Sunteți pe pagina 1din 7

Balance Equations and Conservation of

Mass
Copyright Brian G. Higgins (2004)
Introduction
Balance equations (for mass, momentum, energy, entropy) provide the foundation for much of the physical-based
modeling discussed in transport phenomena courses in chemical engineering. They are also the starting point for develop-
ing qualitative understanding of phenomenological observations in fluid mechanics, heat transfer, mass transfer, and
reaction engineering. Moreover, the mystique of engineering intuition can often be traced back to a general understand-
ing of balance laws. It is therefore imperative that engineering students have a firm understanding of the fundamental
ideas and assumptions used to derive balance laws.
In this tutorial we discuss a general framework for deriving balance laws for single component systems that satisfy the
continuum hypothesis. When the continuum hypothesis (defined in the next section) breaks down, it is necessary to use
concepts from statistical mechanics to derive the necessary balance laws. We address this topic in a future tutorial.
Even though many chemical engineering processes involve multicomponent, multiphase reacting mixtures, it turns out
that studying single component systems is still instructive. The reason is that many of the ideas and principles used to
derive balance laws for single component systems can readily be extended to handle multicomponent, multiphase
systems.
Continuum Hypothesis
The underlying assumption of the continuum hypothesis is that one can define continuous fields for variables such as
density, velocity, stress, temperature, energy, etc. Depending on the variable of interest these fields may be scalar, vector
or tensor functions of position and time. Thus the continuum hypothesis circumvents the need to study the motion of
individual molecules that make up the system under investigation.
When we invoke the continuum hypothesis, the implicit assumption is that the mean free path of molecules (the distance
travelled by molecules between collisions), or the correlation length of molecules (distance over which a given molecule
can influence the behavior of its neighbors) is small compared with the length scale used in the measurement of the
event. For example, suppose we had an instrument that could measure the mean density of water at a point x
0
by
sampling a volume of fluid centered at x
0
. Now, if we construct an imaginary cube with its centroid at x
0
and with sides
of length `, then the mean density of water within this imaginary cube will be
(1) o
`
=
M
`

