B asic C oncepts

Chemical Engineering Fundamentals

Chemical Engineering, R Rawatlal

A study of the conservation equations starts with establishment of the basic concepts: flux, rate,
diffusion and convection. As we’ll find later, these basic ideas can be applied to generate at least four
of the fundamental balances that govern all of chemical engineering. We establish these concepts
through a mass balance to begin with, and then generalise to other balance types later in the course.

1. Fluid and phase concepts

From the more fundamental aspects of chemistry, we have established the concepts
of atoms and molecules as basic units. Of
course, electrons, protons and neutrons
comprise atoms, and even these may be
further expanded upon as subatomic
particles. For the majority of chemical
engineering operations, our focus is on the
metre scale, in which case we have to
acknowledge that our default interest lies in
the behaviour of large groups of molecules.

Noting that 1 mole of any molecule is

6.023 × 1023 molecules, and that, for water,
this is only 18.01 × 10−3 𝑘𝑔, we begin to
realise the staggering number of molecules contained in just one cubic meter.

Therefore, our interest is in what we term the “fluid” or the “phase”, meaning the
body of molecules which can be understood in terms of “macro” state variables like
temperature, pressure and concentration. On the scale of individual molecules, these
concepts have no meaning; on the scale
of large numbers of molecules, these are
useful variables which characterise the
behaviour of the chemicals at hand.

The main point is that we are moving

away from viewing our chemicals as
discrete molecules, and instead as a
continuum of an “averaged” fluid or a
phase. We therefore view the
concentration or the temperature or the
velocity etc. as a field of values (which
may vary in time or across space).

1 | Page
 2. D ensity and concentration
D ensity is defined as the mass per unit volume; it indicates how “intensely” the
mass is packed into a space.
𝑚
𝜌=
𝑉
So even though the mass of a particle might be high, if the particle is small, it can
exhibit high momentum for the same velocity compared with other similarly sized
but lower density particles.

The density of water is close to 1g/cm3, or 1000 kg/m3. That is, 1 ton of water fits in
one cubic metre.

C oncentration is based on the same idea, and counts the same set of objects in a
different way; it is the number of moles “concentrated” into the space:
𝑛
𝑐=
𝑉
Clearly, we can translate between mass and moles through the molar mass: 𝑚 =
𝑀𝑀. 𝑛, so
𝑛 𝑚 𝜌
𝑐= = =
𝑉 𝑀𝑀. 𝑉 𝑀𝑀
Of course, the density and the concentration may vary across time and space:

𝑐 = 𝑐(𝑡, 𝑥, 𝑦, 𝑧)

𝜌 = 𝜌(𝑡, 𝑥, 𝑦, 𝑧)

We may consider a small space (e.g. Δ𝑉 = 1𝑚𝑚3) or a short time range (e.g. Δ𝑡 =
1𝑠). Even in space a small space, the number of molecules is so large (e.g. 1020 ) that
the fluid properties are regarded as “reliable”, or statistically large enough.

2 | Page
 3. C onvection and flux
Even though the number of molecules within a space may be defined and possibly
even constant, the space itself may be moving. Consider, for example, a set of
particles uniformly suspended in a gas, where the gas is in motion at a constant
velocity 𝑢. The particles will adopt the same velocity in that case.

Consider a fluid containing component-𝐴 in motion with linear velocity 𝑢, where 𝑢 is

a vector in 3-space. The concentration in the fluid of 𝐴 is 𝑐𝐴 . Let’s imagine surface
whose normal is in the same direction as 𝑢, and this surface has an area of 𝑎 m2.
After Δ𝑡 = 1𝑠, the number of moles of 𝐴 that will have passed through this surface is

Δ𝑛𝐴 = 𝑢. 𝑐𝐴 . 𝑎. Δ𝑡

The molar flowrate can therefore be written as:

𝑑𝑛𝐴 Δ𝑛𝐴
= lim = 𝑢. 𝑐𝐴 . 𝑎⁡
𝑑𝑡 Δt→0 Δ𝑡

We note, however, that this flowrate is dependent on the choice of area 𝑎. Instead of
first defining an area before computing a flowrate, it is more convenient for us to
introduce a more intrinsic variable, namely, the molar flux:

1 𝑑𝑛𝐴
𝑁𝐴 = = 𝑢. 𝑐𝐴
𝑎 𝑑𝑡
Note that the mass flux 𝑗𝑚 can be computed just as easily:

𝑗𝑚 = 𝑢. 𝜌

These formulations apply to the case where material is in motion due to the bulk
movement of a fluid, and where the material is uniformly suspended in the fluid, i.e.
the actual flowrate can be estimated from the rate of motion of the medium.

