kC A S
∆t ∆z
Taking the limit of ∆t 0 and ∆z 0 and omitting S from both sides give the
following PDE:
∂C A ∂vC A ∂N A (2.172)
=− − − kC A
∂t ∂z ∂z
N A = − DAB
dC A (2.173)
dz
Where D AB is the binary diffusion coefficient. Equation 2.172 can be then written as
R R
follows:
∂C A ∂ ( vC A ) ∂ 2C A (2.174)
=− + DAb − kC A
∂t ∂z ∂z 2
Expanding the derivatives, the last equation can be reduced to:
∂C A ∂C ∂v ∂ 2C A (2.175)
= −v A − C A + DAb − kC A
∂t ∂z ∂z ∂z 2
This equation can be further simplified by using the mass balance equation for
incompressible fluids (Eq. 2.132). We get then:
∂C A ∂C ∂ 2C A (2.176)
= −v A + DAb − kC A
∂t ∂z ∂z 2
The equation is a PDE for which the state variable (C A ) depends on both t and z.R R
The PDE is reduced at steady state to the following second order ODE,
dC A d 2C A (2.177)
0 = −v + DAb − kC A
dz dz 2
The ODE can be solved with the following boundary conditions (BC):
66
BC1: at z = 0 C A (0) = C A0
R R R (2.178)
BC2: at z = L dC A ( z )
=0
(2.179)
dz
The first condition gives the concentration at the entrance of the reactor while the
second condition indicates that there is no flux at the exit length of the reactor.
To T
CAo v(t,z) v(t,z)
CA
Assumptions:
• Kinetic and potential energies are neglected.
• No Shaft work.
• Internal energy is approximated by enthalpy.
• Energy flow will be due to bulk flow (convection) and conduction.
~
(q z A + v Aρ h )∆t| z
R R R R
(2.180)
Energy flow out of the shell:
~
(q z A + v Aρ h )∆t| z+∆z
R R R R
(2.181)
Accumulation of energy:
~ ~
(ρA h ∆z)| t+∆t − (ρA h ∆z)| t R R R R
(2.182)
Heat generation by reaction:
(−∆H r )kC A A ∆z ∆t R R R R
(2.183)
Heat transfer to the wall:
h t (πD∆z)(T − T w )∆t
R R R R
(2.184)
Where h t is film heat transfer coefficient.
R R
Substituting these equations in the conservation law (equation 1.7) and dividing by
Α∆t ∆z give:
~ ~ ~ ~
( ρh ) |t + ∆t −( ρh ) |t ( ρvh ) | z −( ρvh ) | z + ∆z q z | z −q z | z + ∆z πD
= + − ∆H r kC A − ht ( )(T − Tw ) (2.185)
∆t ∆z ∆z A
by:
~
h = C p(T − Tref ) (2.188)
Since the fluid is incompressible it satisfies the equation of continuity (Eq. 2.132).
Substituting these expressions in Eq. 2.186 and expanding gives:
∂T ∂T ∂ 2T πD
ρC p = − ρC pv + kt 2 − ∆H r ko e − E / RT C A − ht ( )(T − Tw ) (2.189)
∂t ∂z ∂z A
68
At steady state this PDE becomes the following ODE,
dT d 2T πD
0 = − ρC pv + kt 2 − ∆H r ko e − E / RT C A − ht ( )(T − Tw ) (2.190)
dz dz A
Similarly to Eq. 2.177 we could impose the following boundary conditions:
B.C1: at z = 0 T(z) = T oR (2.191)
B.C2: at z = L dT ( z )
=0
(2.192)
dz
The first condition gives the temperature at the entrance of the reactor and the second
condition indicates that there is no flux at the exit length of the reactor.