`
3
where M
`
is the mass of water within the cube. What will our density probe measure as we vary `?
Suppose ` is 1 cm. Then the number of water molecules within our 1 cm cube will be approximately 3 10
22
. If we
reduce ` to 10
-7
cm, we will have approximately 30 molecules in our cube. Depending on the thermal fluctuations, the
reading on our density probe will vary and if we continue to reduce `. In short, we will see oscillations in the density
measurement, depending on the number of water molecules we happen to sample. On the other hand, if
` >> 10
-7
cm, our density probe will determine the mean density by averaging over a large ensemble of molecules
and we would anticipate that o
`
will be independent of `. However, if ` is too large, then spatial variations in density
within our cube will become important and o
`
will no longer be independent of `. Thus, if ` is in the range,
L
MFP
` L
s
, we would anticipate that it is possible to represent the liquid density as a continuous function of
position and time, i.e. o = o H x, tL. Here L
MFP
is a measure of the mean free path or correlation length, and L
s
is a
measure of the macroscopic system under study. For multiphase systems, there will be different scales for different
phases. In those circumstances, it is also possible to define a continuum variable such as density but now the averaging is
done over multiple phases.
BalanceLaws.nb 1
Suppose ` is 1 cm. Then the number of water molecules within our 1 cm cube will be approximately 3 10
22
. If we
reduce ` to 10
-7
cm, we will have approximately 30 molecules in our cube. Depending on the thermal fluctuations, the
reading on our density probe will vary and if we continue to reduce `. In short, we will see oscillations in the density
measurement, depending on the number of water molecules we happen to sample. On the other hand, if
` >> 10
-7
cm, our density probe will determine the mean density by averaging over a large ensemble of molecules
and we would anticipate that o
`
will be independent of `. However, if ` is too large, then spatial variations in density
within our cube will become important and o
`
will no longer be independent of `. Thus, if ` is in the range,
L
MFP
` L
s
, we would anticipate that it is possible to represent the liquid density as a continuous function of
position and time, i.e. o = o H x, tL. Here L
MFP
is a measure of the mean free path or correlation length, and L
s
is a
measure of the macroscopic system under study. For multiphase systems, there will be different scales for different
phases. In those circumstances, it is also possible to define a continuum variable such as density but now the averaging is
done over multiple phases.
Control Volumes
The continuum hypothesis is based on the concept that we can measure a field variable for the continuum by averaging
over an ensemble of molecules sufficiently large to ensure that the field variable is a smooth function of position and
time. Another way of stating this is that we can consider, or "cut out", an arbitrary control volume from the continuum in
which we may define a field variable to be a smooth function of position and time. Our length scale ` for the control
volume must satisfy the continuum hypothesis stated in the previous section. Since we are restricting our analysis to
single component, single phase systems, we exclude regions in our control volume in which variations in field variables
occur so rapidly that they can be regarded as surfaces of discontinuity (e.g. interfaces in heterogeneous media).
There are several possibilities for defining a control volume, based on how we specify the kinematics of the surface
enveloping the control volume. Suppose we select an arbitrary control volume HtL, with an enveloping surface
HtL, defined by its outward directed unit normal x. We specify that the surface of the control volume moves through
the fluid with velocity w . We discuss three possibilities:
Material Control Volumes:
For this case we take w v, the local mass average velocity of the continuum, also called the barocentric velocity. Since
the surface of the control volume moves at the velocity of the continuum, there is no mass flux through the surface and
thus our control volume retains the material originally present within its control surface. We denote such a control
volume by
m
HtL, and its control surface by
m
HtL. Material control volumes are also called Lagrangian control
volumes. They are almost never used in the analysis of problems since they move through space, change shape and
deform. They are, however, used routinely in the derivation of balance laws for continua.
Fixed Control Volumes:
For this case we specify that w 0 everywhere on the control surface. The control volume is therefore fixed in physical
space, and there can be mass flux through portions of the control surface that define inflow and outflow boundaries.
Fixed control volumes define open systems when mass flux can enter or leave through the control surface. We denote
such control volumes by and its control surface by .
Moving Control Volumes:
For this case we specify w 0 on the control surface. Since in general w is not equal to the mass average velocity v of
the continuum, moving control volumes will normally have mass flux passing through portions of the control surface
that define inflow and outflow boundaries. We denote moving control volumes by HtL and its control surface by
HtL.
Reynolds Transport Theorems
The Reynolds Transport theorem allows one to determine how a given quantity defined within a control volume changes
with time as the control volume deforms. We consider first a material control volume
m
HtL with surface
m
HtL and
outward directed unit normal x. As noted previously the surface
m
HtL moves with the continuum mass average
velocity v. Consider a scalar quantity , Hx, tL defined per unit volume. The total amount of , present in the control
volume
m
HtL is
BalanceLaws.nb 2
(2)
HtL =

m HtL
, Hx, tL V
The time rate of change of HtL is
(3)

t
=

m HtL
, Hx, tL V
Since the limits of integration (the size and shape of
m
HtL) are functions of time, we cannot take the time derivative
inside the integral directly. The theorem that allows us to move the time derivative inside the integral is called the
Reynolds Transport Theorem (a three dimensional version of Liebnitz' rule for differentiating 1-D integrals) for material
control volumes:
(4)

m HtL
, Hx, tL V =

m HtL
9
,

t
+ Hv ,L= V
An alternative form for the Reynolds Transport Theorem is found by applying the divergence theorem to the last term on
the RHS of (4):
(5)

m HtL
, Hx, tL V =

m HtL
,

t
V +

m HtL
, v x S
If we have an arbitrary moving control volume HtL with surface HtL, such that the surface moves with velocity w,
then the time rate of change of HtL
(6)

t
=

HtL
, Hx, tL V
becomes
(7)

HtL
, Hx, tL V =

HtL
9
,

t
+ Hw ,L= V
This is known as the generalized Reynolds Transport theorem. Note that the transport theorem applies to scalars, vectors
and tensors. For example, if we are required to find how the vector gHx, tL
(8)
G HtL =

HtL
g Hx, tL V
changes with time, then direct application of the generalized Reynolds Transport theorem gives
(9)
G

t
=

m HtL
g Hx, tL V =

HtL
9
g

t
+ Hw gL= V
Balance Principle
In this section we derive the general form for a balance principle. Our staring point is to consider an arbitrary material
control volume
m
HtL by applying the Euler cut principle to our continuum. As before we let x be the outward
directed unit normal to our control volume, and the surface enclosing
m
HtL is denoted by
m
HtL. Recall that since
we are considering a material control volume, the surface velocity of
m
HtLis equal to the mass average velocity of the
continuum v, evaluated at
m
HtL.
The general form for a balance principle, which applies to all physical circumstances in a material control volume is
(10)