C om m ents on dim ensionality

It should be noted that generally speaking, in 3-space, 𝑐𝐴 is a scalar and 𝑢 is a
vector; some texts neglect this under the assumption that all the developments are
within one dimension only. If we wish to develop the balances more generally, we
note then that the flux 𝑁𝐴 is then a vector.

The molar flowrate is then taken as the dot product of the flux and the area:

𝐹𝐴 = 𝑁𝐴 . 𝐴

Which can be taken as the total area multiple with the normal vector from the
surface. In other words, this is the projection of the fluid velocity with the surface

3 | Page
normal vector, i.e. with the area projection onto the area co-linear with the fluid
flow.

In the simulation, we observe that the area of the disc is 𝐴 (scalar); we define the
unit normal vector to the surface as 𝑛. We can write then

⃗ 𝐴 . 𝐴. 𝑛⃗
𝐹𝐴 = 𝑁
=𝑢
⃗ . 𝑐𝐴 . 𝐴. 𝑛⃗

We note that only 𝑢 and 𝑛 are vectors. This dot product provides us with the
projection of 𝑢 onto 𝑛, and is then the fraction of the stream which “is presented” to
the opening with area 𝐴. (Note 𝐹𝐴 , being proportional to a dot product, is a scalar.)

The closer these vectors are to being co-linear, the higher 𝐹𝐴 becomes.

N ote

𝑎. 𝑏⃗ = ∑ 𝑎𝑖 . 𝑏𝑖 = ||𝑎||||𝑏||. 𝑐𝑜𝑠𝜃
𝑖

The reason is based on the area of parallelograms – see maths notes.

4 | Page
Exam ple

Consider a gas stream containing a multi-component mixture at:

linear velocity 𝑢 = 5⁡𝑚. 𝑠 −1 with density 𝜌 = 2⁡𝑘𝑔. 𝑚−3 and mean molar mass

𝑀𝑀 = 54 × 10−3 ⁡𝑘𝑔. 𝑚𝑜𝑙 −1 A dipped tube is introduced to the stream with a total
area of 3𝑐𝑚2.

a) What should the angle be to obtain a flux of 𝑥⁡𝑚𝑜𝑙. 𝑠 −1 . 𝑚−2?

𝜋 𝜋 𝜋 𝜋
b) When the angle between the surface normal and the flow axis is 0, 6 , 4 ⁡, 3 , 2 ,
what will the molar flowrate through the bleed line be?

Solution

The concentration is
𝜌
𝑐=
𝑀𝑀
The molar flux is then (taking the flow to be along the 𝑧-axis):

𝑁𝑧 = 𝑢𝑧 𝑐

For 0, the flowrate is simply the product of the flux and the area:

𝐹 = 𝑁. 𝐴 = 𝑢𝑧 𝑐𝐴
More generally

𝐹 = 𝑢𝑧 𝑐𝐴. 𝑐𝑜𝑠𝜃

5 | Page
Exam ple

Consider a vessel fed by a stream at flowrate 𝑄𝑖𝑛 at concentration of 𝐴 at 𝑐𝐴,𝑖𝑛 . The

cross-sectional area of the vessel is 𝐴𝑐 and the exit flow is gravity driven. The initial
concentration is 𝑐𝐴0 and the vessel is vigorously stirred.

Develop the required material balances and sketch the expected concentration
changes.