69
Chapter 3: Equations of Change
∆y
∆x
(ρvx)x+∆x
(ρvx)x
∆z
z (ρvy)y
y
(ρvz)z
Mass in:
The mass entering in the x-direction at the cross sectional area (∆y∆z) is
(ρv x )| x ∆y∆z∆t
R R R R (3.1)
The mass entering in the y-direction at the cross sectional area (∆x∆z) is
(ρv y )| y ∆x∆z∆t R R R R (3.2)
The mass entering in the z-direction at the cross sectional area (∆x∆y) is
(ρv z )| z ∆x ∆y ∆t
R R R R
(3.3)
Mass out:
The mass exiting in the x-direction is:
(ρv x )| x+∆x ∆y∆z∆t
R R R R
(3.4)
The mass exiting in the y-direction is:
(ρv y )| y+∆y ∆x∆z∆t
R R R R
(3.5)
The mass exiting in the z-direction is:
(ρv z )| z+∆z ∆x∆y ∆t
R R R R
(3.6)
Rate of accumulation:
The rate of accumulation of mass in the elementary volume is:
(ρ)| t+∆t ∆x ∆y ∆z - (ρ)| t ∆x ∆y ∆z
R R R R
(3.7)
Since there is no generation of mass, applying the general balance equation Eq. 1.2
and rearranging gives:
(ρ| t+∆t − ρ| t ) ∆x ∆y ∆z = (ρv x | x − ρv x | x+∆x )∆y∆z∆t + (ρv y | y
R R R R R R R R R R R R R R R R
(3.8)
− ρv y | y+∆y )∆x∆z∆t + (ρv z | z − ρv z | z+∆z ) ∆x ∆y ∆t
R R R R R R R R R R R R
By taking the limits as ∆y,∆x,∆,z and ∆t goes to zero, we obtain the following
equation of change:
∂ρ ∂ρvx ∂ρv y ∂ρvz
=− − − (3.10)
∂t ∂x ∂y ∂z
Expanding the partial derivative of each term yields after some rearrangement:
∂ρ ∂ρ ∂ρ ∂ρ ∂v ∂v ∂v
+ vx + vy + vz = −ρ( x + y + z ) (3.11)
∂t ∂x ∂y ∂z ∂x ∂y ∂z
This is the general form of the mass balance in cartesian coordinates. The equation is
also known as the continuity equation. If the fluid is incompressible then the density
is assumed constant, both in time and position. That means the partial derivatives of ρ
are all zero. The total continuity equation (Eq. 3.11) is equivalent to:
∂vx ∂v y ∂vz
0 = −ρ( + + ) (3.12)
∂x ∂y ∂z
or simply:
∂vx ∂v y ∂vz
0= + + (3.13)
∂x ∂y ∂z
and ρ (kg/m3s) is the density of the mixture. Our objective is to establish the
P P
∆y
nAx|x+∆x
nAx|x
∆z
y
∆x
nAy|y nAz|z+∆z
x
Mass of A in:
The mass of species A entering the x-direction at the cross sectional (∆y∆z) is:
(n Ax )| x ∆y∆z∆t
R R R R
(3.14)
where n Ax kg/m2 is the flux transferred in the x-direction
R R P P
Similarly the mass of A entering the y and z direction are respectively:
(n Ay )| y ∆x∆z∆t
R R R R
(3.15)
(n Az )| z ∆x∆y∆t
R R R R
(3.16)
Mass of A out:
The mass of species A exiting the x, y and z direction are respectively
(n Ax )| x+∆x ∆y∆z∆t
R R R R
(3.17)
(n Ay )| y+∆y ∆x∆z∆t
R R R R
(3.18)
(n Az )| z+∆z ∆x∆y∆t
R R R R
(3.19)
The rate of accumulation is:
ρ A | t+∆t ∆x ∆y ∆z − ρ A | t ∆x ∆y ∆z
R R R R R R R R
(3.20)
(3.22)
n Ay | y )∆x∆z∆t + (n Az | z+∆z − n Az | z )∆x∆y∆t +r A ∆x ∆y ∆z∆t
R R R R R R R R R R R R R R
Dividing each term by ∆x∆y∆z∆t and letting each of these terms goes to zero yields:
∂ρ A ∂n Ax ∂n Ay ∂n Az
+ + + = rA (3.23)
∂t ∂t ∂t ∂t
We know from Section 1.11.1, that the flux n A is the sum of a term due to convection R R
n A = ρA v + j A
R R R R R (3.24)
Substituting the different flux in Eq. 3.23 gives:
∂ρ A ∂ ( ρ A v x ) ∂ ( ρ A v y ) ∂ ( ρ A v z ) ∂j Ax ∂j Ay ∂j Az
+ + + + + + = rA (3.25)
∂t ∂x ∂y ∂z ∂x ∂y ∂z
For a binary mixture (A,B), Fick’s law gives the flux in the u-direction as :
∂wA
j Au = −ρDAB (3.26)
∂u
where w A = ρ A /ρ. Expanding Eq. 3.25 and substituting for the fluxes yield:
R R R R
∂ρ A ∂v ∂v y ∂v z ∂ρ A ∂ρ ∂ρ
+ ρ A x + + + v x + v y A + v z A
∂t ∂x ∂y ∂z ∂x ∂y ∂z
(3.27)
∂ ∂ρD AB w A ∂ ∂ρD AB w A ∂ ∂ρD AB w A
− ( )+ ( )+ ( ) = rA
∂x ∂x ∂y ∂y ∂z ∂z
This is the general component balance or equation of continuity for species A. This
equation can be further reduced according to the nature of properties of the fluid
involved. If the binary mixture is a dilute liquid and can be considered
incompressible, then density ρ and diffusivity D AB are constant. Substituting the
R R
This equation can also be written in molar units by dividing it by the molecular
weight M A to yield:
R R
∂C A ∂C A ∂C A ∂C A ∂ 2C A ∂ 2C A ∂ 2C A
+ v x + vy + vz − D AB + + = RA
∂
t ∂x ∂y ∂z ∂x 2 ∂ 2
∂ 2 (3.29)
y
z
reaction
accumulation Convection Diffusion
−σxx |x+∆x
σxx |x
−σzx |z+∆z
σyx |y
x
z
Since, unlike the mass or the energy, the momentum is a vector that has three
components, we will present the derivation of the equation for the conservation of the
x-component of the momentum. The balance equations for the y-component and the
z-component are obtained in a similar way. To establish the momentum balance for
its x-component we need to consider its transfer in the x-direction, y-direction, and z-
direction.