m HtL
o , V =

m HtL
o s V -

m HtL
j x S
Here , is a physical quantity defined per unit mass (may be a scalar, vector, or tensor-valued quantity); o is the contin-
uum density; s is the rate of supply of , per unit mass per unit time; and -j x is the influx of , per unit area per unit
time. The convention used here is that when the flux vector j is pointing outward from the control volume j x is
positive. The minus sign therefore indicates that there is an influx of y per unit area per unit time.
BalanceLaws.nb 3
Here , is a physical quantity defined per unit mass (may be a scalar, vector, or tensor-valued quantity); o is the contin-
uum density; s is the rate of supply of , per unit mass per unit time; and -j x is the influx of , per unit area per unit
time. The convention used here is that when the flux vector j is pointing outward from the control volume j x is
positive. The minus sign therefore indicates that there is an influx of y per unit area per unit time.
We say we have a conservation law if s = 0, and j x = 0.
The above expression is an integral statement of the balance principle. This statement can be formulated as a point
equation by applying the Reynolds Transport theorem in conjunction with the divergence theorem. The time derivative
on the LHS can be expressed as
(11)

m HtL
o , V =

m HtL
9
o,

t
+ Hv o ,L= V
Applying the divergence theorem to the surface integral in our balance principle gives
(12)
m HtL
j x S =

m HtL
j V
Regrouping all terms in our balance principle under the same integral gives
(13)
m HtL
9
o,

t
+ Hv o ,L - o s + j= V = 0
Since our control volume
m
HtL is arbitrary, it then follows that the integrand must be everywhere zero. Thus
(14)
o,

t
+ Hv o ,L - o s + j = 0
This is the point equation for our balance principle. The above equation holds when , is a scalar, vector, or tensor-val-
ued function. so long as the corresponding forms for s and j are used. For example, suppose , is a vector quantity, say
velocity v. Then our balance principle is of the form:
(15)
o v

t
+ Ho v vL - o s + J = 0
In this case the supply of v is given by a vector quantity s and the influx of v is given by a tensor-valued quantity J.
The reader might notice that the first term on the LHS of the above balance principle is related to the time rate of change
of linear momentum. Thus s and J must some how relate to the supply and influx of momentum per unit time. We will
discuss these terms in more detail when we consider specific forms for y in our balance principle.
Conservation of Mass
If o Hx, tL denotes the local density of our continuum, then the total mass in an arbitrary material control volume

m
HtL is given by
(16)
M HtL =

m HtL
o Hx, tL V
For a material control volume, the time rate of change of the total mass is zero, as no mass can be transferred through the
surface
m
HtL (by definition). Hence
(17)
M

t
=

m HtL
o Hx, tL V = 0
If we apply the Reynolds Transport theorem we obtain
BalanceLaws.nb 4
(18)

m HtL
o Hx, tL V =

m HtL
9
o

t
+ Ho v L= V = 0
The point equation expressing conservation of mass is then
(19)
o

t
+ Ho vL = 0
In fluid mechanics this equation is often referred to as the continuity equation. Since o v = v o + o v, we
can express the continuity equation in an alternative form:
(20)
1

o

Do

Dt
+ v = 0
We have use the definition of the material derivative in deriving (20). That is, we use D Dt t + v . The
form of (20) also provides an interesting physical interpretation of the divergence of the velocity field. From (20) we can
write
(21) v = -
1

o

Do

Dt
When written in this form, it is evident that the divergence of the velocity is equal to the unit rate of decrease of density
in the neighborhood of a fluid particle. Thus v is the rate of expansion of a material region, also known as the
dilation rate. The latter view is made clear by considering a material volume
m
HtL moving through a continuum.
(22)

m
HtL =

m HtL
V
The time rate of change of
m
HtLis
(23)

t
=

m HtL
V =

m HtL
v V
We have used the Reynolds Transport Theorem again. Next we invoke the mean value theorem to write the RHS of (23)
as
(24)
m HtL
v V = H vL
-

m
HtL
The superscript (*) denotes that the integrand is evaluated at a suitable interior point of
m
HtL such that (24) holds.
Combining (24) and (23) gives
(25)
1

t
= H vL
-
If we let
m
HtL shrink to zero, then H vL
-
is evaluated at the point of interest, such that
(26) limit
m HtL0
9
1