Solution

The liquid level is at a height of

ℎ = 𝑉/𝐴𝑐
The volume varies as
𝑑𝑉
= 𝑄𝑖𝑛 − 𝑄𝑜𝑢𝑡
𝑑𝑡
Where

𝑄𝑜𝑢𝑡 = 𝑘√ℎ
The general material balance on component-𝐴 is
𝑑𝑛𝐴
= 𝑐𝐴𝑖𝑛 𝑄𝑖𝑛 − 𝑐𝐴 𝑄𝑜𝑢𝑡
𝑑𝑡
That is
𝑑𝑐𝐴 𝑉
= 𝑐𝐴𝑖𝑛 𝑄𝑖𝑛 − 𝑐𝐴 𝑄𝑜𝑢𝑡
𝑑𝑡
We note that generally both 𝑐𝐴 and 𝑉 are functions of time hence
𝑑𝑐𝐴 𝑑𝑉 𝑑𝑐𝐴
𝑉 + 𝑐𝐴 =𝑉 + 𝑐𝐴 (𝑄𝑖𝑛 − 𝑄𝑜𝑢𝑡 ) = 𝑐𝐴𝑖𝑛 𝑄𝑖𝑛 − 𝑐𝐴 𝑄𝑜𝑢𝑡
𝑑𝑡 𝑑𝑡 𝑑𝑡
That is
𝑑𝑐𝐴 𝑄𝑖𝑛
= (𝑐𝐴𝑖𝑛 − 𝑐𝐴 ) ⁡
𝑑𝑡 𝑉(𝑡)
We note that this equation must be solved simultaneously with
𝑑𝑉
= 𝑄𝑖𝑛 − 𝑄𝑜𝑢𝑡
𝑑𝑡
And the initial conditions 𝑐𝐴 (𝑡 = 0) = 𝑐𝐴0 and 𝑉(𝑡 = 0) = 𝑉0 .

6 | Page
 4. D iffusion
In reality, the motion of a fluid is nothing but the overall motion of all the molecules
in the fluid; indeed, a fluid is the averaged, macroscopic view of the collection of
molecules.

Let’s now consider a fluid at rest.

If we were to consider the motion of specific molecules, we would find there are
departures by individual molecules from the bulk motion of the fluid.

By the definition of temperature, for any temperature above 0⁡𝐾, individual

molecules are in a state of continual random motion. If at some instant in time, there
is a high concentration of a component in some region in space there will be a
relatively high rate of motion of the molecules away from that region when compared
with a region with a lower concentration.

The net result, or the net observation, is that molecules, by virtue of random motion,
will move from a region (𝐴)⁡of high concentration to a region (𝐵) of lower
concentration. We note that if this happens without bound, region 𝐵 will become the
high concentration and region 𝐴 the low. However, this does not happen without
bound; once the concentrations become the same, a rise in concentration in either
region would tend to be result in migration away from that region toward the other
until the two regions are again equal in concentration.

The overall tendency is for the concentrations to become the same, and we may
think of this as a steady state.

Most texts write as if this occurrence was intentional by nature; the observation may
support this view, however we can see it is merely a consequence of the randomness
of the velocities and the fact that high concentrations means simply that there are
more entities in motion and giving the appearance of greater rates of motion.

To formalise this discussion, let’s consider two regions of a continuum separated by

distance Δ𝑥. The first region can be defined to be at location 𝑥, hence the second
region is at location 𝑥 + Δ𝑥. Consider the region at 𝑥 + Δ𝑥 to be higher in
concentration. The net flux will tend to be from 𝑥 + Δ𝑥 to 𝑥:−𝑁𝐴 ∝ (𝑐𝐴,𝑥+Δ𝑥 − 𝑐𝐴,𝑥 ⁡)

We also anticipate that the rate of this observed effect will be inversely proportional
to the distance Δ𝑥, i.e.
𝑐𝐴,𝑥+Δ𝑥 − 𝑐𝐴,𝑥 ⁡
−𝑁𝐴 ∝
Δ𝑥

7 | Page
Note that we’ve used the negative sign, since the movement is from 𝑥 + Δ𝑥 to 𝑥, i.e.
in the direction of the decrease of 𝑥. We’ve therefore adopted the convention that
flux in a particular direction is positive for motion in the direction of the increase
with the direction.

We can now take the limit and introduce a constant of proportionality 𝐷𝐴 :

𝑑𝑐𝐴
𝑁𝐴 = −𝐷𝐴
𝑑𝑥
This is the Fick’s Law of diffusion. It accounts for the flux of molecules that
(apparently) results from a difference of concentrations in space. The sharper the
gradient, the higher the rate of flux; the unfortunately leads many chemical
engineering authors to describe this phenomenon as a “driving force”.

Exam ple

Consider a two-compartment cell containing a liquid with concentrations 𝑐𝐴,1 and 𝑐𝐴,2
in the compartments. A permeable membrane of thickness 2𝑚𝑚 separates the
compartments; the diffusivity of 𝐴⁡through the membrane is 𝐷𝐴 . The two
compartments are stirred vigorously to the extent that the concentration remains
constant within each compartment.

Sketch the expected concentration profiles across the membrane for different
instances in time.

Compute the concentration change as a function of time.