Momentum in:
The x-component of momentum entering the boundary at x-direction, by
convection is:
(ρv x v x )| x ∆y∆z∆t
R R R R R R
(3.30)
The x-component of momentum entering the boundary at y-direction, by
convection is:
(ρv y v x )| y ∆x∆z∆t
R R R R R R
(3.31)
and it enters the z-direction by convection with a momentum:
(ρv z v x )| z ∆x∆y∆t
R R R R R R
(3.32)
The x-component of momentum entering the boundary at x-direction, by molecular
diffusion is:
(τxx )| x ∆y∆z∆t
R R R R
(3.33)
The x-component of momentum entering the boundary at y-direction, by molecular
diffusion is:
(τyx )| y ∆x∆z∆t R R R R
(3.34)
and it enters the z-direction by molecular diffusion with a momentum:
(τzx )| z ∆x∆y∆t R R R R
(3.35)
Momentum out:
The rate of momentum leaving the boundary at x+∆x, by convection is:
(ρv x v x )| x+∆x ∆y∆z∆t R R R R R R
(3.36)
and at boundary y+∆y,:
(ρv y v x )| y+∆y ∆x∆z∆t
R R R R R R
(3.37)
and at boundary z+∆z:
(ρv z v x )| z+∆z ∆x∆y∆t R R R R R R
(3.38)
Using the equation of continuity (Eq. 3.10) for incompressible fluid, Equation (3.46)
is reduced to:
∂v x ∂v ∂v ∂v ∂τ ∂τ ∂τ ∂P
ρ + v x x + v y x + v z x = −( xx + yx + zx ) − + ρg x (3.47)
∂t ∂x ∂y ∂z ∂x ∂y ∂z ∂x
∂v z ∂v ∂v ∂v ∂ 2v ∂ 2v ∂ 2 v ∂P
ρ + ρ v x z + v y z + v z z = µ 2z + 2z + 2z − + ρg z
∂
t ∂x ∂y ∂z ∂x ∂y ∂z ∂
z (3.51)
accumulation
generation
transport by bulk flow transport by viscous forces
∆y
∆x
qx |x+∆x
qx |x
∆z
qy |y+∆y
qz |z
The rate of generation is Φ H where Φ H includes all the sources of heat generation, i.e.
R R R R
reaction, pressure forces, gravity forces, fluid friction, etc. Substituting all these terms
in the general energy equation (Eq. 1.7) and dividing the equation by the term
∆x∆y∆z∆t and letting each of these terms approach zero yield:
∂ ( ρCpT ) ∂ ( ρCpTv x ) ∂ ( ρCpTv y ) ∂ ( ρCpTvz ) ∂qx ∂q y ∂qz
+ + + + + + = ΦH (3.60)
∂t ∂x ∂y ∂z ∂x ∂y ∂z
Using the equation of continuity (Eq. 3.10) for incompressible fluids the equation is
reduced to:
∂T ∂T ∂T ∂T ∂q x ∂q y ∂q z
ρCp + vx + vy + vz + + + = ΦH (3.62)
∂t ∂x ∂y ∂z ∂x ∂y ∂z
The energy balance includes as before a transient term, a convection term, a diffusion
term, and generation term. For solids, the density is constant and with no velocity, i.e.
v = 0, the equation is reduced to:
∂T ∂ 2T ∂ 2T ∂ 2T
ρCp = k 2 + 2 + 2 + Φ
∂ ∂ ∂y ∂z generation
H
t x (3.65)
accumulation
Transport by thermal diffusion