t
= = v
The LHS of (26) is just the local rate of expansion of a material region - called the dilation rate, which we see is equal to
the divergence of the velocity field!
Finally, if the continuum is assumed to be incompressible so that density in the neighborhood of each material particle
remains constant as it moves, then the equation of continuity simplifies to
(27) v = 0
BalanceLaws.nb 5
This is referred to in fluid mechanics as the condition of incompressibility. Please note that in reality no fluid is incom-
pressible, but rather we invoke incompressibility as an assumption. We also find in the literature the term
"incompressible flow". This is a misnomer as a "flow" is never incompressible. Rather, we assume the fluid behaves, at
least approximately, as if it were incompressible.
Balance Law Using Moving Control Volume
In the previous two sections we use of the balance principle to derive a balance law (or conservation law) using a
material control volume
m
HtL. In this section we illustrate how the same point equation for the conservation of mass
is derived by considering an arbitrary moving control volume, HtL bounded by the surface HtL. In this derivation
w e will need to introduce the concept of mass flux. For an arbitrary moving control volume. Our balance principle
applied to the control volume HtL is
(28) 9
Time rate of change
of mass in HtL
= = 9
net influx of mass
through HtL
=
Since we are not restricting our analysis to a material volume, mass can be transferred across the boundary HtL. In
other words, our control volume HtL defines an open system. In mathematical terms our balance law (28) is
(29)

HtL
o V = -

HtL
j x S
where j is the mass flux vector, as yet unspecified. At the surface HtL we define a mass flux vector n Hx, tL in
terms of the mass average velocity of the continuum and the local density:
(30) n Hx, tL = o Hx, tL v Hx, tL
Recall in our control volume HtL a surface element dS is not fixed, but moves with velocity w Hx, tL, relative to a
given reference frame. Thus the mass flux across that surface must also be relative:
(31) n
rel
= o v
rel
where the relative velocity is given by v
rel
= Hv - wL. Thus the local mass flux vector is
(32) n
rel
j = o Hv - wL
The mass flux, or rate of mass outflow, through a surface element of area dS, at a point on the surface of our control
volume, must depend on the orientation of the surface element relative to the mass flux vector. Thus the differential mass
outflux M through dS is
(33) M = j x S
where x is the outward directed unit normal to the surface. Thus the net influx of mass through the surface HtL is
(34)
9
net influx of mass
through HtL
= = -

HtL
j x S
negative sign accounts for our convention that x is the outward directed unit normal to HtL. Combining (32), (34)
with (29) gives
(35)

HtL
o V = -

HtL
j x S = -

HtL
o Hv - wL _ S
BalanceLaws.nb 6
This is the integral statement of our balance law for mass in a moving control volume. It is instructive to compare this
result with the integral statement of our balance law for mass in a material control volume (17) which is given by
(36)

m HtL
o V = 0
The next task is to show how to derive the point equation for our balance law using (35). First, we need to use the
generalized Reynolds Transport theorem to write
(37)

HtL
o V =

HtL
9
o

t
+ Ho wL = V
Next, we use the divergence theorem to express
(38)
HtL
o Hv - wL S =

HtL
o Hv - wL V
Combining (36) and (37) with (35) gives
(39)
HtL
9
o

t
+ Ho wL + o Hv - wL= V = 0
Cancelling out the common term gives
(40)
HtL
9
o

t
+ ov= V = 0
Since HtL is arbitrary, the integrand must vanish to give
(41)
o

t
+ o v = 0
Summary Remarks
What we have shown is that the point equation describing the balance of mass does not depend on the choice of control
volume. We obtained the same result (see eqns (40) and (19)) whether we used a moving control volume or a material
control volume. The obvious advantage for using a material control volume is that the derivation of the conservation
principle (particular the point equation) simplified considerably . It is for this reason that derivations of balance laws
make exclusive use of material control volumes. On the other hand when we need to apply an integral statement of a
balance principle to solve a given problem, it almost always involves using either a stationary or moving control volume.
References
These notes were compiled from a number of sources. The following were most helpful
1. G.K. Batchelor, Introduction to Fluid Dynamics, Cambridge University Press, 1967
2. S. R. de Groot and P. Mazur, Non-Equilibrium Thermodynamics, Dover Publications, 1984
3. L. E. Malvern, Introduction to the Mechanics of a Continuous Medium, Prentice -Hall, Inc, 1969
BalanceLaws.nb 7

S-ar putea să vă placă și