Solution

We estimate the mass flux through the membrane from compartment 1 to 2 as

𝑑𝑐𝐴 𝑐𝐴,1 − 𝑐𝐴,2

𝑁𝐴 = −𝐷𝐴𝑀 ≅ 𝐷𝐴𝑀
𝑑𝑧 Δ𝑧
Where Δ𝑧 is the thickness of the membrane. We may write a balance on each
compartment as follows:

𝑑𝑐𝐴1 𝑁𝐴 𝐴 𝐷𝐴𝑀 𝐴 𝑐𝐴,1 − 𝑐𝐴,2

= =
𝑑𝑡 𝑉1 𝑉1 Δ𝑧
𝑑𝑐𝐴2 𝑁𝐴 𝐴 𝐷𝐴𝑀 𝐴 𝑐𝐴,1 − 𝑐𝐴,2
=− =−
𝑑𝑡 𝑉2 𝑉2 Δ𝑧

Solving these equations in python:

8 | Page
9 | Page
 5. N et flux
We note that this discussion has been for a bulk fluid at rest. The same arguments
hold for a fluid in motion: Even if a bulk fluid is in motion, there may exist
concentration differences across the fluid, in which case, on top of the convective
flux, diffusive flux is also superimposed.

We therefore have

𝑑𝑐𝐴
𝑁𝐴 = 𝑢. 𝑐𝐴 − 𝐷𝐴 .
𝑑𝑥
In more standard terminology, convective flux is written as 𝑗𝑐 and diffusive flux as 𝑗𝑑

𝑁𝐴 = 𝑗𝑐,𝐴 + 𝑗𝑑,𝐴

where
𝑑𝑐𝐴
𝑗𝑐,𝐴 = 𝑢. 𝑐𝐴 and 𝑗𝑑,𝐴 = −𝐷𝐴 . 𝑑𝑥

10 | P a g e
 6. C onvection in m ulticom ponent m ixtures

The total concentration for a multicomponent mix is simply:

𝑐 = ∑ 𝑐𝑖
𝑖

This is a consequence of denominator (the volume) being the same.

𝑛𝑡𝑜𝑡𝑎𝑙 𝑃
For a gas, we have 𝑐 = = 𝑅𝑇
𝑉

We may also define the mole fractions for liquids and gases respectively as
𝑐 𝑐𝑖
𝑥𝑖 = 𝑐𝑖 , 𝑦𝑖 = 𝑐

For a gas, we may also use

𝑝𝑖
𝑅𝑇
𝑐𝑖 𝑝𝑖
𝑦𝑖 = = 𝑃 =
𝑐 𝑅𝑇 𝑃
We also have ∑𝑖 𝑥𝑖 = 1, ∑𝑖 𝑦𝑖 = 1

We can also define the mass fraction as

𝜌𝑖 𝑥𝑖 . 𝑀𝑀𝑖
𝜔𝑖 = =
𝜌 ∑𝑖 𝑥𝑖 . 𝑀𝑀𝑖

We can also calculate mole fractions when given mass fractions as follows:
𝜔𝑖
𝑀𝑀𝑖
𝑥𝑖 = 𝜔
∑𝑖 𝑖
𝑀𝑀𝑖

The molar-average velocity can be obtained as follows

𝑢𝑚𝑜𝑙,𝑎𝑣 ∑ 𝑐𝑖 = ∑ 𝑢𝑖 . 𝑐𝑖
𝑖 𝑖

So

∑𝑖 𝑐𝑖 𝑢𝑖 ∑𝑖 𝑐𝑖 𝑢𝑖
𝑢𝑚𝑜𝑙,𝑎𝑣 = =
∑𝑖 𝑐𝑖 𝑐

The mass averaged velocity for a multicomponent mixture can be obtained as follows

∑𝑖 𝜌𝑖 𝑢𝑖 ∑𝑖 𝜌𝑖 𝑢𝑖
𝑢𝑚𝑎𝑠𝑠,𝑎𝑣 = =
∑𝑖 𝜌𝑖 𝜌

11 | P a g e
 7. D iffusion in m ulticom ponent m ixtures

Consider first a binary mixture in 𝐴 and 𝐵, where 𝐷𝐴,𝐵 is the known diffusivity of 𝐴
through 𝐵. We have
𝑑𝑐𝐴
𝐽𝐴,𝑧 = −𝐷𝐴𝐵
𝑑𝑧
Let’s say the average molar velocity 𝑢𝑧 ⁡was known, and that the velocity specifically
of component 𝐴 was known to be 𝑢𝐴 . Then clearly the diffusion flux can also
alternatively be obtained as

𝐽𝐴,𝑧 = 𝑐𝐴 (𝑢𝐴,𝑧 − 𝑢𝑧 )

Hence

𝑑𝑐𝐴
−𝐷𝐴𝐵 = 𝑐𝐴 (𝑢𝐴,𝑧 − 𝑢𝑧 )
𝑑𝑧
or
𝑑𝑐𝐴
𝑐𝐴 𝑢𝐴,𝑧 = 𝑐𝐴 𝑢𝑧 + (−𝐷𝐴𝐵 )
𝑑𝑧
From the previous section, we can compute the molar average velocity as follows
1
𝑢𝑧 = (𝑐 𝑢 + 𝑐𝐵 𝑢𝐵,𝑧 )
𝑐 𝐴 𝐴,𝑧
That is

𝑐𝐴 𝑢𝑧 = 𝑥𝐴 (𝑐𝐴 𝑢𝐴,𝑧 + 𝑐𝐵 𝑢𝐵,𝑧 )

Substitution yields
𝑑𝑥𝐴
𝑁𝐴,𝑧 = 𝑐𝐴 𝑢𝐴,𝑧 = 𝑥𝐴 (𝑐𝐴 𝑢𝐴,𝑧 + 𝑐𝐵 𝑢𝐵,𝑧 ) + (−𝑐. 𝐷𝐴𝐵 )
𝑑𝑧
More generally, for flow in 3-D,

⃗⃗⃗⃗
𝑁𝐴 = 𝑐𝐴 ⃗⃗⃗⃗
𝑢𝐴 ⁡,
⃗⃗⃗⃗⃗𝐵 = 𝑐𝐵 ⃗⃗⃗⃗
𝑁 𝑢𝐵 ⁡

𝑑 𝑑 𝑑 𝑇
∇= [ , , ]
𝑑𝑥 𝑑𝑦 𝑑𝑧
𝑑𝑥 𝑑𝑥𝐴 𝑑𝑥𝐴
∇. 𝑥𝐴 = 𝑑𝑥𝐴 , ,
𝑑𝑦 𝑑𝑧

Hence

12 | P a g e
 ⃗⃗⃗⃗
𝑁𝐴 = −𝑐𝐷𝐴𝐵 ⃗∇𝑥𝐴 + 𝑥𝐴 (𝑁
⃗⃗⃗⃗𝐴 + ⃗⃗⃗⃗⃗
𝑁𝐵 )

Note that we should recognise 𝑐𝐴 𝑢 ⃗⃗⃗⃗𝐴 + ⃗⃗⃗⃗⃗

⃗ = 𝑥𝐴 (𝑁 ⃗ 𝑥𝐴 is
𝑁𝐵 ) as bulk flux and −𝑐𝐷𝐴𝐵 ∇
diffusion flux, and ⃗⃗⃗⃗
𝑁𝐴 is net flux.

Exam ple

We wish to design a system with a 2𝑐𝑚 diameter 1𝑚 long pipeline which is

immersed in a pool of an infinitely dilute solution. The molar flux from the pipe to
the pool is 3.5 mol.m-2.s-1. We wish to maintain the concentration at the entrance at
32 mol.m-3 and 28 mol.m-3 at the exit, and it may be assumed that the profile is
linear. The diffusivity is 10-5 m2.s-1. What is the molar loss rate from the tube? What
is the velocity?

Solution

The total molar loss rate is from the fluid inside the tube to the outside pool is
through the outer surface of the tube. If this area is 𝐴′ then the molar removal rate
from the tube is

Δ𝑛𝐴̇ = 𝑁𝐴′ 𝐴′
Where

𝐴′ = 𝜋𝐷𝐿
This molar flowrate will reduce the concentration in the tube stream; we can
estimate the mass loss there as follows
𝑑𝑐𝐴 𝑑𝑐𝐴
Δ𝑛̇ 𝐴 = 𝑄(𝑐𝐴,𝑖𝑛 − 𝑐𝐴 ) − 𝐷𝐴 = 𝑢𝐴(𝑐𝐴,𝑖𝑛 − 𝑐𝐴 ) − 𝐷𝐴
𝑑𝑧 𝑑𝑧
Where 𝐴 is the tube cross sectional area:

𝜋𝑑 2
𝐴=
4
If we assume that the gradient difference is constant across the tube (note the
concentration change is somewhat smaller than the absolute mean value) then
𝑑𝑐𝐴 𝑐𝐴2 − 𝑐𝐴1
≅
𝑑𝑧 𝐿
May then solve for 𝑢:
𝑑𝑐𝐴
Δ𝑛̇ 𝐴 + 𝐷𝐴
𝑢= 𝑑𝑧
𝐴(𝑐𝐴,𝑖𝑛 − 𝑐𝐴 )

13 | P a g e
14 | P a g e